Anda di halaman 1dari 184

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review

1st edition, May 2005

This document presents the findings from a systematic review of public health interventions to promote the duration of breastfeeding. It was undertaken by the Health Development Agency (HDA) but published after the functions of the HDA were transferred to NICE on 1 April 2005. Neither this document nor the summary version represent NICE guidance.

This document is also published on the National Institute for Health and Clinical Excellence website at: www.nice.org.uk

Published by: National Institute for Health and Clinical Excellence MidCity Place 71 High Holborn London WC1V 6NA Copies of this publication are available to download from the National Institute for Health and Clinical Excellence website: www.nice.org.uk ISBN 1-84629-000-7 National Institute for Health and Clinical Excellence 2005. All rights reserved. This material may be freely reproduced for educational and not-for-profit purposes within the NHS. No reproduction by or for commercial purposes is allowed without the express permission of the National Institute for Health and Clinical Excellence.

National Institute for Health and Clinical Excellence The National Institute for Health and Clinical Excellence (NICE) is the independent organisation responsible for providing national guidance on the promotion of good health and the prevention and treatment of ill health. NICE produces guidance in three areas of health: Public health guidance on the promotion of good health and the prevention of ill health for those working in the NHS, local authorities and the wider public and voluntary sector Health technologies guidance on the use of new and existing medicines, treatments and procedures within the NHS Clinical practice guidance on the appropriate treatment and care of people with specific diseases and conditions within the NHS.

Authors Mary Renfrew, Lisa Dyson, Louise Wallace, Lalitha DSouza, Felicia McCormick and Helen Spiby Mother and Infant Research Unit University of York Health Services Research Centre University of Coventry Most of the work on this review was carried out while several of the authors worked in the Mother and Infant Research Unit, University of Leeds

ii

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Foreword

n April 2005 the functions of the Health Development Agency (HDA) were transferred to the National Institute for Clinical Excellence to form the National Institute for Health and Clinical Excellence (NICE). The new organisation will continue the work of the HDA on building the evidence base in public health and developing evidence-based guidance for the NHS and others on public health matters. This publication is one of the first public health outputs of the new organisation. It is a systematic review of the literature on public health interventions to promote the duration of breastfeeding. This systematic review is a follow-up to work begun by the HDA when it published an evidence briefing on the promotion of breastfeeding in 2003 (Protheroe et al., 2003). It became apparent when reviewing the literature on breastfeeding that there was a considerable gap in the review-level evidence about the duration of breastfeeding. This systematic review was commissioned to fill this gap. The publication of this review by NICE signals both the continuity of the work of the the old HDA and the new NICE, and also the continuing public health commitment to this important topic. We would be pleased to hear from readers who would like to comment on matters relating to the content of this review. Comments can be made on our website: www.nice.org.uk

he government is fully committed to the promotion of breastfeeding, which is accepted as the best form of nutrition for infants to ensure a good start in life. Breastmilk provides all the nutrients a baby needs and exclusive breastfeeding is recommended for the first six months of an infants life. Breastfeeding also has health benefits for both infants and mothers, as it promotes health and prevents disease in both the long and short term. Despite an increase in breastfeeding initiation rates since 1980, there is still a sharp decline in breastfeeding following birth, with continuation rates lowest among lower socio-economic groups, contributing to inequalities in health and the perpetuation of the cycle of deprivation. This systematic review builds on the earlier Enabling women to breastfeed report (Renfrew et al., 2000) but expands on this work to include public health, health promotion and public policy interventions, as well as clinical interventions and interventions targeted at the health sector. It therefore acknowledges the broad community of practitioners, policy makers and community organisations who wish to make a difference to breastfeeding and ensures that their practice is based on the best available evidence. I know that this review will be a useful resource for breastfeeding mothers and to all those working together to support mothers in continuing to breastfeed their babies.

Professor Michael P. Kelly Director, Centre for Public Health Evidence National Institute for Health and Clinical Excellence

Christine Beasley Chief Nursing Officer

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

iii

Acknowledgements

We are grateful for the help and support of the following colleagues: Health Development Agency: Mike Kelly, Caroline Mulvihill, Antony Morgan, Amanda Killoran and Tricia Younger Advisory group: Phyll Buchanan, Jane Putsey, Christine Carson, Sally Inch, Janet Fyle, Caroline Healy, Felicity Savage-King, Amanda Sowden and Krystyna Henderson Royal Society of Medicine: Beata Kloska and Nicola Howson Friends and associates who have translated papers: Anne Renfrew, Naoko Natsume, Martina Dolapchieva, Lucy Yang Research assistants: Daniel Stringer, Louise Clarke, Heather Marshall Friends and colleagues at the Health Sciences Library, University of Leeds: Martin Gill, Rosemary Campell-Blair, Susan Mottram; and at the British Library, Boston Spa Friends and colleagues at the Mother and Infant Research Unit: Mike Woolridge, who contributed to data extraction and thinking through Section 3; Helen Simpson, who provided her usual expert administrative input; and Stephanie Bridgford, who provided excellent support throughout the project, and managed the production of the final report with skill, dedication and calm. Extracts from Enabling women to breastfeed (Renfrew et al., 2000) have been used in this review, with permission. The Health Development Agency funded this work.

iv

The effectiveness of public health interventions to promote the duration of breastfeeding Evidence briefing 1st edition February 2005

Contents

Foreword Acknowledgments Summary Introduction Breastfeeding and health Breastfeeding rates Exclusive breastfeeding Why do women discontinue breastfeeding? Why are there problems with breastfeeding? Implications for this review Methods Aim and objectives Outcomes Framework for the study Reviews conducted and inclusion criteria 1 Public health interventions 2 Public policy interventions 3 Clinical interventions 4 Health professional and lay breastfeeding educator/counsellor training, education and practice change Methods of the study Developing the sections Presentation of the results Summarising the findings 1 Public health interventions 1.1 Support 1.1.1 Background 1.1.2 Evidence base 1.1.3 Findings 1.1.3.1 Additional postnatal support interventions 1.1.3.2 Breastfeeding support interventions: from health professionals 1.1.3.3 Breastfeeding support interventions: from peer supporters/ volunteer counsellors 1.1.3.4 Breastfeeding support interventions: from peer supporters and health professionals working together

iii iv 1 5 5 6 9 9 10 12 13 13 13 14 14 14 16 17 19 19 20 21 22

24 24 25 26 26 26 27 28

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1.1.4 Discussion 1.1.5 Conclusions 1.1.6 Future research 1.2 Educational interventions 1.2.1 Background 1.2.2 Evidence base 1.2.3 Findings 1.2.3.1 Written information about breastfeeding 1.2.3.2 Breastfeeding self-assessment tool 1.2.3.3 Antenatal paediatric visit 1.2.3.4 Group antenatal sessions 1.2.4 Discussion 1.2.5 Future research 1.3 Multifaceted interventions 1.3.1 Background 1.3.2 Evidence base 1.3.3 Findings 1.3.3.1 Antenatal education and postnatal support 1.3.3.2 Postnatal education and support 1.3.3.3 Education, with or without support, plus incentives 1.3.3.4 Antenatal education and hospital discharge packs 1.3.4 Discussion and conclusions 1.3.5 Future research 1.4 Community based interventions 1.4.1 Background 1.4.2 Evidence base 1.4.3 Discussion and conclusions 1.4.3.1 Employment practices and interventions in the workplace 1.4.3.2 Future research promoting and supporting breastfeeding at work 1.4.3.3 Community and school education interventions 1.4.3.4 Mass media campaigns 1.4.3.5 Interventions to support mothers to breastfeed in public 1.4.3.6 Future research community and school interventions 1.5 The organisation of healthcare provision 1.5.1 Background 1.5.2 Evidence base 1.5.3 Results 1.5.3.1 Postnatal ward organisation 1.5.3.2 Post-discharge care studies where the focus of intervention includes support for breastfeeding 1.5.4 General implications for research 2 Public policy interventions 2.1.1 Background 2.1.2 Evidence base 2.1.3 Findings 2.1.3.1 Commercial marketing of breastmilk substitutes 2.1.3.2 Large area and national evaluation of health service policy

29 32 32 36 36 36 37 37 37 38 39 40 41 44 44 44 45 45 47 47 48 49 49 53 53 54 54 54 55 56 57 57 58 59 59 60 60 60 61 63 66 66 68 68 68 69

vi

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

3 Clinical interventions 3.1 Antenatal preparation for breastfeeding 3.1.1 Background 3.1.2 Evidence base 3.1.3 Findings from previous reviews 3.1.4 Discussion and conclusions 3.1.5 Future research 3.2 Interventions in childbirth 3.2.1 Background 3.2.2 Evidence base 3.2.3 Findings 3.2.3.1 Continuous support in labour 3.2.3.2 Delayed clamping of the umbilical cord 3.2.3.3 Birth by caesarean section and psychosocial outcomes 3.2.3.4 Unintended effects of epidural analgesia during labour: a systematic review 3.2.4 Discussion and conclusions 3.2.5 Future research 3.3 Prevention of clinical problems 3.3.1 Evidence base 3.3.2 Findings 3.3.2.1 Timing of the first feed 3.3.2.2 Kangaroo care and skin-to-skin contact for healthy term newborns 3.3.2.3 Prevention of sore nipples 3.3.2.4 Frequency and duration of breastfeeding 3.3.2.5 Additional fluids for breastfed babies 3.3.2.6 Methods of giving supplementary fluids 3.3.2.7 Pacifier use 3.3.2.8 Breast pumping to establish breastfeeding in women at risk of delayed lactation 3.3.2.9 Postoperative analgesia after caesarean section 3.3.2.10 Growth, weight and food charts 3.4 Solving breastfeeding problems 3.4.1 Evidence base 3.4.2 Findings 3.4.2.1 Painful breasts and nipples 3.4.2.2 Treatment of nipple infection 3.4.2.3 Breast engorgement 3.4.2.4 Mastitis 3.4.2.5 Breast pain 3.4.2.6 Milk insufficiency 3.4.3 Conclusions 3.5 Baby related problems 3.5.1 Evidence base 3.5.2 Findings 3.5.2.1 Hypoglycaemia 3.5.2.2 Jaundice
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

75 75 75 76 76 76 78 78 78 79 79 79 79 80 80 81 83 83 83 83 84 86 88 88 90 91 93 94 94 98 98 98 98 100 101 102 103 103 105 107 107 107 107 108 vii

3.5.2.3 3.5.2.4 3.5.2.5 3.5.2.6

Crying and colic Downs syndrome Ankyloglossia, or tongue tie Cleft lip and palate

109 110 110 112 115 115 117 117 117 118 118 119 119 123 125 127 128 137 137 137 138 138 140 141 144 163 165 165 168 171 172 173 174 175 176

3.6 Maternal problems that may affect breastfeeding 3.6.1 Background 3.6.2 Evidence base 3.6.3 Discussion 3.6.4 Future research 4 Health professional and lay breastfeeding educator/counsellor training, education and practice change 4.1.1 Background 4.1.2 Evidence base 4.1.3 Findings 4.1.4 Discussion and conclusions 4.1.5 Future research Overview Overview tables Limitations and strengths of this review Limitations Strengths Discussion and implications Effective and appropriate care in breastfeeding Problems with the evidence base Future research References Glossary Appendixes 1 2 3 4 5 6 7 8 On 9 10 11 12 13 Search strategy Excluded studies Ongoing studies Studies published after search completion Developed country list Standard postnatal care in the UK UNICEF/WHO Ten Steps to Successful Breastfeeding WHO breastfeeding definitions NICE website Pre-screen form Quality appraisal forms (templates) Data extraction forms (templates) Quality appraisal forms for included studies Data extraction forms for included studies

viii

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Summary

Introduction
Breastfeeding has a major role to play in public health. It promotes health and prevents disease in both the short and long term, for both infant and mother. But breastfeeding, especially the prolonged, exclusive breastfeeding that results in the greatest benefits, is far from universally practised in the UK and other western cultures. Breastfeeding initiation rates in the UK are around the lowest in Europe, with rapid discontinuation rates for those who do start. Further, initiation and continuation rates are lowest among families from lower socio-economic groups, adding to inequalities in health and contributing to the perpetuation of the cycle of deprivation. The reasons for this are multifaceted and include the influence of society and cultural norms, as well as clinical problems, the organisation of health services and the lack of preparation of health professionals and others to support breastfeeding effectively. This paper summarises the findings of a systematic review of interventions to enable women to continue breastfeeding, with special reference to women from disadvantaged groups where rates are lowest. Full details are in NICE (2005). It follows on from the previous Health Development Agency (HDA) review of systematic reviews of interventions to promote the initiation of breastfeeding (Protheroe et al., 2003). This information will be of interest to pregnant women, new mothers, health and social care professionals, and lay advisers supporting women in their decision to start and continue breastfeeding. This systematic review will inform the development of the following NICE guidance:

Antenatal and postnatal mental health: clinical management and service guidance Postnatal care: routine postnatal care of recently delivered women and their babies Intrapartum care: management and delivery of care to women in labour Guidance for midwives, health visitors, pharmacists and other primary care services to improve the nutrition of pregnant and breastfeeding mothers and children in low-income households.

Methods
The identification of studies that can inform practice and policy for the support of breastfeeding among women from disadvantaged groups was a priority for the review. A thorough search, data extraction and analysis were conducted. All included papers were reviewed by at least two members of the team. Around 55,000 citations were identified from the electronic search and a further 8,000 from hand searching; 940 papers were pre-screened and full data extraction was carried out on 138 papers. A total of 80 eligible studies (including three reviews) were finally included. Only 17 studies (21%) examined the needs of women from disadvantaged groups. Only 10 studies (12.5%) were conducted in the UK.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Findings
One of the main findings of this review is the great extent of the evidence gap relating to disadvantaged groups. Ways of raising breastfeeding rates among groups where the rates are lowest remain to be explored further. Although there are evidence gaps identified across all the sections, they are widest in clinical issues, public policy and those that address womens key concerns and problems. There is an urgent need for research into clinical problems, including insufficient milk, sore nipples, engorgement and the breastfeeding needs of babies and mothers with particular health needs. There is very little research to inform any aspect of public policy. Two other important findings emerge across all sections. First, there are effective and ineffective interventions. Second, a gap in the evidence base identified across all the reviews is an understanding of the views of those most involved childbearing women and their families, and the staff who care for them whose voices are largely silent in relation to the interventions that might be effective. Practices and policies that have been shown to be effective/beneficial for enhancing breastfeeding duration Postnatal hospital stay Skilled breastfeeding support, peer or professional, proactively offered to women who want to breastfeed (Dennis et al., 2002; Porteous et al., 2000) Preventing the provision of discharge packs containing formula-feeding information and samples (Bliss et al., 1997) Unrestricted feeding from birth onwards (Renfrew et al., 2000) Unrestricted mother-baby contact from birth onwards (Renfrew et al., 2000) Unrestricted kangaroo care/skin-to-skin care from birth onwards (Renfrew et al., 2000) Avoiding supplementary fluids for babies unless medically indicated (Howard et al., 2003) Regular breast drainage/continued breastfeeding for mastitis (Renfrew et al., 2000) Antibiotics for infective mastitis (Renfrew et al., 2000)

Postnatal care in the community Skilled breastfeeding support, peer or professional, proactively offered to women who want to breastfeed (Porteous et al., 2000) Ongoing care in the community Skilled breastfeeding support, peer or professional (Serafino-Cross and Donovan, 1992) Forms of care/practices/policies that appear to be promising and well grounded in theory for enhancing the duration of breastfeeding In pregnancy Group, interactive, culture-specific education sessions (Rossiter, 1994) Group education sessions on positioning and attachment (Duffy et al., 1997) Antenatal education individually tailored to the needs of low-income women (Brent et al., 1995) Immediate postnatal care Basing prevention and treatment of sore nipples on principles of positioning and attachment (Henderson et al., 2000) Cabbage leaves/extract for treatment of engorgement (Roberts et al., 1995, 1998) Systemic antibiotics for infected nipples (Livingstone and Stringer, 1999) Postnatal care in the community Self-monitoring daily log for women from higher socioeconomic groups (Pollard, 1995) Combination of supportive care, teaching breastfeeding technique, rest and reassurance for women with insufficient milk (Renfrew et al., 2000) Division of the frenulum in infants with signs of congenital ankyloglossia [tongue tie] and breastfeeding difficulties (Ballard et al., 2002; Fitz-Desorgher, 2003; Masaitis and Kaempf, 1996)

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Wider social/political issues National policy of encouraging maternity units to adhere to the UNICEF Baby Friendly Initiative (BFI) (Britten and Broadfoot, 2002) Regionally/nationally determined targets with supporting activities, and/or penalties and/or incentives (Cattaneo and Buzzetti, 2001; Giovannini et al., 2003) Multifaceted interventions (across time periods and types of interventions) Tailored antenatal education combined with proactive postnatal support in hospital and the community (Fredrickson, 1995) Combining antenatal education with partner support, postnatal support and incentives for women in lowincome groups (Sciacca et al., 1995) Forms of care/policies that may be ineffective or harmful for enhancing breastfeeding duration (as shown by good but not conclusive evidence) In pregnancy Self-help manual used alone (Coombs et al., 1998) Antenatal education by a paediatrician (Serwint et al., 1996) Providing materials produced by formula milk companies on infant feeding in early pregnancy (Howard et al., 2000) Immediate postnatal care

Ongoing care in the community Dopamine antagonists for insufficient milk (Renfrew et al., 2000) Forms of care/practices/policies shown to be ineffective or harmful for breastfeeding duration In pregnancy Conditioning nipples in pregnancy (Renfrew et al., 2000) Hoffmans exercises for inverted and non-protractile nipples in pregnancy (Renfrew et al., 2000) Breast shells for inverted and non-protractile nipples in pregnancy (Renfrew et al., 2000) Immediate postnatal care Restricting the timing and/or frequency of breastfeeds (Renfrew et al., 2000) Restricting mother/baby contact from birth onwards (Renfrew et al., 2000) Routine use of supplementary fluids (Howard et al., 2000) Provision of discharge packs containing samples or information on formula feeding (Bliss et al., 1997) Topical agents for the prevention of nipple pain (Renfrew et al., 2000) Breast pumping before the establishment of breastfeeding in women at risk of delayed lactation (Chapman et al., 2001) Multifaceted interventions

Separating mothers and babies for treatment of jaundice (Renfrew et al., 2000) Postnatal care in the community Written educational materials used alone (Hauck and Dimmock, 1994) GP clinic visit at one week postpartum (Gunn et al., 1998) Single home visit by community nurse following early discharge (Gagnon, 2002) Dopamine antagonists for insufficient milk (Renfrew et al., 2000)

Combined antenatal education and limited postnatal telephone support for high-income women and women who intend to breastfeed (existing high rates suggest resources are better spent elsewhere) (Rojjanasrirat, 2000)

Conclusion
The extent of the work needed to change the current patterns of infant feeding should not be underestimated. These patterns have been developed over the past century and are now embedded in the thinking and behaviour of several generations of practitioners and in society as a whole. A coordinated and well-supported programme will 3

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

be needed if real culture and practice change is to occur. To enable women to breastfeed the evidence suggests that the following changes are needed: Coordination of national with local policy so that departmental policy is funded, enabled and monitored at the level of, for example, PCTs, Sure Starts, and acute trusts, with a two-way flow of information to enable both a bottom-up and a top-down approach Ongoing monitoring of rates of variation in infant feeding, with agreed definitions and timing of followup, combined with socio-demographic data. It will also require the wholehearted involvement and support of: Clinical professionals in community and hospital settings Community based workers including Sure Start staff Managers with responsibility for health and social services and staff Those with responsibility for collecting health and health service-related data Educators in the fields of health and social services; schoolteachers and those responsible for the school curriculum in primary and secondary schools Employers in large and small organisations Politicians and policy makers at local, regional and national levels Those with influence over public opinion Families and the public at large.

Including both breastfeeding outcomes and examination of the views of participants in future studies will help to inform the potential for interventions to make a difference in practice Much more information is needed about the content of interventions, the training needed to implement them, the participants, and the settings in which they were used Research is needed to test out the efficacy of interventions based on empirical and theoretically derived research Research is needed to compare policy interventions across different countries as well as research within healthcare systems

Implications for future research


The quality of the included studies was variable. Some were designed to provide reliable evidence of effectiveness of the interventions tested, but others were too small, methodologically flawed or used inappropriate designs. Future research should use appropriate designs and consider carefully the range of factors involved in breastfeeding, and funding agencies should be aware of the pitfalls in this field To address the evidence gap, breastfeeding needs to become a priority for a range of funding bodies in the UK Future studies should include examination of cost effectiveness Studies should examine the effectiveness of interventions among different disadvantaged groups 4
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Introduction

Breastfeeding has a major role to play in public health, as it promotes health and prevents disease in both the short and long term for both infant and mother. But breastfeeding, especially the prolonged, exclusive breastfeeding that results in the greatest benefits, is far from universally practised in the UK and other western cultures. Breastfeeding initiation rates in the UK are around the lowest in Europe, with rapid discontinuation rates for those who do start. Further, initiation and continuation rates are lowest among families from lower socio-economic groups, adding to inequalities in health and contributing to the perpetuation of the cycle of deprivation. The reasons for this are multifaceted and include the influence of society and cultural norms, as well as clinical problems, the organisation of health services and the lack of preparation of health professionals and others to support breastfeeding effectively. This review seeks to identify effective interventions that will enable women to continue breastfeeding, with special reference to women from disadvantaged groups where rates are lowest. It follows on the from the previous HDA review of systematic reviews of interventions to promote the initiation of breastfeeding (Protheroe et al., 2003); together these two reviews form a sound evidence base for policy makers, practitioners, and community organisations who wish to make a difference to breastfeeding.

evidence of a dose-response relationship, whereby exclusive breastfeeding is associated with the lowest rates of these illnesses in the first six months of life (Raisler et al., 1999). Breastfeeding protects against some important childhood diseases, including juvenile-onset insulin-dependent diabetes mellitus (Sadauskaite-Kuehne et al., 2004; Mayer et al., 1988; Virtanen et al., 1991), respiratory disease (Wilson et al., 1998), raised blood pressure (Wilson et al., 1998) and obesity (Fewtrell, 2004; Heinig and Dewey, 1996; Von Kries et al., 1999; Gilman et al., 2001; Arenz et al., 2004). Breastfeeding also protects the baby from developing a range of diseases in later life, including atopic disease (Fewtrell, 2004; Lucas et al., 1990; Saarinen and Kajosaari, 1995) and raised blood pressure (Fewtrell, 2004; Martin et al., 2004), protection which is particularly important where there is an underlying predisposition to disease. Several studies have found a consistent, positive correlation between breastfeeding and cognitive development in children and young adults, an effect that seems to increase with the duration of breastfeeding (eg Anderson et al., 1999; Fewtrell, 2004), although it is difficult to know how far such results are confounded by other factors. In addition to the effects on the baby, breastfeeding has an impact on maternal health (Labbok, 2001). It can act as an effective and safe means of birth spacing if guidelines are adhered to (Kennedy et al., 1989) while also protecting women against certain forms of cancer, including breast cancer (Newcombe et al., 1994; United Kingdom National Case Control Study Group (UKNCCSG), 1993; Beral, 2002) and epithelial ovarian cancer (Gwinn et al., 1990; Rosenblatt and Thomas, 1993). Breastfeeding babies have been shown to wake more often at night, which has in turn been linked to adverse outcomes for womens mental health (Alder, 1984). This effect is likely to be modified by the sleeping arrangements made for 5

Breastfeeding and health


Breastfeeding provides babies with a complete source of nutrition that changes in synchrony with each babys growing needs. It protects babies from common childhood infections such as gastroenteritis and respiratory disease (Howie et al., 1990; Kramer et al., 2001b; Cesar et al., 1999) and urinary tract infection (Marild et al., 2004; Pisacane et al., 1992; Marild et al., 1990). There is some

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

babies; if the baby is in a separate cot in another room the effect of waking at night will be greater than if the baby sleeps in close proximity to the mother. Others have found that encountering breastfeeding problems and stopping breastfeeding before the mother wishes to can result in distress and more frequent unhappiness and mental illness (Houston, 1984; Green et al., 1998). A positive breastfeeding experience, therefore, may have different mental and emotional outcomes than a negative experience. It can also be observed that breastfeeding offers the natural circumstances for the development of close, affectionate ties between the mother and her child. Such closeness may be especially important for those living in circumstances of social deprivation, and where there is low self-esteem. Convincing research is lacking on the impact of infant feeding on the family and on the mothers mental health and her relationship with the child, for which both exploratory and controlled in-depth studies are needed. Replacing breastfeeding is impossible; while it is possible to feed babies with artificial substitutes, many of the attributes of breastfeeding are lost as a result. Breastmilk substitutes differ not only in their nutrient composition and their lack of immune factors, but also in the practical aspects of their manufacture, transport, storage and the feeding vessel used. These aspects have been addressed in studies spanning several decades (eg Miller and Chopra, 1984; Renfrew et al., 2003; Randerson, 2004), and contamination of formula milk with a range of microorganisms has been the subject of a recent WHO workshop (World Health Organization, 2004). This work demonstrates that the impact of feeding babies with substitutes for breastmilk is hazardous and carries the risk of multiple adverse outcomes. There is an important public health question about the costs related to infant feeding. Breastmilk substitutes obviously result in an increased use of resources for families and for the health services. When the larger picture is taken into account, to include issues such as absence from work related to childhood illness and the impact on the health of the population in the long term, there is likely to be an economic impact on wider society, the economy and the environment. Although we still await sound, large-scale studies in which all relevant aspects are taken into account, several studies have demonstrated the increased costs of artificial feeding, even though the costs considered are confined only to the costs of excess ill health on health services (eg Ball and Wright, 1999; Hoey and Ware, 1997; Riordan, 1997). 6

There is widespread international acceptance of the evidence for the health benefits arising from prolonged exclusive breastfeeding and recognition of the far-reaching public health benefits (eg World Health Organization, 1991a; American Academy of Pediatrics, 1997). In the UK, several professional and government bodies have recognised the importance of the issue for many years (eg Standing Committee on Nutrition of the British Paediatric Association,1994; Ministry of Agriculture, Fisheries and Food, 1994; Faculty of Public Health Medicine, 1995; Department of Health, 1999). Such support has not reached throughout the system, however, and it is only in recent years that the important part that breastfeeding has to play in public health has been properly recognised by the NHS. A new target has been set in the Department of Healths Priorities and Planning Framework 2003-2006 for England to deliver an increase of two percentage points per year in breastfeeding initiation rates, focusing especially on women from disadvantaged groups (Department of Health, 2003). This first step in raising initiation rates must be matched by efforts to raise continuation rates, or there is a risk that women may start to breastfeed, encounter problems and stop, with resulting distress, loss of ongoing health benefits and negative feedback to other women. A further policy goal has been adopted, as announced by the public health minister in May 2004, of supporting exclusive breastfeeding up to the age of six months, replacing the previous advice that additional foods should be introduced at four to six months. This raises the question of how to encourage women to breastfeed exclusively for that length of time.

Breastfeeding rates
Care should be taken when interpreting any data on breastfeeding rates. Studies of rates often have a number of problems, which include: The sample studied they may not be representative of the country at large, in socio-economic background, ethnicity or region The definitions of breastfeeding and formula feeding used in data collection these may not differentiate between exclusive breastfeeding, partial breastfeeding, or the continuation of any breastfeeds (Auerbach et al., 1991). There are WHO definitions of breastfeeding

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

which can be used to Figure 1: Prevalence of breastfeeding by mothers socio-economic classification distinguish between these important categories (World Health Organization, 1991b)* The way in which data are collected data collection methods may rely on long-term recall and on methods requiring literacy skills or fluency in one specific language The time at which the data were collected there have been shifts in breastfeeding rates over the past few years in some countries, rendering previous surveys questionable. Source: Hamlyn at al., 2002 Breastfeeding rates in the UK and Ireland are among the lowest in Western Europe, and indeed worldwide (Hamlyn et al., 2002). Data are available on infant feeding rates over the past 25 years in the UK countries from the five-yearly surveys conducted by the governments department of statistics since 1975. England and Wales have been included in these surveys from the start. Scotland was included from 1980 and Northern Ireland from 1995. There has been a slight rise in breastfeeding initiation rates overall since 1980. However, although figures uncorrected for social class from the sample show a real increase this is not apparent in England and Wales if appropriate corrections are applied. Real increases have taken place in Scotland and Northern Ireland. The corrected rate for initiation for the UK overall is 62% (71% uncorrected). There has, however, been no improvement in the sharp decline in breastfeeding following birth in England and Wales. Only 43% of women were still breastfeeding at all at six weeks after birth (Hamlyn et al., 2002) compared to 44% in 1995. This contrasts with a real improvement in Scotland, where rates have risen from 36% in 1995 to 40% in 2000 (Hamlyn et al., 2002). The median rate of breastfeeding in the UK is around one month, as compared with other European countries of five months or over, including Belgium, Germany, Switzerland and Denmark (Nicoll et al., 2002). The high initiation and continuation rates in Scandinavia as a
* The WHO definitions of breastfeeding are given in Appendix 8.

whole are well known initiation rates there are well over 90%. The UK figures are much more comparable to rates in the US, where around 68% of women were reported as starting to breastfeed, with only 59% doing so one week after birth (Ross Laboratories, 2002; Declercq et al., 2002), than to our closest neighbours. An added problem is that in the UK today, and indeed more widely throughout much of western Europe and the US, babies from the least affluent families, who are most at risk of ill health, are those least likely to be breastfed and to continue to breastfeed (Hamlyn et al., 2002). There has been little change in the stark social class differences in rates over the past 25 years; the persistent gradient in both initiation and duration rates mirrors social classification of the babys parents, whether this is measured by classifying the occupation of the father (as was done up until 1995) or the mother (as was done in 2000). The low rates of breastfeeding in the UK have been resistant to the range of international and national initiatives that have taken place over the past 25 years, perhaps due in part to the incomplete implementation of these developments in the UK. International initiatives have included the WHO code of marketing of breastmilk substitutes (World Health Organization,1981b), the Innocenti Declaration on the Protection, Promotion and Support of Breastfeeding (World Health Organization,1991a), the UNICEF Baby Friendly Initiative (launched in the early 1990s) and the WHO Global Statement on Infant Feeding (World Health Organization, 2003). 7

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Factors associated with continuation or early cessation of breastfeeding (Hamlyn et al., 2000) Socio-economic background Lower socio-economic group of the mother and of her partner is associated with lower initiation and continuation rates Younger mothers are least likely to breastfeed, and most likely to discontinue early Mothers of second or later babies breastfeed for longer than mothers of first babies Mothers who breastfed their previous child for longer than six weeks continue to breastfeed their later child for longer Duration of breastfeeding is longest among those who left full-time education after the age of 18, and shortest among those who left school at 16 or less In the UK, white mothers are less likely to initiate breastfeeding and those who do so then continue for a shorter duration than mothers from other ethnic groups. In other countries, patterns of feeding differ; they are affected by the mothers cultural background, and the length of time the mother has spent in that country Breastfeeding mothers who were entirely bottlefed themselves are more likely to stop in the first two weeks after birth Since epidural anaesthetic became common for caesarean section there are no significant differences in breastfeeding according to mode of birth. Women who received a general anaesthetic, however, are less likely to breastfeed The shortest duration of breastfeeding is among women who go back to work when the baby is between six weeks and four months old

Age Parity Previous breastfeeding

Education

Ethnicity/cultural background

Whether mothers were themselves breastfed as babies Mode of birth

Employment status

National initiatives in all four countries of the UK were summarised in Protheroe et al. (2003). They include the appointment of national breastfeeding coordinators in Scotland, Northern Ireland and Wales; the development of national breastfeeding strategies in those three countries; and a Department of Health-led Infant Feeding Initiative and the appointment of an Infant Feeding Advisor in England. It is only in very recent years that any real shift has happened, and that shift is still mainly to be found in Northern Ireland and Scotland, where increases in initiation and duration among women in lower social class groups have occurred. There has been an increase in initiation of breastfeeding among the social class III non-manual group in England and Wales, and a trend in this direction among lower socio-economic groups (III manual, IV and V), but not yet any clear evidence of a shift among lower socio-economic groups in duration (Protheroe et al., 2003). 8

One other encouraging trend is towards more women who previously formula fed choosing to breastfeed a subsequent baby (from 18% in 1995 to 26% in 2000) across all countries in the UK (Protheroe et al., 2003); again, evidence for a shift in duration is lacking as yet. A range of maternal and infant-related factors have been consistently found to be associated with the continuation or early cessation of breastfeeding. Details for the UK can be found in Hamlyn et al. (2002), and these findings are consistent across a wide range of developed countries. They are listed in summary in the table above. Other factors affecting the duration of breastfeeding include the gestation and wellbeing of the baby, social and cultural factors and the healthcare available to families see the following pages for more details.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Percentage of women exclusively breastfeeding in selected countries in Europe Country Austria 1998 Denmark 1992 Finland 1995 Germany 1997 Iceland 1998 Italy 1996 Sweden 1997 UK 2000* One month 92 73 Two months 85 68 42 75 94 29 81 25 Four months Six months 46 44 68 33 48 26 69 16

10

42 10

Sources: Yngve and Sjostrom (2001) and Hamlyn et al. (2002). *Note that the definition of exclusive breastfeeding in this survey (Hamlyn et al., 2002) related only to milk
feeds; the figures given here are therefore of mothers who did not give formula, but they may have been giving solids, water or other non-milk drinks

Exclusive breastfeeding
Figures discussed so far have been for the duration of any breastfeeding. The World Health Organization recommends that wherever possible infants should be fed exclusively on breastmilk until six months of age (World Health Organisation, 2003), and the UK government supports this recommendation. This is a major challenge; at present in the UK, figures suggest that 25% of breastfed babies are breastfed exclusively at six to eight weeks of age, and 16% at three to five months. This contrasts with other European countries, as shown in the table above. These figures are overestimates, however; babies who did not receive formula are the basis of the figures, but they may have received solids or water. They do not, therefore, give a true reflection of exclusive breastfeeding. There are at present two breastfeeding-related policy recommendations in England. The first is the target to increase the rate of initiation of breastfeeding by 2% per annum from April 2004, especially among disadvantaged groups. The other is the recommendation of encouraging women to breastfeed exclusively up to six months of age. Advocating both an increase in initiation rates and an increase in duration of exclusive breastfeeding may be difficult to achieve simultaneously. Some women may wish to breastfeed but not to continue to breastfeed exclusively; a recent study of women living in deprived communities in the UK (Dyson et al., 2003) found that many pregnant women in these communities planned to mix breast and formula feeding, and indeed did continue to do this. The effect of working to increase rates of exclusive breastfeeding on overall prevalence of any breastfeeding

is not known; there is no evidence in the UK about the views of women, or practitioners, on the feasibility of exclusive breastfeeding for six months. It has been suggested by some practitioners that some women may not attempt to breastfeed at all if they believe they must feed exclusively. It is likely to be important, at least in the short term, not to emphasise the exclusivity message over the importance of breastfeeding at all, and it will be essential to gather evidence from women and from practitioners about the impact of these two policy messages.

Why do women discontinue breastfeeding?


Reasons given in the most recent UK national survey of breastfeeding (Hamlyn et al., 2002) for discontinuing breastfeeding are shown in the table overleaf. Of these, the most common and the most consistently reported worldwide are insufficient milk and painful breasts and nipples, accompanied by the category would not suck/rejected the breast, which is often a part of the syndrome which results in both painful nipples and the mothers feeling that she does not have enough milk. These figures do not reflect the true extent of the problem; mothers who continue to breastfeed also experience difficulties. For example, a further 12% of women who continued to feed reported that they had experienced painful nipples and breasts (Hamlyn et al., 2002). It is of course likely that the reasons given for discontinuation in this and other large surveys reflect only one part of the story. Breastfeeding problems are often multifactorial in causation and in detailed studies women 9

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Reasons given for discontinuing breastfeeding (% of women discontinuing), birth-4 months, UK Reason for discontinuing breastfeeding Insufficient milk Painful breasts or nipples Baby would not suck/rejected breast Breastfeeding took too long/tiring Mother was ill Did not like breastfeeding Domestic reasons Baby was ill Difficult to judge how much baby drunk Baby could not be fed by others Returned to work/study
Source: Hamlyn et al. (2002)

Less than one week 26 27 37 12 8 7 3 6 3 1 1

One to two weeks Two to six weeks Six weeks to four months 40 28 18 16 13 3 9 12 2 1 0 53 23 13 24 10 1 11 6 0 4 2 52 8 10 17 5 2 6 4 0 8 19

will describe a sequence of events leading up to their decision to stop in which physical factors interrelate with psychological factors and the support or otherwise that they receive (Houston, 1984). They may also experience a number of different problems at the same time, such as sore nipples, engorgement, slow weight gain as observed in routine weighing of the baby, and anxiety about the baby getting enough milk. It is likely, therefore, that womens reasons for stopping are more complex than can be recorded in large postal surveys such as those undertaken by the Office for National Statistics. Some of these symptoms could reflect a true milk insufficiency. But the prevalence of the diagnosis is not in accord with the fact that the majority of women are physiologically capable of producing enough milk. Such a diagnosis often reflects, therefore, inappropriate care or misplaced concern about milk supply. The underlying reasons for this are described in the next section.

Physiological processes milk synthesis and release


Milk synthesis is influenced by hormonal and autocrine factors, which in turn are influenced by the appetite and feeding patterns of the baby (Hamosh and Goldman, 1986; Woolridge, 1995; Peaker and Wilde, 1995). Together, these factors control the supply and demand mechanism; the more the baby feeds, the more milk is synthesised, and vice-versa. This results in a close interplay between physiology and behaviour the babys behaviour, and any patterns imposed on it by advice given to the mother, or by maternal preference, will have an immediate impact on the milk supply, either to stimulate or suppress. Once milk is synthesised in the breast it is then stored in the alveoli and actively released by the hormone oxytocin; this is the let-down reflex (McNeilly, 1977). Oxytocin is released by the action of the baby feeding and also by psychological stimuli such as the mother hearing her baby cry. It follows that psychological stimuli are likely to also inhibit the production of oxytocin, although experimental evidence for the impact of mild stress on inhibiting milk release in humans is lacking. The number of women who are incapable of synthesising or releasing enough milk to feed their baby is not known accurately. But there is good evidence from societies in which breastfeeding is normal practice, well supported by families and society, that in the right circumstances almost all women are capable of producing an adequate milk supply (Akre, 1989; World Health Organization, 2003). The rapid discontinuation of breastfeeding in the first few weeks after the birth of the baby due to insufficient milk,

Why are there problems with breastfeeding?


Any analysis of this situation, and possible remedies, must be considered within the context of an understanding of the physiological processes underlying breastfeeding, the healthcare practices that may support or interfere with it and the social and psychological influences which affect women, their families, health professionals and indeed anyone in the wider social context (Akre, 1989; World Health Organization, 2003; Lawrence, 1994). These issues are described in Renfrew et al. (2000) and a brief summary only will be given here. 10

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

therefore, is not the result of a widespread physiological inability to produce enough milk, but is often the result of psychosocial/cultural factors and practice. That said, more research is needed to identify specific maternal conditions associated with impaired milk supply, such as polycystic ovarian disease and the sequelae of breast surgery, and to provide additional, effective support and preventive care. This will include challenging the current fashion for cosmetic breast surgery.

having enough milk and experiencing problems such as engorgement, sore nipples and inconsolable babies. This is a profound example of iatrogenic problems resulting simply from the inappropriate care given to mothers and babies. Trust and confidence in the ability of womens bodies to feed babies was lost; this affected women themselves, health professionals, family members and society as a whole. Breastfeeding rates declined to very low levels in the 1950s and 1960s and in some communities virtually disappeared. It was not until the late 1970s that real concern surfaced about breastfeeding. Since that time, efforts have been made to increase rates, but as described above without great success so far in the UK and many other countries. Some countries, notably in Scandinavia, made early and successful efforts to combat the problem. The problem for the UK and other countries where rates remain low is that the experience of several decades of very low rates has resulted in deep-rooted problems for women, families, health and social care professionals, and the loss of a supportive environment in which breastfeeding is seen as the norm. Media portrayals often highlight the sexualisation of the breast and the normality of formula feeding (Henderson et al., 2000). In many communities in the UK, breastfeeding is seen as embarrassing, even socially unacceptable and immoral by some (Dyson et al., 2003), and it can be challenging to do in public settings. For some women, notably the young and those living in deprived circumstances, this includes feeding in their own home in front of family, friends and visitors. Women report that they fear breastfeeding can attract an unwanted sexual voyeurism, and interviews with men substantiate these fears (Dyson et al., 2003). The past century also saw a change in womens role, with opportunities to work outside the home that were new, rewarding and brought women financial freedom and enhanced status. It is often said that this is a major reason for the decline in breastfeeding. There is, however, a lack of evidence to substantiate this. UK studies have not reported that return to work was a major factor in not initiating or in stopping breastfeeding prematurely (Martin and White, 1988; White et al., 1992; Foster et al., 1997; Hamlyn et al., 2002), and a range of international studies have found similar results (van Esterik and Greiner, 1981). High rates of womens employment can co-exist with high rates of breastfeeding (eg Scandinavia, Australia); in Western countries, this is normally associated with 11

Pain free, effective feeding


To feed effectively without hurting the mother, the baby needs to take a large enough mouthful of breast tissue well into the mouth so that the nipple is free from friction and potential for damage. This will also ensure an effective milking action, enabling the baby to release not only the large volume, low fat foremilk, but also the essential low volume, high fat hindmilk (Woolridge, 1986a/b). One consequence of the decline in breastfeeding described in the next section seems to be that many women in Western cultures have not seen a baby being breastfed before they have their own child. They therefore have little opportunity to see the best way to hold, position and attach a baby to the breast before trying it for themselves. A widespread and erroneous belief that breastfeeding is painful in the early days after birth seems to have resulted from this, as women need to be taught the basics of positioning and attachment (see Glossary) that would come more easily to them if they had seen other women breastfeeding.

Healthcare practices, social and cultural influences


Throughout the last century, advice given to breastfeeding women by health professionals and the routine care of women in hospital contributed in large part to the decline of breastfeeding. Routine care was influenced by a fashion in childrearing to keep babies separate from their mothers, which in turn was strengthened by the development of near-universal hospital birth in many countries. There was a concomitant development of easily available breastmilk substitutes and a concern reflected in medical texts and hospital practices to measure infants food intake and control infant behaviour. As a consequence, babies were routinely kept in hospital nurseries, the frequency and length of feeds was restricted, and breastfed babies were given additional fluids. As a result, the supply and demand and let-down mechanisms underlying breastfeeding were compromised, resulting in a huge increase in the number of women identified as not

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

adequate maternity leave to allow this to occur. For women to breastfeed exclusively for the first six months, they require the entitlement to six months paid maternity leave, with provision for expressing if they return to work before that time (as in International Labour Organization (ILO), 2000, Maternity Protection Convention). National policies for supporting breastfeeding working mothers vary widely across Europe (Nicoll et al., 2002). In a recent review of European policies, it was noted that the UK was the EU country with the least compliance with ILO standards, though it has recently extended maternity leave to six months (EU, 2003). This widespread lack of knowledge of and confidence in breastfeeding, and its lack of acceptance as a social norm, (see Glossary) has remained endemic in western culture, and undermines all efforts to support women in breastfeeding. Resultant problems include practical issues such as the lack of weight reference charts for breastfed infants. Charts in current use are based on cohorts of babies who were predominantly formula fed, although it is known that patterns of weight gain of breastfed babies differ from formula-fed babies (Kramer et al., 2002; Cole et al., 2002). This can result in breastfed babies being assessed as faltering in their growth, rather than more accurately assessed as following a normal breastfed growth pattern. Very recently, charts have been developed based on the growth of breastfed babies, though these remain to be tested in routine use (Fry, 2003); and the publication of feeding-specific charts based on international data collected by the World Health Organization is awaited. Other problems include peoples attitudes in a culture where formula feeding is seen as the norm. Although most people will state when asked that breastfeeding is a healthier option for babies, there is a widespread acceptance of formula feeding as equivalent and in some regards better than breastfeeding. These attitudes, prevalent throughout culture and widely represented in the media (Henderson et al., 2000) can affect everyone, including children (one study Gregg, 1989 found such ambivalent attitudes already existed among 11 year old children), childbearing women, their families, health professionals, employers and policy makers alike. An understanding of this situation is crucial to identifying ways to intervene, and to increase breastfeeding rates.

Implications for this review


To examine interventions which may impact on breastfeeding duration, therefore, this review must of necessity be very wide ranging. Supporting or undermining breastfeeding can result from the actions of many different kinds of people and organisations. This will include health professionals in medical, nursing, midwifery and health visiting roles and their managers; social care professionals such as those in the Sure Start programme (see Glossary) and social workers; and the child care sector, including childminders and nurseries. The voluntary sector, including lay support workers and counsellors, also have a role to play, as do families themselves, and the wider public. Those responsible for public policy, in health, education and the environment, also have influence. The activities of the commercial sector will have an impact, resulting from advertisements for breastmilk substitutes and attitudes reflected in the media. Employers and those responsible for public places influence the support or resistance offered to women when working, travelling, shopping and eating out. The review must therefore include public health, health promotion and public policy interventions, as well as clinical interventions and interventions targeted at the health sector. The review must consider not only interventions that are intended to have a supportive effect (eg training health professionals in breastfeeding support skills), but also those that may have a detrimental effect (eg healthcare practices such as separating mothers and babies for routine procedures and the care given to women experiencing problems). There is evidence that antenatal interventions, and interventions in childbirth, can have an impact on breastfeeding duration (eg Fairbank et al., 2000; BernardBonnin et al., 1989), and therefore these periods must be included along with care after birth if this study is to inform appropriate care. Further, it is important to include studies that examine the promotion of the duration of breastfeeding, as well as those intended to support women who have started to breastfeed.

12

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Methods
The factors outlined so far all have important implications for the types of studies to be included, as appropriate research approaches differ across these topic areas. It follows that this is a large review indeed, more a series of linked reviews than one single review which has required careful planning about its scope and methods. This work builds on and develops a number of previous reviews conducted by members of this research team. Most notably, results of a review conducted for the Department of Health (Renfrew et al., 2000) was used as a basis for this review and findings incorporated where appropriate.

3 Clinical interventions i) ii) iii) iv) v) vi) Antenatal care Interventions in childbirth Prevention of problems: routines in care postpartum Solving breastfeeding problems Baby related problems Maternal problems likely to affect breastfeeding

4 Health professional and lay breastfeeding educator/counsellor training, education and practice change

Outcomes
Studies are only included where there are breastfeeding duration outcomes.*/** This is important, as there are a wealth of studies concerning practices that affect breastfeeding but where there are no breastfeeding duration outcomes. Such studies may provide evidence of links in a causal chain, such as studies of the impact of training initiatives on staff attitudes to breastfeeding, but the direct impact on breastfeeding outcomes remains to be demonstrated in controlled studies. Where no studies exist, reference is made to studies (excluded from the review) that include intermediate outcomes, to indicate areas for future research. Breastfeeding up to six months of life (at all, and exclusively) are the main outcomes of interest. Many papers have not reported outcomes for as long as six months after birth. It is considered that an increase in breastfeeding of any length may confer a health gain and therefore all duration outcomes are included, whether presented as mean duration or as numbers discontinuing at specific points in time. Other outcomes of interest include duration of breastfeeding beyond six months, supplementation with

Aim and objectives


The aim of this study is to systematically review interventions relevant to (ie affecting positively or negatively) the continuation of breastfeeding, including promotion, protection and support, and with particular reference to interventions that will work with women from disadvantaged groups.* The care of premature and sick babies has been excluded.** The following areas have been reviewed: 1 Public health interventions i) ii) iii) iv) v) Support for pregnant and postpartum women Antenatal and postnatal education Multifaceted programme-based interventions Interventions targeted at community groups The organisation of healthcare provision

2 Public policy interventions


* With the exception of the sections on clinical problems (Sections 3.3, 3.4 and 3.5), where it has been assumed that outcomes such as painful nipples and engorgement will have a direct impact on the duration of breastfeeding, and where duration outcomes are not always recorded. ** One important methodological issue is that in measuring duration we are limited by the information provided in research reports. Commonly, the figures given are of the percentage of women who started to breastfeed and who are still continuing, reflecting a point prevalence at a specific time, such as one week, four months, etc, after birth. Such figures do not inform the reader of the relationship with initiation rates. Survivorship curves would be a better method of reporting such data, or at least adjusting for incidence in each group.

* Disadvantaged groups is used throughout this document to describe any group experiencing social/material disadvantage as a result of, for example, ethnicity, poverty, age, or regional variation. ** It is acknowledged that the care of this group of babies is especially important in addressing issues of health inequality, as families living in disadvantaged circumstances experience a higher rate of premature birth and growth retardation. The literature relating to feeding for this group of babies is, however, very extensive and could not be included in this present review.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

13

food or fluids and maternal and infant-related outcomes likely to affect breastfeeding duration, including sore nipples and breasts, problems with milk supply and problems with the infants weight gain. The views of women included in the studies, and of their caregivers, have been examined if they were measured in studies in which duration was also measured. Studies that only examined the views of caregivers and women have been excluded.

Dates for searches: 1990 to June 2003.* The publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs. This topic is appropriate for an RCT design and there are a number of good quality RCTs available. Other literature has been drawn on as appropriate in the discussion. Population group studied: pregnant and postpartum women from all groups, with a special interest in women from disadvantaged groups. Countries included: developed countries.** Care systems and circumstances in developing countries are very different; studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for the baby, health outcomes for the baby, etc), relevant health service outcomes including economic outcomes. 1.2 Antenatal and postnatal classes/education intended to promote and support the duration of breastfeeding (ie not general antenatal/postnatal classes) Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. Publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing

Framework for the study


This study comprises a series of linked systematic reviews, which have been then summarised in a comprehensive overview.

Reviews conducted and inclusion criteria


Each section of the review required decisions to be made about studies to be included. Here, we describe each section and the rationale for the framework used for each in terms of the following inclusion criteria: Topic areas and interventions covered Including primary studies or existing reviews Dates for searches Study design of included studies Population group studied (eg mothers, fathers, babies, communities, disadvantaged groups) Types of countries included (eg developed/developing) Outcomes primary, secondary, process The information given below for each section of the review is summarised at the start of the report on each section.

1 Public health interventions


1.1 Support offered to pregnant and postpartum women by individuals and agencies to increase the duration of breastfeeding, including peer support and health professional support Primary studies or reviews: primary studies. Any existing reviews were scrutinised for references and additional information.

* The date at which the search was completed. ** See Appendix 5 for list of countries considered to be developed.

14

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs. This topic is appropriate for an RCT design and there are a number of RCTs available. Other literature was drawn on if appropriate in the discussion. Population group studied: pregnant and postpartum women from all groups, with a special interest in women from disadvantaged groups. Countries included: developed countries. Care systems and circumstances in developing countries are very different; studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for the baby, health outcomes for the baby, etc), relevant health service outcomes including attendance at classes, economic outcomes. 1.3 Multifaceted programme-based social services interventions, including some activities relevant to the promotion and support of breastfeeding duration, targeted at specific, relevant groups, including WIC, Sure Start, Welfare Food Scheme, etc Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. The publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs. This topic is appropriate for an RCT design and there are a number of RCTs available. Other literature was drawn on if appropriate in the discussion.

Population group studied: pregnant and postpartum women offered the support of one or more of these programmes, from all groups, with a special interest in women from disadvantaged groups. Countries included: developed countries. Care systems and circumstances in developing countries are very different; studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for baby, health outcomes for baby, etc), relevant health service outcomes including uptake of programme elements, economic outcomes. 1.4 Interventions intended to promote and support breastfeeding targeted at community groups, such as schools, media campaigns, etc Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. Publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). This topic area may need a non-RCT design under some circumstances as it is difficult to provide large-scale interventions such as this in an RCT framework. Population group studied: all relevant target groups, including schoolchildren, specific population groups and the community at large. Countries included: developed countries. Societal norms and the culture in which infant feeding decisions are made are very different in developing countries, and studies from these countries are less likely to be able to inform care in the UK. 15

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those providing and experiencing the programmes (including attitude change to breastfeeding by participants), relevant process outcomes including uptake of programme elements, economic outcomes. 1.5 The organisation of healthcare provision intended to promote and support breastfeeding duration, including rooming-in, breastfeeding only wards, etc. Interventions with a wider focus than breastfeeding, but which may include an effect on breastfeeding outcomes (eg continuity of care) were excluded Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. Publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). This topic area may need a non-RCT design under some circumstances, as it is difficult to provide large-scale organisational interventions such as this in an RCT framework. Population group studied: all postpartum women included in such studies. Countries included: developed countries. Care systems and circumstances in developing countries are very different, and studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including health service and economic outcomes.

2 Public policy interventions


2.1 The provision of maternity/paternity/parental leave, implementation of the WHO code on the marketing of breastmilk substitutes, mother-friendly work policies, the promotion of formula milk, and restriction of availability of formula milk to prescription-only Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. Publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). These topic areas are likely to need a nonRCT design, as it is difficult to provide large-scale policy interventions such as this in an RCT framework. National case-studies are included where available. Population group studied: countries experiencing relevant policy change. Countries included: developed countries. Care systems and circumstances in developing countries are very different, and studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including economic outcomes.

16

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

3 Clinical interventions
3.1 Antenatal practices, including antenatal expression of colostrum, nipple preparation, and treatment for inverted and non-protractile nipples Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1995 to June 2003. This topic was covered in Renfrew et al. (2000) up to 1997. A two-year overlap was allowed to ensure studies published, but not yet entered on the databases by 1997, were identified. Study design of included studies: RCTs only. Population group studied: pregnant women. Countries included: all countries. There is little reason to expect developed and developing countries to differ greatly in the outcomes of these clinical practices. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes. 3.2 Interventions in childbirth, including place of birth, continuity of care in labour, pain relief in labour, interventions including induction, amniotomy, perineal care, food and drink in labour, and mode of birth where breastfeeding has been an outcome measured Primary studies or existing reviews: existing reviews only have been included. This is a large topic area and resources for this study did not permit a review of primary studies. Dates for searches: 1990 to June 2003. Care in labour changed significantly in the 1990s as a result of Changing Childbirth (Department of Health, 1993) and, together with changes in public health policy, has resulted in work before that time being of less relevance. Study design of included studies: reviews of RCTs, nonRCTs with concurrent controls, and before-after studies. Some topics (eg induction of labour) can be effectively

studied using RCTs. Others (eg place of birth) need different methodological approaches. Population group studied: women in labour/at birth. Countries included: developed countries. Care systems and circumstances in developing countries are very different; studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including economic outcomes. 3.3 Prevention of problems; routines in care postpartum, including skin to skin contact, timing of the first feed, prevention of sore nipples including teaching positioning and attachment to mothers, frequency and duration of breastfeeding, supplementing the baby, use of weight and food charts, methods of giving additional fluids, use of pacifiers Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1995 to June 2003. These topics were covered in Renfrew et al. (2000) up to 1997. A two-year overlap was allowed to ensure studies published, but not yet entered on the databases by 1997, were identified. Study design of included studies: RCTs only. Population group studied: breastfeeding mothers of healthy babies. Countries included: all countries. There is little reason to expect developed and developing countries to differ greatly in the outcomes of these clinical practices. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes. 3.4 Solving breastfeeding problems including sore nipples, infection, sore breasts, engorgement, mastitis, breast abscess, and insufficient milk, and including the 17

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

use of interventions such as nipple shields, breast shells, dressings and applications Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1995 to June 2003. These topics were covered in Renfrew et al. (2000) up to 1997. A two-year overlap was allowed to ensure studies published, but not yet entered on the databases by 1997, were identified. Study design of included studies: RCTs only. Population group studied: breastfeeding mothers of healthy babies. Countries included: all countries. There is little reason to expect developed and developing countries to differ greatly in the outcomes of these clinical practices. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes. 3.5 Baby-related feeding problems, including hypoglycaemia, jaundice, crying and colic, sucking problems, sleep problems, sleepy babies, slow weight gain, Downs syndrome, cleft lip and palate, tongue tie Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1995 to June 2003. These topics were covered in Renfrew et al. (2000) up to 1997. A two-year overlap was allowed to ensure studies published, but not yet entered on the databases by 1997, were identified. Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). The difficulty in this area is being able to conduct studies that are big enough, as there will be relatively small numbers of babies with some of these problems, so methods other than RCTs may be necessary. Inclusion of case studies are considered for some less common problems if no other studies exist. 18

Population group studied: breastfeeding mothers of babies experiencing the problems identified. Countries included: all countries. There is little reason to expect developed and developing countries to differ greatly in the outcomes of these clinical practices. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes. 3.6 Maternal problems that will affect the promotion and support of duration of breastfeeding, including drug and substance abuse, smoking, and maternal nutrition Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1990 to June 2003. Publication of the Health of the Nation in the early 1990s signalled a change in health policy in the UK, with an increasing emphasis on achieving population targets for health outcomes. Saving Lives: Our Healthier Nation (Department of Health, 1999) added to this but with particular reference to addressing inequalities in health. Care provision before the 1990s may be less relevant to the care of disadvantaged women. Study design of included studies: RCTs only. Population group studied: pregnant or breastfeeding mothers experiencing the problems identified, and their babies. Countries included: developed countries. Care systems and circumstances in developing countries are very different, and studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

4 Health professional and lay breastfeeding educator/counsellor training, education and practice change
4.1 Health professional and lay breastfeeding educator/counsellor training, education, and practice change relevant to breastfeeding, including the use of specific training programmes, ways of teaching health professionals about positioning/attachment, and ways of changing health professional practice. Only studies including outcomes of the duration of breastfeeding have been included (ie those only examining attitude, knowledge and behaviour change of health professionals/lay breastfeeding educators have been excluded) Primary studies or existing reviews: primary studies. Any existing reviews were scrutinised for references and additional information. Dates for searches: 1980 to June 2003. This area has not been reviewed before and we wished to capture all relevant studies over the past 20 years. Prior to that, the provision of healthcare and the training of health professionals was conducted in significantly different ways, and we considered little would be gained by seeking earlier studies. Study design of included studies: RCTs and non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). Population group studied: health professionals and lay breastfeeding educators/counsellors working with pregnant/postpartum women. Studies examining premature, small or sick babies are excluded. Countries included: developed countries. Care systems and circumstances in developing countries are very different, as are the role and training of health professionals, and studies from these countries are less likely to be able to inform care in the UK. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes, attitude, knowledge and behaviour change of health professionals and lay breastfeeding educators/counsellors.

Methods of the study


The study was conducted in accordance with the NHS Centre for Reviews and Disseminations guidance for reviews of research on effectiveness (Khan et al., 2001).

Searching, retrieving, appraisal and inclusion


Searches were conducted by the Royal Society of Medicine information services. The search was complex and included identifying clinical, public health, policy and educationrelated studies, including a wide range of methods. The review team advised on search terms and appropriate databases. Each topic area was examined and one overall search strategy was designed. The final search strategy is given in Appendix 1 together with the list of databases searched. The search was completed in June 2003. In addition to this, the review team conducted handsearching of: Journal of Human Lactation, Health Promotion International and Health Education and Behaviour. Two specialist databases were searched: LILACS and MIDIRS.

Initial screening and retrieving publications


The wide strategy resulted in around 55,000 citations for initial screening by the research team. These were downloaded to an Endnote database and de-duplicated and cleaned with the assistance of the University of Leeds library; this system was then used to track all citations and papers throughout the review. A further 8,000 citations were identified by the review team from handsearching and the smaller specialist databases. Each citation was independently reviewed by two members of the team. This process resulted in 940 citations for which full publications were required. Of these about half were identified as already available in the Mother and Infant Research Units in-house collection. The remainder were obtained from several sources, primarily the University of Leeds library and the British Library at Boston Spa. A number of theses were obtained from a range of sources.

Translations
There were no restrictions on language. Citations obtained from the databases allowed us to conduct initial screening in English, Spanish and Portuguese languages the review team has expertise in. In the pre-screening process, 11 translations of abstracts were requested, seven from Japanese and four from Spanish. Twelve translations of full papers were requested: two German, two French, three Italian, two Spanish, two Bulgarian, one Cantonese. The translation process was quality controlled by Felicia 19

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

McCormick, who identified reliable sources for each language, and who did the Spanish translations herself.

Pre-screening of retrieved publications


A pre-screen form was developed and tested, based on previous reviews (Appendix 9, on website). Pre-screening, conducted by two reviewers working together, was carried out on all retrieved publications to identify eligibility. This required checking the different eligibility criteria for each section, and therefore gave rise to discussion among the whole team about the appropriate allocation of papers. As a result of this process, 138 publications were identified for full data extraction.

Comparability of groups Co-factors Withdrawals reported Appropriateness of analysis

For case studies (Section 3.5 only), a form was designed to capture all details of the paper Systematic reviews (Section 4.1 only) Clear review question Inclusion/exclusion criteria relating to primary studies reported Evidence of substantial effort to search for all relevant research in English Primary studies assessed by more than one reviewer Allocation concealment of primary studies reported Blinding outcome of primary studies reported Attrition rates of primary studies reported Sufficient detail of primary studies reported Criteria for grouping studies explained Primary studies summarised adequately. Decisions on whether or not studies and reviews were suitable for inclusion were then made. As with the other aspects of the review, these decisions varied from section to section depending on the topic area. This is explained in the outline of methods for each section.

Data extraction
Data extraction and quality appraisal forms were developed and tested, based on previous reviews. Different versions of the forms were developed for different research methods. Initial data extraction consisted of one member of the review team completing the standard form and the quality appraisal table. This was then independently checked by a second reviewer. The final version of the forms for both data extraction and quality appraisal, shown in Appendix 10 (on website) for quality appraisal and Appendix 11 (on website) for full data extraction for all included studies, was agreed by two reviewers and finally checked by Mary Renfrew.

