Anda di halaman 1dari 79

Editorial

In 2005, WHO urged its member states to commit to


develop their health systems so that all people have
access to essential health services without the nancial
hardship associated with payment. Yet the world is still
a long way from universal health coverage for rich and
poor countries, with more than half the population
lacking any type of social protection, and billions of
people lacking health care.
To guide countries at all development stages to
achieve this goal more quickly and sustain existing
achievements, the World Health Report 2010 is subtitled
Health System Financing: The Path to Universal Coverage.
Margaret Chan, Director General of WHO, launched
the report on Nov 22 at an International Ministerial
Conference in Berlin, Germany, where high-ranking
decision makers gathered to exchange experiences and
discuss strategies for health nancing.
The launch of the report at the ministerial conference
means the beginning of a new push towards universal
coverage in the context of both the economic downturn
and the continuing rise of health-care expenditure. The
overarching message is optimistic and practicalthat
all countries can take steps to move faster to achieve this
goal and maintain their achievements. Many countries are
beginning to embrace universal coverage. The USA and
Chinatwo major economic powers whose health funding
had previously been based on free-market mechanisms
are moving back to some element of universal coverage.
Several low-income and middle-income countries have
recently made great strides in developing their health
systems towards universal coverage. These include Brazil,
Chile, Columbia, Costa Rica, Cuba, Ghana, Kyrgyzstan,
Mongolia, Rwanda, Sri Lanka, Thailand, and Moldova.
There are three barriers to progress. By far the greatest
obstacle is over-reliance on direct payments (out-ofpocket payments at the time people use services), which
exclude 13 billion poor people from accessing health
services and drive an additional 100 million people
into poverty yearly. Such fees encourage health service
overuse by people who can pay and underuse by those
who cannot. Sadly, the poorer the country, the greater
the reliance on direct payments, which represented more
than 50% of total health expenditures in 33 mostly lowincome countries in 2007. Only when the percentage falls
to 1520% does the risk of nancial catastrophe become
www.thelancet.com Vol 376 November 27, 2010

negligible. Other impediments include the availability


of resources and the inecient and inequitable use of
resources. Global annual expenditure on health is about
US$53 trillion, of which it is estimated that 2040% is
wastedmost often by inappropriate use of medicines.
To overcome these obstacles, the report provides feasible
guidance and actions: raising sucient funds, reducing
direct payments through compulsory prepayment and
pooling to spread the nancial risks of the ill across the
population, and spending money more eciently and
equitably.
This timely report sends a message of hope.
Nevertheless, David Evans, one of the principal authors
from WHO, stressed that many questions remain. For
instance, it is not known which people miss out on
which services, long-term nancial hardship or the
impact of direct payments on individuals, and how
much several newly launched global-health initiatives
cost. Also unclear is the proportion of external
nancial assistance that actually gets through to lowincome countries, which rely increasingly on external
assistance to achieve the Millennium Development
Goals and to save more than 3 million additional lives
by 2015. Ironically, while spending on improving the
health of people in most countries has increased at a
historically unprecedented rate, there is not enough
research that assesses health policy and health systems
to guide investment. Health systems research is
underutilised and badly needed, said Julio Frenk, Dean
of the Harvard School of Public Health, in the First
Global Symposium on Health Systems Research held in
Montreux, Switzerland on Nov 1619, 2010.
That universal coverage is so distant for so many
people is unacceptable. Health and universal coverage
were discussed as a legal right in the 1948 UN Universal
Declaration of Human Rights. Universal health coverage
comes down to political will. Eective governance is
the key to health-system reform. Policy makers must
prioritise health in their governmental budgets and
move health nancing for universal coverage to the top
of the political agenda. Meanwhile, the international
community should use all means possible to support
low-income countries to obtain enough aid and use it
more transparently and eciently to improve the health
of their people. The Lancet

UN Photo/Eskinder Debebe

Striving for universal health coverage

For the World Health Report


2010 see http://www.who.int/
whr/2010/en/index.html
For more on Universal Health
Coverage see http://www.
thelancet.com/journals/lancet/
article/PIIS0140-6736
(10)62006-5/fulltext, and http://
www.thelancet.com/journals/
lancet/article/PIIS0140-67
36(10)62058-2/fulltext.
For more on the First Global
Symposium on Health Systems
Research see http://www.
hsr-symposium.org/.

1799

Editorial

United Nations Childrens Fund (UNICEF)

Female genital mutilation and social change

For The dynamics of social change:


Towards the abandonment of
female/genital mutilation/
cutting in ve African countries
see http://www.unicef-irc.org/
publications/618

Changing social expectations is the key to ending the


practice of female genital mutilation or cutting, according
to a new report by UNICEF, The dynamics of social change:
towards the abandonment of female genital mutilation/
cutting in ve African countries. Worldwide, up to 140 million
girls and women are estimated to have undergone some
form of genital mutilationa recognised violation of
human rights and a procedure complicated by severe
haemorrhage, infection, and diculties with delivery
and sexual intercourse. Yet many parents, inuenced by
community expectations, believe that cutting secures
social and economic security for their daughters. For them,
the social harm of not cutting outweighs any physical,
psychological, or even legal risk. It is insucient to simply
provide individual families information of the harm of the
practice says author Francesca Monetti. While cutting is
seen as the only possible social way to act, one of the rst
steps towards abandoning the practice is to promote the
alternative, not cutting.
The report looked at how Kenya, Senegal, Sudan,
Egypt, and Ethiopia have promoted the type of social

change needed for communities to abandon the practice.


Although the national prevalence of genital mutilation
remains high in Egypt (91%) and Sudan (89%), as a
result of community-driven change all ve countries
have reported a decrease in the percentage of women
who think the practice should continue. Successful
approaches include reinforcing the positive aspects
of local culture rather than demonising traditional
practices, using the media to elevate the status of being
uncut, human-rights education linked to local values
and aspirations, and the development of linkages with
neighbouring countries and countries of migration.
Eorts to end female genital mutilation started decades
ago. The reports insights represent an important step
towards ending this and other practices that are damaging
to womens health. Whilst respecting the subtle message
that understanding and changing social expectations
takes time, governments and donors must act quickly
and decisively to support what is working to end female
genital mutilation. Communities ready to adopt the social
expectation not to cut can then do so. The Lancet

McPolicy: bringing you the Big Mac society

Getty Images

If you were a UK Health Secretary faced with soaring rates


of obesity, alcohol misuse, and diet-related diseases,
what would you do? Were you to take an evidence-based
approach, you might consider minimum pricing per
unit of alcohol and restrictions on its availability. You
might look at toughening the regulation of how the
least healthy foods are marketed to children. You could
even demand that manufacturers reformulate their least
healthy products to meet minimum nutritional standards.
Or you could, if your name was Andrew Lansley, dismiss
all of the above and instead invite representatives of
McDonalds, PepsiCo, and the drinks giant Diageo among
others, to submit their policy suggestions on how best to
deal with the UKs public-health crises for a forthcoming
governmental white paper.
After the initial surprise, it can still take a while for
the bizarre reality to sink inthat the companies who
have proted the most from the epidemics of obesity
and alcohol misuse should now be responsible for
setting the agenda on public health simply beggars
1800

belief. Whatever sage wisdom the various captains


of the food and drink industry have to impart, it will
certainly be in the narrow interests of their shareholders,
whose continued wealth is contingent on maintaining
precisely the status quo that brought about the current
public-health crises. Perhaps their feelings of corporate
responsibility will extend to plugging the funding gap
left by Education Secretary Michael Goves decision to
remove 162 million of funding to English schools for
the sports-for-all programme, which tackled low levels
of physical activity in children.
The creeping inuence of corporate power on
public policy is not news to anyone in the UK, but the
breathtaking speed and scale by which the UK coalition
Government is embracing the agenda of business at the
expense of the health of the electorate is an unwelcome
novelty. By putting the interests of big business at the
heart of public-health policy, Lansley is ensuring that
the UKs big society will not be shedding the pounds any
time soon. The Lancet
www.thelancet.com Vol 376 November 27, 2010

Comment

Doctors and climate change

www.thelancet.com Vol 376 November 27, 2010

shoulder with the people of Pakistan as victims of the


use of fossil fuel energy.
Health professionals everywhere have a responsibility
to put health at the heart of climate change negotiations.
Firstly, because climate change already has, and will
continue to have, a major adverse impact on the health
of human populations.5 Secondly, because reducing
greenhouse-gas emissions has unrivalled opportunities
for improving public health.6 Indeed, moving to a low
carbon economy could be the next great public health
advance. The hazards to human health from climate
change are well documented. Strong evidence already
exists that climate change will aect rates of malnutrition,
diarrhoea, malaria, deaths as a result of oods, and
temperature related deaths from cardiovascular disease.5
More recently, the health benets of reducing
greenhouse-gas emissions have been assessed and
quantied. Meeting greenhouse-gas emissions targets
in the transport sector will require substantial increases
in walking and cycling, with corresponding reductions in
car use.7 The available epidemiological evidence linking
physical activity and health has shown that this would
dramatically reduce rates of chronic disease, with around
a 1020% reduction in ischaemic heart disease and
stroke, 12% reduction in breast cancer, 8% reduction in
dementia, and 6% reduction in depression.7 This last
estimate considered the eects of physical activity only.
It did not take into account the mental health benets of

Published Online
November 18, 2010
DOI:10.1016/S01406736(10)62106-X

Corbis

Links between climate policy and health policy must not


be overlooked.
In November 2010, representatives from countries
around the world will meet in Cancn, Mexico, at the
2010 United Nations Climate Change Conference.1 Here
they will attempt to draft a treaty aimed at stabilising
atmospheric greenhouse-gas concentrations at a level
that will prevent catastrophic climate change. What a
pity the meeting had not been scheduled in Pakistan.
Then the anger of those whose livelihoods have been
destroyed by the biblical oods that have washed away
the hopes of a nation would surely have focused the
delegates minds. Alternatively, the meeting could
have been held in Western Russia, where record high
temperatures, wild res, droughts, and crop failures
have precipitated a state of emergency. The conference
might even have been held in Mozambique, where
rapidly rising wheat prices have caused rioting in the
streets. All of these climatic events and their predictable
human aftermath occurred this year and all are made
more probable by climate change, the main cause of
which is the increase in anthropogenic greenhouse-gas
emissions, mostly from the burning of fossil fuels.
But perhaps Mexico is not such a bad location for
the climate conference after all. Mexico is second only
to the US with regard to the prevalence of obesity.
One in four Mexicans is obese.2 If the delegates at the
climate conference think that obesity and climate
change are unrelated, they would be wrong. The planet
is getting hotter, its people are getting fatter, and the
use of fossil fuel energy is the cause of both.3 Large
increases in motor vehicle trac in Mexican towns and
cities have decimated levels of physical activity. This,
combined with increased availability of energy dense
food, has propelled the body-mass index in the entire
population upwards. Mexicans are paying for these
changes in terms of reduced health and wellbeing,
with increased rates of diabetes, heart disease, stroke,
and cancer. Unchecked car use has also conspired with
rapid population growth and topology to make Mexico
City one of the most polluted cities in the world. The
city topped the list in a 2010 IBM poll of commuter
pain, with 22% of commuters spending more than
two hours a day travelling to and from work.4 In this
respect, the people of Mexico stand shoulder to

Wild re near Moscow, Russia on Aug 5, 2010

1801

Comment

For the Climate and Health


Council pledge see http://www.
climateandhealth.org/pledge

urban greening, reduced community severance, reduced


fatness, and less noise.
The project also considered the health eects of
reducing livestock production to limit the cattle
related methane emissions and deforestation that are
contributing to global warming.8 A reduction in animal
products in the diet would reduce consumption of
saturated animal fats and result in a large (about 15%) fall
in ischaemic heart disease. Reducing meat consumption
might also reduce rates of cancer of the colon and
rectum. Colorectal cancer is the second most common
cancer in men after lung cancer and meat consumption
is an established risk factor. Eating less saturated fat
and taking more physical activity will reduce levels of
population fatness. Consuming less animal products will
also reduce food prices because cattle are fed on grain
and high meat consumption forces up world grain prices.
Feeding grain to animals is an inecient use of food
energy in a world where millions of people go hungry.9
A reduction in car use would also aect food prices.
In April 2008, Evo Morales, president of the poor and
increasingly hungry Bolivia, pleaded la vida primero los
autos segundos (life rst, cars second), exhorting the
wealthy world to stop burning food in their cars. He was
objecting to Western governments policies on biofuels
for transport. However, car use and food prices were
linked long before the introduction of biofuels. Car use
drives up food prices because oil is a key agricultural input.
A reduction in car use is essential to prevent starvation.
Responding to climate change could be the most
important challenge that health professionals face.
The Climate and Health Council was established to
enable health professionals around the world to take
personal and collective action against the causes of
climate change, and to insist that global health is central
in climate change negotiations.10 It seeks to provide
information on all aspects of health and climate change

and suggests a range of actions that health professionals


can take. We invite colleagues everywhere to join us
in tackling this major public health scourge of the
21st century. By signing the Climate and Health Council
pledge, you will join the thousands of health professionals
already committed to action. Please contact your health
minister to ensure that the links between climate policy
and health policy are known and fully taken into account
at COP16 in Cancn, COP17 in South Africa next year, and
indeed in all climate change negotiations.
*Ian Roberts, Robin Stott
London School of Hygiene and Tropical Medicine,
London WC1E 7HT, UK (IR); and Climate and Health Council (RS),
on behalf of the Climate and Health Council executive
ian.roberts@lshtm.ac.uk
This Comment is being published simultaneously in the BMJ and
the Finnish Medical Journal.
We declare that we have no conicts of interest. RS is the co-chair and IR is an
executive member of the Climate and Health Council, a component part of the
charity, Knowledge into Action.
1

3
4

5
6
7

9
10

UNFCCC. The United Nations Climate Change Conference in Cancun,


29 November10 December 2010. http://unfccc.int/meetings/
cop_16/items/5571.php (accessed Nov 16, 2010).
Ruiz-Arregui L, Castillo-Martnez L, Orea-Tejeda A, Meja-Arango S,
Miguel-Jaimes A. Prevalence of self-reported overweight-obesity and its
association with socioeconomic and health factors among older Mexican
adults. Salud Pblica Mx 2007. http://www.scielosp.org/scielo.php?script=
sci_arttext&pid=S0036-36342007001000007 (accessed Nov 16, 2010).
Roberts I, Edwards P. The energy glut: the politics of fatness in an overheating
world. Zed Books, 2010.
Goldsmith B. IBM nds worlds worst commutes. Toronto Sun 2010.
www.torontosun.com/life/2010/06/30/14564796.html (accessed
Nov 16, 2010).
McMichael A, Woodruf R, Hales S. Climate change and human health: present
and future risks. Lancet 2006; 367: 85969.
Haines A, Wilkinson P, Tonne C, Roberts I. Aligning climate change and public
health policies. Lancet 2009; 374: 203558.
Woodcock J, Edwards P, Tonne C, et al. Public health benets of strategies to
reduce greenhouse-gas emissions: urban land transport. Lancet 2009;
374: 193043.
Friel S, Dangour AD, Garnett T, et al. Public health benets of strategies to
reduce greenhouse-gas emissions: food and agriculture. Lancet 2009;
374: 201625.
Roberts I. The economics of tackling climate change. BMJ 2008; 336: 16566.
Climate and Health Council. http://www.climateandhealth.org/ (accessed
Nov 16, 2010).

Health co-benets of policies to tackle climate change


In 2009, The Lancet published a Series of groundbreaking
papers presenting evidence for certain actions to tackle
climate change that might improve health.16 These health
co-benetswhich are additional to the global-health
benets that come from mitigationcould partly oset
the costs of tackling climate change. This positive news
1802

contrasts with the widespread belief that policies to


tackle climate change will be socially and economically
demanding. The public health benets of mitigation need
to be more prominent in international negotiations and
domestic policy making. The Series coincided with the last
set of UN climate-change talks in Copenhagen, Denmark;
www.thelancet.com Vol 376 November 27, 2010

Comment

For the Climate and Health


Council pledge see http://www.
climateandhealth.org/pledge

urban greening, reduced community severance, reduced


fatness, and less noise.
The project also considered the health eects of
reducing livestock production to limit the cattle
related methane emissions and deforestation that are
contributing to global warming.8 A reduction in animal
products in the diet would reduce consumption of
saturated animal fats and result in a large (about 15%) fall
in ischaemic heart disease. Reducing meat consumption
might also reduce rates of cancer of the colon and
rectum. Colorectal cancer is the second most common
cancer in men after lung cancer and meat consumption
is an established risk factor. Eating less saturated fat
and taking more physical activity will reduce levels of
population fatness. Consuming less animal products will
also reduce food prices because cattle are fed on grain
and high meat consumption forces up world grain prices.
Feeding grain to animals is an inecient use of food
energy in a world where millions of people go hungry.9
A reduction in car use would also aect food prices.
In April 2008, Evo Morales, president of the poor and
increasingly hungry Bolivia, pleaded la vida primero los
autos segundos (life rst, cars second), exhorting the
wealthy world to stop burning food in their cars. He was
objecting to Western governments policies on biofuels
for transport. However, car use and food prices were
linked long before the introduction of biofuels. Car use
drives up food prices because oil is a key agricultural input.
A reduction in car use is essential to prevent starvation.
Responding to climate change could be the most
important challenge that health professionals face.
The Climate and Health Council was established to
enable health professionals around the world to take
personal and collective action against the causes of
climate change, and to insist that global health is central
in climate change negotiations.10 It seeks to provide
information on all aspects of health and climate change

and suggests a range of actions that health professionals


can take. We invite colleagues everywhere to join us
in tackling this major public health scourge of the
21st century. By signing the Climate and Health Council
pledge, you will join the thousands of health professionals
already committed to action. Please contact your health
minister to ensure that the links between climate policy
and health policy are known and fully taken into account
at COP16 in Cancn, COP17 in South Africa next year, and
indeed in all climate change negotiations.
*Ian Roberts, Robin Stott
London School of Hygiene and Tropical Medicine,
London WC1E 7HT, UK (IR); and Climate and Health Council (RS),
on behalf of the Climate and Health Council executive
ian.roberts@lshtm.ac.uk
This Comment is being published simultaneously in the BMJ and
the Finnish Medical Journal.
We declare that we have no conicts of interest. RS is the co-chair and IR is an
executive member of the Climate and Health Council, a component part of the
charity, Knowledge into Action.
1

3
4

5
6
7

9
10

UNFCCC. The United Nations Climate Change Conference in Cancun,


29 November10 December 2010. http://unfccc.int/meetings/
cop_16/items/5571.php (accessed Nov 16, 2010).
Ruiz-Arregui L, Castillo-Martnez L, Orea-Tejeda A, Meja-Arango S,
Miguel-Jaimes A. Prevalence of self-reported overweight-obesity and its
association with socioeconomic and health factors among older Mexican
adults. Salud Pblica Mx 2007. http://www.scielosp.org/scielo.php?script=
sci_arttext&pid=S0036-36342007001000007 (accessed Nov 16, 2010).
Roberts I, Edwards P. The energy glut: the politics of fatness in an overheating
world. Zed Books, 2010.
Goldsmith B. IBM nds worlds worst commutes. Toronto Sun 2010.
www.torontosun.com/life/2010/06/30/14564796.html (accessed
Nov 16, 2010).
McMichael A, Woodruf R, Hales S. Climate change and human health: present
and future risks. Lancet 2006; 367: 85969.
Haines A, Wilkinson P, Tonne C, Roberts I. Aligning climate change and public
health policies. Lancet 2009; 374: 203558.
Woodcock J, Edwards P, Tonne C, et al. Public health benets of strategies to
reduce greenhouse-gas emissions: urban land transport. Lancet 2009;
374: 193043.
Friel S, Dangour AD, Garnett T, et al. Public health benets of strategies to
reduce greenhouse-gas emissions: food and agriculture. Lancet 2009;
374: 201625.
Roberts I. The economics of tackling climate change. BMJ 2008; 336: 16566.
Climate and Health Council. http://www.climateandhealth.org/ (accessed
Nov 16, 2010).

Health co-benets of policies to tackle climate change


In 2009, The Lancet published a Series of groundbreaking
papers presenting evidence for certain actions to tackle
climate change that might improve health.16 These health
co-benetswhich are additional to the global-health
benets that come from mitigationcould partly oset
the costs of tackling climate change. This positive news
1802

contrasts with the widespread belief that policies to


tackle climate change will be socially and economically
demanding. The public health benets of mitigation need
to be more prominent in international negotiations and
domestic policy making. The Series coincided with the last
set of UN climate-change talks in Copenhagen, Denmark;
www.thelancet.com Vol 376 November 27, 2010

Comment

www.thelancet.com Vol 376 November 27, 2010

The IAMP statement urges that goals for health


improvement should be one of the main criteria in
deciding on mitigating measures for climate change. The
statement advises that the often overlooked co-benets
must be given greater prominence in international
negotiations and that health ministries should become
more involved in decisions on climate change. For this
change to happen, the health, scientic, transport,
agriculture, energy, and industry sectors should
collaborate to study and implement climate-change
mitigation and adaptation measures that benet health.
Furthermore, in view of our increased understanding
that health systems are major sources of greenhouse-gas
emissions,9 professional communities in health should
provide leadership to reduce these emissions.
The 2006 Stern Review10 emphasised the economic
consequences of climate change. The medical science
community should now do the same for health, and the
IAMP statement can contribute to this endeavour. The
growing evidence about the relation between climate
change and health presents another lens through which
climate change can be perceived.11 This evidence also
oers a new political space in which climate change
negotiations and national policy formulation can occur.
Although the eects of mitigation policies on the climate
take time to manifest and are often widely dispersed
worldwide, the health co-benets can be realised more
directly and quickly, thus making them more tangible
and attractive to policy makers and the public.
The IAMP statement provides new impetus to revive
international negotiations and domestic action. Notably,
academies and their international networks, such as
IAMP, recognise an increasing responsibility to provide
independent, expert, evidence-based advice to inform
and inuence policy development and, thereby, to
catalyse change. With extensive expertise in a wide range
of disciplines, national academies, and the networks in
which they participate, oer a rich resource for national
governments and international institutions when faced
with global challenges such as climate change.

Getty Images

however, despite the best eorts of the environmental


and health communities to emphasise the urgency of
action, many see the Copenhagen meeting as having
been a disappointment. No internationally binding
treaty to address climate change was signed to replace
the Kyoto Protocol, which expires in 2012, and health
remained too low on the agenda. What can we hope for
the next set of talks that will take place in Cancun, Mexico,
on Nov 29? Can new allies and evidence be brought to
bear on this substantial public health threat?
The InterAcademy Medical Panel (IAMP), the global
network of academies of medicine, and medical sectors
within science academies have built on work started by
the UK Academy of Medical Sciences by publishing a
statement re-emphasising the importance of the health
co-benets of policies to mitigate climate change, and
stressing the major threat of climate change to health.7,8
This alliance shows the potentially important contribution
of national academies to international aairs, alongside
nation states, charitable foundations, companies, and
non-governmental organisations. Endorsement of the
statement by academies from so many dierent countries
is particularly important in view of the international nature
of climate change, and the shift in world power towards
the eastern and southern regions of the world. The IAMP
statement reviews some of the main examples of how
actions to mitigate climate change can also provide cobenets for health, while acknowledging that there are
complex connections between policies to mitigate climate
change and human health, and there are substantial gaps
in knowledge. For example, increasing the eciency
of household cooking stoves in India could result in
reductions in childhood respiratory infections and in adult
heart and lung disease, and might reduce the emission of
black carbon and other greenhouse pollutants.
A 10-year programme to introduce 150 million
low-emission stoves in India could prevent about
2 million premature deaths. Reduced use of private cars
in urban areas and an increased proportion of more
active alternatives in urban transport, such as cycling
or walking, in New Delhi and London could reduce the
burden of heart disease, diabetes, some cancers, and
depression. These and other co-benets greatly support
arguments for emissions reduction, therefore providing
added incentives for countries to adopt such policies
early, and helping to oset the economic and societal
costs of tackling climate change.

*Detlev Ganten, Andy Haines, Robert Souhami


German National Academy of Sciences Leopoldina,
Halle 06108, Germany (DG); London School of Hygiene and
Tropical Medicine, London, UK (AH); and Academy of Medical
Sciences, London, UK (RS)
detlev.ganten@charite.de

1803

Comment

AH was Chair of the Task Force on Climate Change Mitigation and Public
Health. We declare that we have no conicts of interest.
1
2
3
4
5
6

Wilkinson P, Smith KR, Davies M, et al. Public health benets of strategies


to reduce greenhouse-gas emissions: household energy. Lancet 2009;
374: 191729.
Woodcock J, Edwards P, Tonne C, et al. Public health benets of strategies
to reduce greenhouse-gas emissions: urban land transport. Lancet 2009;
374: 193043.
Friel S, Dangour AD, Garnett T, et al. Public health benets of strategies to
reduce greenhouse-gas emissions: food and agriculture. Lancet 2009;
374: 201625.
Markandya A, Armstrong BG, Hales S, et al. Public health benets of
strategies to reduce greenhouse-gas emissions: low-carbon electricity
generation. Lancet 2009; 374: 200615.
Haines A, McMichael AJ, Smith KR, et al. Public health benets
of strategies to reduce greenhouse-gas emissions: overview and
implications for policy makers. Lancet 2009; 374: 210414.
Smith KR, Jerret M, Anderson HR, et al. Public health benets of
strategies to reduce greenhouse-gas emissions: health implications
of short-lived greenhouse pollutants. Lancet 2009; 374: 2091103.

10
11

Academy of Medical Sciences. Mitigating climate change: improving


global health. November, 2009. http://www.acmedsci.ac.uk/
p99puid168.html (accessed Nov 19, 2010).
InterAcademy Medical Panel. Statement on the health co-benets of
policies to tackle climate change. Nov 12, 2010. http://www.iamp-online.
org/resources/health-benets-of-climate-change-mitigation.pdf/view
(accessed Nov 19, 2010).
NHS Sustainable Development Unit. NHS England carbon emissions
carbon footprinting report. September, 2008. http://www.sdu.nhs.uk/
documents/publications/1263313924_jgyW_nhs_england_carbon_
emissions_carbon_footprinting_r.pdf (accessed Nov 19, 2010).
Stern N. The economics of climate change. http://webarchive.
nationalarchives.gov.uk/+/http://www.hm-treasury.gov.uk/
stern_review_report.htm (accessed Nov 19, 2010).
Gill M, Stott R. Health professionals must act to tackle climate change.
Lancet 2009; 374: 195355.

Childhood pneumonia: a neglected, climate-sensitive disease?

Getty Images

Reports from WHO and the Intergovernmental Panel


on Climate Change (IPCC) list undernutrition, diarrhoea,
and vector-borne disease as the most important health
eects of climate change.1,2 Although these disorders
are of major importance, childhood pneumonia, which
is responsible for 17% of childhood deaths worldwide,3
is rarely mentioned in the context of climate change.
Respiratory infections follow seasonal patterns. In temperate settings, respiratory illness is most common in
winter months.4 However, the epidemiology is quite
dierent in tropical settings, where most childhood
deaths due to pneumonia occur, with the incidence
of lower respiratory-tract infection being generally
increased during the rainy season.510 For example, in The
Gambia, west Africa, the incidence of clinical pneumonia
in children (per 1000 person-years) was about 409 in the
rainy season (95% CI 391427), compared with 243 in the
hot, dry season (229258), and 160 in the cool, dry season
(148173).6 In time-series analyses, the number of days of
rainfall per month in Malaysia was positively associated
with incidence of infection with respiratory syncytial virus
(RSV) in children (p=001).9 Similarly, daily occurrence
of rainfall in Indonesia was positively associated with
incidence of RSV infection in children (incidence ratio
164, 95% CI 113238, p=0009).10 Rainfall models from
the IPCC predict overall increases in tropical rainfall, with
more intense rainy seasons in Asia, Africa, the Pacic, and
in parts of South America than in subtropical and some
temperate regions.11 Such changes are probably associated
with an increase in childhood pneumonia.
1804

The health eects of climate change are very much on


WHOs agenda, exemplied by international meetings
in the lead up to the 16th Conference of the Parties to
the UN Framework Convention on Climate Change
in Cancun, Mexico, from Nov 29, to Dec 10, 2010. In
view of the association between seasonal rainfall and
childhood pneumonia in the tropics, climate change
could potentially increase the incidence of childhood
pneumonia in tropical settings, both directly and
indirectly, and delegates should consider this eect when
advocating for health concerns to be included in policy
initiatives. Many biologically plausible mechanisms might
contribute to an increase in the incidence of pneumonia
as a result of climate change. More time indoors or
under cover because of increased rainfall will increase
crowding and exposure to biomass fuel smoke, and will
reduce exposure to sunlight. Crowding will be further
exacerbated by large-scale population displacement.
Bacterial survival and virus stability in aerosols might be
increased by higher humidity.12 An increase in childhood
undernutrition because of climate change will increase
pneumonia deaths44% of pneumonia deaths in
children under 5 years are caused by undernutrition.13 In
many parts of Africa and Asia, the rainy season coincides
with the preharvest season when food shortages are
most likely to occur, particularly in poorer communities.
If climate change leads to delayed rainy periods, this
seasonal undernutrition might increase and excessive rain
might reduce crop yield and compromise food security.14
Increased temperatures and drier dry seasons are
www.thelancet.com Vol 376 November 27, 2010

Comment

expected to further reduce worldwide food supplies, with


the largest burden on those living in lower latitudes.1,2
Childhood pneumonia is climate sensitive and too
important to ignore. The seasonal drivers of the incidence
of childhood pneumonia in tropical countries should be
elucidated to better predict the health eects of climate
change in tropical settings. Future assessments of the
health eect of climate change for low-income and
middle-income countries in tropical settings should
specically quantify the predicted eect of climate
change on childhood pneumonia.
Stuart Paynter, Robert S Ware, Philip Weinstein,
Gail Williams, *Peter D Sly
School of Population Health (SP, RSW, GW) and Queensland
Childrens Medical Research Institute (RSW, PDS), University of
Queensland, Brisbane, QLD 4029, Australia; and Graduate Research
Centre, University of South Australia, Adelaide, SA, Australia (PW)
p.sly@uq.edu.au
We declare that we have no conicts of interest.
1

Confalonieri U, Menne B, Akhtar R, et al. Human health. In: Parry ML,


Canziani OF, Palutikof JP, van der Linden PJ, Hanson CE, eds. Climate change
2007: impacts, adaptation and vulnerability. 2007. http://www.ipcc.ch/pdf/
assessment-report/ar4/wg2/ar4-wg2-chapter8.pdf (accessed Nov 22, 2010).
WHO. Protecting health from climate change: connecting science, policy
and people. 2009. http://whqlibdoc.who.int/publications/2009/
9789241598880_eng.pdf (accessed Nov 19, 2010).

4
5

8
9

10

11

12

13

14

WHO. The global burden of disease: 2004 update. 2008. http://www.who.


int/healthinfo/global_burden_disease/GBD_report_2004update_full.pdf
(accessed Nov 22, 2010).
Dowell SF, Whitney CG, Wright C, Rose CE Jr, Schuchat A. Seasonal patterns
of invasive pneumococcal disease. Emerg Infect Dis 2003; 9: 57479.
Ye Y, Zulu E, Mutisya M, Orindi B, Emina J, Kyobutungi C. Seasonal pattern
of pneumonia mortality among under-ve children in Nairobis informal
settlements. Am J Trop Med Hyg 2009; 81: 77075.
Enwere G, Cheung YB, Zaman SM, et al. Epidemiology and clinical features
of pneumonia according to radiographic ndings in Gambian children.
Trop Med Int Health 2007; 12: 137785.
Moura FEA, Nunes IFS, Silva GB, Siqueira MM. Short report: respiratory
syncytial virus infections in northeastern Brazil: seasonal trends and
general aspects. Am J Trop Med Hyg 2006; 74: 16567.
Shek LP, Lee BW. Epidemiology and seasonality of respiratory tract virus
infections in the tropics. Paediatr Respir Rev 2003; 4: 10511.
Chan PW, Chew FT, Tan TN, Chua KB, Hooi PS. Seasonal variation in
respiratory syncytial virus chest infection in the tropics. Pediatr Pulmonol
2002; 34: 4751.
Omer SB, Sutanto A, Sarwo H, et al. Climatic, temporal, and geographic
characteristics of respiratory syncytial virus disease in a tropical island
population. Epidemiol Infect 2008; 136: 131927.
Christensen JH, Hewitson B, Busuioc A, et al. Regional climate projections.
In: Solomon S, Qin D, Manning M, et al, eds. Climate change 2007: the
physical science basis. 2007. http://www.ipcc.ch/pdf/assessment-report/
ar4/wg1/ar4-wg1-chapter11.pdf (accessed Nov 19, 2010).
Yusuf S, Piedimonte G, Auais A, et al. The relationship of meteorological
conditions to the epidemic activity of respiratory syncytial virus.
Epidemiol Infect 2007; 135: 107790.
WHO. Global health risks: mortality and burden of disease attributable to
selected major risks. 2009. http://www.who.int/healthinfo/global_burden_
disease/GlobalHealthRisks_report_full.pdf (accessed Nov 19, 2010).
Naylor RL, Battisti DS, Vimont DJ, Falcon WP, Burke MB. Assessing risks
of climate variability and climate change for Indonesian rice agriculture.
Proc Natl Acad Sci USA 2007; 104: 775257.

Delirium treatment: an unmet challenge


Rivastigmine, a cholinesterase inhibitor, has been used
to treat delirium in elderly patients with stroke.1 A
biologically plausible premisethat impaired cholinergic
transmission might either cause or worsen deliriumled
to a randomised, placebo-controlled, double-blind trial by
Maarten van Eijk and colleagues2 in The Lancet in which
they added rivastigmine or placebo to usual treatment
of patients in intensive care. The trial was halted at
104 patients by the drug safety and monitoring board
(DSMB) because of increased mortality (12/54 in the
rivastigmine group, 4/50 in the placebo group; p=007)
and a worse outcome. The rivastigmine group had a longer
median duration of delirium (50 days vs 30 days, albeit
with a p value of 006). Additionally, delirium severity
seemed worse in the rivastigmine group. Because of the
small numbers, long-term outcomes were not assessed.
Two dimensions of this study deserve highlighting. In
science, publication bias lies heavily in favour of studies
that show therapeutic superiority. Studies that do
www.thelancet.com Vol 376 November 27, 2010

not reveal benet are, however, at least as important:


their publication invites caution and serves as a gentle
reminder of the fundamental notion of primum non
nocere. A pragmatic DSMB ensured these principles were
upheld in this study. Second, a therapeutic intervention
for delirium was carefully studied. Delirium has long been
described as harmful, yet trials addressing pharmacological
intervention for delirium in critical care can be counted on
the ngers of one hand. Indeed, there is little evidence
to support any therapeutic approach, in any population
of patients, despite the prevalence of delirium and its
associated morbidity.
This study is not the rst in which critical care physicians
had hoped to ameliorate outcome by correcting a
biological marker of disease. Corticosteroid repletion in
sepsis3 and vasopressin repletion in shock4 are examples
of similar eorts. These studies, like the one reported
here, not only showed no benet but demonstrated
additional harm in some patients. Seeking to restore

Published Online
November 5, 2010
DOI:10.1016/S01406736(10)61998-8
See Articles page 1829

1805

Science Photo Library

Comment

homoeostasis or to normalise associated biological


abnormalities in disease might thus be too simplistic.
Delirium in critical care seems to be complex. The
syndrome is reported with alarmingly dierent incidences
in this population: some studies report greater than 80%,
whereas van Eijk and colleagues study counted less than
10% in the 6724 patients screened. Few psychometrically
validated screening tools exist. Some, like the confusion
assessment method for the intensive care unit used in
this study, suggest that the diagnosis is binary; others,
like the intensive care delirium screening checklist, grade
syndromes from no abnormalities to subsyndromal
delirium5 to delirium, in addition to describing specic
symptoms.6 Risk factors, which might help to stratify
patients and identify responders, dier according to the
group describing them.7 Additionally, environmental
factors (such as early mobilisation), individual clinical
characteristics, and choice of sedative drugs all play a role in
its development and course. Thorough descriptions of the
natural history of the disease and a sound understanding
of the basic underlying mechanisms are lacking.
Challenging questions remain as to the nature of
delirium: is it an acute marker of cerebral dysfunction,
a predictor of long-term cognitive outcome, or both?
Plum and Posner,8 in their classic 1972 publication on the
diagnosis of stupor and coma, described a continuum
between delirium and coma in the diseased brain, but
the same continuum cannot be presumed to exist in
the healthy brain. Whether the two pathologies can be
considered a composite outcome is debatable.9 Moreover,
in intensive care units across continents, healthy brains
1806

are exposed to variable sedation practices, and some to


pharmacologically induced coma. How much of the longterm cognitive morbidity can be attributed to sedation
depth and the drug used,10 or to confusion and delirium11
is thus dicult to establish. Additionally, psychiatric
diagnoses or abnormalities seem to be more common in
the intensive care unit than are any other disturbance.12
Diagnostic categories (delirium, depression, and posttraumatic stress disorder) have not been comprehensively
compared with patients outcomes. Finally, confounders
other than psychiatric disturbances might aect longterm cognition. For instance, postoperative cognitive
dysfunction, a disorder which aects up to 25% of surgical
patients,13 can result in cognitive abnormalities for up
to 3 years after surgery. Long-term outcome studies
of survivors quality of life and burden on caregivers
following intensive care unit admission remain sparse.
The unexpected results of van Eijk and colleagues
study beg the question: why did rivastigmine not seem
to be benecial? Delirium in an elderly patient with
Alzheimers disease but who is not critically ill might be a
dierent disease. Baseline cognitive function is dierent,
as is the degree of systemic inammation, a potential
factor in psychiatric or psychological dysfunction.14
Neurohormonal transmission might vary signicantly
from one individual to the next, as might the response
to drug treatment. Better or more complex explanations
await future investigations from neuroscientists and
bedside clinicians alike.
Yoanna Skrobik
Soins Intensifs, Hpital Maisonneuve Rosemont, Montral, QC,
Canada H1T 2M4
skrobik@sympatico.ca
I declare that I have no conicts of interest.
1

Oldenbeuving AW, de Kort PL, Jansen BP, Kappelle LJ, Roks G. A pilot study
of rivastigmine in the treatment of delirium after stroke: a safe alternative.
BMC Neurol 2008; 8: 34.
van Eijk MMJ, Roes KCB, Honing MLH, et al. Eect of rivastigmine as an
adjunct to usual care with haloperidol on duration of delirium and
mortality in critically ill patients: a multicentre, double-blind,
placebo-controlled randomised trial. Lancet 2010; published online Nov 5.
DOI:10.1016/S0140-6736(10)61855-7.
Sprung CL, Annane D, Keh D, et al, for the CORTICUS Study Group.
Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008;
358: 11124.
Russell JA, Walley KR, Singer J, et al, for the VASST Investigators.
Vasopressin versus norepinephrine infusion in patients with septic shock.
N Engl J Med 2008; 358: 87787.
Ouimet S, Riker R, Bergeron N, Cossette M, Kavanagh B, Skrobik Y.
Subsyndromal delirium in the ICU: evidence for a disease spectrum.
Intensive Care Med 2007; 33: 100713.
Marquis F, Ouimet S, Riker R, Cossette M, Skrobik Y. Individual delirium
symptoms: do they matter? Crit Care Med 2007; 35: 253337.

www.thelancet.com Vol 376 November 27, 2010

Comment

8
9

10

Van Rompaey B, Elseviers MM, Schuurmans MJ, Shortridge-Baggett LM,


Truijen S, Bossaert L. Risk factors for delirium in intensive care patients:
a prospective cohort study. Crit Care 2009; 13: R77.
Plum F, Posner JB. The diagnosis of stupor and coma. Contemp Neurol Ser
1972; 10: 1286.
Girard TD, Kress JP, Fuchs BD, et al. Ecacy and safety of a paired sedation
and ventilator weaning protocol for mechanically ventilated patients in
intensive care (Awakening and Breathing Controlled trial): a randomised
controlled trial. Lancet 2008; 371: 12634.
Pandharipande P, Shintani A, Peterson J, et al. Lorazepam is an independent
risk factor for transitioning to delirium in intensive care unit patients.
Anesthesiology 2006; 104: 2126.

11

12

13

14

Girard TD, Jackson JC, Pandharipande PP, et al. Delirium as a predictor of


long-term cognitive impairment in survivors of critical illness. Crit Care Med
2010; 38: 151320.
Skrobik Y. Praying for healthy minds and healthy bodies in ICU survivors.
Intensive Care Med 2010; published online July 27.
DOI: 10.1007/s00134-010-1975-5.
Moller JT, Cluitmans P, Rasmussen LS, et al, for the ISPOCD investigators.
Long-term postoperative cognitive dysfunction in the elderly ISPOCD1
study. Lancet 1998; 351: 85761.
Dantzer R, OConnor JC, Freund GG, Johnson RW, Kelley KW. From
inammation to sickness and depression: when the immune system
subjugates the brain. Nat Rev Neurosci 2008; 9: 4656.

Mobile phones to improve HIV treatment adherence


In sub-Saharan Africa, access to HIV treatment has
expanded at an unprecedented rate. Nearly 3 million
HIV-infected adults and children in the region have now
started antiretroviral therapy, an increase of 30 times
since 2003.1,2 Despite this undeniable success, there
are growing concerns about long-term outcome in
such patients. Ensuring strict adherence to suppressive
antiretroviral regimens remains a formidable challenge,3
particularly in those who are on therapy for a long time.
Reports show high levels of patients attrition from
antiretroviral programmes, as high as 35% at 3 years.4,5
In The Lancet, Richard Lester and colleagues6 report
the WelTel Kenya1 study, a randomised trial of mobile
(cell) phone text-messaging to improve patients
adherence to antiretroviral therapy. Patients were
randomly assigned to either a short message service
(SMS) interventiondesigned by the investigators and
with use of the extensive mobile-phone network in
Kenyaor to standard care. Patients in the SMS group
received, and were expected to respond to, weekly text
messages that asked about their general wellbeing. If a
patient reported a problem or did not reply within 48 h,
a follow-up phone call was made by a health provider.
Primary outcomes were self-reported adherence to
antiretrovirals (dened as having taken >95% of the
prescribed drugs in the past 30 days) at 6 and 12 months
and HIV-1 viral RNA load suppression (<400 copies
per mL) at 12 months.
In the intention-to-treat analysis, the SMS intervention was associated with higher rates of adherence
(62%) than was the control group (50%; relative
risk for non-adherence 081, 95% CI 069094;
p=0006) and with better rates of virological suppression (57% vs 48%; relative risk for virological
failure 085, 072099; p=004). The primary analysis
www.thelancet.com Vol 376 November 27, 2010

rightly included patients retention and survival as part


of its treatment outcome, because both are potential
collateral benets of the intervention. In a secondary
analysis that censored missing data, the eect of the
SMS intervention on adherence was no longer observed
(91% vs 91%; 100, 094107; p=094), but its eect on
virological suppression persisted (75% vs 66%; 088,
077100; p= 0047).
WelTel was sensibly designed and well executed,
and provides some of the rst randomised data for
mobile-phone-based health interventions. Despite
the huge amount of investment and activity in mobile
health-technology,7 at present there are few outcomes
data available. As policy makers consider bringing the SMS
intervention to scalewhich we think they shouldsome
questions remain. First, why did the intervention work?
The SMS queries were too infrequent to actually remind
patients to take their drugs pill by pill. Possibly the SMS
intervention worked by improving communication and
rapport between health providers and patients (patients
reported during the pilot phase that it feels like someone cares8). A clearer understanding of the mechanism
behind the interventions eectivenessperhaps identied by qualitative researchcould provide insight
into how the intervention might be optimised and
replicated elsewhere.
A second question is: can these ndings be extended
to other settings? The availability of mobile-phone
infrastructure will not be a major hindrance to
widespread use of this technology. In many African
countries, service coverage is nearly ubiquitous, and
almost everyone has a mobile phone or access to a
shared one. In WelTel, for example, only 39 of the
581 (7%) candidates were excluded because of the
absence of such access. Where the availability of mobile

Published Online
November 9, 2010
DOI:10.1016/S01406736(10)62046-6
See Articles page 1838

1807

Comment

8
9

10

Van Rompaey B, Elseviers MM, Schuurmans MJ, Shortridge-Baggett LM,


Truijen S, Bossaert L. Risk factors for delirium in intensive care patients:
a prospective cohort study. Crit Care 2009; 13: R77.
Plum F, Posner JB. The diagnosis of stupor and coma. Contemp Neurol Ser
1972; 10: 1286.
Girard TD, Kress JP, Fuchs BD, et al. Ecacy and safety of a paired sedation
and ventilator weaning protocol for mechanically ventilated patients in
intensive care (Awakening and Breathing Controlled trial): a randomised
controlled trial. Lancet 2008; 371: 12634.
Pandharipande P, Shintani A, Peterson J, et al. Lorazepam is an independent
risk factor for transitioning to delirium in intensive care unit patients.
Anesthesiology 2006; 104: 2126.

11

12

13

14

Girard TD, Jackson JC, Pandharipande PP, et al. Delirium as a predictor of


long-term cognitive impairment in survivors of critical illness. Crit Care Med
2010; 38: 151320.
Skrobik Y. Praying for healthy minds and healthy bodies in ICU survivors.
Intensive Care Med 2010; published online July 27.
DOI: 10.1007/s00134-010-1975-5.
Moller JT, Cluitmans P, Rasmussen LS, et al, for the ISPOCD investigators.
Long-term postoperative cognitive dysfunction in the elderly ISPOCD1
study. Lancet 1998; 351: 85761.
Dantzer R, OConnor JC, Freund GG, Johnson RW, Kelley KW. From
inammation to sickness and depression: when the immune system
subjugates the brain. Nat Rev Neurosci 2008; 9: 4656.

