Effect of zinc supplementation on linear growth has been evaluated previously [3–5]. The first and most widely known meta-analysis evaluating the effect of zinc supplementation on growth was conducted by Brown et al in 2002 . This review included studies with children until pre-puberty and studies from both developing and developed countries. Pooled results from 33 studies showed that zinc supplementation had a highly significant positive effect on linear growth (effect size = 0.350, 95% CI: 0.189, 0.511). In the latest review published by the same authors , more studies have been added to the previous (2002) analysis and the beneficial effect remained significant [effect size = 0.170 (95% CI: 0.075, 0.264)]. In contrast, another meta-analysis by Ramakrishnan et al. based on 43 studies found no significant effect of zinc supplementation on linear growth [effect size = 0.07 (95% CI: -0.03, 0.17)] in children < 5 years of age . Our results from included studies from developing countries and with children < 5 years of age showed that preventive zinc supplementation (zinc ± iron) had a significant effect on linear growth [Effect size = 0.13 (95 % CI 0.04, 0.21)]. If we include results from four excluded studies from developed countries [23–26], the effect size comes to be 0.14 (95% CI 0.05-0.22) and if the results of studies with children > 5 years are included, the estimate becomes 0.15 (95 % CI 0.07, 0.23).
Our results are in concordance with that of Brown et al. 2009  but at variance with the findings of Ramakrishnan et al 2009 . What are the reasons for these contradictory findings and the differences between these reviews? The main differences were in inclusion and exclusion criteria. We and Ramakrishnan et al (5) included studies with children < 5 years, while Brown et al. (3) also included children until pre-pubertal age. Both Brown et al. (3) and we excluded studies where zinc was supplemented in fortified foods, while Ramakrishanan et al. (5) included these as well. Brown et al. (3) also excluded those studies where SD for change in length was not reported in the paper; however we and Ramakrishnan included those (5). Do these differences explain the reasons for contradictory results? If we include zinc fortified studies with our analysis (all countries), the estimated effect size becomes 0.15 (95 % CI 0.06, 0.20), which indicates that excluding zinc fortification studies did not significantly alter the results and direction of effect. Excluding children > 5 years also did not change the results significantly as shown above. We could not find any further explanation for the difference in results of our analysis and these reviews.
Presentation of results in terms of absolute difference in the growth increment is described in previous reviews. In the previous review by Brown et al. in 2002 (4) data were analyzed in two ways i.e. pooled weighted mean difference (results discussed above) and actual increase in length (cm) . In the second analysis, they pooled results from 25 studies to get a net increment in length of 0.73 (±0.98) cm in zinc supplemented group compared to controls. These results were not however described for any particular dose or duration and the meta-views were also not provided to get an idea of the contribution of each study. On the other hand we have described pooled results for all the studies that reported absolute increment in length and also did a subgroup analysis for the most effective dose for a particular duration. This leads to an estimated effect size of 0.37 (±0.25) cm increase in linear growth in children <5 years, with a specific dose of 10 mg zinc/day for a duration of 24 weeks (Fig 3). The p-value for this estimate was 0.005 and quality grade for the pooled evidence was that of ‘moderate’ level. The most prominent contribution to this estimate comes from study by Umeta et al. in stunted children . This is understandable, as zinc seems to have more prominent effect on growth in stunted compared to non-stunted children . If results of this data set are omitted from analysis, the estimate becomes 0.36 (± 0.26) cm. In any case, keeping in mind the statistical significance, quality grade and specificity of estimate, this seems to be the most suitable input to LiST model, for an effect size estimate of zinc supplementation in prevention of stunting in children < 5 years of age in developing countries.