Quality appraisal
As this review was not restricted to one research design, quality appraisal was developed to be appropriate for a range of approaches. This included assessing the following attributes. RCTs Clear inclusion and exclusion criteria A priori sample size calculation Method of allocation Comparability of groups at baseline reported Blinded outcome assessment Objective outcome measures Withdrawals reported Intention to treat analysis

Developing the sections Allocation of included studies to review sections


The process of allocating studies to each section of the review happened in several stages: First, the reviewers responsible for pre-screening agreed which section they considered the study to be eligible for. If there was doubt they consulted the review team Second, the review member responsible for each section examined their allocated papers in depth at the stage of full data extraction. At this stage, it became apparent that some papers were more appropriate for a different section and re-allocation was agreed, again at team meetings. Some papers were excluded at this stage for reasons of ineligibility or lack of relevance. Studies excluded are listed in Appendix 2 For a small number of papers, this was an iterative process, involving several discussions and consultations between section leaders as the work progressed

Before-after studies 20 Sampling frame used for both groups Method of sampling A priori sample size calculation Clear inclusion and exclusion criteria

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Consensus was reached on every paper. In a small number of cases, papers have been referred to in more than one section, although each has been allocated a primary section in which the full details of the study have been discussed Each member of the review team was allocated lead responsibility for at least one section and second reader support for at least one other. In this way, consistency of approach was ensured across the sections. Mary Renfrew worked across all the sections of the review.

well conducted, or outcome assessment has been blinded. Many elements of quality are related both to the intervention itself, eg whether it is appropriate for the group and setting, whether it has been well conducted, or even fully described; and to the measurement of outcomes, eg whether infant feeding has been well defined and measured. We were therefore reluctant to use a scoring system that did not include the appropriateness of these factors as we considered that such a quantitative ranking would demonstrate a spurious accuracy. Instead, we have presented a series of quality safeguards, which include: A summary of methods and findings of all studies, tabulated, and each checked by at least two members of the research team A quality assessment of all studies, each checked by at least two members of the research team A narrative presentation of each study in the text, in which important strengths and limitations are drawn to the attention of the reader. In this way, the reader will have access to our judgements on the studies, and also be able to make their own judgements using the information presented.

Analysis
Results from primary studies and any existing reviews were assessed and summarised in a qualitative synthesis for each intervention. Meta-analysis was not appropriate for any section. A number of methodological issues have been raised by the inclusion of a range of study designs. These have been addressed and discussed by this team previously (Fairbank et al., 2000) and several methodological issues have been discussed in the final overview.

Excluded studies
Studies for which data extraction was considered to be appropriate, but which were subsequently excluded, are listed in Appendix 2.

Ongoing studies and studies published since search


Ongoing studies that we are aware of are listed in Appendix 3 and those published since the search, again that we are aware of, have been listed in Appendix 4 for information. Only those studies for which information was available at the time of the search have been included in the full review. For a small number of studies full papers were published after the search, but we identified material such as a conference abstract or related publication in our search which allowed us to track down a later publication. Where this was the case, this publication has been included.

Presentation of the results


Each section is presented with: The section topic and any definitions required A boxed summary of the section-specific methods An overview of the evidence base identified, including a summary table (with a brief description of each included study). Each included study has a unique study number allocated, shown in [ ]. This number is shown in the summary table in each section, and the data extraction and quality appraisal tables are listed in number order according to this study identifier Examination of the findings of the included studies, including a critique of the quality. Depending on the section and the number of sub-topics examined, this may be divided into several sub-sections. Where no studies were identified, a brief overview of the topic is given. Discussion and conclusions are given for each section and sub-section. Implications for future research are described.

Summarising quality
We elected not to allocate a quantified quality score to studies, for two reasons. First, we have included a range of different study designs across and within sections and we were reluctant to use a single quantified score to assess studies of different designs in different contexts. Second, and perhaps more importantly, the quality of the studies does not simply depend on elements of study design such as whether or not randomisation has been

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

21

Details of all included studies, both quality appraisal and data extraction forms, can be found in Appendices 5 and 6.

Using information from the teams critique of the included studies and overview of specific topic areas throughout this report, the final decisions were decided as follows. For inclusion in Overview tables 1 and 5: The total number of participants in high quality studies needs to be adequate for the question being addressed (ie, one large high quality study could suffice, or two or more smaller studies which together provide adequate numbers, each coming to the same conclusion) Quality assessment includes not just the design elements of the study/studies, but also assessment of salience (the appropriateness of the design to the question being addressed and the appropriateness of the outcomes measured) The intervention/form of care also has to be applicable in a UK setting to be included. For inclusion in Overview tables 2 and 4: Relevant studies are of good enough quality, with consistent findings, with a sound theoretical base. They differ from Overview tables 1 and 5 in that adequate numbers may not have been reached and the overall quality may be less good. Salience and applicability in a UK setting are not required. These tables indicate interventions that are promising but require further monitoring/audit if implemented (Overview 2) or that are likely to be either ineffective or possibly even harmful, where monitoring/audit is required to assess either their continuation or their withdrawal (Overview 4). For inclusion in Overview table 3: Any research question considered to be important by the team following their review of the evidence included in this review. Some (weak) evidence may already exist but requires further testing, or the question reflects an important gap in the evidence base. None of these interventions should be introduced or withdrawn without further evidence. Finally, to be included in these tables, the full agreement of the whole research team had to be reached. Following these summary tables, an overview addresses the broad issues raised by this series of reviews.

Summarising the findings


Each finding was assessed according to a set of criteria and was allocated to one of the following summary tables, adapted from Enkin et al. (1995): 1 Forms of care/practices/policies that have been shown to be effective/beneficial for enhancing breastfeeding duration 2 Forms of care/practices/policies that appear to be promising for enhancing breastfeeding duration, well grounded in theory and with some research to substantiate 3 Forms of care/practices/policies that are unproven in relation to breastfeeding duration, and require further research 4 Forms of care/practices/policies that appear to be ineffective or harmful to breastfeeding duration 5 Forms of care/practices/policies that have been shown to be ineffective or harmful to breastfeeding duration. Decisions about the final allocation to each of these tables were guided by two fundamental principles. These were first elucidated by Enkin et al. (1995: 389), as follows: We worked from two basic principles; first, that the only justification for practices that restrict a womans autonomy, her freedom of choice and her access to her baby, would be clear evidence that these restrictive practices do more good than harm; and second, that any interference with the natural processes of pregnancy and childbirth and breastfeeding (our addition) should also be shown to do more good than harm. We believe that the onus of proof rests on those who advocate any intervention that interferes with either of these principles. We also used as a guide a set of criteria informed by previous literature in the field of assessing public health/health promotion interventions (eg Nutbeam, 1998; Truswell, 2001; Petticrew and Roberts, 2003), and our own experience of conducting reviews in similar fields (eg Green et al., 1998; Fairbank et al., 2000; Snowden et al., 2001; Fairbank et al., 1999).

22

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Results A total of 80 studies* (including three reviews) have been included. The table summarises the way in which these studies were distributed across each section.
*Note: two studies are combined in one data extraction form, so there are 79 study numbers referred to in this review.

Summary of included studies by section


Topic Section number UK studies Other countries 1990-1995 Post-1995 Studies Total targeting studies disadvantaged groups

Section 1: Public health interventions Support Education Multifaceted Community-based Organisation of care Total public health interventions Section 2: Public policy interventions Total public policy interventions Section 3: Clinical interventions Antenatal care Care in childbirth Prevention of problems/routines in care Treatment of problems Baby-related problems Maternal problems affecting breastfeeding Total clinical interventions Section 4: Education and training interventions Total education and training Total 4 3 10 6 70 2 17 7 63 5 17 9 80 3.1 3.2 3.3 3.4 3.5 3.6 0 0 0 0 1 0 1 0 3 13 6 5 0 27 0 0 0 1 0 0 1 0 3 13 5 6 0 27 0 0 0 0 0 0 0 0 3 13 6 6 0 28 2 1 5 0 6 0 6 1.1 1.2 1.3 1.4 1.5 3 0 0 0 2 5 8 9 9 0 6 32 2 3 5 0 4 14 9 6 4 0 4 23 3 4 4 0 1 12 11 9 9 0 8 37

This table illustrates some important issues about the studies identified in this review. Thirty seven studies (46%) addressed public health topics, predominantly support, education or a combination of both. Twelve targeted disadvantaged groups. Only five of these studies were conducted in the UK. Six studies (7%) examined public policy interventions. None targeted disadvantaged groups. Only one study was conducted in the UK; this was in Scotland. Twenty eight studies (35%) examined clinical issues, predominantly the prevention of problems. None examined antenatal care, and none examined maternal problems that might affect breastfeeding. None

specifically targeted disadvantaged groups. Only one (a case study of one infant) was conducted in the UK. Nine studies (11%) examined education and training. Five targeted disadvantaged groups. Three were conducted in the UK. Overall, only 17 studies (21%) specifically examined the needs of women from disadvantaged groups. All of these were related to public health, and to education and training. There was a dearth of studies conducted in the UK; only 10 studies (12%) in total. Of these, two were general interventions to offer support/care to women with new babies, and one was a case study of single infant. 23

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1.1 Public health interventions: support


Topic Support offered by individuals and agencies to pregnant and postpartum women to increase the duration of breastfeeding, including peer support, and health professional support

Summary of methods
Study design of included studies: RCTs only Dates for searches: 1995 to June 2003 Population group included: pregnant and postpartum women Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for baby, health outcomes for baby, etc), relevant health service outcomes including economic outcomes

1.1.1 Background
It has been known for many years that additional support for women who start to breastfeed is an important factor in helping them to continue to breastfeed, either at all or exclusively (eg Houston, 1984; Sikorski et al., 2002). This appears to be the case in all countries, but is perhaps especially necessary in countries where there is a culture where bottlefeeding is considered the norm and in communities where breastfeeding rates are low (see Introduction). Many kinds of breastfeeding support have developed, including support from a range of health professionals (eg midwives, health visitors, public health nurses, GPs, lactation consultants); from voluntary supporters trained by organisations with established programmes and accreditation (such as the Association of Breastfeeding Mothers, Breastfeeding Network, La Leche League and the National Childbirth Trust); and peer supporters, who are normally local women who have themselves breastfed and who may or may not have received some additional training in providing such support. Additional postnatal support in the community, from professionals or ancillary staff, may also enhance breastfeeding duration (see Glossary).

Given the acceptance that support is necessary and can be effective, there remain a number of questions to ask about its provision, which include: What forms of support are most effective to support the ongoing duration of breastfeeding and to support exclusive breastfeeding? What is the preparation and training needed to best support women to breastfeed? When is breastfeeding support best offered? In pregnancy, or after birth? Is different support needed by women who already plan to breastfeed, and those who are undecided? How can support best be provided for women living in circumstances where breastfeeding is not the norm? What is the cost effectiveness of the provision of support to breastfeeding women? Answering these questions will help provide effective and appropriate breastfeeding support. The problem is that existing RCTs describe different forms of support, offered at different times, to different groups of women living in different cultures and communities. Some do not provide enough detail to allow the reader a full understanding of the intervention and participants in their trial. We start by listing the studies that form the evidence base for this review and give their main findings with an indication of

24

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

study quality. We then address the questions outlined above. Because we were looking primarily for evidence of effectiveness of interventions on breastfeeding duration, and support is an intervention that can be tested using an RCT design, we excluded studies that used other methodologies from only this section of our review. Questions about acceptability of support offered to women are very important and can best be addressed using qualitative methods. We included findings about views of women and caregivers where these were reported as part of a trial (see data extraction forms in Appendix 13). We have referred to some of the excluded studies (Appendix 2) in our discussion.

The remaining study, from Australia, was carried out among very disadvantaged young women (Quinlivan et al., 2003 [8]). Provision of maternity care varies between these countries. A similar intervention could be expected to have different effects according to what is already received by or available to women as part of standard care. It must be also be remembered that care may not be completely uniform within as well as between countries, may change over time, and the quality of the care given will vary even if the system is uniform. See Appendix 6 for details of standard care in UK at the present time. Two trials evaluated additional postnatal support interventions offered to women regardless of their infant feeding intention or practice and delivered by professionals (Quinlivan et al., 2003 [8]) or ancillary staff (Morrell et al., 2000a/b [4]). In Quinlivan et al. (2003)s Australian setting, the main aim of postnatal community services was to provide sufficient support to a mother to enable her to safely maintain guardianship of her child. In this study of disadvantaged teenage mothers, the intervention group received five extra structured postnatal home visits from a nurse-midwife, with a range of inputs including breast and bottlefeeding. Morrell et al. (2000a/b) [4] sought to establish the relative cost effectiveness, primarily in terms of benefits to womens general health, of postnatal support in the community in addition to the usual care provided by community midwives in England. The intervention group received up to 10 visits of up to three hours each from a trained community postnatal support worker whose remit on infant feeding was to reinforce the midwifes advice. Nine trials evaluated breastfeeding support interventions. Four of these were structured programmes of home visits and phone calls offered postnatally to breastfeeding women by health professionals with specific breastfeeding training. The particular health profession, where reported in the paper, appears in Table 1.1. Three of these interventions were each offered by one professional who was also the researcher and the fourth (Pugh and Milligan, 1998 [6]) was offered by two professionals. Three trials evaluated peer support/volunteer counsellor interventions. In Dennis et al. (2002) [1] breastfeeding 25

1.1.2 Evidence base


Eleven RCTs evaluating a range of support interventions met the criteria for this review (see Table 1.1 for details). Two papers from a study by Dennis are referenced; Dennis et al. (2002) [1] was our source for information about the trial and Dennis (2002) [1] provided additional information about views of mothers and peer counsellors who took part. We used the full study report (Morrell et al., 2000b [4]) as well as the published paper (Morrell et al., 2000a [4]) to obtain data from Morrells study. The authors of two papers provided us with additional information (Quinlivan et al., 2003 [8]; Winterburn et al., 2003 [10]). Three trials were conducted in the UK, all with women recruited from the general population (Graffy et al., 2004 [2]; Winterburn et al., 2003 [10]; Morrell et al., 2000a/b [4]). Three trials took place in Canada, two among relatively advantaged women (Dennis et al., 2002 [1]; Mongeon and Allard, 1995 [3]) and one with a group of mainly well-educated women who intended to breastfeed and reported having no-one to provide breastfeeding support (Porteous et al., 2000 [5]). Of the four studies from the US, two were conducted among low-income women (Pugh et al., 2002 [7]; Serafino-Cross and Donovan, 1992 [9]), one among what appears to be a group of relatively advantaged women (Pugh and Milligan, 1998 [6]), and one among women who were members of the US armed forces or their dependents, and receiving care at a military hospital (Wrenn, 1997 [11]).

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

women received telephone support from a peer counsellor who contacted them initially within 48 hours of the birth. In Mongeon and Allard (1995) [3], women intending to breastfeed received an antenatal visit from a peer counsellor and postnatal phone calls initiated by the peer counsellor within a week of the birth. In Graffy et al.s UK study (Graffy et al., 2004 [2]), women considering breastfeeding received an antenatal visit from a volunteer counsellor inviting them to access counsellor support postnatally. This support might be phone-based or could include home visits. The remaining two trials evaluated interventions that combined peer and professional support. In Pugh et al. (2002) [7], women receiving financial medical assistance during postpartum hospitalisation were visited in hospital and at home by the community nurse/peer counsellor team, and phoned at home by paid peer counsellors. In Winterburns UK study (2003), pregnant women were asked to identify a close female confidante who could support them to breastfeed. The community midwife visited mother and confidante together, before the birth, to discuss breastfeeding.

breastfeeding duration, but we can say that the effect size would be small and that we would need to test it with a larger group of women to see such an effect if it exists. Morrell et al. (2000a/b) [4] provide information relevant to costing such a test in the UK and this is discussed below. Quinlivan et al. (2003) [8] recruited women under 18 years old attending their first antenatal appointment at an Australian public-care teenage pregnancy clinic for first-time mothers, and evaluated a series of structured postnatal home visits, additional to standard care. They found significantly fewer neonatal adverse events (nonvoluntary foster care of the neonate) and significantly greater knowledge and effective use of contraception at six months in the intervention group. No differences between the groups were found in rates of breastfeeding initiation, breastfeeding at hospital discharge or any breastfeeding at six months.

1.1.3.2 Breastfeeding support interventions: from health professionals


Porteous et al. (2000) [5] (Canada) met nearly all the quality criteria for this review, only lacking evidence of blinded outcome assessment. Numbers needed for the sample size calculation were achieved, so although these were small (n=51), we can have confidence in its finding that rates of any breastfeeding were significantly higher and rates of exclusive breastfeeding were higher in the intervention group at four weeks. This was an intensive intervention of structured, individualised support, including hospital and home visits and phone calls. It was delivered by one person, the midwife-researcher, to relatively advantaged women who were breastfeeding at recruitment and identified themselves as having no-one with breastfeeding experience to provide breastfeeding support. Standard care did not include routine home visits. This high quality trial provides good evidence that intensive, structured postnatal support from one trained professional to women who want breastfeeding support, and lack it from other sources, is likely to be effective at increasing exclusive and any breastfeeding at four weeks. Pugh and Milligan (1998) [6] (US): the report on this study (n=60) did not give clear inclusion or exclusion criteria or a sample size calculation. The methods of randomisation, number of withdrawals if any and whether or not outcome assessment was blinded were not reported. Analysis was by intention-to-treat. Intervention and control groups were stated to be

1.1.3 Findings 1.1.3.1 Additional postnatal support interventions


Neither of the two studies offering additional postnatal support interventions to women regardless of infant feeding intention found any difference between intervention and control groups for breastfeeding duration. Morrell et al. (2000a/b) [4] (n=623) met all the quality criteria for this review and Quinlivan et al. (2003) [8] (n=136) lacked only evidence of blinded outcome assessment. Breastfeeding duration was a secondary outcome in both these high quality trials. Both had sample sizes large enough to meet power calculation requirements for their primary outcomes. Neither was powered for secondary outcomes. We cannot conclude from this that an additional postnatal support intervention delivered to women regardless of infant feeding intention or practice will have no effect on 26

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

comparable but few details of their characteristics were provided. This trial therefore appears to be of low quality and its finding, that US women receiving a carefully structured nursing intervention that included home nursing and lactation support (apparently from two different people, neither of whom was the researcher) breastfed longer than those who did not, cannot be said to be strongly supported by the evidence presented. Wrenn (1997) [11] (US): this thesis (n=186) met some of the quality criteria for this review although it did not report details of the randomisation procedure or whether outcome assessment was blinded. Exclusive and any breastfeeding were found to be higher in the intervention group at six weeks. The author states that these differences were not statistically significant but were clinically significant. This health professional intervention, delivered by the researcher and comprising one hospital visit, one home visit and one follow-up telephone call, although structured, was less intensive than the others in this group. The setting was a US military hospital whose standard care is not described. Participants were all members of the US armed services or their dependents; these women are likely to have been more separated from their extended support networks than non-military women in the general or disadvantaged populations. We can have some confidence that this intervention was not intensive enough to increase breastfeeding significantly in this population, and the findings of this medium quality study have limited value outside its setting. Serafino-Cross and Donovan (1992) [9] (US): this paper also did not report clear inclusion or exclusion criteria, sample size calculation, method of randomisation or whether outcome assessment was blinded. However, the groups were comparable at baseline, there were no withdrawals at two months and analysis was by intention-to-treat. The authors of this small (n=52), medium quality study found a statistically significant increase in breastfeeding for at least half of the infants feedings at two months in the intervention group. The intervention group received structured home visits over a two month period from a trained professional (one of the researchers). Some confidence can be placed in the efficacy of this intervention among disadvantaged women in the US.

significantly more breastfeeding and exclusive breastfeeding at four and 12 weeks and significantly more breastfeeding at eight weeks in the intervention group. This high quality trial was conducted in a region where 60% of mothers were practising some form of breastfeeding at three months postpartum, among relatively advantaged breastfeeding women recruited during postnatal hospitalisation. Details of care available to all women in pregnancy and in hospital were not provided. Peer supporters were women who had breastfed for at least six months and were members of a local volunteer breastfeeding organisation that provided mother-to-mother telephone-based breastfeeding support. Those participating in the trial received 2.5 hours training and received a handbook. Mothers who had enrolled prenatally with this organisation were excluded from the trial. Peer supporters were paired geographically with women who were telephoned within 48 hours of the birth and then as often as the woman deemed necessary. Data about what the peer supporters did were collected from 59% of the peer supporters. Ninety seven per cent of reported connections were by telephone and 3% were face-to-face meetings. Standard postnatal care was available to all participants and included a wide range of hospital and community services although no details are provided. We can have confidence that telephone support from well-matched peer supporters to relatively advantaged women living in a region where breastfeeding rates are high, and who are breastfeeding at postnatal recruitment, is likely to be effective at increasing rates of breastfeeding and exclusive breastfeeding up to three months. Mongeon and Allard (1995) [3] (Canada): this study (n=200) met some of the quality criteria for this review. Sample size calculation and blinding of outcome assessment were not reported. The method of randomisation was not stated and true randomisation is unlikely since there were exactly 100 women in each group. Women in this trial were mainly well educated, partnered and born in Quebec. They were recruited antenatally and intended to breastfeed. Details of standard care in pregnancy and in hospital were not provided. Peer supporters were local women who had breastfed. They received nine hours training from the trial nurse and monthly group supervision thereafter. Peer supporters visited women once before the birth and telephoned weekly for six weeks and then fortnightly for five months, or until the child was weaned. Standard 27

1.1.3.3 Breastfeeding support interventions: from peer supporters/volunteer counsellors


Dennis et al. (2002) [1] (Canada): this study (n=256) met all the quality criteria for this review. It reported

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

postnatal care consisted of a home visit in the first month and then other phone calls and meetings as initiated by the mother. This study found no differences between the groups for breastfeeding duration outcomes or breastfeeding problems. We can have some confidence from this medium quality study that breastfeeding support by phone alone, when it starts up to a week after the birth and is offered to women who may not have started to breastfeed or who have already stopped breastfeeding due to early problems, may not be effective in improving breastfeeding outcomes. Graffy et al. (2004) [2] (UK): this study (n=720) met all the quality criteria for this review. Sample size required by the power calculation was not achieved but the small shortfall would not have modified the findings of no significant differences between the groups in breastfeeding initiation or any or exclusive breastfeeding at six weeks or four months. This high quality trial was conducted among the general population of women attending for antenatal care at 32 general practices in London and south Essex. With regard to infant feeding intention, women considering breastfeeding, ie all women excepting those who expressed the intention to bottlefeed, were included. Women who had planned to contact a breastfeeding counsellor were excluded from the trial. The volunteer breastfeeding counsellors were women who had themselves breastfed and had trained with and achieved breastfeeding counsellor accreditation from the National Childbirth Trust. Matching of counsellors with women is not reported. The intervention was an antenatal home visit with the offer of postnatal support by telephone or further home visits if requested. Data about contacts made was reported for 336/363 women in the intervention group (93%); 269 of these (80%) received the antenatal visit; 126 women (38%) did not make contact with a breastfeeding counsellor after the birth; 143 women (43%) made contact and received telephone support; and 67 (20%) made contact and also received at least one home visit. Women who left school at an earlier age were found to be significantly less likely to arrange a postnatal visit. We can have confidence that an antenatal offer of breastfeeding support to women who may or may not want it and who would have to ask again later in order to get it, from breastfeeding counsellors who may or may not be their neighbours or their social peers, is unlikely to be effective at increasing rates of exclusive or any breastfeeding up to four months. 28

1.1.3.4 Breastfeeding support interventions: from peer supporters and health professionals working together
Pugh et al. (2002) [7] (US): this study (n=41) met some of the quality criteria for this review, specifying randomisation method, comparability of groups at baseline and intention-to-treat analysis. However, inclusion and exclusion criteria were not clear and sample size calculation, blinding of outcome assessment and withdrawals were not reported. Recruits were disadvantaged US women during their postpartum hospitalisation, who by implication were breastfeeding at recruitment, though this is not stated. The intervention was additional to standard care and consisted of daily hospital visits and home visits at least during weeks one, two and four from the community nurse/peer counsellor team, and telephone support from peer counsellors for six months. Which team members made the visits, the content of visits and phone calls, and the training and volunteer/employed status of peer counsellors were not reported. Costs of the intervention, including time to feed infants and amounts of supplementary formula used were reported. Statistical significance of the finding that after week one more mothers in the intervention group were breastfeeding at all time periods was not reported. Interpretation and application of this finding from a small, medium quality study is further hampered by these omissions from the report. Winterburn et al. (2003) [10] (UK): this study (n=72) reported withdrawals and used intention-to-treat analysis, but does not meet the other quality criteria set for this review. Recruitment was from the general population attending antenatal care in one locality in North Trent. Breastfeeding intention is not reported. Women in the intervention group were asked to identify a close female confidante who could support them to breastfeed. Whether the confidante had breastfed is not reported. The midwife visited the woman and her confidante together during the third trimester to discuss breastfeeding. The control group followed the normal policy of care, typically an antenatal breastfeeding discussion with a midwife. Whether women in the control group had confidantes who could or could not support them with breastfeeding is not reported. The study found no significant differences between the groups for initiation or duration of breastfeeding. However, the trial as reported is of low quality and we cannot place confidence in this finding.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1.1.4 Discussion
Having presented the included studies and their main findings, we return to the questions posed in the background to this section. What forms of support are most effective, to support the ongoing duration of breastfeeding and to support exclusive breastfeeding? We found high quality evidence that both health professional support (Porteous et al., 2000 [5]) and peer support (Dennis et al., 2002 [1]) can be effective in supporting both exclusive and any breastfeeding among relatively advantaged women. Evidence of lesser quality suggests that both health professional support (SerafinoCross and Donovan, 1992 [9]) and support from peers and health professionals working together (Pugh et al., 2002 [7]) can be effective in supporting any breastfeeding. These two studies did not present data on exclusive breastfeeding and may not have been looking for this among their population of disadvantaged American women. Morrell et al. (2000a/b) [4] and Quinlivan et al. (2003) [8] provide high quality evidence that additional postnatal support interventions not including additional breastfeeding support were ineffective in supporting breastfeeding. Graffy et al. (2004) [2] provide high quality evidence that a volunteer support intervention was not effective; however, women in this trial had not all chosen to breastfeed and had to ask, after the birth, for the support offered antenatally. Mongeon and Allard (1995) [3] provide medium quality evidence of ineffectiveness of a peer-support intervention where details of support before the baby was a week old are not specified. This indicates that effective support can come from peers and from professionals, working separately or together, if it is specific breastfeeding support and if it is offered to women who actively want breastfeeding support or have decided to breastfeed. It suggests that in developed countries breastfeeding support for relatively advantaged women can result in higher rates of continued exclusive breastfeeding as well as higher rates of any breastfeeding. Among relatively disadvantaged women, support interventions we have identified have aimed to increase any rather than exclusive breastfeeding.

We have not found evidence of effectiveness of interventions to support exclusive breastfeeding among disadvantaged women. Hamlyn et al. (2002) found that UK mothers from ethnic minority groups (who may have patterns of feeding influenced by their cultural background, and who are likely to be relatively disadvantaged), were more likely to breastfeed and breastfeed for longer; in comparison, white mothers who breastfed were more likely to do so exclusively. We have also not found evidence about effects on any breastfeeding of interventions to support exclusive breastfeeding among relatively disadvantaged women. One recent study of women from deprived groups in the UK (Dyson et al., 2003) found that women from some groups, especially Pakistani and African-Caribbean women, voiced a strong intention to mix breast and bottlefeeding from the start. This finding, together with the data from Hamlyn et al. (2002), suggests approaches to encourage exclusive breastfeeding might be counterproductive in some ethnic communities and/or disadvantaged groups, perhaps leading to total abandonment of breastfeeding by women who are happy to mix breast and bottlefeeding but not happy to aim for exclusive breastfeeding. It is likely that moving towards exclusive breastfeeding will require careful planning over several years, and health professionals and policy makers will have to accept that women will wish to mix breast and bottlefeeding as a way of sustaining any breastfeeding. It will be essential, therefore, to examine reasons for the wish to mix breast and bottlefeeding and interventions that might enable women to breastfeed exclusively. What is the preparation and training needed to best support women to breastfeed? Evidence for the care women need to establish breastfeeding is available (see Section 3); however, the nature and content of effective support to extend the duration of breastfeeding is not so well understood. In our included studies we found evidence of varying quality for the effectiveness of support interventions by health professionals (Porteous et al., 2000 [5], Serafino-Cross and Donovan, 1992 [9]). Studies state that most of the professionals delivering interventions had specialised breastfeeding training. Whether health professionals had breastfed is not reported. There was high quality evidence of effectiveness (Dennis et al., 2002 [1]) and 29

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

ineffectiveness (Graffy et al., 2004 [2]) of peer/volunteer support interventions. Most of the peer/volunteer supporters were prepared by having breastfed successfully themselves (Pugh et al., 2002 [7] and Winterburn et al., 2003 [10] do not specify this). The training received by volunteer counsellors in Graffy et al. (2004) [2] is well accredited in the UK, but few details of the content of peer training are reported in Dennis et al.s paper (Dennis et al., 2002 [1]). The papers in this section of our review do not provide evidence to differentiate effectiveness of breastfeeding support interventions by preparation or training of supporter. Peers and professionals in these studies appear to be equally well prepared and trained to be effective where women positively intend to breastfeed and receive knowledgeable support starting soon after the birth. It may be that elements of support from well-matched peers, related to having themselves breastfed successfully in circumstances similar to those of women they are supporting, could be effective in supporting women to continue, after breastfeeding has been well established, up to six months or longer. Most of our papers do not provide good quality breastfeeding outcomes for later time points, so further research would be needed to investigate this question. When is support best offered? In pregnancy, or after birth? Of the nine breastfeeding support interventions we identified, three were offered antenatally (Graffy et al., 2004 [2]; Mongeon and Allard, 1995 [3]; Winterburn et al., 2003 [10]) and none of these were found to be effective. Graffy et al. (2004) [2] was the best quality study among these three. In this study, women considering breastfeeding were visited antenatally by a National Childbirth Trust (NCT) counsellor who offered postnatal breastfeeding support, by phone or home visit. To get this support, women needed to ring and ask for it after the birth. All the interventions in studies of high or medium quality that were found to be effective (Dennis et al., 2002 [1]; Porteous et al., 2000 [5]; Pugh et al., 2002 [7]; SerafinoCross and Donovan, 1992 [9]) were offered to women intending to breastfeed or already breastfeeding after the birth of their baby, where women did not have to ask again to get the support. The remaining two studies also offered support after the birth and provided low quality 30

evidence of effectiveness (Pugh and Milligan, 1998 [6]) and moderate quality evidence of no effect (Wrenn, 1997 [11]). We cannot conclude that women should not be informed antenatally about availability of postnatal breastfeeding support. However, it appears that to be effective, breastfeeding support should be offered to women starting soon after the birth without women having to ask for it at that point. Is different support needed by women who already plan to breastfeed and those who are undecided? The two high quality studies of additional postnatal support interventions that were offered regardless of infant feeding intention were found to have no effect on breastfeeding outcomes (Morrell et al., 2000 a/b [4]; Quinlivan et al., 2003 [8]). All the other studies in this section except for Graffy et al. (2004) [2] were offered to women who intended to breastfeed or were already breastfeeding and all the interventions found to be effective were of this type. Graffy et al.s high quality study included women who were undecided about infant feeding antenatally and showed no effect of the breastfeeding support intervention. Women should be informed in pregnancy about postnatal breastfeeding support that is available. Skilled helpers for women who want to breastfeed should be available at all times on delivery suites and postnatal areas of maternity hospitals to help women establish breastfeeding. Evidence to differentiate effectiveness of types of support to breastfeeding women at later time points has not been identified in this section of the review (see above, Preparation and training). How can support best be provided for women living in circumstances where breastfeeding is not the norm? Of the 11 studies identified for this section, only in Dennis et al. (2002) [1] could breastfeeding be said to have been the norm in the region where this high quality trial was conducted, 60% of mothers were practising some form of breastfeeding at three months postpartum, and so participants in this study were not being countercultural by breastfeeding for at least three months. Quinlivan et al. (2003) [8], testing an additional postnatal support intervention, found no differences between the

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

groups in breastfeeding rates up to six months. However, rates reported in this Australian study were high compared with UK rates for apparently less disadvantaged women. In Quinlivan et al. (2003) [8], breastfeeding initiation was 100/136 (74%), breastfeeding at hospital discharge 90/136 (66%) and any breastfeeding at six months was 32/136 (42%). Most of these teenage women had a low/destitute socio-economic status score; a quarter were depressed (EPDS>13); a quarter were experiencing domestic violence; and smoking, use of alcohol and nonprescription drugs were reported to have been prevalent among these women. Hamlyn et al. (2002) reported the following rates for the year 2000 in England and Wales: among women in lower occupations/never worked, breastfeeding initiation was 60%, breastfeeding at one week was 42% (lower occupations) and 36% (never worked), and breastfeeding at six months was 13%. Among women in England and Wales who left full-time education aged 16 or under, breastfeeding initiation was 54%, breastfeeding at one week was 40% and breastfeeding at six months 11%. Perhaps in Australia breastfeeding is the norm, or these higher rates might be due to differences in standard care, aspects of care offered in the particular public-care teenage pregnancy clinic reported in Quinlivan et al. (2003) [8], factors associated with taking part in the trial, or a combination of these and other unknown factors. UK service providers may be interested to find out more about these, particularly care offered in this clinic. Extended breastfeeding is not a general norm in the UK. One midwife addressed this by collaborating with La Leche League to set up peer counsellor training in three Newcastle estates. She writes that she grew up in Rhodesia (now Zimbabwe) and on coming to England found: I had to learn how to teach women to breastfeed! My African friends laughed at me when I told them about my British experiences. After all, breastfeeding was like breathing something everyone did! (Thompson, 1998). Thompson observed that peer counsellors provided the support women needed to continue to breastfeed, often in the face of opposition from families and friends. Bick et al. (1998) interviewed 906 women who gave birth in Birmingham, 566 of whom breastfed at least once, and found that regular child care from a female relative was an important predictor of early breastfeeding cessation.

Bick suggests this is likely to be due to negative advice from relatives, based on their personal experiences of breastfeeding. One study included in this section of our review aimed to address this problem. In Winterburn et al. (2003) [10], women in the intervention group were asked to identify a close female confidante who could support them to breastfeed. The midwife visited the woman and her confidante together during the third trimester to discuss breastfeeding. This intervention may have potential to help a woman who has a close female confidante able to provide breastfeeding support, to ask her for it and to get it. It could also help women, confidantes, midwives and other health professionals to reduce conflicting advice, and might be a way of identifying peer supporters in a community who could receive training to support other women to breastfeed. An intervention like this could become part of a community midwifes normal work, and we suggest further investigation in a more robustly designed study. What is the cost effectiveness of the provision of support to breastfeeding women? Two of the 11 studies considered costs. Interestingly, Pugh et al. (2002) [7] costed time to feed infants, and amounts of supplementary formula, but the details from her small US study are not directly transferable to the UK. The intervention group in the high quality UK study by Morrell et al. (2000a/b) [4] included 311 women from the general population (control group, n=312, received standard care). The support worker intervention, designed to help women rest and recover, was delivered at a total cost of 48,960. Support workers provided women with up to 10 visits for up to three hours per day in the first 28 postnatal days, and were trained to provide practical and emotional support, including reinforcing midwifery advice on infant feeding. Visit costs comprised staff time (84%), travel expenses (8%), education and training (5%) and equipment (3%). The mean difference in total NHS costs between the groups was 178.61 (95% confidence interval 79.60 to 272.40). Women in the intervention group were very satisfied with the support worker visits; however, no health benefits of the visits compared with traditional community midwifery visiting were found on study measures, and at six months there were no significant differences between the groups in NHS resource except for the support service. We conclude that an additional postnatal support 31

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

intervention, offered to women regardless of infant feeding intention or practice, for the purpose of increasing breastfeeding duration, is very unlikely to be cost effective in the UK.

1.1.5 Conclusions
The findings detailed here are summarised in Overview Tables 1-5. The strong messages from the studies reviewed here are that: Breastfeeding-specific support from both peers and professionals has been shown to be effective at increasing any breastfeeding among women who plan to breastfeed, as long as that support is offered to women soon after the birth Such support has been shown to be effective at increasing exclusive breastfeeding among women from relatively advantaged backgrounds Such support has not been shown to be effective at increasing exclusive breastfeeding among women from disadvantaged backgrounds Additional general postnatal support offered to women regardless of infant feeding intention or practice is unlikely to affect breastfeeding duration In relation to the training and preparation required there is no evidence that professionals without additional, breastfeeding-specific training have been effective in supporting women to breastfeed.

More information is needed to examine ways of supporting women to breastfeed for longer than three months The provision of routine breastfeeding support from health professionals to women on low incomes as part of the national Healthy Start programme provides a unique opportunity to implement widespread, routine delivery of this intervention to women who intend to breastfeed. Such delivery could provide the necessary monitoring and evaluation data to evaluate the effectiveness of local processes and outcomes in different settings in England.

1.1.6 Future research


More information is needed about components of support shown to be effective or ineffective as a basis for recommendations about what works to support women to breastfeed More information is needed about the training and preparation needed to support women to breastfeed. Studies reported here almost all used health professionals with additional, breastfeeding-specific training, or peer/volunteers who had themselves breastfed and had some additional training More information is needed about ways of supporting women from communities where breastfeeding is not the norm, to breastfeed exclusively Studies are needed to examine ways in which informal support (ie a close female confidante or family member) can be used to support women, together with input from health professionals 32
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Table 1.1: Included studies support for breastfeeding women Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Within 48 hours after hospital Any breastfeeding, exclusive breastfeeding and discharge then as often as the mother deemed necessary bottlefeeding at 4, 8 and 12 weeks Timing of breastfeeding duration outcomes

Name of authors and study number Telephone-based peer Primiparous women, mainly support well educated, married, middle income, ethnicity not stated. All breastfeeding at recruitment Number of participants = 256 Women recruited from general population, predominantly white, primiparous. Half were intending to breastfeed for more than 3 months Number of participants = 720 Women mainly expecting first One antenatal visit and postnatal phone calls from a baby, university educated, volunteer partnered, ethnicity not stated. All intended to breastfeed, half for more than 6 months Number of participants = 200 Women recruited from general population, mainly white, multiparous, a third receiving housing benefit. Recruited regardless of infant feeding intention or experience and these are not reported Number of participants = 623 Offer of visits from a trained worker employed to provide practical and emotional support, including reinforcing midwifery advice on infant feeding Home visit in the last month of pregnancy. Phone calls weekly for 6 weeks and two-weekly until infant 5 months old or weaned One antenatal visit from an NCT volunteer counsellor offering postnatal support (phone or home visit) After 36 weeks gestation, before the birth

Date of Country publication

Design

Dennis et al. [1]

2002

Canada

RCT

Graffy et al. [2]

2004

UK

RCT

Any and exclusive breastfeeding at 6 weeks, any breastfeeding at 4 months, any bottle at 7 days, 6 weeks and 4 months

Mongeon and Allard [3]

1995

Canada

RCT

Any breastfeeding at 6 months

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Morrell et al. [4]

2000

UK

RCT

Up to 10 visits for up to 3 hours per day during the first 28 postnatal days

Exclusive breastfeeding, mixed breast and bottlefeeding and formula only at 6 weeks

33

34 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Visits from and phone contacts with the researcher, a practising midwife in the community Daily visits in hospital, home visits in the first week and as required. Phone call within 72 hours of discharge and weekly for 4 weeks Exclusive breastfeeding and breastfeeding with occasional supplementation of expressed breastmilk at 4 weeks Timing of breastfeeding duration outcomes Women mainly educated beyond high school, white, about half multiparous. All intended to breastfeed and reported having no-one to provide breastfeeding support Number of participants = 51 Primiparous women, mainly white, completed high school. By implication, all intended to breastfeed, though this is not stated Number of participants = 60 Two structured postnatal home visits from RN and a phone call from lactation consultant. Did not receive standard home visit. All participants received breastfeeding support before randomisation RN visits on days 3 or 4 and 12 Visits from community nurse/ peer counsellor team and telephone support from peer counsellors Lactation consultant phone call on day 5 Daily visits in hospital, home visits during weeks 1, 2 and 4, and at the teams discretion. Telephone support twice weekly through week 8 and monthly through month 6 6 months Women, all low-income, mainly single, AfricanAmerican, with 12+ years education. Parity/infant feeding experience not reported. By implication, all intended to breastfeed this baby, though this is not stated Number of participants = 41 Structured postnatal home Women, all <18 years old visits from one of two nurseand expecting first baby, midwives almost all with low/destitute socio-economic score, a quarter indigenous Australian. Recruited regardless of infant feeding intention and this is not reported Number of participants = 136 Exclusive breastfeeding at 3 months, exclusive and partial breastfeeding at 6 months 1 and 2 weeks and 1, 2, 4 and 6 months Hospital discharge and 6 months

Table 1.1: Included studies support for breastfeeding women (cont.)

Name of authors and study number

Date of Country publication

Design

Porteous et al. [5]

2000

Canada

RCT

Pugh and Milligan [6]

1998

US

RCT

Pugh et al. [7]

2002

US

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Quinlivan et al. [8]

2003

Australia

RCT

Table 1.1: Included studies support for breastfeeding women (cont.) Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention 5-8 postnatal breastfeeding During the first 2 months, support home visits from the concentrated in the first 2 researcher, a trained and weeks experienced breastfeeding counsellor. All participants received breastfeeding support and education before randomisation Mothers were asked to Visit during third trimester identify a close female confidante who could support them to breastfeed. The researcher, a midwife, visited mother and confidante together Breastfeeding for at least half of the infants feedings, at 2 months, and 6 months for intervention group only Timing of breastfeeding duration outcomes

Name of authors and study number Women, all low-income, mainly educated 9-12 years, primiparous, non-white. All intended to breastfeed for 2 months or longer Number of participants = 52

Date of Country publication

Design

Serafino-Cross and Donovan [9]

1992

US

RCT

Winterburn et al. [10] Women attending antenatal clinics in one locality, mean age 28 years (range 16-37). Other details not reported. By implication all intended to breastfeed, though this is not stated Number of participants = 72 Primiparous women, all members of US armed forces or their dependents, mainly with some college education, white/Hispanic. All planned to breastfeed for at least 6 weeks Number of participants = 186 Structured visits from the researcher, a nurse and lactation consultant, in hospital and at home, followed by home phone call

2003

UK

RCT

10 days, 1 month, 6 weeks, 3 months, and beyond (6 months)

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Wrenn [11]

1997

US

RCT

One visit during hospital stay, one visit 2-4 days after discharge and a follow-up phone call 10-14 days after home visit

Exclusive, partial (total, 20-80% breastmilk, <20% breastmilk), token and no breastfeeding at 6 weeks

35

1.2 Public health interventions: educational interventions


Topic Antenatal and postnatal classes/education intended to promote and support the duration of breastfeeding (ie not general antenatal/postnatal classes)

Summary of methods
Study design of included studies: RCTs Dates for searches: 1990 to June 2003 Population group included: pregnant and postpartum women, from a variety of circumstances or backgrounds Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for baby, health outcomes for baby etc), relevant health service outcomes including economic outcomes plus relevant health service outcomes including attendance at classes, economic outcomes (Pollard, 1998 [18]). Three trials were conducted in Australia, including two studies among women from disadvantaged groups, namely low-income women (Duffy et al., 1997 [14]) and Vietnamese women, mostly on low incomes (Rossiter, 1994 [19]). The other study was on Caucasian women on higher incomes (Hauck and Dimmock, 1994 [15]). The remaining two trials were conducted among women in the general populations of Italy (Curro et al., 1997 [13]) and Ireland (Loh et al., 1997 [17]). Three trials evaluated educational interventions which provided written information about breastfeeding. A selfhelp breastfeeding manual was delivered antenatally in the US study (Coombs et al., 1998 [12]) whereas a breastfeeding booklet was delivered in the postnatal period in the studies conducted in Italy (Curro et al., 1997 [13]) and Australia (Hauck and Dimmock, 1994 [15]). Two studies were evaluations of a breastfeeding selfassessment tool. One study delivered an educational fact sheet and breastfeeding questionnaire in the antenatal period to women in Ireland (Loh et al., 1997 [17]) and the other study delivered a breastfeeding daily log in the postnatal period to women in the US (Pollard, 1998 [18]). One American study evaluated a prenatal paediatric visit, with no written information (Serwint et al.,1996 [20]).

1.2.1 Background
Studies included in this section examined interventions that provide factual or technical information about breastfeeding to a specific target group in a hospital or community setting (see Glossary). Health education interventions are grounded in professional expertise and can be delivered in either the antenatal or postnatal periods. Educational interventions provide education which is additional to that provided in routine care with the aim of promoting or supporting the duration of breastfeeding. Participants may be undecided about their feeding intention, have expressed a planned feeding method or have initiated breastfeeding in the early postpartum period.

1.2.2 Evidence base


A total of nine RCTs evaluating education interventions for pregnant or breastfeeding women have been included in this review (see Table 1.2 for details). Four trials were conducted in the US, three of which were conducted among mostly low-income, black women (Coombs et al., 1998 [12]; Kistin et al., 1990 [16]; Serwint et al., 1996 [20]) and one among higher income, Caucasian women 36

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

The remaining three educational interventions included in this review were evaluations of group education sessions delivered in the antenatal period. One study provided practical antenatal education on positioning and attachment (Duffy et al., 1997 [14]) while the Australian study delivered culture-specific educational materials to Vietnamese women (Rossiter, 1994 [19]). The study conducted by Kistin et al. (1990) [16] compared both group and one-to-one discursive education programmes with routine antenatal education.

1.2.3 Findings 1.2.3.1 Written information about breastfeeding


None of the three studies evaluating the effect of an educational intervention which provided written information in either the ante- or postnatal periods was shown to be effective at achieving a statistically significant increase in the duration of breastfeeding among women on low or high incomes. The self-help manual was developed with guidance from low-income women and included steps to motivate and enable women to breastfeed as well as problem-solve any breastfeeding difficulties (Coombs et al., 1998 [12]). While the intervention was effective at achieving a statistically significant between-group difference in rates of exclusive breastfeeding at 48 hours/one week postpartum, no significant differences were demonstrated for continued breastfeeding at three to four weeks and three months. Confidence in the findings is limited due to the poor quality of the study in some key areas, namely the lack of information on the method of randomisation used to allocate women to comparison groups, comparability of participants across groups and whether the study design included blinded outcome assessment. A further concern regarding the validity of these study findings relates to the voluntary nature of participation in this education intervention. The authors note that the participants across both groups had significantly stronger pro-breastfeeding attitudes than those women who did not volunteer to participate in the study. This confounding variable, together with the other methodological weaknesses of this study, limit the usefulness of its findings. The Italian study by Curro et al. (1997) [13] evaluated the effectiveness of a postnatal 10 minute oral counselling

session on breastfeeding during a paediatric visit and an information booklet including a set of instructions for practical breastfeeding management compared to only oral counselling. The booklet was not demonstrated to have a statistically significant increase in the duration of exclusive or any breastfeeding at six months although 82% of mothers in the intervention group did report the booklet was useful. This study was of moderate size (total sample of 200 participants across two trial arms) and met nearly all the quality criteria for this review. This study does provide confidence in the findings therefore that a postnatal breastfeeding booklet is not likely to be effective at increasing the duration of exclusive breastfeeding at six months among primiparous women, regardless of income. The third study evaluating written breastfeeding literature, in the form of a 33 page illustrated booklet delivered in the postnatal period, also failed to demonstrate effectiveness at increasing the duration of breastfeeding at six, 13, 20 and 52 weeks or increased satisfaction with the breastfeeding experience (Hauck and Dimmock, 1994 [15]). This study was conducted among Australian Caucasian women of higher incomes who were breastfeeding their baby. The findings may be partly explained by the high rates of duration of breastfeeding across both groups compared to national averages, probably reflecting the higher socio-economic status of this sample. In addition, significantly more women in the control group intended to breastfeed for seven to nine months whereas significantly more women in the intervention group intended to breastfeed until four to six months. The lack of comparability across groups for this key confounding variable limits confidence in interpretation of these findings. The poor quality of this study in terms of its randomisation method, poor reporting of losses to the study and blinding of outcome assessment further undermine confidence in its findings.

1.2.3.2 Breastfeeding self-assessment tool


The two evaluations of educational self-assessment tools suggest that this type of intervention may be effective at increasing duration rates of breastfeeding among some groups of women. One study, conducted in the US among women from different socio-economic groups, evaluated the effectiveness of a breastfeeding selfmonitoring daily log book in addition to a breastfeeding education session including a video and a question/answer session (Pollard, 1998 [18]). Women in the control group received the same educational session 37

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

and a notebook without the self-monitoring daily log book. This intervention was delivered in the postnatal period during the womans hospital stay to women who had intended to breastfeed and had initiated breastfeeding. While this intervention was not demonstrated to be effective at increasing the mean duration of breastfeeding across the whole sample, a significant increase in the mean duration of breastfeeding was demonstrated among those participants who adhered to the protocol by completing the self-monitoring forms. These women were older, married, higher income mothers who were committed to breastfeeding. They breastfed their infant three times longer than the low income, single mothers who were uncertain about their breastfeeding decisions and had not completed the selfmonitoring forms. This RCT can be considered to be of good quality with a true randomisation method, an a priori sample size calculation, appropriate reporting of withdrawals and comparability of groups at baseline. Outcome assessment was not blinded, however, and breastfeeding behaviour was self-reported by women. The findings of this trial can be interpreted with some confidence and suggest that a question/answer education session supported by a postnatal daily log of breastfeeding behaviour for completion by the mother can be effective at increasing the duration of breastfeeding among women in higher socio-economic groups who are committed to breastfeeding. The study conducted in Ireland evaluated a simple fact sheet covering eight positive aspects of breastfeeding, followed by a questionnaire designed to reinforce this intervention (Loh et al., 1997 [17]). The intervention took about three minutes and was conducted in individual clinic rooms by a medical student in the late antenatal period. Women in the control group received routine antenatal care only. This self-assessment intervention was demonstrated to achieve a 38% increase in breastfeeding rates at hospital discharge for women in the intervention group compared to women in the control group. Similar increases were also reported for breastfeeding duration rates at four weeks postpartum. Although these findings did not achieve statistical significance, this may be due to the small numbers in the study (total n=193 across two trial arms). The authors 38

have calculated a subsequent study would need to recruit over 300 participants in each group to achieve 80% certainty of detecting a 25% improvement in breastfeeding rates. The study met some of the quality criteria for this review although the method of randomisation was not stated and blinding of outcomes is not evident. In addition, the authors did not identify participants intended feeding method, a confounding variable which could significantly affect the findings. The lack of detail regarding the baseline characteristics of participants also limits interpretation of the findings for further replication among different target groups of women. This may be critical given the differential effect of the self-monitoring log book intervention (Pollard, 1998 [18]) detailed above. Indicative findings of this study do suggest, however, that this relatively simple, low-cost intervention may have a positive effect on increasing the duration rates of any breastfeeding among married women from the general population.

1.2.3.3 Antenatal paediatric visit


A single study evaluated a more formal approach to antenatal education among nulliparous, low income mothers who were mostly African-American (Serwint et al., 1996 [20]). Mothers in the intervention group received a scheduled prenatal visit from the infants future paediatrician at between 32 and 36 weeks gestation at a hospital-based clinic. Paediatricians had received three one-hour training sessions on counselling parents of newborn infants and breastfeeding techniques and promotion. Paediatricians provided oral counselling with no supporting breastfeeding literature on feeding options and the advantages of breastfeeding. This intervention was not demonstrated to be effective at increasing the duration of breastfeeding at 30 and 60 days postpartum among this target group, with mothers in both groups reporting they perceived their paediatrician knew them well. This trial is considered to be of good quality, meeting all the quality criteria except evidence of blinding. The calculated sample size was not achieved (actual sample of 156 participants across two trial arms) although the lack of positive effect demonstrated across the comparison groups does not suggest this is likely to be an explanation of the findings of no effect. This trial appears to provide conclusive evidence that a formal paediatric visit in the antenatal period is not likely to be effective at increasing duration rates among black American women on low incomes.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1.2.3.4 Group antenatal sessions


Three educational interventions included in this review evaluated the effect of group antenatal sessions on the duration of breastfeeding. A study conducted by Kistin et al. (1990) [16] targeted black American women on low incomes of whom approximately one third planned to breastfeed. This three-armed trial compared the effect of at least one 50-80 minute group education session with discussion of the benefits and potential problems of breastfeeding to a similar session delivered for 15-30 minutes on a one-to-one basis to standard clinic care. The group education session(s) demonstrated a statistically significant increase in duration rates of breastfeeding compared to the control group in hospital and at 12 weeks postpartum but not at two and six weeks. The one-to-one education session was found to have a significant increase in duration rates in hospital and at two weeks postpartum but not at six and 12 weeks when compared to the control group. Analysis by feeding intention suggests the strength of the group classes appears to be in helping women who had already considered breastfeeding to maintain their decision to breastfeed at 12 weeks. Conversely, analysis by feeding intention suggests the strength of the individual session appears to be in changing womens minds or getting them to initially consider breastfeeding. This trial had a relatively small sample size with only 159 participants across the three groups. Failure to conduct intention-to-treat analysis may affect the findings of effectiveness although the authors did report there were no significant differences between the characteristics of women completing the study compared to those lost to follow-up. An appropriate randomisation method was used, although only for two of the three groups the control group was not randomly allocated. Blinding of outcome assessment is not evident. This poor quality trial provides indicative findings of the potential of discursive, problem-solving education classes and one-to-one sessions to increase the duration of breastfeeding among black American women on low incomes. Another study evaluated the effect of an antenatal group teaching session on postnatal nipple pain, nipple trauma and breastfeeding rates among Australian nulliparous women who intended to breastfeed and were mostly on low incomes (Duffy et al., 1997 [14]). The one-hour teaching session focused on the correct position and attachment of the baby on the breast for feeding

including demonstration of correct positioning with a doll for each woman in the intervention group. Participants in the control group received standard antenatal hospital education. This intervention was shown to be effective, with a statistically significant between-group difference at increasing the duration of exclusive breastfeeding at six weeks postpartum. Improved position and attachment and reduced nipple pain and nipple trauma were also demonstrated as a result of the intervention. This trial is considered to be of good quality although the nonrepresentative nature of the small sample size limits generalisability of findings to women from all socioeconomic groups. Although the authors did not conduct intention-to-treat analysis, losses to comparison groups were small and equivalent and are considered unlikely to affect study findings. This trial clearly suggests that an antenatal group teaching session on positioning and attachment is likely to be effective at increasing the duration of exclusive breastfeeding at six weeks postpartum among women on low incomes. The third study which evaluated a group educational intervention was a culture-specific education programme to promote breastfeeding among Vietnamese women in Australia (Rossiter, 1994 [19]). The intervention group consisted of a 25 minute videotape followed by a series of three two-hour small group discussion sessions. The programme was conducted in Vietnamese with the assistance of a Vietnamese health interpreter. The control group received standard breastfeeding and childbirth leaflets. This intervention was shown to be effective, with statistical significance, at increasing knowledge, attitudes, intended and actual behaviour about breastfeeding at birth and four weeks postpartum. A significant increase was not demonstrated at six months postpartum although the effect of the intervention remained positive. The authors noted a greater proportion of women had given birth in their own country in the intervention group compared to the control group, with most women who had given birth in their country also breastfeeding their infants. These findings suggest that childbirth and breastfeeding experience in the womans own country may have been influencing factors which contributed to the study findings. The authors did not adjust for these differences. The quality of this trial is considered to be poor with failure to report the method of randomisation, 39

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

no evidence of blinding and failure to report the reasons for losses to the study groups. On reporting outcomes, the authors have not provided denominator data, limiting scope for verification of data validity. The methodological weaknesses in the study design further limit confidence in the study findings. This trial provides indicative but inconclusive evidence therefore of the potential of this culture-specific education programme to be effective at increasing duration rates of Vietnamese women on low incomes.

1.2.4 Discussion
The evidence base to evaluate the effectiveness of health education interventions has been limited to studies utilising a RCT study design to increase confidence in study findings. While many of these studies have some methodological weaknesses and provide indications of effectiveness requiring further research, a small number of good quality trials do provide confidence in their findings for potential implications for policy and practice. Written education materials have not been found to be effective at increasing the duration of breastfeeding among women in different income groups from developed countries (Coombs et al., 1998 [12]; Curro et al., 1997 [13]; Hauck and Dimmock, 1994 [15]). The single, good quality trial by Curro et al. (1997) [13] does provide confidence in its findings. The authors conclude: Although written material is perceived as low cost and unlikely to cause harm, there are substantial costs with promoting strategies of unproved effect. This is supported by the findings of an earlier review on interventions to promote the initiation of breastfeeding (Fairbank et al., 2000), which found that written materials alone did not increase initiation rates of breastfeeding among women in developed country settings, regardless of income group. This evidence base, although limited, does warrant serious caution against further breastfeeding promotion resources being allocated to the provision of written educational materials for the promotion of breastfeeding when used alone. See Section 1.3 for combined use with other strategies. Education interventions delivered through the use of a self-assessment tool appear to have potential to increase duration rates of breastfeeding among some groups of 40

women. The good quality trial evaluating the breastfeeding postnatal self-monitoring daily log (Pollard, 1998) [18] has demonstrated the effectiveness of this intervention to increase duration rates of breastfeeding among American women from higher socio-economic groups. However, widespread implementation of this intervention, regardless of socio-economic group and high self-efficacy to breastfeed, has the potential to increase inequalities in health, resulting in relatively poorer health status for infants from families with low socio-economic backgrounds. Delivery of this intervention to women from lower socio-economic groups who are less committed to breastfeeding requires further research to identify additional intervention components which would increase womens commitment to breastfeed in the antenatal period and provide additional support or incentives to complete the daily breastfeeding log. Furthermore, an appropriate breastfeeding daily log instrument would need to be developed in accordance with the needs of the local population. Further research is required to provide more conclusive evidence of the effectiveness of the use of a short questionnaire to reinforce the educational session delivered by medical staff in the antenatal period (Loh et al., 1997 [17]). In particular, further research should give reference to adequate sample size, analysis by feeding intention and socio-economic characteristics, method of randomisation and blinding. The authors state that many doctors might delegate the job of promotion of breastfeeding to midwives but previous studies have shown that the duration of breastfeeding is longer in mothers who perceived that their medical advisers favoured breastfeeding (Loh et al., 1997 [17]). The value of the research would be further enhanced if consideration were given to the most appropriate, as well as the most effective, health professionals to deliver such an intervention. Didactic prenatal education at a paediatric visit has been clearly shown to be ineffective at increasing duration rates of breastfeeding among African-American women on low incomes (Serwint et al., 1996 [20]). Findings from this good quality trial suggest further breastfeeding promotion resources would not be utilised effectively on replication of this intervention among this group of women. Two of the trials evaluating the effectiveness of group, discursive and/or culture-specific education sessions suggest that these forms of education can be effective at

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

increasing duration rates of breastfeeding among black American women on low incomes (Kistin et al., 1990 [16]) or Vietnamese women on low incomes (Rossiter, 1994 [19]). The poor quality of both of these trials provides indicative evidence only. Further high quality research is required to evaluate the effectiveness of these potentially effective interventions in local settings among similar target groups experiencing disadvantaged circumstances. A group educational session on positioning and attachment of baby at the breast has been clearly shown to be effective at increasing the duration of exclusive breastfeeding at six weeks postpartum among nulliparous Australian women on low incomes who intended to breastfeed (Duffy et al., 1997 [14]). This is consistent with the supporting evidence base regarding a clear association between poor positioning and attachment, nipple pain and early cessation of breastfeeding (see Section 3.3). This high quality trial provides confidence in the study findings to enable widespread replication of this intervention among similar population groups in UK settings.

Breastfeeding self-assessment tools A postnatal breastfeeding self-monitoring daily log book (Pollard, 1998 [18]) An antenatal breastfeeding fact sheet and questionnaire (Loh et al., 1997 [17]). Informal, discursive problem-solving education sessions Repeated one-to-one antenatal sessions (Kistin et al., 1990 [16]) Group antenatal sessions (Kistin et al., 1990 [16]) Culture-specific education programmes (Rossiter, 1994 [19]). A single antenatal group education session on positioning and attachment has been clearly shown to be effective at increasing the duration of exclusive breastfeeding at six weeks postpartum among nulliparous women in Australia who are on low incomes and intend to breastfeed (Duffy et al., 1997 [14]). See Findings and Discussion above for details of the intervention. Although this evidence base is limited to one study, the quality of the evidence provides confidence in its findings and supports replication of this intervention to women of similar circumstances in UK settings. As with Section 1.1, the introduction of the national Healthy Start programme provides a unique opportunity to implement and evaluate widespread, routine provision of education to women on low incomes.

1.2.5 Future research


A number of interventions have been shown to be potentially effective at increasing duration rates of breastfeeding among some target groups. This evidence base is indicative and not conclusive, however, as a result of the poor quality of the available research, including a common failure to account for the key confounding variable of feeding intention. The indicative nature of the existing evidence base suggests that small-scale evaluations at a local level are an appropriate first step to identify the potential for wider replication of specific interventions at a regional or national level. Adequately resourced, high quality research is required therefore to evaluate the effectiveness of these interventions to increase the duration of breastfeeding among the priority group of women from disadvantaged circumstances in different local settings in the UK. Sure Start programmes, committed to the promotion and support of breastfeeding in accordance with local needs, are ideally placed, given the appropriate resources, to conduct the following evaluations of these potentially effective interventions (see Findings and Discussion above for details of interventions and recommended research questions/methodologies):

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

41

42 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention A prenatal self-help manual compared to routine bf antenatal education Antenatal: around 7 weeks before the birth Exclusive bf at 48h/1 week; bf (not defined) at 3-4 weeks and 3 months Timing of breastfeeding duration outcomes Antenatal clinic serving high % of low income women. Participants were mostly single, African-American, primiparous with pro-bf attitudes (n=200) Primiparous women, exclusively bf @10-20 days postpartum (n=200) Oral counselling and Postnatal: immediately after information booklet compared recruitment 10-20 days to oral counselling only postpartum Median duration of exclusive and any bf; prevalence of exclusive and any bf Exclusive bf at 6 weeks Nulliparous women who intended to bf, mostly low income (n=75) bf women, mostly married, higher socio-economic groups, Caucasian (n=150) Black women, low income (n=159) Bf information booklet compared to no booklet Teaching session on position and attachment compared to routine antenatal education Antenatal: after 36 weeks gestation Postnatal: 5-7 days postpartum bf (at least once a day) at 6, 13, 20 and 52 weeks Group prenatal or one-to-one Antenatal prenatal education compared to routine prenatal education Education fact sheet and questionnaire compared to routine antenatal care Antenatal: after 36 weeks gestation bf (at least once a day) in hospital and at 2, 6 and 12 weeks Women of >36 weeks gestation, mostly married and one third primiparous (n=193) Any bf at hospital discharge and 4 weeks

Table 1.2: Included studies public health interventions: educational interventions

Name of authors and study number

Date of Country publication

Design

Coombs et al. [12]

1998

US

RCT

Curro et al. [13]

1997

Italy

RCT

Duffy et al. [14]

1997

Australia

RCT

Hauck and Dimmock [15]

1994

Australia

RCT

Kistin et al. [16]

1990

US

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Loh et al. [17]

1997

Ireland

RCT

Table 1.2: Included studies public health interventions: educational interventions (cont.) Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Education and self-monitoring Postnatal: during hospital intervention compared to stay education only Mean duration of any bf Timing of breastfeeding duration outcomes

Name of authors and study number Primiparous women who planned to, and initiated bf, mostly Caucasian, married and further education (n=86) Ethnic Vietnamese or Vietnamese born women, mostly married, low income and multiparous (n=194) Nulliparous, low-income women, mostly AfricanAmerican (n=156) Routine care plus prenatal paediatric visit compared to routine prenatal care Antenatal: 32-36 weeks gestation Culture-specific video and Antenatal group education compared to standard bf leaflets bf (as the main feeding method) at 4 weeks and 6 months

Date of Country publication

Design

Pollard [18]

1998

US

RCT

Rossiter [19]

1994

Australia

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Serwint et al. [20]

1996

US

RCT

Any bf at 30 and 60 days postpartum

43

1.3 Public health interventions: multifaceted interventions


Topic Multifaceted activities relevant to the promotion and support of breastfeeding duration. Interventions are often delivered within the context of programme-based social services targeted at specific vulnerable groups, for example the US Special Supplemental Feeding Programme for Women, Infants and Children (WIC) Summary of methods
Study design of included studies: RCTs Dates for searches: 1990 to June 2003 Population group included: pregnant and postpartum women, offered the support of one or more of these programmes, from disadvantaged groups Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of women and care providers, relevant clinical outcomes (occurrence of breastfeeding problems, weight problems for baby, health outcomes for baby, etc), relevant health service outcomes including economic outcomes plus relevant health service outcomes including uptake of programme elements, economic outcomes

1.3.1 Background
A multifaceted intervention is defined, for the purposes of this review, as one that includes more than one component; for example, the provision of both breastfeeding education and peer support. In addition, the intervention group(s) must be receiving at least two service components more than the comparison group receiving routine care. For example, in the case of an intervention being delivered to participants in the WIC programme, the control group is likely to be receiving breastfeeding education as routine care. The delivery of an additional peer-support component is not classified as a multifaceted intervention whereas a study evaluating an additional peer support and additional education components do fall into this category. In summary, a multifaceted intervention is an evaluation of several intervention components over and above that being delivered to the comparison group. Multifaceted interventions have been shown to have an important role in increasing the initiation of breastfeeding where the delivery of combined breastfeeding promotion strategies have an incremental effect to achieve overall effectiveness (Fairbank et al., 2000). Such interventions 44

are particularly important in the health promotion arena where behaviour change can be more readily achieved by utilising a variety of strategies to address different barriers to change, for example a lack of knowledge and a lack of capacity to implement new knowledge.