Mobile phones to improve HIV treatment adherence


In sub-Saharan Africa, access to HIV treatment has
expanded at an unprecedented rate. Nearly 3 million
HIV-infected adults and children in the region have now
started antiretroviral therapy, an increase of 30 times
since 2003.1,2 Despite this undeniable success, there
are growing concerns about long-term outcome in
such patients. Ensuring strict adherence to suppressive
antiretroviral regimens remains a formidable challenge,3
particularly in those who are on therapy for a long time.
Reports show high levels of patients attrition from
antiretroviral programmes, as high as 35% at 3 years.4,5
In The Lancet, Richard Lester and colleagues6 report
the WelTel Kenya1 study, a randomised trial of mobile
(cell) phone text-messaging to improve patients
adherence to antiretroviral therapy. Patients were
randomly assigned to either a short message service
(SMS) interventiondesigned by the investigators and
with use of the extensive mobile-phone network in
Kenyaor to standard care. Patients in the SMS group
received, and were expected to respond to, weekly text
messages that asked about their general wellbeing. If a
patient reported a problem or did not reply within 48 h,
a follow-up phone call was made by a health provider.
Primary outcomes were self-reported adherence to
antiretrovirals (dened as having taken >95% of the
prescribed drugs in the past 30 days) at 6 and 12 months
and HIV-1 viral RNA load suppression (<400 copies
per mL) at 12 months.
In the intention-to-treat analysis, the SMS intervention was associated with higher rates of adherence
(62%) than was the control group (50%; relative
risk for non-adherence 081, 95% CI 069094;
p=0006) and with better rates of virological suppression (57% vs 48%; relative risk for virological
failure 085, 072099; p=004). The primary analysis
www.thelancet.com Vol 376 November 27, 2010

rightly included patients retention and survival as part


of its treatment outcome, because both are potential
collateral benets of the intervention. In a secondary
analysis that censored missing data, the eect of the
SMS intervention on adherence was no longer observed
(91% vs 91%; 100, 094107; p=094), but its eect on
virological suppression persisted (75% vs 66%; 088,
077100; p= 0047).
WelTel was sensibly designed and well executed,
and provides some of the rst randomised data for
mobile-phone-based health interventions. Despite
the huge amount of investment and activity in mobile
health-technology,7 at present there are few outcomes
data available. As policy makers consider bringing the SMS
intervention to scalewhich we think they shouldsome
questions remain. First, why did the intervention work?
The SMS queries were too infrequent to actually remind
patients to take their drugs pill by pill. Possibly the SMS
intervention worked by improving communication and
rapport between health providers and patients (patients
reported during the pilot phase that it feels like someone cares8). A clearer understanding of the mechanism
behind the interventions eectivenessperhaps identied by qualitative researchcould provide insight
into how the intervention might be optimised and
replicated elsewhere.
A second question is: can these ndings be extended
to other settings? The availability of mobile-phone
infrastructure will not be a major hindrance to
widespread use of this technology. In many African
countries, service coverage is nearly ubiquitous, and
almost everyone has a mobile phone or access to a
shared one. In WelTel, for example, only 39 of the
581 (7%) candidates were excluded because of the
absence of such access. Where the availability of mobile

Published Online
November 9, 2010
DOI:10.1016/S01406736(10)62046-6
See Articles page 1838

1807

Comment

a programmes success. However, technology-based


approaches represent only one of many eective means
that should be considered by policy makers and health
providers to improve adherence to antiretrovirals. A
comprehensive multipronged approach tailored to the
specic needs of individual local settings must be used
if maximum gains in patients health are to be realised.
*Benjamin H Chi, Jerey S A Stringer
Centre for Infectious Disease Research in Zambia, Lusaka, Zambia
(BHC, JSAS); and University of Alabama School of Medicine,
Birmingham, AL, USA (BHC, JSAS)
bchi@cidrz.org
We declare that we have no conicts of interest.
Corbis

phones is not so complete, programmes could issue free


phones or subsidise the cost of using the phones for the
SMS intervention.
Finally, what are the cost implications of this intervention? By our calculations, the yearly mobile-phone
costs for the national cohort in Kenya on antiretroviral
therapy (almost 400 000 in 20099) would be about
US$26 million, roughly 1% of the support provided by
the US Presidents Emergency Plan for AIDS Relief to
Kenyas national antiretroviral programme.10 At under
$8 per patient per year, this intervention might prove
cost eective, particularly when one considers the cost
and complexity of second-line therapy.11 However, this
aspect still requires formal analyses.
In sub-Saharan Africa, the science of implementationparticularly focusing on patients adherence
and retentionhas understandably lagged behind the
rapid pace of programme expansion. In this regard,
WelTel is an important step forward, one that shows
the promise of technology to assist in settings where
high disease burden and resource constraints threaten

4
5

8
9

10

11

WHO, UNAIDS. Progress on global access to HIV antiretroviral therapy:


a report on 3 by 5 and beyond. March, 2006. http://www.who.int/hiv/
fullreport_en_highres.pdf (accessed Oct 30, 2010).
WHO, UNAIDS, UNICEF. Towards universal access: scaling up priority HIV/
AIDS interventions in the health sector. September 2009 progress report.
Sept 30, 2009. http://www.who.int/hiv/pub/2009progressreport/en
(accessed Oct 30, 2010).
Nachega JB, Mills EJ, Schechter M. Antiretroviral therapy adherence and
retention in care in middle-income and low-income countries: current status
of knowledge and research priorities. Curr Opin HIV AIDS 2010; 5: 7077.
Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy programs
in sub-Saharan Africa: a systematic review. PLoS Med 2007; 4: e298.
Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up
to three years on treatment in sub-Saharan Africa, 20072009: systematic
review. Trop Med Int Health 2010; 15 (suppl 1): 115.
Lester RT, Ritvo P, Mills EJ, et al. Eects of a mobile phone short message
service on antiretroviral treatment adherence in Kenya (WelTel Kenya1):
a randomised trial. Lancet 2010; published online Nov 9. DOI:10.1016/
S0140-6736(10)61997-6.
UN Foundation, Vodafone Foundation. mHealth for development: the
opportunity for mobile technology for healthcare in the developing world.
2009. http://www.globalproblems-globalsolutions-les.org/unf_website/
assets/publications/technology/mhealth/mHealth_for_Development_full.
pdf (accessed Oct 31, 2010).
Lester R, Karanja S. Mobile phones: exceptional tools for HIV/AIDS, health,
and crisis management. Lancet Infect Dis 2008; 8: 73839.
Kenya National AIDS Control Council, Oce of the President. UNGASS
2010: United Nations General Assembly Special Session on HIV and AIDS.
Country report Kenya. http://data.unaids.org/pub/Report/2010/
kenya_2010_country_progress_report_en.pdf (accessed Oct 30, 2010).
United States Presidents Emergency Plan for AIDS Relief. KenyaFY 2009
approved funding by program area, agency and funding source. http://
www.pepfar.gov/about/122582.htm (accessed Oct 30, 2010).
Boyd M, Emery S, Cooper DA. Antiretroviral roll-out: the problem
of second-line therapy. Lancet 2009; 374: 18586.

Paediatric mortality related to H1N1 infection in England


Published Online
October 27, 2010
DOI:10.1016/S01406736(10)61385-2
See Articles page 1846

1808

In The Lancet Nabihah Sachedina and Liam Donaldson1


analyse paediatric mortality related to pandemic inuenza
A H1N1 in England. The investigators used sensitive casending procedures and follow-up ascertainment of H1N1
status to determine population mortality rates for children
of dierent ages and at various pre-existing states of

health. Case-fatality rates in children with H1N1 infection


are dicult to derive and are probably underestimated.2
Furthermore, because some H1N1-related deaths will
have inevitably been missed, it is almost certain that
the 70 deaths reported in this study are a conservative
estimate. Nevertheless, the mortality rates are similar to
www.thelancet.com Vol 376 November 27, 2010

Comment

a programmes success. However, technology-based


approaches represent only one of many eective means
that should be considered by policy makers and health
providers to improve adherence to antiretrovirals. A
comprehensive multipronged approach tailored to the
specic needs of individual local settings must be used
if maximum gains in patients health are to be realised.
*Benjamin H Chi, Jerey S A Stringer
Centre for Infectious Disease Research in Zambia, Lusaka, Zambia
(BHC, JSAS); and University of Alabama School of Medicine,
Birmingham, AL, USA (BHC, JSAS)
bchi@cidrz.org
We declare that we have no conicts of interest.
Corbis

phones is not so complete, programmes could issue free


phones or subsidise the cost of using the phones for the
SMS intervention.
Finally, what are the cost implications of this intervention? By our calculations, the yearly mobile-phone
costs for the national cohort in Kenya on antiretroviral
therapy (almost 400 000 in 20099) would be about
US$26 million, roughly 1% of the support provided by
the US Presidents Emergency Plan for AIDS Relief to
Kenyas national antiretroviral programme.10 At under
$8 per patient per year, this intervention might prove
cost eective, particularly when one considers the cost
and complexity of second-line therapy.11 However, this
aspect still requires formal analyses.
In sub-Saharan Africa, the science of implementationparticularly focusing on patients adherence
and retentionhas understandably lagged behind the
rapid pace of programme expansion. In this regard,
WelTel is an important step forward, one that shows
the promise of technology to assist in settings where
high disease burden and resource constraints threaten

4
5

8
9

10

11

WHO, UNAIDS. Progress on global access to HIV antiretroviral therapy:


a report on 3 by 5 and beyond. March, 2006. http://www.who.int/hiv/
fullreport_en_highres.pdf (accessed Oct 30, 2010).
WHO, UNAIDS, UNICEF. Towards universal access: scaling up priority HIV/
AIDS interventions in the health sector. September 2009 progress report.
Sept 30, 2009. http://www.who.int/hiv/pub/2009progressreport/en
(accessed Oct 30, 2010).
Nachega JB, Mills EJ, Schechter M. Antiretroviral therapy adherence and
retention in care in middle-income and low-income countries: current status
of knowledge and research priorities. Curr Opin HIV AIDS 2010; 5: 7077.
Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy programs
in sub-Saharan Africa: a systematic review. PLoS Med 2007; 4: e298.
Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up
to three years on treatment in sub-Saharan Africa, 20072009: systematic
review. Trop Med Int Health 2010; 15 (suppl 1): 115.
Lester RT, Ritvo P, Mills EJ, et al. Eects of a mobile phone short message
service on antiretroviral treatment adherence in Kenya (WelTel Kenya1):
a randomised trial. Lancet 2010; published online Nov 9. DOI:10.1016/
S0140-6736(10)61997-6.
UN Foundation, Vodafone Foundation. mHealth for development: the
opportunity for mobile technology for healthcare in the developing world.
2009. http://www.globalproblems-globalsolutions-les.org/unf_website/
assets/publications/technology/mhealth/mHealth_for_Development_full.
pdf (accessed Oct 31, 2010).
Lester R, Karanja S. Mobile phones: exceptional tools for HIV/AIDS, health,
and crisis management. Lancet Infect Dis 2008; 8: 73839.
Kenya National AIDS Control Council, Oce of the President. UNGASS
2010: United Nations General Assembly Special Session on HIV and AIDS.
Country report Kenya. http://data.unaids.org/pub/Report/2010/
kenya_2010_country_progress_report_en.pdf (accessed Oct 30, 2010).
United States Presidents Emergency Plan for AIDS Relief. KenyaFY 2009
approved funding by program area, agency and funding source. http://
www.pepfar.gov/about/122582.htm (accessed Oct 30, 2010).
Boyd M, Emery S, Cooper DA. Antiretroviral roll-out: the problem
of second-line therapy. Lancet 2009; 374: 18586.

Paediatric mortality related to H1N1 infection in England


Published Online
October 27, 2010
DOI:10.1016/S01406736(10)61385-2
See Articles page 1846

1808

In The Lancet Nabihah Sachedina and Liam Donaldson1


analyse paediatric mortality related to pandemic inuenza
A H1N1 in England. The investigators used sensitive casending procedures and follow-up ascertainment of H1N1
status to determine population mortality rates for children
of dierent ages and at various pre-existing states of

health. Case-fatality rates in children with H1N1 infection


are dicult to derive and are probably underestimated.2
Furthermore, because some H1N1-related deaths will
have inevitably been missed, it is almost certain that
the 70 deaths reported in this study are a conservative
estimate. Nevertheless, the mortality rates are similar to
www.thelancet.com Vol 376 November 27, 2010

Comment

www.thelancet.com Vol 376 November 27, 2010

This sense of invulnerability together with concerns


about potential risks of the new vaccine against the
H1N1 virus contributed to poor vaccine-uptake in
many countries.6 In Canada, by October, 2009, the
public-health vaccination media campaign had resulted
in public ennui. Only after a healthy 13-year-old boy
collapsed and died did public opinion shift in favour of
vaccination.7 Sachedina and Donaldson provide evidence
that the very young and those with chronic illness were
at increased risk of death from H1N1 infection. They also
provide partial evidence that even healthy individuals
can present with a more rapid onset of severe illness
than previously thought. Although this study identied
paediatric groups to be at an increased risk of death, it
cannot truly provide evidence that vaccination is more
or less eective in, or should be prioritised to, any such
group. Only two of the children who died had received
the H1N1 vaccinetoo late to be eective. Furthermore,
many deaths were in groups with no risk factors.
However, H1N1 vaccine-delivery programmes have been
generally reported as cost eective, and would have
been more so had the vaccine been available earlier and
had uptake by the population been greater.8,9
Criticisms of the response to the H1N1 virus are
dicult for us to hear, especially when the risks of
the pandemic were uncertain and when we were
woefully underprepared.10 Criticisms about our ability to
eectively communicate the balance of pandemic risks
and benets of vaccination and treatment are surely
warranted, and more persuasive approaches need to be
developed that balance logic with emotion. With the

The printed journal


includes an image merely
for illustration

Niall Carson/PA Archive/Press Association Images

those in other reports, and the true value of this study is


the comparison of mortality in various groups of children.
The investigators noted a higher mortality rate in
children with H1N1 infection than in those with seasonal
inuenza (6 vs 21, both per million population3). Careful
screening might have resulted in more H1N1-associated
deaths being recorded in this study than in many studies of
seasonal inuenza.3 Mortality was highest in children with
pre-existing respiratory, neurological, and cardiac diseases.
Certain ethnic groups might have had higher population
mortality rates than others, which is potentially important;
however, with small numbers of deaths, risk estimates
related to ethnic group are uncertain. About one-third of
the children who died presented with cardiorespiratory
arrest, and a fth of all children were previously healthy
(ie, no, or mild, pre-existing disorders). In the children
who died before or at the point of hospital admission
(early deaths), there was a signicantly greater proportion
of previously healthy children than in the group who died
after admission (late deaths).
Sachedina and Donaldsons study is one of the rst
population-wide studies to accord with anecdotal
and published reports of rapid and fulminate clinical
deterioration in previously healthy children.4 Although
this nding is not more important than the identication
of underlying medical risk factors, it represents a persistent
message of the H1N1 pandemicthat children and young
adults who were not expected to become severely ill
because of inuenza did so, and in greater numbers than
with seasonal inuenza. Similarly, in the USA, by the
end of the 200910 inuenza season, there were more
than three times the number of inuenza-associated
paediatric hospital admissions and deaths compared
with other inuenza seasons.5 Among patients who
died, it is not surprising that most required extraordinary
oxygenation and cardiac support in an intensive-care unit.
The aetiology of cardiovascular dysfunction is unclear, but
might involve cardiac-pump failure (such as myocarditis),
vasodilatation (bacterial co-infection was common and
nearly all patients received antibiotics), hypovolaemia
in combination with preload-reducing mechanical
ventilation, or other inammatory or non-inammatory
disorders that are not yet fully described.
Health-care professionals and the public have begun
to appreciate inuenza as an illness that aects elderly
people or the very young, but are not fully aware of how
new strains can seriously harm healthy populations.

1809

Comment

luxury of post-pandemic hindsight and with the ndings


of Sachedina and Donaldson, we now know that the
200910 H1N1 infection was associated with severe
illness and death in greater numbers of children and
young adults than previous inuenza seasons with other
inuenza viruses. Any talk of over-reaction to 2009 H1N1
virus might lead to an underappreciation of the very real
risks of inuenza. The 2009 pandemic was not nearly as
severe as feared, but might have been even less so with
increased vaccination availability and uptake.
*Robert A Fowler, Philippe Jouvet
Department of Critical Care and Department of Medicine,
Sunnybrook Hospital, University of Toronto, Toronto, ON, Canada
M4N 3M5 (RAF); and Le Centre Hospitalier Universitaire
Sainte-Justine, Universit de Montral, Montral, QC, Canada (PJ)
rob.fowler@sunnybrook.ca
We declare that we have no conicts of interest.
1

Sachedina N, Donaldson LJ. Paediatric mortality related to pandemic


inuenza A H1N1 infection in England: an observational population-based
study. Lancet 2010; published online Oct 27. DOI:10.1016/S01406736(10)61195-6.

10

Health Protection Agency. Pandemic (H1N1) 2009 in England: an overview


of initial epidemiological ndings and implications for the second wave.
Dec 2, 2009. http://www.hpa.org.uk/web/HPAwebFile/HPAweb_C/
1258560552857 (accessed Aug 20, 2010).
Bhat N, Wright JG, Broder KR, et al, for the Inuenza Special Investigations
Team. Inuenza-associated deaths among children in the United States,
20032004. N Engl J Med 2005; 353: 255967.
Lister P, Reynolds F, Parslow R, et al. Swine-origin inuenza virus H1N1,
seasonal inuenza virus, and critical illness in children. Lancet 2009;
374: 60507.
Centers for Disease Control and Prevention. 20092010 inuenza season
week 20 ending May 22, 2010. 2010. http://www.cdc.gov/u/weekly
(accessed Aug 20, 2010).
Centers for Disease Control and Prevention. Morbidity and mortality
weekly report (MMWR) interim results: state-specic inuenza A (H1N1)
2009 monovalent vaccination coverageUnited States, October
2009January 2010. April 2, 2010. http://www.cdc.gov/mmwr/preview/
mmwrhtml/mm5912a2.htm (accessed Aug 20, 2010).
Canadian Broadcasting Corporation. A 13-year-old Toronto boy, described
as healthy as can be, has died of the H1N1 virus. Oct 28, 2009. http://
www.cbc.ca/video/player.html?category=News&zone=health&site=cbc.
health.ca&clipid=1310616402 (accessed Aug 20, 2010).
Khazeni N, Hutton DW, Garber AM, Hupert N, Owens DK. Eectiveness and
cost-eectiveness of vaccination against pandemic inuenza (H1N1) 2009.
Ann Intern Med 2009; 151: 82939.
Sander B, Bauch C, Fisman DN, et al. Is a mass immunization program for
pandemic (H1N1) 2009 good value for money? Early evidence from the
Canadian experience. PLoS Curr 2009; 17: RRN1137.
Fowler RA, Lapinsky SE, Hallett D, et al, for the Toronto SARS Critical Care
Group. Critically ill patients with severe acute respiratory syndrome.
JAMA 2003; 290: 36773.

What do children die from in India today?


Published Online
November 12, 2010
DOI:10.1016/S01406736(10)62054-5
See Articles page 1853

1810

In The Lancet, The Million Death Study Collaborators1


make an important contribution to a topic of
great public health signicance: the causes of child
mortality in India and their distribution across sexes
and geographical regions. As the deadline of 2015
for achieving the Millennium Development Goals
(MDGs) looms, the urgency for scaling up interventions
becomes more pressingespecially in India, where
293% of global neonatal deaths and 161% of global
child deaths occur.2 This new study shows the rst
estimates of the leading causes of child mortality from
direct measurement of 23 152 child deaths captured
across the country. The composition of the leading
causes of death was not surprising, and was consistent
with previous estimates.3 The crucial question for
policy makers, however, is: which population groups
are aected the most?
In answer to this question, the study1 quantiesfor
the rst timeregional and sexual patterns in mortality
from the major causes of child death in India. Some
of the ndings are astounding: the authors report a
ve-fold dierence in postneonatal mortality from
pneumonia between girls in central India and boys in

south India, and a four-fold dierence in the number


of deaths from diarrhoeal disease between girls in
central India and boys in west India. Identication
of subnational patterns is a necessary rst step
towards developing focused intervention policies
that can accelerate the reduction of child mortality.
Quantication is useful, but we need to know why
these dierences exist and how they can be alleviated.
More importantly, we need to know about these
dierences now. A crucial limitation of the study is that
it refers to deaths that occurred on average 8 years ago.
Complacency with, and potential misinformation from,
evidence that is not current is pervasive in global health.
Pleas for high-quality, widely available, and recent
information have been made repeatedly.4,5 Timeliness
is especially troublesome for the Million Death Study
Collaborators study, because, between 2002 and 2010,
neonatal mortality in India decreased by 17% and
mortality in children younger than 5 years decreased by
21%.2 Therefore, although the report provides interesting
ndings, it cannot tell which of the major causes of child
mortality have decreased most rapidly and which need
most attention. This result begs the question of how
www.thelancet.com Vol 376 November 27, 2010

Comment

luxury of post-pandemic hindsight and with the ndings


of Sachedina and Donaldson, we now know that the
200910 H1N1 infection was associated with severe
illness and death in greater numbers of children and
young adults than previous inuenza seasons with other
inuenza viruses. Any talk of over-reaction to 2009 H1N1
virus might lead to an underappreciation of the very real
risks of inuenza. The 2009 pandemic was not nearly as
severe as feared, but might have been even less so with
increased vaccination availability and uptake.
*Robert A Fowler, Philippe Jouvet
Department of Critical Care and Department of Medicine,
Sunnybrook Hospital, University of Toronto, Toronto, ON, Canada
M4N 3M5 (RAF); and Le Centre Hospitalier Universitaire
Sainte-Justine, Universit de Montral, Montral, QC, Canada (PJ)
rob.fowler@sunnybrook.ca
We declare that we have no conicts of interest.
1

Sachedina N, Donaldson LJ. Paediatric mortality related to pandemic


inuenza A H1N1 infection in England: an observational population-based
study. Lancet 2010; published online Oct 27. DOI:10.1016/S01406736(10)61195-6.

10

Health Protection Agency. Pandemic (H1N1) 2009 in England: an overview


of initial epidemiological ndings and implications for the second wave.
Dec 2, 2009. http://www.hpa.org.uk/web/HPAwebFile/HPAweb_C/
1258560552857 (accessed Aug 20, 2010).
Bhat N, Wright JG, Broder KR, et al, for the Inuenza Special Investigations
Team. Inuenza-associated deaths among children in the United States,
20032004. N Engl J Med 2005; 353: 255967.
Lister P, Reynolds F, Parslow R, et al. Swine-origin inuenza virus H1N1,
seasonal inuenza virus, and critical illness in children. Lancet 2009;
374: 60507.
Centers for Disease Control and Prevention. 20092010 inuenza season
week 20 ending May 22, 2010. 2010. http://www.cdc.gov/u/weekly
(accessed Aug 20, 2010).
Centers for Disease Control and Prevention. Morbidity and mortality
weekly report (MMWR) interim results: state-specic inuenza A (H1N1)
2009 monovalent vaccination coverageUnited States, October
2009January 2010. April 2, 2010. http://www.cdc.gov/mmwr/preview/
mmwrhtml/mm5912a2.htm (accessed Aug 20, 2010).
Canadian Broadcasting Corporation. A 13-year-old Toronto boy, described
as healthy as can be, has died of the H1N1 virus. Oct 28, 2009. http://
www.cbc.ca/video/player.html?category=News&zone=health&site=cbc.
health.ca&clipid=1310616402 (accessed Aug 20, 2010).
Khazeni N, Hutton DW, Garber AM, Hupert N, Owens DK. Eectiveness and
cost-eectiveness of vaccination against pandemic inuenza (H1N1) 2009.
Ann Intern Med 2009; 151: 82939.
Sander B, Bauch C, Fisman DN, et al. Is a mass immunization program for
pandemic (H1N1) 2009 good value for money? Early evidence from the
Canadian experience. PLoS Curr 2009; 17: RRN1137.
Fowler RA, Lapinsky SE, Hallett D, et al, for the Toronto SARS Critical Care
Group. Critically ill patients with severe acute respiratory syndrome.
JAMA 2003; 290: 36773.

What do children die from in India today?


Published Online
November 12, 2010
DOI:10.1016/S01406736(10)62054-5
See Articles page 1853

1810

In The Lancet, The Million Death Study Collaborators1


make an important contribution to a topic of
great public health signicance: the causes of child
mortality in India and their distribution across sexes
and geographical regions. As the deadline of 2015
for achieving the Millennium Development Goals
(MDGs) looms, the urgency for scaling up interventions
becomes more pressingespecially in India, where
293% of global neonatal deaths and 161% of global
child deaths occur.2 This new study shows the rst
estimates of the leading causes of child mortality from
direct measurement of 23 152 child deaths captured
across the country. The composition of the leading
causes of death was not surprising, and was consistent
with previous estimates.3 The crucial question for
policy makers, however, is: which population groups
are aected the most?
In answer to this question, the study1 quantiesfor
the rst timeregional and sexual patterns in mortality
from the major causes of child death in India. Some
of the ndings are astounding: the authors report a
ve-fold dierence in postneonatal mortality from
pneumonia between girls in central India and boys in

south India, and a four-fold dierence in the number


of deaths from diarrhoeal disease between girls in
central India and boys in west India. Identication
of subnational patterns is a necessary rst step
towards developing focused intervention policies
that can accelerate the reduction of child mortality.
Quantication is useful, but we need to know why
these dierences exist and how they can be alleviated.
More importantly, we need to know about these
dierences now. A crucial limitation of the study is that
it refers to deaths that occurred on average 8 years ago.
Complacency with, and potential misinformation from,
evidence that is not current is pervasive in global health.
Pleas for high-quality, widely available, and recent
information have been made repeatedly.4,5 Timeliness
is especially troublesome for the Million Death Study
Collaborators study, because, between 2002 and 2010,
neonatal mortality in India decreased by 17% and
mortality in children younger than 5 years decreased by
21%.2 Therefore, although the report provides interesting
ndings, it cannot tell which of the major causes of child
mortality have decreased most rapidly and which need
most attention. This result begs the question of how
www.thelancet.com Vol 376 November 27, 2010

Comment

www.thelancet.com Vol 376 November 27, 2010

The printed journal


includes an image merely
for illustration
Sanjit Das/Panos

regional and sexual patterns have changed with increased


economic development and gains in child survival.
Part of the reason for the large lag reported in the
study is attributable to the way in which verbal autopsy
was implemented and analysed. Verbal autopsy,
despite its limitations, has great potential to identify
the main causes of death in a timely manner. For this
potential to materialise, two things need to happen:
rst, questionnaires need to adhere closely to WHOs
recommended guidelines with minimum reliance
on the open narrative; and second, estimates of
cause-specic mortality fractions should be based on
quantitative methods that rely on automated analysis
of responses to symptom questions.68 These two
principles would ensure that verbal autopsies can be
widely implemented and quickly analysed to produce
estimates of causes of death within a few months,
rather than years, of data collection. Moreover, such
methods have the advantage of being less costly and
not subject to doctors biases.7
So where do we go from here? The Million Death
Study Collaborators study shows that, without a
complete vital registration system, it is still possible
to directly capture the major causes of death with
apparent reliability. The Registrar General of India has
invested greatly in the health information system
with a combination of routine surveillance (through
the ongoing Sample Registration System) and also
regular representative surveys at a national, state, and
district scale. Alternative models exist, based on the
broader sample vital registration with verbal autopsy
approach, and can generate useful data for mortality
in populations without full vital registration.9 For
example, some countries are adding verbal autopsy
modules to large representative household surveys
whereas others, such as Mozambique, are doing
verbal autopsy on a sample of deaths identied
through censuses. These various approaches provide
a more comprehensive picture of population health
than did previous research that was focused on
local communities. As verbal autopsy methods are
developed further, the costs of implementing and
analysing data from such systems will continue
to decrease and the turnaround time from data
collection to analysis and publication of ndings
can be shortened. What is urgently needed in most
countries without adequate vital registration systems

is adoption of one of these low-cost approaches that


can yield nationally representative cause-of-death
estimates. The new study shows that this requirement
is feasible; and, if done in a timely way, would be even
more useful for guiding national health policy.
*Emmanuela Gakidou, Alan D Lopez
Institute for Health Metrics and Evaluation, University of
Washington, Seattle, WA 98121, USA (EG); and School of
Population Health, University of Queensland, Brisbane, QLD 4006,
Australia (ADL)
gakidou@u.washington.edu
We declare that we have no conicts of interest.
1

5
6

7
8

The Million Death Study Collaborators. Causes of neonatal and child


mortality in India: a nationally representative mortality survey.
Lancet 2010; published online Nov 12. DOI:10.1016/S01406736(10)61461-4.
Rajaratnam JK, Marcus JR, Flaxman AD, et al. Neonatal, postneonatal,
childhood, and under-5 mortality for 187 countries, 19702010:
a systematic analysis of progress towards Millennium Development Goal 4.
Lancet 2010; 375: 19882008.
Murray CJL, Lopez AD, eds. The global burden of disease: a comprehensive
assessment of mortality and disability from diseases, injuries, and risk
factors in 1990 and projected to 2020. Boston, MA: Harvard School
of Public Health, 1996.
Chan M, Kazatchkine M, Lob-Levyt J, et al. Meeting the demand for results
and accountability: a call for action on health data from eight global health
agencies. PLoS Med 2010; 7: e1000223.
Birnbaum J, Cowling K, Foreman K, et al. Sceptical optimism: a new take
on global health data. Lancet 2009; 374: 173031.
WHO. Verbal autopsy standards: ascertaining and attributing cause of
death. 2007. http://www.who.int/whosis/mort/verbalautopsystandards/
en/index.html (accessed Nov 2, 2010).
Murray CJL, Lopez AD, Feehan D, Peter S, Yang G. Validation of symptom
pattern method for analysing verbal autopsy data. PLoS Med 2007; 4: e327.
Byass P, DAmbruoso L, Oudraogo M, Qomariyah SN. Assessing the
repeatability of verbal autopsy for determining cause of death: two case
studies among women of reproductive age in Burkina Faso and Indonesia.
Popul Health Metr 2009; 7: 6.
Setel PW, Sankoh O, Rao C, et al. Sample registration of vital events with
verbal autopsy: a renewed commitment to measuring and monitoring vital
statistics. Bull World Health Organ 2005; 83: 61117.

1811

Comment

Oine: Urgency and concern about home births

Getty Images

Science Photo Library

Readers may recall the tidal wave of criticism we received


for giving attention to a 2010 paper from the American
Journal of Obstetrics and Gynecology. Joseph Wax and his
colleagues from Maine claimed to have completed a
systematic review on maternal and newborn safety of
planned home versus planned hospital birth. In their
paper, they concluded that less medical intervention
during planned home birth is associated with a tripling
of the neonatal mortality rate. Carl Michal from the
Department of Physics and Astronomy at the University
of British Columbia, together with three academic
midwifery colleagues and a public health scientist
(including some of those who were authors of papers
included in the meta-analysis), have written to the
editors of AJOG asking that the paper be retracted. They
argue that the article is deeply awed. AJOG allegedly
issued a highly misleading press release to advertise
the oending paper. Michal et al now request a press
release from the Journal announcing the retraction.
Their 9-page letter points to mistakes in denitions,
numerical errors, selective and mistaken inclusion
and exclusion of studies, conation of association and
causation, and additional statistical problems. The result
is a paper that is nonsensical. They ask that the editors
of AJOG act with urgency and concern to address the
insupportable conclusion of the article. Their letter is
dated Sep 15, 2010. So far, the editors of the journal have
yet to respond in public.
*

Getty Images

Harold Varmus may have made an uncharacteristically


ill-judged start at the National Cancer Institute. A plan to
take most tobacco control research out of NCI and place
it into a new addictions institute has understandably
aroused anger in the tobacco control community. In a
letter to Francis Collins (the Director of NIH) on Nov 17,
Stanton Glantz writes that such a move will create the
appearance and reality of NCI walking away from tobacco.
Worse, it will send a strong message that NCI does not
think that tobacco control research is a priority for cancer
control. Glantz is one of the worlds foremost scientists
working on tobacco. His contributions to understanding
how population-based interventions can have large
and rapid health benets have accelerated political
commitments to tobacco control measures. Given that
1812

tobacco is the leading cause of preventable cancer deaths,


the Varmus plan therefore makes little sense. Indeed, as
Glantz argues, if the Varmus proposal goes through it
will create heightened opportunities for the tobacco
industry to shut down the kind of research and training
that has made such a strong contribution to reducing
smoking prevalence and consumption. What is the likely
outcome of this dispute? Some observers believe that
Varmus wields too much power to be stopped. As one
commentator has put it, if NIH tried to drop the plan,
Varmus will probably be mad and try to retaliate. The
nal decision, however, rests with Francis Collins. In the
interests of cancer control, population-based tobacco
research needs to stay at NCI.
*
Back in the UK, an arctic wind is blowing through the
corridors of public health. The new LiberalConservative
Government has brought in a radical new ideology,
abandoning the previous administrations commitment
to addressing upstream causes of ill health. Initiatives
that had been fought for and agreed have now been
mothballed (for example, on obesity). Should the public
health community cooperate with the new government
or oppose it? When asked this question, most medical
leaders reply that in order to keep their seat at the
political high table, they cannot publicly disagree with
government. Their task, they say, is to cooperate to make
bad policies less bad. This is the politics of betrayal. There
can be no deal with ministers at the moment. To retain
their credibility, public health scientists must adhere
to the best available science. And that science points to
the importance of social and economic determinants as
critical causes of much poor health in Britain today. As
one leading public health researcher and advocatewho
still sits at the ministerial tableput it to me last week,
the situation could explode at any moment. The best
that public health can do is to organise more eectively
and hold the government accountable for its plans and
actions. A new Public Health White Paper will be published
next month. All of us who have an interest in Britains
health should be marshalling our forces, ready for attack.
Richard Horton
richard.horton@lancet.com

www.thelancet.com Vol 376 November 27, 2010

World Report

Haiti still gripped by cholera as election looms


With the death toll from cholera in Haiti steadily rising, and many blaming the UN for introducing
the disease, tensions are running high ahead of next weeks election. Barbara Fraser reports.
As the ocial cholera death toll in Haiti
rose past 1200, public health experts
repeated the lessons learned from
dozens of outbreaks elsewhere in the
world: cholera can be controlled with
clean water, sanitation, and proper
hygiene. The deaths are preventable.
But with the cholera outbreak in
Haiti coming just 10 months after
a 70-magnitude earthquake that
killed as many as 300 000 people and
left nearly 15 million homeless, it is
perhaps not surprising that Haitis
health-care system is struggling to
keep up as cholera cases spread around
the country.
Containment eorts face the obvious
problems of widespread poverty, poor
infrastructure, a small health-care
budget, lack of safe water and sanitation
services, along with the hazards posed
by crowded tent camps that house
anywhere from dozens of families to
50 000 peopleconditions that chief
medical director for Partners in Health
Joia Mukherjee called a perfect storm
for a massive epidemic of cholera.
There is also a less-visible factora
brain drain that has robbed the
country of most of its medical school
graduates. Of every 100 doctors
who graduate, only 20 stay in the
country, said Jean Hugue Henrys, who
heads the Health Ministrys National
Cholera Crisis Commission. The
rest go abroad to the USA, Europe,
or Africa. There is an impressive
shortage of (health-care) resources
and personnel in this country, with
slightly more than two physicians per
10 000 inhabitants, he said.
Middle-class professionals from other
specialties have also emigrated, making
it even harder for the country to
recover from this years triple whammy
of disastersthe Jan 12 earthquake;
Hurricane Toms, which struck on
Nov 5; and the cholera epidemic, which
www.thelancet.com Vol 376 November 27, 2010

began before the storm but spread with


the ood waters.
After the earthquake, public health
experts warned that the tent camps
that had sprung up in and around Portau-Prince, the area hardest hit, were
at high risk for outbreaks of disease.
However, they were not expecting
cholera. There has been no cholera in
Haiti in at least half a century, and the
island was unaected by the outbreak
that began in Peru in the early 1990s.

Even before the earthquake,


half the people in Port-auPrince had no latrines or other
sanitation systems, and a third
lacked access to tap water.
So when cholera appeared during
the third week of October, not only
was it unexpected, but most of Haitis
health professionals had no experience
in dealing with it. And the rst cases
occurred not in Port-au-Prince, where
health workers were on their guard,
but in a rural area along the Artibonite
River in central Haiti, north of the
capital, which had not been aected
by the quake.
After genetic tests matched the
cholera bacterium with a south
Asian strain, the mayor of Mirabelais,
in the Artibonite valley, blamed
euent from a camp housing UN
peacekeeping troops from Nepal.
UN spokespersons denied any
connection, and some health ocials
said the exact origin of the outbreak
might never be known. Nevertheless,
protests against the UN peacekeeping
forces erupted on Nov 15 in CapHaitien and quickly spread to other
cities, including Port-au-Prince.
Several people were reported killed,
including one apparently shot by a
UN soldier, and several UN troops
and demonstrators were reported

injured. The Pan American Health


Organization (PAHO) said the riots
were interfering with delivery of
medical supplies and treatment.
With presidential elections scheduled for Nov 28, a UN ocial said the
protests were politically motivated,
but they might also have been
prompted by fear and frustration as
cholera spread to most of the regions
provinces, and PAHO predicted
that the country could see between
200 000 and 270 000 cases in the
coming months.
As a reminder that the disease could
spread further, a Haitian migrant to
the Dominican Republic was reported
to be that countrys rst case on
Nov 16, and a case was reported in a
woman who returned to Florida after
visiting Haiti.
Although cholera is easy to treat
with oral rehydration packs, Haiti is a
country that has not seen cholera for
decades, said Robert Quick, a medical
epidemiologist in the Waterborne
Diseases Prevention Branch of the
Centers for Disease Control and
Prevention (CDC) in Atlanta, GA, USA.
You have an immunologically naive
population. Nevertheless, he said,
these deaths are preventable. Both
prevention and treatment, however,
depend on a rapid response, but the
response in Haiti has been hampered
by poor infrastructure and peoples
lack of familiarity with the disease.
Even before the earthquake, half
the people in Port-au-Prince had no
latrines or other sanitation systems,
and a third lacked access to tap water.
Rural residents mainly rely on surface
water drawn directly from the source,
or piped from a river or stream. Even
before the cholera outbreak, diarrhoeal
diseases were common and a major
cause of death in children younger
than 5 years.
1813

World Report

AP

The printed journal


includes an image merely
for illustration

Haitians are mobile, travelling back


and forth between the countryside and
cities. And many people who are infected with cholera are asymptomatic,
making quarantine based on symptoms
impossible, according to Tom Kirsch,
co-director of the Center for Refugee
and Disaster Response at Johns Hopkins
Universitys Bloomberg School of Public
Health in Baltimore, MD, USA.
We think ooding (from the
hurricane) likely accelerated the
spreading
and
expanded
the
geographical range [of the outbreak],
said epidemiologist Silvain Aldighieri,
PAHO incident coordinator for the
cholera crisis. People and patients
were evacuated from risk areas in the
already aected areas and gathered
in crowded shelters, which increased
transmission. The movement of people
from the Artibonite outbreak epicenter
was a key factor of dissemination.
Surveillance has been a challenge,
especially in rural areas, and some
experts say ocial datamore than
1200 deaths and 20 000 cases treated
are probably low. Haitis surveillance
system was weak even before the
earthquake, although PAHO and the
CDC have been working to strengthen
it since January. The early warning
system put in place was strong enough
to have detected the emergence of
cholera fairly quickly, Aldighieri said.
A weakness in the system is that cases
are being seen by numerous NGOs and
other international partners who are
1814

providing health services in dierent


parts of the country, and their reporting
is not necessarily part [of the Health
Ministry system].
As the outbreak spread northward,
then appeared in Port-au-Prince, other
cities and rural areas in the south,
the UN called for US$164 million in
medical supplies, although ocials
said international donors had provided
only 10% of that amount as of Nov
20. Health ocials and aid agencies
scrambled to train community
health workers, set up temporary
treatment centres and distribute oral
rehydration supplies, and launched
publicity campaigns emphasising the
importance of hand washing and safe
water supplies.
Only about a third of Haitis health
facilities are completely governmentrun, according to Henrys. Many of
the others are partly or completely
operated by non-governmental
organisations (NGOs), especially in
rural areas, where more than half the
countrys population lives.
Both the Health Ministry and NGOs
are training community health workers
to teach good sanitation and hygiene
practices, identify cases and provide
rehydration therapy, and the CDC is
providing train-the-trainer courses.
Health ocials reported that 32 cholera
treatment centres and nearly 60 smaller
treatment units had been set up
around the country. Meanwhile, the
CDC is doing epidemiological studies to
understand the spread of the outbreak,
as well as surveys to determine what
steps Haitians are taking to protect
themselves and their families.
Around Port-au-Prince, the aftermath of the earthquake has further
hampered cholera-control measures.
More than 80% of the capitals
hospitals suered signicant damage
in the earthquake and had not yet
been able to rebuild, Kirsch said.
The tent camps that sprouted around
the capital after the earthquake are a
paradox. While the Petionville camp
on Haitis only golf course houses
50 000 people at close quarters, it is

run by aid agencies that provide latrines,


chlorinated water, and medical care.
Its residents could be at less risk than
slum dwellers whose homes were not
damaged, but who have no bathrooms
or running water. Displaced people are
usually at increased risk, due to poor
living conditions. In Haiti, it depends
on the camps. Some have better
water supplies and sanitation than
others, said Hugo Prado, acting PAHO
representative in Haiti. It is true that
urban slums in the capital in some cases
will be in worse condition than some
camps, and therefore at greater risk of
cholera spreading. But once cholera has
set foot inside the camps, if brought in,
the dense conditions will most likely
expedite the infection cycle.
As prevention and treatment
measures take hold, the outbreaks
initially high mortality rate is expected
to drop, but the cholera crisis has
highlighted urgent needs in Haitis
health-care system. If you want public
health, you have to have a public
health system, Mukherjee said. That
requires investment in the government
system, not simply the NGO-ication
of aid, she added. Henrys said that
requires more than money. Although
we continue to train personnel, 80%
are going to emigrate. This is not just
a problem of poverty. The question
is how to create the conditions in the
country for young people to develop
personally and professionally, he said.
Urgent as they are, the problems
require a long-term approach. The
cholera bacterium has a foothold in
the country now, said PAHO deputy
director Jon K Andrus. There must be
long-term commitment on behalf of
the international community, not just
for the control of the outbreak, but for
reversing the extremely poor water
and sanitary conditions that existed
long before the outbreak. Unless safe
water and waste management can be
provided, cholera will continue despite
everyones incredibly hard work to
prevent and treat new cases.

Barbara Fraser
www.thelancet.com Vol 376 November 27, 2010

World Report

Crime and unjust punishment in Russia


A year after the controversial death in a Moscow detention centre of Sergei Magnitskya
37-year-old lawyer who was denied vital medical treatmentRussia is promising an overhaul of
its antiquated prison system. But will the reforms bring real change to health-care provision?
Tom Partt reports.
It was 1830 h on November 16,
2009, when Sergei Magnitsky was
transferred to the Matrosskaya Tishina
detention centre in Moscow. The
37-year-old lawyer had been healthy
when he was arrested a year earlier
on fraud charges that colleagues said
were trumped-up in revenge for his
work for Hermitage, an international
investment fund that passed evidence
about corrupt ocials to Russian
media. Yet within 4 hours of arriving
at Matrosskaya Tishina (Sailors Rest),
Magnitsky was dead.
In the past year the Magnitsky
Aair, as it is known in Russia, has
become emblematic of the countrys
woeful human rights record and its
sometimes wilfulneglect of the sick
in prison. 6 weeks after Magnitsky
was found lifeless in his cell, the
public oversight commission (ONK)
for Moscows pretrial detention
centres published a scathing report
describing the events that led up to
his death.
The watchdog found he had
been diagnosed with gall stones,
pancreatitis,
and
calculous
cholecystitis while in custody, but
prison ocials had consistently
ignored his condition or refused him
adequate treatment. The head of one
facility where the lawyer was held told
the commission: I did not consider
Magnitsky sick. Detainees often try to
pass themselves o as sick in order to
get better conditions.
In the last 3 days of his life, the
commission determined, Magnitsky
was vomiting and suering from acute
abdominal pain. He was inspected
by doctors but they decided to leave
him alone in his cell without medical
supervision. When a guard nally
noticed him lying motionless on the
www.thelancet.com Vol 376 November 27, 2010

oor, orderlies rushed to revive him


but it was already too late.
Unlike the relatives of many
prisoners, Magnitskys colleagues
at Hermitage had the resources
to make a fuss. This is not about
business, this is about the complete
breakdown of the justice system, said
chief executive William Browder in a
television interview shortly after the
death. If you can have an innocent
man on trumped-up charges taken
hostage and then slowly killed through
mistreatment in prison. Its medieval.

...The head of each institution is


a god, a tsar, while the convicts
are treated as inhuman. Until
that culture of sadism ends we
cant talk of improvements.
In the past year, Browder and his
supporters have produced a series of
online lms which describe the alleged
corruption the lawyer exposed.
They claim Magnitskys health was
deliberately neglected as a means
of pressuring him to testify against
Hermitage employees. As a result of
the campaign, US secretary of state
Hillary Clinton and British foreign
secretary William Hague, have both
urged a thorough investigation of the
death. Meanwhile, Russias president,
Dmitry Medvedev responded by ring
20 top managers from the prison
system, ocially known as the Federal
Service for Execution of Punishments
(FSIN). Whether the Magnitsky case
will prompt real reform of health care
in prisons, however, remains a topic of
intense debate.
Russias sprawling prison network
of prisons and corrective colonies
has 131 hospitals, 59 tuberculosis
clinics, nine drug addiction clinics, and

medical stations at every institution,


although many are in poor shape.
More than 90% of the countrys
864 000 inmates are suering health
problems according to a report
published by the prosecutor generals
oce in August. The report found that
60% of clinics were using outdated
equipment, and in total the prison
health-care system was receiving only
24% of necessary funding.
There have been some signs of
tentative change. In spring this
year, the lower house of parliament
approved revisions to the criminal
code that will soften sentences for
defendants convicted of economic
crimes. In September, FSIN produced a
strategy document that dened broad
targets such as increasing funding for
health care and bringing treatment in
prisons into line with civilian healthcare facilities. And earlier this month
the head of the notorious Butyrka
remand centre near Moscowwhere
Magnitsky was denied an ultrasound
scanmade headlines when he
announced prisoners would soon
have access to sunbeds, as well as a
new ultrasound machine and other
medical services.
Also this year, Russia pushed forward
with plans to separate rst-time
oenders from recidivists; a move
that is hoped will reduce attacks on
defenceless new prisoners. However,
critics believe many of the changes
are cosmetic. The prison system
retains its totalitarian spirit, says Lev
Ponomaryov, one of Russias most
respected human rights activists.
Torture is widespread. The head of
each institution is a god, a tsar, while
the convicts are treated as inhuman.
Until that culture of sadism ends we
cant talk of improvements.
1815

World Report

Russian prisoners under observation while they take tuberculosis medication

Ponomaryov says talented doctors


have ed the prison system, high
levels of corruption mean federal
funds for medical equipment are
often embezzled, and drugs brought
to patients by relatives are stolen or
held up by guards.
The death of Vera Trifonova on
April 30 was another indication that
cruel practices are ingrained. Trifonova,
aged 52, a real estate agent accused of
fraud who was suering from severe
diabetes, died in Matrosskaya Tishina
after prison ocials had advised a
court she should not be released for
treatment.
Nadezhda Radnayeva, chief specialist
at the Fund for Protection of Prisoners,
a Moscow-based non-governmental
organisation (NGO), says: we receive
so many complaints from prisoners
who say that whatever their illness
they are told, take this Aspirin, and o
you go.
Radnayeva says doctors consistently
under-record the seriousness of
prisoners illnesses, in order to deny
them the right to hospital treatment
and better food. In one ongoing
case in a prison in Mordovia region,
authorities have refused to give
disabled status to a prisoner who is
paralysed from the waist down. The
prisoner, who cannot negotiate stairs,
paid cigarettes to other inmates to
bring him food from the canteen
to his cell. When the cigarettes ran
out he didnt eat for a week, says
Radnayeva.
1816

Ponomaryov believes the system


needs fundamental revision. The
problem is that doctors are also men
in epaulettes (ocers), he says. At
the end of the day, they do what the
head of the prison says, even if that
means a cover up. As an example,
he cites the case of four men who
died in a prison colony in Kopeysk
in Chelyabinsk region in June,
2008, after guards battered a dozen
inmates for refusing to submit to
humiliations like washing their faces
in a toilet bowl. The four perished
after they were left unattended in
separate cells. Prison administrators
are now undergoing trial for trying
to hide the murders by faking an
uprising in the prison.
In the light of such scandals, the
bosses of FSIN seem to be in favour
of passing medical care to civil
control. The service did not respond
to a request for interview from
The Lancet, but its deputy director,
Vasily Bolshakov, told the Novaya
Gazeta newspaper in July he was all
for the transfer. Unfortunately,
the ministry of health and social
development is against it, he added.
After the death of Magnitsky we
immediately suggested they take
medical care from us, and then a
second time this year. But both times
the ministry turned us down saying
it was unnecessary and unsuitable.
The health ministry has since
denied it received these approaches
from FSIN.
As a compromise, FSIN says it is
conducting an experiment in two
regions to make doctors directly
responsible to the services federal
headquarters, rather than to prison
governors.
Mikhail Volik, director of the
Moscow oce of AIDS Foundation
East West (AFEW), a Dutch NGO
that is a member of WHOs Health
in Prisons programme, says
integration with civil health care is
vital. According to AFEW, 64% of
Russias prison population, or about
55 000 people, are HIV positive.

