Micronutrient powders and diarrhoea risk in infants and young children

Emily Keats and colleagues¹ revisited the evidence base of maternal and child nutrition interventions using updated systematic reviews. The authors recommend micronutrient powders as an effective intervention to reduce childhood anaemia and iron deficiency, in line with WHO recommendations.² Using data from a 2020 systematic review by Tam and colleagues,³ the authors reported a 30% increased risk of diarrhoea (relative risk [RR] 1·30 [95% CI 1·11–1·53) associated with micronutrient powders. This finding contrasts with findings from two Cochrane reviews of micronutrient powders that found no association between micronutrient powder use and diarrhoea in children aged 6–23 months (odds ratio [OR] 1·05 [95% CI 0·82–1·35]) and 2–12 years (RR 0·97 [95% CI 0·53–1·78]).^4,5 We raise several points to explain these differences.

First, there were differences in systematic review methodology. Tam and colleagues included children aged 1 month to 5 years in low-and-middle income countries,³ whereas the two Cochrane reviews on micronutrient powder use included children aged 6–23 months⁴ and 2–12 years⁵ in any geographical setting. The reviews included and excluded different trials. All reviews assessed the longitudinal prevalence of diarrhoea as a secondary outcome. Tam and colleagues included four trials in their diarrhoea sub-analysis. One trial used ready-to-eat fortified maize flour, not micronutrient powders, and was thus not included in the Cochrane reviews.⁶ This trial and two others that Tam and colleagues included found no differences in diarrhoea by treatment group; only one trial by Soofi and colleagues⁷ (overall weight of 92·9%) showed an increase in the proportion of days with diarrhoea. Suchdev and colleagues included five trials in their diarrhoea subanalysis, with only one trial by Soofi and colleagues in common with the Tam review. A total of four trials selected for inclusion in the review by Tam and colleagues and two trials selected for inclusion in the review by Suchdev and colleagues did not have their diarrhoea results extracted. De-Regil and colleagues included two trials, only one of which included children aged 6 months to 5 years,⁸ but this trial was not included by Tam and colleagues.

Statistical analysis procedures also differed. Tam and colleagues and De-Regil and colleagues used a Mantel-Haenszel random effects model and reported risk ratios based on diarrhoea prevalence; Suchdev and colleagues used generic inverse-variance random effects models because two trials presented pre-calculated estimates of treatment effects, and reported ORs given the overall low rates of diarrhoea.

To produce an estimate of risk of diarrhoea from micronutrient powders in children aged 6 months to 5 years using more comprehensive evidence, excluding the non-micronutrient powder trial,⁶ we did an analysis of eight trials included in these three systematic reviews in duplicate in RevMan (version 5.4) and R (version 4.0.2), including 10 335 children aged 6 months to 5 years. We found no effect of micronutrient powder use on diarrhoea prevalence (OR 0·95 [95% CI 0·72–1·25]; appendix). UNICEF reports 16 million children in 46 countries were reached with micronutrient powders in 2019.⁹ Although additional research on the safety and cost-effectiveness of iron interventions in children might be needed, identifying diarrhoea as a significant side-effect associated with micronutrient powder use, when additional evidence suggests that it might not be, could set back global goals to improve child nutrition in high burden settings.