1.3.2 Evidence base


Sixteen studies (based on 18 papers where two studies each had two published papers) were considered for possible inclusion in this review. They includes nine RCTs. Given the strength of the evidence base from these RCTs (albeit of variable quality) inclusion of the remaining studies with a less rigorous study design was not considered beneficial to the findings of this review. On this basis, a total of nine studies (10 papers where two publications are relevant to one study) have been included (see Table 1.3 for details). Eight of the nine studies were conducted in the US with the remaining study carried out in Australia (Redman et al., 1995 [26]). Three studies were targeted at a general population of women on low incomes (Brent et al., 1995 [21]; Grossman et al., 1990 [25]; Sciacca et al., 1995a

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

[29]) with a further two studies targeting black women on low incomes (Finch and Daniel, 2002 [23]; Fredrickson, 1995 [24]). Three studies were evaluations of multifaceted interventions delivered to women on higher incomes (Campbell, 1996 [22]) or women classified as upper-middle class (Rojjanasrirat, 2000 [27]; Schy et al., 1996 [28]). The remaining study was delivered to all women from the general population (Redman et al., 1995 [26]). Five of the studies were specifically targeting women who intended to breastfeed with the aim of achieving a sustained breastfeeding experience for those women (Campbell, 1996 [22]; Grossman et al., 1990 [25]; Redman et al., 1995 [26]; Rojjanasrirat, 2000 [27]; Schy et al., 1996 [28]). Six of the nine RCTs were evaluations of a combined intervention of breastfeeding education and support, five of which comprised antenatal education and postnatal support at varying levels (Brent et al., 1995 [21]; Campbell, 1996 [22]; Redman et al., 1995 [26]; Rojjanasrirat, 2000 [27]; Schy et al., 1996 [28]) and one comprising postnatal education and support (Grossman et al., 1990 [25]). The postnatal support was provided by health professionals in all six studies. Two RCTs evaluated the additional use of incentives to promote the duration of breastfeeding. In one study, incentives were provided to promote the maintenance of breastfeeding until two months postpartum (Finch and Daniel, 2002 [23]); another trial used incentives to promote participation in a partner-supported antenatal education and postnatal peer support programme as well as maintaining breastfeeding up to three months postpartum (Sciacca et al., 1995a [29]). One two-by-two factorial trial evaluated standard versus tailored breastfeeding antenatal education and/or standard versus non-formula hospital discharge packs (Fredrickson, 1995 [24]).

The theory based evaluation conducted by Campbell (1996) [22] was an additional prenatal education class on breastfeeding based on Banduras concept of selfefficacy (Bandura, 1977). The class included observation learning, physical practice, cognitive practice and performing the mechanics of breastfeeding. Participants received class notes, advice about initiating breastfeeding in hospital and encouragement to attend a local breastfeeding support group. New mothers received telephone calls at home at one, two, three and six weeks postpartum for continuous structured breastfeeding assessment, persuasion and gathering of physiological status information. Participants in the control group also received an additional prenatal class, excluding the breastfeeding educational content, and the same postnatal visit schedule. The authors note that postnatal breastfeeding support and assistance was provided to some mothers in the control group during routine telephone contacts due to ethical concerns. Although the mean duration of breastfeeding was similar in both groups (42 days for the intervention group compared to 37 days for the control group), the authors evaluation of breastfeeding success, having adjusted for the pre-treatment measure of self-efficacy of breastfeeding, demonstrated a statistically significant between-group difference (p<0.05). Confidence in these findings is limited, however, due to the very poor quality of the trial. The sample size was small (total n=63) with a poor randomisation method, inappropriate outcome analysis at the level of the individual despite a group method of allocation, and a lack of blinding and intentionto-treat analysis, including poor reporting of group numbers and withdrawals. Replication of the intervention for further, better quality evaluation and/or potential generalisability into practice is also limited by the complex nature of an intervention based on measuring and evaluating a theoretical concept such as self-efficacy. The education and support intervention studied by Schy et al. (1996) [28] comprised a 30-60 minute antenatal education session based on a specifically designed lactation curriculum. The curriculum included physiology of breastfeeding, positioning, feeding patterns, evaluating the adequacy of breastfeeding in the newborn infant, pumping and storing of breastmilk, and bottle supplementation and problem solving. Lactation specialists provided daily postnatal support for the short period of hospital day, usually between two to four days. Women in the control group received routine education 45

1.3.3 Findings 1.3.3.1 Antenatal education and postnatal support


Of the five trials evaluating the combined effect of antenatal education and postnatal support delivered by health professionals, three were conducted among women on higher incomes who intended to breastfeed (Campbell, 1996 [22]; Rojjanasrirat, 2000 [27]; Schy et al., 1996 [28]). None of these trials demonstrated a statistically significant increase in the duration of breastfeeding as a result of the intervention.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

and postnatal care where staff nurses provided breastfeeding assistance with the first breastfeed and as needed. A breastfeeding support telephone helpline was available seven days a week to all participants. Results were not reported by group but the authors note that no statistically significant differences were found between the groups for duration of breastfeeding at four to six months or still breastfeeding at six months. Breastfeeding rates were relatively high in both groups, however, with 29% of women still breastfeeding at six months. Poor quality of the trial regarding data reporting, lack of clarity regarding the randomisation method and lack of baseline comparability between groups limits confidence in the findings. This study provides indicative evidence therefore of the likely lack of effectiveness of this type of intervention to increase duration rates of breastfeeding among upper-middle class women who intend to breastfeed. The third multifaceted antenatal education and postnatal support intervention targeted at upper-middle class women who intend to breastfeed had a particular focus on increasing the duration of breastfeeding among women who intend to return to work within 12 weeks postpartum (Rojjanasrirat, 2000 [27]). This intervention comprised a single two to three hour breastfeeding lecture, including information specific to combining breastfeeding and employment, delivered in the antenatal period by a certified lactation consultant or educator. A panel of experienced, employed breastfeeding women served as role models during the education session and the involvement of a partner or significant other to attend the session was encouraged. Follow-up support included telephone calls at one week after hospital discharge and four to six weeks postpartum, both by a certified lactation consultant. Participants in the control group received routine antenatal breastfeeding education delivered by a mother-baby nurse, nurse educator or lactation consultants and routine follow up by an obstetric nurse or lactation consultant. The study did not demonstrate a statistically significant increase in the duration of breastfeeding at four or 16 weeks postpartum as a result of the intervention. Initiation and duration rates of breastfeeding were high across both groups, however, with over 80% of women still breastfeeding at 16 weeks postpartum. The authors note the findings could be partly explained by the discrepancy between the planned and actual intervention delivered, whereby participants did not fulfil the request 46

to bring their significant other and the follow-up by the certified lactation consultant was reported to be at inconvenient times. Further, routine postpartum followup care for participants in the control group was reported to provide as good a source of support as that provided to the intervention group. The lack of compliance with the intervention protocol and the lack of effectiveness to increase the duration of breastfeeding suggests this education and support intervention does not warrant further investigation or replication. The relatively good quality of this trial provides confidence in this conclusion. Two further antenatal education and postnatal support interventions were delivered to women on low incomes (Brent et al., 1995 [21]) and to a general population of women intending to breastfeed (Redman et al., 1995 [26]). The education delivered to Australian women intending to breastfeed (Redman et al., 1995 [26]) comprised a three hour teaching session for mothers and support persons at 24-28 weeks gestation. The semistructured sessions included discussion on the benefits and management of breastfeeding, positioning, methods for determining whether the baby is receiving enough milk and rest after birth and caesarean section. Women were shown a five minute video and received a pre-birth package including a glossary of breastfeeding terms, a list of breastfeeding services, common question and answer sheets and a wall planner. The postnatal support comprised hospital visits shortly after delivery and during the hospital stay, including receipt of further education materials, a phone call at two to three weeks and three months postpartum by a breastfeeding consultant, a home visit if requested and an optional postnatal discussion group at six to eight weeks. Although the majority of mothers who responded to an evaluation questionnaire reported they found the intervention useful, a statistically significant increase in the number of women exclusively breastfeeding at six weeks or four months was not reported as a result of the intervention. The authors conclude that educational programmes which target motivational variables may not be important in determining breastfeeding rates among this pre-motivated group of women who intend to breastfeed. This well-sized trial is considered to be of good quality, meeting nearly all the quality criteria of this review, providing confidence in the finding that this type of antenatal education and postnatal support intervention is not likely to be effective at increasing duration rates of breastfeeding among women who intend to breastfeed.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

A comprehensive antenatal education and postnatal support programme delivered to American women on low incomes, including approximately one third who had stated a probable intention to breastfeed, did however show significant increases in the duration of breastfeeding as a result of the intervention (Brent et al., 1995 [21]). The prenatal education consisted of two to four individual 10-15 minute sessions with a lactation consultant to discuss the benefits and practice of breastfeeding. The content of the sessions was tailor made to the needs of each participant. After birth, mothers were followed up with daily in-patient rounds by the lactation consultant, were telephoned at home 48 hours after discharge, had a visit to the lactation clinic at one week and a further lactation consultation at each health supervision visit until weaning or when the infant was one year old, whichever came first. Professional education was also directed at the nursing and medical staff who interacted with the breastfeeding dyad. Routine care in this hospital consisted of optional prenatal breastfeeding classes, postpartum breastfeeding instruction by nurses and doctors and outpatient followup in the paediatric ambulatory department. This education and support programme, providing regular and consistent information and assistance throughout the preand extended postnatal periods, was demonstrated to achieve a statistically significant increase in the numbers of women breastfeeding in hospital and at two weeks and two months postpartum. The effect was not significant at six months although this may be due to the relatively small size of the initial sample. Analysis of breastfeeding rates in hospital by feeding intention (those women planning to bottlefeed) also demonstrated a significant effect of the intervention. This trial is considered to be of good quality in key areas such as a true randomisation method, clear inclusion and exclusion criteria, comparability of groups at baseline and analysis for the confounding variable of feeding intention. Design weaknesses include a failure to conduct intentionto-treat analysis and to report withdrawals beyond the initiation of breastfeeding, a lack of blinded outcome assessment and the moderate sample size (total n=123) combined with an absence of an a priori sample size calculation. The quality of this study provides indicative evidence therefore that this type of comprehensive education and support intervention is likely to be effective at increasing the initiation and duration rates of breastfeeding up to at least two months postpartum

among women on low incomes who intend to either breast or bottlefeed.

1.3.3.2 Postnatal education and support


A single study evaluated the effect of a package of education and support delivered in the postnatal period among American women on low incomes who had stated an intention to breastfeed when the baby was born (Grossman et al., 1990 [25]). Women in the intervention group received a 30-45 minute face-to-face meeting in hospital with a lactation counsellor (a registered nurse) shortly after birth. An education booklet was also provided at this meeting. Telephone contact was made on days two, four, seven, 10 and 21 postpartum and a telephone helpline staffed by a nurse or paediatrician was available. A lactation clinic provided back-up support for mothers experiencing breastfeeding problems. The control group received routine postnatal teaching about infant care and feeding by obstetrical nursing staff. This type of multifaceted intervention was not shown to be effective at increasing duration rates of breastfeeding at six weeks, three months or six months or at increasing the median duration of breastfeeding. The methodological quality of this study was fundamentally flawed, however, due to the poor randomisation method used and the reported cross-contamination between groups due to the presence of the lactation specialist on the hospital wards. Confidence in these findings is limited, therefore.

1.3.3.3 Education, with or without support, plus incentives


A trial conducted in the US among low-income women (WIC participants) evaluated the provision of incentives to promote sustained breastfeeding until two months postpartum (Finch and Daniel, 2002 [23]). The intervention consisted of a prenatal breastfeeding education session by a trained counsellor with incentive marketing in the form of a truth or myth activity in small group format. This was followed by instruction and discussion accompanied by handouts. It appears that women in the intervention group gained access to breast pumps and those women who exclusively breastfed until two months received a financial incentive in the form of a $25 mall gift certificate. Women in the control group received routine WIC prenatal breastfeeding education regarding the general benefits of, and barriers to, breastfeeding. Participants in both groups received the usual WIC educational materials, peer and professional 47

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

support and incentives such as a greater quantity of food vouchers and extended programme eligibility for breastfeeding participants. The report suggests, although it is not clear, that all women who exclusively breastfed or who did not receive formula were eligible to receive a food package valued at more than $50 per month. While this intervention was not found to be effective at increasing the duration of any breastfeeding at hospital discharge or at two months postpartum, a statistically significant increase was demonstrated for exclusive breastfeeding at both time points (Finch and Daniel, 2002 [23]). This suggests this specific type of educational approach and/or the incentives of breast pumps and/or the $25 mall gift certificate are likely to influence women on low incomes to exclusively breastfeed for longer in addition to the usual WIC education, support and financial incentives. The lack of clarity on which incentives were provided to either or both comparison groups does warrant caution in interpretation, however. Furthermore, these findings are based on extremely small numbers (total n=48) and do not include the large number of women who withdrew from the intervention group (n=11/30) and the single participant who was lost to follow-up from the control group. The method of randomisation has not been stated for this trial, further limiting confidence in its findings. This study provides indicative evidence therefore of the potential use of incentives to promote the duration of exclusive breastfeeding among low-income women. The study conducted by Sciacca et al. (1995a/b) [29], also among low-income women participating in the WIC programme, evaluated the use of incentives to promote both the participation of women and their partners in a breastfeeding education programme and the maintenance of breastfeeding for at least half of the time at hospital discharge, six weeks and three months postpartum. The single breastfeeding two-hour class covered breastfeeding concerns of both parents, basics of milk production, nursing techniques, typical problems and solutions and pumping and storage information. The importance of early nursing was stressed and the benefits and myths of breastfeeding were addressed. Women received a gift bag and a breast pump and partners received a pair of tickets to a football game. Gift incentives were also given to couples attending the five one-hour childbirth preparation classes. Women in the intervention group were automatically assigned a peer counsellor from the WIC peer support programme 48

and were encouraged via an incentive to contact the counsellor within two days of delivery. Details of the gift incentives offered to breastfeeding mothers are not provided. Women in the control group received the local WIC programme prenatal childbirth preparation classes including one unit on breastfeeding versus formula feeding, the breast-pump rental programme, infant T-shirts with a breastfeeding promotion message, clinic displays of breastfeeding promotion posters, the optional breastfeeding group classes and optional peer support programme for breastfeeding mothers. It appears, although it is not clearly stated, that mothers in the intervention group are eligible for all these components of the WIC programme in addition to those services provided as part of the study intervention. This incentive-based antenatal education and peer support programme was demonstrated to be effective at achieving statistically significant between-group differences in breastfeeding rates for both any and exclusive breastfeeding at hospital discharge, two and six weeks and three months postpartum (Sciacca et al., 1995a/b [29]). This relatively small trial (total n=68) has several methodological weaknesses including failure to report the randomisation method or blinding of outcome assessment and an absence of intention-to-treat analysis. Failure to include the 19% (n=13/68) of participants who withdrew from the study may affect study findings. These limitations warrant some caution in interpreting the findings although this trial does provide indicative evidence of the potential effectiveness of an incentivebased, partner-supported education and support programme to increase the duration of both any and exclusive breastfeeding among women on low incomes.

1.3.3.4 Antenatal education and hospital discharge packs


A two by two trial conducted in the US among lowincome women at a WIC clinic evaluated standard versus tailored breastfeeding antenatal education and/or standard versus non-formula hospital discharge packs (Fredrickson, 1995 [24]). The intervention teaching lasted approximately 10-15 minutes and was tailored to the specific barriers to breastfeeding reported by the mother. The woman subsequently received five letters at weekly intervals addressing the specific knowledge points. The standard WIC education at the local study clinics comprised a query from a WIC nutritionist regarding intended feeding practice and answering of the mothers specific questions in accordance with WIC nutrition

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

education guidelines. The intervention discharge pack was consistent with the WHO Code of Marketing of Breastmilk Substitutes and was of equal commercial value to formula discharge packs (US$6). Contents included a baby blanket, money-off coupon for purchasing a breast pump, a baby care book written at 8th grade reading level, re-usable breast pads and a bilingual breastfeeding how-to leaflet. The formula discharge pack contained advertising material prepared by an infant formula company and bottles of formula or water and teats. A total of 382 participants were recruited for allocation to one of the four comparison groups. This trial is considered to be of good quality, meeting nearly all the quality criteria for this review. The study found that although breastfeeding duration was higher in the three groups receiving either or both interventions than in the group receiving neither (at weeks two, four, 10, 16, 20, 24), the main effects of the interaction did not demonstrate a statistically significant increase in the duration of breastfeeding. Exploratory modelling analyses did suggest, however, that the teaching intervention or non-formula discharge pack had strong effects with a posterior defined sub-groups. In the case of the teaching intervention, breastfeeding duration was highest among mothers with a plan to breastfeed and low breastfeeding knowledge and lowest among both mothers with teaching and breastfeeding knowledge but no plan and mothers without any of the three. Therefore, the presence of a plan to breastfeed appears critical to the potential effectiveness of the teaching intervention. The intervention discharge pack was associated with a significant increase in breastfeeding duration among women with high breastfeeding knowledge. Conversely, formula discharge packs may have significant adverse effects on breastfeeding duration among sub-groups of low-income mothers, particularly those without breastfeeding knowledge. In summary, this trial provides a strong evidence base for the potential effectiveness of a tailored teaching and knowledge-based intervention to increase breastfeeding duration among women on low incomes who intend to breastfeed. Of equal importance, this trial highlights the potential adverse effects on breastfeeding duration which are likely to occur if women on low incomes who do not have knowledge about breastfeeding receive formula discharge packs.

1.3.4 Discussion and conclusions


The combination of antenatal education and limited postnatal telephone support has been tested among high income women and women who intend to breastfeed; it has been shown not to be effective and is probably not worth further investigation. The high rates of breastfeeding duration generally among this group, regardless of the intervention, suggests that breastfeeding promotion and support resources may be better spent where a significant impact is more likely to be achieved, ie disadvantaged groups. One study found that women from low-income groups, with no knowledge of breastfeeding, were adversely affected by being given discharge packs with formula information, whether or not they received additional education and support. This adds to the evidence presented in Section 1.5 that such packs should not be used; at present they are not given to women in the UK. Other studies outlined in this section have provided indicative evidence of a range of ways in which multifaceted interventions may be effective in low-income communities. None of them, however, were conducted in the UK. Sure Start schemes, and the new Healthy Start scheme, offer opportunities for testing some of these interventions in suitable UK settings. Suggestions for such evaluations are outlined here.

1.3.5 Future research


Among women from low-income groups, the comprehensive approach tested in Brent et al. (1995) [21] appears to be effective. This unusual approach included antenatal education tailored to individual womens needs, proactive visits in hospital and at home after birth, and ongoing availability of a lactation consultant. It also included breastfeeding education for all staff caring for mothers and babies. This should be replicated with evaluation in a practice setting, targeting women on low incomes. This could be conducted through Sure Start schemes. Tailored antenatal teaching education was also part of another set of interventions shown to be effective for low-income women who plan to breastfeed. Such tailored education could be replicated more widely as a stand-alone intervention, but it requires monitoring and 49

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

evaluation of effect in UK settings. The new Healthy Start scheme may be one vehicle for testing such an intervention. There is indicative evidence of a positive effect of the combination of education and support with incentives. Again, this is worthy of replication in local English settings among women on low incomes with high quality, well-sized evaluation trials. Sure Start schemes would provide a suitable setting for such replication. One of the before-after studies identified during this review was an evaluation of a truly multifaceted intervention which aims to facilitate both appropriate structural change within the healthcare system and micro-level change within the individual (Wright et al., 1997). The intervention included elements of training of hospital staff and subsequent policy and practice change, culturally appropriate education and peer support programmes, and community media activities, based on the principles of community empowerment from Paulo Frieres work (see Robertson and Minkler, 1994) and social marketing techniques based on Manoff, 1985 (see Glossary). This intervention appeared to be effective at significantly increasing the duration of any and exclusive breastfeeding among native American women. While this study fell outside the scope of this review it does suggest the need for further research to evaluate the effectiveness of such an intervention among different population groups of women in different settings in the UK. When read in conjunction with Sections 1.1 and 1.2, this section suggests that more care is needed to evaluate the components of multifaceted interventions that are most effective. Studies are needed to examine ways in which informal/family support can be used in conjunction with professional/peer support; ways in which women from low-income groups can be supported in exclusive breastfeeding using combinations of education, support, and incentives; and the training and preparation needed for those who work with breastfeeding women, whether peer or professional.

50

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Table 1.3: Included studies public health interventions: multifaceted interventions Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Professional lactation education and support compared to routine hospital education and support From first prenatal visit up to weaning of infant or 1 year old, including follow-up support at 48 hours after discharge, 1 week postpartum and all subsequent health visits Prenatal education class and support during hospital stay, within 72 hours of delivery and at 1, 2, 3 and 6 weeks postpartum Any bf at 6 weeks Any bf in hospital, 2 weeks, 2 and 6 months Timing of breastfeeding duration outcomes

Name of authors and study number Low income, nulliparous, pregnant women (n=123)

Date of Country publication

Design

Brent et al. [21]

1995

US

RCT

Campbell [22] Primiparous, pregnant women who intended to bf. Mostly white, higher education and higher incomes (n=63) Education and support compared to similar education and support programme without bf focus

1996

US

RCT

Finch and Daniel [23] Low income, mostly AfricanAmerican and Hispanic, pregnant women (n=60) Low income, pregnant women, 54% AfricanAmerican (n=382) 2x2 trial of standard versus tailored bf education and standard versus non-formula discharge packs Intensive bf education and support compared to routine nurse Education and support programme compared to routine education and support

2002

US

RCT

Education and additional Prenatal education session incentive compared to routine and incentives at birth and bf education and incentives 2 months postpartum Antenatal education and packs at hospital discharge

Exclusive and any bf at hospital discharge and mean duration of bf Any bf at 2, 4, 10, 16 and 24 weeks

Fredrickson [24]

1995

US

RCT

Grossman et al. [25] Low-income mothers who had stated an intention to bf (n=97) Primiparous pregnant women intending to bf (n=235)

1990

US

RCT

Education after birth and support on days 2, 4, 7, 10 and 21 Antenatal education at 24-28 weeks gestation and postnatal support in hospital and at 2-3 weeks, 6-8 weeks and 3 months postpartum

Any bf at 6 weeks and 3 and 6 months

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Redman et al. [26]

1995

Australia

RCT

Exclusive bf at 6 weeks and 4 months

51

52 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention A nurse education and support programme with partner/significant other compared to routine education and support Antenatal education class and Any and exclusive bf at 4 and 16 weeks support at 1 and 4-6 weeks postpartum Timing of breastfeeding duration outcomes Primiparous pregnant women, enrolled in bf classes, intending to bf and return to work within 12 weeks postpartum. Mostly white, upper-middle class, well educated (n=166) Pregnant women who intended to bf but had no bf experience. Mostly white, upper-middle class and married (n=150) Primiparous, low-income pregnant women with interest in participating in partnersupported programme (n=68) Incentive-based, partnersupported education and peer support programme compared to routine education and optional peer support programme Education and support programme compared to routine hospital care plus bf support telephone helpline Antenatal education class and Mean duration of any bf and daily support during hospital any bf at 4-6 and 6 months stay (2-4 days) Antenatal education class and Any and exclusive bf at hospital discharge, 2 and 6 incentives, postnatal peer weeks and 3 months support contact from 2 days postpartum and incentives at discharge, 6 weeks and 3 months

Table 1.3: Included studies public health interventions: multifaceted interventions (cont.)

Name of authors and study number

Date of Country publication

Design

Rojjanasrirat [27]

2000

US

RCT

Schy et al. [28]

1996

US

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Sciacca et al. (a&b) [29]

1995

US

RCT

1.4 Public health interventions: community based interventions


Topic Interventions intended to promote and support breastfeeding targeted at community groups, such as schools, media campaigns, etc

Summary of methods
Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional) Dates for searches: 1990 to June 2003 Population group studied: all relevant target groups, including schoolchildren, specific population groups, and the community at large Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those providing and experiencing the programmes (including attitude change to breastfeeding by participants), relevant process outcomes including uptake of programme elements, economic outcomes

1.4.1 Background
Breastfeeding is a natural behaviour and one that mothers will learn how to perform through observing others and through instruction. Partners will also learn about breastfeeding from a variety of sources. This learning will be mediated through many potential sources of information from close family, the local community and the media. Opportunities to breastfeed, or to see others breastfeed, will be influenced by the policies of employers and those who manage access of people to public places. For many sub-cultures in England, these opportunities to see breastfeeding are limited (Hoddinott and Pill, 1999) and exposure to bottlefeeding is more frequent, eg among Scottish teenagers (Carter et al., 2001). It may be many years before information such as formal education in school, or health promotion advice imparted through contact with health services such as at a GP well-woman clinic, can be used in practice. Of particular concern in many reports of the perceived barriers to breastfeeding, particularly among socially excluded groups and those whose culture may not be supportive of breastfeeding, is the view that there are pervasive cultural norms against breastfeeding. This

combines with messages that bottlefeeding is at least an equivalently healthy and more convenient option. It also includes messages that breastfeeding as a behaviour is repugnant or is in conflict with the sexualisation of the breast. Henderson et al. (2000) reported a study of media portrayals in British newspapers and television during one month. Media coverage implied breastfeeding is problematic, embarrassing, funny and associated with middle class or celebrity mothers. Bottlefeeding was far more ubiquitous and portrayed as normal, integrated and associated with ordinary families. The health benefits of breastfeeding and hazards of bottlefeeding were all but absent. These messages will be part of the social norm that has been shown to be an important determinant of intention to perform health behaviours (Fishbein and Ajzen, 1975). They affect not only adults but also children. A study in Liverpool, for example, examined the attitudes of schoolchildren and found that by 11 years of age, they already saw breastfeeding as embarrassing and bottlefeeding as more convenient and fashionable (Gregg, 1989). Disrupting this cycle of beliefs will therefore be difficult and will need to involve interventions that affect all age groups. 53

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Research using psychological theory has been used to elucidate the attitudes and beliefs of teens in Scotland as part of the process of building health education for schools (Swanson and Power, 2000; Carter et al., 2001). It was found that the most significant predictors of attitudes to infant feeding were knowledge (generally poor), previous experience or exposure to breast and bottlefeeding, social normative beliefs and perceived barriers to breastfeeding. It is advised that analyses of interventions include gender, age and socio-economic variables to take account of different patterns of attitudes and beliefs. A study in progress in England (Dunn et al., 2003) has found that psychological variables derived from the Theory of Planned Behaviour (Ajzen and Madden 1986; Ajzen, 2001) were predictive of intentions held at six weeks to breastfeed to four months, and to actual breastfeeding behaviour to both four months and six months. These variables added to the ability to predict behaviour beyond demographic variables (such as the age of the mother and her level of education) alone. The predictive power of the psychological variables declined after the six month stage. However, this suggests that there may be an important window of opportunity for interventions aiming to support breastfeeding women between six weeks and six months. Intention in this period is likely to be influenced by beliefs about how to feed the infant and return to employment. Interventions to support continued breastfeeding may be required if more mothers are to achieve the length of exclusive breastfeeding now advised by the World Health Organization. Dunn et al. (2003) suggest that this support could be welcomed as approximately three out of four mothers who stopped between six weeks and four months would like to have breastfed for longer. Influences on mothers and partners intentions to cease breastfeeding, or to move to mixed feeding, will include perceptions of ease of expressing and storing breastmilk, and opportunities to breastfeed with the baby while out of the home. Employment policies concerning mothers returning to work are likely to influence this, as are paternity leave policies aimed at allowing men to share caring at home. The latter may enable some mothers to breastfeed more effectively, while others may choose to return to work while the father provides care. This would adversely affect breastfeeding unless milk expression was well established. 54

Interventions to support breastfeeding by school education, community intervention and media campaigns have been reported in several descriptive studies. This includes evaluation of Sure Start and Sure Start Plus programmes in England and Wales. However, their impact on breastfeeding outcomes requires more rigorous designs. Some studies referred to in other sections, especially those that form part of the US Women, Infants and Children (WIC) programmes (eg Section 1.3), have elements of community intervention. This section is concerned with research where any of these socialeducational interventions have been subject to a controlled study where there are breastfeeding duration data available.

1.4.2 Evidence base


There were no controlled studies found that are related to this section.

1.4.3 Discussion and conclusions


In the absence of suitable controlled studies, we describe some key papers that show areas for development of policy and research. Papers discussed here were noted in the screening phase of this study as relevant background material for this section or were known to the authors.

1.4.3.1 Employment practices and interventions in the workplace


In a culture such as the UK where government policy supports the active participation of women in the workforce, and return to work after childbirth is common, the protection of breastfeeding women in the workplace is a crucial part of any national policy. In recognition of the difficulties of breastfeeding and working, the International Labour Organization (ILO) has for many years set standards for the protection and promotion of breastfeeding, which include a minimum of 14 weeks paid maternity leave, entitlement to one or more paid breastfeeding breaks daily, or daily reduction in working hours without loss of pay, and job protection and non-discrimination for breastfeeding workers (International Labour Organization, 2000). It has been noted in a recent review of employment policy in Europe that the UK seems to be the EU country with the least compliance with ILO standards (EU, 2003), though it has recently extended its paid maternity leave to six months.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Mothers often anticipate that returning to work may be a barrier to continued breastfeeding and may experience additional stress on return to work from conflicts between work and nursing roles (Auerbach and Guss, 1984). The potential adverse effects on breastfeeding from returning to work can be mitigated by ensuring mothers learn to express and store breastmilk before they return and by flexible breaks at work to allow this practice to continue. In the US, this can include on-site lactation advice (Bar-Yam, 1998). In the UK, the Health and Safety Executive has issued guidance to employees and employers. Published work in this area is confined to descriptions of specific programmes, of which some of these with breastfeeding outcomes are described here. Cohen (1994) describes two corporate lactation programmes in the US aimed at women with pregnancy, perinatal and return to work maintenance phases. Breastfeeding data presented are selective and subject to many sources of bias. While 75% of participants who returned to work while breastfeeding continued to do so to at least six months with an average duration of 8.1 months, uptake of the programme was only 24-29% across all eligible mothers, meaning that the programme was highly self-selecting. Cohen et al. (2002) reports on a corporate lactation programme that focused on promoting breastfeeding through male employees of a utility company in the US. The rationale was that many studies show that mothers attribute their infant feeding decisions to, among other influences, the attitudes and actions of the father (Sharma and Petosa, 1997) and that fathers are often ill informed. It is possible that unspoken beliefs and concerns underpin actions that are not supportive. These concerns may include feeling left out of decisions, frustrated sexual urges and insecurity about their new role as fathers (Jordan and Wall, 1990; Jordan and Wall 1993; Rodriguez-Garcia and Frazier, 1995). The programme was offered to 331 eligible men, of whom 128 took part with their partners. They were offered two variants of the same programme: the fathering programme consisted of two 45-minute group classes about breastfeeding, designed to encourage men to participate in childcare and assist breastfeeding by observing positioning and attachment, or one hours individual lactation education including care for and use of a loaned piston pump, milk storage and a book and handout on breastfeeding. In addition, breast pumps

were loaned for home and work use, with on-site advice from a paediatric obstetric nurse consultant and nutritionist/lactation educator accessible 10 hours a day, seven days a week. The outcome for included parents was a mean duration of breastfeeding of eight months. At one month 96% were still breastfeeding and 69% at six months, compared with a national average of 21.7% in 1996.

1.4.3.2 Future research promoting and supporting breastfeeding at work


Any programme for workplace support for breastfeeding will be influenced by the context of national legislation for parental leave and health and safety, as well as corporate policies and practices affecting work-life balance. Interventions could include worksite education and skilled lactation support, awareness raising among managers, personnel and occupational health staff, the provision of breastfeeding and work education for school leavers, and through well man and well women health education programmes. Studies of corporate organisational change have the same challenges as healthcare organisation studies. A study such as the PROBIT trial (Kramer et al., 2001b) used cluster randomisation on a huge scale, a very expensive and high risk design. More feasible are controlled studies with designs that compare variations in one or more feature of a programme on the same worksite and employer, although generalisation will be limited and issues of contamination between conditions need to be minimised. It is likely that effective workplace programmes to encourage breastfeeding initiation and duration requires elements of best practice reviewed in Sections 1.1 and 1.2, such as effective support and education. In addition, workplace factors such as on-site nurseries allowing access to babies to breastfeed, rooms and flexible breaks to allow breastmilk expression and storage, and loan of breast pumps, could all be subject to controlled studies. Researchers would have to ensure that the usual constraints such as systematic sampling and selection, controlled allocation to conditions and blinded assessment with accepted measures of breastfeeding outcomes are adhered to. Following the study by Cohen et al. (2002) of a workplace intervention for fathers, direct comparison of intervention in a workforce with either mothers or fathers, or both (if both are not employed at the workplace, partners could be invited to join as a couple) would be a challenging study, but would 55

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

test the importance of influencing both parents and the workplace conditions. To design effective workplace policies to support breastfeeding among women working away from their baby, research can be conducted in studies within one site to begin to address areas where the assumptions of current programmes are not built on sound evidence. For example, it is not known whether daytime feeds with expressed milk maintains milk supply and effective feeding patterns in infants. Do babies fed expressed milk by bottle daily either feed more often and longer from the breast after the working day/night shift? What work patterns will support effective breastmilk expression (frequency of breaks, comfort and privacy of rooms used)?

embedded in community development and parent support services of local Sure Start projects. None of these met the criteria for inclusion in this review. One, however, has elements of community intervention, and is described here. Battersby (2002) presents a community project which combines the work of two Sure Start projects in Sheffield. Simply the Breast is an evaluation of a peer support programme that combined both paid and voluntary peer supporters working together under the supervision of a Sure Start midwife. Eight peer supporters were trained by two local trainers who had attended the five-day trainingof-trainers programme of La Leche League. Their activities included visiting antenatal clinics, home visits, setting up and running six support groups, phone contact with mothers and input to a schools awareness programme. The neighbouring Sure Start scheme, Worldly Wise, was concerned with a peer support programme. It was run by two paid peer supporters and trained local health professionals. Evaluation data are presented in Battersby (2002) for the combined projects, known as Breastfeeding is Best Supporters (BIBS), which eventually trained a total of 23 peer supporters. The results show improved rates of initiation and duration using a two samples design, with historical audit data at baseline. Community initiatives require coordinated intervention across several agencies with clear means of addressing tasks shared across sectors, such as: Ensuring local educational authorities prioritise infant nutrition and parenting in the curriculum Sure Start schemes that employ or can access professional support for their services on infant feeding Requiring staff in primary care to undertake relevant training in breastfeeding support Hospital services that encourage integration of their health promotion activities with those that are effective in reaching disadvantaged mothers. In England, the Dykes (2003) review of the 79 Department of Health projects found that 26 were primarily peer support, 11 were drop-in services, and eight focused on school education. The report makes recommendations based on the practical lessons learnt in establishing these interventions. Lindenberger and Bryant (2000) describe the social marketing approach to the development of breastfeeding

1.4.3.3 Community and school education interventions


Interventions to influence the wider determinants of breastfeeding such as local cultural norms may include diverse community interventions, including direct support to pregnant and breastfeeding mothers, and health education to mothers and young people. These interventions may seek to simultaneously influence initiation and duration. Few studies include interventions across more than one sector, an exception being a study by Barber-Madden (1990) in the US. The intervention was focused on municipal hospitals in New York. It included staff training, a resource kit to help staff promote breastfeeding in hospital and establishing a breastfeeding coordinator post in each hospital. The community element consisted of guidelines for worksite health and nutrition programmes, and a media campaign for two months. Breastfeeding data were confined to percentages of women breastfeeding at discharge between 1979 and 1988, with the greatest increases (threefold) in municipal hospitals, and smaller increases in private and voluntary hospitals. The design weaknesses are such that there can be no confidence that the intervention was the agent of change or that rates reflect the breastfeeding rates of the whole community. But it serves to illustrate the breadth of interagency work that may be practical to deliver, even if it is a greater challenge to evaluate. Dykes (2003) reviewed the 79 development and research projects funded by the Department of Health Infant Feeding Initiative in England, some of which were 56

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

support within the Women, Infants and Children (WIC) programmes in the US. While no research evidence is presented in this paper, it illustrates how a conceptual framework can be used to integrate interventions across diverse sectors. It takes a programmed approach to development of interventions with a core theme, itself built on consumer research of what messages are important to the client group. One example is the finding that while many WIC participants are attracted to the benefits of breastfeeding many are also deterred by the practical barriers, including personal inconvenience, conflict of values within the family, anticipated pain of breastfeeding and anxiety about milk insufficiency. The programme sought ways to address these and to position breastfeeding as having unique personal emotional benefits that distanced it from the rival product of bottlefeeding. The intervention was combined with the UNICEF Baby Friendly Initiative (see Section 2.1) to improve the supportiveness of hospital services and media campaigns, legislation to support and workplace policies all featured in this multi-faceted community intervention.

unaware of specific benefits. This point is supported by two other studies: Shaw et al. (2003) in a mixed-race sample in Coventry, and an intervention designed to address South Asian grandmothers role in supporting breastfeeding in Bristol (Ingram and Johnson, 2003). The Scottish study found barriers to breastfeeding were both rational and emotional, and many related to embarrassment about feeding in public. Motivations for breastfeeding by mothers and family members were also rational and emotional. Partners and grandmothers cited concern about exclusion of their caring role if the baby was breastfed. Not all of these concerns could be addressed in one advertisement, but designing it so that the key messages were not obscured by negative reactions to parts of the message was felt to be an important reason for conducting small-scale field tests, using discussion groups around draft scripts. It is perhaps worth noting that the target audience research did not address issues of religious or ethnic sub-group beliefs, which may need to be addressed to access those groups least likely to breastfeed, where some culturally determined beliefs may be specific barriers or drivers of breastfeeding (Thomas and Avery, 1997).

1.4.3.4 Mass media campaigns


It is outside the scope of this review to examine the psychosocial impact of mass media campaigns, since no reported studies have included breastfeeding outcomes. However, the use of psychological research in the design of persuasive communications has a long tradition. There is a large volume of research on attitudes and beliefs about breastfeeding that can be drawn on in designing interventions. For example, the national television campaign in support of Breastfeeding Awareness Week in Scotland used both published research and bespoke qualitative research with samples of the target audience to test the concepts and format of an advertisement (Health Education Board for Scotland, 2001). As with the social marketing approach above, it was found to be important to the acceptability of the advertisement with target audiences to determine their response to specific messages. The target audience was intended to include those mothers who were pregnant for the first time, who may have decided not to breastfeed, as well as those who were interested in breastfeeding, or who had successfully breastfed, along with family members. Responses were not uniform across these groups, so the messages needed to be able to encompass different perspectives. An example is that all knew Breast is Best, but many sub-groups were

1.4.3.5 Interventions to support mothers to breastfeed in public


Research on schoolchildren, and various adult samples referred to above, shows that attitudes to breastfeeding vary considerably within national cultures and subcultures. Research with mothers in the UK suggests many women anticipate adverse reactions and some report having experienced unpleasant behaviour by members of the public and by those in charge of the use of public places. Mothers cite embarrassment as being a factor in deciding not to breastfeed (Hamlyn et al., 2002). Newspaper reports have highlighted the hostility shown to mothers who breastfeed in public (eg Martin and OHare, 1999). This evidence has stimulated a Bill in Scotland to make it a legal right for women to breastfeed in public and to make it illegal to request that a breastfeeding mother and baby cease or are asked to leave public places or licensed premises (where children are otherwise permitted). It is seen as an important part of a concerted effort across government to promote breastfeeding as the norm (www.show.scot.nhs.uk).

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

57

1.4.3.6 Future research community and school interventions


There are no included studies that directly address the impact of interventions aimed at community groups such as schools or the public via the media, although some peer support projects in Sections 1.1 and 1.3 have elements of community intervention. There is a need for medium to large longitudinal studies that allow assessment of community initiatives, including media campaigns, on attitudes to breastfeeding among all age groups as well as breastfeeding outcomes among those likely to deliver in the period of evaluation. Given the evidence that schoolchildren of 11 years old already have ambivalent attitudes towards breastfeeding (Gregg, 1989) it is important to investigate ways of promoting positive attitudes in young children. One important issue in designing community based interventions is the potential for raising the rates among higher socio-economic groups who may respond more easily to some health promotion messages, while having no impact among lower socio-economic groups thereby having the paradoxical effect of increasing the disparity in breastfeeding rates and increasing inequalities in health (as seen in Kirk, 1980). Such interventions should therefore be based on existing literature, which is primarily descriptive, but which raises important issues and hypotheses, and on a clear theoretical rationale, which will allow them to be replicated if necessary. Clear description of socio-demographic variables is needed to ensure analyses can be conducted to establish whether sub-groups respond differentially to interventions. Social, and particularly professional, community and familial influences are important in determining breastfeeding intentions and behaviour. These influences are potent at different periods throughout the course of breastfeeding. This is shown in two studies currently in preparation for publication; a prospective study of new mothers from birth to six weeks in Scotland (V. Swanson, University of Stirling, personal communication), and a prospective study of breastfeeding mothers in England over the period from six to 17 weeks when many mothers cease breastfeeding (O. Dunn, University of Coventry, personal communication). This has important implications for the tailoring of health promotion and support services to ensure the most potent influences are working when mothers are most receptive to them. 58
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1.5 Public health interventions: the organisation of healthcare provision


Topic The organisation of healthcare provision intended to promote and support breastfeeding duration, including rooming-in, breastfeeding only wards, etc. Interventions with a wider focus than breastfeeding but which may include an effect on breastfeeding outcomes (eg continuity of care) were excluded Summary of methods
Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional) Dates for searches: 1990 to June 2003 Population group studied: all postpartum women included in such studies Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including health service and economic outcomes

1.5.1 Background
It is self-evident that the organisation of health services to pregnant and postpartum mothers will have direct and indirect effects on initiation and duration of breastfeeding. The research on infant feeding policies in maternity wards (Perez-Escamilla et al., 1994) suggests that many factors in maternity hospital services will impact on breastfeeding. The research underpinning the Baby Friendly Initiative Ten Steps (Saadeh and Akre, 1996; World Health Organization, 1998a; see Appendix 7), covering the wider healthcare system, suggests many ways in which the organisation of services and provider behaviour influence breastfeeding success. For example, the considerable research on interventions which affect the timing of the first feed (see Section 3 in this review, and Renfrew et al., 2000) suggests that services should be organised to ensure mothers can put the baby to the breast soon after birth, normally within an hour if possible. This has implications for staffing in delivery suites and for staff training. The research on the adverse effects of practices such as test weighing babies before and after feeds and breastmilk supplementation has implications for staff training to encourage them to abandon such routines. The need to achieve pain-free, effective feeds and reduce the risk of early

malnourishment and dehydration will have an impact on staffing levels as staff support mothers in learning the skills of positioning and attachment. Unrestricted contact and feeding implies abandoning the use of ward nurseries, again impacting on staffing requirements and skill mix. Early discharge home has an impact on the type of service required for primary care, both ante- and postnatal. Early discharge home may be supportive of mothers who may escape from some aspects of hospital care that are detrimental to breastfeeding, or who may be able to cope better at home than in hospital, as suggested by Margolis and Schwartz (2000) from their analysis of early discharge in hospitals in the US in the 1980s. In other circumstances it may mean the mother is sent to an unsupportive home with no professional and no informal sources of advice. Early discharge has become part of routine care in the UK, as in the US and other industrialised nations, partly as a cost containment measure. There is conflicting evidence that early discharge has unwanted effects on infants and their mothers. Some studies show that infants who are discharged early are at risk from re-admission for rare but sometimes dangerous consequences of dehydration, and breastfed infants are at risk from malnutrition in the first two weeks (eg Rowland and Zori, 1982; Kotagal et al., 59

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

1999) and mothers and carers are at risk from anxiety and breastfeeding failure. Current postpartum care in the UK is described in Appendix 6. It is clear from this description that the organisation of post-discharge care, as for postnatal ward care, has elements that when varied or combined simultaneously may have conflicting impacts on breastfeeding. Designing studies capable of delineating the elements of the organisation of services and provider behaviour that are both potent and amenable to change is challenging.

postnatal ward in Oxford, England. Both arms of the trial were not offered additional feeds. The study number is small (n=40) and although no statistics are reported the effects would not be statistically significant. The study has serious design limitations. Some contamination or merging between the two conditions cannot be ruled out because of design weaknesses. Also, there was no auditing of the compliance with the protocol of care. This means control care is difficult to characterise and replicate. There was no blinding of randomisation nor of assessment. Also, the samples allocated to each arm were dissimilar, hence differences cannot be clearly attributed to the intervention. Waldenstrom and Nilsson (1994) [35]: this Swedish study of 1,230 mothers reported on the comparison of a birthing centre to standard obstetric care on breastfeeding and found no impact on duration, although in mothers allocated to the birth centre there were slightly more problems of sore nipples and milk stasis, possibly due to the birthing centre having less experienced staff. Although randomisation was blind it is unclear whether assessment was blind this would have been be difficult to achieve as mothers would have been aware of the care received, which may have introduced bias. The high rates of exclusive feeding in both groups are notable (mean eight and half months), indicating a well-motivated and supported sample. Watters and Sparrow (1990) [36]: this Canadian study of 412 mothers used a before-after design to examine combined care shared between the postnatal nurse and mother, and routine care. Neither condition is described sufficiently for replication. Control care included more mothers per staff member, confounding both staffing and style of care; the mother is included as a partner in care with midwifery staff in the intervention group. The study found no impact on breastfeeding duration. The study is of low quality so the results are not particularly informative. Watters and Kristiansen (1995) [36]: this Canadian paper reports two before-after design studies of rooming-in. The first study recruited 408 women, the second 412. The second study appears to be that reported in the earlier paper (Watters and Sparrow, 1990 [36]). The intervention in study one consisted of rooming-in for up to 15 hours a day and continuity of care with more staff to mothers in the experimental group than controls. It resulted in no better breastfeeding outcomes than the

1.5.2 Evidence base


Eight studies met the criteria for inclusion in this section. In conducting the search, we were mindful of studies where the focus of intervention is postnatal care but the intervention is wider than, but includes, support for breastfeeding. These were excluded from the review. Also excluded from this review are studies where the focus is solely on length of stay, with no specific aspect of care designed to influence breastfeeding. There is a Cochrane review (Brown et al., 2002) and three further studies where length of stay (LOS) is varied as the sole or main part of the intervention. These were also excluded from the review. Table 1.5 shows there are eight studies that relate to the organisation of hospital and post-hospital services where the main focus of the intervention is on breastfeeding. Of these, four relate to postnatal ward design and four to post-discharge care designed to support breastfeeding.

1.5.3 Results 1.5.3.1 Postnatal ward organisation


These four studies included any changes in the organisation of post-delivery care, such as rooming-in and breastfeeding-only wards or bays, intended to support breastfeeding and where the care did not include changes in post-discharge care. Berry (1994) [30]: this study examined the impact of a sole breastfeeding room (SBG) versus allocation to rooms with breastfeeding and non-breastfeeding mothers in one 60

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

previous type of care. In study two the new form of care was implemented for an eight hour nursing shift and the control care is not described. The studies are of low quality, so again, the results are not particularly informative. Discussion and conclusions: postnatal ward organisation There are no high quality studies of rooming-in, shared breastfeeding rooms and mother-infant combined care. No study showed statistically significant impact on duration. There is insufficient evidence on which to base decisions regarding the types of care described. However, the absence of evidence should not imply evidence of absence of effect. Practices that encourage behaviours in the mother known to support breastfeeding, ie those that encourage unrestricted contact such as rooming-in should be encouraged. Waldenstrom and Nilssons (1994) [35] study of birthing centre versus obstetric unit care showed no differences in breastfeeding outcomes. Design weaknesses were noted above. The trend to less medicalised care offers opportunities to improve breastfeeding provided care is at least of equal quality. Aspects of care that may be important in supporting breastfeeding are the absence of intrusive assessments, such as weighing or taking blood samples from the baby, and a focus by midwives on observing and guiding mothers and engaging the wider family in caring for the baby. Further research: postnatal ward organisation There are many aspects of postnatal ward care organisation not evaluated in these studies, such as bedsharing, which should be susceptible to controlled evaluation. The aspects of care described in these studies are not sufficiently proven to be described as promising for future research. However, new variants of accepted care, such as medically indicated supplementation and breastfeeding-only wards, need to be examined in studies that vary only the relevant characteristics of care: Research on rooming-in is now unnecessary and unethical since its benefits are well known and there is no evidence to support the separation of mothers and babies at any stage. Variants of the nature of care, such as bed-sharing in hospital, or shared care involving the partner sharing the room and baby care, would now be more appropriate

For breastfeeding-only wards, the comparison should be between those who are sharing with breastfeeding mothers and those allocated to wards with both artificial and breastfeeding care. To avoid problems of convergence of care, as some mothers in the breastfeeding-only ward or bay may move to supplementation or artificial feeding, such a study needs to be conducted where there is a virtual absence of supplementation in routine care A controlled study of birthing centre care versus obstetric lead care would require careful documentation of the differences in staffing and care routines (where rooming-in and restricted supplementation would be only a small part of many differences), as well as comparable levels of competence in breastfeeding support skills Staffing levels and skill mix may in themselves be important issues in the context of breastfeeding, regardless of the setting in which staff work.

1.5.3.2 Post-discharge care studies where the focus of intervention includes support for breastfeeding
These four studies examined the organisation of postdischarge care, where length of stay was not also systematically varied as part of the intervention, and where there was a focus on protecting breastfeeding. Emond et al. (2002) [31] presented a study with prospective data on 459 mothers served by seven health visiting nurse bases in a deprived urban area in the Bristol area of England. The aim was to evaluate whether a model of intensive home visits by health visitors, compared to an unspecified standard model, had differential impacts on a number of health and service access outcomes, including breastfeeding status at six weeks. Although initial analyses (not reported) showed a significant benefit (61% breastfeeding at six weeks in the intervention group versus 39% breastfeeding in the control group), this effect was no longer significant when adjusted for confounders (these are not reported) and clustering effects (area of service). As the authors point out, the small number of service area bases being compared means that clustering effects were hiding true differences between the two models of health visiting. Gagnon et al. (2002) [32] compared community nurse home visiting with a hospital nurse clinic visit in a sample of 586 women in Canada discharged less than 36 hours after childbirth and breastfeeding at discharge. This study 61

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

found that there were no statistically significant differences in daily breastfeeding frequency or infant weight gain between the groups. The incidence of insufficient breastfeeding and insufficient weight gain were low and similar in both groups. Gunn et al. (1998) [33]: this Australian study used an RCT design to examine whether an additional GP appointment at one week after discharge was superior to no appointment, with both arms offered follow-up at six weeks with the GP. The study recruited 683 mothers. There were no differences in duration of breastfeeding, nor in prevalence of breastfeeding fully or partially, at three and six months. There were no differences in the number of visits to the GP in the first six weeks, and it is possible that both arms were alerted to the availability of GP services in the early post-hospital period, minimising differences between groups. Steel OConnor et al. (2003) [34]: this Canadian study was designed to examine two forms of care after early discharge (48 hours postpartum) for vaginally delivered primiparous women. The study compared care by a public health nurse through up to two routine home visits for all mothers, or phone contact to screen for the offer of a home visit. A total of 733 mothers were recruited to the RCT. The results showed no difference in breastfeeding duration. However, while analyses were by intention-totreat, there was some convergence across the two arms in delivery. This would reduce the chances of finding a statistically significant effect. Discussion and conclusions post-discharge care studies The four studies showed no significant effects for the types of care described. There is insufficient evidence from these studies to support changes in service in favour of the intervention arm care nor to rule out there being some advantages in the control arm care. The four studies cover service patterns that are not directly comparable across time and countries. Two studies (Steel OConnor et al., 2003 [34]; Gagnon et al., 2002 [32]) are both studies of post-discharge care services designed to protect breastfeeding after short stay (under 48 hours postpartum and under 36 hours, respectively). Breastfeeding is not fully established by the third day postpartum, hence care in this period may influence breastfeeding more than at other times. We advise that maternity services that provide short-stay care 62

should be combined with post-hospital care provided by skilled practitioners who can support breastfeeding mothers, and ways of bridging the gap between hospital and community services should be examined. It is also likely that care may need to be provided over a more sustained period, since many mothers experience problems of soreness or apparent milk insufficiency over the first two or more postpartum weeks. The study by Gunn et al. (1998) [33] evaluated the effect of an earlier (at one rather than six weeks) appointment with a GP. It is possible that the care offered was not the most likely to assist the mother with breastfeeding, since an office visit is not conducive to observation of feeding. This may play a part in determining the advice given and it is not evident that the GP had specific training in breastfeeding. Of interest is the study by Emond et al. (2002) [31] since it was designed to meet the apparent needs of women in a deprived urban area by intensive home visiting. In contrast with the two other studies, the care provided in both conditions was probably more comprehensive and designed to support the mother, baby and family in many ways in addition to support for breastfeeding. Care that supports a mother to achieve a settled baby may indirectly impact on the ease of breastfeeding. Absence of effect on breastfeeding, once confounders are controlled for, does not mean there was no benefit. It is likely that any effects achieved were masked by care conditions that were quite similar. In the UK context mothers routinely receive home visits by midwives up to 10 days postpartum and health visitors make home visits and/or care for mothers at clinics over the subsequent months. It is usual for services to prioritise vulnerable families, but that can mean that health visitors target mothers with mental health or social problems, not routinely providing intensive expert support to all breastfeeding mothers. The implications for practice are that services may need to be more targeted at breastfeeding mothers. Further research: post-discharge care studies The care provided in the four studies was probably too similar in each study arm to achieve detectable impacts on breastfeeding duration. The Gunn et al. (1998) [33] study of a single visit to a GP one week after discharge in the context of care in Australia in which maternal and child health nurses visit mothers at home within a few days of discharge from hospital shows that shifting the routine GP appointment

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

to one week from six weeks had no influence on breastfeeding nor the number of GP visits. The overall level of care provided by the GP and other services also did not differ significantly so it is unsurprising there was minimal impact by shifting the time of appointment. Future studies could more usefully study the content of care offered at this time, which is known to be a critical period in the establishment of breastfeeding. Ways of bridging the gap between hospital and community services should be examined. The Steel OConnor et al. (2003) [34] study of telephone screening to prioritise home visits, compared to ad hoc availability of up to two visits in 14 days after discharge in Canada, where no other healthcare is offered, showed no differences in breastfeeding outcomes. There was considerable convergence of care, with 30% in the telephone screen condition having one home visit, and a third of the control home visit arm refused home visits. Future research should account for mothers choice in how services are provided. An improved design would be to allocate all mothers to telephone screening and compare provision of mother-led home visits to standard schedule visits, or to vary the content of care provided to screened women. The study by Emond et al. (2002) [31] showed some impacts on duration but when the confounders were controlled for these effects were not significant. It would be useful, therefore, especially in the UK context of comprehensive community care by midwives and health visitors, to establish the most effective form of postdischarge care to support breastfeeding. Further studies that seek the views of mothers about the style of service most suited to their needs will help to establish the most effective form of post-discharge care to support breastfeeding in that socio-cultural and healthcare context. The design of such studies must take into account the discharge policies of local maternity units, since studies reported elsewhere (eg Ingram et al., 2002 [76], Section 4.1) show that staffing policies and crises of these services impact on the criteria for discharge. This may also hold for staffing, caseload and skill mix changes in community health services. It is advised that designs control for this through case control or randomisation at the level of the individual rather than time-series or other cohort designs where variation in the service provided may be significantly varied over time.

The types of care described are very specific to the context of usual care in that country. Future studies of the organisation of post-discharge care should isolate those factors likely to have a major impact on breastfeeding duration. These are likely to be care that supports mothers in effective feeding, unrestricted and without supplementation, and which address any problems for the mother or baby that make it difficult to continue. Important variables will be the timing and content of care delivered and the extent this is duplicating or complementing care provided by other professional and lay support services, or informal carers. Studies reviewed have mainly been on low-risk women and babies, and with minimal description of sociodemographic factors known to influence breastfeeding. Future studies should characterise the samples in sufficient detail to enable sub-group analyses by these variables to examine whether there are differential support needs to vulnerable groups, so ensuring more cost-effective targeting of services. Research on organisational interventions, whether postnatal or post-discharge, present the same challenges. Studies of organisational interventions cannot be replicated nor can the essential elements that might impact on breastfeeding be determined unless the care provided is laid out in a protocol. Staff must be trained in adherence to the protocol. This must include ensuring that, in all the conditions of care that are examined, staff have an agreed level of breastfeeding support skills. The care received should be evaluated through audit of protocol compliance by staff and maternal experience of care.

