Intended for healthcare professionals


Vitamin A supplements in newborns and child survival

BMJ 2008; 336 doi: (Published 19 June 2008) Cite this as: BMJ 2008;336:1385
  1. James M Tielsch, professor
  1. 1Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205, USA
  1. jtielsch{at}

Benefit depends on the setting, baseline infant mortality, and vitamin A deficiency

Vitamin A supplementation at 6-72 months of age has become a mainline intervention for improving survival in populations with endemic vitamin A deficiency.1 However, in the same setting, supplementation at 1-6 months of age has little or no effect on mortality, whether given with immunisation or not.2 3

Giving vitamin A supplements to newborns within the first few days of life significantly reduced early infant mortality in Asian populations with endemic maternal vitamin A deficiency andhigh infant mortality.4 5 6 In Africa, however, this strategy had no beneficial effect on early infant survival in an urban setting7 or—as reported in the accompanying paper by Benn and colleagues—in a peri-urban setting.8

How can these findings be reconciled? Some might suggest that a meta-analysis of all the newborn dosing studies would provide a more accurate estimate of the true effect on early infant mortality. But combining these studies would be a fundamental mistake, because the variation in effect on early infant mortality in the five available studies is what would be expected given the variations in the populations enrolled in these studies.

So how can the variation in outcomes be explained? Differences between the study populations are key to understanding this variation. Vitamin A supplements have the greatest benefit in populations with high mortality and endemic vitamin A deficiency. Deaths from causes (such as infectious diseases) that are most likely to be affected by vitamin A supplements will already be low in populations with low mortality. Also, if a population is already receiving sufficient vitamin A from dietary sources, supplements are unlikely to improve vitamin A status enough to decrease case fatality from infectious diseases.

So, which of the five studies of vitamin A dosing in newborns were conducted in populations with high infant mortality and endemic vitamin A deficiency? The studies in India and Bangladesh were done in populations with high infant mortality.5 6 In these studies, rates of maternal night blindness in pregnancy were high enough to classify vitamin A deficiency as being of public health importance.6 9 In contrast, both African studies were done in populations with little, if any, vitamin A deficiency; mortality in the Zimbabwean study was very low7; and the study in Guinea-Bissau reduced mortality by excluding the highest risk infants (those with low birth weight) and giving free care and drugs to sick infants.8

Vitamin A status in newborns is difficult to interpret because all infants are born with low reserves of vitamin A, especially those born prematurely. Newborns depend on adequate supplies from breast milk or appropriate substitutes to satisfy physiological demands in early life.10 The risk of vitamin A deficiency in these situations should be judged by the status of the mothers, not the infants. Despite Benn and colleagues’ claim that differences in the characteristics of the study populations cannot explain the variation in treatment effects, they are clearly the most likely explanation.

The outlier study in this assessment is the Indonesian study, the smallest of the five studies. This study found that giving vitamin A supplements to newborns had a large effect on early infant mortality, even though the mothers’ serum retinol concentrations were relatively normal.4 But vitamin A deficiency is a common problem in women and children in Indonesia,11 and liver reserves may have been low even though serum retinol concentrations were adequate; alternatively, the finding in Indonesia may have been the result of chance.

Benn and colleagues’ study provides no new information on the interactions between vitamin A supplements and vaccines as all children received concurrent BCG and, we assume, the other standard vaccines including diphtheria, pertussis, and tetanus; oral polio; hepatitis B; and measles vaccines. Observational evidence suggests that giving vitamin A supplements to infants may modify the effect of vaccines on early infant mortality.12 However, little is likely to be gained from further observational data as the selection bias associated with the receipt of vaccines is almost impossible to overcome. Further delineating this interaction will require randomised trials of both newborn vitamin A supplementation and immunisation, which are unlikely given current immunisation policies.

Still missing from the evidence base for vitamin A supplementation in newborns are trials from populations in sub-Saharan Africa with endemic vitamin A deficiency and high infant mortality. Such populations exist in areas of eastern and southern Africa, and trials in these areas are urgently needed. Also, we still do not know why vitamin A supplements are beneficial when given in the first few days of life, at least in Asia, but have no effect on mortality when given only a few weeks later at the time of the first diphtheria, pertussis, and tetanus immunisation. Studies exploring the mechanisms of action of early versus later supplementation would be helpful in understanding this interesting discrepancy.

Given current evidence, Benn and colleagues’ conclusion that “a global or regional recommendation of 50 000 IU vitamin A supplementation at birth is unwarranted” is untenable. For populations in Asia who are deficient in vitamin A the evidence for benefit is convincing. Recommendations for Africa await the results of trials conducted in appropriate populations.



View Abstract