89
A. Dale and S. Cornwell
non-use. No such claims are made for synthetic lavender perineal discomfort than a group of mothers not using
oil, which on visual and olfactory inspection is very the preparation.
similar. These claims made for lavender oil, however, lack
any scientific basis. This investigation seeks to remedy
this position. FORMAT OF THE INVESTIGATION
A clinical trial was undertaken using three groups of A blind clinical trial involving three groups of mothers
mothers, one using an extract of lavender oil as a bath
was planned. Current practice makes claims only for pure
additive, one using a synthetic laboratory-produced oil
lavender oil but not for synthetic oil, although the basis
and a final control group using a 'generally regarded as
for both the claim and the distinction is difficult to
safe' (GRAS) compound (2-Methyl 3-Isobutyl Pyrazine)
identify in the literature. It made sense, therefore, to have
diluted 10 - lo in distilled water as an inert but aromatic
two groups, one using pure and one using synthetic
placebo (Bureau de Liaison 1990). The latter was chosen
lavender oil respectively, and another group using an
since the pure and synthetic oils both had odours, and it
inert aromatic compound for which no claims are made,
was felt important that the placebo should not be
as a 'control' group.
odourless.
90
Lavender oil and pm‘neal discomfort
Table 1 Variables included in the study Variables 1and 2 were completed daily by the mother,
3-8 by the attending midwife in hospital, 9-11 by the
1 Daily discomfort VAS score (10days) mother on the loth day and 12-18 by the community
2 Daily mood VAS score (10days)
midwife on the final visit.
3 Group to which mothers were randomly allocated
4 Parity: primiparouslmultiparous
5 Delivery type instrumentdnormal T R IA L PROCEDURE
6 Perineum: 1st to 3rd degree tear, episiotomy (intact Patient consent
excluded)
Approval for the trial was sought and obtained from the
7 Sutures: yeslno
8 Sutured by: doctor/midwife Ethical Committee. Batches of the three bath additives
9 Used additive on 10th day: yeslno were prepared in a laboratory, and delivered to the
10 Obstetric
Used other medication for pain relief during the 10 days: Unit. The authors and staff in the unit were not
yeslno informed of the identity of the three batches until after
11 What helped discomfort most: tablets/ointmentlice the data analysis for the trial was completed. The control
packsbathlother ‘GRAS’ compound was distinguishable from the other
12 Day perineum healed by: day 1-10 two by olfactory inspection. However, mothers entering
13 Use of additive each day: yes/no the trial were informed that the trial was of different bath
14 Perineal bruising apparent: yes/no additives, and not of lavender oil. From this perspective
15 Haemorrhoids: yeslno
Sutures removed: yeslno
the smell of the oil was not especially important. Never-
16
17 Infection presentlabsent theless, in a ’perfumery’ sense the GRAS compound did
18 Antibiotics used: yeslno not smell as ‘pleasant’ as the lavender and the synthetic
lavender oils. The additives did not colour the bath
water.
Antenatal visits of Midwives clinics and to the
The difference of 10 points between means to estimate Maternity Unit were routinely made for those booked for
sample size was selected to reflect a difference which confinement. During the course of these visits women
would be important in a clinical midwifery sense. were made aware that a trial of bath additives might be in
progress at the time of their confinement.
Following delivery, prior to the mother’s first bath, the
The hypothesis trial was explained by the attending midwife. A consent
form was signed if the mother decided to enter the trial.
More formally, the hypotheses tested could be expressed Women were reassured that they could withdraw, with-
thus: out prejudice, from the trial at any time. The mother was
then randomly allocated to one of the three groups. Each
Null hypothesis mother was at this point issued with a 5-ml bottle of one
There is no difference in the mean discomfort scores of of the three additives, and her name was written on it.
women in the three groups using different preparations as The midwife also explained in detail to the mother the
bath additives. method for completing the VAS scale to record her
discomfort and mood daily, half an hour after her daily
bath.