1.5.4 General implications for research


Common problems in methods of existing studies should be addressed: Many of the studies reviewed lack adequate description of the intervention and especially control care Definitional issues concerning feeding status make comparison across studies difficult Many organisational changes concern interrelated practices: early discharge care has implications for postnatal services, yet many studies do not record whether these services were manipulated, or indeed accessed, which may have an over-riding effect on interventions such as short length of stay 63

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Organisational interventions can seldom be delivered in a standard manner since staff may vary in skills and users may vary in motivation to use services offered. This can lead to convergence of interventions. Research should mitigate this through staff training, protocols, monitoring staff and user compliance. The evaluation of organisational interventions is challenging, and researchers should consider new ways of designing such research. For example: The studies of postnatal care are small and with many design flaws, hence there is a need for robust designs for all the innovations likely to impact on breastfeeding including rooming-in, breastfeeding-only rooms or wards, partnership care, birthing centres and skin-toskin contact. Any nationally determined innovations, perhaps driven by IT or telemedicine, could fruitfully be conducted in prospective large-cohort studies Studies of large-scale organisational and policy changes that could impact on breastfeeding, along with other outcomes, should be designed to include measurement of breastfeeding outcomes over a realistic and clinically relevant period, since systemic healthcare changes may have inadvertent negative or positive impacts on breastfeeding Studies of post-hospital care reviewed have been driven mainly by legislation or cost control. Managed care is likely to be a feature of UK healthcare in future. It is suggested that large-scale multi-centre projects using electronic records are established to determine the impact of new protocols of care. To establish detailed understanding of the impact on parents and staff, these can be combined with smaller scale more intensive observational studies. Research into organisational interventions to support breastfeeding should be conducted within the context of the extensive research already conducted into effective organisational change. As indicated above, it is self-evident that the organisational context will influence the support mothers receive for breastfeeding in hospital. There is a large body of health service research into effective change management (Wallace et al., 2001; Grimshaw et al., 2001). This suggests that interventions, particularly those mediated by staff behaviour (Section 4), must be planned as part of whole systems change and resourced to address key organisational culture and personal 64

competence barriers. Research into the development of effective practice, eg through targeted guidelines using psychological models, shows promise (Michie and Johnston, 2004) and models that address specific beliefs and perceived competence can be powerful (Marteau et al., 2002). It is difficult to conduct experimental research on policy development (Janovsky and Cassels, 1996) so unsurprisingly there is scant research on the development and dissemination of local breastfeeding policies (Baby Friendly Initiative Step 1), although some are included in Section 4 where the main intervention is staff training. Research from the field of risk management would suggest that factors such as organisational culture and professional sub-cultures influence judgements of nursing, midwifery and medical staff about using best practice guidelines (Lawton and Parker, 1999). It is recommended that research designs, including experimental designs as used by Lawton and Parker (1999) to test key determinants of intentions to follow guidelines, are used in studies to examine how to introduce staff-mediated organisational change in services to support breastfeeding mothers.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Table 1.5: Included studies organisation of healthcare provision Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Postnatal placement in breastfeeding only rooms Postnatal in hospital Day 3, day 7, 4-6 weeks Timing of breastfeeding duration outcomes

Name of authors and study number Convenience sample of 40 women with uncomplicated deliveries, well babies, breastfed within 2 hrs 459 primiparous women with New health visitor scheme one baby in geographic areas of Bristol. No participant background data reported described as deprived, urban community In third trimester and post hospital over 8 months, up to 2 years for 20% 586 healthy mothers and their healthy infants (women were well-educated and ethnically heterogeneous) Additional appointment with 683 well mothers and babies at discharge, older, multi- and GP at 1 week after discharge primiparous, varied social class vs none 733 primiparous women from 2 hospitals: age, education, gestational age infant controlled in regression Postpartum care provided at home vs in clinic after short length of stay (36 hours after birth) Telephone contact 48 hours after delivery + home visit at 3-4 days postpartum No initiation data Bf at 6 weeks after birth

Date of Country publication

Design

Berry [30]

1994

England

RCT

Emond et al. [31]

2002

UK

Before-after

Gagnon et al. [32]

2002

Canada

RCT

14 days post discharge

Gunn et al. [33]

1998

Australia

RCT

1 week after discharge

3 and 6 months Initiation recorded at recruitment (day 2-3) Bf data on those who were bf at discharge, at 2, 4 and 6 weeks

Steel OConnor 2003 et al. [34]

Canada

RCT

Home visits vs phone screen Post-hospital and selective home visits after early discharge

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Waldenstrom and Nilsson [35]

1994

Sweden

RCT

1230 mothers, data given for Birth centre or obstetric unit age, gestation, parity, marital status, education, intention to bf expressed as group comparability data 412 mothers Combined care with nurse and mother

Antenatal, postpartum

2 months, no initiation data

Watters and Sparrow [36]

1990

Canada

Before-after study

Postnatal in hospital

Initiation in hospital recorded, and at 6 weeks Mother infant (rooming-in) Postnatal in hospital versus separate postnatal care Initiation, duration bf at 1 and 6 weeks. No data reported by group

Watters and Kristiansen [36]

1995

Canada

Before-after studies

408 and 412 mothers

65

2.1 Public policy interventions


Topic Provision of employment leave, implementation of the WHO code on marketing of breastmilk substitutes, workforce policies to support breastfeeding mothers, promotion of formula milk and restrictions on provision of formula by prescription

Summary of methods
Study design of included studies: RCTs, non-RCTs with concurrent controls, and before-after studies (cohort or cross-sectional). National case-studies are included Dates for searches: 1966 to June 2003 Population group studied: countries experiencing relevant policy change Countries included: developed countries. Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including economic outcomes

2.1.1 Background
Public policy can be expected to influence breastfeeding duration both directly and indirectly. Indirect effects may result from health and safety legislation encouraging more generous maternity leave provision. This in turn may impact on initiation and duration of breastfeeding in working mothers through enabling parents to have sufficient time at home with their baby to establish breastfeeding and to maintain it if the mother returns to work outside the home. Interventions reviewed in Section 1.4, ie workplace interventions to support breastfeeding through fathers, have been developed on the rationale that fathers caring role can support mothers to breastfeed. However, in Section 1.4 we found there are no studies published with breastfeeding outcomes. Practices that reduce parental leave directly, or paid benefits, could be expected to influence mothers decisions to return to work, meaning that feeding intentions may become secondary to economic necessity. At a policy level, we could expect that changes in employment law and public policy could influence breastfeeding rates in either direction. It is noteworthy that maternity leave is much longer in Norway and Sweden, countries with far higher rates of 66

breastfeeding initiation and duration. Policies by employers will affect mothers intention and success with maintaining breastfeeding while working, through for example paid breaks and facilities to express and store breastmilk. The International Labour Organization has passed a convention supporting the right of women who work outside the home to take paid breastfeeding breaks (International Labour Organization, 2000). In the UK, the Health and Safety Executive has provided advice leaflets for employers and employees on breastfeeding. A report from the Scottish Breastfeeding Group, Breastfeeding and returning to work, includes a code of practice for employers, along with a summary of current legislation (www.scotland.gov.uk/library2/doc15/bfrw-01.asp). Evaluation of the impact of these policy initiatives with indirect relevance to breastfeeding is not reported in published research. Direct impact can be expected from welfare and nutrition policies. In the UK, the Welfare Food Scheme provides nutritional supplements to childbearing women and children under five in receipt of Working Families Tax Credit, including milk tokens for breastfeeding mothers and a supply of formula for bottlefeeding mothers. This provides an indirect economic incentive to choose artificial feeding and arguably sends inappropriate messages about the equivalent healthiness of artificial

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

feeding with breastfeeding. There are no published evaluation studies of this scheme meeting the criteria of this review. The scheme is being replaced in 2005 by the Healthy Start scheme, which does attempt to redress the imbalance between breast and bottlefeeding mothers. Controlled evaluation of this new scheme would allow lessons to be learned about its impact. A further direct effect of government policy can be expected through endorsement of international controls on marketing of formula feeds. The practice of advertising and marketing formula milk has met with considerable controversy around the world because it is assumed to discourage breastfeeding, to interfere with the demand and supply cycle of breastmilk production, to increase the risk of early weaning and to undermine womens confidence in themselves and their ability to breastfeed their own infants. In countries where potable (drinkable) water is scarce the practice of feeding infants with formula milk rather than breastmilk results in high rates of infant mortality. Formula milk companies advertise their products by providing free samples of not only formula milk but also feeding bottles, teats and pacifiers, all of which have been implicated in the failure of initiation and establishment of breastfeeding and shortening of the duration of breastfeeding. Condemnation of the intense activity of formula milk manufacturing companies in many countries led to the development of the WHO International Code of Marketing of Breast Milk Substitutes (World Health Organization, 1981b) to curtail over-active marketing by formula milk companies with the explicit intent of reversing the decline in breastfeeding rates. The code and subsequent resolutions set down a minimum requirement in law. It prohibits the promotion and distribution of breastmilk substitute feeds to pregnant or nursing mothers and is a central element of the Baby Friendly Initiative (BFI) Ten Steps to Successful Breastfeeding (World Health Organization, 1998a). It has been UK policy to endorse aspects of the code since 1981, supported by Department of Health guidance (Department of Health and Social Security, 1983; Department of Health, 1989). The use of discharge packs containing formula-related products has as a consequence been discontinued in the UK, although it continues in other countries, and restrictions on direct advertising of formula milk to the public have been imposed. The promotion of formula

milk to health professionals and of related products such as follow-on milks to the public continue to be the subject of criticism (IBFAN, 2004). An indirect effect of government endorsement of the code could arise through adoption of BFI standards in some or all services within the jurisdiction of health authorities. Studies of the impact of BFI accreditation have included an RCT in Brazil (Westphal et al., 1995) and a before/after study in Thailand (Buranasin, 1991), both outside the criteria of this review. One paper provides a descriptive account of the WHO code in European countries (Helsing, 1990), including the wider health promotion activities, maternity hospital routines and staff education and training issues encompassed by the code. The use of national surveys and nationally collected health service data enables the evaluation by observational design of changes in the organisation and delivery of services and other policy initiatives. Two studies were identified that gave information about national or regional policy initiatives and breastfeeding data, and which add useful context to the use of national data and the evaluation of national policy, and are briefly described below. But because they did not report comparative figures allowing for assessment of changes over time, these studies were excluded from this review. The development of interventions in Scandinavia that include many elements that may impact on duration was described in Fairbank et al. (2000). Evaluation of interventions on national breastfeeding rates is confined to a paper by Aasheim et al. (2000). It describes a whole country (Norway) approach, with breastfeeding data in 1999 of 97% breastfeeding to one month, 88% to three months and 80% to six months. Interventions include: encouraging lay led breastfeeding groups such as La Leche League, early mother-infant contact, adopting the WHO code for marketing breastmilk substitutes at a national level, state financial support for BFI, parental leave for up to 52 weeks, a one-off maternity grant for non-working mothers, free pregnancy care and provision of community midwifery care by local councils. The experience of one hospital is described including allowing the mother to stay in hospital, with the option also for the family to stay in a nearby hotel, for up to four days after delivery. A sample of women (n=349, selection unclear) showed 100% breastfeeding at discharge from hospital. Clearly it would be most instructive for other European countries if the interventions described were evaluated. 67

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Other national studies with breastfeeding data were reviewed but excluded on design criteria. Kersting and Dulon (2002) describe the SuSe study in Germany, a nationwide survey of 117 maternity hospitals from a random sample of 360/1,120 possible hospitals, with 1,717 mother and infant pairs. The study shows wide variation between hospitals in reported adherence to BFI standards but does not relate these to the breastfeeding prevalence data. The variation in practices could usefully be subjected to controlled evaluation.

2.1.2 Evidence base


Six studies met the criteria for inclusion in this review. Table 2.1 shows two studies were concerned with commercial marketing of breastmilk substitutes including discharge packs, one study was on prenatal formula advertising and a further three studies were on regional or national area evaluations of health service policy. Two studies evaluated the impact of government policy in setting breastfeeding targets and in applying incentives and penalties to breastfeeding data collection and promotion activities. One study used health service data to evaluate BFI-related activities in one large provincial area. There were no studies of the provision of employment leave, workplace policies that support breastfeeding mothers or restrictions on the provision of formula by prescription. National case study data are available from two countries, and regional information from part of one of these countries.

in the US to receive three different discharge packs containing breast pump only, breast pump and formula, or formula only. All enclosures other than specific product instructions were removed. This study found that the contents of the discharge pack did not have an impact on whether the mother engaged in exclusive or partial breastfeeding during the first 16 weeks. Unexpectedly, higher rates of exclusive breastfeeding were found among women receiving only formula in their packs compared to those receiving either breast pumps only or both pump and formula (p<.01). It must be noted that a small number of women were lost to follow-up, most of these being poorer women with less education. Those remaining in the trial until the end of the study were predominantly white middle-class women, most of whom intended to breastfeed their newborns, if not exclusively, then at least predominantly. Another study (Bliss et al., 1997 [37]) similarly sought to determine if breastfeeding duration was influenced by the contents of discharge packs; 1,625 women in the US were randomised to receive a can of powdered formula, or a manual breast pump, or a breast pump and formula, or nothing. The contents of the discharge pack had no influence on feeding method at six weeks, four months and six months, or on breastfeeding duration. Sub-group analyses showed that among mothers who intended to breastfeed for six months or more a small effect of the discharge pack was seen on feeding method at six weeks (p<.05).The difference in full or partial breastfeeding between those receiving formula and pump was significant (p<.003). There was no difference in the age at which juices, cereal and solids were introduced. A subgroup of mothers intending to breastfeed for at least six months were less likely to be breastfeeding at six weeks if they received formula in the discharge pack, while those receiving a breast pump were more likely to be breastfeeding. The two groups did not differ in overall breastfeeding duration. Among mothers who had not returned to work or school by the sixth week, more were partially breastfeeding at six weeks if they received formula compared to those who received a pump. However the two groups did not differ in duration of overall breastfeeding. Overall the study showed that women receiving formula in the discharge pack were more likely to be supplementing early, but there was no effect on breastfeeding duration, and those receiving a pump were less likely to supplement but again there was no effect on overall duration of breastfeeding.

2.1.3 Findings 2.1.3.1 Commercial marketing of breastmilk substitutes


Findings We provide evidence from three RCTs published between 1995 and 2003, two of which investigate the impact of commercial hospital discharge packs and their contents, and one which examines the effect of prenatal formula advertising on the duration of breastfeeding. Dungy et al. (1997) [40] investigated the contents of commercial discharge packs and their impact on breastfeeding duration. They randomised 763 mothers 68

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Howard et al. (2000) [42] compared the effect on breastfeeding duration of providing formula company produced materials about infant feeding at the first outpatient prenatal visit with breastfeeding promotion materials without formula advertising. A total of 547 pregnant women in the US were randomised to receive formula company designed educational packs, or specially designed packs about infant feeding when they attended for their first prenatal visit. Exposure to formula promotion materials significantly increased breastfeeding cessation in the first two weeks. Among women with uncertain goals about infant feeding and those who intended to breastfeed for 12 weeks or less, exclusive, full and overall breastfeeding duration were shortened. It must be noted that the packs were given to women during their first antenatal visit, months before they delivered, and they may have been exposed to commercial packs from other sources and commercial information and hand-outs later during the pregnancy. Findings from previous reviews Renfrew et al., 2000 synthesised evidence from eight trials of discharge packs published between 1983 and 1997. A Cochrane review (Donnelly et al., 2000) synthesised evidence from nine RCTs conducted between 1983 and 1997 involving 3,730 women. Their meta-analysis found that exclusive breastfeeding was reduced at all points in time as a result of commercial discharge packs. There was, however, no evidence that the use of hospital discharge packs shortens the duration of non-exclusive breastfeeding. Both studies on commercial hospital discharge packs described in our review, involving 2,388 women, are included in Renfrew et al., 2000 and Donnelly et al., 2000. Discussion and conclusions Overall the evidence from this and previous reviews suggests that commercial discharge packs are not conducive to exclusive breastfeeding. The trial of prenatal discharge packs has one major confounder women may have received information or samples of formula from other sources anytime during the pregnancy. Despite this difficulty, findings from this study demonstrate that advertising formula in the prenatal period may have an adverse effect on breastfeeding duration among women who are either unsure of the infant feeding plans or those who intend to breastfeed for 12 weeks or less.

Future research commercial marketing of breastmilk substitutes Further research is needed to identify the impact of the range of marketing practices currently used to promote formula milk and related products to pregnant women, especially to those who are uncertain of their infant feeding plans. Both duration and initiation outcomes should be assessed. It would be unethical to conduct this in countries already adhering to the WHO code, given the evidence presented here that some of these activities can adversely impact on duration.

2.1.3.2 Large area and national evaluation of health service policy


Findings Scotland The paper by Britten and Broadfoot (2002) [38] described a survey of all 33 maternity units in Scotland in March 2000, respondents being midwives responsible for breastfeeding promotion and support. Topics included local breastfeeding strategies and targets, staff training, breastfeeding rates, and steps to implement and audit good practice. Breastfeeding duration data at six to seven days postpartum was taken from a national dataset based on data collected at the same time as the routine Guthrie test (see Glossary). Interventions in place were national policy implemented at health board and maternity unit level. Interventions included Scottish Executive Office support for BFI in 1996, and advice on breastfeeding targets to be set locally, with a national target in 1994 of more than 50% women will still be breastfeeding their babies at six weeks by 2005. The paper reports that local progress was good, with all but four maternity units working towards BFI and six achieving the award. Data were not presented by hospital so no conclusions can be drawn about improvement in breastfeeding rates according to the types of interventions (such as BFI achievement). Breastfeeding data were reported as being collected by 30/33 units, but these were at varying time periods. Data on national breastfeeding rates at six weeks were not reported in this paper, which is required to monitor progress towards attainment of the target in 2005. The paper shows that in Scotland significant improvements have been made in duration up to seven days. The data from 1995 to 1999 show a 2.5% increase 69

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

in duration at six-seven weeks postpartum, which may be at least partially attributable to increased maternal age, and to the existence of national breastfeeding duration targets, national government support for BFI, hospitals working towards BFI and local breastfeeding targets. This must be set against national data showing in 1995 that 55% mothers chose to breastfeed at birth, but by seven days only 46% were breastfeeding and by six weeks 36% were breastfeeding. Campbell and Gibson (1997) reported data on a survey of all 14 Scottish Health Boards in 1995, concerning the target setting process, data collection and activities encouraged by health services to promote breastfeeding. As no breastfeeding duration data were included it is excluded from this review. However, it provides insights into the process of implementation of a centrally determined policy that underpins a national breastfeeding policy. There are findings that suggest elements to be included in future centrally planned processes. Target setting was not linked to health promotion activities in 7/14 health board areas; 4/14 health boards adopted data collection strategies that were not linked to the target. There was little evidence of engagement between health boards and provider services in target setting, data collection or planning activities. Planned activities were put in place in only 7/14 areas, and these were not multiagency. Consultation included health service staff but only one board consulted with members of the public. None consulted other public agencies such as local education authorities or the voluntary sector. Actual targets set used non-standard definitions and varied by time of collection, making comparison across time and across areas problematic. The policy of government support for breastfeeding and local targets was only part of the intervention. A wide range of activities have been undertaken throughout Scotland to support this policy over the past decade, including a Scottish NHS Board Breastfeeding Strategy Group, local Joint Breastfeeding Initiative groups, boardlevel strategies or action plans, multi-disciplinary local breastfeeding peer support projects, public awareness campaigns, return to work policies for breastfeeding mothers in the NHS, working towards BFI accreditation, and the appointment of local Infant Feeding Advisers in 5/15 boards. Data from the first two years from national audits of the reported activity by 15 Scottish boards are published on the NHS Scotland website (www.show.scot.nhs.uk) for 2002 and 2003. However, 70

while breastfeeding data are collected and reported by maternity unit, they are reported as the percentage of mothers giving any breastfeeding or breastfeeding at seven days (collected at the same time as the routine Guthrie test) by maternity unit, without an overall summary percentage by national population or health board. It is not therefore possible from this published data to ascertain links between progress on policy implementation revealed in the audit and breastfeeding outcomes. Findings Italy Two studies report large area or national breastfeeding data in the context of the evaluation of national or regional policies, both in Italy. Cattaneo et al. (2001) [39] described the effect of public health authority policy on breastfeeding duration for all babies in the six local authority areas within one regional health authority. This was Friuli Venezia Giulia, NE Italy, serving 1.187 million people. The intervention was the use of financial penalties on the local health authority (LHA). These penalties would depend on the LHA not achieving targets and activities. The targets were breastfeeding data collection objectives (using agreed WHO definitions and the time periods of discharge from hospital and immunisation appointment of 16-19 weeks postpartum in 1998). In 1999, process objectives were agreed (policies, competence of health professionals, information and education to mothers, service coordination and integration). Not all LHAs agreed objectives in all categories. Only two LHAs agreed breastfeeding outcome targets. The results showed that early duration outcomes (discharge from hospital rates) were not significant for non-breastfeeding, but significant improvements in exclusivity (p<0.001) and full breastfeeding (exclusivity plus predominant breastfeeding) (p<0.001) occurred, while there were reductions in predominant breastfeeding (p<0.001) and complementary breastfeeding (p<0.001). At 16-19 weeks duration of exclusivity (p<0.001) and full breastfeeding (p<0.001) were again significantly improved, with no effects on predominant and no breastfeeding and a significant reduction (p<0.05) in complementary breastfeeding. Of the two LHAs that set breastfeeding outcome targets, one produced a significant reduction in non-

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breastfeeding at discharge. The analyses do not show whether differences in categories of feeding were different in these two LHAs, but they show the largest gains in exclusivity, and one of the two LHAs showed a significant increase in breastfeeding (exclusive and any non-milk substitute liquids), along with one other LHA that had not set targets. The paper does not report if the financial penalties were imposed and if so whether it was because breastfeeding or other public health targets were not achieved. A further year of data would be needed to assess whether the actual imposition of penalties, or their avoidance, impacted on the achievement of process activities and breastfeeding rates. Based on these results it is possible to predict that a more powerful approach would be to ensure breastfeeding duration targets were set by all LHAs. The results of this study can be cautiously interpreted as showing that regionally agreed breastfeeding activities and targets by LHAs have had some impact on improving duration of exclusivity and full breastfeeding (including predominant breastfeeding, with liquids other than milk substitutes), at discharge from hospital and 16-19 weeks, but not on overall rates of any breastfeeding. However, the observational design does not exclude the possibility that the changes were due to changes in other policies, levels of service provision in the health services or other sources of support for mothers. The study by Giovannini et al. (2003) [41] describes a two samples cohort design of mothers and infants sampled to be representative of the population across Italy (n=4,051). The time periods were one month in 1995 and 1999. Participation rates at one month were 67% in 1995 and 70% in 1999. The interventions examined were adherence, reported by the mother at one month postpartum, to steps 3-8 of the 10 BFI steps in 1995 and steps 3-9 in 1999 (see Appendix 7 for description of the 10 steps). Concomitant adherence to the overall 10 steps was reported but whether this means actual achievement of formal accreditation is not clear. Breastfeeding data were collected within 48 hours of delivery, discharge and monthly from one to six months by telephone interview, using WHO definitions. Results show improved initiation rates (feeding in first 48 hours) for any breastfeeding, 85.3% (83.6-87.0) versus 91.1% (90.0-92.2) p<10-6; and exclusivity, 29.8% (27.6-32.0) versus 38.7% (36.8-40.6) p<10-6, with differences of similar magnitude at all periods for any breastfeeding. Prevalence of those categorised as exclusively feeding showed significant

differences at discharge (70.0% (68.2-72.6) versus 76.8% (75.1-78.5) p<10-6) and four and six months at the level of p<0.001%, but not at one month. Predictors of any breastfeeding duration were adherence to steps 4 and 9, higher education and non-smoking status of the mother. Predictors of exclusivity were the same mother characteristics and reported adherence to steps 6 and 9. This study also showed that rates of initiation and duration were predicted by both mothers reported experience of the hospitals adherence to BFI standards and socio-demographic characteristics (higher educational level and fewer smokers), indicating that greater impact of the intervention was shown among those known to be more likely to breastfeed. The authors discuss the high conversion rate (87%) of mothers who started to mix feed but were exclusive at discharge. This means that hospital practices may be encouraging supplementation. This also means the babies were not, in fact, exclusively fed according to the strict interpretation of WHO definitions (see Appendix 8) since exclusivity is not possible if the baby ever had other than breastmilk. Analyses are not available on the influence of in-hospital mixed feeding on subsequent duration of any or exclusive breastfeeding at later end points.

Discussion and conclusions National policy can be used to set a framework for health service strategies to support breastfeeding. These should be inter-agency, given the need to involve public agencies with influence, eg employment practices, school education on health, use of public places to encourage breastfeeding. National public health and health services policy can be used to drive strategies that include the use of agreed breastfeeding data definitions, timing and reporting of routine collection for all mother-infant pairs. In the paper by Britten and Broadfoot (2002) [38], standard definitions of breastfeeding were not in use at six weeks. Unless these are agreed and collection occurs for all babies at six weeks, measurement against the target will not be possible. Where these data can be tied to specific geographic locations, services and to mother and infant pairs, this will allow analysis of the influence of socio-demographic factors, and health service or other interventions (such as BFI) on breastfeeding outcomes. 71

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

National public health and health services policy can be used to drive strategies that engage the health service and other professionals and the public in deciding resources for breastfeeding support activities and how these can be evaluated. There is evidence that the use of targets, penalties and incentives can be effective, but they probably could be more effectively used as part of an integrated strategy of health improvement. Future research evaluation of health service policy There are no studies with breastfeeding duration data of policy initiatives such as the provision of employment leave, implementation of the WHO Code on Marketing of Breast Milk Substitutes, workforce policies to support breastfeeding mothers, promotion of formula milk and restrictions on provision of formula by prescription. National case studies can be designed using national data or surveys and could usefully combine both approaches until national data are shown to be collected in line with the principles as follows. Study data collection must be at a unit of analysis to allow differentiation by mother and infant pairs, health service organisation and geographic area, and should include sufficient socio-demographic data to allow investigation of inequalities issues. Data from the Guthrie records (seven day feeding status) for babies in Scotland allow analysis by hospital, postcode and post district. Analyses could usefully be conducted to link breastfeeding rates with the stated interventions reported at health board level, since maternity units could differentiate mothers by postcode to the appropriate board. Analyses by postcode, linked to deprivation indices, would allow for analyses by factors likely to influence breastfeeding outcomes. This would provide observational data on whether there is a correlation between policy implementation and breastfeeding outcomes at both the level of local boards, maternity units and nationally. Work to link evidence of policy implementation to breastfeeding outcomes is in progress (J. Warren, National Breastfeeding Adviser, Scotland, personal communication).

72

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Table 2.1: Included studies public policy interventions Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Discharge packs At the time of discharge from hospital 6 weeks, 4 and 6 months Timing of breastfeeding duration outcomes

Name of authors and study number 1,625 healthy mothers and their healthy newborns (over 1/5th were minority ethnic, ~1/3rd were on 1Medicaid; ~1/4th were single parents)

Date of Country publication

Design

Bliss et al. [37]

1997

US

RCT

Britten and Broadfoot [38] National policy of locally set targets for breastfeeding, within national policy of breastfeeding target for all Scotland

2002

Scotland

Observational Midwifery unit staff study, before- responsible for breastfeeding after design promotion and support in all 33 units in Scotland Baby breastfeeding status at day 7 (Guthrie data) as % of national dataset

Breastfeeding rate at 7 days 1994 Scottish Office issued (Guthrie data), any bf. Data guidance on setting bf targets, bf strategies, and BFI. compared 1995 to 1999 Target in 1994: More than 50% of women to still be breastfeeding their baby at 6 weeks of life by 2005. And set local targets, address local variations. 1995, Scottish Office to be advised of action to address variation. BFI recommended to boards and trusts (1996) 1998: Included bf promotion in targets linked to financial penalties Data collected for 9 months in 1998 and 9 months in 1999, bf at 16-19 weeks Data are % exclusive, full bf, complementary and no bf

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Cattaneo et al. 2001 [39] RHA, 6 LHAs, 10 maternity units, n=5,968 infants (1998), n=6,568 infants (1999)

NE Italy (Friuli Venezia Giulia)

Observational study, beforeafter design using health service returns

RHA policy of penalties/ incentives for LHAs breastfeeding promotion activities. These included policies to remove nurseries in maternity units, disseminate guidelines, improve hospital routines, create support groups, staff training, bf in antenatal care, booklets on bf, home visits for bf problems, increase bf rates at discharge and 4 months

73

74 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Discharge packs At the time of discharge Every 2 weeks until 16 weeks Timing of breastfeeding duration outcomes 763 healthy mothers and their healthy term infants (Most participants white, educated, middle-income; many women lost to followup were low-income, less educated) Adherence to BFI standards in Unspecified period and hospitals activities related to BFI Ten Steps to Successful bf Breastfeeding within 48hr delivery, birth, discharge, and when infants are 3, 9 and 12months 2, 6, 12 and 24 weeks postpartum 444 pregnant women (less than 1% were on Medicaid) Prenatal formula advertising At the first prenatal visit

Table 2.1: Included studies public policy interventions (cont.)

Name of authors and study number

Date of Country publication

Design

Dungy et al. [40]

1997

US

RCT

Giovannini et al. [41]

2003

Italy

Observational Sample of mothers delivering study (before- infants in November 1995 after) (n=1,601) and November 1999 (n=2,450)

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Howard et al. [42]

2000

US

RCT

3.1 Clinical interventions: antenatal preparation for breastfeeding


Topic Antenatal practices, including antenatal expression of colostrum, nipple preparation, and treatment for inverted and non-protractile nipples

Summary of methods
Study design of included studies: RCTs only Dates for searches: 1995 to June 2003 Population group studied: pregnant women Countries included: all countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes

3.1.1 Background
Pregnancy is a time when women experience changes in their breasts, and when they may consider how they wish to feed and their own physical ability to breastfeed. This may raise a range of issues, including psychosocial issues related to their bodies and specifically their breasts; and questions about the physical shape and normality of their breasts and nipples. It used to be common practice for healthcare professionals to examine all womens breasts in pregnancy. As a result of research that has questioned the purpose of such examinations (eg Alexander et al., 1992) this practice is now not universal in the UK, though it is still commonly practised in other countries. A range of practices have been recommended to pregnant women over the years with the aim of increasing the duration of breastfeeding. Two main areas of care have been examined in research studies: antenatal expression of colostrum and nipple preparation. Antenatal expression of colostrum is intended to help to establish early colostrum and milk production and to familiarise women with handling their breasts. Nipple preparation has two purposes: first, to toughen and condition the nipples to prevent pain and damage when the baby starts to feed, by practices which include the

use of lotions, rubbing with towels, and nipple rolling. Second, nipple preparation has been used to treat inverted and non-protractile nipples. Studies have suggested that around 7-10% of women experience problems with nipple protractility (Hytten, 1954; Alexander et al., 1992). Treatments used have included wearing breast shells in pregnancy, and Hoffmans exercises (stretching the tissue around the base of the nipple), the Niplette device (McGeorge, 1994), which attaches by suction to the nipple in pregnancy, and breast surgery. Such preparatory practices are not necessarily benign: they could cause harm by raising maternal anxiety, by causing damage to the nipples if used too vigorously and by damaging breast tissue and ducts if surgery is used.

3.1.2 Evidence base


No studies relating to any aspect of antenatal breast preparation were identified in this review.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

75

3.1.3 Findings from previous reviews


Enabling women to breastfeed (Renfrew et al., 2000) identified four studies on the routine expression of colostrum (Waller, 1946; Blaikeley et al., 1953; Ingelman-Sundberg, 1958; Brown and Hurlock, 1975), three of which were conducted several decades ago. Six studies were identified on nipple conditioning, three of which were methodologically very flawed (Hewat and Ellis, 1987; Storr, 1988; Brown and Hurlock, 1975, Clark, 1985; Whitley, 1978; LEsperance, 1980). Two wellconducted RCTs on treatment for inverted and nonprotractile nipples were identified (Alexander et al., 1992; MAIN Trial Collaborative Group, 1994). In summary, the results of these studies suggested that: Antenatal expression of colostrum may result in an increase in the duration of breastfeeding and a decrease in the incidence of engorgement and sore nipples, though the studies are dated and methodologically flawed There is no evidence to support the conditioning of nipples in pregnancy There is no evidence to support the use of any antenatal treatment for inverted or non-protractile nipples or even to support antenatal breast examination to identify women with this condition.

It could be argued that expectant women should be encouraged to regard touching or handling their breasts as normal and, while antenatal expression may be a way of achieving this, the small number of studies conducted have shown no clear benefit, though these is also no suggestion of harm. In relation to conditioning nipples in pregnancy, there is no evidence to support this practice. Since cineradiographic and ultrasound studies have shown that the well-positioned nipple is protected from friction, and therefore from pain, by being drawn to the back of the babys mouth (Ardran et al., 1958; Smith et al., 1985; Weber et al., 1986), the emphasis has shifted away from toughening or conditioning nipples towards preventing the cause of trauma (Woolridge, 1986b; see Section 3.3). There is no evidence, therefore, to support its use, and some women may find it painful or distasteful. In relation to inverted and non-protractile nipples in pregnancy, both RCTs (Alexander et al., 1992; MAIN Trial Collaborative Group, 1994) found that many women with this condition, in both the control and intervention groups, did continue to breastfeed. This finding in itself indicates that breastfeeding is possible with inverted and non-protractile nipples. Women are still routinely advised to wear a good supporting bra both throughout pregnancy and lactation. To our knowledge this has not been evaluated by a controlled trial although many women will elect to do this for reasons of comfort and personal choice. No studies have examined simple interventions such as the opportunity for a woman to talk through her concerns about her breasts, nipples and ability to breastfeed with a knowledgeable and skilled person, or for her to have information about the wide range of the normal physical appearance of breasts and nipples described and discussed.

3.1.4 Discussion and conclusions


In relation to antenatal expression of colostrum it is perhaps surprising that the encouraging results of the studies identified in Renfrew et al. (2000) have not resulted in further research. The results may be explained by other factors, however; for example, women in one study (Blaikeley et al., 1953) also received other forms of care (general advice on breastfeeding, advice on suitable brassieres, attention to nipples and treatment of them with glass shells where necessary) and it is likely that women in both groups received additional attention from midwives and obstetricians after birth. There was a high rate of engorgement in both Wallers and Blaikeley et al.s studies; these studies took place at a time when feeding times and duration were limited. It could be that antenatal expression combined with postnatal expression is an effective treatment for engorgement, but that prevention of engorgement by flexible feeding patterns is the best solution to that problem. 76

3.1.5 Future research


A well-designed RCT of antenatal expression of colostrum, conducted in the setting of a modern hospital implementing up-to-date practices, would address the methodological problems in the older trials identified.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

There is no indication that further research on routine nipple conditioning would be of benefit. Use of the Niplette device (McGeorge, 1994) and other unevaluated treatments for inverted and non-protractile nipples should be evaluated by an RCT to examine its effectiveness, possible harmful side-effects and the views of women who use it. Descriptive studies are needed to investigate the support and care needed by women who have inverted and nonprotractile nipples in the early days and weeks after birth, to support them in the initiation and maintenance of breastfeeding and to explore ways of helping babies to take the breast effectively. Questions remain about women who may be especially susceptible to developing sore nipples, including women with persistent dry or sensitive skin, or a history of dermatitis or eczema of the nipple and areola. Antenatal treatment of these conditions would arguably avoid the use of pharmacological preparations during the course of breastfeeding but no studies of the antenatal treatment of these conditions have been identified. There is increasing concern for the number of women who have been sexually abused during their adolescence and who may, in consequence, have deeply entrenched feelings about their breasts and about them being handled. Methods for identification of women at risk and the efficacy of counselling in such situations is an important area for research. In the light of the lack of evidence to support any antenatal treatments in pregnancy, a descriptive study of the views of women on the care they would like to receive in pregnancy to prepare them for breastfeeding is needed. Women may, for example, find it helpful to talk about breastfeeding or to become familiar with handling their own breasts and nipples. Given that sexually explicit information is increasingly available in many media it is likely women may have misinformation about the normal range of appearance and size of breasts and nipples, leading to unfounded concerns and unnecessary help seeking or self-treatment. This requires further research.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

77

3.2 Clinical interventions: interventions in childbirth


Topic Interventions in childbirth, including place of birth, continuity of care in labour, pain relief in labour, interventions including induction, amniotomy, perineal care, food and drink in labour, and mode of birth where breastfeeding has been an outcome measured

Summary of methods
Study design of included studies: reviews of RCTs, non-RCTs with concurrent controls, before-after studies Dates for searches: 1990 to June 2003 Population group studied: women in labour/at birth Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, views of those experiencing and providing care, relevant process outcomes including health service and economic outcomes

3.2.1 Background
The management of labour may impact on breastfeeding outcomes, both in respect of initiation and duration (eg Bernard-Bonnin et al., 1989; World Health Organization, 1998a; Righard and Alade, 1990; RansjoArvidson et al., 2001). The effect on breastfeeding may occur in a variety of ways; for example, through the influence of pain-relieving drugs that cross the placenta from the mothers circulation to the babys, or through the impact on both mother and baby of mode of birth, or length of labour. Breastfeeding may be affected by the mothers experience of comfort and pain; following an instrumental birth, perineal pain may make it more difficult for mothers to achieve comfortable positions when they and their baby are ready to initiate breastfeeding. It is possible that, in addition to the immediate postnatal concerns arising from these problems, different patterns of labour and the interventions received may also have more enduring consequences for breastfeeding continuation. These may include physical sequelae such as anaemia and increased risk of infection, as well as important psychosocial issues such as the womans perceptions of involvement in decision making in labour and the extent to which she felt in control over events in labour. 78

Early research related to care in labour focused largely on immediate labour outcomes, such as duration of each stage of labour, mode of delivery, perineal trauma and the condition of the baby at birth. Interest in womens experiences of labour as a research outcome first appeared in the mid-1980s, but breastfeeding was not generally reported as an outcome in childbirth research until later in both UK and international studies. This section of the review differs from the others in that it considers only existing systematic reviews. This is a large topic area and resources did not permit a full-scale review. It was considered to be too important an area, however, simply to omit from consideration.

3.2.2 Evidence base


The search resulted in four reviews, three of which met our criteria. Table 3.2 shows the three reviews that contribute data for discussion in this section. Hodnett et al.s review (2003) [44] examined the provision of continuous support for women during childbirth; Mercers review (2001) [45] explored the impact of early and delayed umbilical cord clamping on breastfeeding and Di Matteo et al. (1996)

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

[43] investigated associations between birth by caesarean section and psychosocial outcomes. The fourth review will be discussed briefly below.

3.2.3.2 Delayed clamping of the umbilical cord


Mercers (2001) [45] review includes randomised clinical trials and controlled trials of early versus delayed cord clamping. Early clamping is defined by Mercer as immediately or within one minute of birth. Delayed cord clamping is defined as three minutes after birth, cessation of pulsations in the umbilical cord or within 10 minutes of birth. A range of outcomes is reported in the review including the duration of breastfeeding. One UK trial included in the review contributes data related to breastfeeding duration. This trial, conducted by the Oxford Research Midwives Group and published in 1991, examined breastfeeding at 10 to 12 days following birth. In this trial, early cord clamping was defined as within one minute of birth (256 women allocated to receive this treatment) and delayed cord clamping three minutes after birth (296 women allocated to receive this form of care). Women who participated in this trial laboured at term following an uncomplicated pregnancy with a baby in a cephalic presentation and had a vaginal birth. Analysis of data was based on the principles of intention-to-treat analysis with 32 women allocated to delayed intervention receiving early cord clamping. Delayed cord clamping appears to have a beneficial effect on breastfeeding duration, as self-reported by women at 10 to 12 days after birth; this reaches a statistically significant difference (p<.05) although original data are not shown in the review.

3.2.3 Findings 3.2.3.1 Continuous support in labour


Hodnett et al.s Cochrane review (2003) [44] included 15 trials conducted in several countries involving 12,791 women. Continuous support was provided by healthcare professionals, doulas, childbirth educators, family members and friends or strangers with no specific training in labour support. Usual care within these trials is defined as intermittent support from nursing or midwifery staff. The possibility of family members or partners being present during the womans labour varied between settings in that review. Only three trials reported data for breastfeeding outcomes. These include one trial (Hofmeyr et al., 1991) excluded by Hodnett et al. (2003) [44] from further consideration in their review due to loss to follow-up of over 20% of the participants. A trial conducted in Mexico City by Langer et al. (1998) is not considered further for this evidence briefing as it falls outside the countries included in this investigation. The remaining trial was conducted by Hodnett and colleagues in North American and Canadian settings and involved 6,915 women of all parities, labouring from 34 weeks of pregnancy onwards with singleton or multiple pregnancy. They received continuous support from a hospital-based nurse who had attended a special two-day training course related to providing continuous support (Hodnett et al., 2002). The outcomes for women who received care from these nurses were compared with those who had received the usual, intermittent attention from nursing staff. Breastfeeding duration outcomes are reported as not breastfeeding at two months. Thus, data from this one well-designed RCT carried out by Hodnett et al. contributes data to this evidence briefing. While the provision of continuous support during childbirth was associated with several other benefits to women including less use of analgesia, fewer operative births and less reported dissatisfaction with childbirth, no difference in breastfeeding rates was detected among women at two months postpartum (RR 1.06 95% CI 0.99, 1.14).

3.2.3.3 Birth by caesarean section and psychosocial outcomes


The third review Di Matteo et al. (1996) [43] compared outcomes following caesarean section (both planned and emergency) with vaginal birth for a range of psychosocial outcomes that included breastfeeding. This review provides information about decision to breastfeed (10 studies), breastfeeding up to three months (eight studies) and breastfeeding beyond three months postpartum (five studies), although data are not always presented in the most accessible form. The methods of data collection used in the primary research are described and vary between interviews and record reviews. Statistical methods are described but there is no information included in the review about the setting, organisation of care, research participants or the care providers. The papers contributing breastfeeding data to Di Matteos review [43] were published between 1982 and 1990 in a range of European and US journals. Women who gave birth by caesarean section were less likely to decide to 79

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breastfeed, with a more pronounced negative effect when the analysis included only unplanned caesarean sections. However, while women who gave birth by caesarean section were less likely to breastfeed, once initiated the type of birth (caesarean or vaginal) was not associated with differences in breastfeeding up to and after three months postpartum.

3.2.3.4 Unintended effects of epidural analgesia during labour: a systematic review


The fourth review identified in the search cites two pieces of primary research where breastfeeding was included as an outcome (Lieberman and ODonoghue, 2002). However, a lack of appropriate data, probable differences with the UK context and lack of information about epidural method preclude further consideration of this review.

We are, therefore, left with two reviews each contributing data from one trial. However, it appears appropriate to include comment on UK-based studies here (from Mercers review [45]). Similarly, the principle of continuous labour support appears generalisable and important to the UK context (from Hodnett et al.s 2003 review [44]), although the continuous support will most commonly be provided by midwives rather than specially trained nurses or doulas. The lack of contextual information in the studies contributing data to Di Matteos review [43] perhaps reflects the strong psychological orientation to that investigation but requires caution in the interpretation of results. The fact that women who give birth by caesarean section are less likely to decide to breastfeed may not be a surprise for those who have tried to assist mothers with early skin-to-skin contact and initiation of breastfeeding following caesarean section. There remains an urgent need for skilled help at this time. Women who give birth by caesarean section can, however, be reassured that from this review there is no evidence that they will not be able to continue breastfeeding, if they wish, as long as women who experienced vaginal birth. There appears little that can be drawn from this review of existing reviews in terms of recommendations for significant changes in practice due to the limitations of the studies discussed previously. There are many reasons why continuous support should be provided for women in labour, so despite the lack of positive impact on duration of breastfeeding in Hodnett et al.s (2002) trial, there is no evidence of a negative effect and thus no reason to suggest that this should not be available to women. While the results of the one UK-based trial in Mercers review [45] have to be interpreted with caution, delaying cord clamping up to three minutes after birth appeared to confer benefit in terms of the proportion of women still breastfeeding at 10 to 12 days after birth. The mechanisms for this are unclear and need further exploration. This trial took place in a setting where active management of the third stage of labour was the norm. In settings where midwifery led care is provided, physiological management of the third stage of labour is more likely to be a real option for women and the midwives who attend them. Physiological management of the third stage of labour has several components that

3.2.4 Discussion and conclusions


Despite a comprehensive search strategy, only three reviews were identified that could contribute data to this important section. Some of the studies that have contributed data to the reviews were published over 10 years ago; however information from Hodnett et al.s (2002) research on continuous support in labour, in particular, can be considered entirely current. The research conducted by the Oxford Midwives Research Group (1991) took place at a time when active management of the third stage of labour and immediate cord clamping were the norm. This remains the method of management of the third stage most commonly but not universally practised in the UK, as some women wishing a non-interventionist approach to labour may opt for physiological management if given the choice, and some midwives may offer this in low-risk situations. There is no information presented in Di Matteos review [43] related to the context or content of care provided to women in the included studies that allows comparison with current UK care provision. In addition, it is frustrating to note that several trials included in the reviews did not measure breastfeeding as an outcome. Further, due to the limitations defined in the protocol to ensure applicability to UK practice, some of the data presented by the reviewers could not be included here.

80

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

include delayed clamping of the umbilical cord, although the time period for this may well exceed that investigated in the Oxford trial. Recent guidelines published by the National Institute for Clinical Excellence recommend that women should have the opportunity for a discussion on the risks and benefits of caesarean section (National Institute for Clinical Excellence, 2004). Women who give birth by caesarean section can be reassured that, once breastfeeding has been initiated, they appear to be no less likely to continue than women who experience a vaginal birth. For women who had perhaps hoped for a vaginal birth and are subsequently disappointed if this does not occur, this information may be of some encouragement as they recover physically from the birth and adjust to being a new mother.

3.2.5 Future research


Future trials and other research into intra-partum care should explore the impact on breastfeeding duration at appropriate standardised times using definitions of breastfeeding that are useful for care providers, women, researchers and commissioners. Reports should include sufficient information to ensure that information about the context and pattern of care is clear and accessible, thus assisting consideration of the applicability of results beyond the original setting. Similarly, there should be adequate information describing the population and sub-groups participating in research for demographic factors including ethnicity, parity and socio-economic group to increase the utility of reviews. In addition, studies should specifically examine the impact of delaying umbilical cord clamping as part of active management and the effect of pain-relieving drugs in labour on breastfeeding. None of these reviews provided adequate information about the effects of pain-relieving drugs, however administered, on breastfeeding, although there are indications from some studies that this is an important problem (Ransjo-Arvidson et al., 2001; Baumgarder et al., 2003; Riordan et al., 2000; Halpern et al., 1999). Such studies should be examined closely and further research developed to examine ways of minimising such drug use in labour and, importantly, of supporting breastfeeding when they have been used.
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

81

82 Country (* publication of review) (** studies contributing data to this review Intervention and timing in studies eligible for this review Timing of breastfeeding duration outcomes Participants and any sub-groups * US ** Not stated Birth by planned and unplanned caesarean section compared to vaginal birth Continuous support during labour by a healthcare professional, doula, childbirth educator, family member, friend or stranger with no special training (n=3,454) versus usual care = intermittent support from a nurse (n=3,461) Delayed cord clamping 1 minute (n=256 women) versus early cord clamping 3 minutes or when pulsations ceased (n=296 women) Women who gave birth by caesarean section (planned or unplanned) and vaginal birth Women in labour, 34 weeks, singleton or multiple pregnancy, nulliparous and multiparous. No sub-groups Decision to breastfeed. Breastfeeding up to and at 3 months Not breastfeeding at 1-2 months postpartum * US ** US-RCT * US ** UK-RCT Women in labour at term, vertex presentation, no antenatal complications, vaginal delivery. No sub-groups 10-12 days

Table 3.2: Included studies interventions in childbirth

Name of authors and study number

Date of Publication (* date of review) (** date of studies contributing data to this review)

Di Matteo et al. [43]

* 1996 **1982-1990

Hodnett et al. [44]

* 2003 **2002

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Mercer [45]

* 2001 **1991

3.3 Clinical interventions: prevention of clinical problems


Topic Strategies to prevent clinical problems resulting in premature discontinuation of breastfeeding. It includes a wide range of postpartum routines: timing of the first feed, prevention of sore nipples, teaching positioning and attachment to mothers, frequency and duration of breastfeeding, skin-to-skin contact/kangaroo care, supplementing breastfeeds, the use of weight and food charts, methods of giving additional fluids, and the use of pacifiers Summary of methods
Study design of included studies: RCTs Dates for searches: 1995 to June 2003 Population group studied: breastfeeding mothers of healthy babies Countries included: all Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, problems encountered, views of those experiencing and providing care, relevant process outcomes including economic outcomes

3.3.1 Evidence base


Thirteen studies including two theses met our inclusion criteria. All 13 were RCTs. Of these three were related to supplementing newborns, two to methods of giving supplements, three to pacifier use, two to kangaroo care, three to nipple care to prevent sore nipples and one each to teaching positioning and attachment, postoperative analgesia and breast pumping. A summary of the studies is shown in Table 3.3.

that the baby is fed within the first half hour after birth (World Health Organization/UNICEF, 1989). The relevant step in the Baby Friendly Initiative (BFI) Ten Steps (step 4, see Appendix 7) has been modified in the UK to read soon after birth rather than the within half an hour of birth that was considered to be unduly restrictive. It remains the step as recommended globally, however. Hamlyn et al. (2002) found that 72% of breastfeeding mothers fed for the first time within one hour in 2000, a slight increase from 68% in 1995. Evidence base

3.3.2 Findings 3.3.2.1 Timing of the first feed


Background Practices in the timing of the first breastfeed vary widely from culture to culture and include withholding colostrum and giving substitute feeds for the first few hours or days (Morse et al., 1990). These practices may still be followed in some centres and communities. In the past few decades, recommendations in the UK about the timing of the first breastfeed have ranged from waiting until six hours after birth (Myles, 1975) to recommending No trials of the timing of the first breastfeed were identified in our search for this review. Findings from previous reviews Renfrew et al., 2000 identified five RCTs (Eppink, 1969; Widstrom et al., 1990; Van den Bosch and Bullough, 1990; Woolridge et al., 1985; Salariya et al., 1978) and several non-randomised studies. Overall, there are advantages in favour of the early initiation of breastfeeding but no research has demonstrated a critical period for the first feed in terms of breastfeeding success, and there are no grounds for interfering with the 83

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

mothers choice, or for interfering with the normal behaviour exhibited by the baby. That review suggests that no restrictions are placed on contact between mother and baby at any stage, or on the time of the first breastfeed. Benefits identified from early feeding include an increased duration of breastfeeding, an enhanced relationship between mother and baby and a reduction in o the number of babies with temperatures below 36.5 C. Discussion and conclusions There is no evidence to support any restrictions on maternal-infant contact at birth and mothers should be encouraged to have as much close and intimate contact with their baby as they wish, and to feed when they are ready to do so. Although every woman should be given the opportunity to feed soon after birth there is no evidence to suggest that if she does not feed immediately she cannot establish effective breastfeeding. Women who have a delayed onset to breastfeeding, therefore, should be reassured that this is not a contraindication to establishing breastfeeding successfully. One explanation for the fact that no trials of early vs late contact have been conducted since 1995 could be that this is an issue where the evidence base is considered to be sufficiently strong that new trials are not warranted. Indeed, it is likely that a trial in which one arm was separation of mother and baby at birth would now be considered to be unethical. It is still the case, however, that in practice mothers are not always allowed to have the close, intimate and ongoing contact with their babies that they may wish; routines in care are still reported to interrupt this contact (eg Hamlyn et al., 2002) Future research timing of the first feed Research questions related to this issue include: What are the effects of birth experience and pain relief in labour on the babys ability to initiate and maintain breastfeeding? See also Section 3.2 Why is there resistance in some settings to allowing mothers and babies to have unrestricted contact and early feeding, and how can staff be trained and supported to enable this?

3.3.2.2 Kangaroo care and skin-to-skin contact for healthy term newborns
Background Kangaroo mother care (KMC) was first developed in Columbia in the late 1970s as an alternative to a strategy for caring for low birth weight babies. KMC is defined as skin-to-skin contact between a mother and her newborn, frequent and exclusive or nearly exclusive breastfeeding, and early discharge from hospital. It has also been used for term, healthy babies. Skin-to-skin contact (SSC), only one of the three main components of KMC, involves placing the naked baby (wearing a nappy if wished) prone on the mothers bare chest at birth or soon afterwards. In practice a wide range of contact time is observed between mother and newborn in the first 24 hours and thereafter. Studies have tested the effectiveness of SSC starting immediately after birth, within 30-45 minutes or within 24 hours. The term kangaroo care or KMC therefore encompasses a range of practices making its assessment problematic. To complicate this assessment further, some older studies of the timing of the first feed (as discussed above) included close contact along with early feeding as their intervention. Evidence base We identified no published studies of kangaroo care or skin-to-skin contact at birth. Two theses investigated the effect of kangaroo care on breastfeeding success in healthy term infants. Findings Punthmatharith (2001) [56] randomised 180 healthy newborns in a hospital in Thailand to receive early kangaroo care or standard care. Early KMC began at about 60 minutes after birth. Mothers lay supine and held their infants (wearing only a nappy) under their clothing between their breasts, allowing self-regulatory breastfeeding in response to infant feeding cues. Mothers were taught the KMC method and recognition of infant feeding cues. They were encouraged to follow KMC for as often and as long as possible until discharge from hospital. Skin-to-skin contact was discontinued temporarily during hospital procedures such as transferring from labour suite to postnatal ward and when the mother desired. Standard care was supportive of breastfeeding but did not include skin-to-skin contact

84

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

in the same manner. Breastfeeding success was defined as degree of exclusivity of breastfeeding; the mothers perceived sufficiency of breastmilk and infant weight gain. No significant differences between the two groups were found in exclusive breastfeeding on day two and at one month. This study was of good quality, inclusion/exclusion criteria were clear, analysis was by intention-to-treat, and the sample size was calculated a priori and was large enough to detect differences in breastfeeding status. It is notable, however, that participants in both groups were at low risk of difficulties related to mother-infant bonding 80% intended to breastfeed for more than three months, none had intended to formula feed, nearly two thirds of mothers had experienced breastfeeding and the hospital was Baby Friendly. It is possible that the findings might have been different in a setting where the rates of intention to breastfeed were lower, rates of intention to formula feed were higher, the hospital was not Baby Friendly and fewer mothers had experience in breastfeeding. Another thesis (Shiau, 1997) [58] investigated the impact of KMC on maternal anxiety, breastmilk maturation, breast engorgement and breastfeeding status. Fifty-eight mothers and their newborn infants were randomised to the KMC group, in which skin-to-skin contact started four hours after birth and consisted of at least eight hours contact each day in hospital; or the standard care group in which there was no rooming-in. Breastfeeding status was based on the index of exclusive breastfeeding adapted from Labbok and Krasovec (1990) in which the lowest score (1) indicates minimal breastfeeding, and the highest score (8) indicates exclusive breastfeeding with no other food or fluids given to the infant. This study found that KMC reduced maternal anxiety and breast engorgement and improved breastfeeding status in the first three days postpartum. This was a small study meeting all quality criteria for an RCT. Given that standard care lacked rooming-in, however, it is difficult to assess whether the results related to the intervention as a whole, or just the improved contact between mother and baby. Findings from previous reviews A Cochrane review (Anderson et al., 2003a) examined this issue in detail. It included 17 studies, ranging in publication date from 1976-1997, and included two

studies that met our criteria: Shiau (1997) [58] described above and Syfrett (1993), which was an unpublished thesis from the US. This review helpfully discusses the problems of definition of kangaroo care and skin-to-skin contact, and the methodological problems in the field. Its conclusions indicate that skin-to-skin/kangaroo care (all definitions included) has a positive impact on long-term breastfeeding, with no negative effects identified. It does not distinguish between the potentially very different effects of immediate, post-birth skin-to-skin contact and early feeding, from kangaroo care introduced at a later stage. Discussion Policies that restrict contact between mothers and their newborns are separation soon after birth, keeping newborns in nurseries and allowing vulnerable babies to sleep separately undisturbed for the first 12 hours or more. Kangaroo care and skin-to-skin care work in the reverse, improving contact between mother and baby. Our review includes two studies of 238 healthy term newborns. One trial found improvements in breastfeeding status in the first three days after birth and reduction in maternal anxiety. The review by Anderson et al. (2003a), which includes a number of older studies, identified a positive impact on long-term breastfeeding and other maternal and infant-related outcomes. There is no evidence that the practice causes any harm. Mothers should be supported and encouraged to have as much close contact with their babies from birth as they wish, including skin-to-skin care. Future research kangaroo care and early skin-to-skin contact Further trials of kangaroo care and early skin-to-skin contact would be helpful to identify if there are subgroups of mothers for whom such care is especially helpful, including mothers in settings where breastfeeding is not well supported, those who have had difficult birth experiences, and those from communities where breastfeeding is not the norm. Care should be taken to clarify the extent and timing of such interventions, and to distinguish between care immediately after birth and ongoing care.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

85

3.3.2.3 Prevention of sore nipples


Background Sore nipples and painful breasts are among the most common reasons women give for abandoning breastfeeding. Underlying the problems of damaged (cracked) and sore nipples is incorrect attachment (relationship of the babys mouth with the mothers breast) and positioning (relationship of the babys body axis to the mothers body axis). The problem of sore nipples could be prevented if the baby fed without causing damage to the mothers nipples. Two different types of damaged nipples were described by Gunther (1945): an erosive or petechial lesion, and a fissure of the nipple. This paper was the first to ascribe severe pain and nipple damage to the position of the baby on the breast. It also described the very high pressure exerted when a baby is at the breast but not able to obtain a good supply of milk and showed an association between the incidence of soreness and the babys birth weight. A good understanding of the importance of attachment of the baby at the breast was later gained from the cineradiographic studies by Ardran et al. (1958) and ultrasound studies conducted by Smith et al. (1985), Bosma et al. (1990) and Weber et al. (1986); this work was summarised by Woolridge (1986a). These studies demonstrate the relationship between the baby taking a good mouthful of breast tissue into his or her mouth and the resultant lack of damage to the nipple with unrestricted milk flow. With proper attachment the nipple is protected from damage at the back of the babys mouth, and the milk ducts under the areola are effectively milked by the babys tongue. The result should be pain-free feeding, effective milk removal and a satisfied baby. Evidence base We include four RCTs in this section, one of positioning education and three on topical applications to the nipple. One of the RCTs (Akkuzu and Taskin, 2000 [46]) does not give breastfeeding outcomes. We include it for the secondary outcome: development of cracked/sore nipples.

Findings Henderson et al. (2001) [49] randomised 160 first time mothers in Australia to the standard care group or the intervention group. The intervention consisted of 30 minutes of intensive one-to-one education within 24 hours of giving birth, consisting of visual, written and verbal information and the use of a model. A hands-off method was used to teach self-positioning and attachment to the intervention group mothers, and cues for the mother to assess her own technique. Attachment and positioning technique were assessed using the LATCH tool (Jensen et al., 1994) and feedback was given to mothers each day of hospital stay. Standard care varied, but usually meant that a midwife was actively (hands-on) attaching the infant to the mother; formal education, assessment of breastfeeding technique and feedback were not included. No significant differences were found between the two groups in breastfeeding duration at six weeks and at three and six months. There was, in fact, a trend towards lower breastfeeding rates in the intervention group. Less nipple pain was reported in the intervention group on days two (p=.004) and three (p=.04), but this was not sustained. No differences were found in nipple trauma at the three assessment points. The main limitation of this study was that the researcher was the midwife delivering the intervention, making the possibility of confounding large. There was also the possibility that midwives delivering care may have modified their practice, knowing that a trial of this nature was being conducted in their unit. Finally, for this study to detect a difference of 8% in breastfeeding rates at six months a sample size of 1,224 would have been required. A larger trial would be needed to reliably assess this intervention. Akkuzu and Taskin (2000) [46] compared the effectiveness of warm wet compresses, application of expressed breastmilk and keeping the nipples clean and dry on the incidence of cracked nipples and nipple pain in breastfeeding mothers. Ninety mothers were randomised to one of the three groups. All mothers were instructed on the benefits of breastfeeding, taught how to breastfeed and were given a brochure with relevant information. More mothers applying warm and wet compresses to their nipples and those applying expressed breastmilk had cracked nipples on days two and three. The duration of cracked nipples was shortest among mothers who applied expressed breastmilk. No

86

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breastfeeding duration outcomes are reported. This study has poor internal validity inclusion/exclusion criteria were not clear, there was no a priori sample size calculation, and it was not clear if randomisation was true or if the analysis was by intention-to-treat. We cannot have confidence in the findings from this study. Centuori et al. (1999) [47] randomised 219 women in Italy to either an intervention group in which there was total avoidance of application of any chemical, physical or mechanical agent to the nipple, or to the control group who were asked to follow the existing hospital routine cleaning nipples with moist gauze before and after each breastfeed, and unrestricted application of either an ointment prepared in the hospital pharmacy or a commercial spray. Rooming-in and breastfeeding on demand were recommended in the hospital but not enforced; breastfeeding management support was not routinely provided to women. This study found no significant differences between the groups in nipple pain or trauma and breastfeeding duration at discharge, two weeks and three months postpartum. As the sample size was not achieved the findings from this study need to be interpreted with caution. Pugh et al. (1996) [55] randomised 176 women to four groups assigned to receive either warm and wet compresses, application of expressed breastmilk, modified lanolin or education only. No significant differences were found between groups in breastfeeding duration at six weeks, or in sucking strength, pain intensity or pain affect. This study was generally of good quality although no rationale was given for the sample size calculation. No information was given about comparability of groups at baseline or if formula feeds were given. Findings from other reviews Renfrew et al. (2000) identified one RCT (Righard and Alade, 1992) and two other studies (Duffy et al., 1997; Long, 1995) on the teaching of positioning. The Swedish RCT, which included 82 breastfeeding women, found a significant effect of instruction for women with incorrect sucking technique on breastfeeding at all and exclusively at four months. Renfrew and colleagues (2000) also identified two RCTs (Slaven et al., 1988; Herd and Feeney, 1986) that examined the routine use of chlorhexidine spray to prevent sore nipples. These studies included a total of 923 women. Slaven et al. (1988) found no differences between the groups, and Herd and

Feeney (1986) found an increase in nipple trauma in the group using the spray. There is no evidence to support the use of chlorhexidene spray to prevent sore nipples. Discussion and conclusions The theoretical and descriptive work described above in the background to this section demonstrates that sore nipples, whether associated with nipple trauma or not, should be preventable with good care. It also suggests that when they do develop the first line of treatment should be to examine and advise on positioning and attachment (see also Section 3.4). Demonstrating this in experimental studies is limited, however, by the widespread lack of experience of health professionals and of women with good positioning and attachment, confounding the studies with the problems related to the skills of the health professionals involved, and also with the teaching techniques used. Painful breastfeeding is more commonplace in societies where prolonged exclusive breastfeeding is achieved only by a minority of women and where breastfeeding in public is rare. This is explained by the difficulties faced by new mothers resulting from lack of experience in observing other women breastfeed. Henderson et al.s (2001) [49] study, set in Australia, suggests that health professionals teaching positioning and attachment may result in fewer women breastfeeding, but this is in contrast to the Swedish study identified in Renfrew et al. (2000). It is likely that such teaching will vary in quality and effectiveness and this needs to be tested in a much larger sample of women to detect a larger than 20% difference in breastfeeding at six weeks, and with a range of ways of teaching these skills, which can be difficult for mothers to learn, and challenging for health professionals to teach. The three studies on care of the nipple in the prevention of sore nipples included 485 women. Two of the three studies (Centuori, 1999 [47]; Pugh et al., 1996 [55]) suggest that no one strategy has advantages over others in the prevention of sore and damaged nipples. The third study was not of sufficiently good quality. No studies have found convincing evidence that topical agents will help to prevent nipple pain and it is possible that such practices may detract attention from helping the mother to achieve a pain-free feed through correct positioning and attachment.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

87

Future research prevention of sore nipples Studies are needed to examine the best ways to enable women to avoid painful and damaged nipples in the early days of breastfeeding, especially in countries where breastfeeding is not commonly observed in public. This will include examining ways of preparing health professionals to help women avoid this painful problem, including the teaching of positioning and attachment, and assessment of other possible causes of sore nipples (see Sections 3.4, 3.5, and 4).

Discussion and conclusions It is likely that the reason no trials have been conducted on this topic for the past 15 years is that it would now be considered unethical to restrict the timing or duration of breastfeeds. Mother and babies should have unrestricted contact and be encouraged to feed as and when they wish to do so, as there is no evidence that restrictions have any benefit, and they can cause harm. Future research frequency and duration of breastfeeding Important questions remain about the timing of breastfeeds in the early days after birth. There is concern, for example, that leaving a sleeping baby undisturbed for lengthy periods in the early days after birth will result in hypoglycaemia (see Section 3.4; Hawdon et al., 1993; Anderson et al., 1993). As a result, policies often stipulate that babies be woken after set periods and fed to prevent this problem. Case reports suggest that sleeping for several hours may be quite normal (Olmsted and Jackson, 1950), and under natural circumstances babies may feed as little as once or twice in the first 48 hours (Yamauchi and Yamanouchi, 1990). Caregivers in the UK and elsewhere (eg Australia) are now leaving babies for very long periods between feeds in the early days, but further research is needed to identify issues related to the effects of pain relief given to the mother in labour and other reasons why healthy term babies may be reluctant to wake and feed.

3.3.2.4 Frequency and duration of breastfeeding


Background Throughout the last century, until the late 1980s, it was very common for babies to be fed by the clock, rather than according to their own need to feed (Fisher, 1986; Woolridge, 1996). Feeds were limited in both frequency and duration. Although this practice has been changing (Garforth and Garcia, 1989; White et al., 1992; Foster et al., 1997) many women and caregivers still seem to be uneasy if the baby wishes to be fed much more, or much less often, than six or seven times a day, or if the baby stays on the breast for five minutes or less, or for 30 minutes or more at each feed. Limiting the time on the breast is likely to have adverse consequences, both for the supply and demand mechanism of breastfeeding and for ensuring a balanced intake of both foremilk and hindmilk. Evidence base This review identified no trials of this topic. Findings from previous reviews Renfrew et al. (2000) reported findings from three RCTs (Illingworth et al., 1952: Salariya et al., 1978: Gale et al., 1989) and several non-randomised studies. They concluded that although some studies had methodological weaknesses there is substantial evidence that the timing and duration of breastfeeds should be responsive to the needs of the baby.

