Measuring the environmental context of child growth in Burkina Faso

Child growth failure, as indicated by low height-for-age z-scores (HAZ), is an important metric of health, social inequality, and food insecurity. Understanding the environmental pathways to this outcome can provide insight into how to prevent it. While other studies have examined the environmental determinants of HAZ, there is no agreed upon best practices approach to measure the environmental context of this outcome. From this literature, we derive a large set of potential environmental predictors and specifications including temperature and precipitation levels, anomalies, and counts as well as vegetation anomalies and trends, which we include using linear, nonlinear, and interactive specifications. We compare these measures and specifications using four rounds of DHS survey data from Burkina Faso and a large set of fixed effects regression models. We focus on exposures from the time of conception through the second year of life and rely on joint hypothesis tests and goodness-of-fit measures to determine which approach best explains HAZ. Our analysis reveals that nonlinear and interactive transformations of climate anomalies, as opposed to climate levels or vegetation indices, provide the best explanation of child growth failure. These results underline the complex and nonlinear pathways through which climate change affects child health and should motivate climate-health researchers to more broadly adopt measures and specifications that capture these pathways.