Variables included in the study
Ward procedure
In adltion to the independent variable (the group to
which the mother was randomly allocated) and the Daily bathing is a routine part of the mother’s care after
dependent variable (mother’s daily VAS score of perineal delivery. For each day the mother bathed in hospital, the
discomfort) it was clear that a variety of extraneous midwife ensured that six drops of the oil from the
variables could potentially influence the mother’s experi- mother’s named bottle was added to the bath. Half an
ence of perineal discomfort. Clearly one such variable hour after having the bath, the mother was asked to
might be the mother’s mood. All the variables included complete the VAS record indicating the degree of dis-
are shown in Table 1. In randomly allocating mothers to comfort being experienced. The state of the perineum
the three groups one hopes to achieve similar distribution was assessed as part of the midwife’s normal daily
of these variables between groups. examination of the mother.
91
A. Dale and 5. Cornwell
Prior to transfer home the mother was prepared by the a distinction between ‘withdrawals’ and ‘non-returns’ in
midwifery staff to continue the bathing procedure, and practice. Essentially, both groups comprised cases from
was given a reference sheet to take home with her. whom no outcome data were received. Particular diffi-
culty was experienced in obtaining returns from mothers
who resided out of the Health District, but chose to
Following transfer home
deliver at Hinchingbrooke Hospital, and were therefore
Daily visits by the community midwife provided some included in the study. These represent LO%, and some-
opportunity to reinforce the procedure with the mother, times 30% of bookings in any month, due to the
and for assessment of the perineum to be continued until popularity of the unit.
the loth day. Women were deemed to have had ’no outcome’ if
At the last visit to the mother the community midwife efforts to retrieve the data records had failed after a
collected completed visual analogue scales and question- period of 2 months following delivery. They were
naires from her, and collated them with the midwifery then replaced, maintaining the random allocation of
records for return to the department. This process women to the three groups. The rate of non-return of
worked well with mothers residing within the Health outcome data did not seem to be related to treatment
District served by the local community services, but less allocation.
well when mothers were resident outside the Health Deciding when to stop recruitment was a difficult
District. decision given the non-completion rate. Attempts were
made to contact mothers whose data sheets had not been
returned. When contact was made the most common
The sample
reason given for non-return was that the mother had not
Mothers entered the trial sample as outlined in the wished to continue bathing daily.
sections above. They were entered during the trial period In making a comparison of mothers who returned no
if they had delivered vaginally. Excluded from the sample data with those who did, no differences were found in
were: terms of bath data. There were no significant differences
in terms of parity (xz=2.219; d.f.=< P=01363), tears/
those delivered by Caesarian section,
episiotomy (x2=0.585; d.f.= I; P=0.4443), whether they
those with active skin disorders such as eczema or
were sutured (x2= 1.466; d.f. = 1; P=0.2260) or whether
psoriasis,
they had a normal or instrumental delivery (x2=0.227;
those having stillbirths or neonates who were criti-
d.f.= I; P=0*6337).
cally ill,
those without bathing facilities at home,
those not wishing to participate.
Statistical methods
Data were analysed using SPSS/PC+ Version 3.1. Analy-
No-outcome subjects
sis of variance was used for the analysis of discomfort and
Table 2 below shows the extent of non-returned data mood scores, which were tested and found to be nor-
during the trial period. The percentage varied between mally distributed and x2 for the analysis of other nominal
groups from 38.0%to 41.5%.It proved difficult to make level variables.
92
Lavender oil and perineal discomfort
Initial variables
Multiparious 581134 (44.3) 601133 (45.1) 671120 (55.8) 0.1011
Instrumental 151134 (I1.2) 141133 (10.5) 71120 (5.8) 0.2842
Episiotomy 581134 (43.3) 521133 (391) 461120 (40.3) 0.8582
Sutured 1301135 (97.0) 132/133 (992) 1201120 (100.0) 0.0869
Sutured by midwife 108/131 (82.4) 144/131 (87.0) 103/120 (85.8) 0.5597
Outcome variables
Used no other medication 601134 (44.8) 631128 (49.2) 591118 (50.0) 0.3718
Bath helped most 791124 (63.7) 791120 (65.8) 671109 (61.5) 0.9914
Bruising 31121 (2.5) 51124 (4.0) 4/107 (3.7) 0.7791
Haemorrhoids 291123 (23.6) 251125 (20.0) 221108 (20.4) 0.7555
Sutures removed 261121 (21.5) 201125 (16.0) 201109 (18.3) 0.5408
Infection 111123 (8.9) 51124 (4.0) 31109 (2.8) 0.0811
Antibiotics 161123 (13.0) 91124 (7.3) 101109 (9.2) 0.3043
Day healed (1-10) (Kruskal-Wallis test=0.138) 0.9331
Used additive loth day 124/134 (92.5) 1171126 (92.9) 101/115 (878) 0.3071
Did not use additive daily 4123 (3.3) 81125 (6.4 14/109 (87.2) 0.0175
93
A. Dale and S. Cornwell
the 10-day period, there are some consistently lower are shown in Table 6. They show very similar small
mean scores in the lavender oil group, particularly values over the 10 days, indicating no differences
between days 3 and 5. Kazis ef af. (1989) have discussed between the groups.