3.3.2.5 Additional fluids for breastfed babies


Background Descriptive studies of breastmilk intake in the first few days of life indicate that babies receive gradually increasing amounts of colostrum and milk when allowed to feed according to their own needs (Casey et al., 1986; Saint et al., 1984; Houston et al., 1983). Infants are able to regulate their own intake of breastmilk in terms of both quantity and quality. Even in hot climates they do not require additional fluids as they can regulate their feeding accordingly (Sachdev et al., 1991; Brown et al., 1986; Goldberg and Adams, 1983; Amroth, 1978). The routine feeding of newborn infants with sugar solution, water or other fluids used to be common practice in the UK and all over the world. Its practice in

88

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

the UK has declined in recent years but the situation is still far from resolved. Critics have long argued that giving infants additional feeds interferes with the supply and demand mechanism of milk production and therefore reduces milk supply (Daly et al., 1993; McNeilly, 1977); that it interferes with the development of normal immunological mechanisms, thus compromising one of the primary benefits of breastfeeding (Matthew et al., 1977); that it can result in the development of allergic conditions in some babies (Halken et al., 1992); that it can cause reactive hypoglycaemia (Heck and Erenberg, 1987); and that it causes nipple confusion in the baby who may come to prefer feeding from a bottle rather than from the breast (Neifert et al., 1995; see below Methods of giving supplementary fluids). It may also reduce the mothers confidence in her ability to breastfeed her baby without help (Neifert and Seacat, 1987) and its use by hospital staff to settle the crying infant can establish a precedent which may become a model for future management of the unsettled infant when the mother is discharged to home. The UNICEF Ten Steps to Successful Breastfeeding advise restriction of any fluids in the neonatal period. Step 6 advises: Give newborns no food and drink unless medically indicated. Findings We include three RCTs providing evidence of the impact of supplementation in the neonatal period on the duration of breastfeeding. Supplements were given if medically indicated, if mothers requested them or if the infant was thought to be anxious after a breastfeed. Howard et al. (2003) [51] investigated the effect of an artificial teat (nipple is the term used in the US) on the duration of breastfeeding in a sample of 700 breastfed newborns in the US. Non-breastmilk supplements were offered to newborn infants in all four groups while in hospital if there were medical indications, or if mothers requested them. Two groups received supplemental feeds through artificial teat and bottle, the other two from a cup, and 481/700 infants participated in the supplemental feeding intervention. One third of supplements were given for medical reasons (hypoglycaemia or weight loss 10%) and half were given as a result of maternal request. No reason was assigned for the remaining participants. At discharge from hospital, infants in each group (those who received supplements and those who did not) were assigned to either early (two to five days) or late (one month or more)

pacifier use. This study showed that supplemental feeds, regardless of whether they were given by cup or bottle, had a detrimental effect on exclusive, full and overall breastfeeding (p<.0001). There was evidence of modification of effect based on the number of supplements. When three or fewer supplements were given there was no significant effect on breastfeeding duration. However, when more than three supplements were given, cup feeding significantly improved the duration of exclusive (p<.001) and full (p<.0002) breastfeeding. Babies who were delivered by caesarean section were more likely to receive two or more supplements. Among this group those fed by bottle tended to be given larger amounts than those fed by cup. We can have confidence in the findings from this large, good quality trial. Kind et al. (2000) [52] and Schubiger et al. (1997) [57] reported on a multi-centre RCT of 602 infants in Switzerland examining the effect of providing supplementary feeds to breastfed infants in the first five days of life on later breastfeeding success. The provision of fluid supplements was restricted in the intervention group (UNICEF group). Infants were given supplements in the form of 10% dextrose-maltose (DM) solution or formula if medically indicated and, when given, either a cup or a spoon was used, not a bottle and teat. Pacifiers were withheld from babies in the intervention group. There was no similar restriction in the standard care group, who were given supplements and pacifiers as and when desired by the mother. If supplements were given to the standard care group infants a bottle and teat was used. Almost all infants in both groups received one or two supplements. There were only small differences between the two groups in the number of DM supplements during the first two days of life. Results showed that there were no significant differences between groups in the prevalence of breastfeeding at five days and two, four or six months after birth. The high proportion (39%) of protocol violations in the intervention group suggests a failure to effectively test the hypothesis. Other features reduce the internal validity of the study. For example, women in the same ward were assigned to either group, leading to contamination. Martin-Calama et al. (1997) [54] conducted an RCT on 180 full-term healthy newborns in Spain. The intervention group (GW) received glucose water from a bottle and 89

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

teat during the first three days of life if the baby became anxious after the breast was offered, while the control group (NGW) received no supplements. At 16 weeks more infants in the NGW group were breastfeeding but this difference disappeared by the end of 20 weeks. Breastfeeding duration was greater in the NGW group (p<.01). This study also found that at one month significantly more infants in the GW group had received formula. The findings from this study need to be interpreted with caution because authors did not state the method of randomisation and whether the outcome assessment was blinded. The analysis was not by intention-to-treat. Findings from previous reviews Renfrew et al. (2000) synthesised evidence on supplements from two RCTs and two non-RCTs. Overall the evidence pointed to a reduction in the duration of any and exclusive breastfeeding as a result of supplementation, although it was noted that the quality of the studies was not high. Discussion and conclusions Our review considers evidence from three further RCTs published after 1995. A total of 1,482 infants participated in these trials. Howard et al. (2003) [51] found that supplementing infants while in hospital had a detrimental effect on exclusive, full and any breastfeeding, with effects being greater if more than three supplements were given. Kind et al. (2000) [52] found that supplementing while in hospital made no difference to breastfeeding at two, four and six months. It must, however, be noted that there was only a small difference in the number of infants in both groups receiving supplements in the first two days. MartinCalama et al. (1997) [54] found that infants who did not receive glucose water supplements breastfed for longer than those who did, although we cannot place full confidence in these findings. Overall the evidence points to providing supplements in the neonatal period only when there are sound medical indications, and that duration of breastfeeding is less likely to be affected if only a small number (two to three) of supplements are given in the first five days.

Future research additional fluids for breastfed babies There are several gaps in existing knowledge related to supplementing in the neonatal period. The following questions need to be addressed: At what levels of plasma glucose, body temperature and weight loss does fluid supplementation in the neonatal period become essential? Are there differences in duration outcomes related to supplementing in settings where culture and attitudes to breastfeeding are different (for example, UK vs Switzerland vs Sweden), or for women from cultures where breastfeeding is not the norm? What should be the ideal constitution of supplementary fluids when they are medically indicated?

3.3.2.6 Methods of giving supplementary fluids


Background There has been debate in recent years about the possibility that feeding infants by bottle in addition to or instead of the breast may give them an artificial sucking experience which is counter-productive to the successful establishment of breastfeeding. The debate arose from the recognition that infants used different sucking techniques while breast and bottlefeeding (BuLock et al., 1990). Evidence from good quality research is absent, however, for the syndrome which has come to be described as nipple confusion (Neifert et al., 1995). Neifert suggests that it would seem prudent for caregivers to identify newborns at risk of nipple confusion and to minimise the use of bottlefeeding in such babies. There is no evidence available as yet to assist in such identification. The use of cup feeding as an alternative to bottlefeeding for babies who require additional fluids has been increasing in recent years in an effort to avoid nipple confusion. Descriptive studies of the effect of this practice have been carried out in neonatal and transitional care units with babies who are small or sick. Spoons and droppers are also often used to feed newborn infants.

90

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Findings We identified two RCTs examining the impact on breastfeeding duration of different methods of feeding supplements to healthy term infants. Howard et al. (2003) [51], described in the earlier subsection, compared the delivery of supplementary fluids by cup and by bottle on breastfeeding duration. Among women who delivered by caesarean section cup feeding of supplements while in hospital in the immediate postpartum period prolonged exclusive breastfeeding by approximately 10 days, full breastfeeding by five weeks and overall breastfeeding by 10 weeks. There was no similar benefit observed in those delivered vaginally. It is notable that the caesarean infants received more supplements (5.94.6) than those who were delivered vaginally (3.02.4); and were more likely to get them for medical reasons. Authors of this paper hypothesise that the larger volume of fluid given by bottle, compared to the smaller volume given by cup, may have had a detrimental effect on breastfeeding duration outcomes, suggesting that aside from the delivery method the volume of fluid supplemented in the early postpartum period may be one of the factors responsible for differences in breastfeeding duration. Cup or bottle assignment did not affect the type of supplement (expressed breastmilk or formula) consumed by those infants who continued to receive supplements after discharge from hospital. Schubiger et al. (1997) [57] first reported on the Swiss multi-centre trial of supplementary fluids and artificial nipples, described later by Kind et al. (2000) [52]. They examine the impact of providing fluid supplements through artificial teats/bottles, and the effect of pacifiers, on sucking and breastfeeding duration. Supplements were mainly 10% dextrose-maltose (DM) solution. A small number were given formula if lactogenesis failed. The intervention (UNICEF) group were given supplements if indicated by spoon or cup, while the standard care group were given supplements by bottle and teat. Pacifiers were withheld from the intervention group. In the intervention group cup/spoon feeding proved to be too difficult in 9.5% cases and in a further 23.8% cases the mother insisted on using a pacifier. This study found that the number of infants receiving supplements in both groups was not statistically significant, although the number of DM supplements was significantly lower in the

intervention group. There were no differences in sucking behaviour or breastfeeding duration at five days and two, four and six months between the intervention (cup group) and the standard care (bottle/teat) group. This study must be interpreted with caution because of the large number of protocol violations and the inadequate description of the study protocol prior to randomisation. Findings from previous reviews Renfrew et al. (2000) identified only one trial, the Schubiger et al. (1997) [57] trial identified here. Discussion and conclusions The two studies included in this review involve a total of 1,302 infants. The only conclusion that we can confidently draw from existing evidence is that cup feeding may have better breastfeeding duration outcomes than bottlefeeding for babies delivered by caesarean section. Howard et al. (2003) [51]) suggest that it may be the larger volume of fluid intake through bottles that had a negative effect on breastfeeding in this sub-set of babies, rather than the delivery method. More research is needed to build on this evidence. There is limited evidence in the existing literature to describe nipple confusion, its prevalence or its management in healthy term babies. Some infants who are successfully breastfeeding refuse bottles in the same way as some infants who have been exposed to the bottle and teat refuse the breast, suggesting a preference of one type of nipple/teat over the other. Woolridge (1996) has coined the term acquired teat preference to explain this tendency. Future research methods of giving additional fluids Further research is needed to describe and explain nipple preference and acquired teat preference. We need to know more about the characteristics of babies who need supplements during the neonatal period and under what conditions cup or bottle and teat are preferable.

3.3.2.7 Pacifier use


Background Pacifiers are used across the world to settle, soothe or otherwise occupy a fretful or distressed infant. The Avon 91

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Longitudinal Study of Pregnancy and Childhood found that nearly 60% of the 10,950 infants whose mothers provided the relevant data had used a pacifier by four weeks of age (North et al., 1999). More bottlefed than breastfed infants use pacifiers. For example, in New Zealand, 9% of breastfed babies use a pacifier compared with 19% of bottlefed babies (Mitchell et al., 1993). The use of pacifiers has been implicated in an increased risk of otitis media (Uhari et al., 1996; Watase et al., 1998), oral candidiasis (Darwazeh and al-Bashir, 1995), poor intellectual attainment (Gale and Martyn, 1996; Barros et al., 1997), poor jaw muscle activity (Sakashita et al., 1996) and abnormal jaw development (Ogaard et al., 1994) and an earlier return of menstruation in the mother (Ingram et al., 2004). Pacifier use has been implicated in shortening the duration of breastfeeding and in early weaning. The first epidemiological studies on pacifier use and breastfeeding duration were conducted less than 10 years ago. Overall, the literature suggests that pacifier use may be more the result of breastfeeding difficulties than the cause of early weaning (Victora et al., 1997). Pacifiers have less of an impact on babies whose mothers are more confident about nursing and in cultures where breastfeeding is the norm. They are also used as a weaning aid by some women who want to wean their babies. Step 9 of the Ten Steps to Successful Breastfeeding (see Appendix 7) advises restriction of artificial teats and pacifiers. One of the criteria for acquiring the Baby Friendly certificate is the restriction of pacifier use on maternity units. This policy requires healthcare staff to avoid giving pacifiers to individual women and for this to be mandated within the hospital policy. Findings We include results from three RCTs of the impact of pacifier use on the duration of breastfeeding. Howard et al. (2003) [51], described above, investigated the impact of pacifiers and teats on the duration of breastfeeding in a sample of 700 breastfed newborns. This study found that pacifier use at four weeks caused a significant decline in exclusive breastfeeding among those exposed to pacifiers, but did not affect overall and full breastfeeding. Early (two to five days) compared to late (more than four weeks) pacifier use significantly shortened the total duration of breastfeeding (p<.02), but 92

not exclusive or full breastfeeding, when adjusted for supplemental feeds and method of giving supplements. Analysis for first-time mothers showed a stronger effect of early pacifier use on exclusive breastfeeding in the first six months (p<.004). The second study, Kramer et al. (2001a) [53], looked at the effect of regular pacifier use on weaning at three months postpartum. A total of 258 mother-infant pairs were randomised either to the intervention group, which included teaching mothers different strategies to soothe and calm their babies while avoiding the use of pacifiers, or to the control group, in which calming strategies included the use of pacifiers. During the study period a high proportion (<60%) of infants in the intervention group were given a pacifier, and not all infants in the control group used a pacifier. Mean pacifier insertions per day were lower in the intervention group. Experimental analyses showed that the use of pacifiers did not increase the risk of weaning at three months and had no effect on crying and fussing behaviour. However, when randomisation was ignored, and data were analysed as if it were an observational study, and a strong association was noted between pacifier use and early weaning. The large number of protocol violations (62.4% in the intervention group and 16% in the control group) reduces the internal validity of this study. The authors suggest that pacifier use may be a marker of breastfeeding difficulties or reduced motivation to breastfeed, rather than a true cause of early weaning. The third study is the Swiss multi-centre trial of supplementary fluids and pacifier use conducted by Schubiger et al. (1997) [57] (see also Kind et al., 2000 [52]). A total of 602 healthy term infants born in maternity units across the country were assigned to either the standard care group, in which pacifiers were permitted at any time and supplements of 10% DM solution were given by bottle after breastfeeding, or the intervention group, following UNICEF guidelines in which pacifiers were avoided and supplements were given only if medically indicated by cup or spoon. Due to the large number of protocol violations pacifier use in both groups was not significantly different at two and four months. There were no differences between intervention and control groups in percentages of fully or partially breastfed infants. Pacifier use had no influence on sucking behaviour during the first five days of life, or on breastfeeding prevalence and duration in the first six

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

months of life. We cannot have confidence in the findings from this study due to the large number of cross-overs. Findings from previous reviews Renfrew et al., 2000 identified four studies, none of them RCTs, all of which demonstrated a negative association between pacifier use and breastfeeding duration. Discussion and conclusions Our review brings together evidence from three new RCTs on a total of 1,560 infants. Schubiger et al. (1997) [57] suggest that pacifier use has no impact on the duration of breastfeeding in the first six months, although the study has poor internal validity. Howard et al. (2003) [51] demonstrated that when figures were adjusted for supplements vs no supplements and cup vs bottle, there were no differences in the duration of full and exclusive breastfeeding between infants who were exposed or not exposed to pacifier use. In the observational analysis of their data, Kramer et al. (2001a) [53] found that pacifier use was associated with higher prevalence of weaning at three months, but experimental analysis failed to confirm this. Renfrew et al. (2000) reported that current studies could not resolve the issue of whether pacifier use was a cause of breastfeeding problems or a marker of existing problems, as mothers resorted to their use as part of an attempt to overcome a problem. Overall, existing evidence points to no effect of pacifier use on breastfeeding duration, although early pacifier use, and use by first-time mothers, has been associated with more negative outcomes, suggesting that there may be circumstances in which breastfeeding is vulnerable to the use of pacifiers. These results need to be interpreted in the light of cultural circumstances in which the trials were conducted, and uncertainties about the validity of one of the three studies. Switzerland has a high prevalence of breastfeeding and a culture that supports breastfeeding, while this is not the case with the US, Canada or the UK. The prevalence of pacifier use may also be cultural. Future research pacifier use Further research is needed to add strength to the evidence from the three RCTs reported in this section to

come to a clear conclusion about the role of early pacifier use in shortening the duration of breastfeeding, to study differences in the effect of pacifier use at different times, for different groups of women, and in different cultures; and to determine if pacifier use can be used as a marker for breastfeeding difficulties or a low motivation to breastfeed.

3.3.2.8 Breast pumping to establish breastfeeding in women at risk of delayed lactation


Background Risk factors for delayed onset of lactation include unscheduled caesarean section, prolonged second stage of labour and maternal obesity (Chapman and PerezEscamilla, 1999). Delayed onset of lactation is linked to a shortened duration of breastfeeding among women who plan to breastfeed for six months or more. Considering the large numbers of women who undergo caesarean section (RCOG, 2001) in the UK, strategies to hasten the onset of lactation and lengthen the duration of breastfeeding are important. Breast pumping is often advised in women at risk of delayed lactogenesis. Findings We include one trial of breast pumping. Chapman et al. (2001) [48] studied the effectiveness of breast pumping before the onset of lactation after caesarean birth. Sixty women who had undergone caesarean section and had healthy term infants were randomised to receive double electric breast pumping for six sessions lasting 10-15 minutes from 24 to 72 hours postpartum, or to the control group in which no active pumping took place. Findings from this study indicate that the intervention group tended to have lower milk transfer than the controls. The median duration of breastfeeding was 5.9 months in the intervention group and 6.7 in the control group. The difference did not reach statistical significance (p=.76). The difference in breastfeeding duration between primiparae in the intervention and control groups was approximately five months, but this again did not reach statistical significance numbers in the groups were very small. Sixteen primiparas would have been needed in each group to demonstrate an adverse effect of pumping. There were no significant differences in median duration of exclusive breastfeeding or in 93

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

percentage of women still breastfeeding at the time of data analysis (1.09 0.42 years), although again, numbers were very small. Findings from previous reviews No previous reviews have examined this issue. Discussion and conclusions

group; perceived pain in mothers in the intervention group was lower, from three hours to two days after the delivery. Breastfeeding duration outcomes were not reported. The internal validity of this study remains in question because the authors do not state if the sample size calculation was a priori, if the method of randomisation was valid and if outcome assessment was blinded. Findings from previous reviews

Breast pumping before the onset of lactation after caesarean delivery of a healthy term infant has not been shown in this one small trial to have any beneficial effects and it may increase the negative short-term effect of decreasing milk transfer in the first 72 hours. Future research breast pumping to establish breastfeeding in women at risk of delayed lactation Research is needed to find out the hormonal impact of breast pumping on breastmilk transfer in the first 72 hours after caesarean and vaginal delivery, and the optimal time for starting breast pumping in women at risk of delayed onset of lactation.

This topic has not been examined in previous reviews. Discussion One fifth of births in the UK are by caesarean section and breastfeeding rates in the country are low. Strategies to help women undergoing caesarean section to breastfeed for a longer duration need to be developed. While Hirose and colleagues study [50] described in this section indicates higher milk intake in newborns of mothers receiving continuous epidural analgesia, there is no definite indication of longer breastfeeding duration. Pain control is an important factor in the establishment of breastfeeding not only in women undergoing caesarean section but also in the case of women who have had other interventions resulting in pain in the immediate postpartum period. Future research postoperative analgesia after caesarean section Further research is needed to examine the long-term effectiveness of continuous epidural analgesia, and other forms of pain relief, on breastfeeding duration.

3.3.2.9 Postoperative analgesia after caesarean section


Background There is some evidence that the presence of pain suppresses breastmilk production in the immediate postpartum period. Narcotics and sedatives can have an effect on newborn infants and interfere with the successful establishment of breastfeeding. Findings We found one study examining the effect of postoperative continuous epidural analgesia on breastfeeding and infant weight gain. Hirose et al. (1996) [50] randomised 30 women who had undergone elective caesarean section under spinal anaesthesia to the intervention, which consisted of continuous epidural analgesia with 0.25% bupivacaine for three days after birth, or to the control group who did not receive such ongoing analgesia. Breastfeeding started after 2.5 0.2 days in the intervention group compared to 2.4 0.2 days in the control group. Breastmilk intake in g/day after withdrawal of the epidural was higher in the intervention 94

3.3.2.10 Growth, weight and food charts


Background No studies were identified on the use of growth, weight or food charts for breastfed babies, either in hospital or in community use. This is despite the fact that growth charts form the basis of much of the decision making in both hospital and community, and are often the focus of attention for both professionals and parents. There has been increasing recognition in recent years that growth charts currently used in community practice are inappropriate for breastfed babies, as they are derived

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

from cohorts of babies who were bottlefed or introduced to food earlier than recommended. Current charts may suggest to professionals and parents that exclusively breastfed babies falter in their growth at some point after eight weeks (Cole et al., 2002), and it is possible therefore that they underlie at least some of the anxiety that women feel about the sufficiency of their milk supply. Observational evidence derived from a group of babies included in an RCT conducted in Belarus (Kramer et al., 2001b) suggests that prolonged and exclusive breastfeeding accelerates weight and length gain in the first few months. In practice, a breastfed baby could therefore appear to gain weight unexpectedly fast using current charts, and some health professionals may suggest inappropriately that breastfeeds should be reduced. As the slow down in weight then happens around three to five months, as described by Kramer et al. (2001b), some health professionals may suggest that breastmilk is no longer adequate and that weaning is required. Such advice could be avoided if accurate charts were used. Weight charts based on a sample of 120 British long-term breastfed babies have been developed by the Child Growth Foundation (Fry, 2003). The World Health Organization is also conducting a multi-country study to develop new international growth charts for breastfed babies. There remains an evidence gap on the ways in which such charts could be used to support, or to undermine, breastfeeding duration.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

95

96 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Warm wet compresses/ Start on day 1 (not clear expressed breastmilk/water and when they stopped) soap to prevent sore nipples No breastfeeding outcomes reported Cracked/sore nipples on days 1-10 At discharge, 2 weeks and 3 months postpartum Timing of breastfeeding duration outcomes 90 healthy primiparae 219 healthy mothers of healthy term infants (64% in intervention group and 62% in control group were educated to secondary school level or university) 57 women delivering healthy term infants by caesarean section Breast pumping before the onset of lactation for women at risk for delayed lactation (example: caesarean section) Day 1 ~24 hours-72 hours (total 2 days) No topical application vs From day 1 while in hospital routine care involving cleaning and application of liniments/ointments to prevent sore nipples At 6 months postpartum and every 4 months until ~1y and 1m Positioning and attachment 160 healthy primiparae and their healthy newborns (most education for women participant mothers had public health insurance, only 16% had university education) 30 women delivering healthy term infants by elective caesarean section under spinal anaesthesia 6 weeks, 3 and 6 months Continuous epidural analgesia Immediately after delivery to 3 days postpartum No breastfeeding outcomes Milk intake, total weight of infant feeding, infant weight gain, dose of diclofenac and pain scores from Day 1-12 2, 5 10, 16, 24, 38 and 52 weeks 700 healthy mothers intending Early and late pacifier use and 2 groups at 2-5 days two methods of giving 2 groups after week 4 to breastfeed their healthy singleton newborns delivered supplements by caesarean section or vaginally (14.5% participants received Federal Assistance)

Table 3.3: Included studies prevention of breastfeeding problems

Name of authors and study number

Date of Country publication

Design

Akkuzu and Taskin [46]

2000

Turkey

RCT

Centuori et al. [47]

1999

Italy

RCT

Chapman et al. [48]

2001

US

RCT

Henderson et al. [49]

2001

Australia

RCT

Hirose et al. [50]

1996

Japan

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Evidence briefing 1st edition February 2005

Howard et al. [51]

2003

US

RCT

Table 3.3: Included studies prevention of breastfeeding problems (cont.) Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Supplementing fluids and pacifier use From immediately after birth Day 5 2, 4 and 6 months Timing of breastfeeding duration outcomes

Name of authors and study number 602 healthy mothers and their healthy mothers delivering by caesarean section or vaginally 258 healthy mothers and their healthy term infants delivered by caesarean section or vaginally 170 healthy newborns Giving glucose water supplements Day 1-day 3 Warm wet compresses/ expressed breastmilk/keeping nipples clean and dry to prevent sore nipples Day 1 Withholding pacifier use Immediately after birth, and repeated at 10 days and 3 weeks 3 months

Date of Country publication

Design

Kind et al. [52]

2000

Switzerland Multi-centre RCT

Kramer et al. [53]

2001a

Canada

RCT

Martin-Calama 1997 et al. [54] 176 healthy primiparae with healthy term infants (mean family income <US $40,000, mean years in education 13.9, 1.8% non-white groups) 196 healthy term newborns and their mothers (> 50% households had a monthly income level with the minimum wage 162 baht/day at 2000 rates) 602 healthy mothers and their healthy term infants delivered by caesarean section or vaginally 58 healthy term newborns and their mothers Early Kangaroo care

Spain

RCT

Until 5 months

Pugh et al. [55]

1996

US

RCT

Breastfeeding at 6 weeks Pain intensity, pain effect and sucking strength on day 1, 4, 7 and 14

Punthmatharith 2001 [56]

Thailand

RCT

Starting at 30-90 minutes after birth until discharge from hospital

2 days and 1 month

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Schubiger et al. [57]

1997

Switzerland Multi-centre RCT

With-holding pacifiers and avoiding use of bottles for giving supplements

From immediately after birth

Day 5

Shiau [58]

1997

Taiwan

RCT

Kangaroo care

Starting at 4 hours after birth until discharge from hospital (day 3)

2, 4 and 6 months Days 1, 2 and 3

97

3.4 Clinical interventions: solving breastfeeding problems


Topic Effectiveness of treatments for breastfeeding problems including sore nipples, nipple infection, sore breasts, engorgement, mastitis, breast abscess, insufficient milk and lactation failure Summary of methods
Study design of included studies: RCTs only Dates for searches: 1995 to June 2003 Population group included: breastfeeding mothers of healthy babies Countries included: all Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, breastfeeding problems, views of those experiencing and providing care, relevant process outcomes including economic outcomes

3.4.1 Evidence base


Six studies met our inclusion criteria (see Table 3.4). Four trials, all RCTs, examined the effectiveness of different applications for the treatment of sore and/or infected nipples including tea bags, cabbage leaves, liniments and ointments, systemic antibiotics and the avoidance of any application on the sore nipple. Two RCTs tested the effectiveness of cabbage leaves in the treatment of breast engorgement. None of the included studies looked specifically at the needs of women from disadvantaged groups; both Australian studies specifically excluded aboriginal women.

of this problem (Hamlyn et al., 2002). It is not known how many women continued to breastfeed while experiencing pain. Related descriptive and theoretical work was described in Section 3.3; this suggested that a fundamental part of managing sore nipples is to ensure proper positioning and attachment of the baby. Sore nipples are a frequent problem among breastfeeding mothers. The most common cause for nipple soreness is incorrect positioning and attachment (see Section 3.3). Other causes include: eczema, impetigo, dermatitis, candida (thrush) infection, Raynauds phenomenon of the nipple and ankyloglossia in breastfeeding infants. There may or may not be clinically verifiable signs in all mothers presenting with complaints of nipple soreness. If the epidermal integrity of the skin is broken as a result of trauma the resulting fissure (crack), bleeding or crusted lesion can be clinically ascertained. Classical signs of candida infection, Raynauds phenomenon and other conditions are sometimes absent or ill-defined, making the precise cause of soreness difficult to pinpoint. It is notable that in the research literature no distinction is made between types of sore and damaged nipples, despite an early study (Gunther, 1945) having examined the different types of nipple damage in detail.

3.4.2 Findings 3.4.2.1 Painful breasts and nipples


Background Painful breasts and nipples are a leading cause of discontinuation of breastfeeding. In the most recent UK survey of infant feeding, 55% of women who stopped breastfeeding prematurely reported that they discontinued in the first two weeks after birth as a result 98

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Sore nipples are likely to reduce the frequency and length of breastfeeds and may result in inadequate emptying of the breast. A number of strategies have been tried in the treatment of sore nipples, of which the application of breastmilk is one. Breastmilk contains antibodies, antiinflammatory cells and other anti-inflammatory factors such as lactoferrin and bifidus factor, which are known to facilitate healing. Findings We identified three studies of topical agents to treat sore nipples. While none of them measure primary outcomes important to this review, these three studies measure secondary outcomes: improvement in nipple pain and/or healing of nipple lesions. Kuscu (2002) [60] compared the effectiveness of topical collagenase, topical dexpanthenol and washing with warm water and soap. A total of 66 mothers in a hospital in Turkey were randomised to receive any of the three regimes and used them four times daily after breastfeeding until the second week after birth. Breastfeeding duration was not reported. Mean scores for nipple pain were similar for all groups at all points (days one and three, weeks one and two except on day three when there was less pain in the group receiving collagenase. The authors did not clarify inclusion and exclusion criteria, the method of randomisation and whether the analysis was by intention-to-treat. We cannot therefore have confidence in the validity of the findings. Brent et al. (1998) [59] compared the effect of hydrogel moist dressings with standard care. They randomised 42 women in the US to the intervention group, which applied non-adhesive hydrogel dressings to sore nipples after the application of expressed breastmilk and airdrying, or the control group, which received standard care starting with the application of expressed breastmilk to the nipple after each feed and air drying, followed by application of lanolin cream and the use of breast shells. Hydrogel dressings were changed after each feed. Both groups received breastfeeding instruction from the same lactation consultant. There was greater improvement in the lanolin and breast shell group, with differences in physician confirmed healing (p<0.1) and self-reported pain (p<0.5) reaching statistical significance. More nipple infections were reported in the hydrogel dressing group.

Two problems make this study difficult to interpret. First, the study was designed to have 40 women in each arm. Due to the high incidence of infections in the treatment group researchers stopped recruiting women and performed an interim analysis, leaving a sample of only 21 in each arm. Second, there was lack of a no-treatment control group. It cannot be said with any degree of confidence whether it was the lanolin cream, the breast shells or the extra attention received, or indeed a combination of the three, that led to greater improvements and fewer infections in the standard care group. Moreover, we have no information from the paper about the degree of soreness and injury to the nipples. It is likely that hydrogel dressings may prove to be of help in cases of severely ulcerated nipples. Lavergne (1997) [61] studied the effectiveness of teabags and warm water on sore nipples. Nipples (rather than women) of 65 women were randomised to receive either tea bags, warm water compresses or no treatment, resulting in three different treatment groups. Tea bags and warm water were found to be equally effective in the early postpartum period compared with no treatment. However, both the study design and the recurrent selfrating of pain at each feed are likely to show a (spurious) effect of treatment over no treatment. Findings from previous reviews Renfrew et al. (2000) found studies examining the use of nipple shields, and breast shells for the treatment of sore nipples. Three studies of nipple shields demonstrated that the use of nipple shields had no beneficial effect on milk transfer, milk volume or breastfeeding duration (Auerbach, 1990; Nicholson, 1985; Woolridge et al., 1980). The review also found one study (Gosha and Tichy, 1988) on breast shells which concluded that these are not effective in the treatment of sore nipples. Discussion Sore nipples are a leading cause of breastfeeding discontinuation internationally and result in distress and pain for women who choose to continue to breastfeed. The lack of high quality, theoretically based, relevant and useful research in this field is striking. Historically, many treatments have been tried for the treatment of painful nipples including nipple shields, breast shells, topical applications of a variety of ointments 99

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

and liniments and warm wet compresses. No one treatment has been shown to be effective; hydrogel dressings may increase infection and it is likely that use of devices such as nipple shields may interfere with the care needed to ensure that the baby has a good mouthful of breast tissue so that feeding does not hurt and nipple damage is avoided. The use of breastmilk applied to the nipples, lanolin and breast shells needs further investigation. The first line of treatment for sore nipples must remain the removal of the most common cause incorrect positioning and attachment which results in direct friction to the nipple (Section 3.3). There is little evidence to inform practice in relation to the range of other possible causes of sore nipples, including eczema, impetigo, dermatitis, tongue tie/ankyloglossia in the baby, or even the contact sensation of a well-positioned baby feeding at the breast, which individual women may perceive as pain. Practices based on clinical experience and a theoretical analysis of the problem have been developed. These include the use of anti-fungal preparations for candida and steroid preparations for dermatitis. But identification of women most at risk of developing these problems, accurate diagnosis and the best regimes for treatment are based on very little information. One ongoing study on candida is listed in Appendix 3. Future research sore nipples Much is yet to be learned about the aetiology of sore nipples, as well as their treatment. Before designing new RCTs it is essential to conduct descriptive studies to examine questions, including a proper descriptive account of the types of sore nipples, to accurately distinguish between different types of damage and pain. This would serve as the basis for future work to distinguish between pain caused by incorrect positioning and attachment, infection, dermatitis, ankyloglossia, etc. Specific questions could then be addressed, to include: How often are sore nipples associated with signs of infection (see below) or problems in the baby such as tongue tie? In such cases, simple treatments such as ointments are unlikely to be effective without addressing the underlying cause of the pain How many women with sore nipples are not helped by correcting attachment and positioning techniques to ensure the baby is not damaging the nipple by friction? If the baby is not effectively attached 100

treatments such as ointments are unlikely to be effective without care to address the positioning problem. Studies of wound healing from other fields may also yield helpful insights into treatments that may be effective. RCTs can then be designed to assess the effectiveness of appropriate treatments used in situations where they may be of use.

3.4.2.2 Treatment of nipple infection


Background Natural barriers to infection are inherent in the normal anatomy of the stratum corneum of the epidermis, rapid cell turn over and low pH (5-6). Breaks in the integrity of the skin predispose it to bacterial and other infection and delay wound healing. Infection of the nipple can cause irritation and inflammation with symptoms ranging from severe pain to itching and burning. Such infections are not always identified. We found one trial of treatments for infected nipples. Findings One study of treatments for nipple infection was identified. Livingstone and Stringer (1999) [62] randomised 84 women with cracked, fissured or ulcerated nipples to receive breastfeeding instruction, topical mupirocin, topical fusidic acid or oral antibiotics. Eight per cent of mothers in the optimal treatment group showed improvement on self-assessment and physician review at five to seven days compared to 16% in the mupirocin group, 29% in the fusidic acid group and 79% in the oral antibiotic group (p<.0001). Mastitis developed in 5% of mothers treated with systemic antibiotics compared to 12-35% of those not treated with systemic antibiotics (p<.005). The trial was abandoned because of poor response in the groups not receiving systemic antibiotics and the high risk of developing mastitis. Systemic antibiotics were added to 46 of the 58 cases that had not improved. Due to the fact that the sample size was not achieved the findings from this study (that antibiotics are effective in the treatment of infected nipples) can at best provide an indication that antibiotics are effective in preventing mastitis among women with infected nipples.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Findings from previous reviews Renfrew et al. (2000) found no relevant trials but provided some evidence on the treatment of infected nipples from case studies and case series (Tanguay et al., 1994; Huggins and Billon, 1993; Amir and Pakula, 1991). These suggested that topical antibiotics and topical antifungals can be useful in treating infected nipples when correctly identified. Discussion Our search identified only one RCT (Livingstone and Stringer, 1999 [62]) which suggests that systemic antibiotics may be more effective in the treatment of infected nipples compared to lactation instruction and topical applications. Future research nipple Infection This important topic needs further study. Investigation of the prevalence of infected nipples is needed, to be followed by well-conducted RCTs to build on the work of Livingstone and Stringer (1999) [62].

treatment on chest circumference. Mothers in both groups felt that the creams were efficacious. The authors of the study stated that breastfeeding was more effective than creams in perceptions of discomfort and hardness of the breast. Roberts et al. (1995) [63] compared the effectiveness of chilled cabbage leaves with room temperature leaves in a RCT of 28 mothers. There were no differences in pain reduction between the groups. The lack of information about whether these two small-scale studies were analysed by intention-to-treat reduces confidence in the findings. Findings from previous reviews Renfrew et al. (2000) identified six studies of treatments for breast engorgement including ultrasound, cabbage leaves and proteolytic enzymes (Roberts, 1995; Roberts et al., 1995; Nikodem et al., 1993; McLachlan, 1991; Kee et al., 1989; Murata et al., 1965). That review found that oral proteolytic enzymes, which are no longer available in the UK, and the local application of cabbage leaves may help. A Cochrane review (Snowden et al., 2001) synthesised evidence of the effectiveness of treatments for breast engorgement during lactation from eight clinical trials published between 1952 and 1998; all studies since 1995 are included in our review. Discussion There is a striking paucity of research examining this condition, which is common and can be very painful. Untreated it may lead to mastitis and often early weaning. Engorgement can be prevented in some women by unlimited feeding by a baby who is well positioned at the breast and by encouraging the woman to give a prolonged feed from one breast. No information is available to help identify the numbers and types of women whose breasts may become engorged even with good preventive care. Our review included two small trials by Roberts and colleagues [63, 64] involving a total of 91 women, and found that cabbage leaf extract seemed to be effective in reducing pain arising from breast engorgement, but there was no advantage in using chilled cabbage leaves over cabbage leaves at room temperature. Women preferred cabbage leaves over gel packs. Together with the findings from previous reviews, these results require confirmation in larger studies.

3.4.2.3 Breast engorgement


Background Despite the gradual move away from rigid feeding regimes to more flexible patterns of feeding and the encouragement of frequent feeding, engorgement and milk retention remain an important cause of maternal discomfort and pain in the postpartum period. Practices that encourage effective milk flow, such as early breastfeeding, correct positioning and attachment and spontaneous breastfeeding assist in the prevention of engorgement. Findings We found three small-scale Australian RCTs of treatment for engorgement; all three tested the effectiveness of cabbage leaves applied to the engorged breast. Roberts et al. (1998) [64] tested the effectiveness of cabbage leaf extract against a placebo in a sample of 39 women. Pain and breast hardness were statistically significantly different (p<0.05) in favour of treatment with cabbage leaf extract. There was no effect of either

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

101

Future research breast engorgement There is an urgent need for further high quality, theoretically based research in this field. The following research questions were among those identified by Renfrew et al. (2000), yet still remain to be addressed: What is the current incidence of engorgement among breastfeeding women in the UK and what would the prevalence be if women were encouraged and supported in carrying out preventive measures? The treatment of symptoms is unlikely to work if effective milk removal measures such as early initiation and frequent and on-demand feeding are not instituted Does antenatal expression of colostrum help in the prevention of engorgement? Does the local application of cabbage leaves help and if so when and what are the best methods of using them? Is it possible to test further the effectiveness of oral proteolytic enzymes even though they are no longer available in the UK? Do the simple measures used in common practice help (singly or in combination)? These include: warmth, rest, breast massage and practical information and breast support, in the form of a well-fitting bra or breast binder analgesia warm baths/showers gentle expression. Many of the above questions could be effectively addressed in RCTs.

plugged ducts and mastitis. Factors implicated in the development of mastitis are missed feeds, irregular breastfeeding patterns, incorrect positioning or sucking, restrictive clothing and increased susceptibility to infections including yeast infection. Mothers returning to work also seem to be more prone to mastitis. Findings Our review found no recent trials testing the effectiveness of treatments for mastitis. Findings from previous reviews Renfrew et al. (2000) described the problem of mastitis in some detail. They identified one RCT (Thomsen et al., 1984), which examined three forms of mastitis distinguished by leucocyte count and quantitative bacterial cultivation: milk stasis, non-infective mastitis and infective mastitis. They found that all forms of mastitis improved with regular breast drainage, suggesting that breastfeeding should continue with good support to ensure effective drainage. Antibiotics were effective in the treatment of infective mastitis. Discussion Mastitis is an important condition which can make the mother feel very unwell and which can rarely develop into an abscess and even septicaemia. Its prevention and treatment are important. In spite of this, very little is known about the condition; there is no information, for example, on the prevalence of mastitis in the UK. Previous work suggests that treatment should include the continuation of breastfeeding with support to ensure effective breast drainage, and antibiotics for infective mastitis. Ibuprofen, an anti-inflammatory drug, has been used with benefit in cases of mastitis with or without signs of infection. One problem, however, is that there is no simple and quick test to distinguish infective from non-infective mastitis. Future research mastitis As outlined in Renfrew et al. (2000), there remain a number of unanswered questions, including: A need to determine both the incidence of mastitis in this country and its prevalence at different stages of lactation

3.4.2.4 Mastitis
Background The term mastitis may be used to refer to any inflammatory condition affecting the tissue of the breast, the presenting signs for which can range from simple focal inflammation, through development of a tender, hot, wedge-shaped segment of the breast, to involvement of the entire breast and advanced cellulitis. It can include the occurrence of systemic symptoms including pyrexia and those mimicking flu (shivers, rigor, fever, hot and cold flushes). It may, rarely, be related to the development of a breast abscess, which may in turn result in septicaemia. Any condition restricting free milk flow and conditions making the mother vulnerable to illness can cause 102

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Development of a simple differential diagnostic test to discriminate infective mastitis from obstructive (noninfective) mastitis in routine clinical practice Based on accurate diagnosis of alternative forms of mastitis, the development of a range of appropriate treatments Can mastitis be prevented by helping the mother to feed effectively from both breasts? Would the routine use of leucocyte and bacterial counts for women with mastitis be cost effective as part of a treatment protocol? Are there other factors which cause mastitis? For example, illness, stress, exhaustion and blocked ducts are associated with its occurrence, and nipple trauma is a possible predisposing factor. What are the mechanisms which might cause this, and could this be prevented? Studies are needed to explore further the association noted with the mothers preferred holding side (Inch and Fisher, 1995) An association has been noted between a family history of allergy and recurrent mastitis premenstrually (Minchin, 1985). Is this the case? Is there a similar association with mastitis during breastfeeding? If so, what are the mechanisms and can it be avoided?

which is not associated with engorgement or mastitis. It is essential, however, that future surveys of problems distinguish between painful nipples and painful breasts, as the underlying causes and treatments are likely to be different. Future research treatment of breast pain Important questions remain to be addressed in relation to breast pain. These include: When do women develop breast pain in the absence of sore nipples, and in what circumstances are sore nipples and painful breasts related? How many women experience deep breast pain during and after feeds? Such descriptive information is essential before developing trials of treatments.

3.4.2.6 Milk insufficiency


Background One of the most problematic areas in the establishment and maintenance of breastfeeding is perceived insufficiency of breastmilk. Women from all parts of the world give insufficient breastmilk as the most common reason for not continuing breastfeeding (World Health Organization, 1981a; Akre, 1989; White et al., 1992; Foster et al., 1997) including 66% of women in the UK who discontinue in the first two weeks after birth (Hamlyn et al., 2002). Not enough is currently known about actual milk insufficiency and if there is a gap between perceived and actual milk insufficiency. Women and their caregivers may consider the condition of milk insufficiency when breastfeeding babies cry after a feed, continue feeding for long periods of time, change their sleep patterns, gain weight slowly or lose weight. Infrequency or changes in the babys stools may also be a cause for concern. A good understanding of the aetiology of milk insufficiency is an important step in identifying interventions to resolve the problem. Milk insufficiency may result from inappropriate restrictions on feeding, unresolved positioning problems, and lack of accurate and consistent breastfeeding advice and support (Woolridge, 1995). Social and psychological factors may play an important role in the development of concepts 103

3.4.2.5 Breast pain


Background In many studies, sore breasts and nipples are presented together in the analysis (for example, Hamlyn et al., 2002). Women, however, experience pain in the breast as well as pain in the nipples, which may or may not be related. Even in the absence of engorgement and mastitis, breast pain can occur during and between feeds. No studies have been identified which address the prevalence, prevention or treatment of sore breasts caused by factors other than engorgement and mastitis. Suggested causes include spasm of the ducts, persistent reaction to nerve trauma, deep candida infection and prolactin-induced mastalgia. Findings from other reviews No studies have been identified in previous reviews. Discussion No recommendations for practice or policy can be made on the basis of the evidence in regard to breast pain

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

about insufficient milk and undermine womens confidence in their ability to feed their infant. Maternal blood loss and drug intake may also have an effect on milk supply. Findings Our review found no papers published in or after 1995 reporting on RCTs of treatments for milk insufficiency and lactation failure. Findings from previous reviews Renfrew et al. (2000) found three descriptive studies of the provision of support for lactation failure or lactation crises (Hillervik-Lindquist, 1991; Mathur et al., 1992; Woolridge, 1995), all of which reported a strong positive effect of supportive care; and several trials of drug treatments of lactation failure and milk insufficiency including metoclopramide, sulpiride, domperidone, TRH, oxytocin, growth hormone and herbal extracts (Aono et al., 1979, 1982; de Gezelle et al., 1983; Friedman and Sachtleben, 1961; Gokhale, 1965; Guzman, et al., 1979; Hofmeyr et al., 1985; Huntingford, 1961; Kauppila et al., 1981a/b, 1988; Lewis et al., 1980; Luhman, 1963; Milsom et al., 1992; Peters et al., 1991; Petraglia et al., 1985; Sharma et al., 1996; Ylikorkala et al., 1982, 1984; Zarate et al., 1976). These studies investigated not only normal healthy women who reported insufficient milk but also women separated from their babies who wished to produce milk, and mothers with a longstanding history of lactation failure. These studies suggest that dopamine antagonists should not be used in routine practice and that there is not enough evidence to support the use of TRH, oxytocin, growth hormone and herbal preparations for treating milk insufficiency. Discussion In the years since the search for Renfrew et al. (2000) was conducted no further studies have been published. The following comments from that report, therefore, still portray an accurate picture of the field. Mothers who fear that their babies are not receiving enough milk are usually highly anxious and in need of support and good care. It is astonishing that there remains such a lack of good evidence to inform women and practitioners, and to guide practice, in relation to the leading cause of breastfeeding failure worldwide. It is still 104

not possible in practice to distinguish between women who have a physiological cause underlying their inability to satisfy their baby and those whose milk supply is affected by social or psychological factors; research in this field is only in early stages of development. Routine measures to prevent insufficient milk are fairly well established, however, and ways of putting them into practice should be more critically examined. The descriptive study by Mathur et al. (1992) in India is the only example of a study that investigates a theoretically based regime of care which combines physical and psychosocial factors. The care they offered women, which appeared to be very successful, suggests that it is possible to successfully treat a very high proportion of women with insufficient milk even those identified as having total lactation failure. In all of the plethora of studies examining drug treatment for this condition, only Mathur et al. (1992) examined the use of a drug in combination with other supportive measures. The other striking aspects of these studies is that all are small and the crucial questions of clinical relevance, such as milk yield, duration of breastfeeding and weight gain in the baby have not been well addressed. In addition, there are obvious distinctions to be drawn between women with reduced lactation for physical or social/emotional reasons, and women with potentially normal lactation who cannot feed their babies and need support to continue under difficult circumstances. No studies have examined this essential question. Future research milk insufficiency There remain many unanswered questions which deserve to be addressed: Work is needed to distinguish the different types of milk insufficiency and to differentiate those which have a true pathophysiological basis from those which are non-organic in origin (ie acquired, iatrogenic or behaviourally induced) Do women who are anaemic or who have a heavy blood loss at birth have subsequent problems with milk production? If so, are there any interventions which alleviate such problems? A theoretically based programme of supportive care (such as that described by Mathur et al., 1992) should be tested in a RCT, and specifically targeted at specific groups such as:

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

women with insufficient milk women with babies in neonatal units women at risk of discontinuation of breastfeeding (eg socio-economically disadvantaged) Would dopamine antagonists (eg domperidone, metoclopramide) enhance lactation for: women in the community who are lacking effective sources of support? women who also receive supportive care? women separated from their babies, who must sustain their lactation by breast expression? Treatments which are often advised for insufficient milk, and which could be evaluated by RCTs, including: prolonged feeding from one breast at each feed breast massage supplementing babies who show slow weight gain increased frequency of breastfeeds Evaluation of effective ways of: restoring lost confidence overcoming crises of confidence, such as peer support schemes Evaluation of the predictive ability of clinical investigations (milk volume measurement, assessment of milk fat content, hormone levels, ultrasound examination of the breast and/or babys mouth) and of their role during follow-up in cases of perceived or substantiated breastmilk insufficiency Can self-help guides targeted at women experiencing perceived milk insufficiency be as effective as personal contact and support?

Unrestricted contact between mother and baby, and unrestricted feeds Checking positioning and attachment to ensure effective, pain-free feeds Restricting routine supplementary feeds. In relation to specific practices to treat specific breastfeeding problems, this review adds no new information.

A coordinated programme of work is needed to address appropriately the interrelated physical, psychological, professional, social and cultural issues.

3.4.3 Conclusions
This section of the review adds information from seven new RCTs, none of which were conducted in the UK. The only clear guidance from practice from this review is that the preventive measures described in Section 3.3 should be used to reduce the numbers of women experiencing problems, and that when these problems do arise those same measures should be checked to ensure that they are the baseline for all care thereafter. This should include:

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

105

106 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention After presentation and diagnosis (mean 30.1 days in control group; 21.3 days in intervention group) No breastfeeding duration outcomes reported Self-reported pain scores, physician-rated healing, pain behaviours, percentage of mothers skipping breastfeeding because of pain reported No breastfeeding outcomes reported Pain scores on days 1 and 3 and weeks 1 and 2 No breastfeeding outcomes reported Nipple discomfort reported to day 5 Within 36 hours of childbirth No breastfeeding duration outcome reported Physician assessment and self-report 5-7 days after start of treatment No breastfeeding duration On development of breast engorgement usually 3 days outcomes reported Pain levels reported 2 hours postpartum after application of cabbage leaves Chilled vs room temperature cabbage leaves for breast engorgement Cabbage leaf extract vs placebo On development of breast engorgement No breastfeeding duration outcomes reported Pain, engorgement, breast hardness measured 4 hours after start of the treatment Timing of breastfeeding duration outcomes 42 breastfeeding women with Hydrogel dressings for sore nipples sore, cracked or bleeding nipples (14% mothers in the control group and 9% in the intervention group were eligible for WIC) 66 primiparous breastfeeding mothers Collagenase, dexpanthenol, warm water for treating and preventing sore nipples Day 1 onwards for 2 weeks 65 primiparous breastfeeding women with sore nipples (annual family income Canadian $35,000) 84 breastfeeding mothers with staphylococcus aureus infected sore nipples Tea bags for sore nipples Systemic antibiotics vs topical On diagnosis of laboratory antibiotics for sore nipples confirmed staphylococcus aureus-infected sore nipples 28 breastfeeding mothers with breast engorgement Aboriginal mothers excluded 39 breastfeeding mothers with breast engorgement Aboriginal mothers excluded

Table 3.4: Included studies solving breastfeeding problems

Name of authors and study number

Date of Country publication

Design

Brent et al. [59]

1998

US

RCT

Kuscu [60]

2002

Turkey

RCT

Lavergne [61]

1997

Canada

RCT

Livingstone and Stringer [62]

1999

Canada

RCT

Roberts et al. [63]

1995

Australia

RCT

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Roberts et al. [64]

1998

Australia

RCT

3.5 Clinical interventions: baby related problems


Topic Breastfeeding problems resulting from neonatal conditions. These include hypoglycaemia, jaundice, crying and colic, sucking problems, sleep problems, sleepy babies, slow weight gain, Downs syndrome, ankyloglossia, cleft lip, cleft palate

Summary of methods
Study design of included studies: any design Dates for searches: 1995 to June 2003 Population group included: breastfeeding mothers of babies experiencing the problems identified Countries included: all Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, problems encountered, views of those experiencing and providing care, relevant process outcomes including economic outcomes

3.5.1 Evidence base


We searched for primary studies and reviews of evaluations of interventions for the baby related feeding problems listed above. As studies of these conditions are rare we included studies of any design, including case studies. The population group included primiparous and multiparous mothers and their infants in any country. Included are evaluations or descriptions of interventions reporting primary outcomes related to duration of breastfeeding and secondary outcomes such as reduction in nipple pain, reduction in breast engorgement and mastitis and improvement in infant weight gain. We found three papers reporting on strategies to address ankyloglossia and three reporting on strategies to overcome feeding problems associated with cleft lip with or without cleft palate. All except one were small-scale descriptive studies. Only one was a withdrawal design trial testing interventions for infants with cleft lip and/or cleft palate.

3.5.2 Findings 3.5.2.1 Hypoglycaemia


Background One of the more controversial areas in the care of the normal, term breastfed baby is the prevention, detection and management of hypoglycaemia. This is an important topic. The prevalence of hypoglycaemia is estimated to be 4.4 per 1,000 total live births (Gutberlet and Cornblath, 1975). Glucose is an immediate source of energy, providing 38 moles of adenosinetriphosphate (ATP) per mole of glucose oxidised. Nearly all the oxygen consumption in the brain depends on glucose. Cerebral glucose uptake is normally dependent on blood glucose concentration. In neonates there is not always an obvious correlation between blood glucose concentrations and clinical signs and symptoms of hypoglycaemia. The absence of clinical manifestations does not indicate that blood glucose concentration is normal and has not fallen below a level that maintains brain metabolism. Moreover, hypoxaemia and ischaemia potentiate the role of ischaemia in causing brain damage and may result in

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

107

permanent neurological damage. Blood glucose in a healthy term newborn, dependent on maternal glucose, is typically 1.9mmol/L in the first one to three hours, thereafter declining and reaching its lowest point at about three hours. It then begins rising as a result of mobilisation of endogenous glucose via gycogenolysis, gluconeogenesis, active lipolysis and ketogenesis reaching ~2.2mmol/L between three and 24 hours and ~2.5mmol/L after 24 hours. The potential for neonatal hypoglycaemia to cause neurological, intellectual and psychological sequelae in later life has led to much debate about the levels of plasma glucose at which to intervene. However, no fixed cut-off value has been agreed. There is no evidence that healthy term babies develop clinical signs of hypoglycaemia simply as a result of underfeeding. Importantly, there is no convincing evidence that hypoglycaemia detected in healthy term infants has persisting or irreversible clinical consequences. Early and unrestricted breastfeeding has the potential to prevent and correct falling levels of plasma glucose in normal term infants. Our search failed to bring up studies investigating strategies to prevent and manage hypoglycaemia in breastfed infants. A group that often causes concern are babies who will not feed or who do not wake to feed. In these circumstances regular review of the babys condition is advisable. Findings In spite of the importance of this problem no recent studies of care for these babies were identified. Findings from previous reviews The literature relating to this issue is extensive and is the subject of a comprehensive review prepared for the World Health Organization (Williams, 1997). That review draws on theoretical descriptive and experimental studies. Discussion Our review identified no relevant studies to inform the evidence base on hypoglycaemia in normal, healthy term breastfed babies, in spite of the controversial and important nature of the topic. There is no evidence, 108

however, that these babies should be tested routinely for low blood sugar as long as they show appropriate vital signs and are asymptomatic. Future research hypoglycaemia We need to know more about: Neurodevelopmental outcomes in healthy term infants with asymptomatic hypoglycaemia The relationship between breastmilk intake, feeding patterns and glucose levels in healthy term babies in the early neonatal period The diagnostic value of ketone body levels in the hypoglycaemic infant If there are differences in ketogenesis between breast and formula-fed neonates The practice of leaving vulnerable newborns, including those having undergone an instrumental delivery, to sleep for 12 hours or more, and its effect on blood glucose levels. It is reported to be a growing practice in some centres. We also need studies to inform optimal care of babies who will not feed or who do not wake to feed.

3.5.2.2 Jaundice
Background All newborns develop elevated serum bilirubin levels in the first week of life as a result of increased red cell breakdown, decreased removal of the products of breakdown due to transient liver enzyme deficiency and increased reabsorption of the breakdown products by the entero-hepatic circulation. The two main types of jaundice which occur in the healthy term baby are the normal, early-onset physiological jaundice which begins around day three and lasts until around day 10; and the much less common, late-onset, so-called breastmilk jaundice which appears later from around day 14 or later and may last for three to four weeks. Physiological jaundice occurs in about 50% of newborns, while breastmilk jaundice occurs in only 1-2% of neonates. Factors likely to make physiological jaundice worse are prematurity, bruising, cephalhematoma, and the delayed passage of meconium. An association has been found with breastfeeding, though it is hard to identify whether the problem is a result of breastfeeding not being well

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

established (as a result of feeding restrictions commonly practised in many countries) or an intrinsic factor in breastmilk. Other causes of neonatal jaundice include Rh and ABO blood group incompatibility, red blood cell enzyme deficiencies such as glucose-6-phosphate dehydrogenase deficiency, polycythaemia, hypothyroidism, intrauterine infections, neonatal sepsis and some metabolic disorders. High levels of serum bilirubin in the neonate may result in kernicterus, with the possibility of irreversible neurodevelopmental damage. It is very rare in healthy, term babies to find such high levels; much more commonly, babies experience slightly raised levels of serum bilirubin. Because of the known outcomes of kernicterus and the difficulties in identifying the exact cause of jaundice, however, clinicians are likely to intervene with a battery of tests followed by phototherapy and close monitoring even at relatively low levels. Unfortunately these practices may result in separation of mother and baby, recourse to scheduled feeding and the provision of additional feeds and water. This course of events may terminate in early weaning. Findings Our review found no studies of interventions for newborns with jaundice (early or late-onset) published after 1995 that had outcomes of interest. Findings from previous reviews Renfrew et al. (2000) identified five studies of early-onset jaundice looking at the links between breastfeeding and jaundice, prevention of jaundice in the newborn and treatment of jaundice. The interventions tested were water supplementation and separation of mother and infant. They found that maintaining unrestricted breastfeeding, not offering routine supplements and keeping mothers and babies together were effective treatments. No studies of late-onset jaundice were identified. Discussion There is a surprising lack of well-conducted research about this condition, which is very common, and which often results in serious disruption of breastfeeding and of the relationship between the mother and her baby. Such research as does exist suggests that breastfeeding should

be supported and mothers and babies kept together. When such jaundice occurs, treatment ought to be in proportion to the problem. For example, in the absence of complicating factors such as infection or prematurity, it is highly unlikely that physiological jaundice will reach levels that require major intervention. If babies do require phototherapy, care should be taken to avoid disrupting the relationship between mother and baby, and to support breastfeeding. No studies were identified that examined the treatment of late-onset or breastmilk jaundice. Future research jaundice Although practices vary in different settings we need to conduct robust studies of interventions limiting the impact of diagnosis and management of jaundice in the hospital setting. Over and above interventions for maintaining breastfeeding in the normal infant, we need to gather evidence from well designed studies of interventions that help to maintain breastfeeding when the diagnosis of jaundice has been made. These studies should distinguish carefully between early and late-onset jaundice, as their aetiology and treatment are different.

3.5.2.3 Crying and colic


Background Almost all babies go through fussy periods. When crying lasts for longer than three hours in a day and is not caused by a medical problem such as hernia or infection it is often called colic. The only difference between crying and colic is the time spent crying. Of interest is the finding from cross-cultural studies that time spent crying varies from culture to culture. It is not possible to say whether this is associated with maternal behaviour, differences in feeding patterns, variations in maternal diet, or other factors. Many experienced mothers are able to tell the difference between crying associated with hunger and crying that is not linked to hunger; even so, they are likely to become anxious about the reason for the inconsolable crying. New and inexperienced mothers are likely to link any crying with feeding and to become anxious about their breastfeeding if their babies cry more than they expect. This is particularly the case with breastfeeding mothers who feel that their babies are not getting enough milk, 109

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

or that the milk is too weak or too rich. It is not surprising that many mothers resort to formula feeding and early weaning. Findings This review found no studies of interventions to prevent or address the problem of crying and colic published after 1995 with outcomes of interest. Findings from previous reviews Renfrew et al. (2000) describe the hypothesis proposed by Woolridge and Fisher (1988). They proposed that infants received little hindmilk if their mothers followed the practice of feeding for 10 minutes on one breast and then moving to the other during a single feed. The resulting over-ingestion of foremilk was, they suggested, a reason for babies suffering colic and lactose malabsorption. This was then tested in two studies, Evans et al. (1995) and Righard et al. (1993). Evans et al. (1995) found a significant improvement in crying while Righard et al. (1993), in a smaller study, did not. Discussion

these babies may have a congenital heart defect. Congestive cardiac failure as a result of such a defect may cause the baby to tire easily with the physical effort of sucking. In addition to these technical difficulties initiation of feeding may be delayed because of the difficulties in breaking the news to the parents and any immediate investigations related to serious heart defects. Findings Our search did not identify studies of interventions promoting breastfeeding in mothers of babies born with Downs syndrome or other similar chromosomal anomalies. Findings from previous reviews Renfrew et al. (2000) found one study (Aumonier and Cunningham, 1983) describing breast and bottlefeeding experiences of mothers of babies born with Downs syndrome. This study confirms that it is possible to breastfeed babies with the chromosomal configuration and resulting physical features if there is adequate support from caregivers. Discussion

As with insufficient milk, the reaction to inconsolable crying will involve a combination of physical, psychological and social factors that will be difficult to disentangle. There is little evidence to guide practice in this important area, which has the potential to cause great disruption and distress for the family. It is notable that standard reviews of colic treatments lack any reference to breastfeeding management. Future research crying and colic Evidence is urgently needed from studies of interventions to maintain breastfeeding and resolve colic, aimed at both babies and their families.

Breastfeeding a baby with Downs syndrome is not impossible, though many families and their caregivers may be anxious about doing so. The possibility of continuing to breastfeed a baby with Downs syndrome will depend on the quality of care they receive and whether or not the baby has any associated abnormalities. Future research Downs syndrome Evidence is needed of the effectiveness of interventions aimed at mothers of Downs syndrome babies, and of interventions specific to the range of deformities preventing successful breastfeeding among these babies.

