Effects of zinc supplementation on child mortality.

Malnutrition, including micronutrient defi ciencies, is the leading risk factor for child mortality in low-income and middle-income countries. About 30% of the world’s population has zinc defi ciency (fi gure). This defi ciency occurs because zinc is mainly contained in foods such as red meat that are expensive and in short supply in developing countries. Moreover, zinc has no tissue reserves, unlike vitamin A and iron, and its turnover is rapid, especially during common gastrointestinal infections. Young children in developing countries who have a poor diet and high exposure to gastrointestinal pathogens are at greatest risk of zinc defi ciency. Many clinical trials have shown zinc supplementation to be of benefi t to children in the treatment and prevention of diarrhoeal diseases and acute respiratory infections, and in improving growth. Evidence for benefi t on morbidity from malaria is less consistent. In view of these results, zinc supplementation was postulated to prevent up to 5% of deaths in children younger than 5 years, in countries that contribute 90% of worldwide child mortality. Recently, two important reports about the eff ect of zinc supplementation on child mortality have been published. In today’s Lancet, James Tielsch and colleagues report the results of a large randomised trial in Nepal; previously, Sazawal and colleagues reported on a similar trial in Zanzibar, a region with holoendemic malaria. Neither trial showed that zinc supplementation reduced mortality in children younger than 5 years. Nevertheless, when considering children older than 12 months only, both studies suggested a benefi t. In Zanzibar, supplementation resulted in a marginally signifi cant 18% reduction in mortality (relative risk 0·82, 95% CI 0·68–1·00). In Nepal, children aged 12–36 months had a slightly higher survival than did those younger than 12 months (hazard ratio 0·80, 95% CI 0·60–1·06), although this result was not statistically signifi cant. Interpretation of the results by looking at confi dence intervals avoids the yes/no dichotomy of hypothesis testing. In studies with results that do not reach statistical signifi cance, if the upper boundary of the confi dence interval includes an important benefi t, the possibility that the treatment still might be worthwhile has not been ruled out. The investigators might have missed a true treatment eff ect, and one should question if the sample size of the study was adequate (type II error). The stronger the non-signifi cant trend in favour of the experimental treatment, the more likely the investigators missed a true treatment eff ect. A pooled analysis of the trials exploring zinc’s eff ect in children older than 12 months shows a signifi cant 18% reduction in total mortality in this subgroup of children (relative risk 0·82, 95% CI 0·70–0·96). The pooled analysis has the advantage of augmenting the sample size and reducing the width of the confi dence intervals. Can we consider these pooled results conclusive? Should we then promote supplementation in this age group? First, we should note that in both trials zinc supplementation led to similar results. The point estimates are nearly the same, confi dence intervals overlap, and heterogeneity between trials is non-signifi cant (test for heterogeneity, I=0). Second, the diff erence in treatment eff ect by age is consistent with previous evidence. Third, an age eff ect is biologically plausible. Infants especially are not thought to be at high risk of zinc defi ciency. All these factors suggest that zinc can reduce mortality in children older than 12 months. However, questions remain. First, what is the size of the protective eff ect of zinc? From the pooled analysis we know that overall the real reduction of mortality in children older than 12 months lies somewhere between 4% and 30%. A 4% mortality reduction would be clinically relevant. Yet, getting this evidence to practice has many challenges. Second, can we generalise these results? Even if a trend suggesting a benefi cial eff ect of zinc supplementation <20% 20–30% 30–40% ≥40%