Evidence for causal links between education and maternal and child health: systematic review

OBJECTIVE: Numerous studies have documented an inverse association between years of schooling attained, particularly by women, and reduced maternal, infant and child mortality. However, if factors affecting educational attainment — many of which are unobservable, e.g. motivation and genetic endowment — also affect the likelihood of engaging in behaviours that enhance health, then assumed effects of schooling will be inflated in analyses that do not address this endogeneity. This systematic review assesses evidence for a causal link between education and maternal and child health in low and middle‐income countries. METHODS: Eligible studies controlled for observable and unobservable factors affecting both education and health. Reported effects were converted into partial correlations. When possible, we also conducted meta‐analyses to estimate mean effects by outcome. RESULTS: Of 4952 papers identified, 16 met the inclusion criteria. The 15 child health papers examined neonatal, infant and child mortality, stunting and wasting. Significant effects of education on infant and child health were observed for 30 of 33 models that did not account for endogeneity. In contrast, only 18 of 46 effects were significant in models that addressed endogeneity. Notably, for only one outcome —child mortality measured dichotomously —was the effect of maternal educational attainment significant in a meta‐analysis. The one maternal morbidity paper found significant effects for the two preventable outcomes considered. CONCLUSION: While we find evidence for a causal link between education and health, effects are weaker in models that address endogeneity compared to naïve models that do not account for unobservable factors affecting both education and health. Advances in women's educational outcomes have undoubtedly played a role in improving health in many settings; however, the effect is not as strong as some researchers and advocates have claimed.