3.5.2.4 Downs syndrome 3.5.2.5 Ankyloglossia or tongue tie


Background Background Newborns with Downs syndrome (trisomy of the 21st chromosome) have several features that may prevent them from sucking successfully, including generalised hypotonia, poor tone in the musculature of the tongue, a small oral cavity and a thick furrowed tongue. Many of 110 Congenital ankyloglossia or short lingual frenulum (commonly known as tongue tie) may inhibit adequate tongue extension in the neonate, causing difficulty in grasping and holding on to a large enough mouthful of

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breast tissue to breastfeed successfully. This may result in nipple pain, poor latching and inadequate transfer of milk. Treatment of congenital ankyloglossia involves a frenulotomy, or division of the frenulum, a quick and easy procedure with minimal blood loss. Some argue that a short frenulum may tear spontaneously over time and therefore requires no intervention. Others say that the condition has little to do with early weaning. The true prevalence of ankyloglossia in the general population is still unknown. Ballard et al. (2002) [65] reported an incidence of 3.2% among 2,763 breastfeeding term infants in a hospital in the US and 12.8% among infants referred to the hospitals lactation centre. Another centre in the US reported a higher incidence (4.8%) from a total of 1,031 newborns in a well-baby nursery (Messner et al. 2000). The diagnostic tools and definitions used in these two studies differed; Messners study included mild cases of ankyloglossia. It is not known what degree of ankyloglossia results in pain for the breastfeeding mother. Findings Three papers met our review criteria. The first (Ballard et al., 2002 [65]) reports on work carried out in a centre in the US. A total of 3,036 breastfeeding infants, of which 273 were outpatients, were screened for congenital ankyloglossia using the Hazelbaker Assessment Tool for Lingual Frenulum Function and observation of the mother-infant dyad during a feed. Infant latch and nipple pain were assessed. Frenulotomy was performed in 123 infants. Latch improved in all cases and maternal pain levels fell from 6.92.31 to 1.21.52 (p=.0001) after the procedure. The study group was selected from a suitable sampling frame, inclusion criteria were clear and analysis was appropriate. Mothers commitment to breastfeeding along with learning and using the correct latching technique are factors that may have affected the outcome of this study. The findings of this large case series are indicative but not conclusive that frenulotomy can be an effective method of improving latch and reducing breastfeeding problems in babies with tongue tie. The second (Masaitis and Kaempf, 1996 [69]) reports on the development of a frenulotomy policy in one medical centre in the US running a mother-baby programme. The programme follows up three to five day old infants and any mothers developing breastfeeding problems in the first two weeks. Infants with short labial frenula causing

injury to maternal nipples, diagnosed on breastfeeding assessment and oral examination, were subjected to frenulotomy. The procedure is described in detail. Of the 36 infants, 32 were breastfeeding at all at one week and 19 at three months after the procedure. The rate of infant growth was appropriate in all infants at three months. The third (Fitz-Desorgher, 2003 [67]) is a case study of one baby with marked ankyloglossia and more than 10% weight loss on day three. After division of the frenulum the infant latched on successfully, the mother perceived a difference in the power of sucking and the infant continued to breastfeed for one year. Our search also identified three older studies, excluded as they fell outside our criteria for date of publication (Berg, 1990; Huggins, 1990; Nicholson, 1991). Findings from previous reviews Renfrew et al. (2000) reviewed two papers including case studies of 15 babies with ankyloglossia (Marmet et al., 1990; Notestine, 1990). All babies undergoing frenulotomy were able to breastfeed successfully. Discussion Congenital ankyloglossia has potential to result in severe pain for the mother, inadequate milk transfer and growth problems for the baby. It is very likely that a proportion of the women who abandon breastfeeding because of nipple pain and trauma, or because of insufficient milk, may have babies with short frenula. Frenulotomy, or division/clipping of the short frenulum, is a procedure that, given the studies described here, is likely to help mothers and their babies continue breastfeeding, reduce maternal pain and prevent poor infant weight gain. Further controlled studies are required to test this procedure before it is adopted in widespread practice. One RCT (Hogan et al., 2004, published since our search; see Appendix 4) compared the division of tongue tie to support from a lactation consultant in 57 newborns in a regional paediatric centre in the UK. This study has shown promising results in favour of division performed by infant feeding specialists. The potential for a simple intervention to make a major difference to breastfeeding mothers makes this an important area for future research.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

111

Future research ankyloglossia Population studies are needed to identify the prevalence of tongue tie in the UK. This should include defining the range of presentations and severity. It is likely that individual variability in the relationship between the babys mouth and the mothers breast will affect the decision making process; for example, a breast which is small and has reservoirs close to the nipple may be easier for a baby with tongue tie to grasp than a large, full breast. Such descriptive work would identify the babies and mothers who might benefit from intervention. The possible benefits and side effects of frenulotomy, and the personnel, training and circumstances required for safe and effective intervention, need further exploration in controlled studies.

could be initiated and maintained. A withdrawal study design was used so that six different combinations were tested: no feeder with no obturator, followed by feeder only, followed by obturator only or obturator and feeder, followed by withdrawal of obturator and finally reintroduction of obturator. Feeding time was shorter, and volume of milk intake and rate of flow higher when infants used the palatal obturator, compared to the times when the Haberman feeder was used alone. Shortened feeding times, lactation instruction and palatal obturator with use of the Haberman feeder led to normal weight gain over time. Three year data show that there were no significant differences in weight gain for babies who continued to receive breastmilk compared to those who received formula. The second paper (Crossman, 1998 [66]) is a case study of one infant with a cleft palate whose mother was a midwife. Several strategies were adopted in a serial fashion to initiate and maintain breastmilk transfer from mother to baby including direct expression of breastmilk into the babys mouth followed by use of a squeeze bottle and soft teat, and then the use of a special artificial teat with reinforced top, soft underside and cross-slash for outflow. A metal plate was used to close the palatal gap from day five onwards. Only expressed breastmilk was fed to the baby from birth onwards. The baby fed well and gained weight until the surgical repair at seven months. Three weeks following surgery the baby was suckling at the breast, abandoning the use of cups and feeders. Throughout the first eight months the mother received positive support from caregivers. This study brings out the importance of sustained efforts and adaptations to continue breastmilk intake in babies affected with cleft palate until surgical repair becomes possible. Kogo et al. (1997) [68] conducted a trial of a Hotz-type plate to breastfeed infants with both cleft lip and cleft palate in Japan. A Hotz plate is typically used for artificial feeding in infants with cleft lip and palate. The addition of a posterior ridge to the plate seals off the oral cavity by apposition of the acrylic ridge and tongue, thus facilitating breastfeeding. Four of the 10 infants whose mothers continued breastfeeding on their return home sucked at the breast immediately after insertion of the plate and there was no weight reduction compared to babies who fed through a bottle. One baby continued breastfeeding at all to 14 months of age. However, all four infants were given supplementary feeds and three

3.5.2.6 Cleft lip and palate


Background Cleft lip results from failure of the medial nasal and maxillary processes of both sides to meet and join during normal fetal development. Cleft palate results from failure of the palatal shelves of both sides to approximate or fuse before birth. Closure of plates normally takes place in the first trimester of pregnancy. Cleft palate may involve the soft palate only or extend to include the hard palate. Studies suggest that both genes and environmental factors may play a role. Drugs, in particular some of the anticonvulsants, infections, maternal illness, maternal smoking and folic acid deficiency have been implicated in the aetiology of cleft lip/palate. The incidence of cleft lip with or without cleft palate is about 1:600 births, and of cleft palate alone is about 1:1,000 births. This group of developmental defects are known to cause difficulties in breastfeeding. Both conditions are corrected by surgical repair. Findings We found three papers describing different strategies to address feeding problems for babies before surgical repair was attempted. Turner et al. (2001) [70] evaluated the effectiveness of a palatal obturator with lactation instruction and a Haberman feeder in eight infants with cleft lip and/or palate in the US. A lactation consultant met each motherinfant dyad in the first days of life so that breast pumping 112

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

gave up breastfeeding after several months (it is not specified how many months). Findings from previous reviews Renfrew et al. (2000) synthesised evidence from two studies (Weatherley-White et al., 1987; Darzi et al., 1996). Both studies examined breastfeeding after surgery. Discussion Mothers wishing to breastfeed babies with cleft lip and/or cleft palate need both psychological and technical support. Our review has introduced other studies of the effectiveness of interventions to establish and maintain the flow of breastmilk until the time of surgery. There is a limited amount of research on this topic but it is encouraging and suggests that interventions can be used to support breastfeeding in these babies. Future research cleft lip and palate Work is needed to examine further the interventions identified here, to examine the possibility of new interventions and to examine the training of staff and the views of parents on feeding babies with these problems.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

113

114 Participants and any sub-groups Intervention (in addition to routine care, unless otherwise stated) Timing of delivery of intervention Division of frenulum After presentation (median age at presentation for nipple pain = 2 days) Breastfeeding outcomes not reported Nipple pain + latch reported after procedure 8 months of age (3 weeks after scheduled surgery) Timing of breastfeeding duration outcomes 123 infants with signs of ankyloglossia 1 infant with cleft lip Breastmilk expression Use of squeeze bottle and soft teat Dental plate Use of specialised cleft-palate teat Support from caregivers Squeeze bottle from day 2 Dental plate from day 5 Cleft palate teat from 14 weeks Division of frenulum Day 8 1 year 1 term infant with signs of ankyloglossia 10 infants with both cleft lip and palate Hotz type plate with posterior Day 1, day 7 and day 50 in 4 At the time of weaning (maximum 14 months) ridge babies breastfeeding after return home Not stated for 6 babies who were bottlefed Division of frenulum Before 4 days of age in 16/19 1 week after the procedure 3 months of age breastfed infants After 4 days of age in 14/17 infants who were fully bottlefeeding at 3 months Breastfeeding duration not reported Infant weight and height at 2 and 3 weeks, 4 months, 1 and 2 years Time to feed Volume of milk intake Lactation instruction + palatal In the 1st month of life obturator Haberman feeder 36 term infants with signs of ankyloglossia 8 infants with cleft lip with/without cleft palate

Table 3.5: Included studies baby related problems

Name of authors and study number

Date of Country publication

Design

Ballard et al. [65]

2002

US

Descriptive

Crossman [66]

1998

Australia

Descriptive

Fitz-Desorgher [67]

2003

UK

Descriptive

Kogo et al. [68]

1997

Japan

Descriptive

Masaitis and Kaempf [69]

1996

US

Descriptive

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Turner et al. [70]

2001

US

Withdrawal design trial

3.6 Clinical interventions: maternal problems that may affect breastfeeding


Topic Deals with maternal problems that will affect the promotion and support of duration of breastfeeding, including drug and substance abuse, use of prescription and over-the-counter medications, smoking, maternal nutrition, and breast surgery. HIV infection was excluded as it was assumed that in the UK, women who are HIV positive should be supported to feed their baby with formula milk Summary of methods
Study design of included studies: RCTs only Dates for searches: 1990 to June 2003 Population group included: pregnant or breastfeeding mothers experiencing the problems identified, and their babies Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes

3.6.1 Background
Pregnancy is a time of increased maternal nutritional requirements, most of which are met by normal physiological processes (Anderson, 2001). Additional energy need for a mother exclusively breastfeeding her infant is 500kcal/day (Dewey, 1997). The recommended increment in protein during this period is 20g/day. There is, however, little evidence to show that maternal nutritional status has any substantial effect on either the quality or the quantity of breastmilk. Despite the fact that existing evidence does not support maternal nutrition supplementation in developed country settings to improve key indicators of health status there is an argument for providing mothers with supplements to prevent weight loss, general fatigue and other maternal morbidities. Pregnant and lactating women are a target group in their own right; nutritional supplements may improve their health and sense of wellbeing, regardless of their childbearing and breastfeeding roles. Maternal weight loss and tiredness may well be associated with early weaning. Morbidity linked to poor or even normal caloric intake during lactation has not been adequately studied. Maternal exhaustion is a frequent problem in the first few months after childbirth. High levels of fatigue are

reported by women suffering from postpartum depression and may result from or contribute to this condition (Corwin et al., 2003). Research has shown that postnatal depression affects 10-15% of mothers in Western societies (Tammentie et al., 2004). Both exhaustion and depression have been implicated in early weaning. Henderson and colleagues found that early cessation of breastfeeding was significantly associated with postnatal depression (Henderson et al., 2003). Other related factors in this complex and so far inadequately understood area of maternal health are maternal confidence in breastfeeding and self-efficacy. A recent study of 300 Australian women found that self-efficacy was significantly related to duration and level of breastfeeding (Blyth et al., 2002). Strategies that help alleviate fatigue and depression in the postnatal period and those that enhance maternal self-efficacy in breastfeeding are likely to be helpful in extending the duration of breastfeeding. The Infant Feeding Survey 2000 found that of the 20% of mothers who smoked during their pregnancy almost all continued to smoke after childbirth and a quarter of those who stopped during pregnancy started again within a few weeks after childbirth (Hamlyn et al., 2002). There are wide variations in breastfeeding rates among women who 115

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

smoke. Several studies have demonstrated the association between smoking during pregnancy and early motherhood, and early weaning. A large study of 36,324 live births found that prenatal tobacco use was significantly associated with failure to exclusively breastfeed at two weeks (Letson et al., 2002). Edwards et al. found that mothers who smoked were more likely to discontinue breastfeeding by 12 weeks (Edwards et al., 1998). Horta and colleagues found that maternal smoking was strongly associated with breastfeeding duration, with those smoking 20 or more cigarettes a day having an OR of 1.94 for breastfeeding less than six months (Horta et al., 1997). Whether this is a physiological association or not is difficult to determine since many women who smoke are also less likely to hold an intention to breastfeed. An epidemiological review of smoking and breastfeeding suggests that psychosocial factors are more likely to be responsible for the lower rates of breastfeeding among women who smoke (Amir and Donath, 2002). Therefore strategies to improve breastfeeding among smoking mothers are more likely to be effective if they are directed at the psychosocial factors at play. We found no RCTs testing the effectiveness of any strategy to help smoking mothers to continue breastfeeding. Smoking cessation strategies during pregnancy and those that address relapse during the postnatal period are likely to be effective in extending the duration of breastfeeding. A systematic review of smoking cessation interventions found that a wide range of strategies work to help women stop smoking during pregnancy (Lumley et al., 2004). Breastfeeding duration outcomes were not reported in any of the studies included in the review. There is limited reliable information for drug misuse among women and this may be due in part to their status as a hidden drug population. One study conducted by researchers at the Northern and Yorkshire Public Health Observatory in the UK estimated that 75 babies were born to drug abusing mothers for every 10,000 live births during 2001 (Northern and Yorkshire Public Health Observatory, 2002). Amphetamines, marijuana, cocaine, heroin and phencyclidine used by breastfeeding mothers may impair production, volume and composition of breastmilk and can have adverse short- and long-term effects on the infant. For example, delta-9-tetrahydrocannabinol (THC) in marijuana transfers and concentrates in the mothers milk (Astley and Little, 1990). Cocaine is reported to cause irritability, vomiting, tremulousness and seizures in neonates and heroin 116

causes tremors, restlessness, vomiting and poor feeding. Babies born to drug-misusing mothers are likely to be separated from them at birth. The immediate postpartum period is a stressful time for all mothers, and among drug abusers the tendency is to use drugs as a means of reducing stress. Practitioners addressing the needs of drug-abusing mothers say that these women need extra postnatal support and information to continue breastfeeding. We found no studies evaluating interventions offering extra support to this group of women, or other strategies to help them continue breastfeeding. In some cultures folklore has it that maternal alcohol consumption acts like a galactogogue, aiding milk let-down, increasing breastmilk production and improving infant health (Mennella, 2001). Scientific evidence suggests that alcohol levels in breastmilk peak between 30 and 90 minutes after consumption and that infants receive less milk during this period. There is no consensus on how much alcohol a woman may safely take while breastfeeding. Alcohol may disrupt the infants sleep-wake pattern and motor development. This information may discourage women from breastfeeding, and if infants get less milk per feed and a disturbed sleep it may start off a vicious cycle leading to early weaning. We found no studies describing or testing interventions discouraging alcohol use in breastfeeding mothers. The vast majority of medications, both prescribed and over-the counter, taken by mothers during lactation are transferred to breastmilk and become bio-available to the breastfeeding infant (Bar-Oz et al., 2003). Most of these drugs have no adverse effects on the infant and only a few are contraindicated. A review of 100 published case reports in the literature concluded that none of the adverse reactions noted in breastfed babies were definite drug reactions, 47% were probable and 53% were possible (Anderson et al., 2003b). A number of factors need to be considered while counselling women about taking a particular medication while breastfeeding: the properties of the drug including onset of action, half life and bio-availability to the infant; the infants gestational age at birth and present age; and frequency of breastfeeding and volume of milk transferred. Healthcare professionals require evidencebased guidance on how to advise women who need to take medication while breastfeeding. However, we found

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

no RCTs of interventions in this area with outcomes on the duration of breastfeeding. The American Academy of Pediatrics has published a detailed statement on transfer of drugs and other chemicals into human milk (American Academy of Pediatrics, 1997), which provides information about specific drugs. The work of Hale (2002) aims to provide information both for health professionals and for women about this issue. This information suggests that medications fall largely into three groups: a small number that are highly toxic, such as chemotherapeutic agents, where breastfeeding should be discontinued; a small number that are concentrated in breastmilk, such as metoclopramide, where alternative medications should be sought; and the majority of drugs that are unlikely to cause problems for the baby. Practitioners should review the available information and make appropriate prescribing decisions. In relation to the drugs most commonly used by women who are breastfeeding there is little evidence of problems. For example, most painkillers are compatible with breastfeeding. Breastfeeding mothers should not be allowed to experience pain; neither should they be made to choose between breastfeeding and analgesia. Breast surgery is becoming increasingly popular in the field of cosmetic surgery and for the treatment of breast cancer. Breast reduction surgery may have an impact on breastfeeding performance and result in early weaning. One study found that women who had undergone breast reduction surgery had significantly shorter periods of exclusive and any breastfeeding (Souto et al., 2003). Although no studies report on breastfeeding duration outcomes, one small study suggests that physiological type breast reduction surgery (where all functioning breast tissue after the reduction mammoplasty should be attached to the nipple in a physiological manner) facilitates future breastfeeding (Marshall et al., 1994). Many women having silicone implants to augment breasts will breastfeed. Concerns have been raised about the effect of silicone implants on the breastfeeding infants (American Academy Pediatrics, 2001). Here again practitioners may advise women who have had these implants to abandon breastfeeding earlier than intended. We found no RCTs testing interventions to extend the duration of breastfeeding either in women having implants or in those who have undergone mammoplasty.

3.6.2 Evidence base


There were no studies found related to this section.

3.6.3 Discussion
When women have problems that may have an impact on breastfeeding the apparently simple solution is to recommend the discontinuation of breastfeeding. This may, however, cause a range of other problems, including removing the childs and the mothers access to the health benefits of breastfeeding and causing distress and anxiety for the mother. Such advice should therefore be based on sound evidence. There is, however, an absence of evidence to inform these decisions. Research is urgently needed to address this gap.

3.6.4 Future research


Interventions addressing some important conditions relating to the mother that may have an impact on breastfeeding duration are discussed above. We need answers to the following questions: Does improving nutritional status and body mass index (BMI) help to extend the duration of breastfeeding among women in developed country settings? What strategies can be successfully used to provide practitioners with key information in advising women at risk of vitamin deficiencies (vegans, anorexics)? What smoking cessation strategies are useful in helping women smokers to continue breastfeeding and to help breastfeeding women stop smoking? What strategies are successful in supporting drug abusing women who breastfeed to reduce infant exposure to drugs such as cocaine and cannabis? What strategies are successful in helping practitioners and pharmacists to advise women who need medication for pain, coughs and colds, depression, anxiety, epilepsy, hormone replacement (thyroid and others) about continuing or stopping breastfeeding? What kinds of breast enhancement and breast reduction surgery are acceptable to women and allow for full breastfeeding in future pregnancies? What strategies are useful in alleviating maternal exhaustion and depression in the postnatal period with the aim of helping women to continue breastfeeding? 117

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

4.1 Health professional and lay breastfeeding educator/counsellor training, education and practice change
Topic Health professional and lay breastfeeding educator/counsellor training, education and practice change relevant to breastfeeding, including the use of specific training programmes, ways of teaching health professionals about positioning and attachment, and ways of changing breastfeeding practice. Only studies including outcomes of the duration of breastfeeding are included (ie those only examining attitude, knowledge and behaviour change of health professionals/lay breastfeeding educators are excluded) Summary of methods
Study design of included studies: RCTs, non-RCTs with concurrent controls, before-after studies (cohort or cross-sectional) Dates for searches: 1980 to June 2003 Population group included: health professionals and lay breastfeeding educators/counsellors working with pregnant/postpartum women Countries included: developed countries Outcomes primary, secondary, process: duration of breastfeeding, duration of exclusive breastfeeding, clinical problems for women and for babies, views of those experiencing and providing care, relevant process outcomes including economic outcomes, attitude, knowledge and behaviour change of health professionals and lay breastfeeding educators/counsellors (studies had to include duration of breastfeeding for other outcomes to be included)

4.1.1 Background
It has previously been demonstrated that support from an appropriately skilled practitioner can have positive effects on womens initiation and experiences of breastfeeding (Section 1.1). There is wide variation in both the amount and scope of the preliminary education available to those healthcare professionals who support breastfeeding women in the UK. Training varies in terms of orientation and philosophical model (predominantly illness/health versus family centred, societal perspectives), duration, educational level, assessment strategy and between different groups of caregivers within the UK. A recent wide-ranging training needs analysis has demonstrated that this preparation can generally be considered inadequate and fragmented (M. Smale et al., University of York, personal communication), and is perhaps most lacking in the pre- and postgraduate training of medical practitioners. There has, until recently, been little incorporation of womens and breastfeeding counsellors 118

perspectives and expertise in education programmes for health professionals, although there are exceptions (Jack et al., 2001), and this appears to be changing slowly. Research investigating ways of achieving practice change among healthcare professionals has generally focused on medical practitioners. While evidence-based practice initiatives have been reported within midwifery that have involved the production of guidelines (eg Spiby and Munro, 2001; Walsh et al., 1999), these have generally focused on labour and have been introduced as a response to change in service provision. While several have been evaluated, appropriate control groups have not always been used. It is also acknowledged that the production of evidence-based guidelines alone is insufficient to achieve practice change (Bero et al., 1998; NHS Centre for Reviews and Dissemination, 1994) and that support with an interactive educational programme is optimal. Work conducted in medical settings reflects a hierarchy of strategies that may be considered effective in

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

achieving practice change. Consistently effective strategies include manual or computerised reminders, multifaceted interventions, outreach visits and interactive educational meetings. Interventions that are moderately effective include audit and feedback and the involvement of local opinion leaders; didactic education and the use of educational materials alone achieve little in terms of practice change (Haines and Donald, 1998). In relation to preparation for becoming a voluntary/lay breastfeeding counsellor or educator, information about the process and curriculum is readily available from national and international organisations active within the UK including the National Childbirth Trust and La Leche League. These organisations have established formal training programmes which involve both theoretical and practical components. These roles have been noted as valuable by professionals and by women themselves (Smale et al., personal communication). An overview and analysis of the scope, impact and effectiveness of breastfeeding practice projects funded over three years by the Department of Health Infant Feeding Initiative has been conducted by Dykes (2003). Projects included educational interventions for healthcare professionals and other workers. The range of interventions and lack of formal evaluation in these projects render interpretation difficult (Dykes, 2003), but recommendations are made related to the development of future interventions. These include: tailoring interventions to the needs of the target audience; use of existing packages as a more efficient use of time and resources than development of new programmes; and appropriate tools should be used that evaluate knowledge, attitudes and skills. Dykes review identified 26 peer-support projects. The author advises against interpretation of causality from data related to these projects but comments that the peer support schemes appeared, generally, to be supportive of improvements in breastfeeding rates. Further, qualitative information from women reflects their positive perceptions of help with difficulties and support. These findings reflect the positive effects of peer support programmes on breastfeeding duration identified in other research (Section 1.1).

4.1.2 Evidence base


The search identified nine papers. No RCTs were identified. All studies focused on education targeted at health professionals and those working in such settings; none were identified that related to breastfeeding counsellors. Three studies were conducted in the UK, two in US/Canada and the remaining four in mainland Europe. Five were relevant to those working with women from disadvantaged backgrounds. Table 4 provides information about the nine studies reviewed for this section.

4.1.3 Findings
Cattaneo and Buzzetti (2001) [71] describe the impact of providing the18 hour UNICEF training to prepare eight hospitals for the Baby Friendly Hospital Initiative in Italy. The study, using a before-after design, was conducted in two groups, each comprising four hospitals, in North and South Italy between 1996-98. UNICEF training was augmented with a two hour WHO counselling training session. In both groups of hospitals, an initial period of assessment was carried out. This was followed by the training for trainers in the first group of hospitals, subsequently cascaded to health professionals in those hospitals. This was then followed by a second assessment phase and implementation of the intervention in the second group of hospitals. Third assessments occurred in both units five months following completion of training. Methods used for data collection included the selfassessment tool of the Baby Friendly Hospital Initiative, a questionnaire for the trainees, interviews with mothers at discharge from hospital, and telephone interviews after three and six months. The results included an increase in compliance with the UNICEF Ten Steps to Successful Breastfeeding. Health professionals knowledge increased following training. Significant differences (p<0.05) were demonstrated following training in the increase in mothers reporting exclusive breastfeeding at discharge, full breastfeeding at three months and any breastfeeding at six months. Changes were seen for both groups of hospitals. It is also of interest to note that following logistic regression, full breastfeeding at three months was significantly associated with exclusive breastfeeding at discharge and previous experience of breastfeeding. The training of health professionals was found to have a positive but not statistically significant influence. 119

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Durand et al. (2003) [72] reported the results of a feasibility study exploring the introduction of an education programme based on the UNICEF Ten Steps to Successful Breastfeeding in a tertiary maternity facility in Grenoble, France. Training was provided as a standardised three-day seminar of theory and practice with 10 participants per group. All maternity care professionals were targeted and 73 members of staff completed training between early 1998 and early 2000. The selection of these individuals and the proportion of the total workforce trained are not reported. Data were collected from the case notes of 50 women, randomly selected prior to introduction of the training programme, and a further 50 women who received care following the introduction of the training programme. The two groups of women gave birth over six months, from July-December 1997 and January-July 2000. In both samples breastfeeding was initiated by a similar proportion of women (76%); no differences were detected between groups for the median duration of breastfeeding (12 weeks). Other changes were observed after the training and these included fewer babies separated from their mothers overnight in hospital and less use of formula in hospital. More women reported feeling informed about breastfeeding positions (p<.05), that advice given by maternity personnel was very/fairly consistent (p<.04), and they were informed about sources of support available after discharge from hospital (p<.01). Gainotti and Pagani (1980) [73] describe work carried out with staff working with mothers and babies in one hospital in Como, Italy. The aim of this work appeared to be to modify the cultural and psychological attitudes of neonatal personnel, and to increase staff awareness of the importance of psychological aspects of breastfeeding and beneficial properties of breastmilk so that staff could support mothers appropriately with their breastfeeding. The authors report that preparatory work was undertaken with staff but it is unclear exactly what that comprised. Emphasis was given to following Leboyer-type approaches to birth, supporting skin-to-skin contact, keeping mother and baby together for the first two hours of life and early breastfeeding (Leboyer, 2002). Data were collected related to breastfeeding outcomes for 50 mothers and babies who received care in 1974 prior to training and 50 mothers who received care in 1979 following staff training. At the time of the study, babies were transferred to a nursery following birth and brought to their mothers for feeding. Care is described as being provided by nurses. A significant increase in 120

exclusive breastfeeding at three months of age was seen (50% vs 30%) following the staff training programme. Grant et al. (2000) [74] report a programme of evidencebased practice change conducted at a teaching hospital serving a deprived community in south London from 1998-99. Evidence-based guidelines were developed that focused on three aspects of care related to breastfeeding. These three aspects of care were selected from the UNICEF UK Baby Friendly Hospital Initiative. They were: antenatal discussion of breastfeeding with women; promotion of skin-to-skin contact with early breastfeeding; and the prevention of neonatal hypoglycaemia and the care of babies with breastfeeding difficulties. This project was located within a suite of evidence-based practice approaches to a range of healthcare priorities using an action research model. Introduction of the guidelines was supported by a programme of educational interventions, audit and feedback, one-to-one and group discussion, newsletters and provision of educational material. The intervention was targeted at staff and facilitated by an experienced midwife working as project leader and change agent. Power calculations were carried out to inform considerations of sample size. Data were collected from 442 women who gave birth in hospital or at home in the area served and who completed postal questionnaires when their babies were 12 weeks old, and also from a similar group of women (n=489) after the active intervention had ceased. The authors note that compared with the overall population served by this hospital, younger women and black women were underrepresented in the survey. Other data collection included audit data, using specific tools. Certain aspects of professional practice changed significantly, including providing women with evidencebased leaflets about infant feeding during pregnancy, antenatal enquiry and discussion about infant feeding, and skin-to-skin contact on the labour ward. No significant differences in breastfeeding initiation or duration were detected. There was no difference in the number of babies readmitted to hospital after the intervention, although there was an increase in readmission due to jaundice. Hartley and OConnor (1996) [75] report an evaluation of the Best Start breastfeeding education programme conducted from 1993-94 in a hospital centre for

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

womens healthcare in Ohio, US. The programme was presented to medical, nursing and secretarial staff at the womens health centre that served an urban area of lower socio-economic status. The training is described by the authors as a didactic approach to the provision of information about breastfeeding in order that staff could answer womens questions appropriately. The second component of that three hour training comprised an introduction to Best Start educational techniques based on the theory of social marketing or promotion of socially beneficial practices. Sample sizes were based on numbers required to detect a doubling of existing breastfeeding rates. The aim was to provide a three-step approach to education about breastfeeding for women during their antenatal visits. These steps comprised enquiry to identify concerns about breastfeeding, acknowledgement of concern, and explanation of the benefits of breastfeeding. Women were specifically asked not to make a decision about feeding method at the first discussion as that could prevent subsequent discussion and make it less likely that a woman would change her plan. During the 24-hour postpartum hospital stay lactation nurses were available to help mothers to initiate breastfeeding and to help with obtaining a breast pump. All low-income mothers received one home visit within the first 72 hours after discharge from hospital. A lactation clinic was also available for mothers with breastfeeding problems. Data were collected from hospital charts of babies and records of the visit to an outpatient department of a childrens hospital two weeks after birth. Ninety mother-infant pairs were included in each of the before-after groups; these were comparable for several demographic factors, including ethnicity, parity and type of payment for care. Women were discharged from hospital 24 hours after birth. Rates of breastfeeding at that point increased from 15% to 31% following the education programme (p<.03). At two weeks postpartum, an increase was still apparent (13% vs 21%, p<0.2) but that did not achieve the difference sought by the researchers. Examination of data for particular sub-groups reveals suggestions of benefit for women under 20 years of age and over 30 years, albeit for small numbers, and for African-American women, that achieved statistically significant differences (12% vs 31% p<.01). It should be noted that this research was conducted in an area of social disadvantage where the Women, Infants and Children Supplemental Nutrition programme was active during both phases of

the study and most of the participants in the study were recipients of that programme. Ingram and colleagues report a study conducted at a teaching hospital in south Bristol (Ingram et al., 2002 [76]). This study examined hospital midwives teaching mothers about positioning and attachment using a handoff approach whereby midwives did not touch the baby or the mothers breasts but instead talked the mother through the process of attaching the baby herself (in contrast to a common technique of putting the baby onto the breast). The question was whether this approach would reduce breastfeeding problems and impact on breastfeeding duration. The population served by the hospital had low rates for initiation and continuation of breastfeeding. Women who lived in south Bristol all received postnatal care on one designated postnatal ward at the start of the study, which took place in four phases between October 1996 and November 1998. The first phase comprised collection of baseline data from mothers prior to any intervention. A research fellow then provided training to the hospital midwives and healthcare assistants. The training included 45 minute workshops incorporating physiology, a rationale for using a hands off technique, a video, and practical work using dolls and pillows. The hands off breastfeeding technique comprised eight practice points and covered topics such as mothers and babys positions for feeding. The training sessions were supplemented by support from the research fellow in the ward area. During phase two, hospital midwives taught women to use these techniques (initial teaching reported to take 30 minutes) and a research midwife assessed a sample of 395 mothers for use of the techniques through observation and use of a scoring system. Phase three comprised a continuation of work in phase two and the research midwife also provided leaflets to mothers reinforcing the technique. In phase four the research midwife withdrew from the ward and the leaflet was provided to mothers by hospital midwives who continued to provide instruction. Data related to duration of infant feeding was collected from women using postal questionnaires (and for a small number follow-up phone calls) at two weeks (1,173 women) and six weeks (1,071 women) postpartum. There were no significant differences between the groups for a range of factors including parity, intention to 121

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breastfeed or breastfeeding within one hour of birth. Information about ethnicity was not sought but it is reported that the women were predominantly white. Significant organisational changes coincidentally occurred within the hospital during phase four of the study, making interpretation of the results more difficult. The authors report information about the percentage of total breastfeeding women recruited at each stage: this was 74%, 68%, 68% and 58% for the four phases of the research. Data collected at two weeks postpartum reflect significant differences in exclusive and any breastfeeding; at six weeks postpartum no significant differences were detected. An RCT of the use of hands-off positioning advice from midwives is noted in the list of ongoing studies in Appendix 3. The Committee on Breastfeeding of the Manitoba Pediatric Society reported work comprising a range of educational activities targeted at both professionals and the general public (Manitoba Pediatric Society, 1982 [77]). This work was carried out during 1979. Doctors, public health nurses and home economists received mailings of guidelines related to preparing for and establishing breastfeeding. Seminars were provided by members of the committee to staff in urban and rural hospitals, health departments, the medical school and at provincial health conferences, although there is no information about the extent to which these were standardised. A leaflet, One to Grow On, was adopted as the agreed and official standard for teaching about infant feeding to professionals. For the general public, the committee contributed to the development of a poster, Come Close to Your Baby Be a Nursing Mother. This poster was displayed in doctors offices, hospitals and other healthcare premises. Junior and senior high schools received information encouraging the incorporation of education on infant feeding into curricula. Public newsletters, press releases and media interviews were also utilised. Rates of breastfeeding before and after the intervention were determined by telephone interview following postal information to parents when babies were approximately six months old. Some follow-up was carried out by public health nursing staff. Feeding method was collected from 277 and 249 mothers in the 1978 and 1979 groups respectively. Mothers provided information about infant feeding on discharge from hospital, reasons for choice of feeding method and discontinuation of breastfeeding and when the latter occurred. Women providing information 122

came from Winnipeg and rural Manitoba, with a small proportion from native Canadian reservations. The proportion who came from each of these areas was similar for the before and after samples. No significant differences were detected in breastfeeding at discharge between the two samples (57% vs 56%). Significantly more infants from urban communities were breastfeeding at two months compared to rural-based infants (50% vs 36%). At two months, increases in breastfeeding were seen after the campaign for Winnipeg residents but fewer of the rural residents were breastfeeding. Numbers are not reported, nor tests of statistical significance. The pattern appears similar for Winnipeg residents at four and six months. For rural settings, breastfeeding was still less frequent after the intervention but the difference was less pronounced. Two primary care nurses report work conducted in their paediatric clinic in Barcelona (Matilla-Mont and RiosJimenez, 1999). The clinic served a socio-economically deprived area of Barcelona and the aim of the breastfeeding promotion programme, provided in 1993, was to increase breastfeeding initiation rates at three months and to increase duration of breastfeeding. The training was provided to healthcare and other staff working at the clinic. The breastfeeding programme included coordination with midwives, contributing to birth preparation classes, changes in the dynamics of paediatric clinic consultations, use of a breastfeeding protocol, coordination with the wider team, forming a group of mothers, contact with other groups that support women with their breastfeeding (eg La Leche League), and coordination with hospitals to which babies are referred from their clinic. Changes in how the paediatric clinic team worked included: meeting women during their pregnancy; a first clinic visit during the first week of life with the offer of a home visit to a mother who is unwell and could not attend the clinic; increasing the frequency of clinic visits and promoting breastfeeding in the community; and liaison with other groups who support women with breastfeeding. Information was collected relating to 96 babies born between October 1993 and September 1994 and 113 babies born between October 1995 and September 1996. However, important contextual information is not available within the paper; for example, whether the function of the clinic is for routine child health surveillance or a secondary level facility; usual patterns of care; and the usual content of clinic visits.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Information about how the education was provided and who designed and delivered it is not provided and there is no information about methods of data collection. Breastfeeding rates increased after the intervention; reported at 0 month (58% vs 73%) and at three months (31% vs 50%). The mean duration of breastfeeding was also reported as four months prior to the intervention and six months following. Concerns about static breastfeeding rates in Oxford in 1991 triggered multi-agency partnership working to support breastfeeding (Stokoe et al., 1994 [79]). A total of 353 women who had given birth in Oxfordshire in March 1993 completed questionnaires offered by health visitors at the first postnatal visit approximately two weeks after birth. These identified problems most commonly experienced by women as difficulties with latching, sore nipples and insufficient milk supply. Eleven training sessions were subsequently provided by the hospitals breastfeeding adviser and targeted at midwives; 90% of hospital midwives attended the training. The survey was repeated in a second sample of women who gave birth in September 1993; 356 women returned questionnaires. Rates of breastfeeding initiation were similar between the two groups. Exclusive breastfeeding at the first postnatal visit was marginally higher in the second group (55% vs 58%) but fewer women reported changing to bottlefeeding-only by that point (19% vs 14%). Reports of breastfeeding problems were higher for the second group of women. There is no information about the content of the training programme and the authors comment that difficulties with staffing levels may have been a factor.

women during pregnancy or following birth. Other studies reflect work that seems to focus on changing philosophies (Gainotti and Pagani, 1980 [73]). One evidence-based practice initiative comprised the introduction of evidence-based guidelines supported by other strategies (Grant et al., 2000 [74]). No studies of training for breastfeeding educators/ counsellors or peer supporters were included in this review. One study involving the impact of a programme called Maternal Infant Health Outreach Workers on rural, very low income, isolated communities was identified in the search but excluded from this section as breastfeeding duration outcomes were not reported (Clinton, 1988). Positive effects on the proportion of women continuing breastfeeding were seen following the introduction of the 18 hour UNICEF training in eight Italian hospitals. However, Durand et al.s pilot study, which appeared to comprise a similar intervention, had no effect on breastfeeding duration (Durand et al., 2003 [72]). Apparently significant increases in the proportion of women breastfeeding at 12 weeks were reported by Gainotti and Pagani (1980) [73]. While improvements to both philosophy and breastfeeding rates appear positive these figures should be interpreted with caution as the context of care and care-giving practices appear very different from current UK practice. Grant et al. (2000) [74] report a well-designed evidencebased practice initiative that, in the context of other research related to professional practice change, should have had a high chance of achieving an improvement in continuation of breastfeeding at three months. The authors ascribe this lack of impact, in part, to issues related to hospital postnatal care and a need for that to be reviewed. However, while no changes in initiation or duration of breastfeeding occurred, significant changes were seen in professional practice that had the potential to improve womens experiences. This study raises the issues that culture, as well as practice, may need to be addressed. Hartley and OConnors intervention did not achieve the doubling in breastfeeding rates that they had set but they did see significant changes in breastfeeding rates at two weeks (Hartley and OConnor, 1996 [75]), particularly among certain sub-groups of women. 123

4.1.4 Discussion and conclusions


A range of educational interventions have been identified with, apparently, varying effects. The majority of these interventions were education for health professionals aimed at increasing knowledge and changing professional practice in support of breastfeeding (Cattaneo and Buzzetti, 2001 [71]; Durand et al., 2003 [72]; Hartley and OConnor, 1996 [75]; Manitoba Pediatric Society, 1982 [77]; Matilla-Mont and RiosJimenez, 1999 [78]; Stokoe et al., 1994 [79]). Some of the educational interventions were also offered to other workers, eg clinic secretaries who come into contact with

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Different directions of effects for urban and rural women were seen in the Manitoba study when duration was reported at two, four and six months (Manitoba Pediatric Society, 1982 [77]). The reason for that difference in direction of effect is unclear. Ingram and colleagues research (2002) [76] reflected that the teaching of breastfeeding using a hands off method by the midwives can be cascaded from a trainer to midwives to women following an approach that seems relatively inexpensive in resource requirements, including staff time. While the results appear encouraging, the different proportions of total breastfeeding women recruited to the research and providing data contribute an ascertainment bias. Replication of this research would merit consideration in settings with socially and demographically diverse populations. While the primary outcome for this review has been the effect on duration of breastfeeding, other changes that appear important overall in achieving a positive breastfeeding culture have been identified in some studies. Thus, while Durand et al. (2003) [72] found no difference in the proportion of women breastfeeding at 12 weeks, changes were observed after the educational intervention in less separation of babies from their mothers overnight, less use of formula and improvements in information to women. Similarly, as described above, Grant et al. (2000) found other significant changes in practice including the provision of information and more skin-to-skin contact that could contribute positively to womens overall perceptions and experiences of childbirth. The studies reviewed above include little information about the preliminary education received by health professionals in non-UK settings. Interpretation of effect must be made with caution when the knowledge expected of practitioners, for example at the point of professional qualification, is not known. Similarly, no information was identified evaluating the effectiveness of different approaches to breastfeeding educators or counsellors education when measured by duration of breastfeeding. Many of the studies reviewed have methodological limitations. This may, in part, be because research resources were not available to care providers keen to try to improve services in their workplace. The majority of studies were not planned using power calculations, data were presented in ways that make interpretation difficult and tests of statistical significance are frequently not 124

included. The point at which duration of breastfeeding was measured varies widely and a key point is the lack of repeated follow-up in many of the studies that would have allowed comment on whether initial impacts were sustained over time. In some studies, there is insufficient evidence about the context and content of services that appear to differ significantly from those provided in current UK settings; this limits the utility of that research. In other studies, incomplete information about the content of education programmes prevents transfer to or replication studies in UK settings. In addition, the use of before-after study designs provides opportunities for biases to occur when selecting participants and with outcome measurement, together with other changes in practice that pass unreported or confounders at the two time points. The methodological limitations of the studies reviewed are particularly disappointing as a significant proportion of these studies included women from disadvantaged groups who are the focus of a current UK target for increasing rates of breastfeeding. There is also no information available related to the direct and indirect costs of providing educational interventions or of the cost effectiveness, from a wider perspective, of such interventions. These are important issues for service providers and commissioners of healthcare and health professional education. Changes in institutional philosophy in a direction that supports breastfeeding are not always accompanied in the short term by increased duration of breastfeeding. The background culture and traditions of a workgroup or institution are powerful and it may be that significant resources (educational, personnel and change management) that address these issues are required to achieve an impact in some settings. One additional study, published after the period of our literature search, investigated the use of evidence-based guidelines as a component of protocol-based postnatal care provided by midwives over an extended duration. One of the 10 guidelines related to breastfeeding. This high quality, cluster RCT, conducted in the UK, aimed to address postnatal health problems that are frequently experienced by new mothers. While other benefits to womens health accrued from the intervention, no differences in breastfeeding continuation were detected at four or 12 months (MacArthur et al., 2003). From the studies identified, there appears no single way that consistently achieves changes in professional practice

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

that support breastfeeding and that impact positively on breastfeeding duration. However, from the apparently methodologically sounder studies, Cattaneo and Buzzetti (2001) [71] provide evidence of impact of the UNICEF training that could be applied in practice. Replication in contemporary UK settings would be useful, especially if carried out within a controlled design. It does not appear appropriate to recommend implementation of other interventions in UK practice as other studies with positive impacts either lack methodological rigour or are contextually so different that they cannot be translated into UK practice. The Healthy Start initiative requires health professionals (midwives, health visitors) to provide support to women related to breastfeeding. At this stage, an appropriate educational approach to equip them for that aspect of practice does not exist. There is therefore an urgent need for research to provide the essential underpinning to equip healthcare professionals to deliver on this aspect of central policy. What is surprising is the absence of reference to general education models or models of education optimal for adult learners in studies of educational interventions. While some of the interventions may have been underpinned by educational theory this was not made explicit. In addition to methodological rigour, it would appear important for future studies of educational interventions to incorporate the evidence base from adult education. There is also a surprising lack of development of rigorous methods of assessing breastfeeding knowledge and support skills. This is true of studies included here and the far greater number of studies without breastfeeding outcomes but which have some education outcomes (eg knowledge, attitude, behaviour in practice).

4.1.5 Future research


As described above, further work exploring the impact of UNICEF training in UK settings, using rigorous designs, would be useful. Research is required to identify the most appropriate methods of education for health professionals who will be supporting women receiving care following the implementation of the Healthy Start policy.

Evidence from generic educational theory and adult education methods should form the base for the development of new approaches to teaching healthcare professionals and others who support breastfeeding women. The outcomes of educational programmes for healthcare professionals should be compared when delivered by different trainers; for example, experienced breastfeeding counsellors and educators with a healthcare background. Further work exploring the Best Start approach from the US, possibly with modification of the training package to avoid didactic teaching, using rigorous methodologies, would allow comment for UK practice and allow further exploration of the promising findings for certain sub-groups of women with follow-up extended beyond two weeks. Further evaluation of the training of midwives in the hands off technique for supporting breastfeeding women would be appropriate. Evaluating the effects of interventions that are mediated through another individual or professional group is challenging. Further evaluation of interventions aimed at educating those who provide clinical care or advice to women should adopt rigorous designs, preferably RCTs with appropriate pre-determined sample sizes. Future research should include collection of information related to the costs of providing educational interventions and, if possible, costeffectiveness studies using whole systems perspectives. Many of the studies above took place before evidencebased practice was widely adopted within healthcare systems. It would seem appropriate that future research investigating different approaches to changing professional practice should incorporate emerging theory relating to evidence-based practice and achieving cultural change to have the greatest chance of success. Such an approach has been recommended by others (Grimshaw et al., 2004). Educational and psychological research can be used to develop reliable and valid assessment measures which are also pedagogically appropriate, ie linked to the expected knowledge achieved through training. Changes in breastfeeding educator/counsellor training, or the development of new roles or programmes for this group, should also take place within well-designed research programmes that include assessment of impact on breastfeeding duration.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

125

126 Participants and any sub-groups 8 hospitals (3 general and 1 teaching) in 2 groups, UNICEF 18 hour training course (training for North and South Italy. 571 health workers; 2,669 trainers) + 2 hour WHO counselling training mother-infant pairs, babies >2,000g 12 weeks 3 months and 6 months Intervention and timing in studies eligible for this review Timing of breastfeeding duration outcomes Population peasant farmers, craftspeople, artisans, Increasing staff awareness of the psychological and biological aspects of breastfeeding teachers, 325 mother-infant pairs, healthy neonates; nursery staff not described Evidence-based guidelines supported by education sessions, newsletters, audit and feedback and educational materials 3 months 12 weeks 90 mother-infant pairs in a disadvantaged community Best Start, a breastfeeding education programme 2 weeks for health professionals and clerical staff comprising didactic education Hands off technique taught to hospital midwives 2 weeks and 6 weeks and healthcare assistants in 45 minute training workshops and subsequently taught to mothers 6 weeks Education provided by the committee of the Manitoba Pediatric Society including seminars to physicians, public health nurses and home economists; guidelines provided plus information to general public 3 months 277 and 279 mother-infant pairs (before-after intervention) in Manitoba (urban, rural and native Canadian reservations) A paediatric clinic in a socio-economically deprived 2 public health nurses raised awareness of the importance of breastfeeding among clinic staff area of Barcelona n=209 women and changed clinic organisation to improve coordination and access for mothers All Oxfordshire women who gave birth during two 1 month periods n=1,007 11 training sessions for midwives led by the 2 weeks hospital breastfeeding adviser; 90% of midwifery workforce attended. Content addressed problems identified by women in first survey

Table 4: Included studies educational interventions for health professionals

Name of Country Date of authors and publication study number

Study design

Cattaneo [71] 2001 and Buzzetti

Italy

Before-after

Durand et al. [72]

2003

France

Before-after survey Teaching hospital. 100 women (50 women before, Standardised 3 day seminar total 73 staff Observational attended: midwives, paediatricians 50 after who gave birth to a live baby, predominantly European

Gainotti and Pagani [73]

1980

Italy

Before-after

Grant et al. [74]

2000

UK

Before-after A hospital serving a deprived area of South incorporated quasi- London n=931 women experiment audit, action research

Hartley and OConnor [75]

1996

US

Before-after crosssectional

Ingram et al. [76]

2002

UK

Before-after non1,400 South Bristol mothers on 1 postnatal ward, randomised, population with low uptake of breastfeeding; age prospective, cohort 16-44 years, living in lower socio-economic area of city, predominantly white, babies 35-43 weeks

Manitoba Pediatric Society [77]

1982

Canada

Before-after cross-sectional

Matilla-Mont and RiosJimenez [78]

1999

Spain

Before-after

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Stokoe et al. [79]

1994

England

Before-after cross-sectional

Overview

Summary of results
This section provides a summary of the results of the reviews in a set of tables. Criteria for the final allocations to these tables were described in detail in the Introduction. This summary derives from the sections in the review and each intervention shown here is related to the section in which it is analysed. There is inevitably an element of subjectivity in the allocation of interventions to these tables; the reader can scrutinise the information given in each section and the data tables in Appendix 13 (on website) if they wish to reconsider our allocation. Each table is organised according to the timing of the intervention from pregnancy, through childbirth, to postpartum and also includes wider social and community interventions, multifaceted interventions, and educational interventions and professional practice change.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

127

Overview table 1: Forms of care/practices/policies that have been shown to be effective/beneficial for enhancing breastfeeding duration Timing of form of care/ practice/policy Pregnancy Labour and birth Postnatal hospital stay Skilled breastfeeding support, peer or professional, proactively offered to women who want to breastfeed Preventing the provision of discharge packs containing formula feeding information and samples Unrestricted feeding from birth onwards* Unrestricted mother-baby contact from birth onwards* Kangaroo care/skin-to-skin care from birth onwards* Avoiding additional supplementary fluids unless medically indicated Regular breast drainage/continued breastfeeding for mastitis* Antibiotics for infective mastitis* Postnatal care in community Ongoing community care Wider policy interventions Multifaceted interventions (across time periods and types of interventions) Educational interventions and changing professional practice
* Denotes that the evidence base for this is described in Renfrew et al. (2000)

Form of care/practice/policy

Use in deprived groups examined?

Section

No No Yes Yes Yes No No No No No

1.1 2.1 3.3 3.3 3.3 3.3 3.4 3.4 1.1 1.1

Skilled breastfeeding support, peer or professional, proactively offered to women who want to breastfeed. Skilled breastfeeding support, peer or professional

128

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Overview table 2: Forms of care/practices/policies that appear to be promising, well grounded in theory, and with some research to substantiate, for enhancing breastfeeding duration Timing of form of care/ practice/policy Pregnancy Form of care/practice/policy Group, interactive, culture-specific education sessions Group education session on positioning and attachment Antenatal education individually tailored to womens needs, for low-income women Labour and birth Immediate postnatal care Basing prevention and treatment of sore nipples on principles of positioning and attachment Cabbage leaves/extract for treatment of engorgement Systemic antibiotics for infected nipples Not separating mothers and babies for treatment of jaundice* Postnatal care in community Self-monitoring daily log for women from higher socio-economic groups Combination of supportive care, teaching breastfeeding technique, rest and reassurance for women with insufficient milk* Division of the frenulum in infants with signs of congenital ankyloglossia and breastfeeding difficulties Ongoing community care Staff education and training Wider policy interventions National policy of encouraging maternity units adherence to BFI Regionally/nationally determined targets with supporting activities, and/or penalties and/or incentives Multifaceted interventions (across time periods and types of interventions) Tailored antenatal education, combined with proactive postnatal support in hospital and community Combining antenatal education with partner support, postnatal support and incentives for women in low-income groups Educational interventions and changing professional practice
* Denotes that the evidence base for this is described in Renfrew et al. (2000)

Use in deprived groups examined? Yes Yes Yes No No No No No No

Section 1.2 1.2 1.3 3.3, 3.4 3.4 3.4 3.5 1.2 3.4

No

3.5

No No Yes

2.1 2.1 1.3

Yes

1.3

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

129

Overview table 3: Forms of care/practices/policies that are unproven, and require further research; and other important research questions, in relation to impact on breastfeeding duration Timing of form of Form of care/practice/policy care/practice/policy Pregnancy Giving negative messages about mixed breast and formula feeding to women from deprived groups Testing group, interactive, culture-specific education sessions for women from low-income groups as part of Sure Start schemes Short questionnaire to reinforce education session on positive aspects of breastfeeding Impact of current forms of promotion of formula milk on duration Section 1.1 Comment Some studies suggest that exclusive breastfeeding is especially challenging for women from deprived groups, and promoting exclusive breastfeeding may be counterproductive Opportunity to monitor and evaluate such provision, locally designed and delivered in different ways in different localities, as part of national programme

1.2

1.2 2.1 Of especial interest in regard to women who are unsure of their feeding plans. It would be unethical to conduct this in countries already adhering to the WHO code

Antenatal expression of colostrums Niplette device for inverted and non-protractile nipples Surgery for inverted nipples General support and care for women with inverted and non-protractile nipples Antenatal treatments for women with history of nipple problems, eg dermatitis, eczema Care for women with negative feelings about breastfeeding related to history of abuse Views of women on care needed to advise on breastfeeding in pregnancy Labour and birth Birth centre care Inclusion of breastfeeding duration outcomes in intrapartum research Immediate post-birth care for breastfeeding for women having caesarean section Effects of birth experience and pain relief in labour on the babys feeding behaviour and on the duration of breastfeeding Delayed clamping of the umbilical cord (up to three minutes after the birth) Immediate postnatal care Self-monitoring daily log

3.1, 3.4 3.1 3.1 3.1 3.1 3.1 3.1 1.5 3.2 3.2 3.2, 3.3 See Overview table 5 in relation to other treatments for this condition See Overview table 5 in relation to other treatments for this condition

3.2 1.2

Needs to be examined in relation to management of the third stage more widely Needs to be tested for use with additional strategies to achieve compliance from women from deprived groups shown to be of use in higher socio-economic groups

Mother and baby bed-sharing/clip on cots Partnership care including baby in hospital room

1.5 1.5

130

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Overview table 3: Forms of care/practices/policies that are unproven, and require further research; and other important research questions, in relation to impact on breastfeeding duration (cont.) Timing of form of Form of care/practice/policy care/practice/policy Immediate postnatal care (cont.) Breastfeeding-only rooms/wards Staffing levels and skill mix in hospital to enable breastfeeding Length of hospital stay and post-discharge support Ways of bridging gap between hospital and community support for breastfeeding women Use of cup or spoon instead of bottle if additional fluids are necessary Continuous epidural analgesia for women following caesarean section Other forms of pain relief for women following caesarean section Contribution of kangaroo care/skin-to-skin contact in vulnerable sub-groups When are additional fluids medically indicated? Reasons for, sequaelae of, reluctance to wake and feed in healthy babies, and testing of appropriate interventions Are some babies especially susceptible to feeding problems if fed by bottle/teat, and how can they be identified? In what form should additional fluids be given, if medically indicated? Does supplementation have different sequelae in different countries/communities/sub-groups? Ways in which weight and feed charts can be used to support or undermine breastfeeding Are some sub-groups especially vulnerable to the use of pacifiers? Identification of causes of sore nipples, and testing appropriate interventions Breastmilk, lanolin, hydrogel dressings, tea bags, breast shells to treat sore nipples Oral proteolytic enzymes for engorgement Simple measures to treat engorgement eg warmth, rest, analgesia, breast binder, massage, expression Identification of causes of sore breasts in absence of engorgement and mastitis, with testing of appropriate treatments Section 1.5 1.5 1.5 Breastfeeding protective care should be examined in relation to existing post-discharge support and socio-economic group Gap between hospital and community midwives, and also between community midwives and health visitors Comment

1.5

3.3 3.3 3.3 3.3 3.3 3.3 Care needed to define intervention and its timing

3.3

3.3 3.3 3.3 3.3 3.4 3.4 3.4 3.4 Not available in UK at present All used in current practice

3.4

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

131

Overview table 3: Forms of care/practices/policies that are unproven, and require further research; and other important research questions, in relation to impact on breastfeeding duration (cont.) Timing of form of Form of care/practice/policy care/practice/policy Immediate postnatal care (cont.) Work to distinguish different types of milk insufficiency, with testing of appropriate treatments Relationship between blood loss and milk production problems Factors underlying development of mastitis, with development of appropriate interventions, to include differential diagnosis of infective vs noninfective Examination of skill mix, training and settings for frenotomy for ankyloglossia Neurodevelopmental outcomes in babies with asymptomatic hypoglycaemia Studies to inform care of babies who are reluctant to feed, or who do not wake to feed Lactation instruction combined with palatal obturator and a Haberman feeder for babies with cleft lip with or without a cleft palate Modified Hotz plate for babies with both cleft lip and palate Interventions to limit the impact of diagnosis and treatment of jaundice Examination of late onset jaundice and relationship with breastfeeding duration Postnatal care in community Telephone vs face-to-face support for women in all groups Section 3.4 Comment

3.4 3.4

3.5 3.5 3.5 3.5 Large RCTs will not be possible in this field

3.5 3.5 3.5 1.1

Large RCTs will not be possible in this field

Need to examine the timing of such support (eg can telephone support work in the absence of early face to face support?). Does one form or the other work best in deprived groups? Is a combination of both essential? This could be examined as part of breastfeeding support in Healthy Start programme implementation of different forms of education/support to be developed locally, and monitored/tested across the national programme

Ways of supporting women in communities where breastfeeding is not the norm, to breastfeed exclusively

1.1

Intensive, comprehensive home visiting by health visitors Additional postnatal support interventions (offered to women regardless of their infant feeding intention or practice)

1.5 1.1 Existing studies were not powered to examine breastfeeding outcomes; although no differences have been found in relation to breastfeeding, this may be different in larger studies, or differently designed general postnatal interventions Studies need to examine the content of this care, as well as its timing Must take into account the current services provided by hospital and community

Care by GP at one week after discharge Mother-led postnatal care Theory based research to examine the postnatal care needs of specific sub-groups

1.5 1.5 1.5

132

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Overview table 3: Forms of care/practices/policies that are unproven, and require further research; and other important research questions, in relation to impact on breastfeeding duration (cont.) Timing of form of Form of care/practice/policy care/practice/policy Postnatal care in community (cont.) Ways in which growth charts can be used to support or undermine breastfeeding All forms of care for insufficient milk, especially in vulnerable groups Interventions to resolve inconsolable crying/colic, while supporting breastfeeding, aimed both at babies and families Interventions to support breastfeeding in babies with Downs syndrome, including interventions specific to the range of deformities associated with this syndrome Ongoing community care Wider social/political issues Ways of supporting women from all groups to breastfeed, at all and exclusively, for longer than three months Workplace support for breastfeeding mothers on-site nurseries, flexible breaks, facilities for expression, loan of breast pumps Patterns of milk expression and feeding to support breastfeeding when mothers have returned to work Workplace education and support for fathers Impact of interventions aimed at community groups including schools and public that may influence initiation and duration Impact of media campaigns that may influence initiation and duration Interventions to enable and support breastfeeding in public Interventions to promote positive attitudes to breastfeeding among children Impact of employment leave Section 3.3 3.4 3.5 Comment

3.5

Large RCTs will not be possible in this field

1.1

Many existing studies do not measure outcomes beyond three months

1.4

1.4

1.4 1.4 Must be theoretically based. Avoid interventions that might raise rates among more affluent, thereby increasing inequalities Must be theoretically based. Avoid interventions that might raise rates among more affluent, thereby increasing inequalities

1.4

1.4 1.4 2.1 It would be unethical to conduct a study in which current policy provision is reduced. Such studies can only be conducted in countries where provision allows for significant enhancement eg increase in employment leave It would be unethical to conduct a study in which current policy provision is reduced. Such studies can only be conducted in countries where provision allows for significant enhancement eg implementation of aspects of the WHO code not yet accepted It would be unethical to conduct a study in which current policy provision is reduced. Such studies can only be conducted in countries where provision allows for significant enhancement eg offering policy of flexible breaks to all women

Impact of implementation of the WHO code

2.1

Impact of workforce policies to protect breastfeeding women

2.1

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

133

Overview table 3: Forms of care/practices/policies that are unproven, and require further research; and other important research questions, in relation to impact on breastfeeding duration (cont.) Timing of form of Form of care/practice/policy care/practice/policy Wider social/political issues (cont.) Impact of restrictions on provision of formula, eg by prescription Section 2.1 Comment It would be unethical to conduct a study in which current policy provision is reduced. Such studies can only be conducted in countries where provision allows for significant enhancement ie where restrictions on formula provision do not yet exist It would be unethical to conduct a study in which current policy provision is reduced. Such studies can only be conducted in countries where provision allows for significant enhancement ie where formula promotion is currently allowed in any form (including follow-on milk promotion) This could be tested as part of Sure Start schemes, and/or as part of the Healthy Start programme This could be tested as part of Sure Start schemes

Impact of promotion of formula milk

2.1

Multifaceted intervention, across time periods and types of interventions

Ways in which informal/family support can be used in conjunction with professional and peer support, especially for women from disadvantaged groups Ways in which women from low-income groups can be supported in exclusive breastfeeding using combinations of education, support and incentives Inter-agency support for breastfeeding

1.1, 1.3

1.3

1.4, 4

Including combined hospital/community/Sure Start/media/workplace interventions and professional education Built on consumer research to identify important messages Will require monitoring of current practice, and cross-region, cross-country evaluations

Social marketing approaches that may influence initiation and duration Impact of national/regional breastfeeding strategies Ways in which women who smoke and/or use recreational drugs can be encouraged to continue to breastfeed Ways in which women who take prescribed or over-the-counter medications can be encouraged to continue to breastfeed Ways in which women who have undergone breast surgery can be encouraged to continue to breastfeed Educational interventions and changing professional practice How to teach/support women to encourage painfree, effective feeding

1.4, 4 2.1 3.6

3.6

Will include examination of prescribing behaviour

3.6

Will include cosmetic and therapeutic procedures. Should include development of surgical techniques to preserve breastfeeding ability Includes questions of timing and type of input, and skill mix, and further exploration of midwives using hands off technique to teach positioning and attachment Theory based approaches to education needed Theory based approaches to education needed Theory based approaches to practice change needed Theory based approaches to practice change needed

3.3, 4

Effectiveness of UNICEF training in UK settings, hospital and community Education and training for peer supporters and breastfeeding counsellors Ways of changing staff behaviour to encourage early feeding and close contact Ways of changing health service culture to support breastfeeding Ways of preparing staff to work in culturally appropriate ways

4 4 3.3, 4 4 1.2, 4

134

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Overview table 4: Forms of care/practices/policies that may be ineffective or harmful for enhancing breastfeeding duration (as shown by good, though not conclusive, evidence) and should not be used without further evidence of effectiveness Timing of form of care/ practice/policy Pregnancy Form of care/practice/policy Self-help manual used alone Antenatal education by paediatrician Providing formula company-produced materials on infant feeding in early pregnancy Labour and birth Postnatal hospital stay Separating mothers and babies for treatment of jaundice* No 3.5 Use in deprived groups examined? Yes Yes No Section 1.2 1.2 2.1

Postnatal care in community

Written educational materials used alone GP clinic visit at one week postpartum Home visit by community nurse following early discharge Dopamine antagonists for insufficient milk*

No No No No

1.2 1.5 1.5 3.4

Ongoing community care Wider policy interventions Multifaceted interventions, across time periods and types of intervention Educational interventions and changing professional practice

Dopamine antagonists for insufficient milk*

Bold font denotes practices that are likely to be harmful * Denotes that the evidence base for this is described in Renfrew et al. (2000)

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

135

Overview table 5: Forms of care/practices/policies that have been shown to be ineffective or harmful for breastfeeding duration and should be abandoned or not introduced Timing of form of care/ practice/policy Pregnancy Form of care/practice/policy Conditioning nipples* in pregnancy Hoffmans exercises for inverted and non-protractile nipples* in pregnancy Breast shells for inverted and non-protractile nipples* in pregnancy Labour and birth Immediate postnatal care Restricting the timing and/or frequency of breastfeeds* Restricting mother/baby contact from birth onwards* Routine use of supplementary fluids Provision of discharge packs containing samples or information on formula feeding Topical agents for the prevention of nipple pain* Breast pumping before the establishment of breastfeeding in women at risk of delayed lactation Postnatal care in community Ongoing community care Wider policy interventions Multifaceted interventions, across time periods and types of interventions Combined antenatal education and limited postnatal telephone support for high income women and women who intend to breastfeed (existing high rates suggest resources better spent elsewhere) No 1.3 Yes Yes No Yes No No 3.3 3.3 3.3 1.3, 2.1 3.3 3.3 Use in deprived groups examined? No No No Section 3.1 3.1 3.1

Bold font denotes practices that are likely to be harmful * Denotes that the evidence base for this is described in Renfrew et al. (2000)

136

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Limitations and strengths of this review

This series of linked reviews has attempted to examine the range of studies relevant to assessing ways of enabling women to breastfeed.

