Weakening association of parental education: analysis of child health outcomes in 43 low-and middle-income countries

Background: Parental education has been suggested to be an effective instrument for improving child health in low- and middle-income countries. Both education and child health have improved, however, as well as related factors. These changes may have implications for the observed association. Methods: We used Demographic and Health Surveys conducted in 43 countries at two points in time, between 1991 and 2016, to test if the association of parental education with child health has changed over time. We explored how changes relate to commonly cited confounders and pathways, including fertility, household living standards, health care use, urbanicity and geographical clustering. We used linear probability models, Gelbach decomposition, and assessed a range of sensitivity specifications. Results: The point estimate for an additional year of maternal education has attenuated by 0.27% points (56%) for under-5 mortality, 0.34% points (15%) for child stunting, 0.42% points (30%) for child underweight and 0.09% points (24%) for child wasting. The point estimate for paternal education has attenuated by 0.20% points (53%) for under-5 mortality, 0.15% points (8%) for child stunting, 0.28% points (24%) for underweight and 0.06% points (19%) for wasting. Changes in confounding and mediation by fertility, household living standards and urban-rural differences explain to a large extent the attenuations. Geospatial clustering increasingly drives the association of parental education with child health. Conclusions: The role of parental education in child health has attenuated considerably over time in low-resource settings. Decision makers should take into account this weakening association when designing policies aimed at improving child health.