That means more than one in ten of


Russias 540 000 ocially registered
HIV suerers are behind bars.
You have to remember that good
prison health is good public health
because all these inmates will be
released one day, says Volik. So
there needs to be more attention
to HIV in prisons. Despite inmates
accounting for more than 10% of the
countrywide epidemic only 1% of the
federal AIDS budget is dedicated to
care for inmates.
Yet the answer is not only about
improving care and diagnostics on
the inside. Most people with HIV
in jail arrive with it on the rst day,
says Volik. Only improved harm
reduction in wider society and a
change of attitudes will help stem
the problem, he adds. Injecting drug
use (the main cause of HIV because
of shared needles) is still seen as
more of a crime than a disease.
As it is, the ow of HIV suerers
into jail has punishing results. Their
weak immune systems make them
vulnerable to rampant tuberculosis,
one of the chief causes of death in
those with the virus.
A nal battleground in the ght
for prison health is the make-up of
public oversight commissions, or
ONKs. Moscows commission showed
its teeth last year when it issued
its damning report on the death
of Magnitsky. But Zoya Svetova, a
member of the Moscow commission,
told reporters this month that
military and security veterans are
trying to seize control of the ONK.
She said bodies controlled by the law
enforcement agencies planned to
propose and conrm the veterans
as members, with the aim of stiing
the commissions work to expose
violations in prison.
This is an attempt to transform
public control into a mere decoration,
devoid of any real substance,
Ponomaryov, the human rights
defender, told The Lancet.

Tom Partt
www.thelancet.com Vol 376 November 27, 2010

Perspectives
WEBVIDEO Caption: High Society
at Wellcome Collection

Exhibition
Enjoying the high lifedrugs in history and culture

www.thelancet.com Vol 376 November 27, 2010

Kava, a narcotic drink with powers of


intoxication, purication, sedation,
and healing (it is particularly eective
for fungal infections and rheumatism).
In 1982, Queen Elizabeth II drank it
on her state visit. We are not told her
verdict. There are photographs from
the 1930s of the Huichol people of
Mexico making an annual pilgrimage
to collect peyote, a small cactus rich in
mescaline. We are told that it is a rite
that has persisted almost unchanged
since pre-Columbian times.

The message of this exhibition


is the curious uidity in the
identication and classication
of drugs.
European fascination with drugs is
similarly mesmerising. A simple glass
bottle, labelled Forced March, was
marketed as a useful gift for friends
in need of a boost. The bottle was
sold by Burroughs Wellcome in the
early 20th century and contained
a powerful mixture of cocaine and
caeine. If taken once an hour when
undergoing continued mental strain
or physical exertion, it promised to
prolong powers of endurance and
curb appetites. No wonder both Ernest
Shackleton and Robert Scott found
Forced March a useful addition
to their packs on their calamitous
expeditions to the South Pole. Soldiers
during World War I also imbibed these
convenient pills, perhaps helping
them endure the rigours of trench
warfare, but not endearing them
to folks back home who were less
than keen to have crazy soldiers
cavorting in public squares. In 1916,
the government attempted to curb
the use of cocaine by soldiers on leave
from war service, but cocaine was only
banned completely by the Dangerous
Drugs Act of 1920.
The message of this exhibition is the
curious uidity in the identication

and classication of drugs. Drugs are


both glamorous and appalling, and
their status can shift with bewildering
abruptness. It is dicult for us today to
imagine a world in which cocaine was
routinely added to elixirs, cigarettes,
nerve tonics, and drinks. John
Pembertons original 1886 recipe for
Coca Cola insisted on adding cocaine.
Medication sold over the counter by
inuential pharmaceutical company
Parke-Davis contained substantial
quantities of the drug, which was
widely believed to supply the place
of food, make the coward brave,
the silent eloquent, and render the
suerer insensitive to pain. When
Ethel Merman appeared on Broadway
in 1934 singing I Get A Kick Out Of You,
many people in the audience would
have agreed with her claim that
cocaine and alcohol were secondary
thrills compared to love. But they
might also have agreed on the simpler
pleasures of returning home to a
cigarette, tumbler of whiskey, and, if
feeling under the weather, an opiumpacked throat pastille.

High Society
Wellcome Collection, London,
UK, showing until Feb 27, 2011.
http://www.wellcomecollection.
org/whats-on/exhibitions/
high-society.aspx

See Online for webvideo

Joanna Bourke
j.bourke@bbk.ac.uk

Blotter Art, Constance Little/Wellcome Collection

In 1934, Cole Porters Broadway


musical Anything Goes opened with
the song I Get a Kick Out of You. In it,
audiences were told that some get
a kick from cocaine. To enable the
song to be aired on the radio, Porter
agreed that singers might change the
objectionable reference to cocaine:
they could substitute some like a
whi of Guerlain. The line that mere
alcohol doesnt thrill me at all excited
no opposition.
David Nutt, the former UK Government drugs adviser, was widely cited
recently for pointing out in this journal
that alcohol is more harmful than
any other drug. Today, each person
on average drinks more than twice
the volume of alcohol than they did
in 1945. Consumption of cocaine
is also increasing; according to the
EUs drug agency, the UK is at the
top of the European league table
for cocaine use. Internationally, the
drugs market is estimated to be worth
about 200 billion a year. So it would
seem a perfect time for the Wellcome
Collections High Society exhibition,
which seeks to explore the universal
human desire to use mood-altering or
mind-altering substances. It does so
with panache and elegance; informing
rather than moralising, and instilling in
visitors a sense of reverence for peoples
powers of self-deception as well as
ingenuity and creativity.
There are so many fascinating
objects that its dicult to know where
to begin. Even seemingly modest
artifacts tell remarkable stories. On
entering the exhibition, there is a
large wall of intriguing pipes, syringes,
capsules, gourds, vaporisers, cups, betel
boxes, nut cutters, cannabis bongs,
cigarettes, powders, and mushrooms.
Drug taking is global. We are catapulted
from opium dens in Patna (India) to
colourful bazaars in Constantinople. An
intricate engraving from 1785 shows
the King of Tonga accepting a bowl of

1817

Perspectives

Book
Debating drug policy and the path to change

Drug Policy and the Public Good


Thomas Babor, Jonathan
Caulkins, Grith Edwards,
Benedikt Fischer, David Foxcroft,
Keith Humphreys, Isidore Obot,
Jrgen Rehm, Peter Reuter,
Robin Room, Ingeborg Rossow,
John Strang. Oxford University
Press, 2010. US$5795. Pp 368.
ISBN 0199557128.

Cannabis Policy: Moving


Beyond Stalemate
Robin Room, Benedict Fischer,
Wayne Hall, Simon Lenton,
Peter Reuter. Oxford University
Press/The Beckley Foundation,
2010. Pp 300. US$5995.
ISBN 0199581487.

1818

As a historian of drug policy, my


natural inclination is to turn to the
past. An encounter in the mid-19th
century Cambridge market place
came to mind. A character in Charles
Kingsleys novel Alton Locke relates
what the druggists shop was selling:
youll see the little boxes, doozens
and dozens a ready on the counter
Opium, bor alive, opium! Opium was
on open sale in the 19th century; after
1868 pharmacists were in charge with
minimal regulation. In the absence of
much by way of eective therapeutics,
the drug was central to medical practice
and a mainstay of self-medicationthe
aspirin or paracetamol of its day.
Cannabis was a dierent matter. Its
widespread use in the Far East was never
replicated in the home country. Queen
Victoria did not, despite recent claims,
use cannabis in childbirth, although
her physician, William OShaughnessy,
wanted to introduce the drug into
medical practice. Uncertainty of its
action limited its use and dierentiated
cannabis from opium, whose alkaloids,
codeine, morphine, and later heroin,
gained it a central role in developing
professional therapeutics.
It is a far cry from the minimal regulation of the 19th century, to the world
in which these two books operate.
Drug Policy and the Public Good has been
written by an impressive team led by
Thomas Babor and aims to evaluate
critically the available research on drug
policy, and to present it in a way which
informs both the policy maker and
the scientic community. Its scope
is intended to be comprehensive and
international, to inform the debate in
countries where research is thin on the
ground as well as in those that produce
more of it. The books contributors
write about why people use drugs, who
uses drugs, and trends in use. Illicit drug
use is associated with a range of harms,
disease, disability, mortality, criminality,
and other social harms. However, as the

authors point out, for most countries,


the burdens, harms, and costs of illicit
drugs are less than those attributable
to alcohol and tobacco. We learn
about how and why drug markets
operate and their eect on price.
Strategiesprevention, services for
drug users, supply control, prescription
regimes, and criminal sanctionsare
carefully examined. Drugs operate
within a system of international
control that, despite increasingly vociferous attempts from civil society
organisations in recent years, shows
little sign of change. National policies

for most countries, the burdens,


harms, and costs of illicit drugs are
less than those attributable to
alcohol and tobacco
must conform to these international
principles, although the national
experience does dier; country studies
from Nigeria to Sweden illustrate
the point. The book ends with ten
conclusions from the evidence: these
range from the assurance that there is
no magic bullet for drug problems, to
the importance of a countrys pharmacy
system in regulation. Perhaps not so
far, then, from the 19th century.
Cannabis Policy: Moving Beyond Stalemate is from a similar stable led by Robin
Room. Again the model is one of policy
formed by science and evidence, but in
this case with an agenda that accepts
the need for change. About 80% of
illegal drug users in the world are
cannabis consumers. It is not dicult
to show the illogicality of this situation;
the authors point out that in practice,
cannabis control diers signicantly
at the national level and also on the
ground, with categories ranging from
full and partial prohibition, through depenalisation to decriminalisation. The
report supports more variety, perhaps
through a new convention concerned
with cannabis. State licensing or
other forms of state control of bodies

producing, wholesaling, and retailing


the drug are the optimum way to go.
This, of course, was the model in some
Eastern countries in the past.
Both books bring together a huge
amount of research across the many
elds of enquiry with which drug use
intersects. They will be of great value
to those seeking accessible summaries
of evidence. Yet two issues are mostly
absent. One is politics. The books draw
attention to the pervasive inuence of
the USA in international drug control
since World War II, and its opposition,
through UN drugs agencies, to WHO
recommendations for the scheduling
of cannabis for medical use. But the
politics of drug control could be given
more attention: the concept of path
dependency in policy comes to mind.
The second issue is an understanding
of how change takes place. Take the
move from heroin to methadone in
the UK during the 1970s, or the rise of
harm reduction with the emergence
of HIV/AIDS in the late 1980s. It would
be valuable to have some discussion of
general principles or models of change,
drawn from case studies.
Increasingly, illicit drugs are
discussed within a framework that
encompasses tobacco and alcohol.
Attitudes to tobacco since the 1950s
provide one model of change. How
were more restrictive policies adopted?
What was the interaction between
changing cultures of use and policy
responses? The attention paid in both
books to the Framework Convention
on Tobacco Control implies that the
authors see the connection. The
Cambridge market place in the 1850s
would be inconceivable today, but
the path from then to now was far
from rational. History tells us that the
reality of change is often messier, but
no less illuminating.

Virginia Berridge
Virginia.Berridge@lshtm.ac.uk

www.thelancet.com Vol 376 November 27, 2010

Perspectives

On Reection
A Chinese puzzle

Lunch with The Lancet


Peter Goldsworthy

20 km outside Shanghai, the Huachao Community


Health Centre is a showpiece clinic providing
modern western and traditional Chinese medicine to
1000 outpatients a day. With its polished oors and
white coated sta, the array of services it oers would
be the envy of many British health centres. But there is
one department you would never nd in the UK.
In a large room in the centre of the three-storey
building, up to 100 patients receive intravenous
infusions simultaneously, the plastic bags of liquid
suspended above their heads. The charge for the
treatment is 7 yuan (about US$1) which sounds
reasonableuntil the cost of the drug is added in. One
52-year-old woman is being treated for a cold with a
mixture of antibiotics and glucose. Another, who has a
badly bruised face after a fall following a stroke, is being
treated with Gingko Biloba, the Chinese herb, which is
apparently good for dissolving blood clots. According
to Donald Li, president of the Asian division of the
World Organisation of Family Doctors, the Chinese
value drugs much higher than the doctors advice. The
belief is that there is a therapeutic value from the drug
but no value in telling me what I should do. Injections
and intravenous treatment are liked. This is a culture
that seeks intravenous treatment.
Economics plays a part too. Charges for treatment
levied by hospitals and clinics are set deliberately low by
local councils, often too low to provide the service. They
have to make up the dierence by raising the prices of
drugs and diagnostic tests and encouraging patients to
have more. The situation is aggravated by doctors poor
rates of pay. They depend on bonuses paid by hospitals
from the income generated from drugs and tests. All the
incentives in the system are thus aligned, but towards
the wrong objectives.
The Chinese Government is committed to health
reform but changing the focus of the system from
prot to patients welfare is a major challenge. A
limited social insurance scheme exists, which is now
claimed to cover 90% of residents in Shanghai and
other urban centres (a separate scheme launched for
rural areas is aimed to cover 80%). However, cover is
restricted and many patients nd the co-payments
unaordable. There is a deep cultural belief in China
that there is nothing better than living longer. It is
going to be expensive to sustain.

Peter Goldsworthy walks into the Italian restaurant like a man


at home. After a smile and a ciao bello from the patroness,
he explains that he comes here most days for coee. His
actual home is just down the road; once, when his dog
disappeared, it turned up outside this corner establishment,
waiting for a waitress to take it home. A family doctor in the
small Australian city of Adelaide, Goldsworthy is also in the
top league of Australian writers. His novels, poems, short
stories, and plays are both popular and critically lauded.
A reviewer in The Australian newspaper called him the
Chekhov of his time and place. As we order our foodpork
for him, barramundi for mehe explains that writing
came before medicine. At 10 or 11 I used to write science
ction stories, about myself and my group of friends
cavorting around the galaxy. At university, while studying
medicine, he wrote poetry. By his early 30s, his collection
of poems, Readings from Ecclesiastes had begun winning
awards. Somerset Maugham said that the best education
for a writer was a medical degree, Goldsworthy says. I
think the two are very complementary, for many reasons.
Medicine trains you to observe human nature. Chekhov
is a good illustration, he says. He basically put himself
through medical school by writing comic sketches and
the vision just gets darker and darker really, and theres no
doubt thats through medical practice. Like Chekhov, but
unlike some other medics who take up writing, Goldsworthy
has never given up practice. Since nishing training he has
always split his days between the two: morning writing,
afternoon medicine. Its an arrangement that helps with
productivity. If I have to go and see patients at two oclock,
it concentrates the mind.
Medical themes turn up regularly in Goldsworthys work
from the dying child in Jesus Wants Me For a Sunbeam, to the
medical mnage trois in Three Dog Night. His work is more
than medical literature, but theres no doubt that exploring
how people deal with extremis is his bread and butter. On the
ip side, writing has also aected his medicine, he says. To
begin with I was the science nerd-type doctor, liked diagnosis
and didnt like people, you know. Ive become more patient,
although some of my patients might disagree.
Lunch over, Goldsworthy is heading to the practice, a
routine that provides a rich seam of inspiration. Medicine
is full of high stakes stories. As a writer you need those.
Stories like the one he tells of the elderly lady, close to death
and in need of morphine. As Goldsworthy brought up the
needle, he said Its only a little prick. It doesnt matter
dear, she replied, you have nice eyes.

Jeremy Laurance
J.Laurance@independent.co.uk

Stephen Pincock
stephen.pincock@journalist.co.uk

www.thelancet.com Vol 376 November 27, 2010

1819

Perspectives

The art of medicine


Neuron overload and the juggling physician
Patients often complain that their doctors dont listen.
Although there are probably a few doctors who truly are
tone deaf, most are reasonably empathic human beings,
and I wonder why even these doctors seem prey to this
criticism. I often wonder whether it is sheer neuron overload
on the doctor side that leads to this problem. Sometimes it
feels as though my brain is juggling so many competing
details, that one stray request from a patienteven one
that is quite relevantmight send the delicately balanced
three-ring circus tumbling down.
One day, I tried to work out how many details a doctor
needs to keep spinning in her head in order to do a
satisfactory job, by calculating how many thoughts I have
to juggle in a typical oce visit. Mrs Osorio is a 56-year-old
woman in my practice. She is somewhat overweight. She
has reasonably well-controlled diabetes and hypertension.
Her cholesterol is on the high side but she doesnt take any
medications for this. She doesnt exercise as much as she
should, and her last DEXA scan showed some thinning of
her bones. She describes her life as stressful, although shes
been good about keeping her appointments and getting
her blood tests. Shes generally healthy, someone whod
probably be described as an average patient in a medical
practice, not excessively complicated.
Here are the thoughts that run through my head as I
proceed through our 20-min consultation.
Good thing she did her blood tests. Glucose is a little better.
Cholesterol isnt great. May need to think about starting a
statin. Are her liver enzymes normal?
Her weight is a little up. I need to give her my talk about ve
fruits and vegetables and 30 min of walking each day.
Diabetes: how do her morning sugars compare to her
evening sugars? Has she spoken with the nutritionist lately?
Has she been to the eye doctor? The podiatrist?
Her blood pressure is good but not great. Should I add
another BP med? Will more pills be confusing? Does the
benet of possible better blood pressure control outweigh
the risk of her possibly not taking all of her meds?
Her bones are a little thin on the DEXA. Should I start a
bisphosphonate that might prevent osteoporosis? But now
Im piling yet another pill onto her, and one that requires
detailed instructions. Maybe leave this until next time?
How are things at home? Is she experiencing just the usual
stress of life, or might there be depression or anxiety disorder
lurking? Is there time for the depression questionnaire?
Health maintenance: when was her last mammogram? PAP
smear? Has she had a colonoscopy since she turned 50? Has
she had a tetanus booster in the past 10 years? Does she
qualify for a pneumonia vaccine?

1820

Ms Osorio interrupts my train of thought to tell me that


her back has been aching for the past few months. From
her perspective, this is probably the most important item in
our visit, but the fact is that shes caught one of my neurons
in mid-re (the one thats thinking about her blood sugar,
which is segueing into the neuron thats preparing the
diet-and-exercise discussion, which is intersecting with the
one thats debating about initiating a statin). My instinct
is to put one hand up and keep all interruptions at bay. Its
not that I dont want to hear what she has to say, but the
sensation that Im juggling so many thoughts, and need
to resolve them all before the clock runs down, that keeps
me in moderate state of panic. What if I drop onewhat
if one of my thoughts evaporates while I address another
concern? Im trying to type as fast as I can, for the very
sake of not letting any thoughts escape, but every time I
turn to the computer to write, Im not making eye contact
with Mrs Osorio. I dont want my patient to think that
the computer is more important than she is, but I have to
keep looking toward the screen to get her lab results, check
her mammogram report, document the progress of her
illnesses, order the tests, rell her prescriptions.
Then she pulls a form out her of bag: her insurance
company needs this form for some reason or another. An
innocentand completely justiedrequest, but I feel that
this could be the straw that breaks the camels back, that
the precarious balance of all that Im keeping in the air will
be simply unhinged. I nod, but indicate that we need to do
her physical examination rst. I barrel through the basics,
then quickly check for any red-ag signs that might suggest
that her back pain is anything more than routine muscle
strain. I return to the computer to input all the information,
mentally running though my checklist, anxious that nothing
important slips from my brains holding bay.
I want to do everything properly and cover all our bases,
but the more eort I place into accurate and thorough
documentation, the less time I have to actually interact with
my patient. A glance at the clock tells me that weve gone well
beyond our allotted time. I stand up and hand Mrs Osorio her
prescriptions. What about my insurance form, she asks. It
needs to be in by Friday, otherwise I might lose my coverage.
I clap my hand against my forehead; Ive completely forgotten
about the form shed asked about just a few minutes ago.
Studies have debunked the myth of multitasking in human
beings. The concept of multitasking was developed in the
computer eld to explain the idea of a microprocessor doing
two jobs at one time. It turns out that microprocessors are in
fact linear, and actually perform only one task at a time. Our
computers give the illusion of simultaneous action based
on the microprocessor scheduling competing activities in
www.thelancet.com Vol 376 November 27, 2010

Perspectives

www.thelancet.com Vol 376 November 27, 2010

The printed journal


includes an image merely
for illustration

Dean, Graham/The Bridgeman Art Library

a complicated integrated algorithm. Like microprocessors,


we humans cant actually concentrate on two thoughts at
the same exact time. We merely zip back and forth between
them, generally losing accuracy in the process. At best, we
can juggle only a handful of thoughts in this manner.
The more thoughts we juggle, the less we are able to
attune fully to any given thought. To me, this is a recipe
for disaster. Today I only forgot an insurance company
form. But what if Id forgotten to order her mammogram,
or what if Id relled only ve of her six medicines? What if
Id forgotten to fully explain the side-eects of one of her
medications? The list goes on, as does the anxiety.
At the end of the day, my mind spins as I try to remember
if Ive forgotten anything. Mrs Osorio had seven medical
issues to consider, each of which required at least ve
separate thoughts: thats 35 thoughts. I saw ten patients
that afternoon: thats 350. Id supervised ve residents that
morning, each of whom saw four patients, each of whom
generated at least ten thoughts. Thats another 200 thoughts.
Its not to say that we cant handle 550 thoughts in a working
day, but each of these thoughts potentially carries great risk if
improperly evaluated. If I do a good job juggling 98% of the
time, that still leaves ten thoughts that might get lost in the
process. Any one of those lost thoughts could translate into a
disastrous outcome, not to mention a possible lawsuit. Most
doctors are reasonably competent, caring individuals, but the
overwhelming swirl of thoughts that we must keep track of
leaves many of us in a perpetual panic that something serious
might slip. This is what keeps us awake at night.
There are many proposed solutionscomputergenerated reminders, case managers, ancillary services. To
me, the simplest one would be time. If I had an hour for each
patient, Id be a spectacular doctor. If I could let my thoughts
roll linearly and singularly, rather than simultaneously and
haphazardly, I wouldnt fear losing anything. I suspect that
it would actually be more ecient, as my patients probably
wouldnt have to return as frequently. But realistically,
no one is going to hand me a golden hour for each of my
patients. My choices seem to boil down to entertaining
fewer thoughts, accepting decreased accuracy for each
thought, giving up on thorough documentation, or having
a constant headache from neuronal overload.
These are the choices that practising physicians face
every day, with every patient. Mostly we rely on our clinical
judgment to prioritise, accepting the trade-o that is
inevitable with any compromise. We attend to the medical
issues that carry the greatest weight and then have to let
some of the lesser ones slide, with the hope that none of
these seemingly lesser ones masks something grave.
Some computers have indeed achieved the goal of
true multitasking, by virtue of having more than one
microprocessor. In practice, that is like possessing an
additional brain that can function independently and thus
truly simultaneously. Unless the transplant eld advances

Graham Dean, Small Echo (2004)

drastically, there is little hope for that particular deus ex


machina. In some cases, having a dedicated and competent
clinical partner such as a one-on-one nurse can come close
to simulating a second brain, but most medical budgets
dont allow for such stang indulgence.
As it stands, it seems that we will simply have to continue
this impossible mental high-wire act, juggling dozens of
clinical issues in our brains, panicking about dropping a
critical one. The resultant neuronal overload will continue
to present a distracted air to our patients that may be
interpreted as us not listening, or perhaps not caring.
When my computer becomes overloaded, it simply
crashes. Usually, I reboot in a fury, angry about all my lost
work. Now, however, I view my computer with a tinge
of envy. It has the luxury of being able to crash, and of a
reassuring, omniscient hand to press the reboot button.
Physicians are permitted no such extravagance. I pull out
the bottle of paracetamol tablets from my desk drawer and
set about disabling the childproof cap. Its about the only
thing I truly have control over.

Danielle Ofri
Department of Medicine, New York University School of Medicine,
New York, NY 10016, USA

Further reading
Ofri D. Incidental ndings:
lessons from my patients in the
art of medicine. Boston, MA:
Beacon Press, 2006.
Ofri D. Medicine in translation:
journeys with my patients.
Boston, MA: Beacon Press, 2010.
Ofri D. Singular intimacies:
becoming a doctor at Bellevue.
Boston, MA: Beacon Press, 2009.

1821

Obituary

David Geraint James


Internationally acclaimed specialist in
sarcoidosis. Born on Jan 22, 1922, in Treherbert,
UK, he died on Oct 22, 2010, in London, UK.
As many doctors will testify, an early encounter with an
inspiring gure can engender a lifetime preoccupation with
a particular disorder. It was sarcoidosis that caught and held
David (Gerry) Jamess attention; the man who introduced
him to it was Professor John Scadding, renowned specialist
in respiratory medicine at what was then, in the late 1940s,
the Brompton Chest Hospital in London, UK. The features
of Jamess career that were soon to emerge as so distinctive
were the extent to which he not only acted as a focal point
for others who shared his enthusiasm, but acquired an
international reputation while so doing.
James trained at Cambridge University and Londons
Middlesex Hospital, where his nal clinical year coincided
with the wartime blitz on the city. With his rst house jobs
completed he joined the Royal Navy and served as doctor to a
small eet of vessels sweeping the English Channel for mines.
It was on demobilisation and his return to London that he
went to work for Scadding at the Brompton. At this hospital
and also at another in London, the Hammersmith, Scadding
had accumulated a number of patients with sarcoidosis, and
he instilled in James his own interest in the disorder. On the
personal level too the Hammersmith Hospital had something
to oer James: another young doctor called Shelia Sherlock,
later to establish an international reputation of her own in
hepatology. They married in 1951.
8 years later James was appointed consultant physician at
Londons Royal Northern Hospital. Within months he had
1822

started the worlds rst properly organised sarcoidosis clinic.


Here he encouraged clinicians, pathologists, epidemiologists,
and radiologists to collaborate in attempts to reach a better
understanding of the disease. For three decades these
clinics were what a couple of Jamess sometime colleagues,
Professors Alimuddin Zumla and Om Sharma, have described
as a Mecca for international sarcoidologists. They came
for three reasons, according to Zumla, now professor of
infectious diseases and international health at University
College, London. The rst was simply a question of numbers.
Clinics devoted to sarcoidosis were exceptional, and James
soon began to receive referrals from all over the country,
especially of patients with intractable disease. Specialists
had an unrivalled opportunity to encounter a whole range of
patients with sarcoidosis in one place. Then there was Jamess
unrivalled experience in treating these patients; he ran a
number of drug trials from the clinic. Finally, there was the
man himself. Gerry was an excellent teacher, says Zumla. If
you listened to his lectures you would never forget them. He
had a way of imparting knowledge that was special.
Why the fascination with this particular disease? It was
the mystery, the enigma, according to Zumla. Gerry liked
challenges. When he rst became interested in sarcoidosis it
was viewed as a rare chronic disease of unknown aetiology.
In the course of his career James did much to unravel its
pathophysiology and pathogenesis. He started by classifying
the disease into its dierent types and treating them
accordingly. Then he investigated the immunology of the
illness. He didnt have a well-funded lab, says Sharma, who
spent 3 years with James in the 1960s before going to the
University of Southern California where he is now a professor
of medicine. But he had a knack of working with people in
other specialties. Sharma recalls that whenever he needed
access to expertise not available at the Royal Northern, James
always knew the right person with whom to collaborate.
Zumla rst met James when he began working for him
as a registrar in 1984, and went on to become a friend. He
speaks of his old boss with deep aection. He was a kindhearted, seless, loving, and gentle soulbut one who could
be rm with the people he was teaching to be sure they
focused on what they were doing and got the most out of
it. He was full of praise. He supported and encouraged. He
reassured patients. They called him the King of Sarcoid. In
1958, James organised the rst ever international conference
on sarcoidosis at the Brompton Hospital, which led to the
formation of the World Association of Sarcoidosis and
other Granulomatous Disorders (WASOG). James became
its founder President. In 1999, James and Zumla co-edited
The Granulomatous Disorders that rapidly became a standard
text. It will continue to serve as one of his more tangible
memorials. James is survived by two daughters.

Geo Watts
geo@scileg.freeserve.co.uk

www.thelancet.com Vol 376 November 27, 2010

Correspondence

As doctors, reading the Editorial


Chinese doctors are under threat
(Aug 28, p 657)1 was such a
consolation. The Chinese media
certainly has an important role in the
demonisation of doctors and nurses.
However, there is a saying in China:
you can never make applause with
one palm. The reasons are not always
from the outside; some originate from
the doctors and nurses themselves.
The low quality of medical service in
China is universal. The access qualication for a physician is not as strict
as in western countries. One can
apply for a physicians licence even
without formal college education,
and this is granted if that person can
pass the qualication examination,
which is not actually dicult. For a
nurse, the education period is shorter
and the qualication examination
easier. In previous years, a nurse, even
one without a medical education
background, could work as a doctor
without the need for any qualication
examination. Many of these doctors
are still working in their positions
today.
The Editorial cites a few Chinese
media reports that have misled public
opinion. Nevertheless, some reports
highlight real cases of malpractice,
and the most frequent reason is
the doctor or nurses lack of professional ethics. Without the deserved
salary and respect, deterioration of
professional ethics is perhaps to be
expected in other professions. But a
medical workers lack of ethics directly
harms patients. So it is no surprise
that patients are usually treated as the
victims in the publics opinion when
medical disputes occur.
Moreover, for a patient subjected
to malpractice, it is very hard in
the current justice system to win a
legal claim against a big hospital.
If a patient cannot nd protection
from the law, revenge via violence
is conceivable.
www.thelancet.com Vol 376 November 27, 2010

Although we agree that Chinas


health-system reforms cannot be
successful without reforming the social
and economic status of doctors,1
we believe that improvement of the
quality of medical workers and rening
of medical practice laws are essential
as well.

with doctorpatient relationships in


China. To face up to the threat in China,
we should full the Hippocratic oath,
learn how to put ourselves at patients
disposal, and earn their trust.

We declare that we have no conicts of interest.

zhao_jc120@126.com

Jiwei Huang, Lvnan Yan, *Yong Zeng


zengyongmd@gmail.com
Department of Hepato-Biliary-Pancreatic Surgery,
West China Hospital, Sichuan University,
Chengdu 610041, Sichuan, China
1

The Lancet. Chinese doctors are under threat.


Lancet 2010; 376: 657.

As young doctors in China, we feel


strongly about the grim situation in
our country, which was mentioned
in the Editorial entitled Chinese
doctors are under threat.1 Improper
health-care system reforms did not
acquire Chinese doctors a deserved
improvement in social and economic
status. Owing to misleading media
reports and public misunderstanding
of the medical profession, Chinese
doctors have become progressively
demonised.
We are particularly excited by the
Editorials call for Chinese doctors to
be involved more in shaping health
policy, by giving voice to their own
experiences and constructive ideas
about the health system. Fortunately,
the new health-care system reforms
include more nancial input from
government and abrogating reversal
of the burden of proof in medical
disputes, which had been long
requested by medical personnel. We
can expect a promising future.
On the other hand, young trainee
doctors such as ourselves should be
quite clear what a good doctor is, and
how we can become one, especially in
the eyes of patients. Patients priorities
for general practice care include humaneness, competence and accuracy,
involvement in decisions, and time for
care.2 Eective communication and
correct assumptions about patients
preferences are important for dealing

We declare that we have no conicts of interest.

AFP/Getty Images

Facing up to the threat


in China

Yi Yang, *Ji-Chun Zhao, Yu-Pei Zou,


Lu-Nan Yan
Department of Liver and Vascular Surgery
(YY, JCZ, LNY) and Department of Geriatrics (YPZ),
West China Hospital, Sichuan University,
Chengdu 610041, Sichuan, China
1
2

The Lancet. Chinese doctors are under threat.


Lancet 2010; 376: 657.
Wensing M, Jung HP, Mainz J, Olesen F, Grol R.
A systematic review of the literature on
patient priorities for general practice care.
Part 1: description of the research domain.
Soc Sci Med 1998; 47: 157388.

We would like to oer our thanks to


The Lancet for the Editorial depicting
the threatened life of Chinese doctors.1
It has become an immediate topic
of discussion here. To gain further
insight into the attitudes of Chinese
doctors, we did a survey on Ding Xiang
Yuan, the most popular biomedical
website in China with a registered user
base of over 2 million. By Oct 2, 2010,
14 577 doctors had participated in the
survey, including 5710 residents, 5132
attending physicians, 2256 associate
chief physicians, and 609 chief
physicians.
When asked whether they were
concerned about the health-system
reforms in China and what their
primary concerns were, 67% of doctors
said that they were strongly concerned
about the reforms, with 65% choosing
safe medical treatments and 53%
choosing cure of the patients as their
primary concern.
When asked about the main
reasons for the increased tension
between doctors and patients, 66%
said that their hospitals encountered
one to three medical disputes per
month; 78% blamed it on a lack of
government funding to hospitals and
70% accused the public media for
negative reports, with 86% of doctors
believing that negative reports were

For the Ding Xiang Yuan website


see http://www.dxy.com

Submissions should be
made via our electronic
submission system at
http://ees.elsevier.com/
thelancet/

1823

Correspondence

used to increase audience ratings. In


terms of false media reports, 49% of
doctors took them as deliberate, and
37% thought that the media needed
qualied scientic gatekeepers for
their medical news.
Finally, 91% of doctors strongly
agreed that Chinas health-system
reforms could not be successful without reforming the social and economic status of doctors.
We declare that we have no conicts of interest.

*Danghui Yu, Tiantian Li


alexshing@yahoo.com.cn
Academic Journal of Second Military Medical University,
Shanghai 200433, China (DHY); and Ding Xiang Yuan
Biomedical Forum, Hangzhou, Zhejiang, China (TTL)
1

The Lancet. Chinese doctors are under threat.


Lancet 2010; 376: 657.

Is antiretroviral therapy
modifying the HIV
epidemic?
Julio Montaner and colleagues
(Aug 14, p 532)1 present data from
British Columbia, Canada, on the
association between rates of HIV
diagnoses, coverage of antiretroviral
therapy (ART), and average HIV viral
load. They are incorrect in describing
their study as a population-based
cohort study. It is an ecological study,
because the data on exposure (ART
and viral load) and outcome (HIV
diagnoses) do not come from the same
couples. This is a crucial distinction
because population associations
found in ecological studies often fail
to reect individual-level biological
eects.2
Montaner and colleagues do not
consider the most important confounding eectnamely, the rate
of unsafe injecting in injecting drug
users. Montaners group has reported
data which show that policy changes
led to a greater than 50% decline in
syringe borrowing, and reductions in
HIV incidence in British Columbia.3
Reduction in unsafe injecting has been
acknowledged as an important reason
1824

for decreased HIV incidence in this


population.
HIV diagnoses have increased over
the past decade in men who have
sex with men in developed-country
settings, including Canada.4 In Australia,
HIV diagnoses among such men have
increased despite very high levels of HIV
testing and treatment, and declining
average viral load.5 Thus, ecological
data from comparable settings do not
support the idea that lower viral load
has led to decreased HIV transmission
in men who have sex with men.
Ecological studies can only suggest
hypotheses, they cannot conrm
them.2 Ultimately, the results of
randomised controlled trials are
needed to dene the role of ART in
HIV prevention.
We declare that we have no conicts of interest.

*Andrew E Grulich, David P Wilson


agrulich@nchecr.unsw.edu.au
National Centre in HIV Epidemiology and Clinical
Research, University of New South Wales, Sydney,
NSW 2010, Australia
1

Montaner JS, Lima VD, Barrios R, et al.


Association of highly active antiretroviral
therapy coverage, population viral load, and
yearly new HIV diagnoses in British Columbia,
Canada: a population-based study. Lancet
2010; 376: 53239.
Rothman KJ, Greenland S, Lash TL. Modern
epidemiology, third edn. Philadelphia:
Lippincott Williams and Wilkins, 2008.
Kerr T, Small W, Buchner C, et al. Syringe sharing
and HIV incidence among injection drug users
and increased access to sterile syringes.
Am J Public Health 2010; 100: 144953.
Sullivan PS, Hamouda O, Delpech V, et al.
Reemergence of the HIV epidemic among men
who have sex with men in North America,
Western Europe, and Australia, 19962005.
Ann Epidemiol 2009; 19: 42331.
Falster K, Gelgor L, Shaik A, et al. Trends in
antiretroviral treatment use and treatment
response in three Australian states in the rst
decade of combination antiretroviral
treatment. Sex Health 2008; 5: 14154.

treatment reduces population-level


incidence. The appropriate variables
were not measured, timing of the
associated changes lacks biological
plausibility, and their ndings have
better alternative explanations.
First, their surrogate for transmissionnew diagnosesis not
new infections. Many new diagnoses
represent infections that occurred
years previously.2 Notably, the main
decline in diagnoses Montaner and
colleagues saw occurred at the outset
of highly active antiretroviral therapy
(HAART) in the late 1990s, when
HAART was too new and too limited
to appreciably aect new diagnoses.
Yes, numbers on treatment increased
as new diagnoses declined, but
such declines are common in new
programmes. The easiest are reached
rst; the diminishing residual are
reached with more diculty.
Second, Montaner and colleagues do
not report true community viral loads,
but from among those who reach
testing and continue on treatment.
Importantly, they omit people with
high-viral-load acute infection, so
crucial to HIV propagation3 and who
receive no HAART. Rather than true
declines in community viral loads,
their declines probably reect a
treatment cohort eect. Initially, such
viral loads represent individuals not
yet or just starting treatmentwith
relatively higher viral loads. But over
time, successively more tests represent
individuals under ongoing treatment
with much lower viral loads.
Antiretroviral drugs oer promise
for prevention intervention, especially
in some targeted application, but we
need solid research.
We declare that we have no conicts of interest.

Antiretroviral drugs have clear


potential to reduce HIV transmissibility. However, the ecological
association Julio Montaner and
colleagues1 nd between increased
individuals on treatment, declines
in viral loads, and declining new HIV
diagnoses in British Columbia, Canada,
does not provide good evidence that

*James D Shelton, Myron Cohen,


Matthew Barnhart, Timothy Hallett
jshelton@usaid.gov
Bureau for Global Health, US Agency for
International Development, Washington, DC 20523,
USA (JDS); University of North Carolina, Chapel Hill,
NC, USA (MC); UNICEF, New York, NY, USA (MB);
and Imperial College, London, UK (TH)

www.thelancet.com Vol 376 November 27, 2010

Correspondence

Montaner JSG, Lima VD, Barrios R, et al.


Association of highly active antiretroviral
therapy coverage, population viral load, and
yearly new diagnoses in British Columbia,
Canada: a population-based study. Lancet
2010; 376: 53239.
White PJ, Ward H, Garnett GP. Is HIV out of
control in the UK? An example of analyzing
patterns of HIV spreading using incidence-toprevalence ratios. AIDS 2006; 20: 1898901.
Pilcher CD, Tien HC, Eron JJ, et al. Brief but
ecient: acute HIV infection and the sexual
transmission of HIV. J Infect Dis 2004;
189: 178592.

Authors reply
We acknowledged in our paper that
ours was an ecological study, and as
such the results could not be taken as
denitive proof of causality. We also
indicated that the association between
increasing coverage of highly active
antiretroviral therapy (HAART) in HIVinfected individuals who met contemporary treatment guidelines and
decreasing yearly new HIV diagnoses
occurred against a background of
increased yearly HIV testing, as well
as improved risk ascertainment due
to mandatory HIV reporting, and
increased rates of sexually transmitted
infections.
Notably, our results were internally
reproducible. We recorded decreases
in yearly new HIV diagnoses during
two distinct periods of HAART expansion, which were separated by a stable
period of HAART use. The latter
two periods were characterised on a
prospective basis and were entirely
consistent with the predictions of our
previously published mathematical
models.1 Finally, we were able to
relate the reductions in community
plasma viral load during HAART
expansion with the decrease in new
HIV diagnoses, providing a plausible
biological mechanism to account
for the association, as proposed by
others.2
We also acknowledged that the use
of yearly new HIV diagnoses represented a limitation of the study;
however, it is also clear that there is
no widely accepted gold standard to
estimate HIV incidence, particularly
in population-wide studies. In this
www.thelancet.com Vol 376 November 27, 2010

regard, the consistency of our results


with our previous report looking at
the correlation between community
concentrations of HIV-1 RNA in
plasma and HIV incidence in a well
characterised cohort of injection drug
users is highly reassuring.3
Andrew Grulich and David Wilson
express concerns about our previous
report of declines in unsafe injecting.
However, that report was based on
a small study within Vancouvers
Downtown Eastside, and the present
study reports on all of the province of
British Columbia, where it has been
estimated that the number of sterile
needles and syringes distributed remains inadequate.4
Grulich and Wilson call for
randomised controlled trials (RCTs) to
further characterise the associations
described in our paper. However, the
feasibility of RCTs within the segment
of HIV-infected individuals eligible for
treatment could prove ethically dicult
if this requires withholding HAART
against current standard of care. RCTs
might be more realistic within the
segment of HIV-infected individuals not
eligible for therapy, as proposed by the
test and treat strategy, yet the most
recent guidelines5 are so inclusive that
this segment has become very small
in our setting. Alternative approaches,
such as a modied delayed-start or
randomised-start design, might be
successfully adapted to existing HAART
rollout initiatives.
In the absence of such data, the
available evidence strongly points
to a substantial preventive benet
that can be derived from aggressively
rolling out HAART to all those in
medical needthe actual focus of our
report.
We declare that we have no conicts of interest.

*Julio S G Montaner,
P Richard Harrigan, Thomas Kerr,
Evan Wood, Patricia Daly
jmontaner@hivnet.ubc.ca
British Columbia Centre for Excellence in HIV/AIDS,
Vancouver, BC V6Z 1Y6, Canada (JSGM, PRH, TK,
EW); and Vancouver Coastal Health Authority,
Vancouver, BC, Canada (PD)

Lima VD, Johnston K, Hogg RS, et al. Expanded


access to highly active antiretroviral therapy: a
potentially powerful strategy to curb the
growth of the HIV epidemic. J Infect Dis 2008;
198: 5967.
Das M, Chu PL, Santos G-M, et al. Decreases in
community viral load are accompanied by
reductions in new HIV diagnoses in San
Francisco. PLoS One 2010; 5: e11068.
Wood E, Kerr T, Marshall B, et al. Longitudinal
community plasma HIV-1 RNA concentrations
and incidence of HIV-1 among injecting drug
users: prospective cohort study. BMJ 2009;
338: b1649.
Harvard SS, Hill WD, Buxton JA. Harm reduction
product distribution in British Columbia.
Can J Public Health 2008; 99: 44650.
Thompson M, Aberg J, Cahn P, et al.
Antiretroviral treatment of adult HIV
infection2010 recommendations of the
International AIDS SocietyUSA Panel. JAMA
2010; 304: 32133.

Licence for Botox in


so-called chronic
migraine
Science Photo Library

As experts in headache classication


and in the methodology of clinical
trials in migraine and other headaches,
we were surprised to read that the UKs
Medicines and Healthcare Products
Regulatory Agency had extended the
licence for botulinum A toxin (Botox)
to include prophylactic treatment for
headache in adults who have chronic
migraine.
Botulinum A toxin has been studied
in numerous randomised controlled
trials in episodic migraine and in
tension-type headache with negative
results.1 After these indications were
given up, two large trials (n=1384) were
done in so-called chronic migraine.2,3
However, the diagnosis was not correct
since 65% had medication overuse,24
which precludes the diagnosis of
chronic migraine according to the
international headache classication.5
Furthermore, it is surprising that 35%
of the patients2,3 had never before
received any pharmacological prophylaxis. Is treatment of such patients
with botulinum A toxin really indicated
when they would probably benet
from, for example, propranolol?
Most importantly, the published
data are weak. One of the two trials
1825

Correspondence

Montaner JSG, Lima VD, Barrios R, et al.


Association of highly active antiretroviral
therapy coverage, population viral load, and
yearly new diagnoses in British Columbia,
Canada: a population-based study. Lancet
2010; 376: 53239.
White PJ, Ward H, Garnett GP. Is HIV out of
control in the UK? An example of analyzing
patterns of HIV spreading using incidence-toprevalence ratios. AIDS 2006; 20: 1898901.
Pilcher CD, Tien HC, Eron JJ, et al. Brief but
ecient: acute HIV infection and the sexual
transmission of HIV. J Infect Dis 2004;
189: 178592.

Authors reply
We acknowledged in our paper that
ours was an ecological study, and as
such the results could not be taken as
denitive proof of causality. We also
indicated that the association between
increasing coverage of highly active
antiretroviral therapy (HAART) in HIVinfected individuals who met contemporary treatment guidelines and
decreasing yearly new HIV diagnoses
occurred against a background of
increased yearly HIV testing, as well
as improved risk ascertainment due
to mandatory HIV reporting, and
increased rates of sexually transmitted
infections.
Notably, our results were internally
reproducible. We recorded decreases
in yearly new HIV diagnoses during
two distinct periods of HAART expansion, which were separated by a stable
period of HAART use. The latter
two periods were characterised on a
prospective basis and were entirely
consistent with the predictions of our
previously published mathematical
models.1 Finally, we were able to
relate the reductions in community
plasma viral load during HAART
expansion with the decrease in new
HIV diagnoses, providing a plausible
biological mechanism to account
for the association, as proposed by
others.2
We also acknowledged that the use
of yearly new HIV diagnoses represented a limitation of the study;
however, it is also clear that there is
no widely accepted gold standard to
estimate HIV incidence, particularly
in population-wide studies. In this
www.thelancet.com Vol 376 November 27, 2010

regard, the consistency of our results


with our previous report looking at
the correlation between community
concentrations of HIV-1 RNA in
plasma and HIV incidence in a well
characterised cohort of injection drug
users is highly reassuring.3
Andrew Grulich and David Wilson
express concerns about our previous
report of declines in unsafe injecting.
However, that report was based on
a small study within Vancouvers
Downtown Eastside, and the present
study reports on all of the province of
British Columbia, where it has been
estimated that the number of sterile
needles and syringes distributed remains inadequate.4
Grulich and Wilson call for
randomised controlled trials (RCTs) to
further characterise the associations
described in our paper. However, the
feasibility of RCTs within the segment
of HIV-infected individuals eligible for
treatment could prove ethically dicult
if this requires withholding HAART
against current standard of care. RCTs
might be more realistic within the
segment of HIV-infected individuals not
eligible for therapy, as proposed by the
test and treat strategy, yet the most
recent guidelines5 are so inclusive that
this segment has become very small
in our setting. Alternative approaches,
such as a modied delayed-start or
randomised-start design, might be
successfully adapted to existing HAART
rollout initiatives.
In the absence of such data, the
available evidence strongly points
to a substantial preventive benet
that can be derived from aggressively
rolling out HAART to all those in
medical needthe actual focus of our
report.
We declare that we have no conicts of interest.