Strengths
The strengths of this series of reviews include: The individual approach to each section of the review, where methods were designed to be appropriate for the topic. In this way, we have tried to avoid a one size fits all approach to very different topic areas and to include the range of appropriate study designs The transparent approach to quality assessment. We avoided any attempt to use a quantitative scoring system for the included studies, instead developing a multifaceted approach to assessing each study. This approach is explained in the section on data quality in the Introduction. These reviews have looked across the whole spectrum of potential interventions that can have an impact on breastfeeding, including public health, public policy, clinical issues and staff training. Such an integrated approach is not common but it is essential in addressing this topic. It is only by joining up these different areas that real differences will be made. The reasons for breastfeeding rates being low are complex and multifaceted, and intervening to raise those rates will require an integrated approach. It is also clear that there are real synergies between these topic areas and that progress will only be made in some areas if others are addressed at the same time. For example, developing peer counselling in communities where rates are very low is likely to be more effective if there are supportive local media campaigns to normalise breastfeeding; if care given in the local hospital is evidence-based and harmful routines are abandoned; if public policy is supportive of breastfeeding in public; and if employers are sympathetic and supportive. The review team included the professional disciplines of public health nutrition, community paediatrics, midwifery,

Limitations
The review methods used have limitations. One limitation is that for Section 3 (with the exception of 3.2) searches were conducted only after 1995 and we relied on previous reviews, especially Renfrew et al. (2000), up to that point. Material from that review has been incorporated (with permission). This was to avoid the duplication of work and to keep the current review within the resources available. This was felt to be an appropriate approach, as Renfrew et al. (2000) has had positive feedback from colleagues in practice, policy and research settings in this country and others. It does introduce the issue that the work in Section 3 has not all been conducted over the same time period, to exactly the same specification, and by exactly the same team. Interested readers can access data extraction tables and full critique from that previous review. This review was intended primarily to inform those working in UK settings. Accordingly, we have used examples of UK problems and systems throughout. It is likely, however, that the findings are appropriate for most developed countries. The material we reviewed was from all developed countries for Sections 1, 2, and 4, and from all countries for Section 3; searches were not limited by language and 12 papers were fully translated, resulting in three translated papers being included. It is hoped therefore that despite the obvious UK focus in the discussion, this work will be of relevance to colleagues in other countries.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

137

and clinical psychology, and the academic disciplines of social science, health services research and anthropology, and allowed us to draw on a rich theory base in the analysis of this wide range of studies. Members of the team also have experience of conducting research using a wide range of methods including large RCTs, non-RCTs, before-after studies, clinical and laboratory research and in-depth qualitative studies. They have also conducted studies with women and families from deprived groups, and breastfeeding development work in areas where breastfeeding rates are low. This experience allows us to understand from first-hand knowledge the challenges of research in this complex field and to make recommendations for the ways in which future research studies should be conducted. We have worked in several different countries and have experience of working to create change in local community and hospital settings, in relation to childbirth, maternal and child nutrition including breastfeeding, and other aspects of public health. We have also conducted several systematic reviews of this and related topics using different methods over many years and we draw on this experience throughout the review. We welcome discussion on any aspect of the methods and presentation of this work.

There is an urgent need for a programme of research into clinical problems, including insufficient milk, sore nipples, engorgement and the breastfeeding needs of babies and mothers with particular health needs. The prevention and treatment of clinical problems is of course inextricably linked with the preparation and skills of health professionals. The evidence identified here suggests that work is urgently needed to identify effective ways of improving knowledge, skills and practices for all those who work with breastfeeding women. In relation to public policy, there remains a range of important questions, perhaps the most important of which is identifying ways of enabling women to breastfeed without embarrassment wherever and whenever they need and wish. In spite of the importance of this topic area to public health and to the work of the health service, few of the studies reviewed included outcomes related to costs for families, employers or the health services. From the material reviewed here it is not possible to inform the debate on cost effectiveness of interventions to increase the duration of breastfeeding. Two other important findings emerge from these reviews when looked at as a whole. First, there are effective interventions and ineffective interventions (summarised in the overview tables), yet care based on this evidence is not widespread. Second, a gap in the evidence base identified across all the sections is an understanding of the views of those most involved childbearing women and their families, and the staff who care for them whose voices are largely silent. This overview attempts to address these two issues and concludes with recommendations for future research.

Discussion and implications


The identification of studies that can inform practice and policy for the support of breastfeeding among women from disadvantaged groups was a priority for this series of reviews. It is regrettable that one of the main findings of this review is the great extent of the evidence gap relating to these groups. Ways of raising breastfeeding rates among groups where the rates are lowest remain to be further explored. Although there are evidence gaps identified across all the sections, it is perhaps in Section 3, clinical issues, and in Section 2, public policy, that the gap seems widest (indeed, only one of the studies included in each section was conducted in the UK). This review, like its predecessor (Renfrew et al., 2000), has identified the fact that not only has clinical research not addressed womens key concerns and problems, but clinical care, by imposing unnecessary and harmful routines, has actively interfered with the successful establishment of breastfeeding. 138

Effective and appropriate care in breastfeeding


There are interventions that are likely to be necessary for everyone (though not sufficient in themselves). These factors include the large-scale policy developments that are very difficult to evaluate but have been associated with positive change in countries where real change has occurred, such as adequate maternity leave and employment protection provision. They also include the assurance that care from the health service, in hospital and community settings, is evidence-based and appropriate. This will require a major shift in education and training for staff, and widespread, wholehearted support for the Baby Friendly Initiative.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

One simple yet clear message from this series of reviews is that many interventions will need to be designed to be appropriate for the setting and participants. A uniform approach is unlikely to work, whether the intervention is peer support (see Section 1.1), antenatal education (see Section 1.2), or hospital-based postnatal care (see Sections 3.3 and 3.4). This should not be surprising; breastfeeding rates vary so widely across socio-economic and cultural groups that interventions are likely to have differential effects. One important implication of this is that by using the evidence base presented here local staff and families can plan a what is likely to work approach for their setting. This should then be monitored to assess the outcomes. It is also clear that there may for some time yet be a tension between achieving the target for raising the rates of any breastfeeding and the goal of increasing the duration of exclusive breastfeeding. This is in part because the rates of any breastfeeding are so low in some communities that it will take time to make a difference. Many women will choose to mix breast and formula feeding as part of their strategy to breastfeed at all, and women from some cultural groups will practise mixed feeding from the start. Enabling exclusive breastfeeding is likely to require much more large-scale social change than is likely to occur in the short term (see Sections 1.4 and 2.1), and policy makers and practitioners will need to be realistic about the timeframe for achieving such changes. It is striking that the interventions found to be effective or promising differ from the normal care currently provided. It is known, for example, that breastfeeding women need additional, breastfeeding-specific support (Section 1.1); this is far from being widely available. Studies suggest that education sessions should be conducted in small groups, using informal, participative methods, and culturally appropriate approaches. This is not the norm in current parentcraft education and student health professionals are not always taught these skills. It has been shown that routine supplementation with other fluids is harmful to the duration of breastfeeding, but the practice continues. Providing evidence-based care to enable women to continue to breastfeed will require widespread changes to the current systems of care. Such changes will need a change in attitude to breastfeeding from senior policy makers through to field-level practitioners. For as long as

infant feeding is seen as lifestyle choice, no real change will be made. But to see infant feeding in this light is to ignore the extensive literature on factors affecting feeding, and the constraints on womens choices in many countries, including the UK, which mean that to breastfeed women need support, protection, encouragement and information. This means that in addition to supporting the Baby Friendly Initiative, services which impact on breastfeeding, such as GP services, and the practices of the NHS as a model employer, need to be addressed so that breastfeeding is the norm across the health service. The increasing literature on the public health hazards of formula feeding when compared with breastfeeding, even in developed countries, demonstrates that this is a crucial issue for healthcare systems and for society as a whole. This review has demonstrated that there are ways of overcoming the problems women have with continuing to breastfeed. The real challenge will be implementing the necessary changes. The extent of the work needed to change the current patterns of infant feeding should not be underestimated. These patterns have been developed over the past century and are now embedded in the thinking and behaviour of several generations of practitioners, and of society as a whole. A coordinated and well-supported programme is needed if real change is to occur. Our analysis of the evidence base suggests that the following are required to create essential change in enabling women to continue to breastfeed: National level coordination of policy, across health, social services, school education, employment, food, and environment Coordination of national with local policy, so that departmental policy is funded, enabled and monitored at the level of, for example, primary care trusts, Sure Starts and acute trusts. Change can be catalysed from bottom up as well as top down, in that users and professionals with front-line experience have valuable knowledge and skills which could be harnessed to suggest areas for policy and practice development, as well as to evaluate their impact Ongoing monitoring of rates of variation in infant feeding, with agreed definitions and timing of follow up, combined with socio-demographic data. Implementing such policies and practices will need the wholehearted support of:

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

139

Clinical professionals in community and hospital settings such as paediatricians, midwives, health visitors, GPs, obstetricians, public health physicians, dietitians, pharmacists, practice nurses, nursery nurses, and neonatal nurses Community based workers such as Sure Start staff, peer supporters, child carers in nursery and home settings, and development workers Managers with responsibility for health and social services and staff Those with responsibility for implementing the Healthy Start scheme, at national and local levels for example, early warning of such major policy developments, and support and lead-in time to prepare staff to deliver the scheme effectively, are all likely to maximise their effectiveness Those with responsibility for strategies to implement changes in wider maternal and child health services, including the National Service Framework for Children (including the maternity module); those developing guidelines including the forthcoming NICE guidelines on postnatal care; and those conducting reviews of topic areas that may impact on breastfeeding, such as care in labour and postnatal care Those with responsibility for collecting health and health service-related data, including audit staff, public health observatories, etc Educators in the fields of health and social services, in health and social service and university settings Schoolteachers and those responsible for the school curriculum in primary and secondary schools Employers in large and small organisations Politicians and policy makers at local, regional and national levels who have influence over issues such as the use of public spaces, community regeneration, addressing inequalities in health and access to care Those with influence over public opinion, including politicians and those working in all kinds of media: newspapers, magazines, television, local/national radio Families and the public at large.

Factors underlying the scarcity of relevant, high quality research include: Studies have not addressed some of the key practical issues in the field Many studies are too small to be useful, or are methodologically flawed Studies examining biomedical questions have seldom considered the fundamental contribution of support, information and advice, the effects of womens views and feelings, and the social and cultural context of breastfeeding Many clinical studies have taken place in a culture when breastfeeding practices and routines were such that a large proportion of problems were a result of mismanagement. As a result, some very simple questions have not been addressed. When examining the field as a whole, it becomes apparent that research in breastfeeding is affected by the same bottlefeeding culture that was discussed in the Introduction in relation to breastfeeding in practice. For example, the inconsistency in definitions of breast and formula feeding suggests a problem with the understanding of the dynamic nature of breastfeeding; the lack of rigour of even large-scale RCTs as compared with trials in other fields suggests a lack of attention to their design; and the predominance of small-scale studies suggests that major research funds are not available for this topic area. These all contribute to problems with this evidence base. Further, there is a common and interesting inversion in the testing of interventions. For example, allowing mothers and babies to have close contact at birth is seen as the intervention group in some studies while separating them at birth is seen as the control. The same is true of routine supplementation and pacifier use, suggesting that such practices are considered so normal that not to do them is the intervention that requires evidence. Perhaps the largest manifestation of the bottlefeeding culture in relation to research is the lack of studies in areas where real problems exist for breastfeeding mothers and babies; the topics have apparently not been considered to be important enough for study. This includes the most common cause of breastfeeding discontinuation worldwide, insufficient milk; baby related problems, including tongue tie/ankyloglossia,

Problems with the evidence base


The volume of research covered by this review might give the erroneous impression that there is adequate evidence available to inform practice and policy. In fact, as has been noted previously (Protheroe et al., 2003; Renfrew et al., 2000; Guise et al., 2003), the most striking finding is the paucity of good, well-designed research to inform an area of healthcare which has profound short- and long-term effects on such large numbers of people. 140

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

jaundice, and inconsolable crying; as well as problems women have with breastfeeding related to their own health, wellbeing, and lifestyle, such as smoking, use of prescription and non-prescription drugs, and breast surgery. Future research should use appropriate designs and consider carefully the range of factors involved in breastfeeding, and funding agencies should be aware of the pitfalls in this field. The lack of relevant research is a problem internationally, but it is a profound problem in the UK. Only 12% of the 80 studies included in this review were conducted in the UK, indicating that the funding of research into ways of enabling women to breastfeed is not a research priority. The plethora of small, descriptive, often unfunded studies in this field, in contrast, indicates that it is important to practitioners and to women. Such enthusiasm could be better harnessed if practitioner-based research were supported by those with research skills. To address the evidence gap, breastfeeding needs to become a priority for a range of funding bodies in the UK.

are much less common, such as cleft lip and palate and Downs syndrome, but which often result in distress and discontinuation of breastfeeding. Strategies need to be developed to encourage the continuation of breastfeeding in three particular groups of women: Those who practise unhealthy behaviours such as cigarette smoking and use of recreational drugs Those who take prescribed or over-the-counter medications for a range of medical conditions Those who have undergone breast enhancement or breast reduction surgery for cosmetic or other reasons. While there is a growing body of literature looking into the associations between early weaning and some of the conditions, there have been no trials of interventions to address the problems. Future research should take a problem-based approach; studies should be developed to address the most relevant and important issues. Assessment of this should include examination of the literature (such as in these reviews), and also the views of those most closely affected by the issues: childbearing women and their families, and the practitioners who work them. It is notable, in a field in which the behaviour and views of women, their families, and society are so crucial, that very few studies have examined both views and experiences, and the outcomes related to breastfeeding duration. Including both breastfeeding outcomes and examination of such views in future studies will help to inform the potential for interventions to make a difference in practice. Studies should include the views of women and men of childbearing age, children, teenagers and older people. In many papers included in these reviews, the description of the interventions used were insufficient to replicate in research or to implement in practice. Much more information is needed about the content of interventions, the training needed to implement them, the participants, and the settings in which they were used. Very few studies included in this series of reviews examined cost outcomes for families, health services, employers or others. The health outcomes related to infant feeding are extensive and expensive, and the design and implementation of interventions to increase 141

Future research
Specific problems with research and gaps in the evidence have been identified in the appropriate sections throughout this report. In addition to these, the following issues need to be addressed. Gaps in the evidence are especially apparent in relation to the services and care needed for women from disadvantaged groups, where the rates of breastfeeding are lowest. Studies should examine the effectiveness of interventions among different disadvantaged groups. This will require careful attention to ways of incorporating people who may speak languages other than English or who have problems with literacy, and it may require methodological development. Ways of preparing and training those who work with women from these groups are also essential. There are also important gaps in the evidence in relation to the leading causes of breastfeeding discontinuation: insufficient milk, and sore nipples and breasts. Future research into clinical problems would benefit from dividing the issues into two major areas: issues that are common and widespread, affecting large numbers of women and babies, such as engorgement, insufficient milk, sore nipples and crying babies; and issues which

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

breastfeeding must be set against this ongoing expenditure. One problem is that costs and savings will not necessarily come from the same budget; savings on health service treatment costs arising from increasing breastfeeding duration may not easily be set against the increased need for in-service training for staff to work effectively with women, for example. Information is needed, from audit and from future research, to inform the cost effectiveness of interventions to increase the duration of breastfeeding. Research in the field is very largely atheoretical. That is, assumptions are implicit but not articulated and not tested, and in some cases interventions have clearly not been thought through from a theoretical perspective. An example of this in the studies of clinical problems is work on drugs to enhance milk supply. Drugs that are known to stimulate prolactin have been tested on groups of women with insufficient milk without evidence to suggest that raising prolactin above a threshold level will have any impact on milk production. Such drugs may have a use in women with prolactin deficiency but they appear to be of no benefit in those with insufficient milk syndrome and there is no theoretical reason to expect that they should have. Similarly, theory based research is scarce in studies examining services and specific breastfeeding support interventions. Many of the interventions reviewed rely at some level on a combination of knowledge, attitude, belief and behaviour change. Psychological models have not been used to inform design of interventions, yet prospective studies referred to (Section 1.4) show that it is possible to pinpoint periods in the course of breastfeeding where women and families may be more susceptible to intervention, and the type of intervention (skills based, or emotional support for example) can also be predicted. In the field of education and practice change, no studies have used emerging theory relating to evidence-based practice and achieving organisational cultural change and none have used theory based approaches to adult education. Research is needed to test out the efficacy of interventions based on empirical and theoretically derived research. This will assist in developing a more rigorous understanding of appropriate interventions; for example, in differentiating those who may need additional support and those who may benefit from specific types of support. Research on the effective transfer of evidence into practice still shows resistance by staff to using evidence. 142

There is therefore a need to apply psychological models capable of describing why staff do not follow best practice and to conduct in-depth studies to examine the real-life use of research evidence in hospital and community settings (Marshall et al., 2003). Breastfeeding is in itself a complex behaviour and interventions to support its continuation are therefore likely also to be complex. The evaluation of complex interventions, especially in health promotion, has been the subject of much debate (eg Nutbeam, 1998; Medical Research Council, 2000). A recent paper (Hawe et al., 2004) suggests a model that we believe could address several of the problems in community based studies of complex interventions in breastfeeding. Using this model, studies would address interventions in different sites that standardised the function and process, rather than the form, of the intervention. For example, rather than using a standard form of information-giving such as a leaflet or video for women from deprived groups, this model suggests that local practitioners could devise an approach, tailored for their population, to give them information about breastfeeding. In this way, RCTs could be devised that take into account local differences in the organisation of services and in population groups. It is an approach that offers scope for further debate. Such research as exists suggests that media representations of infant feeding in the UK largely promote bottlefeeding as the norm and breastfeeding as problematic. Research on developing effective mass media health promotion is scant and could usefully be informed by both the psychology of persuasive communications and the growing body of interdisciplinary media studies, using models such as social marketing. There are many ways that changes in the organisation of health services can influence breastfeeding. But it is clear that most will be implemented without evaluation of the impact on breastfeeding. Where changes have included breastfeeding outcomes, eg in relation to shorter duration of hospital stay, study designs have been inappropriate or measurement too crude to detect changes. Yet large-scale changes in health service organisation are inevitable, eg Patient Choice and managed care packages, and these would be suitable for large-scale studies across many sites, allowing simultaneous testing of system changes and differential impact on potentially vulnerable sub-groups. Health

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

equity audit presents a further opportunity to explore how local organisational levers can be used to ensure the most disadvantaged are accessed and supported. Developing effective policy, with government guidance on implementation and performance management of policy driven initiatives, has only just begun to be explored in developed countries. In England and Wales the new public health targets are one approach. Research is needed to compare policy interventions across different countries, as well as research within healthcare systems, to establish the most promising means of integrating action across sectors and between government and front-line services.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

143

References
(The table number of included studies is shown at the end of the relevant references)

Aasheim, V., Dalhaug, R. E. and Vik, B. (2000). Breastfeeding in Norway. MIDIRS Midwifery Digest 10 (3): 372-4. Ajzen, I. and Madden, T. (1986). Prediction of goaldirected behavior: attitudes, intentions and perceived behavioral control. Journal of Experimental Social Psychology 22: 453-74. Ajzen, I. (2001). Nature and operation of attitudes. Annual Review of Psychology 52: 27-58. Akkuzu, G. and Taskin, L. (2000). Impacts of breast-care techniques on prevention of possible postpartum nipple problems. Professional Care of Mother and Child 10 (2): 38-9. [Table 46] Akre, J. (1989). Infant feeding. The physiological basis. Bulletin of the World Health Organization 67: 108. Alder E. (1984). Postpartum changes in mood and sexuality, and some implications for care. In: Houston, M. J. (ed) Recent advances in nursing No 9 pp70-91. Maternal and infant health care. Edinburgh: Churchill Livingstone. Alexander, J. M., Grant, A. M. and Campbell, M. J. (1992). Randomised controlled trial of breast shells and Hoffman's exercises for inverted and non-protractile nipples. British Medical Journal 304 (6833): 1030-2. American Academy of Pediatrics (1997). Breastfeeding and the use of human milk. Pediatrics 100 (6): 1035-9. Amir, L. H. and Donath, S. M. (2002). Does maternal smoking have a negative effect on breastfeeding? The Epidemiological Evidence. Birth 29 (2): 112-23.

Amir, L. H. and Pakula, S. (1991). Nipple pain, mastalgia and candidiasis in the lactating breast. Australian and New Zealand Journal of Obstetrics and Gynaecology 31 (4): 378-80. Amroth, S. G. (1978). Water requirements of breast-fed infants in a hot climate. American Journal of Clinical Nutrition 31: 1154-7. Anderson, S., Shakya, K. N., Shrestha, L. N. and Costello, A. M. (1993). Hypoglycaemia: a common problem among uncomplicated newborn infants in Nepal. Journal of Tropical Pediatrics 39 (5): 273-7. Anderson, J. W., Johnstone, B. M. and Remley, D. T. (1999). Breast-feeding and cognitive development: a meta-analysis. American Journal of Clinical Nutrition 70 (4): 525-35. Anderson, A. S. (2001). Symposium on nutritional adaptation to pregnancy and lactation. Pregnancy as a time for dietary change? Proceedings of the Nutrition Society 60 (4): 497-504. Anderson, G. C., Moore, E., Hepworth, J. and Bergman, N. (2003a). Early skin-to-skin contact for mothers and their healthy newborn infants. The Cochrane Database of Systematic Reviews, Issue 2. Anderson, P. O., Pochop, S. L. and Manoguerra, A. S. (2003b). Adverse drug reactions in breastfed infants: less than imagined. Clinical Pediatrics 42 (4): 325-40. Aono, T., Shioji, T., Aki, T., Hirota, K., Nomura, A. and Kurachi, K. (1979). Augmentation of puerperal lactation by oral administration of sulpiride. Journal of Clinical Endocrinology and Metabolism 48: 478-82.

144

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Aono, T., Aki, T., Koike, K. and Kurachi, K. (1982). Effect of sulpiride on poor puerperal lactation. American Journal of Obstetrics and Gynecology 143 (8): 927-32. Ardran, G. M., Kemp, F. H. and Lind, J. (1958). A cineradiographic study of breastfeeding. British Journal of Radiology 31: 156-62. Arenz, S., Ruckerl, R., Koletzko, B. and Von Kries R. (2004). Breastfeeding and childhood obesity: a systematic review. International Journal of Obesity Related Metabolic Disorder 28 (10): 1247-56. Astley, S. J. and Little, R. E. (1990). Maternal marijuana use during lactation and infant development at one year. Neurotoxicology and Teratology 12 (2): 161-8. Auerbach, K. and Guss, E. (1984). Maternal employment and breastfeeding: a study of 567 womens experiences. American Journal of Diseases of Children 138: 958-60. Auerbach, K. G. (1990). The effect of nipple shields on maternal milk volume. Journal of Obstetric, Gynecologic and Neonatal Nursing 19 (5): 419-27. Auerbach, K. G., Renfrew, M. J. and Minchin, M. A. (1991). Infant feeding comparisons: a hazard to infant health? Journal of Human Lactation 7: 63-71. Aumonier, M. E. and Cunningham, C. C. (1983). Breast feeding in infants with Downs syndrome. Child: Care, Health and Development 9 (5): 247-55. Ball, T. M. and Wright, A. L. (1999). Health care costs of formula-feeding in the first year of life. Pediatrics 103 (4): 870-6. Ballard, J. L., Auer, C. E. and Khoury, J. C. (2002). Ankyloglossia: assessment, incidence, and effect of frenuloplasty on the breastfeeding dyad. Pediatrics 110 (5): e63. [Table 65] Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change. Psychological Review 84: 191-215. Barber-Madden, R. (1990). Design and implementation of a citywide breastfeeding promotion program: the New York City approach. Family and Community Health 12 (4): 71-8.

Bar-Oz, B., Bulkowstein, M., Benyamini, L., Greenberg, R., Soriano, I., Zimmerman, D., Bortnik, O., and Berkovitch, M. (2003). Use of antibiotic and analgesic drugs during lactation. Drug Safety 26 (13): 925-35. Barros, F. C., Victora, C. G., Morris, S. S., Halpern, R., Horta, B. L. and Tomasi, E. (1997). Breast feeding, pacifier use and infant development at 12 months of age: a birth cohort study in Brazil. Paediatric and Perinatal Epidemiology 11 (4): 441-50. Bar-Yam, N. B. (1998). Workplace lactation support, part II: working with the workplace. Journal of Human Lactation 14 (4): 321-5. Battersby, S. (2002). The Breastfeeding is Best Supporters project (BIBS): spreading the word. A community initiative to promote breastfeeding awareness (p40). Sheffield: University of Sheffield. Baumgarder, D. J., Muehl, P., Fischer, M. and Pribbenow, B. (2003). Effect of labour epidural anesthesia on breastfeeding of healthy full-term newborns delivered vaginally. Journal of the American Board of Family Practice 16 (1): 7-13. Beral, V. (2002). Breast cancer and breastfeeding: collaborative reanalysis of individual data from 47 epidemiological studies in 30 countries, including 50,302 women with breast cancer and 96,973 women without the disease. Lancet 360: 187-95. Berg, K. L. (1990). Two cases of tongue-tie and breastfeeding. Journal of Human Lactation 6 (3): 124-6. Bernard-Bonnin, A. C., Stachtchenko, S., Girard, G. and Rousseau E. (1989). Hospital practices and breastfeeding duration: a meta-analysis of controlled trials. Birth 16 (2): 64-6. Bero, L., Grilli, R., Grimshaw, J., Harvey, E., Oxman, A. D. and Thomson, M. A. (1998). Closing the gap between research and practice: an overview of systematic reviews of interventions to promote implementation of research findings by health care professionals. British Medical Journal 317: 465-8. Berry, H. (1994). Postnatal placement of breast-feeding mothers. Midwives Chronicle 107 (1278): 278-82. [Table 30] 145

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Bick, D. E., MacArthur, C. and Lancashire, R. J. (1998). What influences the uptake and early cessation of breast feeding? Midwifery 14: 242-7. Blaikeley, J., Clarke, S., Mackeith, R. and Ogden, K. M. (1953). Breastfeeding: factors affecting success. Journal of Obstetrics and Gynaecology of the British Empire 60: 657-69. Bliss, M. C., Wilkie, J., Acredolo, C., Berman, S. and Tebb K. P. (1997). The effect of discharge pack formula and breast pumps on breastfeeding duration and choice of infant feeding method. Includes commentary by Howard, C, and Howard, F. M. (pp98-101). Birth 24 (2): 90-7. [Table 37] Blyth, R., Creedy, D. K., Dennis, C. L., Moyle, W., Pratt, J. and De Vries, S. M. (2002). Effect of maternal confidence on breastfeeding duration: an application of breastfeeding self-efficacy theory. Birth 29 (4): 278-84. Bosma, J. F., Hepburn, L. G., Josell, S. D. and Baker, K. (1990). Ultrasound demonstration of tongue motions during suckle feeding. Developmental Medicine and Child Neurology 32: 223-9. Brent, N. B., Redd, B., Dworetz, A., DAmico, F and Greenberg, J. J. (1995). Breast-feeding in a low-income population. Program to increase incidence and duration. Archives of Pediatrics and Adolescent Medicine 149 (7): 798-803. [Table 21] Brent, N., Rudy, S. J., Redd, B., Rudy, T. E. and Roth, L. A. (1998). Sore nipples in breast-feeding women: a clinical trial of wound dressings vs conventional care. See comment (1999) 153 (6): 658. Archives of Pediatrics and Adolescent Medicine 152 (11): 1077-82. [Table 59] Britten, J. and Broadfoot, M. (2002). Breastfeeding support in Scotland. British Journal of Midwifery 10 (5): 292-6. [Table 38] Brown, M. S. and Hurlock, J. T. (1975). Preparation of the breast for breastfeeding. Nursing Research 24 (6): 448-51. Brown, K. H., Creed de Kanashiro, H., del Aguila, R., Lopez de Romana, G. and Black, R. (1986). Milk consumption and hydration status of exclusively breastfed infants in a warm climate. Pediatrics 108: 677-80. 146

Brown, S., Small, R., Faber, B., Krastev, A. and Davis, P. (2002). Early postnatal discharge from hospital for healthy mothers and term infants. The Cochrane Database of Systematic Reviews, Issue 3. BuLock, F., Woolridge, M. W. and Baum, J. D. (1990). Development of coordination of sucking, swallowing and breathing: ultrasound study of term and preterm infants. Developmental medicine and child neurology 32: 669-78. Buranasin, B. (1991). The effects of rooming-in on the success of breastfeeding and the decline in abandonment of children. Asia-Pacific Journal of Public Health 5 (3): 217-20. Campbell, S. (1996). Breastfeeding self-efficacy: the effects of a breastfeeding promotion nursing intervention. Thesis. University of Rhode Island. [Table 22] Campbell, H. and Gibson, A. (1997). Health targets in the NHS: lessons learned from experience with breast feeding targets in Scotland. British Medical Journal 314 (7086): 1030-3. Carter, H., Power, K., Shepherd, K. and Swanson V. (2001). Knowledge and attitudes towards breastfeeding and bottle-feeding: a theoretically based assessment among secondary school pupils in Scotland. Scottish Executive, Chief Scientist Office. Casey, C. E., Neifert, M. R., Seacat, J. M. and Neville, M. C. (1986). Nutrient intake by breastfed infants during the first five days after birth. American Journal of Diseases of Childhood 140: 933-6. Cattaneo, A. and Buzzetti, R. (2001). Effect on rates of breast feeding of training for the baby friendly hospital initiative quality improvement report. British Medical Journal 323 (7325): 1358-62. [Table 71] Cattaneo, A., Borgnolo, G. and Simon, G. (2001). Breastfeeding by objectives. European Journal of Public Health 11 (4): 397-401. [Table 39] Centuori, S., Burmaz, T., Ronfani, L., Fragiacomo, M., Quintero, S., Pavan, C., Davanzo, R. and Cattaneo, A. (1999). Nipple care, sore nipples, and breastfeeding: a randomized trial. Journal of Human Lactation 15 (2): 125-30. [Table 47]

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Cesar, A. J., Cesar, G. V., Barros, F. C., Santos, I. S. and Flores, J. A. (1999). Impact of breast feeding on admission for pneumonia during postneonatal period in Brazil: nested case-control study. British Medical Journal 318: 1316-22. Chapman, D. J. and Perez-Escamilla, R. (1999). Identification of risk factors for delayed onset of lactation. Journal of the American Dietetic Association 99 (4): 450-4. Chapman, D. J., Young, S., Ferris, A. M. and PerezEscamilla, R. (2001). Impact of breast pumping on lactogenesis stage II after Caesarean delivery. Pediatrics 107: e94. [Table 48] Clark, M. (1985). A study of four methods of nipple care offered to post partum mothers. New Zealand Nursing Journal 78 (6): 16-8. Clinton, B. (1988). Promoting maternal and child health in the context of rural poverty. Annual meeting of the American Public Health Association. Available from Educational Resources Information Center (ERIC) Document Reproduction Service (EDRS). Cohen, R. (1994). The impact of two corporate lactation programs on the incidence and duration of breastfeeding by employed mothers. American Journal of Health Promotion 8 (6): 436-41. Cohen, R., Lange, L. and Slusser W. (2002). A description of a male-focussed breastfeeding promotion corporate lactation program. Journal of Human Lactation 18 (1): 61-5. Cole, T. G., Paul, A. A. and Whitehead, R. G. (2002). Weight reference charts for British long-term breastfed infants. Acta Pdiatrica 91: 1296-1300. Coombs, D. W., Reynolds, K., Joyner, G. and Blankson, M. (1998). A self-help program to increase breastfeeding among low-income women. Journal of Nutrition Education 30 (4): 203-9. [Table 12] Corwin, E. J., Bozoky, I., Pugh, L. and Johnston, N. (2003). Interleukin-1beta elevation during the postpartum period. Annals of Behavioural Medicine 25 (1): 141-7.

Crossman, K. (1998). Breastfeeding a baby with a cleft palate: a case report. Journal of Human Lactation 14 (1): 47-50. [Table 66] Curro, V., Lanni, R., Scipione, F., Grimaldi, V. and Mastroiacovo, P. (1997). Randomised controlled trial assessing the effectiveness of a booklet on the duration of breast feeding. Archives of Disease in Childhood 76 (6): 500-4. [Table 13] Daly, S. E. J., Owens, R. A. and Hartmann, P. E. (1993). The short-term synthesis and infant-regulated removal of milk in lactating women. Experimental Physiology 78: 209-20. Darwazeh, A. M. and al-Bashir, A. (1995). Oral candidal flora in healthy infants. Journal of Oral Pathology and Medicine 24 (8): 361-4. Darzi, M. A., Chowdri, N. A. and Bhat, A. N. (1996). Breast feeding or spoon feeding after cleft lip repair: a prospective, randomised study. British Journal of Plastic Surgery 49 (1): 24-6. de Gezelle, H., Ooghe, W., Thiery, M. and Dhont, M. (1983). Metoclopramide and breast milk. European Journal of Obstetrics, Gynecology and Reproductive Biology 15 (1): 31-6. Declercq, E. R., Sakala, C., Corry, M. P., Applebaum, S. and Risher, P. (2002). Listening to mothers: report of the first national US survey of womens childbearing experiences. New York: Maternity Center Association. Dennis, C. (2002). Breastfeeding peer support: maternal and volunteer perceptions from a randomized controlled trial. Birth 29 (3): 169-76. [Table 1] Dennis, C., Hodnett, E., Gallop, R. and Chalmers, B. (2002). The effect of peer support on breast-feeding duration among primiparous women: a randomized controlled trial. Canadian Medical Association Journal 166 (1): 21-8. [Table 1] Department of Health (1989). World Health Organizations International Code of Marketing of Breastmilk Substitutes. Health Circular. London: Department of Health.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

147

Department of Health (1993). Changing Childbirth. London: HMSO. Department of Health (1999). Saving Lives: Our Healthier Nation. London: Stationery Office. Department of Health (2003). Priorities and planning framework 2003-2006. London: Department of Health. Department of Health and Social Security (1983). International code of marketing of breastmilk substitutes. Heywood, Lancashire: Department of Health and Social Security. Dewey, K. G. (1997). Energy and protein requirements during lactation. Annual Review of Nutrition 17: 19-36. Di Matteo, M. R., Morton, S. C., Lepper, H. S. and Damush, T. M. (1996). Cesarean childbirth and psychosocial outcomes: a meta-analysis. Health Psychology 15 (4): 303-14. [Table 43] Donnelly, A., Snowden, H. M., Renfrew, M. J. and Woolridge, M. W. (2000). Commercial hospital discharge packs for breastfeeding women. The Cochrane Database of Systematic Reviews, Issue 2. Duffy, E. P., Percival, P. and Kershaw, E. (1997). Positive effects of an antenatal group teaching session on postnatal nipple pain, nipple trauma and breast feeding rates. Midwifery 13 (4): 189-96. [Table 14] Dungy, C. I., Losch, M. E., Russell, D., Romitti, P. and Dusdieker, L. B. (1997). Hospital infant formula discharge packages. Do they affect the duration of breast-feeding? Archives of Pediatrics and Adolescent Medicine 151 (7): 724-9. [Table 40] Dunn, O. M., Alder, E. M. and Wallace, L. M. (2003). Psychological factors predicting sustained breastfeeding in UK first time mothers. In: 17th European Health Psychology Society Conference, Kos, Greece. For information email: l.wallace@coventry.ac.uk Durand, M., Labarere, J., Brunet, E. and Pons, J. C. (2003). Evaluation of a training program for healthcare professionals about breast-feeding. European Journal of Obstetrics, Gynecology and Reproductive Biology 106 (2): 134-8. [Table 72]

Dykes, F. (2003). Infant Feeding Initiative: a report evaluating the breastfeeding practice projects 1999-2002 executive summary. London: Department of Health. Dyson, L., McMillan, B., Woolridge, M., Green, J., Conner, M., Clarke, G., Bharj, K. and Renfrew, M. J. (2003). Looking at infant feeding today. Reducing inequalities in health among socio-economically disadvantaged and ethnic groups by increasing breastfeeding uptake: an examination of intentions and outcomes. Final report to the Department of Health. Mother and Infant Research Unit: University of York. Edwards, N., Sims-Jones, N. and Brethaupt, K. (1998). Smoking in pregnancy and postpartum: relationship to mothers choices concerning infant nutrition. Canadian Journal of Nursing Research 30 (3): 83-98. Emond, A., Pollock, J., Deave, T., Bonnell, S., Peters, T. J, and Harvey, I. (2002). An evaluation of the First Parent Health Visitor Scheme. Archives of Disease in Childhood 86 (3): 150-7. [Table 31] Enkin, M., Keirse, M. J. N. C., Renfrew, M. J., Neilson, J. (1995). A guide to effective care in pregnancy and childbirth (2nd edition). Oxford: Oxford University Press. Eppink, H. (1969). An experiment to determine a basis for nursing decisions in regard to time of initiation of breastfeeding. Nursing Research 18 (4): 292-9. EU (2003). Protection, promotion and support of breast feeding in Europe: current situation. Luxembourg: European Commission, Directorate for Public Health and Risk Assessment. Evans, K., Evans, R. and Simmer, K. (1995). Effect of the method of breast feeding on breast engorgement, mastitis and infantile colic. Acta Paediatrica Scandinavica 84 (8): 849-52. Faculty of Public Health Medicine (1995). Promoting breastfeeding (guidelines) pp1-10. London: Royal Colleges of Physicians. Fairbank, L., Lister-Sharpe, D., Renfrew, M. J., Woolridge, M. W., Sowden, A. J. S. and OMeara, S. (1999). Interventions for promoting the initiation of breastfeeding. The Cochrane Database of Systematic Reviews, Issue 3.

148

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Fairbank, L., OMeara, S., Renfrew, M. J., Woolridge, M., Sowden, A. J. and Lister-Sharp, D. (2000). A systematic review to evaluate the effectiveness of interventions to promote the initiation of breastfeeding. Health Technology Assessment 4 (25): 1-171. Fewtrell, M. S. (2004). The long-term benefits of having been breast-fed. Current Paediatrics 14: 97-103. Finch, C. and Daniel, E. L. (2002). Breastfeeding education program with incentives increases exclusive breastfeeding among urban WIC participants. Journal of the American Dietetic Association 102 (7): 981-4. [Table 23] Fishbein, M. and Ajzen, I. (1975). Belief, attitude, intention and behaviour: an introduction to theory and research. Reading: Addison-Wesley. Fisher, C. (1986). On demand breastfeeding: aspects of midwifery. Midwife Health Visitor and Community Nurse 22: 194-8. Fitz-Desorgher, R. (2003). All tied up. Tongue tie and its implications for breastfeeding. Practising Midwife 6 (1): 20-2. [Table 67] Foster, K., Lader, D. and Cheesborough, S. (1997). Infant Feeding 1995: a survey of infant feeding practices in the United Kingdom carried out by the Social Survey Division of the Office for National Statistics on behalf of the Department of Health, the Scottish Office Department of Health, the Welsh Office and the Department of Health and Social Services in Northern Ireland. London: Stationery Office. Fredrickson, D. D. (1995). Breastfeeding promotion among WIC-participating women in North Carolina. Thesis. University of North Carolina, Chapel Hill. [Table 24] Friedman, E. A. and Sachtleben, M. R. (1961). Oxytocin in lactation. American Journal of Obstetrics and Gynaecology 82: 846-55. Fry, T. (2003). The new Breast is Best growth charts. Journal of Family and Health Care 13 (5): 124-6. Gagnon, A. J., Dougherty, G., Jimenez, V. and Leduc, N. (2002). Randomized trial of postpartum care after hospital discharge. Pediatrics 109 (6): 1074-80. [Table 32]

Gainotti, V. and Pagani, G. (1980). Promotion of breastfeeding: Experience with 325 healthy infants. Minerva Pediatrica 32 (19): 1133-44. [Table 73] Gale, C. R. and Martyn, C. N. (1996). Breastfeeding, dummy use, and adult intelligence. Lancet 347: 1072-5. Gale, R., Dollberg, S., Branski, D., Stevenson, D. and Seidman, D. S. (1989). Breast-feeding of term infants. Three-hour vs four-hour non-demand. A randomized controlled reappraisal of hospital-based feeding schedules. Clinical Pediatrics 28 (10): 458-60. Garforth, S. and Garcia, J. (1989). Breast feeding policies in practice no wonder they get confused. Midwifery 5 (2): 75-83. Gilman, M. W., Rifas-Shiman, S. L., Camargo, C. A., Berkey, C. S., Frazier, A. L., Rockett, H. R. H., Field, A. E. and Colditz, G. A. (2001). Risk of overweight among adults who are breastfed as infants. Journal of the American Medical Association 285: 2461-7. Giovannini, M., Banderali, G., Radaelli, G., Carmine, V., Riva, E. and Agostoni, C. (2003). Monitoring breastfeeding rates in Italy: national surveys 1995 and 1999. Acta Paediatrica 92 (3): 357-63. [Table 41] Gokhale, M. (1965). A herbal drug that stimulates lactation. Indian Practitioner 58: 514-8. Goldberg, N. M. and Adams, E. (1983). Supplementary water for breast-fed babies in a hot and dry climate not really a necessity. Archives of Disease in Childhood 58: 514-8. Gosha, J. L. and Tichy, A. M. (1988). Effect of a breast shell on postpartum nipple pain: an exploratory study. Journal of Nurse Midwifery 33 (2): 74-7. Graffy, J., Taylor, J., Williams, A. and Eldridge, S. (2004). Randomised controlled trial of support from volunteer counsellors for mothers considering breastfeeding. British Medical Journal 328: 26-9. [Table 2] Grant, J., Fletcher, M. and Warwick, C. (2000). The South Thames Evidence Based Practice (STEP) project: supporting breastfeeding women, pp43 (plus 18 appendices). London: South Bank University, Kings Healthcare, Kingston University, St Georges Hospital Medical School. [Table 74] 149

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Green, J. M., Coupland, V. A. and Kitzinger, J. V. (1998). Great expectations: a prospective study of womens expectations and experiences of childbirth. Hale, Cheshire, England: Books for Midwives Press. Gregg, J. E. (1989). Attitudes of teenagers in Liverpool to breast feeding. British Medical Journal 299 (6692): 147-8. Grimshaw, J. M., Shirran L., Thomas, R., Mowatt, G., Fraser, C., Bero, L., Grilli, R., Harvey, E., Oxman, A. and OBrien, M. A. (2001). Changing provider behaviour. Medical Care 39 (2, supplement): 2-45. Grimshaw, J. M., Thomas, R. E., MacLennan, G., Fraser, C., Ramsay, C. R., Vale, L., Whitty, P., Eccles, M. P., Matowe, L., Shirran, L., Wensing, M., Dijkstra, R. and Donaldson, C. (2004). Effectiveness and efficiency of guideline dissemination and implementation strategies. Health Technology Assessment 8 (6): 84. Grossman, L. K., Harter, C., Sachs, L. and Kay, A. (1990). The effect of postpartum lactation counselling on the duration of breast-feeding in low-income women. American Journal of Diseases of Children 144 (4): 471-4. [Table 25] Guise, J. M., Palda, V., Westhoff, C., Chan, B. K. S., Helfand, M. and Lieu, T. A. (2003). The effectiveness of primary care based interventions to promote breastfeeding: systematic evidence review and metaanalysis for the US preventative task force. Annals of Family Medicine 1 (2): 70-8. Gunn, J., Chondros, P. and Young, D. (1998). Does an early postnatal check-up improve maternal health: results from a randomised trial in Australian general practice. British Journal of Obstetrics and Gynaecology 105 (9). [Table 33] Gunther, M. (1945). Sore nipples: causes and prevention. Lancet 2: 590-3. Gutberlet, R. and Cornblath, M. (1975). Neonatal hypoglycaemia revisited. Pediatrics 58: 10-7. Guzman, V., Toscano, G., Canales, E. S. and Zarate A. (1979). Improvement of defective lactation by using oral metoclopramide. Acta Obstetrica et Gynecologica Scandinavica 58: 53-5.

Gwinn, M. L., Lee, N. C., Rhodes, R. H., Layde, P. M. and Rubin, G. L. (1990). Pregnancy, breastfeeding and oral contraceptives and the risk of epithelial ovarian cancer. Journal of Clinical Epidemiology 43: 559-68. Haines, A. and Donald, A. (1998). Getting research findings into practice. London: BMJ Publishing Group. Hale, T. (2002). Medication and mothers milk. Pharmasoft Medical Publishing. Halken, S., Host, A., Hansen, L. G. and Osterballe O. (1992). Effect of an allergy prevention programme on incidence of atopic symptoms in infancy. A prospective study of 159 high-risk infants. Allergy 47 (5): 545-53. Halpern, S. H., Levine, T., Wilson, D. B., MacDonell, J., Katsiris, S. E. and Leighton, B. L. (1999). Effect of labour analgesia on breastfeeding success. Birth Issues in Perinatal Care 26 (2): 83-8. Hamlyn, B., Brooker, S., Oleinikova, K. and Wands S. (2002). Infant feeding 2000. A survey conducted on behalf of the Department of Health, Social Services and Public Safety in Northern Ireland. London: Stationery Office. Hamosh, M. and Goldman, A. S. (1986). Human lactation 2: maternal and environmental factors. New York: Plenum Press. Hartley, B. M. and OConnor, M. E. (1996). Evaluation of the Best Start breast-feeding education program. Archives of Pediatrics and Adolescent Medicine 150 (8): 868-71. [Table 75] Hauck, Y. l. and Dimmock, J. E. (1994). Evaluation of an information booklet on breastfeeding duration: a clinical trial. Journal of Advanced Nursing 20 (5): 836-43. [Table 15] Hawdon, J. M., Platt, M. P. and Aynsley-Green, A. (1993). Neonatal hypoglycaemia blood glucose monitoring and baby feeding. Midwifery 9 (1): 3-6. Hawe, P., Shiell, A. and Riley T. (2004). Complex interventions: how out of control can a randomised controlled trial be? British Medical Journal 328: 1561-3. Health Education Board for Scotland (HEBS) (2001). Development of mass media breastfeeding campaign. Edinburgh: HEBS.

150

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Heck, I. J. and Erenberg. (1987). Serum glucose levels in term neonates during the first 48 hours of life. Journal of Pediatrics 110: 119-22. Heinig, M. J. and Dewey, K. G. (1996). Health advantages of breast feeding for infants: a critical review. Nutrition Research Reviews 9: 89-110. Helsing, E. (1990). Supporting breastfeeding: what governments and health workers can do European experiences. International Journal of Gynecology and Obstetrics. 31 (1): 69-76. Henderson, A., Kitzinger, J. and Green, J. (2000). Representing infant feeding: content analysis of British media portrayal of bottle feeding and breast feeding. British Medical Journal 311: 1196-8. Henderson, A., Stamp, G. and Pincombe, J. (2001). Postpartum positioning and attachment education for increasing breastfeeding: a randomized trial. Birth 28 (4): 236-42. [Table 49] Henderson, J. J., Evans, S., Straton, J. A. Y., Priest, S. and Hagan, R. (2003). Impact of postnatal depression on breastfeeding duration. Birth 30 (3): 175-80. Herd, B. and Feeney, J. G. (1986). Two aerosol sprays in nipple trauma. Practitioner 230: 31-8. Hewat, R. J. and Ellis, D. J. (1987). A comparison of the effectiveness of two methods of nipple care. Birth: Issues in Perinatal Care and Education 14 (1): 41-5. Hillervik-Lindquist, C. (1991). Studies on perceived breast milk insufficiency. A prospective study in a group of Swedish women. Acta Paediatrica Scandinavica Supplement 376: 1-27. Hirose, M., Hara, Y., Hosokawa, T. and Tanaka, Y. (1996). The effect of postoperative analgesia with continuous epidural bupivacaine after cesarean section on the amount of breast feeding and infant weight gain. Anesthesia and Analgesia 82 (6): 1166-9. [Table 50] Hoddinott, P. and Pill, R. (1999). Qualitative study of decisions about infant feeding among women in east end of London. British Medical Journal 318: 30-4.

Hodnett, E., Gates, S., Hofmeyr, G. J. and Sakala C. (2003). Continuous support for women during childbirth. Cochrane Database of Systematic Reviews, Issue 1. [Table 44] Hodnett, E. D., Lowe, N. K., Hannah, M. E., Willan, A. R., Stevens, B., Weston, A., Ohlssen, A., Gaffni, A., Muir, H. A., Myhr T. L. and Stremler, R. for the Nursing Supportive Care in Labour Trial Group (2002). Effectiveness of nurses as providers of birth labour support in North American hospitals: a randomized controlled trial. Journal of the American Medical Association 288 (11): 1373-81. Hoey, C. and Ware, J. L. (1997). Economic advantages of breast-feeding in an HMO setting: a pilot study. American Journal of Managed Healthcare 3 (6): 861-5. Hofmeyr, G. J., van Iddekinge, B. and Blott, J. A. (1985). Domperidone: secretion in breastmilk and effect on puerperal prolactin levels. British Journal of Obstetrics and Gynaecology 92: 141-4. Hofmeyr, G. J., Nikodem, V. C., Wolman, W. L., Chalmers, B. E. and Kramer, T. (1991). Companionship to modify the clinical birth environment: effects on progress and perceptions of labour, and breastfeeding. British Journal of Obstetrics and Gynaecology 98 (8): 756-64. Hogan, M., Westcott, C. and Griffiths, D. M. (2004). A randomised controlled trial of division of tongue-tie in infants with feeding problems. Archives of Disease in Childhood 89 (Suppl 1) (A5). Horta, B. L., Victora, C. G., Menezes, A. M. and Barros, F. C. (1997). Environmental tobacco smoke and breastfeeding duration. American Journal of Epidemiology 146 (2): 128-33. Houston, M. J. (1984). Home support for the breast feeding mother. In: Houston, M. J. (ed) Maternal and infant health care. 1st edition. Edinburgh: Churchill Livingstone. Houston, M. J., Howie, P. W. and McNeilly, A. S. (1983). Factors affecting the duration of breast feeding: 1. Measurement of breast milk intake in the first week of life. Early Human Development 8 (1): 49-54.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

151

Howard, C., Howard, F., Lawrence, R., Andresen, E., DeBlieck, E. and Weitzman, M. (2000). Office prenatal formula advertising and its effect on breast-feeding patterns. Obstetrics and Gynecology 95 (2): 296-303. [Table 42] Howard, C. R., Howard, F. M., Lanphear, B., Eberly, S., DeBlieck, E. A., Oakes, D. and Lawrence, R. A. (2003). Randomized clinical trial of pacifier use and bottlefeeding or cup feeding and their effect on breastfeeding. Pediatrics 111 (3): 511-8. [Table 51] Howie, P. W., Forsyth, J. S., Ogston, S. A., Clark, A. and Florey, C. du V. (1990). Protective effect of breast feeding against infection. British Medical Journal 300: 11-6. Huggins, K. (1990). Ankyloglossia one lactation consultants personal experience. Journal of Human Lactation 6 (3): 123-4. Huggins, K. E. and Billon, S. F. (1993). Twenty cases of persistent sore nipples: collaboration between lactation consultant and dermatologist. Journal of Human Lactation 9 (3): 155-60. Huntingford, P. J. (1961). Intranasal use of synthetic oxytocin in management of breastfeeding. British Medical Journal 2: 709-11. Hytten, F. E. (1954). Clinical and chemical studies in human lactation. British Medical Journal 1: 175-82. IBFAN (2004), Breaking the rules, stretching the rules. Evidence of violations of the International Code of Marketing of Breastmilk Substitutes and subsequent resolutions: pp4-8. Malaysia: International Baby Food Action Network. Illingworth, R. S., Stone, D. G. H., Jowett, G. H. and Scott, J. F. (1952). Self-demand feeding in a maternity unit. Lancet 1: 683-7. Inch, S. and Fisher, C. (1995). Mastitis: infection or inflammation. Practitioner 239 (1553): 472-6. Ingelman-Sundberg, A. (1958). The value of antenatal massage of nipples and expression of colostrum. Journal of Obstetrics and Gynaecology of the British Empire 65: 448-9.

Ingram, J. and Johnson, D. (2003). South Asian grandmothers influence on breast feeding in Bristol. Midwifery 19: 318-27. Ingram, J., Johnson, D. and Greenwood, R. (2002). Breastfeeding in Bristol: teaching good positioning, and support from fathers and families. Midwifery 18 (2): 87-101. [Table 76] Ingram, J., Hunt, L., Woolridge, W. and Greenwood R. (2004). The association of progesterone, infant formula use and pacifier use with the return of menstruation in breastfeeding women: a prospective cohort study. European Journal of Obstetric Gynecology and Reproductive Biology 114 (2): 197-202. International Labour Organization (2000). C183 Maternity Protection Convention: concerning the revision of the Maternity Protection Convention (Revised), 1952. www.ilo.org/ilolex/cgi-lex/convde.pl?C183 Jack, C. M., Elison, J. E., Winder, V., Day, J. and Blenkinsip, L. (2001). Education issues: breast-feeding: continuing professional development. British Journal of Midwifery 9: 162-6. Janovsky, K. and Cassels, A. (1996). Health policy and systems research: issues, methods, and priorities. In: Janovsky, K. (ed) Health policy and systems development: an agenda for research. Geneva: World Health Organization. Jensen, D., Wallace, S. and Kelsay, P. (1994). LATCH: a breastfeeding charting system and documentation tool. Journal of Obstetric, Gynecologic and Neonatal Nursing 23: 27-32. Jordan, P. and Wall, V. (1990). Breastfeeding and fathers: illuminating the darker side. Birth 17: 210-3. Jordan, P. and Wall, V. (1993). Supporting the father when an infant is breastfed. Journal of Human Lactation 9: 31-4. Kauppila, A., Kivinen, S. and Ylikorkala, O. (1981a). A dose response relation between improved lactation and metaclopramide. Lancet 1: 1175-7.

152

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Kauppila, A., Kivinen, S. and Ylikorkala, O. (1981b). Metoclopramide increases prolactin release and milk secretion in puerperium without stimulating the secretion of thyrotropin and thyroid hormones. Journal of Clinical Endocrinology and Metabolism 52: 436-9. Kauppila, A., Anunti, P., Kivinen, S., Koivisto, M. and Ruokonen A. (1988). Metoclopramide and breast feeding: efficacy and anterior pituitary responses of the mother and the child. European Journal of Obstetrics, Gynecology, and Reproductive Biology 19 (1): 19-22. Kee, W. H., Tan, S. L., Lee, V. and Salmon, Y. M. (1989). The treatment of breast engorgement with serrapeptase (Danzen): a randomised double-blind controlled trial. Singapore Medical Journal 30 (1): 48-54. Kennedy, K., Rivera, R. and McNeilly, A. (1989). Consensus statement on the use of breast-feeding as a family planning method. Contraception 39 (5): 477-96. Kersting, M. and Dulon, M. (2002). Assessment of breastfeeding promotion in hospitals and follow-up survey of mother-infant pairs in Germany: the SuSe Study. Public Health Nutrition 5 (4): 547-52. Khan, K., ter Riet, G., Glanville, J., Sowden, A. and Kleijnen, J. (2001). Undertaking systematic reviews of research on effectiveness. CRD Report Number 4 (2nd Edition). University of York: NHS Centre for Reviews and Dissemination. Kind, C., Schubiger, G., Schwarz, U. and Tonz, O. (2000). Provision of supplementary fluids to breast fed infants and later breast feeding success. Advances in Experimental Medicine and Biology 478: 347-54. [Table 52] Kirk, T. R. (1980). Appraisal of the effectiveness of nutrition education in the context of infant feeding. Journal of Human Nutrition 34 (6): 429-38. Kistin, N., Benton, D., Rao, S. and Sullivan, M. (1990). Breast-feeding rates among black urban low-income women: effect of prenatal education. Pediatrics 86 (5): 741-6. [Table 16] Kogo, M., Okada, G., Ishii, S., Shikata, M., Iida. S. and Matsuya. T. (1997). Breast feeding for cleft lip and palate patients, using the Hotz-type plate. Cleft Palate Craniofacial Journal 34 (4): 351-3. [Table 68]

Kotagal, U. R., Atherton, H. D., Eshett, R., Schoettker, P. J. and Perlstein, P. H. (1999). Safety of early discharge for Medicaid newborns. Journal of the American Medical Association 282: 1150-6. Kramer, M. S., Barr, R. G., Dagenais, S., Yang, H., Jones, P., Ciofani, L. and Jane, F. (2001a). Pacifier use, early weaning, and cry/fuss behavior: a randomized controlled trial. Journal of the American Medical Association 286 (3): 322-6. [Table 53] Kramer, M. S., Chalmers, B., Hodnett, E. D., Sevkovskaya, Z., Dzikovich, I., Shapiro, S., Collet, J. P., Vanilovich, I., Mezen, I., Ducruet, T., Shishko, G., Zubovich, V., Mknuik, D., Gluchanina, E., Dombrovskiy, V., Ustinovitch, A., Kot, T., Bogdanovich, N., Ovchinikova, L. and Helsing, E. for the PROBIT Study Group. (2001b). Promotion of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. See comments. Journal of the American Medical Association 285 (4): 413-20. Kramer, M. S., Guo, T., Platt, R. W., Shapiro, S., Collet, J. P., Chalmers, B., Hodnett, E., Sevkovskaya, Z., Dzikovich, I. and Vanilovich, I. for the PROBIT Study Group (2002). Breastfeeding and infant growth: biology or bias? Pediatrics 110 (2): 343-7. Kuscu, N. K., Koyuncu, F. and Lacin, S. (2002). Collagenase treatment of sore nipples. International Journal of Gynecology and Obstetrics 76 (1): 81-2. [Table 60] Labbok, M. H. (2001). Effects of breastfeeding on the mother. Pediatric Clinics of North America 48 (1): 143-58. Labbok, M. H. and Krasovec, K. (1990). Towards consistency in breastfeeding definitions. Studies in Family Planning 21: 226-30. Langer A., Campero, L., Garcia, C. and Reynoso, S. (1998). Effects of psychosocial support during labour and childbirth on breastfeeding, medical interventions, and mothers wellbeing in a Mexican public hospital. British Journal of Obstetrics and Gynaecology 105: 1056-63. Lavergne, N. A. (1997). Does application of tea bags to sore nipples while breastfeeding provide effective relief? [comment]. Journal of Obstetric, Gynecologic and Neonatal Nursing 26 (1): 53-8. [Table 61]

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

153

Lawrence, R. A. (1994). Breastfeeding: a guide for the medical profession. 4th ed. St. Louis: Mosby. Lawton, R. and Parker D. (1999). Procedures and the professional: the case of the British NHS. Social Science and Medicine 48: 351-61. Leboyer, F. (2002). Birth without violence. Inner Traditional International. LEsperance, C. M. (1980). Pain of pleasure: the dilemma of early breastfeeding. Birth Family Journal 7: 21-6. Letson, G. W., Rosenbergm K. D. and Wu, L. (2002). Association between smoking during pregnancy and breastfeeding at about 2 weeks of age. Journal of Human Lactation 18 (4): 368-72. Lewis, P. J., Devenish, C. and Kahn, C. (1980). Controlled trial of metoclopramide in the initiation of breast-feeding. British Journal of Clinical Pharmacology 9: 217-9. Lieberman, E. and ODonoghue, C. (2002). Unintended effects of epidural analgesia in labour: a systematic review. American Journal of Obstetric Gynecology 186 (5): S31-S68. Lindenberger, J. H. and Bryant, C. A. (2000). Promoting breastfeeding in the WIC program: a social marketing case study. American Journal of Health Behavior 24 (1): 53-60. Livingstone, V. and Stringer, L. J. (1999). The treatment of staphyloccocus aureus infected sore nipples: a randomized comparative study. Journal of Human Lactation 15 (3): 241-6 (erratum:16 (2):179). [Table 62] Loh, N. R., Kelleher, C. C., Long, S. and Loftus, B. G. (1997). Can we increase breast feeding rates? Irish Medical Journal 90 (3): 100-1. [Table 17] Long, L. (1995). Breastfeeding workshops: a focus on knowledge, skills and attitudes. British Journal of Midwifery 3 (10): 540-4. Lucas, A., Brooke, O. G., Morley, R., Cole, T. J. and Bamford, M. F. (1990). Early diet of preterm infants and the development of allergic or atopic disease: randomised prospective study. British Medical Journal 300: 837-40.

Luhman, L. A. (1963). The effect of intranasal oxytocin on lactation. Obstetrics and Gynecology 21: 713-7. Lumley, J., Oliver, S., Chamberlain, C. and Oakley, L. (2004). Interventions for preventing smoking cessation during pregnancy. The Cochrane Database of Systematic Reviews, Issue 3. MacArthur, C., Winter, H. R., Bick, D. E., Lilford, R. J., Lancashire, R. J., Knowles, H., Braunholtz, D. A., Henderson, C., Belfield, C. and Gee, H. (2003). Redesigning postnatal care: a randomised controlled trial of protocol-based midwifery-led care focused on individual womens physical and psychological needs. Health Technology Assessment 7 (37): 97. MAIN Trial Collaborative Group (1994). Preparing for breast feeding: treatment of inverted and non-protractile nipples in pregnancy. Midwifery 10 (4): 200-14. Manitoba Pediatric Society and Committee on Breastfeeding (1982). Breast-feeding promotion in Manitoba. Canadian Medical Association Journal 126 (6): 639-42. [Table 77] Manoff, R. (1985). Social marketing. New York: Praeger. Margolis, L. H. and Schwartz, J. B. (2000). The relationship between the timing of maternal postpartum hospital discharge and breastfeeding. Journal of Human Lactation 16 (2): 121-8. Marild, S., Jodal, U. and Hanson, L. A. (1990). Breastfeeding and urinary tract infection. Lancet 336: 942. Marild, S., Hansson, S., Jodal, U., Oden, A. and Svedberg, K. (2004). Protective effect of breastfeeding against urinary tract infection. Acta Paediatrica Scandinavica 93 (2): 164-8. Marmet, C., Shell, E. and Marmet, R. (1990). Neonatal frenotomy may be necessary to correct breastfeeding problems. Journal of Human Lactation 6 (3): 117-21. Marshall, D. R., Callan, P. P. and Nicholson, W. (1994). Breastfeeding after reduction mammoplasty. British Journal of Plastic Surgery 47 (3): 167-9.

