Double blind, cluster randomised trial of low dose supplementation with vitamin A or βcarotene on mortality related to pregnancy in NepalBMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7183.570 (Published 27 February 1999) Cite this as: BMJ 1999;318:570
All rapid responses
Our most valid estimate of mortality impact derives from comparing
supplement-specific rates based on all deaths and pregnancies during a
common follow-up period, adjusted for design effect. This analysis showed
that vitamin A and carotene reduced mortality of mothers through 3
months post-partum by 40% and 49%.1 Each effect was statistically
significant from placebo. Undo reliance on lay verbal autopsy findings
may lead to considerable misclassification of deaths by cause;2 thus,
stratified analyses enrich interpretation and raise questions, but do not
invalidate intent-to-treat findings. Ronsmans and Collumbien are correct
in surmising that the pregnancy-related effect of supplementation is not
offset by a qualitative change in adult female mortality (unpublished
data). These results are being reported separately.
We combined vitamin A and -carotene effects for several reasons:
-carotene is the most efficiently converted provitamin A carotenoid and
the major dietary source of vitamin A in South Asia.3 Typically, an
improved diet comprises both precursor carotenoids and preformed vitamin
A. Expectedly, both supplements raised serum retinol, although -carotene
was less effective than vitamin A. Finally, mortality effects of vitamin
A and -carotene were comparable and statistically indistinguishable,
allowing us to also pool estimates given their overlapping nutritional
Lack of a linear relationship between mid-pregnancy serum retinol and
risk reduction in maternal mortality should not be surprising. Once
normalized, circulating retinol is homoeostatically controlled.4 Animal
studies show mortality reversal with small amounts of added vitamin A,
before tissue concentrations respond.5 A greater impact of -carotene on
mortality, given less impact on circulating retinol,1 night blindness6 and
perhaps other vitamin A-related outcomes may have been due to chance or
could reflect benefit of -carotene beyond its provitamin A role. An
antioxidant hypothesis seems plausible in women continuously exposed to
dietary deficiencies, infection, multiple pregnancies, heavy work,
sunlight, smoke through cigarettes and cooking fires, and other
environmental stresses to oxidative balance.
Two commentors are concerned that 157 of 20,119 women lost to follow-
up ~2 weeks after birth may have died, distorting the vitamin A effect.
More likely, these women emigrated to distant relatives. Sachdev's charge
that 10% died, supported by analysis of "adjusted data" in his table, is
untenable given an expectation of <½ death in the entire group based
on observed rates.1
We agree it is too early to divine policy from the Nepal findings and
that replicative work is needed. Still, improving vitamin A or -carotene
intakes may enhance pregnancy-related survival.
Keith P. West Jr., DrPH
Joanne Katz, ScD
Subarna Khatry, FRCS
1. West KP, Jr., Katz J, Khatry SK, LeClerq SC, Pradhan EK, Shrestha
SR, et al. Double blind, cluster randomised trial of low dose
supplementation with vitamin A or b-carotene on mortality related to
pregnancy in Nepal. B.M.J. 1999;318:570-575.
2. Ronsmans C, Vanneste AM, Chakraborty J, van Ginneken J. A
comparison of three verbal autopsy methods to ascertain levels and causes
of maternal deaths in Matlab, Bangladesh. Int.J.Epidemiol. 1998;27:660-
3. Gopalan C, Rama Sastri BV, Balasubramanian SC, Narasinga Rao BS,
Deosthale YG, Pant KC. Nutritive Value of Indian Foods. New Delhi,
India: ICMR Offset Press, 1989.
4. Green MH, Green JB. Dynamics and control of plasma retinol. In:
Blomhoff R, (ed.), Vitamin A in Health and Disease. New York: Marcel
Dekker, Inc., 1994:119-133.
5. West KP, Jr. Dietary vitamin A deficiency: effects on growth,
infection, and mortality. Food Nutr.Bull. 1991;13:119-131.
6. Christian P, West KP, Jr., Khatry SK, Katz J, LeClerq SC, Pradhan
EK, et al. Vitamin A or B-carotene supplementation reduces but does not
eliminate maternal night blindness in Nepal. J.Nutr. 1998;128:1458-1463.
Competing interests: No competing interests
There Are No Magic Pills for Reducing Pregnancy Related Mortality
The conclusion drawn from this trial is likely to influence policy
adopt a "magic pill" approach for a dramatic reduction of pregnancy
mortality. I urge extreme caution.
The impact of supplements on the hypothesised outcome - infection
ascribable mortality-and obstetric causes was non-existent. The two
interventions however succeeded dramatically for injuries (burns,
snakebite and hanging), miscellaneous and uncertain causes. The authors
related this discrepancy to the relative crudeness of "verbal autopsy"
technique. However, better (not actual) relative risks can be determined
safely excluding deaths with a reliably determined cause which can not be
biologically related to supplementation, for example, injury (Table 1).
correction alone results in statistically comparable outcomes in the
and vitamin A groups. Similar adjustments for chronic illness (leukaemia,
asthma) and differential loss to follow up (assuming 10% loss was
related) strengthen the conclusion that vitamin A supplementation does not
reduce pregnancy related mortality.