*Julio S G Montaner,
P Richard Harrigan, Thomas Kerr,
Evan Wood, Patricia Daly
jmontaner@hivnet.ubc.ca
British Columbia Centre for Excellence in HIV/AIDS,
Vancouver, BC V6Z 1Y6, Canada (JSGM, PRH, TK,
EW); and Vancouver Coastal Health Authority,
Vancouver, BC, Canada (PD)

Lima VD, Johnston K, Hogg RS, et al. Expanded


access to highly active antiretroviral therapy: a
potentially powerful strategy to curb the
growth of the HIV epidemic. J Infect Dis 2008;
198: 5967.
Das M, Chu PL, Santos G-M, et al. Decreases in
community viral load are accompanied by
reductions in new HIV diagnoses in San
Francisco. PLoS One 2010; 5: e11068.
Wood E, Kerr T, Marshall B, et al. Longitudinal
community plasma HIV-1 RNA concentrations
and incidence of HIV-1 among injecting drug
users: prospective cohort study. BMJ 2009;
338: b1649.
Harvard SS, Hill WD, Buxton JA. Harm reduction
product distribution in British Columbia.
Can J Public Health 2008; 99: 44650.
Thompson M, Aberg J, Cahn P, et al.
Antiretroviral treatment of adult HIV
infection2010 recommendations of the
International AIDS SocietyUSA Panel. JAMA
2010; 304: 32133.

Licence for Botox in


so-called chronic
migraine
Science Photo Library

As experts in headache classication


and in the methodology of clinical
trials in migraine and other headaches,
we were surprised to read that the UKs
Medicines and Healthcare Products
Regulatory Agency had extended the
licence for botulinum A toxin (Botox)
to include prophylactic treatment for
headache in adults who have chronic
migraine.
Botulinum A toxin has been studied
in numerous randomised controlled
trials in episodic migraine and in
tension-type headache with negative
results.1 After these indications were
given up, two large trials (n=1384) were
done in so-called chronic migraine.2,3
However, the diagnosis was not correct
since 65% had medication overuse,24
which precludes the diagnosis of
chronic migraine according to the
international headache classication.5
Furthermore, it is surprising that 35%
of the patients2,3 had never before
received any pharmacological prophylaxis. Is treatment of such patients
with botulinum A toxin really indicated
when they would probably benet
from, for example, propranolol?
Most importantly, the published
data are weak. One of the two trials
1825

Correspondence

was negative for the primary endpoint


(headache episodes)2 and one positive
for the primary endpoint (headache
days).3 Although the combined results
for headache days were signicant,4
the dierence between the active and
placebo groups was only around 10%
for most variables and there was
no dierence between groups in
the reduction of intake of acute
medicines.24
The published papers give much
emphasis to the dierence between
baseline and treatment, which is indeed
impressive, but is equally so in the
placebo group.2,3 The pooled analysis4
of the two large trials2,3 states that the
ecacy of the medication overuse
stratum has been analysed and will be
published,4 but what is interesting is
the possible magnitude of the eect
in the patients without medication
overuse.
Finally, no assessment of the eectiveness of blinding was done. Blinding
seems likely to have been far from
optimal since botulinum A toxin
weakens muscles and placebo does not.
The facial expression is therefore likely
to change after botulinum A toxin.
This presumed lack of blinding alone
can probably explain a 10% dierence
between the active drug and placebo.
JO has received grants or research support from, has
been a consultant or scientic adviser for, and/or
has been on the speakers bureaux of Merck and
Company, UCB Pharma, Bristol-Myers Squibb,
NeurAxon, Zogenix, and H Lundbeck. PTH declares
that he has no conicts of interest.

Jes Olesen, *Peer Tfelt-Hansen


For the MHRA assessment
procedure see http://www.
mhra.gov.uk/home/groups/par/
documents/websiteresources/
con093969.pdf

ptha@glo.regionh.dk
Danish Headache Centre, University of Copenhagen,
Glostrup Hospital, 2600 Glostrup, Denmark
1

1826

Evers S, Olesen J. Botulinum toxin in headache


treatment: the end of the road? Cephalalgia
2006; 26: 76971.
Aurora SK, Dodick DW, Turkel CC, et al.
OnabotulinumtoxinA for treatment of chronic
migraine: results from the double-blind,
randomized, placebo-controlled phase
of the PREEMPT 1 trial. Cephalalgia 2010;
30: 793803.
Diener HC, Dodick DW, Aurora SK, et al.
OnabotulinumtoxinA for treatment of
chronic migraine: results from the
double-blind, randomized, placebocontrolled phase of the PREEMPT 2 trial.
Cephalalgia 2010; 30: 80414.

Dodick DW, Turkel CC, DeGryse RE, et al.


OnabotulinumtoxinA for treatment for
treatment of chronic migraine: pooled
results from the double-blind, randomized,
placebo-controlled phases of the PREEMPT
clinical program. Headache 2010; 50: 92136.
Headache Classication Committee of the
International Headache Society. The
international classication of headache
disorders, 2nd edn. Cephalalgia 2004;
24 (suppl 1): 9160.

Response from MHRA


Jes Olesen and Peer Tfelt-Hansen
comment on the decision of the
Medicines and Healthcare Products
Regulatory Agency (MHRA) to grant
an extension of indication for Botox
(botulinum A toxin) in the prophylaxis
of headaches in adults with chronic
migraine (headaches on at least
15 days per month of which at least
8 days are with migraine).
In December, 2009, Allergan
Limited applied for an extension
to the licence for Botox to include
an indication for the prophylaxis of
headaches in adults with chronic
migraine, which was eventually
approved in July, 2010. The decision
was based on the evidence submitted
by the company, mainly data from
two pivotal double-blind placebocontrolled phase 3 trials1,2 and their
open-label extension; these data were
considered adequate to support this
extension. Furthermore, the decision
of the MHRA was endorsed by its
advisory body, the Commission on
Human Medicines with its experts.
Full details of the assessment
procedure, including a transparent
discussion of the study results and
a critical assessment of the benet/
risk balance of the medicinal product
in this new indication, are publicly
available on the MHRA website.

Jennifer Kyne
jennifer.kyne@mhra.gsi.gov.uk
Medicines and Healthcare Products Regulatory
Agency, London SW8 5NQ, UK
1

Aurora SK, Dodick DW, Turkel CC, et al.


OnabotulinumtoxinA for treatment of chronic
migraine: results from the double-blind,
randomized, placebo-controlled phase of the
PREEMPT 1 trial. Cephalalgia 2010;
30: 793803.

Diener HC, Dodick DW, Aurora SK, et al.


OnabotulinumtoxinA for treatment of
chronic migraine: results from the doubleblind, randomized, placebo-controlled phase
of the PREEMPT 2 trial. Cephalalgia 2010;
30: 80414.

Improving surgery
service delivery in
context
Luke Funk and colleagues (Sept 25,
p 1055)1 estimate that 2 billion people
lack access to surgery worldwide.
This estimate conrms anecdotal
experiences of substantial unmet
surgical need by those who work with
colleagues in low-income countries.
As Paul Myles and Guy Haller note
in their accompanying Comment,2
although the availability of pulse
oximetry was used as a proxy for access
in Funk and colleagues study, many
other factors such as human resource
constraints (numbers of skilled
surgical and anaesthesia providers),
health-system factors (public vs
private insurance, cost of care), and
other patient-driven factors aect
the actual delivery of care. Given the
complex interplay of these factors on
the ground, it is dicult to identify the
rate-limiting step to service delivery
for any given context.
A substantial (and unknown)
proportion of patients with surgically
correctable disorders do not reach
care at all. Specically, trauma
accounts for the greatest surgical
burden in poor countries, and most
deaths have been shown to occur in
the prehospital setting of countries
with no emergency system.3 In one
such setting (Uganda), we developed
and implemented a basic, contextappropriate prehospital curriculum
for lay rst-responders (police, taxi
drivers, public ocials) and showed
that this can improve care at minimal
cost. This intervention was coupled
with hospital-based trauma training,
and such an approach might also be
applicable in other austere settings.4
www.thelancet.com Vol 376 November 27, 2010

Correspondence

We declare that we have no conicts of interest.

*Doruk Ozgediz,
Jacqueline Mabweijano,
Cephas Mijumbi, Sudha Jayaraman,
Michael Lipnick
dozgediz@hotmail.com
University of Toronto, Toronto, ON M5G 2J8,
Canada (DO); Makerere University, Kampala,
Uganda (JM, CM); University of California at San
Francisco, San Francisco, CA, USA (SJ); and Global
Partners in Anesthesia and Surgery, San Francisco,
CA, USA (DO, JM, CM, SJ, ML)
1

2
3

Funk L, Weiser TG, Berry WR, et al. Global


operating theatre distribution and pulse
oximetry supply: an estimation from reported
data. Lancet 2010; 376: 105561.
Myles PS, Haller G. Global distribution of access
to surgical services. Lancet 2010; 376: 102728.
Mock CN, Jurkovic GJ, Nii-Amon Kotei D, et al.
Trauma mortality patterns in three nations of
dierent economic levels: implications for
global trauma system development. J Trauma
1998; 44: 80414.
Jayaraman S, Mabweijano JR, Lipnick M, et al.
First things rst: eectiveness and scalability
of a basic prehopital program for lay rst
responders in Kampala, Uganda. PLoS One
2010; 4: 17.
McQueen KA, Ozgediz D, Riviello R, et al.
Essential surgery: integral to the right to health.
Health Human Rights J 2010; 12: 13752.

Cancer funding in
developing countries:
the next health-care
crisis?
We welcome the call to expand
cancer control in low-income and
middle-income countries (LMIC) by
Paul Farmer and colleagues (Oct 2,
p 1186).1 The Global Fund to ght
AIDS, Tuberculosis and Malaria was
formed in 2002 after a call in The
Lancet by Amir Attaran and Jerey
Sachs2 for a special funding stream for
HIV/AIDS. Can we expect the same for
cancer?
In less developed countries alone,
more people develop (71 million) and
die from cancer (48 million) each year3
www.thelancet.com Vol 376 November 27, 2010

than are infected or die from HIV/AIDS


throughout the world (an estimated
27 million and 2 million, respectively,
in 2008).4 Cancer, unlike HIV/AIDS,
is also often curable, particularly if
diagnosed early.
Although there is much to be
learned from the achievements of the
focused approaches to AIDS, malaria,
and tuberculosis, cancer care requires
a much more individualised approach,
based on the type and stage of cancer
both of which require expertise to
determine. Notwithstanding the
mention by Farmer and colleagues of
the provision of cancer chemotherapy
in rural regions in Rwanda and Malawi
by non-specialists, successful cancer
treatment mostly requires a team
of skilled specialists to ensure good
diagnosis and the availability of
appropriate surgery, radiotherapy,
systemic therapy, and supportive care.
Doubtless, the complexities and cost
of cancer care have much to do with
the low priority assigned to cancer
control, which has created a kind of
triple penalty in which poor access
to care results in patients presenting
with advanced disease, greatly
increasing both the cost5 and toxicity
of treatment (including more surgical
mutilation) and resulting in much
lower survival rates. In turn, the poor
results back up the assumption that
poor countriesand wealthy donors
cannot aord to include cancer care on
their health agendas.
Funding will initially be required
to generate evidence on what
works and what doesnt in lowresource settings, coupled with
the development of sustainable
cancer control programmes and
an improvement in the results of
existing cancer treatment centres
through education of all stakeholders
with respect to what is possible when
diagnosis is early and basic cancer
care is accessible. This should lead
to an increase in available funding
from a wide range of sources, and a
gradual amelioration of the situation,
although the challenges involved

should not be underestimated, nor


too much expected as a result of the
successes achieved with HIV/AIDS.
We declare that we have no conicts of interest.

*Ian Magrath, Pierre Bey, Aziza Shad,


Simon Sutclie
International Network for Cancer Treatment and
Research (INCTR), 1180 Brussels, Belgium (IM);
Alliance Mondiale Contre le Cancer (INCTR France),
Paris, France (PB); INCTR USA, Chevy Chase, MD,
USA (AS); and INCTR Canada, Vancouver, BC,
Canada (SS)
1

Farmer P, Frenk J, Knaul FM, et al. Expansion of


cancer care and control in countries of low and
middle income: a call to action. Lancet 2010;
376: 118693.
Attaran A, Sachs J. Dening and rening
international donor support for combating the
AIDS pandemic. Lancet 2001; 357: 5761.
International Agency for Research on Cancer.
Globocan database. http://globocan.iarc.fr
(accessed Nov 17, 2010).
Avert. Global HIV/AIDS estimates, end of
2008. http://www.avert.org/worldstats.htm
(accessed Nov 9, 2010).
Kerr DJ, Midgley R. Can we treat cancer for a
dollar a day? Guidelines for low income
countries. N Engl J Med 2010; 363: 9.

Migration, health, and


care in French overseas
territories
France was recently reprimanded by
a UN human rights body1 concerned
about discriminatory political discourse and an increase in acts and
expressions of racism and xenophobia
in this country. The report also called
attention to the increasing diculties
faced by certain inhabitants of [French]
overseas territories in accessing health
care without discrimination.
In Mayotte, especially, access to
health care is very precarious, and it
is further hampered by security and
nancial concerns. Located in the
Comoros Archipelago, this French
island, with a population of 200 000,
of whom nearly a third do not have
regular residence status, holds
the record for a policy of expelling
undocumented foreigners. During the
rst quarter of 2010, there were 12 300
expulsions from Mayotte compared
with 14 700 from mainland France,

The printed
journal
includes an
image merely
for illustration
Reuters

Funk and colleagues study underscores the urgent need for resources to
implement and assess interventions
that reduce global disparities in access
to essential surgical care, which should
be seen as a basic human right.5

1827

Correspondence

Published Online
November 15, 2010
DOI:10.1016/S01406736(10)62108-3

or about 62 and 02 expulsions,


respectively, per 1000 population.2
Since 2005, so as not to create an
open invitation for illegal immigration,
these people have been denied health
coverage and, with the exception of
rare emergency situations, access to
free health care. Yet, a study that we
did among a representative sample
of the islands inhabitants3 found that
the proportion of foreign migrants
who had emigrated for health reasons
(88%) was low compared with the
proportion of those who had done
so for other reasons (the main one
being economic: 494%). A similar
study in French Guiana,4 bordering
Brazil and Surinam and, like Mayotte,
a destination of signicant migratory
movements, shows that despite more
favourable legislation (people without
regular residence status can theoretically obtain medical coverage under
the same conditions as in mainland
France), the rate of immigration for
health reasons is even lower (1%).
The fact that there is no public
medical assistance for people without
regular residence status in Mayotte
constitutes an impediment to the
right to health protection guaranteed
by the French constitution and several
international treaties ratied by
France. This situation was condemned
by an independent French public
authority, the Haute Autorit de
Lutte contre les Discriminations
(HALDE), on March 1, 2010,5 after
pressure from several organisations,
but so far to no avail. It has also
asked the government to make social
security available to unaccompanied
minors and children of persons
without regular residence status. We,
too, urge the French national and
local authorities to distinguishin
overseas territories particularly
health and health-care policies from
immigration policy.

Science Photo Library

We declare that we have no conicts of interest.

Anne Jolivet, Sophie Florence,


Jacques Lebas, *Pierre Chauvin
chauvin@u707.jussieu.fr

1828

*Inserm, U707, Research Team on Social


Determinants of Health and Health Care,
75012 Paris, France (AJ, SF, JL, PC); Universit
Pierre-et-Marie-Curie Paris-6, UMR S 707, Paris,
France (AJ, SF, JL, PC); and AP-HP, Hpital SaintAntoine, Paris, France (JL, PC)
1

Committee on the Elimination of Racial


Discrimination. France: concluding
observations. Geneva: Oce of the United
Nations High Commissioner for Human
Rights, 77th session, 2010, report CERD/C/
FRA/CO/17-19.
Rpublique Francaise, Prfecture de Mayotte.
Rsultats de la lutte contre limmigration
clandestine et la travail illgal: premiere
semestre, 2010. Dzaoudzi: Prfecture de
Mayotte, 2010.
Florence S, Lebas J, Parizot I, et al. Migration,
health and access to care in Mayotte Island in
2007: lessons learned from a representative
survey. Rev Epidemiol Sante Publique 2010;
58: 23744.
Jolivet A, Cadot E, Carde E, et al. Migration,
sant et soins en Guyane. Paris: Agence
Franaise de Dveloppement, 2010 (in press).
HALDE. Dlibration relative aux conditions
daccs aux soins des trangers en situation
irrgulire et de leurs enfants ainsi que des
mineurs trangers isols, rsidant Mayotte.
Paris: Haute Autorit de Lutte contre les
Discriminations et pour lEgalit, 2010,
dlibration n2010-87 du 01/03/2010.

metastatic cancer would be much


more likely. Such an attitude can lead to
erroneous treatment decisions.
Physicians need to be aware of
this all too human tendency towards
compassionate optimism and aim to
correct it by means of a more balanced,
bias-free interpretation of the facts.
I declare that we have no conicts of interest.

Ami Schattner
amiMD@clalit.org.il
Department of Medicine, Kaplan Medical Center,
POB 1, Rehovot 76100, Israel
1

Eisenberg JM. Sociologic inuences on


decision-making by clinicians. Ann Intern Med
1979; 90: 95764.
Landon BE, Reschovsky J, Reed M,
Blumenthal D. Personal, organizational, and
market level inuences on physicians practice
patterns: results of a national survey of
primary care physicians. Med Care 2001;
39: 889905.
Logan RL, Scott PJ. Uncertainty in clinical
practice: implications for quality and costs of
health care. Lancet 1996; 347: 59598.
Nanda S, Prakash Bhatt S, Steinberg D,
Volk SA. Unusual cause of generalized
osteolytic vertebral lesions: a case report.
J Med Case Reports 2007; 1: 33.

Compassionate
optimism
Many non-clinical factors can interfere
with the pure, evidence-based process
of physicians decision making.1
They include diverse personal factors
such as physicians reactions to fear
of litigation and poor tolerance of
common clinical uncertainty, but also
overcondence and hidden nancial
incentives.2,3 A previously unreported
personal factor aecting objective
evidence-based decision making
by physicians might be termed
compassionate optimism.
Compassionate optimism might
occur when a physicians empathy and
compassion towards his or her patient
causes erroneous interpretation of
clinical data to favour a more benign, less
ominous diagnosis, contrary to the facts
and clinical probability. For example,
a clinician who strongly sympathises
with a young mother who has osteolytic spine lesions might adhere to a
false belief that they represent a treatable granulomatous disease,4 whereas

Department of Error
Alwan A, MacLean DR, Riley LM, et al. Monitoring
and surveillance of chronic non-communicable
diseases: progress and capacity in high-burden
countries. Lancet 2010; 376: 186168In the
Summary and line 7 of the Mortality section of
this Series (Nov 27), the proportion of deaths
occurring in people younger than 70 years
should have been 47%. This correction has
been made to the online version as of Nov 15,
2010, and to the printed version.
Fernald LCH, Gertler PJ, Neufeld LM. 10-year
eect of Oportunidades, Mexicos conditional
cash transfer programme, on child growth,
cognition, language, and behaviour: a
longitudinal follow-up study. Lancet 2009;
374: 19972005In this Article (Dec 12,
2009), the last sentence of the Findings
section in the Summary should have read: An
additional 18 months of the programme
before age 3 years for children aged 810 years
whose mothers had no education resulted in
improved child growth of about 15 cm
assessed as height-for-age Z score ( 023
[0023044] p=0029), independently of cash
received. This correction has been made to
the online version as of Nov 26, 2010.

www.thelancet.com Vol 376 November 27, 2010

Articles

Eect of rivastigmine as an adjunct to usual care with


haloperidol on duration of delirium and mortality in
critically ill patients: a multicentre, double-blind,
placebo-controlled randomised trial
Maarten M J van Eijk, Kit C B Roes, Marina L H Honing, Michael A Kuiper, Attila Karakus, Mathieu van der Jagt, Peter E Spronk, Willem A van Gool,
Roos C van der Mast, Jozef Kesecioglu, Arjen J C Slooter

Summary
Background Delirium is frequently diagnosed in critically ill patients and is associated with adverse outcome. Impaired
cholinergic neurotransmission seems to have an important role in the development of delirium. We aimed to establish
the eect of the cholinesterase inhibitor rivastigmine on the duration of delirium in critically ill patients.
Methods Patients (aged 18 years) who were diagnosed with delirium were enrolled from six intensive care units in
the Netherlands, and treated between November, 2008, and January, 2010. Patients were randomised (1:1 ratio) to
receive an increasing dose of rivastigmine or placebo, starting at 075 mL (15 mg rivastigmine) twice daily and
increasing in increments to 3 mL (6 mg rivastigmine) twice daily from day 10 onwards, as an adjunct to usual care
based on haloperidol. The trial pharmacist generated the randomisation sequence by computer, and consecutively
numbered bottles of the study drug according to this sequence to conceal allocation. The primary outcome was the
duration of delirium during hospital admission. Analysis was by intention to treat. Duration of delirium was censored
for patients who died or were discharged from hospital while delirious. Patients, medical sta, and investigators were
masked to treatment allocation. Members of the data safety and monitoring board (DSMB) were unmasked and did
interim analyses every 3 months. This trial is registered with ClinicalTrials.gov, number NCT00704301.
Findings Although a sample size of 440 patients was planned, after inclusion of 104 patients with delirium who were
eligible for the intention-to-treat analysis (n=54 on rivastigmine, n=50 on placebo), the DSMB recommended that the
trial be halted because mortality in the rivastigmine group (n=12, 22%) was higher than in the placebo group (n=4, 8%;
p=007). Median duration of delirium was longer in the rivastigmine group (50 days, IQR 27142) than in the
placebo group (30 days, IQR 1093; p=006).
Interpretation Rivastigmine did not decrease duration of delirium and might have increased mortality so we do not
recommend use of rivastigmine to treat delirium in critically ill patients.
Funding ZonMw, the Netherlands Brain Foundation, and Novartis.

Introduction
Delirium is an acute neuropsychiatric syndrome,
characterised by disturbances of attention and other
cognitive functions.1 Delirium is very common in patients
in intensive care units, with reported frequencies up
to 80%,2,3 and is associated with increased complications,
length of hospital stay, and mortality.35 Antipsychotics
and benzodiazepines have become the standard drug
treatment for delirium in critically ill patients,6 although
few placebo-controlled trials have been done for
antipsychotics7,8 and none for benzodiazepines.9
Furthermore, haloperidol might lead to extrapyramidal
side-eects and ventricular arrhythmia,10 whereas
benzodiazepines might even induce delirium.11
Several lines of evidence suggest that impaired
cholinergic neurotransmission has an important role in
the development of delirium.12 Serum anticholinergic
activity was found to be increased in patients with
delirium,13 and drugs with anticholinergic eects can
www.thelancet.com Vol 376 November 27, 2010

cause delirium, particularly in elderly patients.14 Persistent


delirium can be treated successfully with cholinesterase
inhibitors, such as rivastigmine, and in case series of
elderly patients, low-dose rivastigmine (ie, 36 mg daily)
was sucient to resolve symptoms of delirium, usually
within 4872 h.15,16 Rivastigmine is an inhibitor of
acetylcholinesterase and butyrylcholinesterase, and has
been approved for the symptomatic treatment of
Alzheimers disease, Parkinsons disease dementia, and
Lewy-body dementia.17 On the basis of these considerations,
we postulated that rivastigmine added to usual care would
reduce the duration of delirium in critically ill patients.

Methods
Patients

Lancet 2010; 376: 182937


Published Online
November 5, 2010
DOI:10.1016/S01406736(10)61855-7
See Comment page 1805
Department of Intensive Care
Medicine (M M J van Eijk MD,
Prof J Kesecioglu MD,
A J C Slooter MD), and Julius
Centre for Health Sciences and
Primary Care
(Prof K C B Roes PhD), University
Medical Centre, Utrecht,
Netherlands; Department of
Intensive Care, Medical Centre
Alkmaar, Alkmaar, Netherlands
(M L H Honing MD);
Department of Intensive Care,
Medical Centre Leeuwarden,
Leeuwarden, Netherlands
(M A Kuiper MD); Department
of Intensive Care,
Diakonessenhuis Utrecht,
Utrecht, Netherlands
(A Karakus MD); Department of
Intensive Care, Erasmus MC
University Medical Centre,
Rotterdam, Netherlands
(M van der Jagt MD);
Department of Intensive Care,
Gelre Hospitals (Lukas site),
Apeldoorn, Netherlands
(P E Spronk MD); Department of
Neurology, Academic Medical
Centre, Amsterdam,
Netherlands
(Prof W A van Gool MD); and
Department of Psychiatry,
Leiden University Medical
Centre, Leiden, Netherlands
(Prof R C van der Mast MD)
Correspondence to:
Dr Arjen J C Slooter, Room
F06.149, Department of
Intensive Care Medicine,
University Medical Centre,
Heidelberglaan 100, 3584 CX,
Utrecht, Netherlands
a.slooter-3@umcutrecht.nl

Patients were enrolled from six hospitals in the


Netherlands and treated between November, 2008, and
January, 2010. In the six participating intensive care
units, patients were assessed daily by trained nurses with
1829

Articles

For the study protocol see


http://www.umcutrecht.nl/
subsite/intensive-care/Verblijf/
Wetenschappelijk-onderzoek

the confusion assessment method for the intensive care


unit (CAM-ICU)18 to identify occurrence of delirium,
which was conrmed daily by research nurses. If the
treating physician in the intensive care unit had any
doubts about the delirium diagnosis, a psychiatrist,
geriatrician, or neurologist was consulted to conrm the
diagnosis before the patient was included in the study.
This diagnosis could overrule a negative score on the
CAM-ICU for patients with clinical signs indicating
delirium. Only the rst event of delirium was investigated;
if patients did not meet eligibility criteria on the rst
event of delirium, they could not be assessed for inclusion
again on a subsequent event of delirium.
Patients were eligible for inclusion in the study if they
were aged 18 years or older, had been admitted to the
intensive care unit, scored positive on the CAM-ICU, and
were expected to remain in the intensive care unit for at

6724 patients

6076 did not have delirium

648 had delirium

Randomisation and masking


539 did not meet eligibility criteria
146 diagnosis uncertain
141 no informed consent
65 renal replacement therapy
22 hepatic encephalopathy
15 unable to receive enteric drugs
11 bradycardia with haemodynamic
consequences
139 other reasons
31 could not speak Dutch or English
22 expected to be in intensive care
unit for <48 h
79 logistical problems
7 not specied

109 enrolled and randomised

55 allocated to receive rivastigmine


1 withdrawn by family

least 48 h. Patients were ineligible if they met any of the


following criteria: pregnant or lactating; unable to receive
enteric drugs; receiving renal replacement therapy; liver
failure with hepatic encephalopathy; known allergy to
rivastigmine; or second or third degree atrioventricular
block or bradycardia with haemodynamic consequences,
without a functioning pacemaker. On the basis of
previous reports on the use of rivastigmine with
Alzheimers disease, Parkinsons disease dementia, and
Lewy-body dementia,19,20 we had no reason to assume that
rivastigmine would have a clinically signicant interaction
with other drugs, so patients were allowed to continue on
existing drug regimens.
The study was done in accordance with Good Clinical
Practice guidelines of the International Conference
on Harmonisation of Technical Requirements for
Registration of Pharmaceuticals for Human Use, the
Declaration of Helsinki, and local laws and regulations.
The study protocol was approved by the ethics committees
of all participating centres. All legal representatives of
participating patients gave written informed consent.
Patients who recovered from delirium could withdraw
consent during the follow-up interview.

54 allocated to receive placebo

Patients were randomised (1:1 ratio) to receive either


placebo or rivastigmine as adjunct to existing treatment.
The randomisation sequence was computer generated by
the trial pharmacist, and was stratied by study centre.
All centres received batches of ten identical bottles, ve
of which contained a solution of the study drug and ve
contained a placebo solution. The solutions had identical
colour, smell, taste, and viscosity. The trial pharmacist
consecutively numbered the bottles according to the
randomisation sequence to conceal allocation of the next
patient in the sequence. Once eligibility of patients was
conrmed, the investigator used bottles of the study drug
consecutively to mask every patient and their family,
medical sta, and investigators from treatment allocation.
The leading pharmacist held a list of study codes, which
could be broken at any time if deemed necessary by the
treating physician.

4 withdrawn by family

Procedures
54 included in intention-to-treat analysis

12 died

42 reached endpoint of end of delirium or


discharge from hospital

6 died

36 completed 90 days of follow-up

Figure 1: Trial prole

1830

50 included in intention-to-treat analysis

4 died

46 reached endpoint of end of delirium or


discharge from hospital

7 died

39 completed 90 days of follow-up

Legal representatives of patients completed a questionnaire on premorbid cognitive functioning,21 comorbidity


(Charlson comorbidity index22), alcohol use (more than
15 glasses of alcohol per week in men or more than ten
glasses of alcohol per week in women; one standard glass
contains 99 g ethanol), tobacco use (any smoking directly
before hospital admission, irrespective of frequency),
recreational drug use, visual impairment (use of glasses
or presence of blindness), and hearing impairment
(use of hearing aids or presence of deafness). The
questionnaire was completed after informed consent was
obtained but before the rst dose of study drug was given
to the patient. Psychiatric diagnosis (any past psychiatric
www.thelancet.com Vol 376 November 27, 2010

Articles

diagnosis, or admittance to a psychiatric ward or hospital


before inclusion in the study) was registered from the
questionnaire and from the patients medical record. The
acute physiology and chronic health evaluation
(APACHE) II score23 and drug use before admission
to the intensive care unit (at home or in a dierent
hospital department) were registered from the patients
medical record.
Placebo and rivastigmine were packaged in 120 mL
brown glass bottles. Placebo was manufactured according
to international good manufacture practice by the trial
pharmacy of University Medical Centre Utrecht,
Netherlands. Rivastigmine was manufactured by Novartis
(Huningue, France) and delivered in a 2 mg/mL solution.
Patients received the study drug (rivastigmine or placebo),
in doses of 075 mL (15 mg rivastigmine) twice daily
on days 13, 15 mL (3 mg rivastigmine) twice daily on
days 46, 225 mL (45 mg rivastigmine) twice daily on
days 79, and 3 mL (6 mg rivastigmine) twice daily
thereafter. The dosing schedule of the study drug diered
from the regimen used for patients with Alzheimers
disease. Our dosing schedule was based on that used in a
pilot study by Oldenbeuving and colleagues,16 but we
used a more cautious approach, with incremental
increases every 3 days instead of every 2 days, and the
same starting, incremental, and maximum dose. This
regime was tolerated well when rivastigmine was
prescribed o-label before the onset of our trial.16
Transdermal application of rivastigmine could have been
an advantage, but was not chosen because skin perfusion
might not be sucient in critically ill patients.
Once patients had reached the end of delirium or were
discharged from hospital, the dose regimen was reversed
and the study drug was tapered o within 3 days. If a
possible side-eect occurred during treatment, the study
drug was reduced until the side-eect had resolved or
was stopped if the side-eect persisted for longer than
3 days. Patients with persistent side-eects were followed
up until an endpoint was reached (end of delirium,
discharge from hospital, or death). In the occurrence of a
serious adverse event, the study drug was discontinued
and the lead physician (AJCS) was notied.
During treatment with the study drug and for 3 days
thereafter, patients were assessed daily for 7 days a week
by a trial nurse or physician with the Richmond agitation
sedation scale (RASS),24 the CAM-ICU to detect delirium
during stay in the intensive care unit, the CAM25 to detect
delirium during stay in a ward with a lower level of care
than in the intensive care unit, and the delirium severity
index (DSI).26 We also noted signs of delirium daily, as
per the patients medical record. When the medical
record suggested delirium, or in doubtful cases, a
psychiatrist, geriatrician, or neurologist was consulted,
whose conclusion could overrule the CAM-ICU. Possible
side-eects of the study drug, serious adverse events,
other drug use, physical restraints, self-removed
catheters, and sequential organ failure assessment
www.thelancet.com Vol 376 November 27, 2010

Rivastigmine (n=54)

Placebo (n=50)

Age (years)

680 (114)

700 (122)

Male sex

38 (70%)

29 (58%)

APACHE II score

203 (89)

196 (79)

SOFA score

56 (23)

55 (31)

Charlson comorbidity index

26 (23)

23 (23)

RASS score
+2

12 (22%)

9 (18%)

+1

20 (37%)

14 (28%)

3 (6%)

7 (14%)

10 (19%)

11 (22%)

9 (17%)

9 (18%)

Admitting specialty
Internal medicine

17 (31%)

15 (30%)

General surgery

21 (39%)

16 (32%)

Cardiology or cardiothoracic surgery

13 (24%)

18 (36%)

Neurology or neurosurgery
Emergency admission to intensive care unit

3 (6%)

1 (2%)

46 (85%)

32 (64%)

33 (61%)

36 (72%)

Sensory handicap
Visual
Auditory

7 (13%)

12 (24%)

IQCODE score

515 (63)

529 (59)

Alcohol and tobacco use


Alcohol

11 (20%)

6 (12%)

Smoking

15 (28%)

10 (20%)

5 (9%)

5 (10%)

Benzodiazepines

7 (13%)

6 (12%)

Antipsychotics

6 (11%)

3 (6%)

Analgesics

5 (9%)

Psychiatric diagnosis
Drug use before admission to intensive care unit

Estimated duration of delirium before inclusion (h)*

120 (58360)

120 (65225)

Data are mean (SD), number (%), or median (IQR). APACHE=acute physiology and chronic health evaluation.
SOFA=sequential organ failure assessment. RASS=Richmond agitation sedation scale. IQCODE=informant
questionnaire on cognitive decline in the elderly. *Data were missing for three patients on rivastigmine.

Table 1: Demographic and clinical characteristics of patients at baseline

(SOFA) scores were recorded daily.27 A trial nurse


interviewed all surviving patients by telephone at 90 days
after inclusion in the study.
All included patients received usual care, which included
frequent orientation, physical therapy, and exercise.
Patients aged 70 years or older received 1 mg haloperidol
intravenously three times a day, and those aged 69 years or
younger received 25 mg haloperidol intravenously three
times a day. Patients were also allowed to receive 1 mg
lorazepam intravenously at night (2200 h). The treating
physician could adjust treatment with haloperidol or a
benzodiazepine, similar to usual care. In case of persistent
severe agitation, rescue haloperidol (25 mg in patients
aged 70 years or older, and 5 mg in those aged 69 years or
younger) was recommended, to be repeated every 30 min
if needed. If haloperidol was not eective and patients
were agitated, they were sedated with propofol at
1 mg/kg per h intravenously; the dose was raised until the
1831

Articles

Rivastigmine (n=54)

Placebo (n=50)

p value

Primary outcome
Delirium duration (days)

50 (27142)

30 (1093)

006

Endpoint of end of delirium (n=35 vs n=34)

40 (20160)

25 (1058)

006

Endpoint of hospital discharge (n=7 vs n=12)

60 (35115)

60 (30215)

095

Endpoint of death (n=12 vs n=4)

95 (48118)

80 (1090)

029

109 (51327)

Secondary outcomes
Total dose of study drug (mL)

64 (45225)

012

Fixation days

10% (0043)

10% (0023)

090

Number of self-removed catheters

0 (01)

0 (01)

067

Median of mean DSI scores for individual patients

23 (2031)

20 (1825)

00040
030

Median of mean use of psychotropic drugs for individual patients (mg per day)
Haloperidol

32 (2656)

31 (1853)

Lorazepam

02 (0006)

00 (0005)

028

Propofol

00 (001043)

00 (00237)

022

Number of study days


Comatose (RASS score 4 or 5)
Non-comatose (RASS score 3 or higher)

69/659 (10%)

16/459 (3%)

<00001

590/659 (90%)

443/459 (97%)

<00001

Length of stay (days)


Intensive care unit

15 (930)

8 (317)

<00001

Hospital

29 (1755)

25 (1739)

006

Mortality
During treatment with study drug

12 (22%)

4 (8%)

007

At 90 days of follow-up

18 (33%)

11 (22%)

014

Data are median (IQR) or number (%), unless otherwise indicated. DSI=delirium severity index. RASS=Richmond
agitation and sedation scale.

Table 2: Primary and secondary outcomes

patient was calm, but was tapered every 12 h. If propofol


was contraindicated, because of haemodynamic instability
for example, the patient was sedated with midazolam at
5 mg per h intravenously, the dose was raised until the
patient was calm, and the dose was also tapered every 12 h.
During sedation, the study drug was continued.
The primary outcome was the duration of delirium
during hospital admission (ie, in the intensive care unit
and in the hospital ward combined). Secondary outcomes
were the percentage of xation days (proportion of study
days for which the patient was restrained by their arms
or legs, or both), number of self-removed catheters
(bladder catheters, arterial catheters, intravenous
catheters, or ventilation tubes), severity of delirium,
psychoactive drug use, and length of stay in the intensive
care unit or hospital. Other secondary outcomes were
total dose of study drug, number of study days, and
mortality during treatment and at 90 days of follow-up.
Members of the data safety and monitoring board
(DSMB) were not masked to treatment allocation and
had access to all data on the primary outcome and all instudy serious adverse events. DSMB members were
independent and had no conict of interest with
investigators, the sponsor, or the manufacturer of the
study drug. According to a predened plan, interim
analyses were done every third month of enrolment. The
1832

dierence in occurrence of serious adverse events


between the two intervention groups was sequentially
monitored by the DSMB against an overall signicance
level of 10% to indicate when stopping of the trial should
be considered. Monitoring was done in accordance with
Good Clinical Practice regulations.
In a sample of ten consecutive patients who met eligibility
criteria but were not included in the trial, we recorded a
mean of seven serious adverse events per patient per week,
and more than double that amount of adverse events.
Reporting of all adverse events according to denitions of
the US Food and Drug Administration would lead to a
large number of adverse events, with the possibility of
under-reporting. Therefore, in consultation with the central
medical ethics committee and the DSMB, only serious
adverse events resulting in death or clinically signicant
injury were reported and used in interim analyses.

Statistical analysis
On the basis of an average duration of delirium in
patients in intensive care of 34 days (SD 19),28 at 80%
power with =005, 220 patients per group and
440 patients in total were needed to detect a clinically
signicant reduction in the average duration of delirium
of 15% or half a day.
The primary outcome was compared across the two
study groups in an intention-to-treat analysis with a
Mann-Whitney U test and Cox regression analysis.
Duration of delirium was censored for patients who died
or were discharged from hospital while delirious to avoid
underestimation of the duration of delirium in these
patients. We used Cox regression analysis because it is
more appropriate for analysis of censored data than is
linear regression analysis of logarithmically transformed
data. Hazard ratios (HRs) were adjusted for sex and
emergency admissions to the intensive care unit at
baseline. The primary outcome was also analysed in
subgroups of patients meeting three endpoints: end of
delirium (CAM-ICU negative for 48 h), discharge from
hospital, or death during treatment with the study drug.
Mortality rates were compared between the two groups
with a Kaplan-Meier curve and log-rank statistics.
Secondary outcomes were compared across the two study
groups in an intention-to-treat analysis with Mann-Whitney
U tests. Delirium severity was assessed as the cumulative
DSI score in each patient divided by number of days during
which the patient had delirium. Disease severity was
estimated as the cumulative SOFA score in each patient
divided by the number of study days. We calculated p values
with Mann-Whitney U tests or tests as appropriate.
After interruption of the study, a post-hoc exploratory
analysis of several clinical indicators was done to identify
an underlying biological mechanism. To this end, standard
haemodynamic indicators (heart rate, arterial blood
pressure, and urine production), which were used as
surrogate variables that could represent cardiac output,
and indicators of infections (number of leucocytes and
www.thelancet.com Vol 376 November 27, 2010

Articles

Duration
of study
(days)

Dose of
study drug
(mL)*

Intervention Short case history

Autopsy

10

Rivastigmine

Type A aortic dissection followed by asystole due to recurrent heart tamponade


for which several thoracotomies were done; long stay in intensive care unit;
died of heart failure due to several myocardial infarctions

Yes

22

15

34

075

Rivastigmine

Acute pancreatitis, several events of septic shock, and abdominal surgery;


died of multiorgan failure, including disseminated intravascular coagulation
and intestinal ischaemia

Yes

70

30

13

075

Rivastigmine

Chronic renal failure and abdominal aortic aneurysm for which endograft
placement was done, but became infected with Staphylococcus aureus,
resulting in septic shock; persistent endograft leakage resulting in death

Yes

Female

69

16

12

Placebo

Abdominal aortic endograft with rupture of anastomotic aneurysm;


postoperative septic shock due to intestinal ischaemia; died of rupture of
recurrent aneurysm

Yes

Patient 5

Male

77

19

10

225

Rivastigmine

Amputation because of severe peripheral arterial disease; respiratory failure


and heart failure; died of pneumonia

No

Patient 6

Male

78

26

15

Rivastigmine

Colorectal carcinoma with peritoneal carcinomatosis, ileus, and abdominal


sepsis; died of septic shock following bowel perforation during surgery

No

Patient 7

Female

59

20

14

075

Rivastigmine

Subarachnoid haemorrhage followed by hydrocephalus (ventricular drain


and secondary meningitis); died of respiratory failure due to acute respiratory
distress syndrome

Yes

Patient 8

Male

82

19

18

Placebo

Coronary artery bypass graft and aortic valve replacement, anticoagulation,


recurrent rectal bleeding; cardiac arrest followed by renal failure and decubitus
ulcers; died of recurrent sepsis

No

Patient 9

Male

74

11

13

11

Rivastigmine

Aortic valve replacement complicated by persistent bleeding and recurrent


thoracotomy; died of multiorgan failure

No

Patient 10 Male

88

26

11

075

Placebo

Gastric ulceration complicated by pneumonia; died of severe sepsis with acute


kidney failure

No

Patient 11 Male

57

40

12

075

Rivastigmine

Multiple myeloma complicated by pneumonia, atrial brillation, and heart


failure; no life sustaining treatment was started because of unfavourable
prognosis; died of multiorgan failure

No

Patient 12 Male

55

23

16

15

Placebo

Chemotherapy for acute myeloid leukaemia; admitted to intensive care unit


for pneumonia resistant to antibiotic treatment; died of multiorgan failure

No

Patient 13 Female

75

17

075

Rivastigmine

Pancreatoduodenectomy for pancreatic carcinoma; postoperative acute septic


shock based on anastomosis dehiscence; died of septic shock

Yes

Patient 14 Female

83

12

13

225

Rivastigmine

Ovarian carcinoma for which surgical debulking was done; postoperative


anastomotic leakage from bowel with sepsis for which several abdominal
surgeries were done; died of recurrent septic events and multiorgan failure

No

Patient 15 Female

62

38

18

15

Rivastigmine

Ileus; several abdominal operations complicated by ischaemic colon and sepsis;


long stay in intensive care unit with recurrent septic events; died of
multiorgan failure

No

Patient 16 Female

80

15

Rivastigmine

Hemicolectomy for colon carcinoma; postoperative hypovolaemic shock;


subsequent operation led to septic shock and multiorgan failure, resulting
in death

No

Sex

Age
(years)

APACHE II
score

Patient 1

Female

66

26

Patient 2

Male

74

Patient 3

Male

Patient 4

SOFA
score*

APACHE=acute physiology and chronic health evaluation. SOFA=sequential organ failure assessment. *At 24 h before death.

Table 3: Case series of patients who died during treatment

C-reactive protein concentrations) were analysed by use of


linear regression and mixed model analysis for repeated
measurements in a large part of the study population
(n=65). As the objective was to explore whether any of
these indicators diered systematically between the two
study groups, no corrections for multiplicity were applied.
This trial is registered with ClinicalTrials.gov, number
NCT00704301.

to submit for publication. Novartis supplied the study


drug, information about the study drug, and empty bottles
for placebo, but had no role in the decision to do the study
or to stop the study early, study design, data collection,
data analysis, data interpretation, writing of the report, or
the decision to submit for publication. The lead authors
(MMJvE, KCBR, and AJCS) had full access to all data in
the study, and the corresponding author (AJCS) had nal
responsibility for the decision to submit for publication.

Role of the funding source


ZonMw and the Netherlands Brain Foundation did not
contribute to study design, data collection, data analysis,
data interpretation, writing of the report, or the decision
www.thelancet.com Vol 376 November 27, 2010

Results
The DSMB recommended that the trial be halted after
the fourth interim analysis and inclusion of 109 patients
1833

Articles

100

Placebo
Rivastigmine

Survival (%)

80

60

40

20

0
0
See Online for webappendix

Patients at risk
Placebo 50
Rivastigmine 54

30
60
Time (days after inclusion)
44
40

41
37

90
39
36

Figure 2: Kaplan-Meier survival curve

(gure 1). Five patients were withdrawn from the study


by their families (one on rivastigmine and four on
placebo), leading to a modied intention-to-treat analysis
of 54 patients on rivastigmine and 50 on placebo. Data
were censored for 16 patients who died and 19 patients
who were discharged from hospital while still delirious.
Mortality during treatment with the study drug seemed
to be higher in the rivastigmine group (n=12, 22%) than
in the placebo group (n=4, 8%; p=007, based on
sequential testing). This result met predened criteria to
consider stopping the trial, and, according to the DSMBs
recommendation, the study was immediately halted.
Seven centres were expected to participate in the trial as
per the protocol, but recruitment at the seventh centre
had not yet begun when the trial was stopped prematurely,
so data are supplied for six centres only.
Demographic and clinical characteristics recorded at
baseline, including age, APACHE II score, SOFA score,
and Charlson comorbidity index, were well balanced
across treatment groups, although the rivastigmine
group had a slightly higher proportion of men and
more emergency admissions than did the placebo
group (table 1). Post-hoc analyses showed that the mean
SOFA score was similar for the rivastigmine group
(66, SD 29) and the placebo group (63, SD 28) on
the 5 days before inclusion.
Duration of delirium was longer in the rivastigmine
group than in the placebo group, but the dierence
between the groups did not reach signicance (table 2).
Furthermore, delirium was of signicantly higher
severity in the rivastigmine group than in the placebo
group (table 2), despite similar RASS scores at baseline
(table 1) and thus similar DSI scores. Compared with
patients in the placebo group, those in the rivastigmine group, on average, received more haloperidol,
lorazepam, and propofol, stayed in the intensive care
unit for signicantly longer, and were signicantly
longer in coma (table 2). The HR for delirium duration
associated with use of rivastigmine was 072 (95% CI
1834

044117), and did not change substantially after


adjustment (077, 047126). In a post-hoc analysis,
our ndings did not dier substantially when data were
censored for discharge from hospital but not death, with
an HR of 080 (95% CI 051125; 084, 053132
after adjustment).
Analysis of mean SOFA scores recorded daily during
treatment with the study drug showed no dierence
between the rivastigmine group (53, SD 37) and the
placebo group (45, SD 50). Furthermore, post-hoc
exploratory analyses of estimates of cardiac output and
infection did not show dierences between the two
groups (webappendix pp 12).
Table 3 presents a short case history for each of the
patients who died during treatment. None of the
104 patients included in the intention-to-treat analysis
was lost to follow-up up to 90 days after inclusion. At
90 days follow-up, mortality was higher in the
rivastigmine group than in the placebo group (table 2;
gure 2; log-rank test p=023). The HR for mortality at
90 days follow-up associated with use of rivastigmine
was 158 (95% CI 074337), and did not change
substantially after adjustment (180, 080407).
Mortality was evenly balanced between participating
centres. Protocol-specied analyses, such as follow-up
analyses of cognitive functioning, activities of daily living,
and costs of care were not done because the trial ended
early so the sample size was too small.