154

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Marshall, J. L., Renfrew, M. J. and Godfrey, M. (2003). Understanding the use of research evidence in practice: the support given to breastfeeding women by community midwives and health visitors. Poster presentation at Research in Midwifery and Perinatal Health Conference, Birmingham MIRU. Marteau, T. M., Sowden, A. J. and Armstrong, D. (2002). Implementing research findings into practice: beyond the information deficit model. In: Haines, A. and Donald, A. (eds) Getting research findings into practice (second edition) pp68-76. London: BMJ Books. Martin, I. and OHare, I. P. (1999). Breastfeeding mother in bus ban. Scotland on Sunday 8 August. Martin, J. and White, A. (1988). Infant feeding survey, 1985. Office of Population Census and Surveys, Social Survey Division. London: HMSO. Martin, R. M., Ness, A. R., Gunnell, D. and Davey Smith, G. for the ALSPAC study team (2004). Does breastfeeding in infancy lower blood pressure in childhood? Circulation 109: 1259-66. Martin-Calama, J., Bunuel, J., Valero, M.T., Labay, M., Lasarte, J. J., Valle, F. and Miguel, C. (1997). The effect of feeding glucose water to breastfeeding newborns on weight, body temperature, blood glucose, and breastfeeding duration. Journal of Human Lactation 13 (3): 209-13. [Table 54] Masaitis, N. S. and Kaempf, J. W. (1996). Developing a frenotomy policy at one medical center: a case study approach. Journal of Human Lactation 12 (3): 229-32. [Table 69] Mathur, G. P., Chitranshi, S., Mathur, S., Singh, S. B. and Bhalla, M. (1992). Lactation failure. Indian Paediatrics 29 (12): 1541-4. Matilla-Mont, M. and Rios-Jimenez, A. (1999). Nursing and maternal breast feeding [Spanish]. Enfermeria Clinica 9 (3): 93-7. [Table 78] Matthew, W. H., Taylor, B., Norman, A. P., Turner, M. W. and Soothill, J. F. (1977). Prevention of eczema. Lancet 1: 321-4.

Mayer, E. J., Hamman, R. F. and Gay, E. C. (1988). Reduced risk of IDDM among breastfed children: the Colorado IDDM registry. Diabetes 37: 1625-32. McGeorge, D. D. (1994). Ideas and innovations the Niplette: an instrument for the non-surgical correction of inverted nipples. British Journal of Plastic Surgery 47 (1): 46-9. McLachlan, Z. (1991). Ultrasound treatment for breast engorgement: a randomised double blind trial. Australian Journal of Physiotherapy 37: 23-8. McNeilly, A. (1977). The physiology of human lactation. Journal of Biosocial Sciences Supplement 4: 5-21. Medical Research Council (2000). A framework for the development and evaluation of randomised controlled trials for complex interventions to improve health. London: MRC. Mennella, J. (2001). Alcohols effect on lactation. Alcohol Research and Health 25 (3): 230-4. Mercer, J. S. (2001). Current best evidence: a review of the literature on umbilical cord clamping. Journal of Midwifery and Women's Health 46 (6): 402-14. [Table 45] Messner, A. H., Lalakea, M. L., Aby, J., Macmahon, J. and Bair, E. (2000). Ankyloglossia incidence and associated feeding difficulties. Archives of Otolaryngology Head Neck Surgery 126 (1): 36-9. Michie, S. and Johnston, M. (2004). Changing clinical behaviour by making guidelines specific. British Medical Journal 328: 343-5. Miller, S. A. and Chopra, J. G. (1984). Problems with human milk and infant formulas. Pediatrics Supplement 74 (2): 639-47. Milsom, S. R., Breier, B. H., Gallaher, B. W., Cox, V. A., Gunn, A. J. and Gluckman, D. (1992). Growth hormone stimulates galactopoiesis in healthy lactating women. Acta Endocrinologica 127: 337-43. Minchin, M. (1985). Breastfeeding matters. Sydney: George Allen and Unwin.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

155

Ministry of Agriculture, Fisheries and Food (1994). The British diet: finding the facts. Food Surveillance Paper. HMSO. Mitchell, E. A., Taylor, B. J., Ford, R. P., Stewart, A. W., Becroft, D. M., Thompson, J. M., Scragg, R., Hassal, I. B., Barry, D. M., Allen, E. M. et al. (1993). Dummies and the sudden infant death syndrome. Archives of Disease in Childhood 68 (4): 501-4. Mongeon, M. and Allard, R. (1995). A controlled trial of regular telephonic support given by volunteers on the progress and outcome of breast-feeding. Original: Essai controle dun soutien telephonique regulier donne par une benevole sur le deroulement et lissue de lallaitment. Canadian Journal of Public Health 86 (2): 124-7. [Table 3] Morrell, C. J., Spiby, H., Stewart, P., Walters, S. and Morgan, A. (2000a). Costs and effectiveness of community postnatal support workers: randomised controlled trial. British Medical Journal 321 (7261): 593-8. [Table 4] Morrell, C. J., Spiby, H., Stewart, P., Walters, S. and Morgan, A. (2000b). Costs and benefits of community postnatal support workers: a randomised controlled trial. Health Technology Assessment 4: (6). [Table 4] Morse, J. M., Jehle, C. and Gamble D. (1990). Initiating breastfeeding: a world survey of the timing of postpartum breastfeeding. International Journal of Nursing Studies 27 (3): 303-13. Murata, T., Hunzow, M. and Nomura, Y. (1965). The clinical effects of protease complex on postpartum breast engorgement. Journal of Japanese Obstetric Gynaecological Society 12 (3): 139-47. Myles, M. F. (1975). Textbook for midwives. Eighth edition. Edinburgh: Churchill Livingstone. National Institute for Clinical Excellence (2004). Caesarean section Quick reference guide. Clinical Guidelines 13. London: NICE. Neifert, M., Lawrence, R. and Seacat, J. (1995). Nipple confusion: toward a formal definition. Journal of Pediatrics 126 (6): S125-9.

Neifert, M. R. and Seacat, J. M. (1987). Lactation insufficiency: a rational approach. Birth 14: 182-8. Newcombe, P. A., Storer, B. E. and Longneckr, M. P. (1994). Lactation and a reduced risk of premenopausal breast cancer. New England Journal of Medicine 330: 81-7. NHS Centre for Reviews and Dissemination (1994). Implementing clinical practice guidelines: can guidelines be used to improve clinical practice? Effective Health Care 8: 1-12. York: University of York. Nicholson, W. L. (1985). Cracked nipples in breast feeding mothers: a randomised trial of three methods of management. Nursing Mothers Association of Australia 21 (4): 7-10. Nicholson, W. L. (1991). Tongue-tie (ankyloglossia) associated with breastfeeding problems. Journal of Human Lactation 7 (2): 82-4. Nicoll, A., Thayaparan, B., Newell, M. and Rundall, P. (2002). Breast feeding policy, promotion and practice in Europe. Results of a survey with non-governmental organizations. Journal of Nutritional and Environmental Medicine 12 (3): 255-64. Nikodem, V. C., Danziger, D., Gebka, N., Gulmezoglu, A. M. and Hofmeyr, G. J. (1993). Do cabbage leaves prevent breast engorgement? A randomized, controlled study. Birth 20 (2): 61-4. North, K., Fleming, P., Golding, J. and the ALSPAC study team (1999). Pacifier use and morbidity in the first six months of life. Pediatrics 103 (3): e34. Northern and Yorkshire Public Health Observatory (2002). Drug misuse in pregnancy in the Northern and Yorkshire Region. Occasional paper no. 6. Available at: www.nepho.org.uk Notestine, G. E. (1990). The importance of the identification of ankyloglossia (short lingual frenulum) as a cause of breastfeeding problems. Journal of Human Lactation 6 (3): 113-5. Nutbeam, D. (1998). Evaluating health promotion: progress, problems and solutions. Health Promotion International 13: 27-44.

156

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Ogaard, B., Larsson, E. and Lindsten, R. (1994). The effect of sucking habits, cohort, sex, intercanine arch widths, and breast or bottle feeding on posterior crossbite in Norwegian and Swedish 3-year-old children. American Journal of Orthodontics and Dentofacial Orthopedics 106 (2): 161-6. Olmsted, R. W. and Jackson, E. B. (1950). Self-demand feeding in the first week of life. Pediatrics 6: 396-401. Oxford Midwives Research Group (1991). A study of the relationship between the delivery to cord clamping interval and the time of cord separation. Midwifery 7 (4): 167-76. Peaker, M. and Wilde, C. J. (1995). Autocrine regulation of milk secretion by a protein in milk. Biochemical Journal 305: 51-8. Perez-Escamilla, R., Pollitt, E., Lonnerdal, B. and Dewey, K. G. (1994). Infant feeding policies in maternity wards and their effect on breast-feeding success: an analytical overview. American Journal of Public Health 84 (1): 8997. Peters, F., Schulze-Tollert, J. and Schuth, W. (1991). Thyrotrophin-releasing hormone a lactation-promoting agent? British Journal of Obstetrics and Gynaecology 98 (9): 880-5. Petraglia, F., de Leo, V., Sardelli, S., Pieroni, M. L., DAntona, N. and Genazzani, A. R. (1985). Domperidone in defective and insufficient lactation. European Journal of Obstetrics, Gynecology, and Reproductive Biology 19 (5): 281-7. Petticrew, M. and Roberts, H. (2003). Evidence, hierarchies, and typologies: horses for courses. Journal of Epidemiology and Community Health 57 (7): 527-9. Pisacane, A., Graziano, L., Mazzarella, G., Scarpellino, B. and Zona, G. (1992). Breastfeeding and urinary tract infection. Journal of Pediatrics 120: 87-9. Pollard, D. L. (1998). Effect of self-regulation on breastfeeding in primiparous mothers. Thesis. University of Pittsburgh. [Table 18]

Porteous, R., Kaufman, K. and Rush, J. (2000). The effect of individualized professional support on duration of breastfeeding: a randomized controlled trial. Journal of Human Lactation 16 (4): 303-8. [Table 5] Protheroe, L., Dyson, L., Renfrew, M. J., Bull, J. and Mulvihill, C. (2003). The effectiveness of public health interventions to promote the initiation of breastfeeding. Evidence briefing. 1st edition. London: Health Development Agency. Pugh, L. C. and Milligan, R. A. (1998). Nursing intervention to increase the duration of breastfeeding. Applied Nursing Research 11 (4): 190-4. [Table 7] Pugh, L. C., Buchko, B. L., Bishop, B. A. and Cochran, J. F. (1996). A comparison of topical agents to relieve nipple pain and enhance breastfeeding. Birth: Issues in Perinatal Care 23 (2): 88-93. [Table 55] Pugh, L. C., Milligan, R. A., Frick, K. D., Spatz, D. and Bronner, Y. (2002). Breastfeeding duration, costs, and benefits of a support program for low-income breastfeeding women. Birth 29 (2): 95-100. [Table 6] Punthmatharith, B. (2001). Randomized controlled trial of early kangaroo (skin-to-skin) care: effects on maternal feelings, maternal-infant interaction and breastfeeding success in Thailand. Thesis. Case Western Reserve University. [Table 56] Quinlivan, J. A., Box, H. and Evans, S. F. (2003). Postnatal home visits in teenage mothers: a randomised controlled trial. Lancet 361 (9361): 893-900. [Table 8] Raisler, J., Alexander, C. and OCampo, P. (1999). Breast feeding and infant illness: a dose-response relationship? American Journal of Public Health 89 (1): 25-30. Randerson, J. (2004). Baby food could trigger meningitis. New Scientist 182 (16). Ransjo-Arvidson, A. B., Matthiesen, A. S., Lilja, G., Nissen, E., Widstrom, A. M. and Uvnas-Moberg, K. (2001). Maternal analgesia during labour disturbs newborn behavior: effects on breastfeeding, temperature, and crying. Birth 28 (1): 5-12. Comment: 28 (1): 20-1.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

157

RCOG (2001). The National Sentinel Caesarean Section Audit Report. London: RCOG. Redman, S., Watkins, J., Evans, L. and Lloyd, D. (1995). Evaluation of an Australian intervention to encourage breast feeding in primiparous women. Health Promotion International 10 (2): 101-3. [Table 26] Renfrew, M. J., Woolridge, M. W., Ross McGill, H. (2000). Enabling women to breastfeed. A review of practices which promote or inhibit breastfeeding with evidencebased guidance for practice. Norwich: Stationery Office. Renfrew, M. J., Ansell, P. and Macleod, K. L. (2003). Formula feed preparation helping reduce the risks; a systematic review. Archives of Disease in Childhood 88: 855-8. Righard, L. and Alade, M. O. (1990). Effect of delivery room routines on success of first breast-feed. Lancet 336 (8723): 1105-7. Righard, L. and Alade, M. O. (1992). Sucking technique and its effect on success of breastfeeding. Birth 19 (4): 185-9. Righard, L., Flodmark, C-E., Lothe, L. and Jakobsson, I. (1993). Breastfeeding patterns: comparing the effects on infant behavior and maternal satisfaction of using one or two breasts. Birth 20 (4): 182-5. Riordan, J. M. (1997). The cost of not breastfeeding: a commentary. Journal of Human Lactation 13 (2): 93-7. Riordan, J., Gross, A., Angeron, J., Krumwiede, B., Melin, J. (2000). The effect of labour pain relief medication on neonatal suckling and breastfeeding duration. Journal of Human Lactation 16 (1): 7-12. Roberts, K. L. (1995). A comparison of chilled cabbage leaves and chilled gel-packs in reducing breast engorgement. Journal of Human Lactation 11: 117-20. Roberts, K. L., Reiter, M. and Schuster, D. (1995). A comparison of chilled and room temperature cabbage leaves in treating breast engorgement. Journal of Human Lactation 11 (3): 191-4. [Table 63]

Roberts, K. L., Reiter, M. and Schuster, D. (1998). Effects of cabbage leaf extract on breast engorgement. Journal of Human Lactation 14 (3): 231-6. [Table 64] Robertson, A. and Minkler, M. (1994). New health promotion movement: a critical examination. Health Education Quarterly 21 (3): 295-312. Rodriguez-Garcia, R. and Frazier, L. (1995). Cultural paradoxes relating to sexuality and breastfeeding. Journal of Human Lactation 11: 111-5. Rojjanasrirat, W. (2000). The effects of a nursing intervention on breastfeeding duration among primiparous mothers planning to return to work. Thesis. University of Kansas. [Table 27] Rosenblatt, K. A. and Thomas, D. B. (1993). WHO collaborative study of neoplasia and steroid contraceptives. International Journal of Epidemiology 22: 192-7. Ross Laboratories (2002). Breastfeeding trends through 2000. US: Ross Laboratories. Rossiter, J. C. (1994). The effect of a culture-specific education program to promote breastfeeding among Vietnamese women in Sydney. International Journal of Nursing Studies 31 (4): 369-79. [Table 19] Rowland, T. W. and Zori, R. T. (1982). Malnutrition and hypernatremic dehydration in breastfed infants. Journal of the American Medical Association 247: 1016-7. Saadeh, R. and Akre, J. (1996). Ten steps to successful breastfeeding: a summary of the rationale and scientific evidence. Birth 23 (3): 154-60. Saarinen, U. M. and Kajosaari, M. (1995). Breastfeeding as prophylaxis against atopic disease: prospective followup study until 17 years old. Lancet 346: 1065-9. Sachdev, H. P. S., Krishna, J., Puri, R. K., Satyanarayana, L. and Kumar, S. (1991). Water supplementation in exclusively breastfed infants during summer in the tropics. Lancet 337 (8747): 929-33.

158

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Sadauskaite-Kuehne, V., Ludvigsson, J., Pagaiga, Z., Jasinskiene, E. and Samuelsson, U. (2004). Longer breastfeeding is an independent protective factor against development of type 1 diabetes mellitus in childhood. Diabetes Metabolic Research Review 20 (2): 150-7. Saint, L., Smith, M. and Hartman, P. E. (1984). The yield and nutrient content of colostrum and milk of women giving birth to one month postpartum. British Journal of Nutrition 52: 87-95. Sakashita, R., Kamegai, T. and Inoue, N. (1996). Masseter muscle activity in bottle feeding with the chewing type bottle teat: evidence from electromyographs. Early Human Development 45: 83-92. Salariya, E. M., Easton, P. M. and Cater, J. I. (1978). Duration of breast-feeding after early initiation and frequent feeding. Lancet 2 (8100): 1141-3. Schubiger, G., Schwarz, U. and Tonz, O. (1997). UNICEF/WHO baby-friendly hospital initiative: does the use of bottles and pacifiers in the neonatal nursery prevent successful breastfeeding? European Journal of Pediatrics 156 (11): 874-7. [Table 57] Schy, D. S., Maglaya, C. F., Mendelson, S. G., Race, K. E. and Ludwig-Beymer, P. (1996). The effects of in-hospital lactation education on breastfeeding practice. Journal of Human Lactation 12 (2): 117-22. [Table 26] Sciacca, J. P., Phipps, B. L., Dube, D. A. and Ratliff, M. I. (1995a). Influences on breast-feeding by lower-income women: an incentive-based, partner-supported educational program. Journal of the American Dietetic Association 95 (3): 323-8. [Table 29] Sciacca, J. P., Phipps, B. L., Dube, D. A. and Ratliff, M. I. (1995b). A breast feeding education and promotion program: effects on knowledge, attitudes, and support for breast feeding. Journal of Community Health 20 (6): 473-90. [Table 29] Serafino-Cross, P. and Donovan, P. (1992). Effectiveness of professional breastfeeding home support. Journal of Nutrition Education 24 (3): 117-122. [Table 9]

Serwint, J. R., Wilson, M. E., Vogelhut, J. W., Repke, J. T. and Seidel, H. M. (1996). A randomized controlled trial of prenatal pediatric visits for urban, low-income families. Pediatrics 98 (6 pt 1): 1069-75. Comment: 1195-6. [Table 20] Sharma, M. and Petosa, R. (1997). Impact of expectant fathers in breast-feeding decisions. Journal of the American Dietetic Association 97: 1311-3. Sharma, S., Ramji, S., Kumari, S. and Bapna, J. S. (1996). Randomized controlled trial of asparagus racemosus (shatavari) as a lactogogue in lactational inadequacy. Indian Pediatrics 33: 675-7. Shaw, R. L., Wallace, L. M. and Bansal, M. (2003). Health professionals perceptions of young mothers and young womens experiences of feeding their babies: the Coventry young mothers infant feeding study. Community Practitioner 76 (8): 299-303. Shiau, S. H. H. (1997). Randomised controlled trial of kangaroo care with full term infants: effects on maternal anxiety, breastmilk maturation, breast engorgement and breastfeeding status. Thesis. Case Western Reserve University. [Table 58] Sikorski, J., Renfrew, M. J., Pindoria, S. and Wade, A. (2002). Support for breastfeeding mothers. The Cochrane Database of Systematic Reviews, Issue 1. Slaven, S., Craft, I, and Harvey, D. R. (1988). A doubleblind controlled trial of chlorhexidine in aerosol spray for the prevention of sore nipples. Unpublished. Smith, W. L., Erenburg, A., Nowak, A. and Franken, E. A. (1985). Physiology of sucking in the normal term infant using real-time ultrasound. Radiology 156: 379-81. Snowden, H. M., Renfrew, M. J. and Woolridge, M. W. (2001). Treatments for breast engorgement during lactation. The Cochrane Database of Systematic Reviews, Issue 2. Souto, G. C., Guigliani, E. R., Giugliani, C. and Schneider, M. A. (2003). The impact of breast reduction surgery on breastfeeding performance. Journal of Human Lactation 19 (1): 43-9. Spiby, H. and Munro, J. (2001). Evidence into practice for midwifery-led care. British Journal of Midwifery 9: 550-2.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

159

Standing Committee on Nutrition of the British Paediatric Association (1994). Is breastfeeding beneficial in the UK? Archives of Disease in Childhood 71: 376-80. Steel OConnor, K. O., Mowat, D. L., Scott, H. M., Carr, P. A., Dorland, J. L. and Young Tai, K. F. W. (2003). A randomized trial of two public health nurse follow-up programs after early obstetrical discharge: an examination of breastfeeding rates, maternal confidence and utilization and costs of health services. Canadian Journal of Public Health 94 (2): 98-103. [Table 34] Stokoe, B., McClarey, M. and Dakin, S. (1994). Failure breeds success. Health Visitor 67 (5): 170. [Table 79] Storr, G. B. (1988). Prevention of nipple tenderness and breast engorgement in the postpartal period. Journal of Obstetric, Gynecologic and Neonatal Nursing 17 (3): 203-9. Swanson, V. and Power, K. G. (2000). A theoretically based, cross-cultural study of infant feeding in new mothers and their partners. In: Watson, J. and Platt, S. (eds) Researching health promotion. London: Routledge. Syfrett, E. B. (1993). Very early and virtually continuous kangaroo care for 34-36 week gestation preterm infants: effects on temperature, breastfeeding, supplementation and weight. Thesis. University of Florida. Tammentie, T., Tarkka, M. T., stedt-Kurki, P., Paavilainen, E. and Laippala, P. (2004). Family dynamics and postnatal depression. Canadian Journal of Nursing Research 11 (2): 141-52. Tanguay, K. E., McBean, M. R. and Janin, E. (1994). Nipple candidiasis among breastfeeding mothers. Casecontrol study of predisposing factors. Canadian Family Physician 40: 1407-13. Thomas, M. and Avery, V. (1997). Infant feeding in Asian families: early feeding practices and growth. London: Stationery Office. Thompson, B. (1998). Breakthrough in breastfeeding. Practising Midwife 1 (5): 35-7.

Thomsen, A. C., Espersen, T. and Maigaard, S. (1984). Course and treatment of milk stasis, noninfectious inflammation of the breast, and infectious mastitis in nursing women. American Journal of Obstetrics and Gynecology 149 (5): 492-5. Townsend, J., Wolke, D., Hayes, J., Dave, S., Rogers, C., Bloomfield, L., Quist-Therson, E., Tomlin, M. and Messer, D. (2004). Routine examination of the newborn: the EMREN study. Evaluation of an extension of the midwife role including a randomised controlled trial of appropriately trained midwives and paediatric senior house officers. Health Technology Assessment 8 (14). Truswell, A. S. (2001). Levels and kinds of evidence for public-health nutrition. Lancet 357 (9262): 1061-2. Turner, L., Jacobsen, C., Humenczuk, M., Singhal, V. K., Moore, D. and Bell, H. (2001). The effects of lactation education and a prosthetic obturator appliance on feeding efficiency in infants with cleft lip and palate. Cleft Palate Craniofacial Journal 38 (5): 519-24. [Table 70] Uhari, M., Mantysaari, K. and Niemala, M. (1996). A meta-analytic review of the risk factors for acute otitis media. Clinical Infectious Disease 1996 (22): 6. United Kingdom National Case Control Study Group (1993). Breastfeeding and risk of breast cancer in young women. British Medical Journal 307: 17-20. Van den Bosch, C. A. and Bullough, C. H. (1990). Effect of early suckling on term neonates core body temperature. Annals of Tropical Paediatrics 10 (4): 347-53. Van Esterik, P. and Greiner, T. (1981). Breastfeeding and womens work: constraints and opportunities. Studies in Family Planning 12 (4): 184-97. Victora, C. G., Behague, D. P., Barros, F. C., Olinto, M. T. and Weiderpass, E. (1997). Pacifier use and short breastfeeding duration: cause, consequence, or coincidence? Pediatrics 99 (3): 445-53. Virtanen, S. M., Fasanen, L. and Aro, A. (1991). Infant feeding in Finnish children, 7yrs of age with newly diagnosed IDDM. Diabetes Care 14: 415-7.

160

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Von Kries, R., Koletzko, B., Sauerwald, T., von Mutius, E., Barnetrt, D., Grunert, V. et al. (1999). Breastfeeding and obesity: cross sectional study. British Medical Journal 319: 147-50. Waldenstrom, U. and Nilsson, C. A. (1994). No effect of birth centre care on either duration or experience of breast feeding, but more complications: findings from a randomised controlled trial. Midwifery 10 (1): 8-17. [Table 35] Wallace, L. M., Freeman, T., Latham, L., Walshe, K. and Spurgeon, P. (2001). Organisational strategies for changing clinical practice; how Trusts are meeting the challenges of clinical governance. Quality in Health Care 10: 76-82. Waller, H. (1946). The early failure of breastfeeding. Archives of disease in childhood 21: 1-12. Walsh, D., Harris, M. and Shuttlewood, S. (1999). Midwifery birthing practice: change through audit. British Journal of Midwifery 7 (7): 432-5. Watase, S., Mourine, A. P. and Tipton, G. A. (1998). An analysis of malocclusion in children with otitis media. Pediatric Dentistry 20 (5): 327-30. Watters, N. and Sparrow, B. (1990). Combined care: as good as they say? Canadian Nurse 86 (2): 28-32. [Table 36] Watters, N. E. and Kristiansen, C. M. (1995). Two evaluations of combined mother-infant versus separate postnatal nursing care. Research in Nursing and Health 18 (1): 17-26. [Table 36] Weatherley-White, R. C., Kuehn, D. P., Mirrett, P., Gilman, J. I. and Weatherley-White, C. C. (1987). Early repair and breast-feeding for infants with cleft lip. Plastic and Reconstructive Surgery 79 (6): 879-87. Weber, F., Woolridge, M. W. and Baum, J. D. (1986). An ultrasonographic study of the organisation of sucking and swallowing by newborn infants. Developmental Medicine and Child Neurology 28: 19-24. Westphal, M. F., Taddei, J. A., Venancio, S. I. and Bogus, C. M. (1995). Breast-feeding training for health professionals and resultant institutional changes. Bulletin of the World Health Organization 73 (4): 461-48.

White, A., Freeth, S. and OBrien, M. (1992). Infant Feeding 1990. London: HMSO. Whitley, N. (1978). Preparation for breastfeeding. Journal of Obstetric, Gynaecologic and Neonatal Nursing 7: 44-8. Widstrom, A. M., Wahlberg, V., Matthiesen, A. S., Eneroth, P., Uvnas-Moberg, K., Werner, S. and Winberg, J. (1990). Short-term effects of early suckling and touch of the nipple on maternal behaviour. Early Human Development 21 (3): 153-63. Williams, A. (1997). Hypoglycaemia of the newborn: a review. Bulletin of the World Health Organization 75 (3): 261-90. Wilson, A. C., Forsyth, J. S., Greene, S. A., Irvine, L., Hau, C. and Howie, P. W. (1998). Relation of infant diet to childhood health: seven year follow up of cohort of children in Dundee infant feeding study. British Medical Journal 316 (7124): 21-5. Winterburn, S., Jiwa, M. and Thompson, J. (2003). Maternal grandmothers and support for breastfeeding. Journal of Community Nursing 17 (12): 4-9. [Table 10] Woolridge, M. W. (1986a). The anatomy of infant sucking. Midwifery 2 (4): 164-71. Woolridge, M. W. (1986b). Aetiology of sore nipples. Midwifery 2 (4): 172-6. Woolridge, M. W. (1995). Breastfeeding: physiology into practice. In: Davies, D. P., (ed) Nutrition in Child Health. First edition. pp13-31. London: Royal College of Physicians Press. Woolridge, M. W. (1996). Problems of establishing lactation. Food and Nutrition Bulletin 17 (4): 316-23. Woolridge, M. W. and Fisher C. (1988). Colic overfeeding and symptoms of lactose malabsorption in the breastfed baby: a possible artefact of feeding management? Lancet 2: 382-4. Woolridge, M. W., Baum, J. D. and Drewett, R. F. (1980). Effect of a traditional and of a new nipple shield on sucking patterns and milk flow. Early Human Development 4 (4): 357-64.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

161

Woolridge, M. W., Greasley, V. and Silpisornkosol, S. (1985). The initiation of lactation: the effect of early versus delayed contact for suckling on milk intake in the first week post-partum. A study in Chiang Mai, Northern Thailand. Early Human Development 12: 269-78. World Bank (2001). World Development Report. Oxford: Oxford University Press. World Health Organization (1981a). Report on the WHO collaborative study on breast-feeding: contemporary patterns of breast-feeding. pp31-46. Geneva: World Health Organization. World Health Organization (1981b). Code of marketing of breast milk substitutes. Geneva: World Health Organization. World Health Organization (1991a). Innocenti Declaration on the protection, promotion and support of breastfeeding. Geneva: World Health Organization. World Health Organization (1991b). Indicators for assessing breastfeeding practices. Geneva: World Health Organization. World Health Organization (1993). Breastfeeding counselling: a training course. Geneva: World Health Organization. World Health Organization (1998a). Evidence for the Ten Steps to Successful Breastfeeding. Geneva: World Health Organization. World Health Organization (1998b). HIV and infant feeding: guidelines for decision makers. Geneva: World Health Organization. World Health Organization (2003). Global strategy for infant and young child feeding. Geneva: World Health Organization. World Health Organization (2004). Enterobacter sakazakii and other micro organisms in powdered infant formula: meeting report. MRA Series 6. Geneva: World Health Organization.

World Health Organization/UNICEF (1989). Protecting, promoting and supporting breast-feeding: the WHO collaborative study on breastfeeding. Geneva: World Health Organization. Wrenn, S. S. (1997). Effects of a model-based intervention on breastfeeding attrition. Thesis. University of Texas [Table 11] Wright, A. L., Naylor, A., Wester, R., Bauer, M. and Sutcliffe, E. (1997). Using cultural knowledge in health promotion: breastfeeding among the Navajo. Health Education and Behavior 24 (5): 625-39. [Table 36] Yamauchi, Y. and Yamanouchi, I. (1990). The relationship between rooming-in/not rooming-in and breast-feeding variables. Acta Paediatrica Scandinavica 79 (11): 1017-22. Ylikorkala, O., Kauppila, A., Kivinen, S. and Viinikka, L. (1982). Sulpiride improves inadequate lactation. British Medical Journal 285 (6337): 249-51. Ylikorkala, O., Kauppila, A., Kivinen, S. and Viinikka, L. (1984). Treatment of inadequate lactation with oral sulpiride and buccal oxytocin. Obstetrics and Gynecology 63 (1): 57-60. Yngve, A. and Sjostrom, M. (2001). Breastfeeding in countries of the European Union and EFTA: current and proposed recommendations, rationale, prevalence, duration and trends. Public Health Nutrition 4 (2B): 631-45 [Erratum 4 (6):1306]. Zarate, A., Villalobos, H., Canales, E. S., Soria, J., Arcovedo, F. and MacGregor, C. (1976). The effect of oral administration of thyrotrophin-releasing hormone on lactation. Journal of Clinical Endocrinology and Metabolism 43: 301-5.

162

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Glossary

Attachment The relationship of the babys mouth to the mothers breast (see Positioning). Counseling/counselling see Education. Discharge packs Educational materials and samples of maternity/paediatric products given to mothers in hospital, usually at the time of discharge home. Education Provision of factual or technical information, usually grounded in professional expertise and delivered in written or verbal form. In the US studies in this review, the term counseling/counselling is often used for education interventions as defined here. Guthrie test A screening test for phenylketonuria (PKU) and congenital hypothyroidism performed on all UK babies, normally on the 6th day of life. In Scotland infant feeding status is included among the data collected at Guthrie testing. Medicaid US government funded health service provision for socially disadvantaged groups. Rooming-in Having the baby stay with the mother on the postnatal ward rather than be placed in a separate nursery. Social marketing The use of marketing skills from business to advance social change. Social norm Expected pattern of behaviour. Positioning The relationship of the babys body to the mothers body (see Attachment)

Support Forms of breastfeeding support described in this review include peer support, voluntary support and health professional support. Components of breastfeeding support were not always reported in the studies included in this review, and differed from study to study. We identify here the main forms of support offered in UK settings: Peer support: normally support from local women, who have themselves breastfed, and who are normally paid for their work by a community or statutory health organisation, including Sure Start schemes and primary care trusts (PCTs). Peer supporters may or may not have received additional training in providing breastfeeding support Voluntary support: normally support from volunteer breastfeeding counsellors, who have themselves breastfed, and who have received breastfeeding support training from organisations with established programmes and accreditation. In the UK these organisations include the Association of Breastfeeding Mothers, Breastfeeding Network, La Leche League and the National Childbirth Trust Health professional support: normally describes support from a range of health professionals who work with mothers and babies, and who have may have undertaken specific additional training in breastfeeding support/lactation consultancy. In the UK these include midwives, health visitors, general practitioners, public health nurses, and paediatric and nursery nurses. Lactation consultants are a relatively new professional group who specialise in the care of breastfeeding women and babies. Around the world they are at different stages in the process of establishing themselves as a separate profession in their countries. In the UK they are at an early stage of development and although there are 163

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

around 150 qualified lactation consultants (with the internationally-recognised IBLCE qualification) there are few opportunities for lactation consultants to be appointed to remunerated posts. In our search for this review, we did not find any studies that examined breastfeeding support from crche workers or nannies, or others working with mothers and babies. We also included two studies, one from the UK and one from Australia, of additional postnatal support that included support for infant feeding, regardless of method of feeding. In the Australian setting (Quinlivan et al., 2003) additional postnatal support had the main aim of enabling each mother to safely maintain guardianship of her child. In the UK setting (Morrell et al., 2000a/b), the aim of additional postnatal support in the community was to help women rest and recover after childbirth, and the outcomes measured were womens general health and risk of postnatal depression. Sure Start UK government-funded schemes to support families with children under five years old with a range of health and social services.

164

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

APPENDIX 1 Search strategy


Breastfeeding common to Areas 1-4 breast feeding/ lactation/ milk ejection/ milk, human/ (breastfeed$ or breastfed).tw. (breast feed$ or breast fed).tw. (breastmilk or breast milk or baby milk).tw. (breast adj4 (fed or milk or feed$)).tw. (baby adj2 milk).tw. (lactation or lactating).tw. (nursing adj2 (mother or mothers)).tw. (nursing adj2 (baby or babies)).tw. (nursing adj2 (infant or infants)).tw. infant food/ infant feeding.tw. infant formula.tw. formula milk.tw. mixed feed$.tw. Public health interventions covering Areas 1, 2 and 4 exp health promotion/ health education/ patient education/ health fairs/ public health/ public health practice/ public health nursing/ preventive health services/ parental leave/ legislation/ attitude of health personnel/ peer group/ self-help groups/ social support/ community networks/ community health aides/ community health services/ community health nursing/ primary health care/ child health services/ infant welfare/ maternal health services/ maternal welfare/ maternal-child nursing/ maternal-child health centers/ infant care/ rooming-in care/ prenatal care/ postnatal care/ community adj2 (worker$1 or personnel).tw. health adj2 (worker$1 or visitor$1).tw. nurse midwives/ midwifery/ midwife$.tw. or midwives.tw. or birth attendant$1.tw. physicians, family/ general practitioner$1.tw. or general physician$1.tw. or family physician$1.tw. gp.tw. or gps.tw. or generalist$1.tw. nurses/ nurse practitioners/ counseling/ or directive counseling/ house calls/ world health organization/ united nations/ persuasive communication/ mass media/ delivery of health care/ delivery of health care, integrated/ health behavior/ public policy/ health policy/ nutrition policy/ government programs/ government agencies/ national health programs/ social control policies/ teaching/ (health adj4 promot$).tw. (health adj4 educat$).tw. ((antenatal or postnatal or prenatal or postpartum or post partum) adj2 (class$ or educat$ or care or practice$1)).tw. (campaign$ or policy or policies).tw. (promoting or promotion or promote$).tw. (educat$ or training).tw. (program$ or programme$).tw. (intervention$ or scheme$).tw. process evaluation$.tw. "process assessment (health care)"/ 165

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

"outcome and process assessment (health care)"/ (public adj2 health).tw. (public adj2 educat$).tw. (practice adj2 chang$).tw. (parental leave).tw. or (maternity leave).tw. or (paternity leave).tw. ((peer$1 or lay or professional$1 or community or agenc$) adj2 support).tw. ((peer$1 or lay or community) adj2 (group$1 or network$1)).tw. ((mother adj2 mother) adj support).tw. ((home or health or followup or follow up) adj visit$).tw. ((healthcare or care) adj provision).tw. rooming in.tw. counsel$.tw. (encourage$ or motivate$ or support or supportive or guidance).tw. (legislation or legal).tw. ((maternal or infant) adj4 (health or nursing or welfare or care)).tw. health sector initiatives.tw. mother friendly.tw. baby friendly.tw. la leche league.tw. wic.tw. sure start.tw. welfare food scheme.tw. unicef.tw. maternity alliance.tw. national childbirth trust.tw. donor milk.tw. milk banks/ medical audit/ or nursing audit/ audit$.tw. quality assurance, health care/ quality adj2 (assurance or assessment$1).tw. guidelines/ or practice guidelines/ or health planning guidelines/ guideline.pt. or practice guideline.pt. or guideline$1.tw. (opinion or professional) adj2 leader$1.tw. project adj2 champion$1.tw. interactive adj2 educat$.tw. workshop$1.tw. or work shop$1.tw. world health organization/ world health organisation.tw. world health organization.tw. who code.tw. 1 or 2 or 3 or 4 ((breastmilk or breast milk) adj substitut$).tw. 5 and 6 166

Clinical interventions covering Area 3 exp delivery, obstetric/ episiotomy/ or episiotom$.tw. home childbirth/ parturition/ exp labor, induced/ exp labor, obstetric/ exp labor onset/ exp labor presentation/ birthing centers/ delivery rooms/ childbirth.tw. or child birth.tw. or birth.tw. labour$1.tw. or labor$1.tw. caesarean$1.tw. or cesarean$1.tw. aquanatal.tw. or aqua natal.tw. prenatal adj2 water adj2 exercis$.tw. (amniotom$ or induction or inducing or induce$).tw. artificial$ adj2 membrane$1 adj2 (ruptur$ or punctur$).tw. (perineal adj2 care).tw. ((skin adj2 skin) adj contact).tw. early adj2 contact.tw. (timing adj4 (breastfeed$ or feed$)).tw. first feed$1.tw. ((frequency or duration) adj4 (breastfeed$ or feed$)).tw. supplement$.tw. (feeding adj4 (regime$ or schedule$1 or routine$1 or pattern$1)).tw. (food chart$1).tw. or (weight chart$1).tw. pacifiers/ or pacifier$1.tw. or dummy$1.tw. or dummies.tw. discharge pack$1.tw. exp lactation disorders/ (hypogalactia or galactorrhea or galactorrhoea or mastitis).tw. breast diseases/ or mastitis/ nipples/ ((preparing or prepare$ or preparation) adj2 nipple$).tw. ((sore or crack$ or fissure$ or size or small or large) adj2 nipple$).tw. ((sore or painful) adj2 breast$1).tw. ((flat or invert$ or retract$ or protract$ or non protract$) adj2 nipple$).tw. ((insufficient or poor or absence or lack or failure) adj2 (milk or supply)).tw. ((excess$ or spontaneous$) adj2 (milk or supply)).tw. (breast$1 adj2 (engorge$ or infection$ or abscess$ or abces$ or inflam$)).tw. object attachment/ (position$ or attachment or bonding).tw.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

(nipple adj2 confusi$).tw. (nipple adj2 shield$).tw. (breast adj2 pump$).tw. (hand adj2 express$).tw. (antenatal adj4 express$).tw. or (colostrum adj4 express$).tw. (breast adj2 (shell$1 or dressing$1 or application$1)).tw. infant, newborn, diseases/ hypoglycaemia/ or hypoglycaemi$.tw. or hypoglycemi$.tw. jaundice/ or exp jaundice, neonatal/ or hyperbilirubinemia/ or jaundice.tw. crying/ or colic/ or colic.tw. sucking behavior/ (suck$3 adj2 (problem$1 or difficult$ or dysfunction$)).tw. feeding behavior/ (feeding adj2 (problem$1 or difficult$)).tw. exp sleep disorders/ (sleep adj2 (problem$1 or difficult$ or dysfunction$)).tw. (sleepy adj2 (baby or babies or infant$1 or newborn$1)).tw. ((poor or slow or inadequate) adj2 (weight or gain)).tw. down syndrome/ cleft lip/ or cleft palate/ or harelip.tw. or cleft lip.tw. or cleft palate.tw. tongue tie.tw. drug therapy/ ointments/ surgery/ surgical procedures, operative/ womens health/ maternal behavior/ maternal nutrition/ prenatal nutrition/ ((maternal or prenatal or pregnancy) adj2 nutrition$).tw. exp smoking/ smoking cessation/ or tobacco use cessation/ tobacco use disorder/ tobacco smoke pollution/ smoking.tw. exp substance-related disorders/ alcohol drinking/ exp alcohol-related disorders/ ((drug or substance or alcohol or alcoholic) adj2 (abuse or dependence or intoxicat$ or withdraw$)).tw. Excluding letters and news items

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

167

APPENDIX 2 Excluded studies


A total of 54 studies for which data extraction was considered to be appropriate at the pre-screening stage, but which were subsequently excluded for reasons of ineligibility or lack of relevance, are listed here. Excluded from Section 1.1 because the study was not an RCT (n=14) Basha, M. Q. (1993). A study to examine the effectiveness of the postpartum parent support program (PPSP) on a group of primiparous mothers at four to six weeks postpartum. Master of nursing thesis, Memorial University of Newfoundland. Beake, S. (2002). Evaluation of the use of health care assistants to support disadvantaged women breastfeeding in the community. Report to the Department of Health, Infant Feeding Initiative. Fendrick, S. M., Major, A. L. and Brown, F. R. (1994). Nursing mothers service: a community breastfeeding program. Pediatric Nursing 20 (3): 241-4. Jovani Roda, L., Gutierrez Culsant, P., Aguilar Martin, C., Navarro Caballe, R., Mayor Pegueroles, I. and Jornet Torrent, Y. (2002). Influence of primary care personnel on breastfeeding duration. Anales Espanoles de Pediatria 57 (6): 534-9. Kistin, N., Abramson, R. and Dublin, P. (1994). Effect of peer counsellors on breastfeeding initiation, exclusivity, and duration among low-income urban women. Journal of Human Lactation 10 (1): 11-6. Long, D. G., Funk-Archuleta, M. A., Geiger, C. J., Mozar A. J. and Heins, J. N. (1995). Peer counselor program increases breastfeeding rates in Utah Native American WIC population. Journal of Human Lactation 11 (4): 279-84. Martens, P. J. (2002). Increasing breastfeeding initiation and duration at community level: an evaluation of Sagkeeng First Nations Community Health Nurse and Peer Counselling programs. Journal of Human Lactation 18 (3): 236-46. Mason, H. F., May, R. B., Gold, B., Tenney, B. L., Low, R. B. and Garbing, D. (1993). Effect of lactation educator on breastfeeding rates. Clinical Research 41 (4): 723-723. Quarles, A., Williams, P. D., Hoyle, D. A., Brimeyer, M. and Williams, A. R. (1994). Mothers intention, age, education and the duration and management of breastfeeding. Maternal-Child Nursing Journal 22 (3): 102-8. Rajan, L. and Oakley, A. (1990). Infant feeding practice in mothers at risk of low birth weight delivery. Midwifery 6 (1): 18-27. Schafer, E., Vogel, M. K., Viegas, S. and Hausafus C. (1998). Volunteer peer counselors increase breastfeeding duration among rural low-income women. Birth 25 (2): 101-6. Shaw, E. and Kaczorowski, J. (1999). The effect of peer counseling programme on breastfeeding initiation and longevity in a low-income rural population. Journal of Human Lactation 15 (1): 19-26. Thompson, B. (1998). Breakthrough in breastfeeding. Practising Midwife 1 (5): 35-7. Vari, P. M., Camburn, J. and Henly, S. J. (2000). Professionally mediated peer support and early breastfeeding success. Journal of Perinatal Education 9 (1): 22-30. Excluded from Section 1.2 because the study was not an RCT (n=8) Barwick, M., Chambers, L. and Jones, A. (1997). Positive intervention with breastfeeding in an Asian population. East Berkshire Community Health NHS Trust. Cervantes Pardo, A., Montoya Martinez, I., Hernandez Cuevas, J., Hernandez Requena, C., Talave Lopez, J., Lopez Sanchez Solis de Querol, M. and Canteras Jordana, M. (1993). Breast feeding promotion: is health education effective? [Spanish]. Anales Espanoles de Pediatria 39 (6): 529-34. Kramer, J. (1992). The relationship of prenatal education to successful breastfeeding. Master of science thesis. San Jose State Univeristy.

168

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Reifsnider, E. and Eckhart, D. (1997). Prenatal breastfeeding education: its effect on breastfeeding among WIC participants. Journal of Human Lactation 13 (2): 121-5. Sheehan, A. (1999). A comparison of two methods of breastfeeding education. Midwifery 15 (4): 274-82. Swanwick, L. A. (1992). Should mothers know more about breastfeeding? Midwives Chronicle 105 (1252):122-4. Torres Diaz, A., Lopez Tellez, A. and de Ramon Garrido, E. (1996). Importance of health education in the practice of breastfeeding [Spanish]. Atencion Primaria 18 (4): 164-7. Woods, A., Dykes, F. and Bramwell, R. (2002). An intervention study using a breastfeeding positioning and attachment tool. Clinical Effectiveness in Nursing (6): 134-42. Excluded from Section 1.3 because the study was not an RCT (n=10) Ahluwalia, I. B., Tessaro, I., Grummer-Strawn, L. M., MacGowan, C. and Benton-Davis, S. (2000). Georgias breastfeeding promotion program for low-income women. Pediatrics 105 (6): E85. Barber-Madden R. (1990). Design and implementation of a citywide breastfeeding promotion program: the New York City approach. Family and Community Health 12 (4): 71-8. Battersby S. (2002). Community breastfeeding scheme. A research and evaluation report for the Foxhill and Parson Cross Sure Start. www.sheffield.ac.uk/surestart/publns.html Bruce, N. G. A. (1995). Usefulness of a non-experimental study design in the evaluation of service developments for infant feeding in a general hospital. Social Science and Medicine 40 (8): 1109-16. Caulfield, L. E., Gross, S. M., Bentley, M. E., Bronner, Y., Kessler, L., Jensen, J., Weathers, B. and Paige, D. M. (1998). WIC-based interventions to promote breastfeeding among African-American Women in Baltimore: effects on breastfeeding initiation and continuation. Journal of Human Lactation 14 (1): 15-22.

Cohen, R., Lange, L. and Slusser, W. (2002). A description of a male-focused breastfeeding promotion corporate lactation program. Journal of Human Lactation 18 (1): 61-5. Gross, S. M., Caulfield, L. E., Bentley, M. E., Bronner, Y., Kessler, L., Jensen, J. and Paige, D. M. (1998). Counselling and motivational videotapes increase duration of breast-feeding in Africa-American WIC participants who initiate breast-feeding. Journal of the American Dietetic Association 98 (2): 143-8. McInnes, R. J., Love, J. G. and Stone, D. H. (2000). Evaluation of a community-based intervention to increase breast feeding prevalence. Journal of Public Health Medicine 22 (2): 138-45. Pugh, L. C., Milligan, R. A. and Brown, L. P. (2001). The breastfeeding support team for low-income, predominantly-minority women: a pilot intervention study. Health Care for Women International 22 (5): 501-5. Tuttle, C-R. and Dewey, K. G. (1995). Impact of a breastfeeding promotion program for Hmong women at selected WIC sites in Northern California. Journal of Nutrition Education 27 (2): 69-74. Excluded from Section 1.5 because the intervention had a wider focus than breastfeeding (n=8) Bussolati, G., Gambini, L., Musetti, M., Braibanti, S. and Capuano C. (2000). Early discharge and breast-feeding [Italian]. Acta Bio Medica de lAteneo Parmense 1: 681-5. Escobar, G. J., Braveman, P. A., Ackerson, L., Odouli, R., Coleman-Phox, K., Capra, A. M., Wong, C. and Lieu, T. A. (2001). A randomized comparison of home visits and hospital-based group follow-up visits after early postpartum discharge. Pediatrics 108 (3): 719-27. Leone, A., Bertino, E., Gennari, E., Monz, H., Olivetti, M., Bandelloni, A. M., Cavo, L., Coscia, A., Garzena, E., Soldi, A., Simonitti, A., Buttafuoco, V., Costa, S., Visentin, L., Colla, F., Renosio, M., DAmbrosio, L., Tosetto, D., Possidente, D. and Fabris, C. (2000). Early protected discharge of the mother and the neonate: case-control study [Italian]. Acta Bio Medica de lAteneo Parmense 1: 695-9.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

169

Lieu, T. A., Wikler, C., Capra, A. M., Martin, K. E., Escobar, G. J. and Braveman, P. A. (1998). Clinical outcomes and maternal perceptions of an updated model of perinatal care. Pediatrics 102 (6): 1437-44. Lieu, T. A., Braveman, P. A., Escobar, G. J., Fischer, A. F., Jensvold, N. G. and Capra, A. M. (2000). Randomized comparison of home and clinic follow-up visits after early postpartum hospital discharge. Pediatrics 105 (5): 1058-65. Madden, J. M., Soumerai, S. B., Lieu, T. A., Mandl, K. D., Zhang, F. and Ross-Degnan, D. (2003). Effects on breastfeeding of changes in maternity length-of-stay policy in a large health maintenance organization. Pediatrics 111 (3): 519-24. Svedulf, C., Engberg, I., Berthold, H. and Hoglund, I. (1998). A comparison of the incidence of breast feeding two and four months after delivery in mothers discharged within 72 hours and after 72 hours post delivery. Midwifery 14 (1): 37-47. Winterburn, S. and Fraser, R. (2000). Does the duration of postnatal stay influence breast-feeding rates at one month in women giving birth for the first time? A randomized control trial. Journal of Advanced Nursing 32 (5): 1152-7. Excluded from Section 1.5 after translation because it became clear that the study had no concurrent controls (n=1) Vazquez, A. M., Toral, I., Sanchez Ramos, J. L. and Galvez, M. (1992). Effect of post-natal visit on the involvement in the activities of the mother-child program [Spanish]. Atencion Primaria 10 (4): 712-8. Excluded from Section 3.3 because the study was not an RCT (n=1) Aliperti, L. C. and MacAvoy, S. (1996). Effect of hospital breastfeeding practice changes on breastfeeding duration. Journal of Perinatal Education 5 (3): 21-7. Excluded from Section 3.4 because the study was not an RCT (n=3) Dener, C. and Inan, A. (2003). Breast abscesses in lactating women. World Journal of Surgery 27 (2): 130-3. Roberts, K. L. (1995). A comparison of chilled cabbage leaves and chilled gel-packs in reducing breast engorgement. Journal of Human Lactation 11: 117-20.

Singla, S., Bishnoi, P., Kadian, Y. and Pawanjit Jindal, O. (2002). Evaluation of the role of antibiotics in the surgical management of breast abscess. Tropical Doctor 32 (3): 165-6. Excluded from the review because no numerical breastfeeding outcomes were reported (n=7) Aosheim, V., Dalhaug, R. E. and Vik, B. (2000). Breastfeeding in Norway. MIDIRS Midwifery Digest 10 (3): 372-4. Black, M. M., Siegel, E. H., Abel, Y. and Bentley, M. E. (2001). Home and videotape intervention delays early complementary feeding among adolescent mothers. Pediatrics 107 (5): E67. Clinton, B. (1988). Promoting maternal and child health in the context of rural poverty. Annual meeting of the American Public Health Association. Available from Educational Resources Information Center (ERIC) Document Reproduction Service (EDRS). Cox. S. and Turnbull, C. (1998). Developing effective interactions to improve breastfeeding outcomes. Part 2: antenatal empowerment of mothers for postnatal success in breastfeeding. Breastfeeding Review 6 (2): 17-22. Janssen, P., Klein, M., Harris, S., Soolsma, J. and Seymour, L. (2000). Single room maternity care and client satisfaction. Birth 27 (4): 235-43. Martens, P. J. (2000). Does breastfeeding education affect nursing staff beliefs, exclusive breastfeeding rates and BFHI initiative and compliance? Journal of Human Lactation 16 (4): 309-18. Michaels, M. (1993). Breastfeeding promotion in the Utah WIC program. Journal of Human Lactation 9 (3): 206-7. Excluded from the review because the paper was not available within our timeframe (n=2) Grunstra, E. A. and Rowe, S. (1992). The effects of a prenatal breastfeeding class on breastfeeding success and maternal perception of the infant: a replication. Masters thesis, Grand Valley State University. Moreno Manzanares, L., Cabrera Sanz, M. and Garcia Lopez, L. (1997). Breast-feeding [Spanish]. Revista de Enfermeria 20 (227-8): 79-84.

170

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

APPENDIX 3 Ongoing studies


These studies were ongoing when identified in our systematic search, and at the time we prepared our report their results were not available for inclusion. Carfoot, S., Williamson, P. and Dickson, R. (2005). A randomised controlled trial in the north of England examining the effects of skin-to-skin care on breastfeeding. Midwifery 21: 71-9. Inch, S., Law, S. and Wallace, L. (2003). Hands off! The Breastfeeding Best Start Project (1). The Practising Midwife 6 (10): 17-9. Inch, S., Law, S. and Wallace, L. (2003). Hands off! The Breastfeeding Best Start Project (2). The Practising Midwife 6 (11): 24-5. Muirhead, P. (ongoing). A randomized controlled study of the effect of organised peer support on the duration of breast feeding; and the consequences for infant morbidity. Oxenward Surgery, Kilwinning, Ayrshire, Scotland. Thomson, E. (ongoing). Breastfeeding and thrush research programme. Incidence of thrush and other infections as a cause of nipple/breast pain during breastfeeding. St Georges Hospital NHS Trust.

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

171

APPENDIX 4 Studies published since the search


Our systematic search was done in June 2003. We could not include studies published after this date because we did not identify them by repeating the systematic search. Studies listed here have come to our attention in various unsystematic ways. They were all published after June 2003, and may be relevant to areas of this review, but their results have not been included. Bertini, G., Perugi, S. and Dani C. (2003). Maternal education and the incidence and duration of breastfeeding: a prospective study. Journal of Pediatric Gastroenterology and Nutrition 37 (4): 447-52. Dodd, V. and Chalmers, C. (2003). Comparing the use of hydrogel dressings to lanolin ointment with lactating mothers. Journal of Obstetric, Gynecologic and Neonatal Nursing 32 (4): 486-94. Donath, S., Amir, L. and ALSPAC Study Team (2003). Relationship between prenatal infant feeding intention and initiation and duration of breastfeeding: a cohort study. Acta Paediatrica 92 (3): 352-6. Dykes, F., Hall Moran, V., Burt, S. and Edwards, N. (2003). Adolescent mothers and breastfeeding: experiences and support needs an exploratory study. Journal of Human Lactation 19 (4): 391-400. Dykes, F. (2003). Infant Feeding Initiative: a report evaluating the breastfeeding practice projects 1999-2002 executive summary. London: Department of Health. Forster, D., McLachan, H., Lumley, J., Beanland, C. and Waldenstrom, U. (2004). Two mid-pregnancy interventions to increase the initiation and duration of breastfeeding: a randomised controlled trial. Birth 31 (3): 176-82. Forster, D., McLachan, H., Lumley, J., Beanland, C., Waldenstrom, U., Harris, H., Earl, D. and Dyson, K. (2003). ABFAB. Attachment to the breast and family attitudes to breastfeeding. The effect of breastfeeding education in the middle of pregnancy on the initiation and duration of breastfeeding: a randomised controlled trial. BMC Pregnancy Childbirth 3 (1): 5. 172
The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

Heck, K., Schoendorf, K., Chavez, G. and Braveman, P. (2003). Does postpartum length of stay affect breastfeeding duration? A population-based study. Birth 30 (3): 153-9. Hogan, M., Westcott, C. and Griffiths, D. M. (2004). A randomised controlled trial of division of tongue-tie in infants with feeding problems. Archives of Disease in Childhood 89 (Suppl 1): A5. MacArthur, C., Winter, H. R., Bick, D. E., Lilford, R. J., Lancashire, R. J., Knowles, H., Braunholtz, D. A., Henderson, C., Belfield, C. and Gee, H. (2003). Redesigning postnatal care: a randomised controlled trial of protocol-based midwifery-led care focused on individual womens physical and psychological needs. Health Technology Assessment 7 (37): 97. Reeve, J., Gull, S., Johnson, M., Hunter, S. and Streather, M. (2004). A preliminary study on the use of experiential learning to support womens choices about infant feeding. European Journal of Obstetrics and Gynecology and Reproductive Biology 113 (2): 199-203. Ullah, S. and Griffiths, P. (2003). Does the use of pacifiers shorten breastfeeding duration in infants? British Journal of Community Nursing 8 (10): 458-63. Wolfberg, A., Michels, K., Shields, W., OCampo, P., Bronner, Y. and Bienstock, J. (2004). Dads as breastfeeding advocates: results from a randomised controlled trial of an educational intervention. American Journal of Obstetrics and Gynecology 191: 708-12.

APPENDIX 5 Countries meeting review criteria for developed country


Adapted from World Development Report (World Bank, 2001) Australia Austria Belgium Canada Denmark Finland France Germany Greece Iceland Ireland Italy Japan Luxembourg Netherlands New Zealand Norway Portugal Spain Sweden Switzerland United Kingdom United States

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

173

APPENDIX 6 Standard postnatal care in the UK


In the UK, standard NHS maternity care, funded by general taxation and free at the point of delivery, includes community based and hospital-based care. Almost all babies (97%) are born in hospital, so almost all women experience hospital delivery and postnatal care. Postnatal care is provided mainly by midwives, possibly with the support of ancillary staff, and with a routine check for all babies before discharge, usually conducted by a paediatrician (Townsend et al., 2004). The length of stay after a normal, uncomplicated birth is now short, ranging from around 6-48 hours. Variations in breastfeeding support exist between hospitals. Some have breastfeeding specialists on staff, but this is not universal. Some have achieved Baby Friendly status and others are aiming for this. Scotland has the highest rate of Baby Friendly Hospitals in the UK; England has the lowest (www.babyfriendly.org.uk). Most hospitals have implemented parts of the Baby Friendly programme, eg rooming-in. A strong network of primary healthcare professionals including general practitioners, midwives and health visitors provide community based antenatal and postnatal care. Midwifery home visits to mother and baby up to at least the 10th postnatal day, sometimes continuing until the 28th day, and home and clinic-based care from a health visitor until the child is five years old, are a standard part of UK maternity care. Community based midwives normally provide both antenatal and postnatal care to women living in a local area and/or registered with particular GPs/health centres. Hospital-based and community based midwives have their employer, management and accountability structures and policies for care in common. A small number of independent, self-employed midwives care for their own private caseload.

174

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

APPENDIX 7 UNICEF/WHO Ten Steps to Successful Breastfeeding


Every facility providing maternity services and care for newborn infants should: 1 2 3 4 5 Have a written breastfeeding policy that is routinely communicated to all healthcare staff Train all healthcare staff in skills necessary to implement this policy Inform all pregnant women about the benefits and management of breastfeeding Help mothers initiate breastfeeding within a half-hour of birth (modified in the UK to read soon after birth Show mothers how to breastfeed, and how to maintain lactation even if they should be separated from their infants Give newborn infants no food or drink other than breastmilk, unless medically indicated Practice rooming-in allow mothers and infants to remain together 24 hours a day Encourage breastfeeding on-demand Give no artificial teats or pacifiers (also called dummies or soothers) to breastfeeding infants

6 7 8 9

10 Foster the establishment of breastfeeding support groups and refer mothers to them on discharge from the hospital or clinic

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005

175

APPENDIX 8 Definitions of breastfeeding and replacement feeding terms World Health Organization
In 1991, the World Health Organization convened an informal meeting to establish definitions and indicators for assessing breastfeeding practices in household surveys. The main purpose of promoting the use of the indicators was to have a common set of measures to assess practices and monitor the progress of breastfeeding promotion programmes. Several of the following definitions are taken from the Innocenti Declaration on the protection, promotion and support of breastfeeding (World Health Organization, 1991a); others are taken from Code of marketing of breastmilk substitutes (World Health Organization, 1981b), Breastfeeding counselling: a training course (World Health Organization, 1993) and HIV and infant feeding: guidelines for decision makers (World Health Organization, 1998b). Breastfeeding The child has received breastmilk (direct from the breast or expressed) Breastfeeding practices may be further described according to timing and frequency. In terms of timing, breastfeeding may be described as on-demand (by the child) or on schedule (determined by a schedule or work/ separation demands of the mother). Exclusive breastfeeding The infant has received only breastmilk from his/her mother or a wet nurse, or expressed milk and no other liquids, or solids with the exception of drops or syrups consisting of vitamins, mineral supplements, or medicines A child may be exclusively breastfed with expressed human milk from his/her mother, a breastmilk donor or from a milk bank. Predominant breastfeeding The infants predominant source of nourishment has been breastmilk. However the infant may also have received water or water-based drinks (sweetened or 176 Any food, whether manufactured or locally prepared, suitable as a complement to breastmilk or to infant formula, when either become insufficient to satisfy the nutritional requirements of the infant. Such food is commonly called weaning food or breastmilk supplement. Partial breastfeeding Giving a baby some breastfeeds, and some artificial feeds, either milk or cereal, or other food. Replacement feeding The process of feeding a child who is not receiving any breastmilk with a diet that provides all the nutrients that the child needs. During the first six months this should be with a suitable breastmilk substitute commercial formula, or home-prepared formula with micronutrient supplements. After six months it should preferably be with a suitable breastmilk substitute, and complementary foods made from appropriately prepared and nutrientenriched family foods, given three times a day. If suitable breastmilk substitutes are not available, appropriately prepared family foods should be further enriched and given five times a day. flavoured waters, teas, infusions, etc); fruit juice; oral rehydration salts (ORS); drop and syrup forms of vitamins, minerals and medicines, and folk fluids (in limited quantities). With the exception of fruit juice and sugarwater, no food based fluid is allowed under this definition Folk fluids are liquids used for non-nutritional purposes; oil to relieve constipation, tea for relief of colic, etc. Breastmilk substitute Any food being marketed or otherwise presented as partial or total replacement from breastmilk, whether or not suitable for that purpose Complementary feeding

The effectiveness of public health interventions to promote the duration of breastfeeding Systematic review 1st edition May 2005