the fact that statistical significance may occur but have
little clinical relevance and equally items of clinical
significance may not be demonstrably statistically signifi- Secondary analysis
cant. They propose that an analysis of effect size can help
to standardize the magnitude of differences between A second analysis of data was undertaken along the lines
mean scores. In general terms an effect size of 0.2 is suggested by Matthews ef al. (1990) and published after
considered small, one of 0.5 moderate and one of 0.8 to this study began. They argue that fuller use of data for a
be large. The effect size of differences between day 1 and set of measures gathered over time can be made by
day 10 within each group are shown in Table 5. calculating summary measures for individuals in the
Clearly the effect size increases over time within each study. In the case of the data gathered in the course of
group consistently. This pattern is not unexpected as this study the main issue is whether the decrease in
perineal discomfort decreases over time. The pattern of discomfort scores occurs more speedily in one group of
effect sizes is also consistent within each group. How- mothers than another. Measuring a fall to a percentage of
ever, more important is the effect sizes between the the individual's initial score seemed particularly appropri-
lavender oil group and the GRAS (control) group. These ate, and a fall of 50% seemed a natural level to choose
94
Lavender oil and perineal discomfort
Table 5 Wect size within groups over 10 days, using day 1 as Table 7 Distribution of cases with and without a 50%
a baseline reduction in discomfort scores, by group
Table 6 Effect sizes between lavender oil and GRAS (control) 1 (baseline) - - -
groups 2 14 11 13
3 13 13 13
Lavender oil GR4S 4 20 11 13
5 9 17 14
6 17 13 12
1 0.05 0.04
7 8 14 11
2 0.02 0.02
8 11 17 7
3 0.22 0.20
9 11 6 8
4 0.23 0.2I
10 6 5 5
5 0.23 0.21
6 0.09 0.08 Total 108 105 95
7 0.04 0.03
8 0.01 001
9 - 0.05 - 004 Mean days 4.5 4.8 4.5
10 - 0.01 - 0.01 SD 2.4 2.2 2.3
since a reduction of this magnitude occurred in 80% of cases where the day 1 value was missing; for such cases
the cases. the day 2 value was taken as the baseline. The results are
Some cases had to be excluded from this secondary summarized in Table 8.
analysis because measurements were recorded on two The difference between the means is not significant
few days, or a discomfort score on day I was zero. This using a one-way analysis of variance (F0.58).
accounted for I3 cases. There were a further 66 cases
where discomfort scores failed to fall to 50% of their
Side-effe d s
initial value. These include nine with one measurement
missing and one with two missing. The remaining cases No side-effects were identified by the women or mid-
were examined to determine whether those cases which wives during the trial. Those using the additives smelling
did not achieve a 50% reduction in discomfort scores of lavender made gratuitous comments indicating the
were unevenly distributed between groups. The result is substances were pleasurable to use. This was less so for
shown in Table 7, and there is no significant difference those using the control (GRAS)additive.
between groups.
The data were then analysed to determine whether this
DISCUSSION
was a difference in the mean time to reach a 50%
reduction in discomfort scores in the cases where this did The current surge in interest in aromatherapy may create
occur. Care was taken in making this calculation over the a climate in which new practices may be introduced
95
A. Dale and S. Cornwell
without adequate critical appraisal. This study has exam- Berkow R. et al. (1980) The Merck Index 9th edn. Merck,
ined the practice of adding six drops of pure lavender oil Rahway, New Jersey.
to bath water daily for 10 days after childbirth to reduce Bureau de Liaison (1990) Flavour and Ertract Manufacfurers
perineal discomfort. Association FEMA 3133). Bureau de Liaison, Brussels.