Discussion
We have shown that in critically ill patients, rivastigmine
did not decrease duration of delirium when it was added
to standard treatment with haloperidol (panel).
Furthermore, rivastigmine was associated with a more
severe type of delirium, longer stay in the intensive care
unit, and higher mortality than was placebo, although the
dierence in mortality between treatment groups was not
signicant. The DSMB advised that the trial should be
halted prematurely because of increased mortality in the
rivastigmine group compared with the placebo group.
Other important arguments were the fact that
demographic and clinical characteristics of the treatment
groups were well balanced at baseline, and rivastigmine
was not associated with any benecial eects. The
investigators had no counter arguments to the advice of
the DSMB so the trial was halted prematurely.
This large randomised placebo-controlled trial of a
cholinesterase inhibitor is one of the very few randomised
controlled trials of treatment for delirium in critically ill
patients.7,8,29,31 This trial benets from the multicentre
design, which contributes to generalisability of our
ndings. Although we excluded 141 patients with no
informed consent and 139 for other reasons before
randomisation, the other reasons included inability to
speak Dutch or English, expected stay in the intensive
care unit of less than 48 h, and logistical problems.
Therefore, we believe that our study population can be
www.thelancet.com Vol 376 November 27, 2010

Articles

generalised to the population of patients with delirium in


the intensive care unit. Other strengths of the study
include the strict and complete follow-up during the
study (7 days a week), which ensured that patients were
safe and data of good quality was retrieved, and the
interim analyses that were done every 3 months by an
unmasked and independent DSMB to ensure the safety
of patients.
The study population was not homogeneous because
patients with all types of disease who were in intensive
care could be included. Thus, we assumed that impaired
cholinergic neurotransmission was the nal common
pathway for delirium in all critically ill patients. However,
assessment of reduced cholinergic neurotransmission is
dicult while patients are alive, and, in clinical practice,
various critically ill patients receive drugs to treat
delirium, including haloperidol, irrespective of the
underlying cause of delirium.
A potential concern is the method of delirium
assessment. Preferably, the diagnosis of delirium should
be made by a psychiatrist, neurologist, or geriatrician,
according to the diagnostic and statistical manual of
mental disorders, fourth edition.1 Although a psychiatrist,
neurologist, or geriatrician was consulted if the diagnosis
of delirium was in doubt, patients were not routinely
diagnosed by this method. However, we do not believe
that our decision not to use this method will have led to
frequent misdiagnoses because the trial nurses and
physicians used the CAM-ICU, which has good positive
(range 73100%) and negative (8398%) predictive
values.18,30 The CAM was used after discharge from the
intensive care unit and also has excellent positive (94%)
and negative (100%) predictive values.25 Another concern
is use of the DSI to measure severity of delirium because
this instrument has been used in only one study26
previously and has not been externally validated.
Rivastigmine was given as an adjunct to haloperidol.
Comparison of an antipsychotic with a cholinesterase
inhibitor is not feasible because the ethics committee for
our study believed that an investigation in which patients
do not receive usual care would contradict national and
international ethics guidelines. Adjustments to the
doses of haloperidol were left to the discretion of the
treating physicians in accordance with clinical practice;
no sound evidence is available on the duration and
dose of haloperidol in critically ill patients. Pharmacological databases, such as the Summary of Product
Characteristics, report no possible interactions between
rivastigmine and haloperidol.32
Our ndings could have alternative explanations. First,
the causes of death during treatment with the study drug
were heterogeneous, so the patients who were randomised
to receive rivastigmine might have been sicker than were
those in the placebo group, resulting in increased
duration of delirium and mortality. However, this
explanation seems unlikely since we did not identify
dierences in age, comorbidities, or disease severity
www.thelancet.com Vol 376 November 27, 2010

Panel: Research in context


Systematic review
We searched Medline, Embase, the Cochrane Library, and
ClinicalTrials.gov with keywords for intensive care unit,
delirium, and cholinesterase inhibitor. All articles identied
were screened for inclusion by two authors (MMJvE and AJCS)
and assessed for quality by use of levels of evidence from
University of Oxford Centre for Evidence-Based Medicine.
Seven case reports, two reviews, two observational studies,
and three trials were identied. All three trials2830 were
underpowered, and the ndings showed some evidence of
decreased duration of delirium with a cholinesterase
inhibitor, but, importantly, no safety concerns.

For more on University of


Oxford Centre for EvidenceBased Medicine see http://
www.cebm.net/?o=1025

Interpretation
In our trial, the cholinesterase inhibitor rivastigmine had no
benecial eect for treatment of delirium in critically ill
patients, and might have increased mortality. These results,
combined with the ndings of previous studies, do not
support the use of cholinesterase inhibitors to treat delirium
in critically ill patients.

between the treatment groups at inclusion. Second,


patients in the rivastigmine group might have received
higher doses of benzodiazepines, on average, than did
those in the placebo group, resulting in increased
frequency of coma and thereby longer duration of
delirium,11 longer stay in the intensive care unit,35 and
increased mortality. Such a mechanism was assessed
with Cox regression analyses on duration of delirium and
on mortality. After adjustments for the dose of lorazepam
or propofol, rivastigmine use was still associated with
increased duration of delirium and mortality (data not
shown). Therefore, our ndings cannot be explained by a
coincidental dierence in the dose of benzodiazepines.
To explain the higher mortality rate in the rivastigmine
group than in the placebo group, we did a further posthoc analysis based on reported side-eects of rivastigmine
that could lead to death because of reduced cardiac
output, such as bradycardia. Additionally, we explored
indicators of infection, which is a frequent cause of death
in critically ill patients. Findings of these post-hoc
analyses did not show dierences between the two
groups, and did not support a detrimental eect of
rivastigmine. Therefore, we cannot exclude that the
recorded dierence in mortality is due to chance. The
most important nding, however, is the absence of
evidence for any benecial eects of rivastigmine on the
primary or secondary outcomes. Median duration of
delirium was 2 days longer in the rivastigmine group
than in the placebo group, and this dierence was close
to signicance.
Cholinesterase inhibitors are mentioned as possibly
benecial in guidelines on deliriumeg, the American
Psychiatric Associations guidelines6and are prescribed
in many centres to patients in whom delirium is
1835

Articles

refractory to antipsychotics. The o-label use of


cholinesterase inhibitors is supported by case series,15,16
but not by trials of cholinesterase inhibitors for
prophylaxis or treatment of delirium, although these
trials might have been underpowered or used an
insucient dose.33,34 Since ndings of our study show
that rivastigmine does not decrease duration of delirium
and might increase mortality in critically ill patients, we
do not recommend treatment of delirium with
rivastigmine for patients in intensive care.

Contributors
MMJvE and AJCS conceived and designed the study with input from
MAK, WAvG, RCvdM, and JK. MMJvE, MLHH, MAK, AK, MvdJ, PES,
JK, and AJCS undertook the trial and collected data. KCBR supervised
all statistical analyses and was responsible for the post-hoc data
analysis. MMJvE, KCBR, and AJCS interpreted the data. MMJvE and
AJCS drafted the report with input and editing from all authors.

Participating centres
Department of Intensive Care Medicine, University Medical Centre,
Utrecht, Netherlands (MMJvE [coordinator], JK [director], AJCS
[principal investigator], n=60 patients included in analysis); Department
of Intensive Care, Medical Centre Alkmaar, Alkmaar, Netherlands
(MLHH [investigator], n=11); Department of Intensive Care, Medical
Centre Leeuwarden, Leeuwarden, Netherlands (MAK [investigator],
n=11); Department of Intensive Care, Diakonessenhuis Utrecht,
Utrecht, Netherlands (AK [investigator], n=8); Department of Intensive
Care, Erasmus MC-University Medical Centre, Rotterdam, Netherlands
(MvdJ [investigator], n=8); and Department of Intensive Care, Gelre
Hospitals (Lukas site), Apeldoorn, Netherlands (PES [investigator], n=6).

DSMB members
R J van Marum (Department of Geriatrics, Jeroen Bosch Ziekenhuis,
Den Bosch, Netherlands [chair]); I van der Tweel (Department of
Biostatistics, Julius Centre for Health Sciences and Primary Care,
University Medical Centre, Utrecht, Netherlands); P Eikelenboom
(Department of Neurology, Academic Medical Centre, and Department
of Psychiatry, VU Medical Centre, Amsterdam, Netherlands);
J M A Sitsen (Utrecht Institute for Pharmaceutical Sciences, Utrecht
University, Utrecht, Netherlands); E de Jonge (Department of Intensive
Care Medicine, Leiden University Medical Centre, Leiden, Netherlands);
and D Zandstra (Department of Intensive Care Medicine, Onze Lieve
Vrouwe Gasthuis, Amsterdam, Netherlands).

14

Conicts of interest
RCvdM has received payment and travel expenses for presentations and
attendance at committee meetings from the following professional
organisations: the Netherlands Society of Psychiatry, the Netherlands
Society of Intensive Care, the Health Care Inspectorate, and the
Academic Medical Centre, University of Amsterdam. All other authors
declare that they have no conicts of interest.
Acknowledgments
We thank all patients and their legal representatives for participation in
this trial; all research nurses in participating centres; medical and nursing
sta in participating centres who cared for patients and collected data;
data managers who participated in this trial, especially Frank Leus and
Joost Schotsman; the pharmacists in all participating centres for their
help; Michel Bots and Ardine de Wit for their contribution to the design
of the trial; and DSMB members. This study was funded by ZonMw, the
Netherlands organisation for health research and development (grant
number 80-82305-98-08109), the Netherlands Brain Foundation (grant
number 2008(1).30), and Novartis, Basel, Switzerland (unrestricted grant).
References
1
American Psychiatric Association. Diagnostic and statistical manual
of mental disorders, 4th edn. Washington, DC: American
Psychiatric Association, 2000.
2
Ouimet S, Kavanagh BP, Gottfried SB, Skrobik Y. Incidence, risk
factors and consequences of ICU delirium. Intensive Care Med 2007;
33: 6673.

1836

10
11

12

13

15
16

17

18

19

20

21
22

23

24

25

Ely EW, Shintani A, Truman B, et al. Delirium as a predictor of


mortality in mechanically ventilated patients in the intensive care
unit. JAMA 2004; 291: 175362.
Thomason JW, Shintani A, Peterson JF, Pun BT, Jackson JC,
Ely EW. Intensive care unit delirium is an independent predictor
of longer hospital stay: a prospective analysis of 261 non-ventilated
patients. Crit Care 2005; 9: R37581.
Lin SM, Liu CY, Wang CH, et al. The impact of delirium on the
survival of mechanically ventilated patients. Crit Care Med 2004;
32: 225459.
Practice guideline for the treatment of patients with delirium.
American Psychiatric Association. Am J Psychiatry 1999;
156 (suppl): 120.
Devlin JW, Roberts RJ, Fong JJ, et al. Ecacy and safety of
quetiapine in critically ill patients with delirium: a prospective,
multicenter, randomized, double-blind, placebo-controlled pilot
study. Crit Care Med 2010; 38: 41927.
Girard TD, Pandharipande PP, Carson SS, et al. Feasibility, ecacy,
and safety of antipsychotics for intensive care unit delirium: the
MIND randomized, placebo-controlled trial. Crit Care Med 2010;
38: 42837.
Lonergan E, Luxenberg J, Areosa Sastre A. Benzodiazepines for
delirium. Cochrane Database Syst Rev 2009; 4: CD006379.
Settle EC Jr, Ayd FJ Jr. Haloperidol: a quarter century of experience.
J Clin Psychiatry 1983; 44: 44048.
Pandharipande P, Shintani A, Peterson J, et al. Lorazepam is an
independent risk factor for transitioning to delirium in intensive
care unit patients. Anesthesiology 2006; 104: 2126.
van Gool WA, van de Beek D, Eikelenboom P. Systemic infection
and delirium: when cytokines and acetylcholine collide. Lancet 2010;
375: 77375.
Golinger RC, Peet T, Tune LE. Association of elevated plasma
anticholinergic activity with delirium in surgical patients.
Am J Psychiatry 1987; 144: 121820.
Han L, McCusker J, Cole M, Abrahamowicz M, Primeau F, Elie M.
Use of medications with anticholinergic eect predicts clinical
severity of delirium symptoms in older medical inpatients.
Arch Intern Med 2001; 161: 1099105.
Fischer P. Successful treatment of nonanticholinergic delirium with
a cholinesterase inhibitor. J Clin Psychopharmacol 2001; 21: 118.
Oldenbeuving AW, de Kort PL, Jansen BP, Kappelle LJ, Roks G.
A pilot study of rivastigmine in the treatment of delirium after
stroke: a safe alternative. BMC Neurol 2008; 8: 34.
European Medicines Agency. Rivastigmine 1A pharma. http://www.
ema.europa.eu/ema/index.jsp?curl=pages/medicines/human/
medicines/001181/human_med_001210.jsp&murl=menus/
medicines/medicines.jsp&mid=WC0b01ac058001d125 (accessed
Oct 13, 2010).
Ely EW, Inouye SK, Bernard GR, et al. Delirium in mechanically
ventilated patients: validity and reliability of the confusion
assessment method for the intensive care unit (CAM-ICU). JAMA
2001; 286: 270310.
McKeith I, Del Ser T, Spano P, et al. Ecacy of rivastigmine in
dementia with Lewy bodies: a randomised, double-blind, placebocontrolled international study. Lancet 2000; 356: 203136.
Darreh-Shori T, Jelic V. Safety and tolerability of transdermal and
oral rivastigmine in Alzheimers disease and Parkinsons disease
dementia. Expert Opin Drug Saf 2010; 9: 16776.
Jorm AF. The informant questionnaire on cognitive decline in the
elderly (IQCODE): a review. Int Psychogeriatr 2004; 16: 27593.
Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method
of classifying prognostic comorbidity in longitudinal studies:
development and validation. J Chronic Dis 1987; 40: 37383.
Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II:
a severity of disease classication system. Crit Care Med 1985;
13: 81829.
Ely EW, Truman B, Shintani A, et al. Monitoring sedation status
over time in ICU patients: reliability and validity of the Richmond
Agitation-Sedation Scale (RASS). JAMA 2003; 289: 298391.
Inouye SK, van Dyck CH, Alessi CA, Balkin S, Siegal AP,
Horwitz RI. Clarifying confusion: the confusion assessment
method. A new method for detection of delirium. Ann Intern Med
1990; 113: 94148.

www.thelancet.com Vol 376 November 27, 2010

Articles

26

27

28

29

30

Milbrandt EB, Deppen S, Harrison PL, et al. Costs associated with


delirium in mechanically ventilated patients. Crit Care Med 2004;
32: 95562.
Vincent JL, de Mendona A, Cantraine F, et al. Use of the SOFA
score to assess the incidence of organ dysfunction/failure in
intensive care units: results of a multicenter, prospective study.
Working group on sepsis-related problems of the European
Society of Intensive Care Medicine. Crit Care Med 1998;
26: 1793800.
Ely EW, Gautam S, Margolin R, et al. The impact of delirium in the
intensive care unit on hospital length of stay. Intensive Care Med
2001; 27: 1892900.
Overshott R, Vernon M, Morris J, Burns A. Rivastigmine in the
treatment of delirium in older people: a pilot study. Int Psychogeriatr
2010; 22: 81218.
van Eijk MM, van Marum RJ, Klijn IA, de Wit N, Kesecioglu J,
Slooter AJ. Comparison of delirium assessment tools in a mixed
intensive care unit. Crit Care Med 2009; 37: 188185.

www.thelancet.com Vol 376 November 27, 2010

31

32

33

34

Bourne RS, Tahir TA, Borthwick M, Sampson EL. Drug treatment


of delirium: past, present and future. J Psychosom Res 2008;
65: 27382.
UK National Health Service National Electronic Library for
Medicines. Rivastigmine. http://www.nelm.nhs.uk/en/MedicinesA-Z/Medicine/?vtm=323368009 (accessed Oct 13, 2010).
Gamberini M, Bolliger D, Lurati Buse GA, et al. Rivastigmine for
the prevention of postoperative delirium in elderly patients
undergoing elective cardiac surgerya randomized controlled trial.
Crit Care Med 2009; 37: 176268.
Liptzin B, Laki A, Garb JL, Fingeroth R, Krushell R. Donepezil in
the prevention and treatment of post-surgical delirium.
Am J Geriatr Psychiatry 2005; 13: 110006.

1837

Articles

Eects of a mobile phone short message service on


antiretroviral treatment adherence in Kenya
(WelTel Kenya1): a randomised trial
Richard T Lester, Paul Ritvo, Edward J Mills, Antony Kariri, Sarah Karanja, Michael H Chung, William Jack, James Habyarimana,
Mohsen Sadatsafavi, Mehdi Najafzadeh, Carlo A Marra, Benson Estambale, Elizabeth Ngugi, T Blake Ball, Lehana Thabane, Lawrence J Gelmon,
Joshua Kimani, Marta Ackers, Francis A Plummer

Summary
Lancet 2010; 376: 183845
Published Online
November 9, 2010
DOI:10.1016/S01406736(10)61997-6
See Comment page 1807
Department of Medical
Microbiology, University of
Nairobi, Nairobi, Kenya
(R T Lester MD, A Kariri BSc,
S Karanja BSc, E Ngugi PhD,
L J Gelmon MD, J Kimani, MBChB);
Department of Medical
Microbiology, University of
Manitoba, Health Sciences
Centre, Winnipeg, MB, Canada
(R T Lester, T B Ball PhD,
L J Gelmon, J Kimani,
Prof F A Plummer MD); Division
of Infectious Diseases,
Department of Medicine,
University of British Columbia,
Vancouver, BC, Canada
(R T Lester); School of
Kinesiology and Health
Sciences, Department of
Psychology, York University,
York, ON, Canada (P Ritvo PhD);
Faculty of Health Sciences,
University of Ottawa, Ottawa,
ON, Canada (E J Mills PhD);
Department of Global Health,
University of Washington,
Seattle, WA, USA
(M H Chung MD); Department of
Economics (W Jack DPhil) and
Georgetown Public Policy
Institute (J Habyarimana PhD),
Georgetown University,
Washington, DC, USA;
Collaboration for Outcome
Research and Evaluation,
Faculty of Pharmaceutical
Sciences, University of British
Columbia, Vancouver, BC,
Canada (M Sadatsafavi MD,
M Najafzadeh MSc,
C A Marra PharmD); University
of Nairobi Institute of Tropical
and Infectious Diseases,
Nairobi, Kenya
(B Estambale MBChB);
Department of Clinical
Epidemiology and Biostatistics,
McMaster University, Hamilton,
ON, Canada (L Thabane PhD);

1838

Background Mobile (cell) phone communication has been suggested as a method to improve delivery of health
services. However, data on the eects of mobile health technology on patient outcomes in resource-limited settings
are limited. We aimed to assess whether mobile phone communication between health-care workers and patients
starting antiretroviral therapy in Kenya improved drug adherence and suppression of plasma HIV-1 RNA load.
Methods WelTel Kenya1 was a multisite randomised clinical trial of HIV-infected adults initiating antiretroviral
therapy (ART) in three clinics in Kenya. Patients were randomised (1:1) by simple randomisation with a random
number generating program to a mobile phone short message service (SMS) intervention or standard care. Patients
in the intervention group received weekly SMS messages from a clinic nurse and were required to respond within
48 h. Randomisation, laboratory assays, and analyses were done by investigators masked to treatment allocation;
however, study participants and clinic sta were not masked to treatment. Primary outcomes were self-reported ART
adherence (>95% of prescribed doses in the past 30 days at both 6 and 12 month follow-up visits) and plasma HIV-1
viral RNA load suppression (<400 copies per mL) at 12 months. The primary analysis was by intention to treat. This
trial is registered with ClinicalTrials.gov, NCT00830622.
Findings Between May, 2007, and October, 2008, we randomly assigned 538 participants to the SMS intervention (n=273)
or to standard care (n=265). Adherence to ART was reported in 168 of 273 patients receiving the SMS intervention
compared with 132 of 265 in the control group (relative risk [RR] for non-adherence 081, 95% CI 069094; p=0006).
Suppressed viral loads were reported in 156 of 273 patients in the SMS group and 128 of 265 in the control group,
(RR for virologic failure 084, 95% CI 071099; p=004). The number needed to treat (NNT) to achieve greater than
95% adherence was nine (95% CI 50295) and the NNT to achieve viral load suppression was 11 (582273).
Interpretation Patients who received SMS support had signicantly improved ART adherence and rates of viral
suppression compared with the control individuals. Mobile phones might be eective tools to improve patient
outcome in resource-limited settings.
Funding US Presidents Emergency Plan for AIDS Relief.

Introduction
Health programmes that use mobile communication
technologies are emerging with the aim of strengthening health systems.13 The United Nations Joint
Programme on HIV/AIDS (UNAIDS) and WHO have
added wireless communication technologies to their
strategic plans.4,5 However, at present no published
clinical trial has reported the use of mobile health
technologies to improve patient-centred outcomes in
developing countries.
Present eorts to control the HIV/AIDS pandemic
include treatment with antiretroviral therapy (ART),
targeted prevention strategies, and treatment as
prevention measures (ie, prevention of HIV spread by
treating HIV positive people and thereby reducing the
risk of onward transmission).6,7 However, widespread
progress at controlling the pandemic is restricted by poor
infrastructure and increasing health-system costs.

The number of mobile (cell) phone users is rapidly


expanding (45 billion mobile phone subscribers are
expected worldwide by 2012),8 mainly because of free
market forces (ie, capitalism) and the demand for rapid
wireless communications for personal use and to aid
multi-sector economic development (eg, trade, tourism,
and infrastructure); thus, mobile technology has the
potential to be used in health systems worldwide. A
wide range of medical services could be improved by
providing patient-focused support and management
through the health-care system.
Maximum adherence to ART in patients with HIV
improves health outcomes and prevents drug resistance.9
Adherence is also important for programme cost
containment.10 If mobile phone use does improve health
outcomes in resource-limited settings, this mobile health
technology could thus be included in health-system
strategies and help improve health development goals.11
www.thelancet.com Vol 376 November 27, 2010

Articles

In this trial, we aimed to assess whether mobile phone


communication between health-care workers and
patients initiating ART in Kenya2,12 improved drug
adherence and suppression of plasma HIV-1 RNA load.

Methods
Patients
Patients initiating ART were recruited from three
dierent HIV clinics that are involved in intense ART
provision scale-up. The University of Nairobi Pumwani
Clinic serves a very low-income population in Nairobi13
and the Coptic Hope Center for Infectious Diseases
operates out of a faith-based hospital located in a higherincome area of Nairobi.14 The Kajiado Clinic is a
government health centre in a large rural district. We
chose these three locations because they should represent
the regional diversity of health settings.
Patients were eligible for study participation if they
were over 18 years old, initiating ART for the rst time,
and able to access a mobile phone on a near-daily basis
and communicate via short message service (SMS).
People who did not own mobile phones were eligible if
they had shared access (with corroborative agreement by
the phone owner), and illiterate patients were eligible if
assisted by a literate partner. Participants used existing
mobile phone services; phones and network airtime
credit were not provided.
Patients provided written or verbal informed consent at
enrolment in a language they understood. The study
protocol was approved by the University of Manitoba and
Kenyatta National Hospital ethics review boards.
This trial is registered with ClinicalTrials.gov,
NCT00830622.

Randomisation and masking


WelTel Kenya1 was an individually randomised, parallel,
multisite controlled trial. Patients were randomly
assigned (1:1) by simple randomisation15 to the SMS
intervention or to standard care (control group). A project
statistician generated the randomisation numbers with a
random number generating program. Written allocation
of assignment was sealed in individual opaque envelopes
marked with study identication numbers, which were
distributed to all three study clinics. Target enrolment
was estimated at a ratio of about 2:2:1 across clinics, with
the urban centres having the higher enrolment. We did
not do block randomisation because of the unpredictable
scale-up rates at each clinic. Randomisation, laboratory
assays, and analyses were done by investigators masked
to treatment allocation; however, study participants and
clinic sta could not be masked because the intervention
required overt participation.

Procedures
Antiretroviral drugs were provided by the government of
Kenya with support from the US Presidents Emergency
Plan for AIDS Relief (PEPFAR), and consisted primarily
www.thelancet.com Vol 376 November 27, 2010

of three drug combinations containing zidovudine or


stavudine, plus lamivudine, plus efavirenz or nevirapine
as rst-line drugs. Initiation of ART was in accordance
with national guidelines (CD4 count <250 cells per L or
WHO stage III or IV).16
Typically, the study site in Kajiado provided one
counselling session at ART initiation and the two sites in
Nairobi provided two counselling sessions before and
one session 1 month after ART initiation. Disclosure of
HIV status, pairing up with a treatment adherence
partner, and participation in support groups was
encouraged but not insisted upon. Additional brief
counselling was provided at each site during dispensation
of the drugs in the clinic or pharmacy.
We thought that regular, structured mobile phone
communication between health-care workers and
patients could improve patient outcomes by both
reminding patients to take their ART and by providing
support to the patients. The intervention was planned in
consultation with investigators, clinic sta, and patients
(in unstructured focus group sessions) with the goals of
low cost and widespread usability.2,17
All intervention participants received brief training for
use of the SMS intervention from the study clinicians.
They were informed that the SMS support service did not
replace existing adherence counselling or emergency
services. On Monday morning of each week, the site
nurse or clinical ocer sent a text message via SMS to
patients in the intervention group to inquire about their
status and thus to remind them about the availability of
phone-based support. Typically, the slogan Mambo?
was sent, which is Kiswahili for How are you? The
health workers used multiple recipient (bulk) messaging
functions to improve eciency. Patients in the intervention group were instructed to respond within 48 h
that either they were doing well (Sawa) or that they had

Global AIDS Program, US


Centers for Disease Control and
Prevention, Nairobi, Kenya
(M Ackers MD); and National
Microbiology Laboratory, Public
Health Agency of Canada,
Winnipeg, MB, Canada
(F A Plummer)
Correspondence to:
Dr Richard T Lester, British
Columbia Centre for Disease
Control, 655 West 12th Avenue,
Vancouver, BC, V5Z 4R4, Canada
rlester.id@gmail.com

581 patients enrolled

43 ineligible
39 had inadequate
phone access
4 declined participation

538 randomised

273 assigned to SMS


intervention

7 withdrew

265 assigned to
standard care

3 withdrew

273 included in
primary analysis

265 included in
primary analysis

Figure 1: Trial prole

1839

Articles

Women
Age (years)

SMS group (n=273)

Control group (n=265)

177 (65%)

174 (66%)

367 (85, 190650)

366 (79, 220840)

Clinic
University of Nairobi Pumwani Clinic

120 (44%)

131 (49%)

Coptic Hope Centre for Infectious Diseases

117 (43%)

92 (35%)

Kajiado Clinic

36 (13%)

CD4 cell count per L*

1676 (1217, 208870)

42 (16%)
1609 (1414, 10 15220)

WHO stage
1

52 (23%)

67 (29%)

59 (25%)

101 (44%)

103 (43%)

9 (4%)

13 (5%)

HIV-1 plasma RNA load (log10 copies per mL)

459 (105)

62 (26%)

483 (096)

Language literacy
English only
Kiswahili only
Both
Other only

1 (0%)

2 (1%)

48 (18%)

36 (14%)

213 (79%)

215 (81%)

9 (3%)

12 (5%)

Education
10 (4%)

14 (5%)

Primary

None

108 (40%)

86 (32%)

Secondary

106 (39%)

124 (47%)

49 (18%)

41 (16%)

Post-secondary
Monthly income
<2000 KES (<US$1/day)

72 (29%)

64 (28%)

114 (47%)

98 (43%)

10 00040 000 KES ($520/day)

46 (19%)

61 (27%)

>40 000 KES (>$20/day)

13 (5%)

7 (3%)

Owns

239 (88%)

225 (85%)

Shares

34 (12%)

40 (15%)

<200010 000 KES ($15/day)

Mobile phone access

Residence status
Rural

51 (19%)

50 (19%)

Urban

222 (81%)

215 (81%)

Data are number (%), mean (SD, range), median (IQR), mean (SD). Percentages do not add up to 100% in some cases
because of rounding. KES=Kenyan shillings. *Data missing for one patient in the SMS group and four in the control
group. Data missing for 44 patients in the SMS group and 28 in the control group. Data missing for 21 patients in
the SMS group and 25 in the control group; for 19 patients in the SMS group and ten in the control group, baseline
viral load was below the limit of detection (400 copies per mL). Log of viral load for these patients was given as
log10(400). Data missing for two patients in the SMS group. Data unavailable for 28 patients in the SMS group and
35 in the control group.

Table 1: Demographics and baseline characteristics

a problem (Shida). The clinician then called patients


who said they had a problem or who failed to respond
within 2 days. Participants were instructed that healthcare workers were available to respond during clinic
hours only. All mobile phone communications between
the health-care workers and patients were recorded in
the study log.
The primary outcomes were self-reported adherence and
suppression of plasma HIV-1 viral load. Self-reported
adherence is the most practical method of assessing
adherence because it closely represents the regional
1840

standard care;18 however, this method of assessment has


been reported to be an overestimate of adherence.19 At
6 and 12 months follow-up, we asked participants how
many pills they missed in the past 30 days; they were
classed as adherent if they reported that they had taken
more than 95% of the provided pills at both follow-up
visits.9 We measured adherence at two timepoints to assess
the durability of adherence and to increase the sensitivity
to detect adherence failures, because of the high levels of
adherence rates recorded by self-report regionally.
Viral load is an important composite endpoint for
monitoring adherence and takes into account pharmacological, biological, and sociobehavioural factors.
Participants were classed as virologically suppressed if
their plasma HIV-1 RNA load at their 12-month visit was
400 copies per mL or less. Patients who did not achieve
this outcome were classed as virologic failures. Plasma
was taken from patients at the 12-month visit for
assessment of HIV-1 RNA viral load (Amplicor, Roche
Diagnostics, Mannheim, Germany) and was stored and
analysed at a later timepoint in batches. Laboratory assays
(CD4 count and HIV-1 RNA load) were all done at a
central laboratory (University of Nairobi Institutes for
Tropical and Infectious Diseases). CD4 count testing
(FACScan, Becton Dickinson, Sunnyvale, CA, USA) was
part of routine care at the urban sites and was provided
for study purposes at the rural site.
Secondary outcomes included the rate of attrition (not
having a nal visit at 12 months) and rates of several
categories of attrition (mortality, withdrawal from the
study, transfer to non-study clinics, and loss to follow-up
without identiable cause). Other predened secondary
endpoints, including quality of life and social and
economic outcomes, will be reported separately. We also
assessed the eects of the SMS intervention in
prespecied subgroups of patients subdivided by sex,
urban or rural residence, clinic attended, disease stage,
and whether they owned or shared a phone.20 Finally, at
one urban site and the rural site we ran semistructured
focus group sessions with ten to 20 participants before,
during, and after the start of the trial and asked for
participant feedback.20

Statistical analysis
We calculated that a sample size of at least 534 would be
required to detect a 10% improvement in adherence, with
80% power and 005 level of signicance.20,21 Demographic
and covariate information were recorded at baseline
(month 0) and at scheduled visits at 6 and 12 months. Selfreported adherence to ART was assessed by a standardised
questionnaire at each follow-up visit. Study sta
maintained a study register to record all SMS responses
and other mobile phone communications with patients.
Patients dened as lost to follow-up were those unable to
be traced within 3 months of the study end date.
Detailed description of the analysis methods can be
found in the trial protocol.20 Briey, we analysed the
www.thelancet.com Vol 376 November 27, 2010

Articles

primary outcomes with the test. The analysis of primary


outcomes was by intention to treat. The primary analyses
were not adjusted, as prespecied and recommended.20,22
Relative risk (RR) was reported for non-adherence and
virologic failure, with an RR less than 1 suggesting better
outcome for the intervention group.
As a measure of absolute eect size, we also calculated
the number needed to treat (NNT) and its associated
95% CI for both unadjusted primary outcomes.23 We also
did a per-protocol (complete-case) analysis of the primary
outcomes, in which only participants who had complete
primary outcome data (self-reported adherence at 6 and
12 months and viral load at 12 months) were included.
We also did adjusted analyses by tting a logistic
regression model to the primary outcomes with adjustments for sex, age, baseline viral load, baseline CD4
count, and baseline WHO stage. We used multiple
imputation (PROC MI and MIANALYZE in SAS, with
ve sets of imputations) to impute the missing values
of covariates.
Secondary outcomes were compared with the test.
We used the Fishers exact test for outcomes that were
reported in ve patients or less to estimate p values. In
the subgroup analysis in the intention-to-treat population,
we compared the intervention groups within each
subgroup of patients with the test. Heterogeneity of
the eect of the intervention across subgroups was
assessed by comparing logistic regression models with
and without interaction term between treatment
allocation and subgroup-dening variables (using the
likelihood-ratio test).
The criterion for signicance was set at =005.
Bonferroni correction with resulting p values of 001 (for
ve tests) was used for predened exploratory secondary
and subgroup analyses. For all models, the results are
expressed as an estimate of eect size, with 95% CIs
and p values. The response type throughout the study
period was analysed by categorising time since random
allocation to the rst 3 months and afterwards, and doing
a test between time and response type. All statistical
analyses were done with SAS (version 9.2, 64-bit edition).

Role of the funding source


The sponsors of the study had no role in the design of
the original study protocol, data collection, data analysis,
data interpretation, writing of the report, or decision to
submit the manuscript for publication. RTL, TK, SK,
MHC, MS, MN, and CM had full access to all the data in
the study, RTL had nal responsibility for the decision to
submit for publication, and all authors approved the
decision to submit.

Results
Between May, 2007, and October, 2008, we enrolled
581 participants (gure 1). Consecutive enrolment was
attempted; however, one site enrolled alternate patients
into separate studies. After screening, 39 patients were
www.thelancet.com Vol 376 November 27, 2010

SMS group
(number [%])

Control group
(number [%])

RR (95% CI)*

p value

Primary outcome
Intention-to-treat analysis
Self-reported adherence (>95%)

168 (62%)

132 (50%)

081 (069094)

0006

Viral suppression
(<400 copies per mL)

156 (57%)

128 (48%)

085 (072099)

004

Complete-case analysis
Self-reported adherence

168 (91%)

132 (91%)

100 (094107)

094

Viral suppression

156 (75%)

128 (66%)

088 (077100)

0047

Total attrition (missing)

53 (19%)

61 (23%)

124 (082189)

031

Loss to follow-up

17 (6%)

27 (10%)

169 (091323)

0094

Mortality

Secondary outcomes

25 (9%)

30 (11%)

127 (072222)

042

Withdrawal

7 (3%)

3 (1%)

226 (059867)

034||

Transfer out

4 (1%)

1 (0%)

025 (019217)

038||

Percentages do not add up to 100% in some cases because of rounding. *For non-adherence or virologic failure.
273 patients in the SMS group and 265 in the control group. Because the intention-to-treat analysis classed all patients
with missing data as non-adherent or having viral failure, the number of adherent patients and number of patients with
viral suppression are the same here as in the intention-to-treat analysis. 185 patients in the SMS group and 145 patients
in the control group. 208 patients in the SMS group and 194 patients in the control group. ||Fishers exact test.

Table 2: Primary and secondary outcomes

excluded because they had inadequate phone access and


four declined participation. Accordingly, 538 patients
were randomly assigned: 273 to the SMS intervention
and 265 to standard care. Ten participants (seven in the
SMS group and three in the control group) withdrew
from the study after random allocation for personal
reasons. Table 1 reports the demographics and baseline
characteristics of both groups, which are generally similar
to those of the AIDS epidemic in Kenya, including
predominant female sex.24
More patients in the SMS intervention group than in
the control group had self-reported adherence of over
95% at both visits (table 2). By contrast, in the completecase analysis, adherence was not dierent between
groups. After adjusting for baseline covariates, selfreported adherence remained signicantly better in the
SMS group than the control group (odds ratio [OR] 057,
95% CI 040083; p=00028). The NNT for adherence
was nine (95% CI 50295).
More patients in the SMS group than in the control
group had suppressed viral loads below the level of
detection (<400 copies per mL) at 12 months (table 2). In
the complete-case analysis there was also higher viral
suppression in the SMS group than the control group.
After adjustment, the intention-to-treat analysis showed
weak evidence of improved suppression of viral load in the
SMS group compared with the control group (OR 071,
95% CI 050101; p=0058). The NNT to achieve HIV-1
viral suppression was 11 (95% CI 582273).
No secondary outcomes showed signicant
associations with the intervention (table 2). Figure 2
shows the results of the subgroup analysis and the
heterogeneity of the intervention eect. Male sex, urban
1841

Articles

Adherence Number of events (n/N [%])

OR (95% CI)

p value

Viral load

Number of events (n/N [%])

OR (95% CI)

p value

Sex

Sex
Women
Men

112/177 (63%)

070 (046107)

56/96 (58%)

049 (027088)

070

Women

101/177 (57%)

066 (043100)

55/96 (57%)

080 (045142)

Men

073

Mobile phone

Mobile phone
Shares

19/34 (56%)

053 (021133)

Owns

149/239 (62%)

064 (044093)

086

Shares

22/34 (65%)

036 (014094)

Owns

134/239 (56%)

078 (054112)

036

Clinic

Clinic
Pumwani

82/120 (68%)

064 (038108)

Pumwani

75/120 (63%)

054 (033089)

Kajiado

16/36 (44%)

056 (022142) 094

Kajiado

20/36 (56%)

054 (022134) 032

Coptic

70/117 (60%)

059 (034102)

Coptic

61/117 (52%)

105 (061181)

WHO stage

WHO stage
1

36/52 (69%)

070 (032153)

31/52 (60%)

094 (044198)

45/67 (67%)

054 (026111)

42/67 (63%)

058 (028117)

54/101 (53%)

062 (036108)

56/101 (55%)

058 (033100)

7/9 (78%)

024 (004166)

5/9 (56%)

128 (023719)

Rural

30/51 (59%)

051 (023112)

Urban

126/222(57%)

075 (052110)

Overall

174/265 (66%)

070 (050098)

002

006

Residence

Residence
Rural

29/51 (57%)

051 (023112)

Urban

139/222 (63%)

065 (044095)

Overall

168/273 (62%)

062 (044088)

0125

025

05

1
Favours
SMS
group

081

0125

025

Favours
control
group

05

1
Favours
SMS
group

057

2
Favours
control
group

Figure 2: Subgroup analyses


Estimated intervention eects are reported as ORs and 95% CIs for risk of failure. p values for heterogeneity were derived from likelihood ratio statistics. OR=odds ratio.

08

Proportion of weekly SMS responses

07
06
05
04
03
02

Sawa (ne)
No response
Shida (problem)

01
0
0

6
Months

10

11

12

Figure 3: SMS intervention response rates


The graph is truncated at 12 months follow-up. A line is drawn at 3 months indicating the usual transition to HIV
disease stability and reduction in toxicities after initiation of antiretroviral therapy.

residence, and mobile phone ownership all favoured


adherence compared with the control group. However,
none was signicant after Bonferroni correction.
We recorded 11 983 valid responses to the SMS
inquiries from intervention participants. Figure 3
reports the patient responses over time. There were
7812 Sawa responses, 391 Shida responses, and
there were no responses or responses were not received
on time on 3780 occasions. The proportion of Shida
responses decreased from 61% in the rst 3 months
1842

after recruitment to 20% afterwards (p<00001). The


most common reason for a Shida response was a
medical issue. Non-responders most frequently cited
cost and other logistical factors or forgetting, rather
than health issues, as the main reasons for not
responding on time.
No adverse event directly attributable to the mobile
phone SMS communication, such as breaches of
condentiality (eg, if non-participants found out the
participants HIV status in an unintentional way) or
injury (eg, caused by driving or riding a bike whilst
texting), was reported in the weekly study logs or during
follow-up visits with health-care workers. At the end of
the study, 191 of 194 patients in the intervention group
reported they would like the SMS programme to
continue, of whom 188 (98%) said they would recommend
it to a friend. In the focus group sessions, many patients
in the intervention group also reported that they thought
the SMS support service was valuable (data not shown).

Discussion
This study shows that mobile health innovations can
improve HIV treatment outcomes. Patients who received
the SMS support were more likely to report adherence to
ART and were more likely to have their viral load
suppressed below detection levels than patients who
received the standard care alone.
The primary analysis classed all-cause attrition as
treatment failures. Thus, the higher follow-up rates and
www.thelancet.com Vol 376 November 27, 2010

Articles

lower mortality reported in the intervention group


contributed to the positive intervention eect for the
primary outcomes. When only available data were included
in the complete-case analyses, a signicant reduction was
preserved in viral suppression but not in self-reported
adherence. This could be because of a recall or social
desirability bias in self-reporting adherence among those
who were followed up. Alternatively, patients might have
been less likely to respond to follow-up if they had not
adhered to ART. Nonetheless, we chose an intention-totreat analysis for the primary outcome because
contributions of loss to follow-up are important indicators
for the durable implementation of ART programmes.2527
Our reported adherence and viral suppression rates
seem to be lower than in other studies;28,29 however, our
detection of adherence failures might have been more
sensitive because we used two timepoints for adherence,
and comparative data on viral suppression outside of
heavily researched settings is limited.30 However, the
actual mean adherence rates reported by participants in
this study was quite high, and because the SMS
intervention was well received by patients and the study
consent process informed participants of potential
benets of mobile phone communication, the study
might have underestimated the actual intervention eect
compared with the regional standard care. More mobile
phone usage might have been promoted by the study
overall; study nurses reported a higher number of calls
from study control individuals compared with non-study
clinic attendees (data not shown). Additionally, several
participants reported forwarding their weekly text
messages to non-intervention participants to share
support. Adherence rates reported among participants in
this study were higher than non-study participants within
the same clinics as well as non-study clinics operated by
the same ART programme providers (ART adherence
data provided by individual programme managers; data
not shown). The SMS intervention was well received by
patients, many of whom reported that they felt like
someone cares; most recommended for the SMS
programme to continue (data not shown).
This is, to our knowledge, the rst eectiveness trial
assessing the ability of a mobile health technology
intervention to inuence HIV outcomes in a resourcelimited setting (panel). However, pilot studies of mobile
health technologies are emerging.3134 In developed
countries, mobile health technology interventions are
gaining a clear evidence base for management and
prevention of a broad range of disorders.3,35 Two
randomised trials in the USA assessed a counselling
intervention by landline telephone for patients taking
ART and reported a signicant benet for adherence but
not for virologic control.36,37 However, in our study the
intervention included regular SMS communications
without additional counselling, so human resource and
training requirements were minimal. Because only 33%
of the weekly text messages identied a denitive
www.thelancet.com Vol 376 November 27, 2010

Panel: Research in context


Systematic review
We searched Medline, Cochrane CENTRAL, and Embase (from
inception to Oct 24, 2010) using the MeSH terms cellular
phone and HIV. We did not identify any randomised
controlled trials that have assessed mobile technology for
treatment support in HIV patients.
Interpretation
This is, to our knowledge, the rst randomised controlled trial
of mobile technology to support adherence and viral
suppression in patients with HIV/AIDS. We reported a
statistically signicant and clinically important increase in
self-reported adherence and viral suppression.

requirement for follow-up (Shida), one nurse could


potentially manage 1000 patients by SMS and expect to
call only 33 patients per week. Additionally, the patient
follow-up seemed to become more ecient over time,
because the proportion of Shida responses from patients
decreased after the rst 3 months on ART; however, this
decrease was after the period in which the most disease
instability would be expected.38
Although some features of health care and communications are universal, developing and developed economic
settings can have dierent challenges and opportunities.
Developing countries have resource challenges but might
benet from bypassing less advanced technologies and
instead using the most advanced methods available, which
are more uniformly taken up and that allow fresh
innovation. Kenya, for instance, led the world in mobile
phone money transfers, allowing people who never
previously accessed banks to safely transfer funds
routinely.39 Mobile phones could also be used to support
and track patients who transfer between ART provision
sites. The political turmoil in Kenya after the 2007
presidential elections when hundreds of thousands of
civilians were internally displaced because of ethnic
violence40 occurred during our study period. Although
some mobile phones were lost, others were used to request
assistance from clinic sta, which was frequently met with
counselling support and directions to new, safe locales
where drug rells could be obtained. Thus, the SMS
service seemed to be durable in a crisis.41
Overall, this study has implications for policy makers
and global funders of ART programmes. First, this is one
of the rst adherence interventions to confer a reduction
in virologic failures.42 ART needs to be taken lifelong,
thus optimal adherence is crucial to the prevention of
antiretroviral drug resistance. Instances of drug resistance
make future treatment options more challenging and
progressively more expensive to deliver. Additionally,
reducing viral replication through ART can decrease
transmission of HIV-1 to new partners43 and thus can
play a preventive role at the population level to reduce the
number of new infections.44,45
1843

Articles

The SMS intervention is inexpensive (each SMS costs


about US$005, equivalent to $20 per 100 patients
per month, and follow-up voice calls averaged $375
per nurse per month) and the mobile phone protocol uses
existing infrastructure. This protocol is also probably less
expensive than in-person community adherence
interventions,30,46,47 on the basis of travel costs alone. Thus,
the intervention could be both cost eective and cost
saving. The estimated NNT of 11 patients for each
additional patient with viral suppression in the SMS group
over the standard care group could theoretically translate
into huge health and economic benets if the programme
was successfully scaled up. For example, if hypothetically
applied to the 297 800 people who received ART in Kenyas
PEPFAR programme in 2009 (and assuming the eect
went beyond 1 year), the SMS intervention could have
resulted in 26 354 additional people with fully suppressed
viral loads. Innovations in management of automated text
messaging and publicprivate partnerships with mobile
health technology developers could improve programme
eciency and scalability.
The applicability of this study to other countries and
other diseases remains to be assessed. Factors that
inuence adherence are often common within Africa and
other global settings.48,49 Although the uptake of wireless
telecommunication devices is becoming ubiquitous,
introduction of mobile health initiatives is variable.1,50 We
believe that the patient-centred communication eect, in
particular the timely support of a patient by a health
professional, is universal and can be improved by mobile
telecommunication.
Contributors
RTL conceived the study and was the principal investigator of the funded
public health evaluation. PR, AK, SK, MHC, WJ, JH, and EN designed
the study and BE, LJG, and JK helped with implementation. RTL, MHC,
LJG, JK, and FAP are holders of grants that broadly support the ART
clinics and research programmes. EJM and LT provided statistical
expertise in clinical trial design. MS and MN, supervised by CAM, did
the primary statistical analysis. All authors contributed to renement
of the study protocol and approved the nal manuscript.
Conicts of interest
MA is an employee of the US Centers for Disease Control and
Prevention (CDC). All other authors declare that they have no conict
of interest.
Acknowledgments
The trial was funded by the US Presidents Emergency Plan for AIDS
Relief (PEPFAR) through the CDC cooperative agreement 5U62PS024510
as a public-health evaluation (PHE:KE07.0045). The Coptic Hope Center
for Infectious Diseases is supported by PEPFAR through the CDC
cooperative agreement (U62/CCU024512-04). Kajiado District Hospital
Comprehensive Care Clinic is supported by the Government of Kenya
Ministry of Health and APHIA II programme. RTL was supported by an
unrestricted research award through Association of Medical Microbiology
and Infectious DiseasesBristol-Myers-Squibb (Canada) and a Canadian
Institute for Health Research (CIHR) training award, and from the
University of Manitoba hosted CIHR/ICID (International Center for
Infectious Diseases) National Training Program in Infectious Diseases.
EJM is a tier 2 Canada Research Chair in global health. MHC was
supported by a K23 grant from the National Institutes of Health
(5K23AI065222-04). FAP is a tier 1 Canada Research Chair in resistance
and susceptibility to infections. The ndings and conclusions in this
paper are those of the authors and do not necessarily represent the

1844

ocial position of the funding agencies. We thank the health-care sta


and patients who participated in the study. We dedicate our greatest
thanks to our Kenyan statistician, Rosemary Ngugi, who died of cancer
during preparation of this manuscript.
References
1
Kaplan WA. Can the ubiquitous power of mobile phones be used to
improve health outcomes in developing countries?
Global Health 2006; 2: 9.
2
Padma TV. Developing solutions. Nature 2010; 466: S1617.
3
Krishna S, Boren SA, Balas EA. Healthcare via cell phones:
a systematic review. Telemed J E Health 2009; 15: 23140.
4
Vitoria M, Granich R, Gilks CF, et al. The global ght against
HIV/AIDS, tuberculosis, and malaria: current status and future
perspectives. Am J Clin Pathol 2009; 131: 84448.
5
UNAIDS. Telecom: tools connecting the world and communicating
about HIV. 2009 Oct 5. http://www.unaids.org/en/Knowledge
Centre/Resources/FeatureStories/archive/2009/20091005_telecom.
asp (accessed May 20, 2010).
6
Remien RH, Berkman A, Myer L, Bastos FI, Kagee A,
El-Sadr WM. Integrating HIV care and HIV prevention:
legal, policy and programmatic recommendations. AIDS 2008;
22 (suppl 2): S5765.
7
Cohen MS, Hellmann N, Levy JA, DeCock K, Lange J. The spread,
treatment, and prevention of HIV-1: evolution of a global pandemic.
J Clin Invest 2008; 118: 124454.
8
Cellular-news. Global Mobile Phone Subscribers to Reach 4.5
Billion by 2012. http://www.cellular-news.com/story/29824.php
(accessed June 15, 2010).
9
Wood E, Hogg RS, Yip B, Harrigan PR, OShaughnessy MV,
Montaner JS. Eect of medication adherence on survival of
HIV-infected adults who start highly active antiretroviral therapy
when the CD4+ cell count is 0200 to 0350 10(9) cells/L.
Ann Intern Med 2003; 139: 81016.
10 Bennett DE, Bertagnolio S, Sutherland D, Gilks CF. The World
Health Organizations global strategy for prevention and assessment
of HIV drug resistance. Antivir Ther 2008; 13 (suppl 2): 113.
11 Rabkin M, El-Sadr WM, De Cock KM. The impact of HIV scale-up
on health systems: a priority research agenda.
J Acquir Immune Dec Syndr 2009; 52 (suppl 1): S611.
12 Lester RT, Kariri A. Mobilizing cell phones to improve antiretroviral
adherence and follow-up in Kenya: a randomized controlled trial in
progress. WHO Essential Medicines Monitor 2009; 2: Article 1.
13 Kreiss JK, Koech D, Plummer FA, et al. AIDS virus infection
in Nairobi prostitutes. Spread of the epidemic to East Africa.
N Engl J Med 1986; 314: 41418.
14 Chung MH, Drake AL, Richardson BA, et al. Impact of prior
HAART use on clinical outcomes in a large Kenyan HIV treatment
program. Curr HIV Res 2009; 7: 44146.
15 Schulz KF, Grimes DA. Unequal group sizes in randomised trials:
guarding against guessing. Lancet 2002; 359: 96670.
16 National AIDS and STI Control Programme (NASCOP) MoH.
Guidelines for antiretroviral drug therapy in Kenya, 3rd edn. 2005.
http://www.aidskenya.org/public_site/webroot/cache/article/le/
ART_GUIDELINES.pdf (accessed June 15, 2010).
17 Lester RT, Gelmon L, Plummer FA. Cell phones: tightening the
communication gap in resource-limited antiretroviral programmes?
AIDS 2006; 20: 224244.
18 Oyugi JH, Byakika-Tusiime J, Charlebois ED, et al. Multiple
validated measures of adherence indicate high levels of adherence
to generic HIV antiretroviral therapy in a resource-limited setting.
J Acquir Immune Dec Syndr 2004; 36: 110002.
19 Simoni JM, Kurth AE, Pearson CR, Pantalone DW, Merrill JO,
Frick PA. Self-report measures of antiretroviral therapy adherence:
a review with recommendations for HIV research and clinical
management. AIDS Behav 2006; 10: 22745.
20 Lester RT, Mills EJ, Kariri A, et al. The HAART cell phone
adherence trial (WelTel Kenya1): a randomized controlled trial
protocol. Trials 2009; 10: 87.
21 Wools-Kaloustian K, Kimaiyo S, Diero L, et al. Viability and
eectiveness of large-scale HIV treatment initiatives in sub-Saharan
Africa: experience from western Kenya. AIDS 2006; 20: 4148.
22 Perneger TV. Adjusting for multiple testing in studies is less
important than other concerns. BMJ 1999; 318: 1288.

www.thelancet.com Vol 376 November 27, 2010

Articles

23
24

25

26

27

28

29

30

31

32

33

34

35

36

Altman DG. Condence intervals for the number needed to treat.


