LAUSR

Background Governments have endorsed global targets to reduce childhood undernutrition as part of the Sustainable Development Goals. Understanding the socioeconomic differences in childhood undernutrition has the potential to be helpful for targeting policy to reach these goals. Methods We specify a logistic regression model with the Composite Index of Anthropometric Failure (CIAF) as the outcome and indicator variables for wealth quartiles, maternal education categories and a set of covariates as explanatory variables. Wealth and education variables are interacted with a period indicator for 1990–2000 compared with 2001–2014 to observe differences over time. Based on these regressions we calculate predicted CIAF prevalence by wealth and education categories and over time. Results The sample included 146 surveys from 39 low-income and lower-middle-income countries with an overall sample size of 533 217 children. CIAF prevalence was 47.5% in 1990–2000, and it declined to 42.6% in 2001–2014. In 1990–2000 the CIAF prevalence of children with mothers with less than primary education was 31 percentage points higher than for mothers with secondary or higher education. This difference slightly decreased to 27 percentage points in 2001–2014. The difference in predicted CIAF prevalence of children from the highest and lowest wealth quartiles was 21 percentage points and did not change over time. Conclusions We find evidence for persistent and even increasing socioeconomic inequalities in childhood undernutrition, which underlines the importance of previous calls for equity-driven approaches targeting the most vulnerable to reduce childhood malnutrition.