What does an extra gain of 0.37 cm means clinically? An average gain of this much in height would not be a huge effect but we need to take into account that that a single micronutrient would not be expected to result in such a substantial benefit at the first place. The results of this review confirms that preventive zinc supplementation indeed has an effect in promotion of growth of young children but this has to be connected with more comprehensive approaches that improve the diets of small children in general to get a more substantial effect. These efforts should especially focus on first two years of life and there should be a special attention to promote exclusive breastfeeding and practices of complementary feeding in addition to correcting micronutrient deficiencies. An analysis published in lancet undernutrition series in 2008 showed that education and counseling of caretakers in food-secure populations can improve growth in height (WMD 0.25; 95% CI: 0.01, 0.49) and providing complementary food, with or without education and counseling, can improve height in food insecure populations (WMD 0.41; 95% CI: 0.05, 0.76) . In an updated analysis for this series, we have shown that provision of complementary food (±nutrition counseling) lead to an extra gain of 0.54 cm (±0.38) and education of mothers about complementary feeding can lead to an extra gain of 0.49 cm (±0.50) in the intervention group compared to control . Thus, in order to get full advantage of relatively small benefit of correcting zinc deficiency, we must, at the same time, focus on interventions to improve complementary feeding and general nutritional status if the child. Future research should focus on strategies where correction of micronutrient deficiencies should be a part of a more general approach to improve the nutritional status of the child in general.
Do zinc and iron interact significantly when supplemented simultaneously? It has been demonstrated from several studies that iron and zinc have similar absorption and transport mechanisms . Experimental studies have also shown that simultaneous supplementation of iron and zinc may inhibit zinc absorption in these cells especially at high ratios of iron to zinc [14, 15, 73]. A review by Walker et al. (64) on interaction of zinc and iron in supplementation trials showed that combined supplementation of zinc and iron did not affect the biochemical status of zinc; however the data were not clear regarding morbidity and growth outcomes. In our review, subgroup analysis excluding studies in which zinc was co-administered with iron showed an increase in overall effect size confirming the likelihood of interaction (Fig 2). When we separately pooled these data sets (zinc + iron), the pooled effect size showed a negative trend -0.10 [95 % CI -0.21, 0.01]. These results were significantly different from overall estimate (p=0.0007). These analyses strongly suggest that addition of iron decreases the positive effect of zinc supplementation on linear growth through potential interference with absorption or bio-availability. We have, therefore, presently restricted our recommendation for zinc supplementation to studies with zinc supplementation alone.
We used rigorous eligibility criteria for inclusion of studies. For example, the minimum period of supplementation should be ≥8 wk because shorter periods may be insufficient for detecting a linear growth response. Furthermore, we excluded studies of premature infants and those suffering from chronic disease or severe protein-energy malnutrition as zinc requirements of these children might differ considerably from those of unaffected children [74, 75]. We also excluded studies in which zinc was supplemented in fortified food. Although tracer-element experimental studies had shown that zinc fortification of food increase total zinc absorption [76–78], relatively few community studies have found positive impacts of zinc fortification on serum zinc concentrations or functional indicators of zinc status . There is also insufficient evidence for ideal food vehicle for zinc fortification and also interaction of zinc with other micronutrient when fortified in a single food .
Preventive zinc supplementation seems to be a safe intervention. It has been suggested that high levels of zinc intake may interfere with normal iron and copper metabolism . Although we did not specifically look at adverse effects, results from previous reviews showed that preventive zinc supplementation in physiological doses do not significantly affect the indicators of iron (i.e. hemoglobin and/or ferritin level) and/or copper metabolism .
Our review has certain limitations. As our recommendations are limited to zinc supplementation alone vs. control/placebo, these results may not be readily applicable to countries where there are on-going national supplementation programs of iron-folate, for example, India. We expect that policy makers will assess local contexts and conditions while considering the feasibility of zinc supplementation programs. We were unable to identify any significant predictor of substantial heterogeneity in the pooled data. There may be factors, not examined by us, that might explain the observed differences. These include initial HAZ score (prevalence of stunting), mean initial age, baseline zinc deficiency, gender and HIV prevalence [4, 79–81]. Although there are relatively large number of studies and a funnel plot for zinc alone supplementation was relatively symmetrical (data not shown), there may be possible publication bias.
In conclusion, our review suggests that zinc supplementation has a positive effect on linear growth, especially when supplemented alone. Zinc supplementation in a dose of 10 mg/day for duration of 24 weeks led to an increase gain in length by 0.37 (± 0.25) cm among children < 5 years of age in developing countries compared to controls.