BMJ 1998; 317: 130912.
National AIDS and STI Control Programme, Ministry of Health,
Kenya. July 2008. Kenya AIDS indicator survey 2007: preliminary
report. Nairobi, Kenya. (http://www.aidsportal.org/repos/KAIS%
20-%20Preliminary%20Report_July%20291.pdf) (accessed June 15,
2010).
Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy
programs in sub-Saharan Africa: a systematic review. PLoS Med
2007; 4: e298.
Brinkhof MW, Pujades-Rodriguez M, Egger M. Mortality of patients
lost to follow-up in antiretroviral treatment programmes in
resource-limited settings: systematic review and meta-analysis.
PLoS One 2009; 4: e5790.
Losina E, Toure H, Uhler LM, et al. Cost-eectiveness of preventing
loss to follow-up in HIV treatment programs: a Cote dIvoire
appraisal. PLoS Med 2009; 6: e1000173.
Kunutsor S, Evans M, Thoulass J, et al. Ascertaining baseline levels
of antiretroviral therapy adherence in Uganda: a multimethod
approach. J Acquir Immune Dec Syndr 2010; 55: 22124.
Barth RE, van der Meer JT, Hoepelman AI, et al. Eectiveness
of highly active antiretroviral therapy administered by general
practitioners in rural South Africa. Eur J Clin Microbiol Infect Dis
2008; 27: 97784.
Chang LW, Kagaayi J, Nakigozi G, et al. Eect of peer health
workers on AIDS care in Rakai, Uganda: a cluster-randomized trial.
PLoS One 2010; 5: e10923.
Curioso WH, Kurth AE. Access, use and perceptions regarding
Internet, cell phones and PDAs as a means for health promotion
for people living with HIV in Peru. BMC Med Inform Decis Mak
2007; 7: 24.
Homan JA, Cunningham JR, Suleh AJ, et al. Mobile direct
observation treatment for tuberculosis patients: a technical
feasibility pilot using mobile phones in Nairobi, Kenya.
Am J Prev Med 2010; 39: 7880.
Haberer JE, Kiwanuka J, Nansera D, Wilson IB, Bangsberg DR.
Challenges in using mobile phones for collection of antiretroviral
therapy adherence data in a resource-limited setting. AIDS Behav
2010; published online June 8. DOI:10.1007/s10461-010-9720-1.
Mukund Bahadur KC, Murray PJ. Cell phone short messaging
service (SMS) for HIV/AIDS in South Africa: a literature review.
Stud Health Technol Inform 2010; 160: 53034.
Strandbygaard U, Thomsen SF, Backer V. A daily SMS reminder
increases adherence to asthma treatment: a three-month follow-up
study. Respir Med 2010; 104: 16671.
Collier AC, Ribaudo H, Mukherjee AL, Feinberg J, Fischl MA,
Chesney M. A randomized study of serial telephone call support
to increase adherence and thereby improve virologic outcome in
persons initiating antiretroviral therapy. J Infect Dis 2005;
192: 1398406.

www.thelancet.com Vol 376 November 27, 2010

37

38

39

40

41

42

43

44
45

46

47

48
49

50

Reynolds NR, Testa MA, Su M, et al. Telephone support to improve


antiretroviral medication adherence: a multisite, randomized
controlled trial. J Acquir Immune Dec Syndr 2008; 47: 6268.
Braitstein P, Brinkhof MW, Dabis F, et al. Mortality of HIV-1infected patients in the rst year of antiretroviral therapy:
comparison between low-income and high-income countries.
Lancet 2006; 367: 81724.
Rice X. Kenya sets world rst with money transfers by mobile.
The Guardian (London) March 20, 2007: http://www.guardian.co.
uk/money/2007/mar/20/kenya.mobilephones (accessed
Nov 4, 2010).
Vreeman RC, Nyandiko WM, Sang E, Musick BS, Braitstein P,
Wiehe SE. Impact of the Kenya post-election crisis on clinic
attendance and medication adherence for HIV-infected children
in western Kenya. Con Health 2009; 3: 5.
Lester R, Karanja S. Mobile phones: exceptional tools for
HIV/AIDS, health, and crisis management. Lancet Infect Dis 2008;
8: 73839.
Rueda S, Park-Wyllie LY, Bayoumi AM, et al. Patient support and
education for promoting adherence to highly active antiretroviral
therapy for HIV/AIDS. Cochrane Database Syst Rev 2006;
3: CD001442.
Donnell D, Baeten JM, Kiarie J, et al. Heterosexual HIV-1
transmission after initiation of antiretroviral therapy: a prospective
cohort analysis. Lancet 2010; 375: 209298.
De Cock KM, Gilks CF, Lo YR, Guerma T. Can antiretroviral therapy
eliminate HIV transmission? Lancet 2009; 373: 79.
Montaner JS, Lima VD, Barrios R, et al. Association of highly active
antiretroviral therapy coverage, population viral load, and yearly
new HIV diagnoses in British Columbia, Canada: a populationbased study. Lancet 2010; 376: 53239.
Williams AB, Fennie KP, Bova CA, Burgess JD, Danvers KA,
Dieckhaus KD. Home visits to improve adherence to highly active
antiretroviral therapy: a randomized controlled trial.
J Acquir Immune Dec Syndr 2006; 42: 31421.
Behforouz HL, Farmer PE, Mukherjee JS. From directly observed
therapy to accompagnateurs: enhancing AIDS treatment outcomes
in Haiti and in Boston. Clin Infect Dis 2004; 38 (suppl 5): S42936.
Ware NC, Idoko J, Kaaya S, et al. Explaining adherence success in
sub-Saharan Africa: an ethnographic study. PLoS Med 2009; 6: e11.
Mills EJ, Nachega JB, Buchan I, et al. Adherence to antiretroviral
therapy in sub-Saharan Africa and North America: a meta-analysis.
JAMA 2006; 296: 67990.
Ybarra ML, Bull SS. Current trends in Internet- and cell
phone-based HIV prevention and intervention programs.
Curr HIV/AIDS Rep 2007; 4: 20107.

1845

Articles

Paediatric mortality related to pandemic inuenza A


H1N1 infection in England: an observational
population-based study
Nabihah Sachedina, Liam J Donaldson

Summary
Lancet 2010; 376: 184652
Published Online
October 27, 2010
DOI:10.1016/S01406736(10)61195-6
See Comment page 1808
Department of Health, London,
UK (N Sachedina MBBS,
Prof L J Donaldson MD); and
National Patient Safety
Agency, London, UK
(L J Donaldson)
Correspondence to:
Prof Liam J Donaldson, National
Patient Safety Agency,
48 Maple Street, London
W1T 5HD, UK
liam.donaldson@npsa.nhs.uk

Background Young people (aged 018 years) have been disproportionately aected by pandemic inuenza A H1N1
infection. We aimed to analyse paediatric mortality to inform clinical and public health policies for future inuenza
seasons and pandemics.
Methods All paediatric deaths related to pandemic inuenza A H1N1 infection from June 26, 2009, to March 22, 2010
in England were identied through daily reporting systems and cross-checking of records and were validated by
conrmation of inuenza infection by laboratory results or death certicates. Clinicians responsible for each
individual child provided detailed information about past medical history, presentation, and clinical course of the
acute illness. Case estimates of inuenza A H1N1 were obtained from the Health Protection Agency. The primary
outcome measures were population mortality rates and case-fatality rates.
Findings 70 paediatric deaths related to pandemic inuenza A H1N1 were reported. Childhood mortality rate was
6 per million population. The rate was highest for children aged less than 1 year. Mortality rates were higher for
Bangladeshi children (47 deaths per million population [95% CI 17103]) and Pakistani children (36 deaths per million
population [1864]) than for white British children (4 deaths per million [36]). 15 (21%) children who died were
previously healthy; 45 (64%) had severe pre-existing disorders. The highest age-standardised mortality rate for a preexisting disorder was for chronic neurological disease (1536 per million population). 19 (27%) deaths occurred before
inpatient admission. Children in this subgroup were signicantly more likely to have been healthy or had only mild
pre-existing disorders than those who died after admission (p=00109). Overall, 45 (64%) children had received
oseltamivir: seven within 48 h of symptom onset.
Interpretation Vaccination priority should be for children at increased risk of severe illness or death from inuenza.
This group might include those with specied pre-existing disorders and those in some ethnic minority groups. Early
pre-hospital supportive and therapeutic care is also important.
Funding Department of Health, UK

1846

Introduction

Methods

Seasonal inuenza can cause infection and treatment


in hospital in children, but mortality is low and the
viruses predominantly aect people older than 65 years
of age.1 However, children have been disproportionately aected by pandemic inuenza A H1N1
compared with older age-groups,2 and this infection
has caused severe disease and death in a minority
of children.3,4
Despite national and global reports of the complications
of pandemic inuenza A H1N1, a detailed analysis of
paediatric mortality has not been provided. Following
the outset of the pandemic in England in April, 2009,
we initiated a condential investigation into all resulting deaths.5 This investigation has provided a real-time
and comprehensive system of national surveillance,
which we have used to examine paediatric deaths in
depth. We aimed to provide important evidence to
strengthen clinical and public health policies for
children during forthcoming inuenza seasons and
future pandemics.

Study population
Mandatory daily reporting systems were established for
all suspected and conrmed deaths from pandemic
inuenza A H1N1 in England. Further deaths were
identied through cross-checking of records held by the
Regional Directors of Public Health and by the Health
Protection Agencys inuenza reference centres. For all
reported cases, a member of the Chief Medical Ocers
clinical team contacted the responsible senior physician
to obtain further details. A death was conrmed as related
to pandemic inuenza A H1N1 if there was laboratory
evidence of H1N1 infection or if H1N1 infection (or
synonym) was recorded on the death certicate. Further
details of the method of death ascertainment have been
reported previously.5

Study design
All validated deaths in children aged less than 18 years
were extracted from this dataset. The clinician responsible
for each individual child was interviewed by telephone by
www.thelancet.com Vol 376 November 27, 2010

Articles

60 000

Number of deaths
Midpoint case estimates

Number of deaths in children

40 000

6
5

30 000
4
3

20 000

2
10 000

Midpoint case estimates (under 18 years)

50 000
7

1
0

22 29 6 13 20 27 3 10 17 24 31 7 14 21 28 5 12 19 26 2 9 16 23 30 7 14 21 28 4 11 18 25 1 8 15 22 1 8 15 22
July,
August,
September,
October,
November,
December,
January,
February,
March,
June,
2009
2009
2009
2009
2009
2009
2010
2010
2010
2009

Date (week beginning)

Figure: Estimated incidence of pandemic inuenza A H1N1 cases and conrmed deaths in children in England, by week
Data for case estimates provided by the Health Protection Agency.

a paediatrician working for the Chief Medical Ocer. The


data collection proforma is available in the webappendix.
Information gathered included: demographic characteristics, pre-existing disorders and past medical history,
and presenting symptoms and clinical course. If results
of microbiological, haematological, and radiological
investigations were not immediately available from
clinicians, they were obtained from hospital or public
health laboratories. Information was cross-checked with
coroners reports. In cases for whom past medical
information was unavailable, supplementary information
was obtained from the general practitioner or local
hospital. When the presence or absence of a symptom or
pre-existing disorder was not specied, we assumed that
none was present.
Collection and processing of data was undertaken
under the Health Service (Control Of Patient Information)
Regulations 2002 (Statutory Instrument 1438) for the
purpose of communicable disease control in England.
Therefore, consent from patients was not needed.
The primary outcome measures were population
mortality rates (by age-group and for children with preexisting disorders) and case-fatality rates. Secondary
measures included presenting symptoms and signs, preexisting disorders, and subsequent complications.
The physical status classication scale of the American
Society of Anaesthesiologists6 was used to establish the
general health of the children before acute illness.
Children were classied as being healthy (grade one) or
having mild (grade two), moderate (grade three), or
severe (grade four or ve) pre-existing systemic disease.
Specic pre-existing disorders (eg, developmental delay)
and the severity of presenting symptoms (eg, dyspnoea)
were classied as absent, mild, moderate, or severe by
the clinicians who cared for the child.
www.thelancet.com Vol 376 November 27, 2010

Recurrent hospital admission was dened as an


inpatient stay for longer than 6 weeks, or greater than
four admissions in the previous year. Bacterial sepsis was
conrmed if bacterial growth was identied on laboratory
culture after clinical suspicion of bacterial infection.
Bacterial sepsis was presumed by clinicians if systemic
features of bacterial infection were present (ie, rising
inammatory markers or signs of focal infection) but
bacterial cultures were negative.
Age-stratied cases of pandemic inuenza A H1N1 were
estimated every week by the Health Protection Agency
with methods described previously.5 Population numbers
for England by age-group and ethnic origin are estimated
by the Oce for National Statistics. The most recent
estimates were used to provide denominators for population mortality rates stratied by age7 and ethnic origin.8
To assess the risk of death associated with pre-existing
disorders that would confer an increased risk of seasonal
inuenza, data from the Department of Health vaccine
monitoring system were used to estimate the number of
people in England aected by these disorders. This
system uses primary care records to estimate the number
of patients with specic grouped disorders by age band.
These estimates are available only for children aged at
least 6 months. Deaths in those aged less than 6 months
were therefore excluded from this analysis.

See Online for webappendix

For more on these Regulations


see http:www.legislation.gov.uk/
UKsi/2002/1438/contents/made

Statistical analyses
Case-fatality rates were calculated for every age-group
with Health Protection Agency midpoint case estimates.
The upper and lower estimates by age were used to
calculate lower and upper estimates for the case-fatality
rates, respectively. A 95% condence interval was
calculated around these estimates, to account for the
uncertainty around the number of cases. The case-fatality
1847

Articles

Number
of deaths

Fatality rate per 100 000 cases of pandemic


inuenza A H1N1 (range)

Mortality rate per


million population
(95% CI)

< 1 year

151 (34635)

14 (626)

14 years

15

33 (9114)

6 (310)

59 years

22

17 (554)

8 (512)

1015 years

15

10 (335)

4 (27)

1617 years

26 (6102)

7 (313)

017 years

70

19 (751)

6 (58)

Table 1: Case-fatality rates for pandemic inuenza A H1N1 infection and population mortality rates by
age in England from June 26, 2009, to March 22, 2010

Number of deaths
Fever

60 (86%)

Cough

47 (67%)

Dyspnoea

46 (66%)

Dehydration

18 (26%)

Altered mental status or encephalopathy

17 (24%)

Diarrhoea

15 (21%)

Nausea or vomiting

14 (20%)

Seizure or convulsion

5 (7%)

Headache

4 (6%)

Sore throat

4 (6%)

Table 2: Frequency of presenting symptoms and signs among children


who died from pandemic inuenza A H1N1 infection

ranges presented refer to the upper 95% condence


interval of the upper case-fatality rate estimate and the
lower 95% condence interval of the lower estimate.
IQRs were calculated for all other ranges other than casefatality rates. Population mortality rates were calculated
by age and ethnic group. Mortality rates for specied preexisting disorders were calculated and age standardised
to the English population.
Fishers exact test or the test with Yates correction
were used to test signicance of data for categorical data
(eg, ethnic origin, those with bacterial sepsis, or those on
pre-hospital antivirals). The MannWhitney U test was
used for non-parametric data. Analyses were done with
SPSS (version 12.0) and R (version 2.9.2).

Role of the funding source


This work was done as part of the Department of Healths
public health response to the inuenza pandemic in
England. No additional funding was sought. The funding
source had no direct involvement in study design; in the
collection, analysis, and interpretation of data; in the
writing of the report; or in the decision to submit the paper
for publication. Both authors had full access to all the data
in the study and the corresponding author had nal
responsibility for the decision to submit for publication.

Results
A total of 70 paediatric deaths related to pandemic
inuenza A H1N1 occurred in England during the study
1848

period (gure). All cases were laboratory conrmed.


Complete demographic and clinical details were obtained
for all deaths. Deaths were evenly distributed between
male (31) and female (39) children. Median age at death
was 7 years (range 02517 years; IQR 212 years). Casefatality and population mortality rates were highest in
children aged less than 1 year (table 1).
Bangladeshi or British Bangladeshi children (47 deaths
per million population [95% CI 17103; n=6]) and
Pakistani or British Pakistani children (36 [1864; n=11])
had the highest population mortality rates, compared
with white British children (4 ([36; n=37]). Pre-existing
health status did not dier between ethnic groups.
Calculation of case-fatality rates by ethnic group was not
possible because estimates of case numbers were not
available by ethnic origin.
Predominantly respiratory symptoms were reported on
presentation in 53 of 70 (76%) deaths. Fever, cough, and
dyspnoea were the most frequent symptoms on
presentation (table 2). The extent of dyspnoea was rated
by clinicians as moderate in 28 (40%) and severe in
12 (17%) deaths. Mainly gastrointestinal symptoms were
described in two (3%) deaths. Among hospital deaths,
most cases (42 of 67; 63%) initially presented to the
emergency department with symptoms of inuenza and
16 (24%) presented in cardiorespiratory arrest. The rest
presented in shock (ve; 7%), status epilepticus (three;
4%), or with non-specic symptoms (one; 1%).
Deaths were categorised into those who deteriorated
rapidly and died before, or at the point of, hospital
admission (early deaths [19 of 70; 27%]), and those who
died after hospital admission (late deaths [51; 73%]). The
early deaths group had a signicantly greater proportion
of children with no, or mild, pre-existing disorders
(table 3). Of the early deaths, nine of 19 (47%) had
consulted a general practitioner during their illness.
Before presentation, fever (47 of 51 vs 13 of 19, p=0020
[Fishers test]) and dyspnoea (40 of 51 vs six of 19, p=0002)
were more common among late deaths than among early
deaths. Nausea and vomiting were less common among
late deaths (seven of 51 vs seven of 19, p=0045). Frequency
of other symptoms did not dier between the two groups.
Children with severe pre-existing disorders accounted
for 64% (45 of 70) of deaths. However, 21% (15 of 70) of
children who died were previously healthy. Mild preexisting disorders were present in 3% (two) of deaths and
moderate disorders in 11% (eight) of deaths. For those
with data available, 36% (23 of 62) had a history of
recurrent hospital admissions. For most deaths in
children aged less than 2 years, moderate or severe preexisting disorders had been reported (87%; 13 of 15). In
those older than 2 years, 73% (40 of 55) had moderate or
severe pre-existing disorders.
Chronic neurological, gastrointestinal, or respiratory
disease were all present in more than half of all deaths
(table 4). Of those with chronic neurological disease,
19 had spastic quadriplegic cerebral palsy. Severe
www.thelancet.com Vol 376 November 27, 2010

Articles

gastro-oesophageal reux was reported in 11 of the


70 (16%) deaths. Regular gastrostomy or nasogastric
feeding took place in 41 of 70 (59%) deaths. A diagnosis
of asthma (requiring any treatment) was reported in
ve of 70 (7%) deaths. Complex congenital cardiac
disease was reported in eight of 70 (11%) deaths. Almost
a quarter of all children had been born prematurely
(17 of 70; median gestation 32 weeks, IQR 27535).
48 (69%) children who died had developmental delay,
which had been severe in 31 (44%). Of the 46 children
of school age (518 years), 23 had attended a school for
children with special needs.
Death rates from pandemic inuenza A H1N1 for
children with pre-existing disorders were higher than for
healthy children. Chronic neurological disease and
kidney disease conferred the highest risk of any of the
disorders studied (table 5). The absence of denominator
data prevented the calculation of an age-standardised
mortality rate for gastrointestinal disease. Two children
with severe underlying disorders received the pandemic
inuenza A H1N1 vaccine 48 h before developing their
terminal illness. No other child was vaccinated against
pandemic inuenza A H1N1.
Median duration from symptom onset to admission (or
death if not formally admitted to hospital) was 2 days
(IQR 15). Five children with inter-current infection
contracted during long-term hospital stays were not
included in this calculation. Median time from symptom
onset to death was 75 days (IQR 313). Of the children
who died, 40 were admitted to critical care units (38 to
intensive care; two to high-dependency units). Of these,
27 were transferred from other hospitals for specialist care.
Transfer to critical care occurred a median of 1 day (IQR
02) after hospital admission. In the children admitted to
critical care, death occurred a median of 6 days later (IQR
211). Within the hospital, the critical care facility was most
commonly the place of death (40 of 67; 60%), followed by
the emergency department (17 of 67; 25%).
Organ failure occurred frequently. 40 (57%) children
developed respiratory failure that needed mechanical
ventilation: seven needed non-invasive ventilation, ten
invasive positive pressure ventilation, 21 high-frequency
oscillation, and two extra-corporeal membrane
oxygenation. High ventilation pressures were usually
needed and pneumothoraces occurred in six of the
40 (15%) children, all of whom had insertion of a chest
drain. Intracerebral haemorrhage arose secondary to extracorporeal membrane oxygenation in one child. Circulatory
failure needing inotropic support occurred in 30 of
70 (43%) children. Acute renal failure necessitated renal
replacement therapy (haemoltration or haemodialysis)
in ve of 70 (7%) children. Acute haematological
abnormalities were recorded in 25 (36%) children before
death. These abnormalities were thrombocytopenia (16;
23%), lymphopenia (14; 20%), neutrophilia (three; 4%),
neutropenia (two; 3%), pancytopenia (two; 3%), and
disseminated intravascular coagulation (two; 3%).
www.thelancet.com Vol 376 November 27, 2010

Early death group


(n=19)
Median age (years; IQR)

6 (214)

Late death group


(n=51)

p value

7 (2512)

09761*

Ethnic origin
White British
Asian or Asian British (any)

12 (63%)

25 (49%)

6 (32%)

21 (41%)

07547

Black or Black British (any)

1 (5%)

3 (6%)

Healthy or mild pre-existing disorders

9 (47%)

8 (16%)

00109

Presumed or conrmed bacterial sepsis

9 (47%)

19 (37%)

05842

Pre-hospital antivirals

2 (10%)

1 (2%)

Duration from symptom onset to death (days;


median, IQR)

2 (125)

10 (55165)

<00001*

Duration from symptom onset to rst medical


consult (days; median, IQR)

1 (0252)

2 (0375)

04533*

01770

Data are number of patients (%) unless otherwise stated. *Mann-Whitney U test; 2 with Yates correction; Fishers test.

Table 3: Characteristics of children presenting with rapid deterioration leading to death before, or at the
point of, hospital admission (early death) and of those who died after hospital admission (late death)

Number of deaths with Median age at death


pre-existing disorders (years; IQR)
(n=70)
Any chronic respiratory disease

37 (53%)

7 (212)

Chronic lung disease

13 (19%)

4 (15)

Home oxygen

12 (17%)

4 (0548)

Scoliosis with restrictive lung disease

10 (14%)

12 (9512)

Pulmonary hypertension

6 (9%)

14 (06511)

Tracheostomy

6 (9%)

6 (2251025)

Asthma needing preventive therapy

4 (6%)

11 (814)

Home ventilation

3 (4%)

11 (5611)

Any chronic cardiac disease

17 (24%)

5 (114)

Congenital cardiac disease

12 (17%)

Cardiomyopathy

3 (4%)

16 (07710)
16 (1216)

Any chronic gastrointestinal disease or feeding problem

41 (59%)

6 (1510)

Exclusive gastrostomy or nasogastric feeding

33 (47%)

6 (29)

Gastro-oesophageal reux

30 (43%)

7 (898)

Unsafe swallow

24 (34%)

7 (389)

Any chronic neurological disease

38 (54%)

7 (298)

Cerebral palsy

22 (31%)

7 (49)

Epilepsy

22 (31%)

8 (612)

Any chronic endocrine disease

7 (10%)

6 (2713)

Any chronic kidney disease

4 (6%)

Any chronic haematological disease

5 (7%)

Any chromosomal disorder

5 (7%)

14 (816)

Any malignant disease

1 (1%)

16*

Other
75 (1427)
16 (1316)

*No IQR as only one child in this group.

Table 4: Frequency of pre-existing disorders in paediatric deaths related to pandemic inuenza A


H1N1 infection

45 of 70 (64%) children received oseltamivirseven


within 48 h of symptom onset (median 5 days [IQR 3775]
between symptom onset and treatment with antiviral
drugs) and three before admission. Oseltamivir resistance
was recorded in two childrenone ante-mortem. Viral
1849

Articles

Deaths

Population prevalence Age-standardised mortality rate per


estimates (thousands) million population (95% CI)

Chronic respiratory disease

33

392

Chronic cardiac disease

16

72

232 (132377)

Chronic neurological disease

35

31

1536 (9882242)

Immunosuppression

38

166 (66343)

Chronic kidney disease

449 (851333)

Diabetes mellitus
None

13

10 125

167 (89268)

0 ()
13 (0722)

Table 5: Age-standardised mortality rate for deaths related to pandemic inuenza A H1N1 for children
aged between 6 months and 18 years with specic pre-existing disorders

pneumonitis was reported in 29 of 70 (41%) children and


bacterial pneumonia in 20 of 70 (29%). Bacterial coinfection was clinically presumed in 14 (20%) deaths and
conrmed in a further 14. The most common sites of
bacterial infection were the lung (20), blood (six), and
cerebrospinal uid (two). Pathogens most frequently
isolated were group A Streptococcus (three cases),
S pneumoniae (three cases), S aureus (three cases), and
Haemophilus inuenzae (three cases). Excluding longterm inpatients, pre-hospital antibiotics had been given
to 23 of the 65 children who died (35% of deaths). All
inpatients received antibiotics. Most patients admitted
received intravenous antibiotics within 48 h of admission
(43 of 45; 96%) or within 48 h of symptom onset for longterm inpatients (three of ve; 60%). One case of
myocarditis was reported. According to clinical reports or
death certication, acute respiratory distress syndrome
occurred in 20 children. Intravenous corticosteroids were
given in 18 cases.
Death certicates were available for 66 cases. In
46 (70%) cases, pandemic inuenza A H1N1 or a synonym
was recorded as the direct cause of death. In seven (11%)
cases, pandemic inuenza A H1N1 or a synonym was
recorded as contributing to death. There was no record of
pandemic inuenza A H1N1 on 13 death certicates; in
ve of which, positive virological test results were
received after the death certicate was issued.

Discussion
Our report of the 70 deaths in children in England related
to pandemic inuenza A H1N1 has shown that mortality
disproportionately aects ethnic minorities and those
with pre-existing disorders. Many deaths occurred before
hospital admission, especially in healthy children or
those with only mild pre-existing disorders.
The overall childhood mortality rate for pandemic
inuenza A H1N1 reported here (six per million
population) is close to that for the Netherlands (ve per
million population for children under 14 years9) but lower
than that in Argentina (11 per million population).3 A high
disease burden was reported in the southern hemisphere
because the pandemic coincided with the inuenza
season in that region. Delayed presentation and
unrecognised illness might have contributed further to
1850

the high mortality rate in Argentina.3 Mortality from


seasonal inuenza in children is lower than that reported
for pandemic inuenza A H1N1. Extrapolated data from
population mortality estimates in England show a
mortality rate from seasonal inuenza of two per million
population for children aged less than 14 years,10 which is
much the same as for international estimates.11 The occurrence of 70 deaths from pandemic inuenza A H1N1 in
children in 1 year in England is greater than the number
of deaths in children every year from leukaemia, and this
high childhood mortality was last seen for a single
infectious disease (meningococcal disease) in 2001.12
We identied the highest population mortality and
case-fatality rates in children aged less than 1 year. A
similar pattern has been reported from other countries.3
The high population mortality rates that we identied
among Bangladeshi and Pakistani groups are consistent
with a US report that ethnic minority groups were
disproportionately aected by severe illness.13 This
nding might be attributable to clustering of pandemic
inuenza A H1N1 cases in areas of England with high
ethnic minority populations (such as London and the
West Midlands),2 although other areas with lower ethnic
minority populations such as the East Midlands and
Yorkshire were also greatly aected. An increased
occurrence of pre-existing disorders might exist in ethnic
minority children, although no reliable data are available
to assess this claim. The extent to which ethnic origin
itself, through genetic or cultural factors, confers
increased susceptibility to severe illness or death remains
unclear, and further investigation is needed.
We identied two distinct clinical patterns for deaths in
children from pandemic inuenza A H1N1. The rst
group had early deterioration with death before hospital
admissionmainly children with no or mild pre-existing
disorders. Although these patients had not presented late
to medical services, the threshold for hospital admission
of these children is likely to have been higher than for
those with severe pre-existing disorders. These children
might have been cared for at home in the expectation
that their illness would be self-limiting. Children dying
early did not have more co-existing bacterial infection
than those who died later, implying that these children
had severe acute viral infection. A fulminant course has
previously been described in a cohort of children with
pandemic inuenza A H1N1 infection who were admitted
to critical care after presenting with shock.14 The
mechanism of organ failure in these children is unclear.
However, they did have a substantially greater frequency
of nausea or vomiting at or before presentation, which
could indicate extensive viral replication, leading to early
deterioration and death.1517
The second group of deaths was children with severe
underlying disorders who deteriorated during hospital
admission. Those with neurological disease, particularly cerebral palsy, were at high risk, as described in the
USA.4 However, we also noted a high frequency of
www.thelancet.com Vol 376 November 27, 2010

Articles

gastro-oesophageal reux, gastrostomy feeding, and


unsafe swallowing (when patients do not have a protective
gag reex) that frequently co-exist with severe cerebral
palsy. The increased risk associated with chronic
neurological disease might be explained partly by
secondary aspiration and low respiratory reserves in these
children. Because of small case numbers and no
denominator data, the relative risk of death for cerebral
palsy in the absence and presence of associated respiratory
and gastrointestinal complications could not be assessed.
Good-quality data for chronic disease prevalence
derived from the National Health Service primary care
services enabled the calculation of disease-specic
mortality rates. Our analysis showed the highest
population mortality rate in children with chronic
neurological disease, consistent with the high proportion
of children with neurological disease among the deaths.
However, although absolute numbers of deaths from
chronic kidney disease were low, the population rate of
death was fairly high (although with wide condence
intervals) because of a low population prevalence.
We noted that asthma was present in only 7% of
paediatric deaths. The prevalence of childhood asthma in
England is estimated at up to 14%.18 Other investigators
have suggested that asthma is associated with increased
mortality from pandemic inuenza A H1N1 in children3
and adults.9 Our ndings contradict this notion. Although
there were deaths in children with asthma, the high
population prevalence of asthma will lower the diseasespecic mortality rate. An explanation for our observations
might be that asthma does not increase the risk of death.
Alternatively, there might have been a low threshold for
admission and aggressive therapy in these patients because
of clinical and parental concern; this explanation is lent
support by the high numbers of children with asthma who
were admitted to hospital and critical care during the
pandemic.3,19 If this explanation is true, a wider use of early
aggressive treatment measures could reduce mortality.
Our study conrms that most children who died
presented with typical respiratory symptoms rather than
atypical features that could aid prognosis. Thrombocytopenia and relative lymphopenia, however, might be
early predictors of severe illness in children, which lends
support to similar ndings in adults20 and children treated
in hospital.21 Conrmed bacterial infection was present in
20% of cases in our studymainly pneumonia, which is
in accord with data for this pandemic in the USA4 and for
seasonal inuenza,22 with similar pathogens described.
Early treatment with antibiotics in children with clinical
deterioration is important.23
Only 10% of children in this study received antiviral
therapy within 48 h of symptom onset, which is a similar
number to reports from Argentina in children treated in
hospital.3 Antiviral use for the treatment of inuenza in
children is controversial.24 However, these drugs are most
eective if given within 48 h of symptom onset.25 Our
ndings suggest that children with pre-existing disorders
www.thelancet.com Vol 376 November 27, 2010

are at an increased risk of death. For many of these


children, antiviral therapies are one of only a few
therapeutic options available if the children are not
vaccinated. Although our study was not designed to assess
antiviral use, early treatment with antiviral therapy might
maximise the eectiveness of this treatment. This approach
is especially important in primary care because 27% of
children in our study died before admission to hospital.
Further investigation into the contribution of pre-hospital
antivirals to the outcome of aected children is needed.
Our ndings support the vaccination of children
against pandemic inuenza A H1N1. Children at greatest
risk of severe illness or death should be prioritised. Our
data indicate that risk groups include children with preexisting illness (including chronic neurological or
gastrointestinal disease) and those in ethnic minority
groups (including Bangladeshi and Pakistani children).
However, our ndings also suggest that protection cannot
be conned to risk groups as 21% of deaths in our cohort
occurred in healthy children.
Complete death ascertainment is dicult to achieve.
However, we believe our reporting system achieved a
high level. We have classied deaths according to those
occurring before and subsequent to inpatient admission.
This classication will not capture deaths occurring
within hours of hospital admission; therefore, the true
early death group might be larger. By using formal
hospital admission, we have been able to examine deaths
occurring too rapidly to enable stabilisation.
Case-fatality rates are subject to limitations.5 Case
estimates in England take into account laboratoryconrmed cases in primary care, in addition to estimates
of the proportion of symptomatic individuals who do not
seek medical attention. The estimates of case fatality in
this study have wide ranges, indicating uncertainty about
those not seeking medical help. If this proportion is
higher than estimated, our case-fatality rates might be
an overestimate. However, this method of case estimation
is likely to provide a more meaningful denominator for
case fatality than use of laboratory-conrmed cases.
Few countries hold comprehensive prevalence data for
chronic diseases in the population. The National Health
Service has a strong primary care system and a
longstanding vaccination programme that relies on
pooled data for the occurrence of risk factors. Although
these data are derived from a sample of clinical practices,
we believe that they provide reliable estimates of
prevalence. Additionally, we have interpreted derived
mortality rates comparatively rather than in absolute
terms. More extensive comparisons of risk would be
possible with a control group.
Although this is a large study of paediatric deaths
related to pandemic inuenza A H1N1, more detailed
analysis would be possible with higher case numbers.
An international study that pooled data on childhood
deaths from pandemic inuenza A H1N1 would be
greatly informative.
1851

Articles

Contributors
LJD planned this study, oversaw its design, and contributed to writing
the paper. NS planned and designed the data collection process, obtained
and analysed all data, and drafted the manuscript.
Conicts of interest
LJD was the Chief Medical Ocer for England from 1998 to May, 2010.
In this role he advised the government on public health policy, including
the management of the pandemic. NS supported him in this task from
2009 to 2010. Both authors declare that they have no additional conicts
of interest.
Acknowledgments
We thank the many clinicians in England who supplied the clinical
information about their patients; Matthew Mak, Emma Stanton,
Paul Rutter, and Tara Fajeyisan for their work in the validation of all
deaths; Oliver Mytton for assistance with data analysis; and Iain Yardley
for assistance in editing the manuscript. This work was done as part of
the public health response to pandemic inuenza in England. The costs
of the study were small and met from the Chief Medical Ocers budget
within the Department of Health without the need for additional funding.
References
1
Pitman RJ, Melegaro A, Gelb D, Siddiqui MR, Gay NJ,
Edmunds WJ. Assessing the burden of inuenza and other
respiratory infections in England and Wales. J Infect 2007;
54: 53038.
2
Miller E, Hoschler K, Hardelid P, Stanford E, Andrews N,
Zambon M. Incidence of 2009 pandemic inuenza A H1N1
infection in England: a cross-sectional serological study.
Lancet 2010; 375: 110008.
3
Libster R, Bugna J, Coviello S, et al. Pediatric hospitalisations
associated with 2009 pandemic inuenza A (H1N1) in Argentina.
NEJM 2010; 362: 4555.
4
CDC. Surveillance for pediatric deaths associated with 2009
pandemic inuenza A (H1N1) virus infectionUnited States,
AprilAugust 2009. MMWR Morb Mortal Wkly Rep 2009; 58: 94147.
5
Donaldson LJ, Rutter PD, Ellis BM, et al. Mortality from pandemic
A/H1N1 2009 inuenza in England: public health surveillance
study. BMJ 2009; 339: b5213.
6
Saklad M. Grading of patients for surgical procedures.
Anaesthesiology 1941; 2: 28184.
7
Oce for National Statistics. Mid-2008 population estimates:
England and Wales. http://www.statistics.gov.uk/statbase/product.
asp?vlnk=15106 (accessed Feb 25, 2010).
8
Oce for National Statistics. Mid-2007 population estimates
ethnicity tables: England and Wales. http://www.statistics.gov.uk/
statbase/product.asp?vlnk=14238 (accessed Feb 25, 2010).
9
van t Klooster TM, Wielders CC, Donker T, et al. Surveillance of
hospitalisations for 2009 pandemic inuenza A(H1N1) in the
Netherlands, 5 June31 December 2009. Eurosurv 2010;
15: pii=19461.
10 Pitman RJ, Melegaro A, Gelb D, Siddiqui MR, Gay NJ, Edmunds WJ.
Assessing the burden of inuenza and other respiratory infections in
England and Wales. J Infect 2007; 54: 53038.

1852

11

12

13

14

15
16

17

18

19

20

21

22
23

24

25

Bhat N, Wright JG, Broder KR, et al. Inuenza-associated deaths


among children in the United States, 20032004. NEJM 2005;
353: 255967.
Oce for National Statistics. Annual mortality statistics: deaths
registered in 2001-2008, England and Wales. ONS 2002-9.
http://www.statistics.gov.uk/statbase/product.asp?vlnk=15096
(accessed Feb 25, 2010).
Centers for Disease Control and Prevention. Deaths related to 2009
pandemic inuenza A (H1N1) among American Indian/Alaska
natives12 states, 2009. MMWR Morb Mortal Wkly Rep 2009;
58: 134144.
Lister P, Reynolds F, Parslow R, et al. Swine-origin inuenza virus
H1N1, seasonal inuenza virus, and critical illness in children.
Lancet 2009; 374: 60507.
To KK, Chan KH, Li IW, et al. Viral load in patients infected with
pandemic H1N1 2009 inuenza A virus. J Med Virol 2009; 82: 17.
Turner SJ, Brown LE, Doherty PC, Kelso AQA. What have we
found out about the inuenza A (H1N1) 2009 pandemic virus?
J Biol 2008; 8: 69.
Maines TR, Jayaraman A, Belser JA, et al. Transmission and
pathogenesis of swine-origin 2009 A(H1N1) inuenza viruses in
ferrets and mice. Science 2009; 325: 48487.
Butland BK, Strachan DP, Crawley-Boevey EE, Anderson HR.
Childhood asthma in South London: trends in prevalence and use
of medical services 19912002. Thorax 2006; 61: 38387.
ORiordan S, Barton M, Yau Y, Read SE, Allen U, Tran D. Risk
factors and outcomes among children admitted to hospital with
pandemic H1N1 inuenza. CMAJ 2010; 182: 39-44.
Cunha BA, Pherez FM, Schoch P. Diagnostic importance of relative
lymphopenia as a marker of swine inuenza (H1N1) in adults.
Clin Infect Dis 2009; 49: 145456.
Cao B, Li XW, Mao Y, et al. Clinical features of the initial cases of
2009 pandemic inuenza A (H1N1) virus infection in China.
NEJM 2009; 361: 250717.
Johnson BF, Wilson LE, Ellis J, et al. Fatal cases of inuenza A in
childhood. PLoS One 2009; 4: e7671.
Wright PF, Kirkland KB, Modlin JF. When to consider the use of
antibiotics in the treatment of 2009 H1N1 inuenza-associated
pneumonia. NEJM 2009; 361: e112.
Shun-Shin M, Thompson M, Heneghan C, Perera R, Harnden A,
Mant D. Neuraminidase inhibitors for treatment and prophylaxis of
inuenza in children: systematic review and meta-analysis of
randomised controlled trials. BMJ 2009; 339: b3172.
WHO. WHO guidelines for pharmacological management of
pandemic inuenza A(H1N1) 2009 and other inuenza viruses.
Part 1: recommendations. http://www.who.int/csr/resources/
publications/swineu/h1n1_guidelines_pharmaceutical_mngt.pdf
(accessed Sept 19, 2010).

www.thelancet.com Vol 376 November 27, 2010

Articles

Causes of neonatal and child mortality in India: a nationally


representative mortality survey
The Million Death Study Collaborators*

Summary
Background More than 23 million children died in India in 2005; however, the major causes of death have not been
measured in the country. We investigated the causes of neonatal and child mortality in India and their dierences by
sex and region.
Methods The Registrar General of India surveyed all deaths occurring in 200103 in 11 million nationally
representative homes. Field sta interviewed household members and completed standard questions about events
that preceded the death. Two of 130 physicians then independently assigned a cause to each death. Cause-specic
mortality rates for 2005 were calculated nationally and for the six regions by combining the recorded proportions for
each cause in the neonatal deaths and deaths at ages 159 months in the study with population and death totals from
the United Nations.
Findings There were 10 892 deaths in neonates and 12 260 in children aged 159 months in the study. When these
details were projected nationally, three causes accounted for 78% (079 million of 101 million) of all neonatal deaths:
prematurity and low birthweight (033 million, 99% CI 031 million to 035 million), neonatal infections
(027 million, 025 million to 029 million), and birth asphyxia and birth trauma (019 million, 018 million to
021 million). Two causes accounted for 50% (067 million of 134 million) of all deaths at 159 months: pneumonia
(037 million, 035 million to 039 million) and diarrhoeal diseases (030 million, 028 million to 032 million). In
children aged 159 months, girls in central India had a ve-times higher mortality rate (per 1000 livebirths) from
pneumonia (209, 194226) than did boys in south India (41, 3056) and four-times higher mortality rate from
diarrhoeal disease (177, 162193) than did boys in west India (41, 3055).