The study provides no evidence to support the prac- Duncan G., Bushnes M. & Gilles J. (1989)Comparison of verbal
tice as it stands. However, comments collected show that and visual analogue scales for measuring the intensity and
most mothers using the oil found it a pleasant experience. unpleasantness of experimental pain. Pain 37(3), 295-303.
There seem to be few, if any, previous systematic Hutchins C., Ferrciza C. & Norman Taylor J. (1985) A
comparison of local agents in the relief of discomfort after
studies which evaluate the clinical use of lavender oil (or
episiotomy. Journal of Obstetrics and Gynuecology 6(1), 45-46.
other oils used in aromatherapy) against a control. It is
Kazis L., Anderson J. & Meenan R.F. (1989) Effect size for
difficult, therefore, to place the results of this study in any
interpreting changes in health status. Medical Care 27(3)
context. We would suggest that further research is (suppl). 178-189.
therefore required. Lautie R. & Passebecq A. (1984) Aromatherapy and the Use of
In particular, further research may confirm or refute the Plant Essences in Healing. Thorsons, Wellingborough.
results of this study, and could valuably explore the effect Matthews J., Altman D., Campbell M. & Royston P. (1990)
of varying the amount of oil used as an additive and/or Analysis of serial measurements in medical research. British
the mode of application. Medical Journal 300(6719), 230-235.
Mulgrew M. (1989) Mothers’ Experience of Post Natal Care.
Internal Report, Hinchingbrooke Midwifery Service,
Acknowledgements
Huntingdon, Cambridgeshire.
The authors would like to thank the midwifery staff in Poco& S. (1986) Clinical Trials: A Practical Approach. John
Huntingdon Health Authority for their co-operation in Wiley, Chichester, pp. 123-130.
this study, and the consultant obstetric staff for their Reynolds J. et al. (eds) (1982) Martindale, The ,Extra Phartnaco-
support. Without such co-operation the study would poeia 29th edn. The Pharmaceutical Press, London.
have been difficult to complete. We are grateful, too, to Sleep J. & Grant A. (1988a) Relief of pain following cluldbirth:
D r George Dodd of the Olfactory Research Unit at a survey of midwifery practice. Midwifery 4(3), 118-122.
Warwick University for his specialized support and the Sleep J.& Grant A. (1988b) Effects of salt and Savlon bath
preparation of the additive in his laboratory. W e are also concentrate post partum. Nursing Times 84(21), 55-57.
indebted to Paul Smith, principal statistician at the East Sleep J., Grant A., Garcia J., Elboume D., Spencer J. & Chalmers
1. (1984) West Berkshire perineal management trial. British
Anglian Regional Health Authority for this advice and
Medical Journal 289(6445), 587-590.
help in analysing the data. Thanks are also due to Clive
Harries in the Research and Development Department of
technical support, and to Jenny Martin for her unflagging Bibliography
secretarial services.
Finally we must acknowledge the grant from the East Amould-Taylor W. (1981) Aromatherapy for the Whole Person.
Thornes, Cheltenham.
Anglian Regional Health Authority’s Locally Organized
Research Committee which made the work possible. Davis P. (1988) Aromatherapy: A n A-Z. Daniel, Saffron
Walden.
Jackson L. (1987)Aromatherapy. Dorling Kindersley, London.
References Stead C. (1986) The Power of Holistic Aromutherapy. Javelin,
Aitken R. (1969)Measurement of feeling using visual analogue Poole, Dorset.
scales. Proceedings of the Royal Society of Medicine 63(lo), 989. Tisserand R. (1987) The Art of Aromatherapy. Daniel, Saffron
Banos J., B o d F., Canellas M., Bassoles A., Ortega F. & Walden.
Bigorra J. (1989) Acceptability of visual analogue scales Van Toller G. & Dodd G. (eds) (1988)Perfumery. Chapman and
in the clinical setting: a comparison with verbal rating Hall, London.
scales. Methods and Findings in Erperimental and Clinical
P h a ~ C o h g y11(2), 123-127.
96