In the two intervention groups, biochemical data suggests a trend
the expected vitamin A dose response relationship. Paradoxically, beta
carotene group with a greater mortality reduction (RR 0.51) had
lower (p<_0.002 serum="serum" retinol="retinol" levels="levels" than="than" the="the" vitamin="vitamin" a="a" group="group" rr="rr" _0.60.="_0.60." further="further" beta="beta" carotene="carotene" in="in" placebo="placebo" and="and" groups="groups" were="were" identical="identical" but="but" significantly="significantly" lower="lower" those="those" group.="group." light="light" of="of" this="this" biochemical="biochemical" heterogeneity="heterogeneity" it="it" is="is" inadvisable="inadvisable" to="to" combine="combine" two="two" different="different" intervention="intervention" compute="compute" relative="relative" risks="risks" which="which" have="have" spuriously="spuriously" narrow="narrow" confidence="confidence" intervals.="intervals." p="p"/> There is no quantification of vital determinants of mortality -
obstetric care provided in 97% of home deliveries, comparability of
ill" referrals and anaemia. Data cited elsewhere reveals that anaemia
a surprising 45 per cent lower in the women receiving supplements who were
not infected with hookworm". Inexplicably the authors omitted this
Other trials conducted by donors and individuals with no conflict of
are imperative for validation, particularly in settings with at least
obstetric services. Meanwhile from this study it can not be concluded that
vitamin A supplementation lowers pregnancy related mortality. Additional
arguments against adopting a narrow approach of supplementation include
logistics of weekly distribution, long term safety, cost effectiveness and
benefits for foetal or early infant mortality.
Finally, I question the motive and ethics of global dissemination of
study findings and conclusions more than one year prior to publication
peer reviewed journal. Interestingly, even advocacy was initiated - the
page of the Report stated "Vitamin A supplementation is helping bolster
disease resistance in children and may soon become an important measure
in helping reduce maternal deaths around the globe". I pray that the
concerned authorities are not mesmerised into operationalising the "magic
approach which will certainly compete with budgets allocated for providing
essential obstetric care.
Professor and Incharge,
Division of Clinical Epidemiology,
Department of Pediatrics,
Maulana Azad Medical College,
New Delhi 110 002, India.
1. West KP Jr, Katz J, Khatry SK, LeClerq SC, Pradhan EK, Shrestha
al. Double blind, cluster randomised trial of low dose supplementation
vitamin A or beta carotene on mortality related to pregnancy in Nepal. BMJ
2. United Nations Children's Fund. Vitamin A supplements save
women's lives. In: The State of the World's Children 1998. New York,
Oxford University Press, 1998; pp12-3.
Table 1. Recalculated impact of supplementation on mortality related to
pregnancy up to 12 weeks post partum.
Placebo Vitamin A Beta carotene No. of pregnancies 7241 7747 7201 Excluding injury No. of deaths 46 33 25 Relative risk 1.00 0.67 0.55 95% CI 0.41-1.10 0.32-0.93 p value 0.112 0.027 Excluding injury and chronic illness No. of deaths 43 33 23 Relative risk 1.00 0.72 0.54 95% CI 0.44-1.18 0.31-0.94 p value 0.194 0.028 Excluding injury and chronic illness and adjusted for loss to follow up No. of deaths 47 40 27 Relative risk 1.00 0.80 0.58 95% CI 0.51-1.26 0.35-0.98 p value 0.337 0.040
Competing interests: Placebo Vitamin A Beta caroteneNo. of pregnancies 7241 7747 7201Excluding injuryNo. of deaths 46 33 25Relative risk 1.00 0.67 0.5595% CI 0.41-1.10 0.32-0.93p value 0.112 0.027 Excluding injury and chronic illnessNo. of deaths 43 33 23Relative risk 1.00 0.72 0.5495% CI 0.44-1.18 0.31-0.94p value 0.194 0.028Excluding injury and chronicillness and adjusted for lossto follow upNo. of deaths 47 40 27Relative risk 1.00 0.80 0.5895% CI 0.51-1.26 0.35-0.98p value 0.337 0.040
I have a question.
What if the control groups had actually included a group that had
carrots, a group that educated women about what they need to eat during
pregnancy or even a group where the women recieved small business loans
and better education? We already know that better food, health education
and raising the status of women give better maternal and infant health
The first could probably be done within the time limits of such a
study (although it would be quite hard to think of how you could design a
placebo for a carrot), the other two would take much longer, yet they are
the ones that would be most likely to have a sustainable impact in the
The impact of such a large trial based on supplements goes way beyond the
outcomes of the trial. The large local health workforce involved has been
inculcated with the view that nutritional problems are best solved with a
pill. The health system in Nepal will also have got this message. Finally,
the donors who now determine the health priorities of many developing
countries will find the rationale very appealing.
I am not denying the importance of identifying the quantifiable
effects of specific micronutrients on maternal and infant mortality, nor
of conducting large blinded trials with measurable outcomes. However, the
discussion ends by saying the issue should be addressed by
supplementation or dietary measures. Note the order in which these are
put. Let us not forget that the medically accepted treatment of choice for
malnutrition is food, as part of a balanced diet. And the ways to ensure
that this happens are also well known, although they lack the "quick fix"
appeal of a tablet.
Competing interests: No competing interests