Lancet 2010; 376: 185360


Published Online
November 12, 2010
DOI:10.1016/S01406736(10)61461-4
See Comment page 1810
*Writing committee listed at end
of Article and members in
webappendix p 7
Correspondence to:
Prof Prabhat Jha, Centre for
Global Health Research,
Li Ka Shing Knowledge Institute,
St Michaels Hospital and Dalla
Lana School of Public Health,
University of Toronto, Toronto,
ON M5B 1W8, Canada
prabhat.jha@utoronto.ca

Interpretation Five avoidable causes accounted for nearly 15 million child deaths in India in 2005, with substantial
dierences between regions and sexes. Expanded neonatal and intrapartum care, case management of diarrhoea and
pneumonia, and addition of new vaccines to immunisation programmes could substantially reduce child deaths in India.
Funding US National Institutes of Health, International Development Research Centre, Canadian Institutes of Health
Research, Li Ka Shing Knowledge Institute, and US Fund for UNICEF.

Introduction
Yearly child mortality rates in India have fallen between
17%1 and 23%2 in the past two decades. Despite this
decrease, the United Nations (UN) estimates that about
235 million children died in India in 2005. This gure
corresponds to more than 20% of all deaths in children
younger than 5 years worldwide, which is more than in
any other country.1,3 Large dierences in overall child
survival between Indias diverse regions have been
previously documented.4,5 However, no direct and nationally
representative measurement of the major causes of death
in neonates (<1 month) and at ages 159 months has been
done,6 and how these causes of death vary across Indias
regions is unknown. Social preference for boys is strong,
as noted by widespread selective abortion of female fetuses7
and by lower immunisation rates in girls.8 The
consequences of boy preference on child mortality remain
undocumented. Understanding of the causes of child
death might, therefore, help to guide the use of widely
practicable interventions for neonatal and child survival.3,9
Most deaths in India, including of children, are not
medically certied since most occur at home, in rural
www.thelancet.com Vol 376 November 27, 2010

areas, and without attention by a health-care worker.10


Thus, other sources of information are needed to help to
establish the probable underlying causes of death. During
the past decade the Registrar General of India (RGI)
has introduced an enhanced form of verbal autopsy
called RHIMEor routine, reliable, representative, resampled household investigation of mortality with medical
evaluation11into its nationally representative sample
registration system (SRS), which covered about 63 million
people and monitored all deaths in 11 million homes.5
This mortality survey is part of the Million Death Study,
which seeks to assign causes to all deaths in the SRS areas
during the 13 years from 2001 to 2013.1115 In this report we
present the results of the causes of child deaths in India,
separately for the neonatal period and at ages 159 months,
for boys and girls, and for each of six major regions
of India.

Methods
Study setting and procedures
Details of the design, methods, and preliminary results of
the Million Death Study have been previously published.1115
1853

Articles

North
194 million births
013 million deaths
Jammu and
Kashmir
Central
1028 million births
111 million deaths

East
619 million births
054 million deaths

Himachal
Pradesh
Punjab
Haryana

Uttarakhand*
Delhi
Sikkim

Rajasthan*

Uttar Pradesh

Assam*

Nagaland

Bihar*
Meghalaya
Jharkhand*

Tripura

Manipur
Mizoram

Ch

ha

tti

sg

arh

Madhya Pradesh*

Gujarat

West
Bengal

randomly, on the basis of the language of the narrative,


to two of 130 collaborating physicians trained in disease
coding who, working independently, assessed the
probable underlying cause of death and assigned a
three-character code from the International Classication
of Diseases tenth revision (ICD-10)17 with use of
structured guidelines for each major disease group.18
Disagreements about the ICD-10 codes assigned were
resolved by anonymous reconciliation (ie, asking each
physician to reconsider); persisting dierences were
adjudicated by a third physician. Separate classication
systems were developed for the causes of neonatal
deaths and at ages 159 months, based on input from
the Child Health and Epidemiology Reference Group9
(webappendix pp 25).

Orissa*
Northeast
099 million births
009 million deaths

Maharashtra

West
344 million births
021 million deaths

Andhra Pradesh

Kamataka
South
446 million births
027 million deaths
Tamil Nadu
la

Kera

Figure 1: Yearly number of livebirths and deaths in children aged 04 years in India, by region, 2005
*These lower-income states are known as the Empowered Action Group plus Assam (EAGA) states.

See Online for webappendix

India was divided into about 1 million areas for the


1991 census, each with about 1000 inhabitants. The RGI
chose 6671 of these areas randomly for the SRS in 1993; in
each area all individuals and their household characteristics
were documented and subsequent births and deaths (but
not cause of death) were documented every month by a
part-time enumerator resident in that area, and independently surveyed twice a year by one of 800 full-time
RGI surveyors (trained non-medical graduates). Each of
these RGI surveyors has visited, since 2002, each SRS area
periodically to record from families (or other informants) a
written narrative in the local language describing the
events that preceded the death, in addition to answers to
standard questions about key symptoms. Separate forms
were used for neonatal deaths and deaths in children aged
between 1 month and 14 years, on the basis of a WHO
multicountry validation study of verbal autopsy for
common causes of childhood deaths.16 Forms were pretested in about 500 child deaths in India.11 Random resampling and other eldwork quality control methods
were used.11

Central medical coding of causes of death


Each of the local language narratives and corresponding
symptom data were electronically scanned and sent
1854

National and subnational mortality rates


The age-specic and sex-specic proportions of each
cause of death were calculated (weighted according to
the SRS sampling fractions in the rural and urban areas
of each state5). We applied the proportions of each cause
of death to the independent UN Population Division
estimates of deaths (235 million) and livebirths
(273 million) in India in 2005,1 to calculate age-specic
and sex-specic mortality rates (per 1000 livebirths) and
absolute deaths by cause. The UN totals were used
because the SRS slightly underestimates child mortality
rates19 and because about 12% of the SRS-enumerated
deaths were not interviewed, mostly because of
migration by the family. The UN totals for 2005 were
used because these data were most complete, could be
compared with the available Indian Census projections
for 2006, and were collected before the implementation
of a new national health programme to reduce child
mortality.9 Application of the 200103 proportions to the
2005 total deaths did not introduce major biases since
there was little change in the yearly distribution of
causes of deaths in our survey, or between 2001 and
2004 in an independent survey of medically certied
causes of death from selected urban hospitals.20
To calculate subnational mortality rates, we partitioned
the UN total births and child deaths into 140 strata
(35 states, rural and urban areas, and both sexes) with the
Census of India 2005 population,21 relative SRS birth and
death rates,5 or smaller demographic surveys4 when SRS
data were not available (webappendix pp 12). Subnational
results were produced for the six regions (north, south,
west, central, northeast, and east),12 and for the lowerincome states with historically higher child mortality
rates and poverty levels (Assam, Bihar, Chhattisgarh,
Jharkhand, Madhya Pradesh, Orissa, Rajasthan,
Uttarakhand, and Uttar Pradesh) and the remaining
states (gure 1). 99% CIs for each cause of death
proportion or mortality rate were calculated with Taylor
linearisation22 on the basis of the survey design and the
observed sample deaths in the Million Death Study. All
statistical analyses were done in Stata (version 10.1).
www.thelancet.com Vol 376 November 27, 2010

Articles

Role of the funding source


The sponsor of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. The corresponding author had full
access to all the data in the study and had nal
responsibility for the decision to submit for publication

Results
Of the 24 841 child deaths surveyed, 93% (23 152) were
double-coded by physicians and included in the study
(table). Reasons for exclusion were missing information
about age or sex (n=191), and non-legible forms,

improper scanning of narrative, or incorrect language


code (n=1498). Respondents for the 23 152 child deaths
were the father (n=5117; 22%), mother (8103; 35%),
siblings and other relatives (5047; 22%), grandparents
(3612; 16%), or a neighbour or non-relative (1273; 5%).
Most child deaths occurred in rural areas (table)
irrespective of the cause of death. Only 17%
(3877/23 152) of children died in a health facility, with
large variations between rural and urban areas and
between states (webappendix p 11). Physicians agreed
on the cause of death initially for 62% (14 410/23 152) of
all deaths.23

Study deaths, 200103


Boys

Girls

All India, 2005

Total

Rural area

Died in a
Two coders
health facility immediately
agreed

Mortality rate per 1000


livebirths

Total deaths (thousands)

Boys

Girls

Total

Boys

Girls

Total (99%CI)*

Neonatal (<1 month)


Prematurity and low birthweight

2012

1619

3631

3265

988

2381

130

108

120

185

142

327 (309345)

Neonatal infections

1544

1339

2883

2694

346

1804

103

94

99

145

123

268 (253286)

Birth asphyxia and birth trauma

1219

854

2073

1869

631

946

80

59

70

113

77

190 (176206)

Other non-communicable diseases

316

243

559

502

118

251

20

16

18

28

21

49 (4258)

Congenital anomalies

213

146

359

304

139

202

14

10

12

20

13

33 (2842)

Diarrhoeal diseases

175

162

337

318

26

227

12

12

12

17

15

32 (2640)

Tetanus

149

115

264

255

14

180

13

10

12

18

14

32 (2639)

Injuries

27

20

47

43

15

02

01

02

5 (38)

414

325

739

665

147

329

27

25

24

39

33

72 (6181)

6069

4823

10 892

6335
(582%)

401

335

369

568

440

Pneumonia

1542

1890

3432

3146

404

2546

112

160

135

159

210

369 (348390)

Diarrhoeal diseases

1184

1532

2716

2480

293

2146

89

134

111

126

176

302 (283323)

Measles

308

450

758

687

64

374

25

42

33

36

56

92 (79104)

Other non-communicable diseases

418

433

851

772

142

490

30

35

32

42

46

88 (77100)

Other causes
All causes

9915
(910%)

2417
(222%)

1008

159 months

Injuries

400

357

757

689

91

673

29

29

29

42

38

80 (6892)

Malaria

262

325

587

562

43

354

17

24

20

24

32

56 (4765)

Meningitis/encephalitis

232

209

441

396

94

183

19

19

19

27

25

52 (4362)

Nutritional diseases

141

201

342

303

18

190

11

19

15

16

25

41 (3451)
38 (3147)

Acute bacterial sepsis and severe infections

147

213

360

324

50

117

11

18

14

15

23

Other infectious diseases

143

182

325

298

43

120

10

15

12

14

19

33 (2741)

Other causes

847

844

1691

1490

218

882

64

72

69

91

95

186 (170203)

5624

6636

12 260

11 147
(909%)

1460
(119%)

8075
(659%)

417

567

489

592

745

1337

11 693

11 459

23 152

21 062
(910%)

3877
(167%)

14 410
(622%)

818

902

858

1160

1185

2345

All causes
04 years
All causes

Total livebirths (2005): 273 million; 142 million boys, 131 million girls. Mortality estimates exclude stillbirths, cancelled reports (ie, not coded), and children with missing information about sex or age. The
percentage of deaths that could not be coded was 65% in boys, 56% in girls, 80% in urban areas, and 58% in rural areas. *99% CIs are provided for the causes of death but not for the UN totals for all-cause
child deaths.For prematurity: mortality rate for boys 95 (99% CI 89101), estimated total deaths 135 000; for girls 74 (6980), 97 000 deaths; and for both 85 (8189), 232 000 deaths. Low birthweight:
mortality rate for boys 35 (3139), estimated total deaths 50 000; for girls 34 (3039), 45 000 deaths; and for both 35 (3238), 95 000 deaths. These two conditions are combined because of the diculty in
dierentiating them in verbal autopsies. Infections category includes neonatal pneumonia, sepsis, and CNS infections (about 2000 deaths every year). For neonatal pneumonia: mortality rate for boys
60 (99% CI 5565), estimated total deaths 85 000; for girls 58 (5363), 76 000 deaths; and for both 59 (5563), 161 000 deaths. For sepsis: mortality rate for boys 42 (3847), 60 000 deaths; for girls
36 (3240), 47 000 deaths; and for both 39 (3642), 107 000 deaths. These three conditions are combined because of the diculty in dierentiating them in verbal autopsies. Sample-weighted percentage
of deaths: 871% occurred in a rural area and 166% occurred in a health facility.

Table: Causes of death in neonates and at ages 159 months in this study and estimated national totals

www.thelancet.com Vol 376 November 27, 2010

1855

Articles

The mortality rate in children younger than 5 years was


858 per 1000 livebirths (818 for boys and 902 for girls).
Five causes accounted for 62% (146 million of 235 million)
of all child deaths: pneumonia, prematurity and low
birthweight, diarrhoeal diseases, neonatal infections, and
birth asphyxia and birth trauma (table). The sex (gure 2)
and regional (gure 3) distribution of the leading causes of

child deaths varied in neonates (gure 4) and at ages


159 months (gure 5). Estimates of deaths and mortality
rates for the leading causes of death for the major states of
India are shown in webappendix pp 920.
Three causes accounted for 78% (079 million of
101 million) of all neonatal deaths in India: prematurity
and low birthweight; neonatal infections, comprising

India
(235 million deaths; MR=858)
Pneumonia (16%)
Prematurity and
low birthweight (14%)
Birth asphyxia and
birth trauma (8%)

Neonatal deaths
43%

Other infectious
diseases (11%)

Neonatal
infections* (12%)
Tetanus (1%)

Other (17%)

Other (7%)
(13%)
Boys
(116 million deaths; MR=818)

Diarrhoeal
diseases

(1%)

Girls
(119 million deaths; MR=902)

Pneumonia (14%)

Pneumonia (18%)
Prematurity and
low birthweight (12%)

Prematurity and
low birthweight (16%)
Other infectious
diseases (9%)

Neonatal deaths
49%

Birth asphyxia and


birth trauma (10%)

Neonatal deaths
37%

Birth asphyxia and


birth trauma (7%)

Other infectious
diseases (13%)

Neonatal
infections* (10%)
Tetanus (1%)

Other (17%)

Neonatal
infections* (13%)

(11%)

Diarrhoeal
diseases

Tetanus (2%)
Other (7%)
(1%)

Higher-income states
(071 million deaths; MR=610)

Other (6%)
Other (17%)
(15%)
Lower-income states
(164 million deaths; MR=1041)

Pneumonia (12%)

Pneumonia (17%)

Prematurity and
low birthweight (11%)

Prematurity and
low birthweight (20%)
Other infectious
diseases (10%)

(1%)

Diarrhoeal
diseases

Neonatal deaths
48%
Birth asphyxia and
birth trauma (10%)

Other (21%)

Birth asphyxia and


birth trauma (7%)
Neonatal deaths
41%

Other infectious
diseases (13%)

Neonatal
infections* (13%)
Tetanus (2%)

Neonatal
infections* (8%)

Other (6%)

Other (15%)
Other (9%)
(9%)

Diarrhoeal
diseases

(1%)

(14%)

Diarrhoeal
diseases

(2%)

Figure 2: Causes of death in children aged 04 years in India, by sex and by state income, 2005
MR=mortality rate in children younger than 5 years. *Includes neonatal pneumonia, sepsis, and CNS infections. Lower-income states are Assam, Bihar, Chhattisgarh,
Jharkhand, Madhya Pradesh, Orissa, Rajasthan, Uttarakhand, and Uttar Pradesh; higher-income states are the remaining 26 states/union territories.

1856

www.thelancet.com Vol 376 November 27, 2010

Articles

neonatal pneumonia, neonatal sepsis, and CNS infections;


and birth asphyxia and birth trauma (gure 4). The
proportion of neonatal deaths to total child deaths was
higher in boys than in girls, and in higher-income than in
lower-income states (table and gure 2). The proportion of
neonatal deaths to total child deaths was greatest in areas
with lowest mortality rates in children younger than 5 years

(gure 3). The all-cause neonatal mortality rate was about


20% higher in boys (401) than in girls (335). Furthermore,
neonatal mortality rates were higher for most causes in
boys than in girls, although neonatal mortality rates were
similar between sexes for diarrhoeal diseases (table). The
proportion of total child deaths caused by neonatal
infections was higher in the lower-income states than in
South
(027 million deaths; MR=607)

West
(021 million deaths; MR=601)
Pneumonia (13%)

Pneumonia (8%)
Prematurity and
low birthweight (23%)

Other infectious
diseases (10%)

Prematurity and
low birthweight (20%)

Other infectious
diseases (12%)

Neonatal deaths
50%

Neonatal deaths
46%
Birth asphyxia and
birth trauma (10%)

Birth asphyxia and


birth trauma (10%)

Neonatal
infections* (6%)

Other (24%)

Other (19%)
Neonatal
infections* (9%)
(8%)

Diarrhoeal
diseases

(10%)

Diarrhoeal
diseases

(1%)

East
(054 million deaths; MR=873)

North
(013 million deaths; MR=656)
Pneumonia (15%)

Prematurity and
low birthweight (13%)

Birth asphyxia and


Neonatal deaths birth trauma (12%)
47%

Other infectious
diseases (7%)

Other (9%)

Other (7%)
(1%)

Prematurity and
low birthweight (12%)

Pneumonia (18%)

Birth asphyxia and


birth trauma (7%)
Neonatal deaths
39%

Other infectious
diseases (12%)

Neonatal
infections* (12%)

Neonatal
infections* (12%)

Other (19%)

Other (17%)

Other (9%)
(12%)

(1%)

Diarrhoeal
diseases

(14%)

Northeast
(009 million deaths; MR=916)
Pneumonia (16%)

Tetanus (1%)
Other (6%)

(1%)

Diarrhoeal
diseases

Central
(111 million deaths; MR=1079)
Prematurity and
low birthweight (10%)

Pneumonia (17%)

Prematurity and
low birthweight (12%)

Birth asphyxia and


birth trauma (8%)
Neonatal deaths
34%
Other infectious
diseases (19%)

Neonatal
infections* (9%)

Birth asphyxia and


birth trauma (8%)
Neonatal deaths
43%

Other infectious
diseases (12%)

Neonatal
infections* (12%)

Other (6%)

Tetanus (2%)
Other (15%)

Other (7%)

Other (16%)
(15%)

Diarrhoeal
diseases

(1%)

(13%)

Diarrhoeal
diseases

(2%)

Figure 3: Causes of death in children aged 04 years in India, by region, 200103


MR=mortality rate in children younger than 5 years. *Includes neonatal pneumonia, sepsis, and CNS infections.

www.thelancet.com Vol 376 November 27, 2010

1857

Articles

Study deaths
200103

Mortality rate per 1000 livebirths (99% CI)

Total deaths
(thousands)
in 2005 (99% CI)

(7196)
(78108)
(96111)
(115136)
(118135)
(133158)

16
10
65
56
130
50

Prematurity and low birthweight


North
229
83
Northeast
212
92
East
865
104
South
525
126
Central
1191
126
West
609
145
Subtotal (A)

3631

120 (115124)

(1419)
(811)
(5969)
(5161)
(121139)
(4654)

327 (309345)

Neonatal infections*
South
160
West
226
North
212
Northeast
184
East
798
Central
1303
Subtotal (B)

38
51
76
79
100
145

2883

(3146)
(4361)
(6589)
(6694)
(93108)
(136154)

17
17
15
8
62
149

99 (94103)

(1421)
(1521)
(1317)
(79)
(5867)
(140158)

268 (253286)

Birth asphyxia and birth trauma


South
245
West
256
East
515
Northeast
158
North
208
Central
691

58
59
64
72
75
81

Subtotal (C)

2073

70 (6674)

190 (176206)

Total (A+B+C)

8587

289 (283292)

785 (773797)

(5068)
(5069)
(5871)
(5987)
(6488)
(7489)

26
20
40
7
14
83

10

(2230)
(1724)
(3644)
(69)
(1217)
(7691)

15

Figure 4: Mortality rates for the three leading causes of neonatal death in India, by region, 2005
*Includes neonatal pneumonia, sepsis, and CNS infections.

the higher-income states (gure 2). Prematurity and low


birthweight formed a greater proportion of all child deaths
in the west and south India than in other regions (gure 3).
Tetanus was a notable cause of death in central and east
India (gure 3).
We recorded substantial regional variation in mortality
rates for the three leading causes of death in neonates
(gure 4). The mortality rate for prematurity and low
birthweight was highest in the west and lowest in the
north (gure 4). The mortality rate for neonatal infections
in central India was nearly four-times higher than that in
south India, and mortality rates for birth asphyxia and
birth trauma were highest in central India and lowest in
the south (gure 4).
Pneumonia and diarrhoeal diseases accounted for 50%
(067 million of 134 million) of all deaths in children
aged 159 months (gure 5). All-cause mortality rate in
children aged 159 months was about 36% higher in
girls (567) than in boys (417) and most of the leading
causes of death were between 12% and 72% higher in
girls than in boys, with the exception of injuries and
meningitis/encephalitis (table). Pneumonia and
diarrhoeal diseases accounted for about two-thirds
(010 million) of the 015 million more deaths from all
causes in girls aged 159 months (table). Pneumonia and
diarrhoeal diseases accounted for a greater proportion of
1858

total child deaths in lower-income than in higher-income


states (gure 2), and their proportion of total child deaths
decreased as rates of mortality in children younger than
5 years fell (gure 3).
The mortality rate from pneumonia in central India
was four times that in the south, and the mortality rate
from diarrhoeal diseases in central India was three times
that in the west (gure 5). Dierences were even greater
when sex was taken into account. Girls in central India
had a ve-times higher mortality rate from pneumonia
than did boys in south India, and had a four-times higher
mortality rate from diarrhoeal disease than did boys in
the west (gure 5).

Discussion
More than three-fths of all 23 million child deaths in
India in 2005 were from ve causes: pneumonia,
prematurity and low birthweight, diarrhoeal diseases,
neonatal infections, and birth asphyxia and birth trauma.
Each of the major causes of neonatal deaths can be
prevented or treated with known, highly eective
and widely practicable interventions such as improvements
in prenatal care, intrapartum care (skilled attendance,
emergency obstetric care, and simple immediate
care for newborn babies), postnatal family-community
care (preventive postnatal care, oral antibiotics, and
management of pneumonia),24 and tetanus toxoid immunisation.25 Concern has been raised that neonatal death
rates in India are not falling fast enough.9 However, our
results suggest that almost half of Indias neonatal deaths
are caused by birth asphyxia and birth trauma, sepsis,
pneumonia, and tetanusmost of which can be avoided
by increases in delivery and postnatal care.26
The substantial regional dierences in cause-specic
mortality, even in girls (webappendix p 12), could indicate
the existence of some underlying social, behavioural, or
biological risk factors for child deaths.10 However, at ages
159 months, girls in every region die more commonly
than do boys, and inequities in access to care, rather than
biological or genetic factors, are a more plausible
explanation for these recorded dierences between
sexes.7,8 Household surveys4,27 show little dierence
between sexes in the rate of respiratory symptoms and
diarrhoeal disease, whereas our study and previous
analyses28 have shown substantial sex dierences in
mortality. Integrated management of child illnesses
increases care seeking for illnesses,29 and reduces child
deaths,30 but in India, boys use such programmes more
than girls.9 Fewer girls than boys are vaccinated in health
facilities.8 However, outreach programmes that visit
households immunise a greater proportion of girls than
do facility-based vaccination programmes.9 Addition of
vaccines against pneumonia (pneumococcal conjugate,
Haemophilus inuenzae type B) and diarrhoeal diseases
(rotavirus) to outreach home-based immunisation
programmes would reduce child deaths and narrow the
gap in child mortality in India between sexes.9,31
www.thelancet.com Vol 376 November 27, 2010

Articles

Our study shows that boy social preference probably


aects survival for girls. States with higher mortality
rates in girls than in boys aged 159 months were also
those with lower female-to-male sex ratio for second
births after a boy (a measure of selective abortion of
girls)7 (Pearsons correlation coecient 047, p=00004;
data not shown). This nding implies that less frequent
use of health services by girls than by boys occurs in the
same states in which selective abortion of female fetuses
is common. Moreover, a relative gap in mortality
between girls and boys at ages 159 months is recorded
in urban areas, in more educated groups, and in
states with lower mortality rates (webappendix p 12).4,32
However, the excess of 015 million deaths in girls aged
159 months is largely oset by the excess of 013 million
deaths in male neonates. Thus, the most plausible
explanation for the dierence of 6 million between boys
and girls in the 2001 census21 (which recorded 76 million
girls and 82 million boys aged 06 years) is probably
selective abortion of female fetuses,7 and less so the
greater mortality in girls.
The main uncertainty in our estimates arises because
verbal autopsy misclassies some causes of death,11,16,23 and
because our estimates relied mostly on family reports of
deaths occurring in rural homes. Previous studies
comparing hospital-based deaths with home-based verbal
autopsy (which formed the basis for the eld instrument
used in the Million Death Study) have reported reasonable
agreement for the symptoms used to assign the ve major
causes of death that we report in this study.16 Hospital
deaths should not be regarded as a gold standard because
there are probably important dierences in the distribution
of causes of child deaths, treatment patterns, and, for
infectious causes, in their underlying pathogens,33 between
hospital deaths (mostly in urban areas) and rural,
unattended deaths in the home.13 Misclassication of
causes can aect our estimates of the total number of
deaths from each cause,34 but misclassication is unlikely
to be greatly dierent across sex, areas, and regions, and is
unlikely to substantially aect our estimates of dierences
between sexes. The missing deaths or deaths that
physicians were unable to code, although sizeable, are
mostly random and unlikely to have aected the overall
substantial variation by sex and region that we recorded.
Similarly, there is also uncertainty in the UN total estimates
of yearly child deaths (235 million deaths in 2005, ranging
between 226 million and 246 million);35 however, such
uncertainty would probably raise or lower the overall
mortality rates, but would not aect the recorded sex or
regional variation in these mortality rates.
Our results correspond to deaths before the widescale introduction of Indias National Rural Health
Mission in 2006. That programme reports increases in
institutional deliveries9 and in coverage of existing
vaccines, and therefore might have reduced child
mortality in India. Our study also suggests that specic
interventions might be priorities for dierent regions
www.thelancet.com Vol 376 November 27, 2010

Study deaths
200103

Mortality rate per 1000 livebirths (99% CI)

Total deaths
(thousands)
in 2005 (99% CI)

Pneumonia
South
West
North
Northeast
East
Central

129
262
304
281
860
1596

47
76
96
142
153
180

Subtotal (A)

3432

135 (130141)

369 (348390)

Boys/south*
Girls/central*

66
896

41 (3056)
209 (194226)

9 (713)
103 (95111)

49
59
81
126
133
145

18
26
16
79
13
150

22
27
20
15
97
188

(3859)
(6488)
(84109)
(120167)
(141164)
(170191)

(1726)
(2230)
(1621)
(1217)
(88102)
(175196)

Diarrhoeal diseases
176
West
153
South
250
North
720
East
227
Northeast
1190
Central

(3959)
(4972)
(6993)
(116137)
(111158)
(135155)

(1420)
(2232)
(1418)
(7285)
(1116)
(139159)

Subtotal (B)

2716

111 (105116)

302 (283323)

Boys/west
Girls/central

84
697

41 (3055)
177 (162193)

7 (510)
87 (8095)

6148

246 (239252)

Total (A+B)

671 (653688)
5

10

15

20

25

Figure 5: Mortality rates for the two leading causes of death in children aged 159 months in India, by
region, 2005
*Boys from the south region have the lowest mortality rate for pneumonia by sex and region (at age
159 months), and girls from the central region have the highest. Boys from the west region have the lowest
mortality rate for diarrhoeal diseases by sex and region (at age 159 months), and girls from the central region
have the highest.

(eg, expanded case management and introduction of


newer vaccines into immunisation programmes would
be particularly needed in central India, especially for
girls). The changes in the sex-specic and regionspecic rates and causes of neonatal mortality and
mortality at ages 159 months will continue to be
monitored and reported by the RGI,12 and should thus
help to assess the eectiveness of the National Rural
Health Mission and other eorts to reduce child
mortality in India.
Contributors
The academic partners in India (Million Death Study collaborators;
webappendix p 7) planned the Million Death Study in close collaboration
with the Oce of the RGI. RK, SA, DGB, and PJ planned the child
mortality study. DGB and PJ did the analyses. All authors were involved
with data interpretation, critical revisions of the paper, and approved the
nal version. PJ is the guarantor of this report.
Million Death Study collaborators
Writing committee Diego G Bassani, Rajesh Kumar, Shally Awasthi,
Shaun K Morris, Vinod K Paul, Anita Shet, Usha Ram,
Michelle F Gaey, Robert E Black (chair of the Child Health
Epidemiology Reference Group), and Prabhat Jha (Principal Investigator
for the Million Death Study). India CHERG group Rajesh Kumar (chair),
Shally Awasthi, Diego G Bassani (facilitator), Robert E Black,
Prabhat Jha, Bhaskar Mishra, Vinod K Paul, Usha Ram, Siddarth Ramji,
Anita Shet, and Mani Subramaniyam.
Conicts of interest
We declare that we have no conicts of interest.

1859

Articles

Acknowledgments
The Registrar General of India established the SRS in 1971, has
continued it ever since, and is collaborating with several of the authors
on the ongoing Million Death Study. External funding is from the
Fogarty International Centre of the US National Institutes of Health
(grant R01 TW0599101), Canadian Institute of Health Research
(CIHR; IEG-53506), International Development Research Centre
(Grant 102172), Li Ka Shing Knowledge Institute and Keenan Research
Centre at St Michaels Hospital, University of Toronto, and the US Fund
for UNICEF (via a grant from the Bill & Melinda Gates Foundation for
CHERG; subgrant 50140). PJ is supported by the Canada Research Chair
programme. SKM is a Fellow of the Pediatric Scientist Development
Program. The opinions expressed in this paper are those of the authors
and do not necessarily represent those of the Government of India. We
thank Joy Lawn, Colin Mathers, Mikkel Oestergaard, Prem Mony, and
Alvin Zipursky for comments; and Maya Kesler, Brendon Pezzack,
Chinthanie Ramasundarahettige, Peter Rodriguez, and
Wilson Suraweera for data support.
References
1
UN Population Division. World population prospects (2008
revision). April, 2009. http://esa.un.org/peps/peps_interpolateddata.htm (accessed June 14, 2010).
2
Rajaratnam JK, Marcus JR, Flaxman AD, et al. Neonatal,
postneonatal, childhood, and under-5 mortality for 187 countries,
19702010: a systematic analysis of progress towards the
Millennium Development Goal 4. Lancet 2010; 357: 19882008.
3
Black RE, Cousens S, Johnson HL, et al. Global, regional, and
national causes of child mortality in 2008: a systematic analysis.
Lancet 2010; 375: 196987.
4
International Institute for Population Sciences (IIPS) and Macro
International. National Family Health Survey (NFHS-3), 200506:
India. Mumbai: IIPS, 2008.
5
Registrar General of India. Sample Registration System. New Delhi:
Oce of the Registrar General of India, 2004.
6
Baqui AH, Darmstadt GL, Williams EK, et al. Rates, timing and
causes of neonatal deaths in rural India: implications for
neonatal health programmes. Bull World Health Organ 2006;
84: 70613.
7
Jha P, Kumar R, Vasa P, Dhingra N, Thiruchelvam D,
Moineddin R. Low male-to-female sex ratio of children born in
India: national survey of 11 million households. Lancet 2006;
367: 21118.
8
Corsi DJ, Bassani DG, Kumar R, et al. Gender inequity and
age-appropriate immunization coverage in India from 1992 to 2006.
BMC Int Health Hum Rights 2009; 9 (suppl 1): S3.
9
Jha P, Laxminarayan R. Choosing health: an entitlement for all
Indians. May, 2009. http://cghrindia.org/images/choosing-health.
pdf (accessed Sept 10, 2010).
10 Jha P. Avoidable mortality in India: past progress and future
prospects. Natl Med J India 2002; 15 (suppl 1): 3236.
11 Jha P, Gajalakshmi V, Gupta PC, et al. Prospective study of one
million deaths in India: rationale, design, and validation results.
PLoS Med 2006; 3: e18.
12 Registrar General of India, Centre for Global Health Research.
Causes of death in India, 20012003: Sample Registration System.
New Delhi: Government of India, 2009.
13 Dhingra N, Jha P, Sharma VP, et al, for the Million Death Study
Collaborators. Adult and child malaria mortality in India: a
nationally representative mortality survey. Lancet 2010; published
online Oct 21. DOI:10.1016/S0140-6736(10)60831-8.
14 Jha P, Jacob B, Gajalakshmi V, et al. A nationally representative
case-control study of smoking and death in India.
N Engl J Med 2008; 358: 113747.
15 Jha P, Kumar R, Khera A, et al. HIV mortality and infection in
India: estimates from nationally representative mortality survey of
11 million homes. BMJ 2010; 340: c621.

1860

16

17

18

19

20
21
22
23

24

25

26

27

28

29

30

31

32

33
34

35

Anker M, Black RE, Coldham C, et al. A standard verbal autopsy


method for investigating causes of death in infants and children.
Geneva: World Health Organization, 1999.
WHO. International statistical classication of diseases and related
health problems, tenth revision (ICD10). Geneva: World Health
Organization, 1994.
Sinha D, Dikshit R, Kumar V, Gajalakshmi V, Dhingra N, Seth J.
Technical document VII: health care professionals manual for
assigning causes of death based on RHIME household reports.
Toronto: Centre for Global He lth Research; University of Toronto,
2006.
Bhat MPN. Completeness of Indias Sample Registration System:
an assessment using the general growth balance method.
Popul Stud 2002; 56: 11934.
Registrar General of India. Medically-certied causes of death,
statistical report: 2000. New Delhi: Government of India, 2005.
Registrar General of India. Census of India 2001. New Delhi: Oce
of the Registrar General & Census Commissioner, 2001.
Wolter KM. Introduction to variance estimation, 2nd edn.
New York: Springer, 2007.
Morris SK, Bassani DG, Kumar R, Awasthi S, Paul VK, Jha P.
Factors associated with physician agreement on verbal autopsy of
over 27000 childhood deaths in India. PLoS One 2010; 5: 18.
Darmstadt GL, Bhutta ZA, Cousens S, Adam T, Walker N,
de Bernis L, for the Lancet Neonatal Survival Steering Team.
Evidence-based, cost-eective interventions: how many newborn
babies can we save? Lancet 2005; 365: 97788.
Ronsmans C, Chowdhury ME, Alam N, Koblinsky M, El Arifeen S.
Trends in stillbirths, early and late neonatal mortality in rural
Bangladesh: the role of public health interventions.
Paediatr Perinat Epidemiol 2008; 22: 26979.
Kumar V, Mohanty S, Kumar A, et al. Eect of community-based
behaviour change management on neonatal mortality in Shivgarh,
Uttar Pradesh, India: a cluster-randomised controlled trial. Lancet
2008; 372: 115162.
International Institute for Population Sciences (IIPS). District
Level Household and Facility Survey (DLHS-3), 2007-08. Mumbai:
IIPS, 2010.
Bassani DG, Jha P, Dhingra N, Kumar R. Child mortality from
solid-fuel use in India: a nationally-representative case-control
study. BMC Public Health 2010; 10: 49199.
Arifeen SE, Hoque DME, Akter T, et al. Eect of the Integrated
Management of Childhood Illness strategy on childhood mortality
and nutrition in a rural area in Bangladesh: a cluster randomised
trial. Lancet 2009; 374: 393403.
Ali M, Asefaw T, Byass P, Beyene H, Pedersen FK. Helping
northern Ethiopian communities reduce childhood mortality:
population-based intervention trial. Bull World Health Organ 2005;
83: 2733.
Zaman K, Anh DD, Victor JC, et al. Ecacy of pentavalent rotavirus
vaccine against severe rotavirus gastroenteritis in infants in
developing countries in Asia: a randomised, double-blind,
placebo-controlled trial. Lancet 2010; 376: 61523.
Registrar General of India. Special Fertility & Mortality Survey,
1988. A report of 1.1 million Indian households. New Delhi:
Registrar General, 2005.
Berkley JA, Lowe BS, Mwangi I, et al. Bacteremia among children
admitted to a rural hospital in Kenya. N Engl J Med 2005; 1: 3947.
Maude GH, Ross DA. The eect of dierent sensitivity, specicity
and cause-specic mortality fractions on the estimation of
dierences in cause-specic mortality rates in children from studies
using verbal autopsies. Int J Epidemiol 1997; 26: 1097106.
Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL, eds.
Global burden of disease and risk factors. New York: Oxford
University Press, 2006.

www.thelancet.com Vol 376 November 27, 2010

Series

Chronic Diseases: Chronic Diseases and Development 5


Monitoring and surveillance of chronic non-communicable
diseases: progress and capacity in high-burden countries
Ala Alwan, David R MacLean, Leanne M Riley, Edouard Tursan dEspaignet, Colin Douglas Mathers, Gretchen Anna Stevens, Douglas Bettcher

The burden of chronic, non-communicable diseases in low-income and middle-income countries is increasing. We
outline a framework for monitoring of such diseases and review the mortality burden and the capacity of countries
to respond to them. We show data from WHO data sources and published work for prevalence of tobacco use,
overweight, and cause-specic mortality in 23 low-income and middle-income countries with a high burden of noncommunicable disease. Data for national capacity for chronic disease prevention and control were generated from a
global assessment that was done in WHO member states in 200910. Although reliable data for cause-specic
mortality are scarce, non-communicable diseases were estimated to be responsible for 234 million (or 64% of the
total) deaths in the 23 countries that we analysed, with 47% occurring in people who were younger than 70 years.
Tobacco use and overweight are common in most of the countries and populations we examined, but coverage of
cost-eective interventions to reduce these risk factors is low. Capacity for prevention and control of noncommunicable diseases, including monitoring and surveillance operations nationally, is inadequate. A surveillance
framework, including a minimum set of indicators covering exposures and outcomes, is essential for policy
development and assessment and for monitoring of trends in disease. Technical, human, and scal resource
constraints are major impediments to the establishment of eective prevention and control programmes. Despite
increasing awareness and commitment to address chronic disease, concrete actions by global partners to plan and
implement cost-eective interventions are inadequate.

Lancet 2010; 376: 186168

Introduction

Non-communicable Diseases
and Mental Health
(A Alwan MD,
Prof D R MacLean MD);
Department of Chronic
Diseases and Health
Promotion (L M Riley MSc);
Tobacco Free Initiative
(E Tursan dEspaignet PhD,
D Bettcher PhD); and Mortality
and Burden of Disease,
Department of Health
Statistics and Informatics
(C D Mathers PhD,
G A Stevens PhD), World Health
Organization, Geneva,
Switzerland

Globally, around 57 million people died in 2008, and


33 million (58%) of these deaths were due to chronic
(non-communicable) diseases (mainly cardiovascular
disease, diabetes, cancer, and chronic respiratory
diseases).1 The burden of such diseases in low-income
and middle-income countries is rapidly increasing and
already has major adverse social, economic, and health
eects.2,3 Progress towards prevention has not kept pace
with the rising burden of chronic diseasesa failing that
is chiey one of the politics of priority setting and not of
the absence of eective interventions.4 A previous Lancet
Series5 estimated that potentially 32 million deaths from
chronic diseases could be averted in 10 years in 23 lowincome and middle-income countries that have a high
burden of such diseases if global partners were to support
widespread adoption of three cost-eective interventions
(on tobacco and salt intake and hypertension). Mortality
in these 23 countries (which are listed in gure 1)
accounted for around 234 million (80%) of deaths from
non-communicable diseases in all low-income and
middle-income countries in 2005, and 50% of the total
burden of disease.1 The previous Series5 called for support
for prevention of chronic diseases from global partners
such as WHO, and actions were outlined for rapid
implementation of cost-eective interventions.
In 2000, the World Health Assembly endorsed a global
strategy7 for the prevention and control of noncommunicable diseases. The plan had three key
components: surveillance to track and monitor the major
risk factors; promotion of health to reduce these risk
www.thelancet.com Vol 376 November 27, 2010

factors; and improved management to promote access to


health care. The World Health Assembly subsequently
endorsed the WHO framework convention on tobacco
control (FCTC),8 and the global strategy on diet and
physical activity and health,9 along with proposed scalingup of technical support to countries taking part in

Key messages
Low-income and middle-income countries are
undergoing a rapid rise in the burden of
non-communicable diseases with major adverse
social, economic, and health outcomes
We provide a surveillance framework to quantify and
monitor non-communicable diseases and their
determinants
Prevalence of tobacco use in men and overweight in men
and women are high in many of the 23 low-income and
middle-income countries with high rates of
non-communicable disease
Age-specic death rates for non-communicable diseases
are higher in the 23 low-income and middle-income
countries than it is in high-income countries
Health-system capacity in the 23 countries with a high
burden of non-communicable diseases, including
surveillance, is inadequate to reduce the disease burden
and urgently requires strengthening
Disease surveillance should be integrated into national
health information systems and regular monitoring of
actions to prevent non-communicable diseases is needed

This online publication


has been corrected. The
corrected version rst
appeared at thelancet.com
on November 15, 2010
Published Online
November 11, 2010
DOI:10.1016/S01406736(10)61853-3
See Comment Lancet 2010;
376: 161921
See Online/Comment
DOI:10.1016/S01406736(10)61856-9, and
DOI:10.1016/S01406736(10)61891-0
This is the fth in a Series of
ve papers about chronic diseases

Correspondence to:
Dr Ala Alwan,
Non-communicable Diseases
and Mental Health, World Health
Organization, 20 Avenue Appia,
1211 Geneva, Switzerland
alwana@who.int

For the 2007 Lancet Chronic


Diseases Series see http://www.
thelancet.com/series/
chronic-diseases

1861

Series

100

Men
Women

Prevalence (%)

80
60
40
20

Ar
ge
Ba nti
ng na
lad
es
h
Br
az
i
Bu l
rm
a
C
DR hina
Co
ng
o
Eg
y
Et pt
hi
op
ia
In
In dia
do
ne
sia
Ira
M n
ex
ic
Ni o
ge
Pa ria
k
Ph ista
ilip n
pi
ne
Po s
lan
d
So Rus
ut sia
hA
fr
T h ic a
ail
an
Tu d
rk
Uk ey
ra
Vi ine
et
na
m

Figure 1: Estimated prevalence of daily tobacco use in adults aged 15 years or older, 2005
Data from WHOs report on the global tobacco epidemic.6 No data for tobacco use were available for Colombia.

surveillance of risk factors. In 2008, WHO developed an


action plan10 in collaboration with its member states to
convert the strategies into actions.
Surveillance and monitoring are crucial to provide
countries with the information needed about development
of policies and programmes for non-communicable
diseases, and to support the assessment of these
initiatives and monitor progress. The capacity to
undertake surveillance for these purposes varies greatly
around the world and was judged inadequate in many
low-income and middle-income countries.11 We present a
framework for surveillance and monitoring of noncommunicable disease in 23 high-burden countries,
covering major risk factors, cause-specic mortality, and
the capacity of countries for prevention.

Methods
As part of the implementation of its non-communicable
diseases action plan,10 WHO established an expert
epidemiology reference group in 2009 to help to develop a
national disease-surveillance framework. Components and
the core indicators were adopted on the basis of existing
WHO techniques and data sources that are used for
periodic reporting (every few years) of health issues, such
as WHO updated mortality estimates, global tobacco
surveillance system,12 the STEPwise approach to chronic
disease risk factor surveillance,13 Global NCD InfoBase,14
and other relevant databases.
WHO periodically estimates deaths globally, regionally,
and nationally by cause.15 We draw on provisional
estimates for the year 2008, which are presently under
country consultation. These estimates use standard
methods16 that draw on updated estimates of all-cause
mortality,17 the latest available death-registration data,18
WHO and UNAIDS programme estimates for some
specic causes (tuberculosis, HIV-1/AIDS, and malaria),
updated estimates of cause-specic deaths in children
younger than 5 years,19 and estimates of cancer deaths by
site for 2008 from the IARC Globocan database.20
For 12 of the 23 countries we assessed, we calculated
mortality rates from recent death-registration data, which
were projected to the year 2008 if 2008 data were not
1862

available. Additional country-specic data sources or


studies were used to update estimates for India,21,22
China,23 Thailand,24 and Vietnam.25 For the seven other
countries (four in Asia and three in Africa), deaths by
age, sex, and cause were estimated for 2008 with causeof-death models as described elsewhere,15 together with
cause-specic estimates as noted previously.
WHO published mortality projections from 2004 to
2030 that were calculated with equivalent methods to
those applied in the original global burden of disease
study.26 A set of simple models were used to project
future health trends, chiey on the basis of projections
of economic and social development, and with the
historically observed relations of these projections with
cause-specic mortality rates. The projections we
suggest here for the 23 countries have been revised to
take into account the updated base estimates for 2008,
updated projections of real change in income per head
(which take into account the eects of the global
nancial crisis),2729 and updated projections for tobaccorelated deaths (which take into account present trends
in tobacco consumption).
We used a common core of indicators (ie, tobacco,
body-mass index, and mortality) for comparability of
information across countries and regions. Lack of
comparability of data is one of the major diculties faced
by global surveillance and monitoring. It is also crucial
that the core indicators can be technically measured in all
resources settings.
Comparable estimates for tobacco use and overweight
(body-mass index 25) for 2005 as key risk factors for
non-communicable diseases are shown. The data used
to calculate these estimates were aggregated data
provided to WHO by countries and obtained through a
review of published and unpublished work. These data
were captured in the WHO Global InfoBase,14 which is
a continuously updated repository of all countryreported data for eight key risk factors for chronic
diseases and is maintained by WHO.14 The inclusion
criteria for our estimation analysis included data that
had come from a random sample of the general
population, with clearly described survey methods
(including sample sizes) and risk factor denitions. We
made adjustments to enhance comparability between
the standard denitions of risk factors, standard set of
age groups for reporting, standard reporting year, and
representativeness of population. We used regression
modelling to produce crude and age-specic rates of a
hypothetical standard indicator for all 23 countries. Full
descriptions of the methods used to produce these
estimates have been published elsewhere.30,31
To review progress in tobacco control, we did an
analysis of the eectiveness of the implementation of
selected demand-reduction measures of WHO FCTC.
Summarised by WHO in the acronym MPOWER, these
measures refer to monitoring of tobacco use and
prevention policies; protection of people from tobacco
www.thelancet.com Vol 376 November 27, 2010

Series

www.thelancet.com Vol 376 November 27, 2010

100

Men
Women

Prevalence (%)

80
60
40
20
0

Ar
g
Ba enti
ng na
lad
es
h
Br
az
i
Bu l
rm
a
C
Co hina
lo
m
DR bi
Co a
ng
o
Eg
Et ypt
hi
op
ia
I
In ndia
do
ne
sia
Ira
M n
ex
i
Ni co
ge
Pa ria
Ph kista
ilip n
pi
ne
Po s
lan
d
So Ru
ut ssia
hA
f
Th rica
ail
an
Tu d
rk
Uk ey
ra
Vi ine
et
na
m

smoke; oers of help to quit; warnings about the dangers


of tobacco; enforcement of bans on advertising,
promotion, and sponsorship; and raising of taxes on
tobacco products. A detailed description of this
assessment has been published elsewhere.6
Data for health-system capacity were drawn from a
global questionnaire assessment of country capacity for
prevention and control of non-communicable diseases,
which was done between November, 2009, and April, 2010
(hereafter referred to as the 2010 country capacity
assessment). Our assessment was designed to measure
individual country capacity to respond to prevention and
control of chronic diseases. Specic areas of assessment
were the public health infrastructure for noncommunicable diseases; the status of relevant policies,
strategies, action plans, and programmes; healthinformation systems, surveillance, and surveys; healthsystem capacity for early detection, treatment, and care of
non-communicable diseases; and health promotion,
partnerships, and collaboration. Country questionnaire
surveys asked participants about the existence of a unit,
branch, or department in their Ministry of Health with
responsibility for non-communicable diseases, and
whether the policy was integrated and operational, had
available funding for surveillance, monitoring, and
assessment, and whether guidelines, protocols, or
standards existed (see webappendix pp 840). 22 of 23 highburden countries responded to the questionnaire, with
only Colombia not responding in 2010.
Our assessment included a set of detailed instructions
to complete the questionnaire and a glossary of terms we
used (webappendix pp 3840). The questionnaire was
translated into Spanish, French, and Russian to encourage
completion by all countries. The nal set of questions
and instructions were provided through use of an
electronic Microsoft Excel questionnaire method, which
was completed by professionals nationally and led by the
focal point for non-communicable diseases within the
Ministry of Health of every country and approved by
Ministry of Health senior ocials before submission to
our group at WHO.
The major components of our proposed national
surveillance framework for non-communicable diseases
encompassed key risk factors (consisting of behavioural,
dietary, physiological, and metabolic factors), outcomes
(mortality and morbidity), health-system response
interventions, and health-system capacity. For every
component of the proposed surveillance framework, a
core and expanded set of indicators was proposed for
global and national monitoring (see webappendix pp 17).
Core indicators had to have an established evidence-base
and policy relevance, and be modiable through costeective interventions, feasible and aordable to obtain,
achievable within a countrys technical capacity, and
measurable in a consistent manner with valid denitions
and techniques. We show several core indicators for the
23 high-burden countries.

