Association of Early Interventions With Birth Outcomes and Child Linear Growth in Low-Income and Middle-Income Countries

Systematic review and Bayesian network meta-analyses of 169 trials in LMICs, with 302,061 participants randomized to 461 arms. The evidence base included pregnant women, mother-infant dyads, and children during pregnancy, exclusive breastfeeding, and complementary feeding periods.

This Bayesian network meta-analysis evaluated early-life nutrition and related interventions in LMICs, including multiple micronutrients, iron, folic acid, zinc, calcium, lipid-based nutrient supplements, fortified flour and local foods, formula, deworming, WASH, and maternal education. Regimens varied widely, from LNS 118 kcal (20 g) to 746 kcal, flour 270-340 kcal, fortified flour 750 kcal, local food 185-260 kcal, fortified local food 597-850 kcal, formula 335 kcal, and deworming as a 1-dose intervention.

Overall, maternal and child nutritional interventions were associated with improved birth outcomes and linear growth, with pregnancy-based interventions generally showing stronger effects than interventions given later in life. Multiple micronutrients in pregnancy reduced preterm birth (OR 0.54, 0.27-0.97) and were associated with better child linear growth, including LAZ 0.20 (0.03-0.35) and HAZ 0.14 (0.02-0.25). Other pregnancy supplements also showed favorable associations, such as zinc for preterm birth (OR 0.53, 0.28-0.93), iron for preterm birth (OR 0.55, 0.31-0.90), and iron plus calcium for preterm birth (OR 0.16, 0.03-0.87). Balanced energy-protein or food supplements generally improved birth weight, for example LNS 118 kcal increased birth weight by 0.11 (0.02-0.21), while deworming, WASH, and maternal education were less consistently beneficial.

The evidence combined many different interventions, doses, delivery periods, and outcomes, which increases heterogeneity and makes direct comparisons difficult. Many estimates were imprecise or null, especially for non-nutritional interventions and later-life outcomes, and the network meta-analysis relied on indirect as well as direct comparisons. Generalizability may also be limited by variation across LMIC settings and trial designs.