Figure 2: Estimated prevalence of overweight (body-mass index of 25 or more) in adults aged 15 years or
older, 2005

Ukraine
Russia
South Africa
Egypt
Nigeria
Poland
Thailand
Burma
Indonesia
DR Congo
India
Bangladesh
Argentina
Brazil
Pakistan
Philippines
China
Ethiopia
Turkey
Mexico
Vietnam
Cardiovascular diseases and diabetes
Cancers
Chronic repiratory diseases
Other chronic diseases

Iran
Colombia
All 23 countries
0

100

200

300

400

500

600

700

800

Death rate per 100 000 adults aged 1569 years

Figure 3: Death rates from non-communicable diseases per 100 000 adults aged 1569 years in
23 high-burden countries

Risk factors

See Online for webappendix

We estimated that the prevalence of tobacco use in 2005


was high in men in most of the 23 countries that we
analysed compared with prevalence for low-burden
countries, and that the highest rates were reported in
eastern Europe and Asia (gure 1); prevalence was highest
in China (571%), Indonesia (584%), and Russia (649%).
Prevalence of tobacco use in men was lower, at around
10% or less, in some African countries (Ethiopia, Nigeria,
and Democratic Republic of the Congo) compared with
other high-burden countries, although South Africa was
1863

Series

an exception, with a prevalence of 213%. In general,


fewer women use tobacco than do men; for example, in
China the prevalence for women was 37%.
Figure 2 shows the 2005 estimates for adult prevalence
of overweight in the 23 countries with a high burden of
non-communicable diseases. Prevalence of overweight
in men ranged from 4% in Vietnam to 73% in Argentina,
and in women from 3% in Ethiopia to 74% in Egypt. In
15 of 23 countries, prevalence of overweight in women
was higher than it was in men, and in three countries it
was more than twice as frequent (DR Congo, Indonesia,
and Vietnam). In ten of 23 countries, more than 50% of
the population was overweight in 2005.

Mortality
Deaths from chronic disease in the 23 high-burden
countries accounted for 234 million (41%) of global
deaths from all causes and 80% of deaths from noncommunicable diseases in all low-income and middleincome countries. Deaths from non-communicable
diseases in people aged younger than 70 years in the
23 countries accounted for 47% of all non-communicable
disease deaths in these countries and for 71% of such
deaths in people younger than 70 years globally.
Integrated NCD policy

Policy addressing specic risk factors

Developed

Operational

Alcohol

Unhealthy diet Physical inactivity Tobacco

Yes

No

No

Yes

Yes

Yes

Bangladesh Yes

Yes

Yes

Yes

Yes

Yes

Brazil

Yes

Yes

Yes

Yes

Yes

Yes

Burma

Yes

No

Yes

Yes

Yes

Yes

China

No

NA

No

No

No

No

DR Congo

Yes

No

Yes

Yes

Yes

Yes

Egypt

Yes

Yes

NR

Yes*

NR

Yes*

Ethiopia

Yes

No

Yes

Yes

Yes

Yes

India

Yes

Yes

Yes

Yes

Yes

Yes

Indonesia

Yes

Yes

Yes

Yes

Yes

Yes

Iran

No

NA

No*

Yes*

Yes*

Yes*

Mexico

Yes

Yes

Yes

Yes

Yes

Yes

Nigeria

Yes

Yes

Yes

Yes

Yes

Yes

Pakistan

No

NA

No*

No*

No*

No*

Philippines

Yes

Yes

Yes

Yes

Yes

Yes

Poland

Yes

Yes

Yes

Yes

Yes

Yes

Russia

Yes

Yes

Yes

Yes

Yes

Yes

South Africa Yes

Yes

Yes

Yes

Yes

Yes

Thailand

Yes

Yes

Yes

Yes

Yes

Yes

Turkey

No

NA

Yes*

Yes*

Yes*

Yes*

Ukraine

No

NA

No*

No*

No*

Yes*

Vietnam

Yes

Yes

No

No

No

No

Overall

17/22 (77%) 13/22 (59%)

17/22 (77%)

19/22 (86%)

Argentina

15/22 (68%) 18/22 (82%)

Data were reported to WHO as part of the NCD country capacity assessment in 2010. Colombia did not respond to the
capacity assessment, so no data were available. NCD=non-communicable disease. NA=not applicable. NR=not
reported. *Country did not report risk factor as part of their integrated NCD policy, but did report a standalone policy.

Table 1: Presence of an integrated non-communicable disease policy, operational status, and inclusion of
specic risk factors for 22 countries with high burdens of non-communicable diseases

1864

Age-specic death rates were higher in many lowincome and middle-income countries than they were in
high-income countries. For the 14 countries with death
registration data for cause of death, the overall agestandardised death rates from non-communicable
diseases were 711 per 100 000 for men (58% higher than
that for men in high-income countries in 2008) and
508 per 100 000 for women (69% higher).
Of the 23 countries shown in gure 3, death rates from
chronic diseases in people aged 1569 years were highest
in eastern European countries such as Ukraine and
Russia, and in some countries of the African continent
such as South Africa, Egypt, and Nigeria. Although agespecic death rates for most non-communicable diseases
are projected to decline with increasing rates of economic
development, the ageing of the populations of these
countries will lead to a substantially increased overall
number of deaths. Overall mortality from noncommunicable diseases for people younger than 70 years
is projected to rise from 108 million in 2010 to
154 million in 2050 for the 23 countries we examined.
Whereas deaths from infectious disease in people
younger than 70 years are projected to decline by around
2% per year during the next 40 years, the number of
deaths from cardiovascular disease are projected to
increase by 07% per year, and from cancer by 11% per
year in high-burden countries.

Country capacity
Of the 23 high-burden countries we selected to investigate,
only one country (Colombia) did not participate in the
2010 assessment. In the 2010 survey, all countries, apart
from Poland, reported the existence of a unit, branch, or
department in their Ministry of Health with responsibility
for non-communicable disease. However, of these
countries, only 17 reported having an integrated policy,
strategy, or action plan in place, which were reported as
being operational in only 13 countries (table 1). In this
context, operational means that the country believes that
the policy is being actively implemented. With respect to
policies, strategies, or plans of action to address individual
diseases or risk factors, 18 countries reported plans to
address cancer (country data not shown), 19 countries
reported having a tobacco control policy, and only
18 countries reported policies or programmes for diet or
physical activity (table 1).
For national surveillance systems for non-communicable
diseases, most countries had funding for surveillance,
monitoring, and assessment (table 2). 20 countries reported
that mortality data were contained in their national health
reporting system, but only ten of these countries stated
that their mortality data were population-based (table 2)
and only seven reported regular provision of reliable data
for cause-specic mortality to WHO. With respect to risk
factors for non-communicable diseases, 17 countries
reported that their national health-reporting system
detailed risk factors, and 11 of these systems included
www.thelancet.com Vol 376 November 27, 2010

Series

Mortality

Cancer registry

Risk factors

Yes*

Yes

Yes*

Bangladesh

No

Yes

Yes

Brazil

Yes*

Yes*

Yes*

Argentina

Burma

Yes

Yes

Yes

China

Yes*

Yes

Yes*

DR Congo

Yes

No

No

Egypt

Yes*

Yes

Yes*

Ethiopia

Yes

No

No

India

Yes

Yes

Yes

Indonesia

Yes

Yes

Yes

Iran

Yes*

Yes*

Yes*

Mexico

Yes*

Yes

Yes*

Nigeria

Yes

Yes

Yes*

Pakistan

No

No

No

Philippines

Yes

Yes*

Yes*

Poland

Yes*

Yes*

Yes*

Russia

Yes*

Yes

No

South Africa

Yes*

Yes

Yes*

Thailand

Yes

Yes

Yes

Turkey

Yes

Yes*

Yes*

Ukraine

Yes*

Yes

Yes

Vietnam

Yes

Yes

No

Overall

20/22 (91%)

19/22 (86%)

17/22 (77%)

Data were reported to WHO as part of the NCD country capacity assessment in 2010.
Colombia did not respond to the capacity assessment, so no data were available.
NCD=non-communicable disease. *Population-based data.

Table 2: Presence of surveillance data in national health-reporting systems


in 22 countries with high burdens of non-communicable diseases

population-based data; for disease registers, only ve


countries reported having a population-based cancer
registry (table 2).
Respondents were asked whether their countries had
guidelines, protocols, or standards available for the
management of risk factors and conditions leading to
non-communicable disease, particularly for primary
care. We contend that guidelines are generally available,
but implementation is low. Much the same was noted
about availability of essential medicines for treatment
and management of chronic diseases, countries
reported high availability but low aordability (country
data not shown).

2000 versus 2010 country capacity assessments


The 2010 country capacity survey was done on the basis of
a similar survey by WHO in 2000.11 Of 22 high-burden
countries that participated in the 2010 assessment, 20 also
contributed to the 2000 assessment (with the exceptions
of Ukraine and South Africa), allowing comparison of
responses to common questions between the two
assessments. A question about implementation was added
to the 2010 assessment because the rst survey was limited
by an absence of dierentiation between the presence of a
policy or programme and its implementation.
www.thelancet.com Vol 376 November 27, 2010

Dedicated NCD
oce in Ministry
of Health

Cardiovascular disease policy

Cancer policy

2000

2010

2000

2010

Operational
in 2010

2000

2010

Operational
in 2010

Argentina

Yes

Yes

Yes

Yes

No

Yes

Yes

Yes

Bangladesh

No

Yes

No

Yes

Yes

Yes

Yes

Yes

Brazil

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Burma

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

China

Yes

Yes

Yes

No

No

Yes

Yes

Yes

DR Congo

Yes

Yes

No

No

NA

No

No

NA

Egypt

No

Yes

No

NR

NA

No

Yes

Yes

Ethiopia

Yes

Yes

No

Yes*

No

No

Yes*

No

India

Yes

Yes

NR

Yes*

Yes

Yes

Yes*

Yes

Indonesia

No

Yes

No

Yes

Yes

No

Yes

Yes

Iran

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Mexico

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Nigeria

Yes

Yes

No

Yes*

Yes

No

Yes

No

Pakistan

No

Yes

No

No

NA

Yes

No

NA

Philippines

Yes

Yes

Yes

Yes*

Yes*

Yes

No

NA

Poland

Yes

No

Yes

Yes

Yes

Yes

Yes

Yes

Russia

Yes

Yes

Yes

Yes*

Yes

No

Yes*

Yes

Thailand

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Turkey

Yes

Yes

No

Yes

No

No

Yes

No

Vietnam

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Overall

14/20
(70%)

19/20
(95%)

11/20
(55%)

16/20
(80%)

13/20
(65%)

13/20
(65%)

17/20
(85%)

14/20
(70%)

Data were reported to WHO as part of NCD country capacity assessments in 2000 and 2010. Colombia, South Africa, and
Ukraine did not respond to the capacity assessment in either 2000 or 2010, so no data were available. NA=not applicable.
NCD=non-communicable disease. NR=Country did not respond to this question. *Integrated NCD policy (an integrated
policy addresses more than one risk factor or more than one NCD, in this case cardiovascular disease or cancer).

Table 3: Countries reporting key elements of non-communicable disease prevention capacity, 2000
and 2010

The existence of a unit, branch, or department with


responsibility for non-communicable diseases within a
countrys Ministry of Health had increased from 70% in
2000 to 95% in 2010 (table 3). Similarly, more countries
report the existence of a policy, strategy, or action plan for
a range of chronic disease problemsie, for
cardiovascular disease the proportion increased from
55% to 80% and for cancer from 65% to 80%. Respondents
suggested that there was an increase in the availability of
guidelines for management of diabetes and hypertension
during the decade, with an increase from 35% to 80% for
diabetes and 40% to 80% for hypertension.
However, despite the presence of a unit for noncommunicable diseases within a Ministry of Health, only
a few countries have implemented their guidelines for
management of such diseases. Equally, although a high
rate of surveillance for risk factors was reported in 2010,
our assessment does not provide accurate information
about the quality and regularity of data collection, nor
does it clarify whether standardised methodsfor
example, the WHO STEPwise13 approach to surveillance
of risk factorswere used.
1865

Series

WHO framework convention on tobacco control

Number of
countries (%)

Article

Criteria

20

Recent (data for 2003 or later), representative (survey sample


representative of the national population), and periodic
(occurring at least every 5 years) data for adults (15 years or
older) and adolescents (1315 years)

4 (17%)

Protection from 8
tobacco smoke

All public places wholly smoke-free (or at least 90% of the


population covered by complete subnational smoke-free
legislation)

3 (13%)

Oers of help to 14
quit

National quit telephone line, with nicotine-replacement


therapy and some cessation services cost-covered; smoking
cessation support available in health clinics or other
primary-care facilities, hospitals, oces of health
professionals, or the community

2 (9%)

Warnings about 11, 12


dangers

4 (17%)
50% (average of the front and back of the cigarette pack),
including pictures or pictograms and appropriate characteristics
such as specic health warnings mandated, appearing on
individual packages and on any outside packaging and labelling
used in retail sale, describing specic harmful eects of tobacco
use on health; warnings are large, clear, visible, legible, rotate,
and are written in all principal language(s) of the country

Enforced bans

13

Ban on all forms of direct and indirect advertising

5 (22%)

Raised taxes

>75% of retail price is tax

1 (4%)

Monitoring of
tobacco use

Table 4: Number of countries with a high burden of chronic, non-communicable diseases that had
implemented demand-reduction measures in December, 2008

Country capacity for implementation of tobacco control


WHOs report of the global tobacco epidemic6 noted that
20 of the 23 countries with a high burden of noncommunicable diseases had joined the FCTC by 2009.
The large number of high-burden countries now in this
organisation is much the same as that reported globally,
in which 170 countriesor about 90% of the world
populationare parties to the convention.
Our analysis also assessed the implementation of various
measures of the MPOWER plan of action, which is
designed to reduce demand for tobacco (table 4). The most
widely implemented strategies were enforcement of bans
on all forms of direct and indirect advertising (Egypt, Iran,
Burma [Myanmar], South Africa, and Thailand),
monitoring of tobacco use (Argentina, India, Iran, and
Poland), and warnings about the dangers (Brazil, Egypt,
Iran, and Thailand). Only three countries have made public
places entirely smoke-free (Colombia, Iran, and Turkey).
Only two countries (Brazil and Iran) have implemented
adequate measures to help smokers to quit, and only one
country (Poland) has increased taxes on tobacco.

Discussion
Our analysis shows that, in high-burden countries, the
capacity to eectively deal with the existing and projected
burden of non-communicable diseases is inadequate.
Chronic diseases are a leading health and developmental
challenge.32 The situation is especially serious in lowincome and middle-income countries that are undergoing
a rapid rise in premature mortality and an increasing
burden on their health systems despite availability of
cost-eective interventions.33 Such diseases are
1866

compromising sustainable development eorts and


poverty reduction initiatives.32,34
A key component of any national strategy to address
this challenge is to quantify and monitor noncommunicable diseases and their determinants through
sustainable surveillance schemes that are integrated into
the national information system. Our analysis provides a
step towards the development and implementation of
national surveillance systems by oering a framework
for surveillance and monitoring of countries against
which results of key core indicators can be assessed in
the future.
We appreciate the limitations of the data presently
available to monitor non-communicable diseases.
Concerted eorts are in progress to improve the coverage
and quality of mortality data. Adjustments made to risk
factors by WHO yield point estimates and 95% CI,
although these results should not be overinterpreted.
Condence intervals provide the range of values
consistent with the true population means or proportions,
and, together with point estimates, are especially
meaningful when they are produced on the basis of
strong, nationally representative surveys. There are
substantial limitations with respect to the quality of data
presently available on a global scale to monitor noncommunicable diseases. Concerted eorts are in progress
to improve the coverage and quality of data for mortality
and risk factors of such diseases. Because of the large
adjustments needed to be made to these data by WHO,
results should be interpreted with caution. WHO needs
to do these complex statistical modelling exercises
because too many countries presently have inadequate or
incomplete data.
The ideal situation would be for countries to do regular
surveys at a national scale with standardised methods,
including indicators, questionnaires, and age groups,
which would obviate the need for the adjustments
undertaken by WHO. The same limitations apply to our
monitoring of country capacity. Assessment of national
capacity for non-communicable disease prevention is in
development. In particular, the validity of the data we
show requires further assessment. Publication of these
data will be a stimulus for countries to examine and
improve their data. Future assessments need to include
details of chronic disease funding as a percentage of total
health budget and details of the sta employed nationally
and regionally to undertake prevention and control of
these diseases.
In the absence of reliable data for risk factors and causespecic mortality, accurate assessment of the eects of
policy and programmes is dicult. Countries need to
award higher priority to strengthening of their core data
for non-communicable diseases than exists at present
and improve vital registration, which is essential for
achievement of high-quality mortality data.35 However, in
regions where vital registration is not available or
inadequate from a coverage or quality perspective,
www.thelancet.com Vol 376 November 27, 2010

Series

countries Ministries of Health should consider


establishment of alternative methods such as verbal
autopsy36,37 as an interim measure before improvements
to their vital registration system can be made. Capacity
building is key, and there is an urgent need for certication
training for physicians and other health workers to
reliably establish causes of death with the WHO
international certicate and apply the rules and procedures
of the International Classication of Diseases.38
We noted high rates of tobacco use for many of the
23 countries that we examined. The WHO FCTC provides
the international legal foundation for countries to
implement measures for tobacco control. However, the
review of progress in implementation of selected
demand-reduction measures suggests that progress is
slow and there is still much work to be done in these
23 countries; tobacco control remains an immediate
priority for prevention of non-communicable diseases in
most countries.
Overweight and obesity are of public health concern in
most of the 23 countries. In 2004, WHO member states
endorsed the global strategy for diet, physical activity,
and health to address two major risk factorsunhealthy
diet and physical inactivity. Although the strategy has
been in force for 6 years, the multisectoral policy response
required to address obesity has been slow.
From the 2010 capacity assessment and comparison
with the 2000 survey, we suggest that progress has
occurred over the past decade. However, major gaps
remain in the accuracy, quality, standardisation of risk
factor data, and reporting of non-communicable disease
outcomes such as cause-specic mortality in a large
proportion of countries. Concerns about the quality and
comparability of data remain, which are major challenges
to overcome if comprehensive global monitoring and
surveillance for such diseases is to occur.

Conclusions
This dearth of capacity, which includes important gaps in
surveillance, is a major challenge to global health
development. The negative eects on health and
socioeconomic development are increasingly understood
and require a strong response from all countries. Recent
developments, including the UN General Assembly
resolution on non-communicable diseases, call for a high
priority to be given to these diseases, not only within the
health sector, but also within the national and global
development agendas.39 It calls for a high-level meeting
of the General Assembly on non-communicable diseases
in September, 2011, with the participation of heads of
state and government. The resolution requests the UN
Secretary General to submit a report on the global status
of these diseases. One area of emphasis during the
discussions leading to the high-level meeting in 2011 will
therefore cover the need for strengthening of global
monitoring of trends and approaches, for integration of
non-communicable disease surveillance into national
www.thelancet.com Vol 376 November 27, 2010

health information systems. A key outcome of this


meeting will be positioning of the prevention of such
diseases as central to country development.40
The UN resolution39 provides an unprecedented
opportunity for countries, UN agencies, international
development organisations, civil society, and the private
sector to work jointly around a common vision and a
road map to address the enormous chronic-disease
burden that low-income and middle-income countries
are increasingly facing. Challenges to prevention of
non-communicable diseases, such as strengthened
surveillance, more eective intersectoral action, and
improved access to basic health care (including essential
medicines and technologies), are key issues that
are pivotal to include in the discussions over the
coming months.
Contributors
AA and DRM conceptualised and designed the study, interpreted the
data, drafted and revised the report, gave nal approval for publication,
and are the guarantors of the paper. LMR interpreted the data and
helped to draft the report, with substantial input to revisions, including
the published version. ETdE, GAS, CDM, and DB helped to interpret the
data and draft sections of the report.
Conicts of interest
We declare that we have no conicts of interest.
References
1
Abegunde DO, Mathers CD, Taghreed A, Ortegon M, Strong K.
The burden and costs of chronic diseases in low-income and
middle-income countries. Lancet 2007; 370: 192938.
2
WHO. Preventing chronic diseases: a vital investment. Geneva:
World Health Organization, 2005.
3
Horton R. Chronic diseases: the case for urgent global action.
Lancet 2007; 370: 188182.
4
Geneau R, Stuckler D, Stachenko S, et al. Raising the priority of
preventing chronic diseases: a political process. Lancet 2010;
published online Nov 11. DOI:10.1016/S0140-6736(10)61414-6.
5
Beaglehole R, Ebrahim S, Reddy S, Vote J, Leeder S, on behalf of
the Chronic Disease Action Group. Prevention of chronic diseases:
a call to action. Lancet 2007; 370: 215257.
6
WHO. WHO report on the global tobacco epidemic, 2009:
implementing smoke-free environments. Geneva: World Health
Organization, 2009.
7
WHO. Global strategy for the prevention and control of
noncommunicable diseases (WHAA53/14). Geneva: World Health
Organization, 2000.
8
WHO. Framework convention on tobacco control (WHA56.1).
Geneva: World Health Organization, 2003.
9
WHO. Global strategy on diet, physical activity and health
(WHA57.17). Geneva: World Health Organization, 2004.
10 WHO. Action plan for the global strategy for the prevention and
control of noncommunicable diseases. Geneva World Health
Assembly: A61/8. Geneva: World Health Organization, 2008.
11 Alwan A, Maclean DR, Mandil A. Assessment of national capacity
for the prevention and control of noncommunicable diseases: the
report of a global survey, WHO/MNC/01.2. Geneva: World Health
Organization, 2001.
12 WHO surveillance and monitoring. http://www.who.int/tobacco/
surveillance/en (accessed Feb 15, 2010).
13 WHO. WHO STEPS surveillance manual: the WHO STEPwise
approach to chronic disease risk factor surveillance. Geneva: World
Health Organization, 2005.
14 WHO Global InfoBase. https://apps.who.int/infobase
(accessed Feb 15, 2010).
15 WHO. The global burden of disease: 2004 update. Geneva: World
Health Organization, 2008.
16 Mathers CD, Boerma T, Ma FD. Global and regional causes of
death. Br Med Bull 2009; 92: 732.

1867

Series

17
18
19

20

21

22
23
24

25

26
27
28
29

1868

WHO. World health statistics 2010. Geneva: World Health


Organization, 2010
WHO Mortality Database 2008. http://www.who.int/healthinfo/
morttables/en/index.html (accessed Feb 10, 2010).
Black RE, Cousens S, Johnson HL, et al, for the Child Health
Epidemiology Reference Group of WHO and UNICEF. Global,
regional, and national causes of child mortality in 2008:
a systematic analysis. Lancet 2010; 375: 196987.
Ferlay J, Shin H, Bray F, Foreman D, Mathers CD, Parkin DM.
Estimates of worldwide burden of cancer in 2008: Globocan 2008.
Int J Cancer 2010; published online June 17. DOI:10.1002/ijc.25516.
Jha P, Gajalakshmi V, Gupta PC, et al. Prospective study of one
million deaths in India: rationale, design, and validation results.
PLoS Med 2006; 3: e18.
Registrar General of India. Causes of death in India in 20012003.
New Delhi: Registrar General of India, Government of India, 2009.
Ministry of Health. Report on the third national retrospective survey
on mortality. Beijing, China: Ministry of Health, 2008.
Porapakkham Y, Rao C, Pattaraarchachai J, et al. Estimated causes
of death in Thailand, 2005: implications for health policy.
Popul Health Metr 2010; 8: 14.
Ngo AD, Rao C, Hoa NP, et al. Mortality patterns in Vietnam, 2006:
ndings from a national verbal autopsy survey. BMC Res Notes 2010;
3: 78.
Mathers CD, Loncar D. Projections of global mortality and burden
of disease from 2002 to 2030. PLoS Med 2006; 3: e442.
World Bank. Global economic prospects, 2009. Washington, DC:
The World Bank, 2009.
World Bank. Global economic prospects, 2010. Washington, DC:
The World Bank, 2010.
University of Denver. International futures: exploring alternative
global possibilities, 2009. http://www.ifs.du.edu (accessed Sept 18,
2010).

30
31

32

33
34

35

36
37
38
39
40

Strong K, Guthold R, Yang J, et al. Tobacco use in the European


region. Eur J Cancer Prev 2008; 17: 16268.
WHO Global InfoBase Team: the SuRF report 2. Surveillance of
chronic disease risk factors: country-level data and comparable
estimates. Geneva: World Health Organization, 2005.
WHO. Discussion paper: noncommunicable diseases, poverty and
the development agenda (July 2009) ECOSOC high-level segment
2009. http://www.who.int/nmh/publications/discussion_paper_
ncd_en.pdf (accessed Aug 27, 2010).
Gaziano TA, Galea G, Reddy KS. Scaling up interventions for
chronic disease prevention: the evidence. Lancet 2007; 370: 193946.
World Economic Forum. Global risks 2010: a global risk network
report. http://www.weforum.org/pdf/globalrisk/globalrisks2010.pdf
(accessed Aug 27, 2010).
Mahapatra P, Shibuya K, Lopez AD, et al, on behalf of the
Monitoring Vital Events (MoVE) writing group. Civil registration
systems and vital statistics: successes and missed opportunities.
Lancet 2007; 370: 165363.
Baiden, F, Bawah B, Biai S, et al. Setting international standards for
verbal autopsy. Bull World Health Organ 2007; 85: 569648.
WHO. Verbal autopsy standards: ascertaining and attributing cause
of death. Geneva: World Health Organization, 2007.
WHO. International statistical classication of diseases, 10th
revision, 2nd edition. Geneva: World Health Organization, 2005.
UN General Assembly. Resolution on noncommunicable diseases,
A/RES/64/265. Geneva: United Nations General Assembly, 2010.
Alleyne G, Stuckler D, Alwan A. The hope and promise of the UN
resolution on noncommunicable disease. Global Health 2010; 6: 15.

www.thelancet.com Vol 376 November 27, 2010

Viewpoint

Universal access to malaria medicines: innovation in


nancing and delivery
Olusoji Adeyi, Rifat Atun

In recent years, several constraints have impeded access


to eective treatments for malaria due to
Plasmodium falciparum. First, the parasite has become
increasingly resistant to established cheap drugs, such as
chloroquine and sulfadoxine-pyrimethamine. Second,
development assistance has been routed largely
through public channels, whereas aected individuals
seek treatment mostly through the private sector.
Finally, new artemisinin-based combination treatments
(ACTs), recommended by WHO for uncomplicated
falciparum malaria,1 are too expensive for many people
who seek treatment in the private sector. These three
restrictions have resulted in low coverage of ACTs and
persistent use of oral artemisinin monotherapy,
thereby increasing the risk of widespread parasite
resistance to artemisininthe only widely eective rstline treatment.
The three constraints noted above have prompted
exploration of new approaches to resolve diculties of
low coverage of combination regimens and continued
use of inappropriate monotherapy. The Aordable
Medicines Facilitymalaria (AMFm) is an innovative
nancing mechanism to expand access to aordable
ACTs through the public and private sectors
and non-governmental organisations (NGOs) and,
crucially, to displace oral artemisinin monotherapies
from the market.2,3
AMFm has the potential to transform the way universal
access to new malaria drugs and similar technologies is
nanced. Managed by the Global Fund, it aims to reduce
the cost of ACTs sold in the private sector, from up to
US$11 per treatment at present to the same price as
chloroquine or sulfadoxine-pyrimethamine (about $050)
and to less than the cost of oral artemisinin monotherapy
(about $37). Patients who receive malaria treatment
through public-sector clinics and not-for-prot services
will also benet from increased access to free or
low-cost ACTs.
The origins of AMFm have been described previously,4,5
and ndings of two small pilot projects indicate that its
basic design works in practice.6,7 Here, we present an
interim perspective on implementation of AMFm and
potential lessons for the architecture of nancing
universal access to lifesaving health technologies.
The design of AMFm incorporates three elements:
(1) price reductions through negotiations with
manufacturers of ACTs; (2) a buyer subsidy, via a
co-payment at the top of the global supply chain; and
(3) support of interventions to promote appropriate use
of ACTs.8 The key innovation is the combined approach
to reduce prices substantially by negotiation with
www.thelancet.com Vol 376 November 27, 2010

manufacturers and by global subsidy. This objective


entails payment by AMFm of a large part of the
post-negotiation price (the co-payment) on behalf of
eligible rst-line buyers from the public and private
sectors and NGOs, who all purchase ACTs directly from
the manufacturer.
AMFm has two funding streams. A rst co-payment
fund of US$216 millionnanced by the Bill &
Melinda Gates Foundation, the UK Government, and
UNITAIDcovers the subsidies. A second allocation of
US$127 million from the Global Fund nances
supporting interventions.
After Global Fund board approval in November, 2009,
phase one of AMFm started in mid-2010 and is
scheduled to last for 2 years. It will be implemented in
eight countries: Cambodia, Ghana, Kenya, Madagascar,
Nigeria, Niger, Tanzania, and Uganda. AMFm will
provide co-payments for rst-line buyers of ACTs in
these countries, but it will only co-pay for the purchases
of products that meet quality criteria of the Global
Funds quality assurance policy.9
AMFm has concluded master supply agreements with
six pharmaceutical companies that met its quality criteria
for supply of ACTs to rst-line buyers, namely: Ajanta
Pharma, Cipla, Guilin Pharmaceutical, Ipca Laboratories,
Novartis, and Sano-Aventis. This achievement is
important because, in a departure from previous
practices, manufacturers will sell ACTs to rst-line buyers
from the private sector at the same reduced prices as they
sell to public-sector buyers. Manufacturers have agreed
to not market oral artemisinin monotherapy. AMFm has
set a maximum acceptable price that manufacturers may
charge rst-line buyers for each formulation pack size.
The subsidy is then applied to each quoted price in the
form of a xed co-payment.
Four challenges to implementation of AMFm, which
are most directly related to its design, are: (1) passing the
subsidy on to patients at the retail level; (2) learning the
most eective ways to expand access to diagnostic tests
for malaria; (3) reaching poor and remote populations
with ACTs;10,11 and (4) nding the appropriate approach to
evaluation and benchmarking of phase one. We examine
each of these challenges in turn.
First, the risk with a high-level subsidy is that some of
its benets might be captured by rst-line buyers and
middlemen. To mitigate this risk, the Global Fund
requires every rst-line buyer to sign an undertaking to
pass on to those further down the supply chain the
benets of the buyer subsidy that they (the rst-line
buyers) will enjoy. First-line buyers commit to addition of
no more than a reasonable margin, dened as the margin

Lancet 2010; 376: 186971


Published Online
October 11, 2010
DOI:10.1016/S01406736(10)61189-0
Global Fund to Fight AIDS,
Tuberculosis and Malaria,
Geneva, Switzerland
(O Adeyi DrPH,
Prof R Atun FFPHM); and
Imperial College London,
London, UK (Prof R Atun)
Correspondence to:
Dr Olusoji Adeyi, Global Fund to
Fight AIDS, Tuberculosis and
Malaria, Chemin de
Blandonnet 8, CH-1214 Vernier,
Geneva, Switzerland
Olusoji.Adeyi@Theglobalfund.
org

1869

Viewpoint

they would normally add to other antimalarials bought at


prices comparable with the subsidised ACTs. Findings of
a pilot study of ACT subsidies in Tanzania showed that
the approach worked without price gouging by
middlemen.6 Nevertheless, there is no guarantee that
every participant in the supply chain will honour all
agreements at all times.
The Global Fund has commissioned an AMFm-specic
logo that countries are using for branding and marketing
of co-paid ACTs. The logo also has a potential benet of
enabling buyers to identify subsidised drugs that should
cost substantially less than unsubsidised products.
With respect to the second challenge, in 2009, the
Global Fund encouraged applicant countries to include
expansion of access to diagnostic tests for malaria as a
supporting intervention in their AMFm applications,
including operational research to inform wider use in the
private sector. These diagnostic tests will not be nanced
from the co-payment fund for subsidies but from grants
for supporting interventions.
WHO updated its guidelines in 2010 for parasitological
conrmation of malaria.12 Universal access to diagnostic
tests will help restrict use of ACTs to only patients with
parasitaemia. A large proportion of presumptive
diagnosis and treatment of malaria takes place in the
private sector. For example, in the Democratic Republic
of Congo, the private sector dominates the market for
antimalarials, distributing 856% of all drugs of this
type. In Nigeria, the private sector accounts for
about 95% of all antimalarial drug distribution.13
In 2006, these two countries alone comprised just over
a third of all cases of suspected malaria in the WHO
Africa region. Formal public-sector services are unlikely
to cover all cases of suspected malaria in the near-tomedium term. Therefore, universal access to diagnostic
tests requires that all patients with suspicion of malaria
in the private sector be tested by a suitable method. Yet,
knowledge is modest of scalable approaches to achieve
this goal in the private sector. This phase one
implementation provides an opportunity to learn about
scalable approaches before a potential global roll-out of
AMFm in 2012.
The third challenge relates to the speed at which
subsidised ACTs will reach the poorest and most remote
populations. Improvement of access in distant locations
is the focus of some supporting interventions and
operations research in AMFm phase one. The new
nancing architecture of AMFm can benet any service
delivery model because service delivery organisations
will be able to buy ACTs at reduced prices, but any one
approach to service delivery is unlikely to work universally
and uniformly on a large scale.
The nal challenge of implementation relates to
evaluation and benchmarking of AMFm phase one.
Universal access to eective treatments is a goal of the
Roll Back Malaria Partnership,14 but despite ocial
commitments and substantial increases in nancing,
1870

this objective remains elusive. In 2009, the conclusion of


the 5-year evaluation of the Global Fund stated:
The ndings related to ACTs are the most perplexing
and worrisome of the four primary malaria interventions
because they show the least improvement. While there
are data showing that most countries have purchased
large amounts of ACT, there is little or no evidence of a
corresponding increase in the use of ACT for treatment
of children [p ES-38].15

This observation could be accounted for by a lag


between nancial commitments and implementation of
activities. However, reports based on more recent data
indicate little progress in overall access to ACTs, with
price remaining a major barrier to access.13 In Africa,
only 14 countries reported distribution of enough ACTs
to treat at least 50% of reported malaria cases in the
public sector, and only ve countries reported allocation
of sucient ACTs to treat all reported cases of malaria
in 2008.16
Evidence of the success of AMFm phase one, to be
ascertained by independent assessment of the pilot
projects, will inform a decision by the Global Fund
board on the future of AMFm. Endpoint evaluation data
will be gathered after about 1 year of implementation.
Two complementary approaches could be used,
inter alia, to assess success. The rst entails
measurement of how far AMFm progresses towards
achievement of its stated objectivesto reduce prices,
increase availability of ACTs, gain market share, and
augment use of ACTsin view of the period of
implementation. The second compares AMFm with
other nancing platforms that have similar objectives,
whether such nancing platforms seek to achieve those
objectives through the public sector or through several
sectors. Such direct comparisons would provide (for the
rst time and with similar indicators) independent and
veriable comparisons of performance of nancing
platforms. An obvious comparator is the existing Global
Fund malaria grant nancing scheme, because if the
Global Fund had additional resources without AMFm,
such resources would normally be channelled through
those grants. Other potential comparators might
include non-Global Fund programmes, such as the
US Presidents Malaria Initiative17 and the World Banks
Booster Program for Malaria Control.18 Such
comparisons, which make data, methods, and results
available in the public domain, would promote a culture
of independent and contestable monitoring and
evaluation of performance in an era of tight budgets,
with increased emphasis on performance and value
for money.19
Time is a relevant factor in assessment of the
performance of AMFm phase one; the report of the
5-year evaluation of the Global Fund20 is instructive in
setting expectations of achievements within the duration
of AMFm phase one. Another important principle is to
www.thelancet.com Vol 376 November 27, 2010

Viewpoint

compare like with like. Finally, the scale of comparators


is relevant. Unless these factors are taken into
consideration, there will be a high risk of conating a
nancing mechanism such as AMFm, which benets all
approaches to service delivery, with specic schemes.
Comparison of a short-lived proof-of-concept such as
AMFm with initiatives that have had much longer periods
to mature will be a further risk.
In our view, an appropriate expectation for the short
duration of the pilot phase is that the innovative aspects
of AMFm are shown to work, in terms of direction of
change in price and availability of ACTs and in initial
displacement of oral artemisinin monotherapies from
the market. These achievements can be assessed at
outlets in the public and private sectors and compared
with independent evaluations of other nancing
mechanisms after similar periods of implementation.
Expectations of attributable and rapid increases in
measures of service delivery at the household level, which
are neither new nor unique to AMFm, are inappropriate
and unrealistic within the duration of the pilot studies.
Crucially, they are incompatible with the key nding of
the 5-year evaluation study of the Global Fund in relation
to time needed for maturation of new approaches.20
In conclusion, available evidence suggests that the
traditional approach to development assistance for
malaria treatment, which puts most resources through
the public sector alone, will not achieve by 2015
Millennium Development Goal 6, of universal access to
malaria treatment. AMFm will test a new and
complementary architecture of nancing and
development assistance for malaria drugs. Assessment
of its rst phase provides an opportunity to learn how
well AMFm works and how it compares with traditional
models of nancing.
Contributors
Both authors had the original idea for and wrote the paper.
Conicts of interest
OA is director of AMFm and was formerly coordinator of public-health
programmes at the World Bank, where he led the project to develop
AMFm. RA is director of the strategy, performance and evaluation
cluster at The Global Fund.
References
1
WHO. Guidelines for the treatment of malaria. Geneva: World
Health Organization, 2006.
2
Arrow K, Panosian C, Gelband H. Saving lives, buying time:
economics of malaria drugs in an age of resistance. Washington:
National Academies Press, 2004.
3
Laxminarayan R, Over M, Smith DL. Will a global subsidy of new
antimalarials delay the emergence of resistance and save lives?
Health A (Millwood) 2006; 25: 32536.

www.thelancet.com Vol 376 November 27, 2010

4
5

10

11

12

13

14

15

16
17

18
19

20

Laxminarayan R, Gelband H. A global subsidy: key to aordable


drugs for malaria? Health A (Millwood) 2009; 28: 94961.
Frost L, Reich MR, Pratt BA, Guyer AL. Process evaluation of the
project on Dening the Architecture and Management of a Global
Subsidy for Antimalarial Drugs. March 31, 2009. http://site
resources.worldbank.org/INTMALARIA/Resources/
AMFmProcessEvaluation.pdf (accessed Jan 11, 2010).
Sabot OJ, Mwita A, Cohen JM, et al. Piloting the global subsidy:
the impact of subsidized artemisinin-based combination therapies
distributed through private drug shops in rural Tanzania.
PLoS One 2009; 4: e6857.
Talisuna A, Grewal P, Rwakimari JB, Mukasa S, Jagoe G, Banerji J.
Cost is killing patients: subsidising eective antimalarials.
Lancet 2009; 374: 122426.
AMFm Task Force of the Roll Back Malaria Partnership. Aordable
Medicines Facilitymalaria (AMFm): technical design.
November, 2007. http://www.rollbackmalaria.org/partnership/tf/
globalsubsidy/AMFmTechProposal.pdf (accessed Aug 5, 2010).
The Global Fund. Aordable Medicines Facilitymalaria: frequently
asked questions. March 29, 2010. http://www.theglobalfund.org/
documents/amfm/AMFmFAQs_en.pdf (accessed May 15, 2010).
Gwatkin DR, Wagsta A, Yazbeck AS. Reaching the poor with
health, nutrition, and population services: what works, what
doesnt, and why. 2005. http://siteresources.worldbank.org/
INTPAH/Resources/Reaching-the-Poor/complete.pdf (accessed
Aug 5, 2010).
The Economist. A spoonful of ingenuity: new ideas for raising
money for medical careand spending it. Jan 7, 2010.
http://www.economist.com/node/15213715 (accessed Sept 16, 2010).
WHO. Guidelines for the treatment of malaria, second edition.
2010. http://whqlibdoc.who.int/publications/2010/9789241547925_
eng.pdf (accessed Aug 5, 2010).
ACTwatch Group. Availability, volumes, price and use of
antimalarials in 7 malaria-endemic countries. Presentation at the
5th Pan African Malaria Conference; Nairobi, Kenya;
November, 2009. http://www.actwatch.info/results/overview.asp
(accessed Sept 16, 2010).
Roll Back Malaria Partnership. The Global Malaria Action Plan:
for a malaria-free world. 2008. http://www.rollbackmalaria.org/
gmap/gmap.pdf (accessed Jan 4, 2010).
The Global Fund. Global Fund ve-year evaluation:
study area 3the impact of collective eorts on the reduction of the
disease burden of AIDS, tuberculosis and malaria. May, 2009.
http://www.theglobalfund.org/documents/terg/TERG_SA3.pdf
(accessed Aug 5, 2010).
WHO. World malaria report 2009. 2009. http://whqlibdoc.who.int/
publications/2009/9789241563901_eng.pdf (accessed Aug 5, 2010).
Presidents Malaria Initiative. About the Presidents Malaria
Initiative. http://www.ghtingmalaria.gov/about/index.html
(accessed Aug 5, 2010).
The World Bank. Counting malaria out! June 4, 2010.
http://go.worldbank.org/IQFZ26ZIU0 (accessed Aug 5, 2010).
Lim SS, Stein DB, Charrow A, Murray CJL. Tracking progress
towards universal childhood immunisation and the impact of global
initiatives: a systematic analysis of three-dose diphtheria, tetanus,
and pertussis immunisation coverage. Lancet 2008; 372: 203146.
Global Fund. The ve-year evaluation of the Global Fund to Fight
AIDS, Tuberculosis, and Malaria: synthesis of study
areas 1, 2 and 3. March, 2009. http://www.theglobalfund.org/
documents/terg/TERG_synthesis_report.pdf (accessed
Sept 16, 2010).

1871

Case Report

A hearty sneeze
Harry A Roselle, Marc Herman
Lancet 2010; 376: 1872
Department of Medicine
(H A Roselle MD) and
Department of Radiology
(M Herman MD), Englewood
Hospital and Medicine Centre,
New Jersey, USA
Correspondence to:
Dr Harry A Roselle, Department
of Medicine, Englewood
Hospital and Medical Centre,
309 Engle St, Englewood,
New Jersey 07631, USA.
rosellesrhj@aol.com

In February, 2009, an 84-year-old woman was admitted to


our clinic with a fractured left arm. The fracture was treated
conservatively with splinting. She had left maxillary
swelling and ecchymosis. Facial radiography showed no
fracture. During discharge, she sneezed and had instant
loss of vision in the left eye. The left orbital area had acute
swelling, (gure A) and crepitus. A CT of the orbit showed
orbital air, but no overt fracture (gure B). Otolaryngology
and ophthalmology examination showed normal eye and
nasopharyngeal function. She was treated with neosynephrine nasal spray and antibiotics. The patient was
advised not to blow her nose.
Orbital emphysema is uncommon, occurring secondary
to air entering the tissues of the orbital area. The condition
has been seen in cases of sinus infection, facial trauma, or
surgery.1 A paper-thin area of the orbital wall (lamina
papyracea), the most common site of the bony defect,
allows air to pass from the medial paranasal sinuses to the
orbit. Our patient had normal eye movements and vision
on subsequent eye examinations. She had lateral maxillary
ecchymosis and there were no fractures seen on CT. An
undetectable anatomical defect might have been present
in the orbital wall and sneezing opened it up for air to ow
into the orbital tissue. She was discharged, and 10 days
later, on follow-up, there was complete resolution.
Orbital emphysema is most commonly seen with
traumatic orbital fracture, but is also seen in dental,
otolaryngeal procedures, infection, breath holding, and
COPD with bronchopleural stula.2,3 Nose blowing, and
sneezing, can lead to sudden increase in intranasal
pressure causing dehiscence of the lamina papyracea,
which allows air to enter the surrounding orbital tissue.
A

Orbital emphysema can cause ischaemic optic neuritis


and central retinal artery occlusion leading to blindness.
When signs of pressure eect occur, such as restricted
ocular motion, pupillary sluggishness, disc oedema, or
decreased or double vision, trapped air can be removed
by underwater drainage with a 24 gauge needle or by
lateral canthotomy.4 Treatment includes cold packs, oral
antibiotics, and nasal spray decongestants. Although the
role of antibiotics is unclear, prophylactic use is common,
especially in the presence of sinusitis, dental caries, or
other infectious sources. The patient should avoid nose
blowing and have otolaryngology and ophthalmology
follow-up. According to most studies, most cases have a
benign outcome.13 However, the risk of complications
and medico-legal issues are important considerations for
the managing doctor.5
Contributors:
HAR looked after the patient and wrote the report. MH contributed the
images and the critical review of the report. Both authors approved the
nal manuscript. Written consents to publish were obtained.
References:
1
Chiu W, Lih M, Huang T, Ku W, Wang W, Spontaneous orbital
subcutaneous emphysema after sneezing. Am J Emerg Med 2008;
26: 381 82.
2
Bolognini A, Delehage E, Case M, Cosso L. Barometric orbital
emphysema of rhinogenic origin in a breath-hold diver. A case
report. Undersea Hyperbaric Med 2008; 35: 16367.
3
Upadhyay D, Sporn P. Orbital emphysema in COPD with
bronchopleural stula. Int J Chron Obstruct Pulmon Dis 2007;
2: 18788.
4
Benharbit M, Karim A, Mohcine Z. Emergency treatment of post
traumatic Orbital Emphysema. A case report. J Fr Ophthalmol 2003;
26: 95759.
5
Wright J, Opperman G. Negligent management of orbital
emphysema leading to blindness AvMA Med Legal J 2008; 14: 15759.

Figure: Acute orbital emphysema of the left eye


(A) Extensive bruising and swelling of left orbital area, and (B) CT shows presence of orbital air in the left orbit but no overt fracture.

1872

www.thelancet.com Vol 376 November 27, 2010