Long-term drought and risk of infant mortality in Africa: A cross-sectional study

Abstract

Background: As extreme events such as drought and flood are projected to increase in frequency and intensity under climate change, there is still large missing evidence on how drought exposure potentially impacts mortality among young children. This study aimed to investigate the association between drought and risk of infant mortality in Africa, a region highly vulnerable to climate change that bears the heaviest share of the global burden.

Methods and findings: In this cross-sectional study, we obtained data on infant mortality in 34 African countries during 1992-2019 from the Demographic and Health Surveys program. We measured drought by the standardized precipitation evapotranspiration index at a timescale of 24 months and a spatial resolution of 10 × 10 km, which was further dichotomized into mild and severe drought. The association between drought exposure and infant mortality risk was estimated using Cox regression models allowing time-dependent covariates. We further examined whether the association varied for neonatal and post-neonatal mortality and whether there was a delayed association with drought exposure during pregnancy or infancy. The mean (standard deviation) number of months in which children experienced any drought during pregnancy and survival period (from birth through death before 1 year of age) was 4.6 (5.2) and 7.3 (7.4) among cases and non-cases, respectively. Compared to children who did not experience drought, we did not find evidence that any drought exposure was associated with an increased risk of infant mortality (hazard ratio [HR]: 1.02, 95% confidence interval [CI] [1.00, 1.04], p = 0.072). When stratified by drought severity, we found a statistically significant association with severe drought (HR: 1.04; 95% CI [1.01, 1.07], p = 0.015), but no significant association with mild drought (HR: 1.01; 95% CI [0.99, 1.03], p = 0.353), compared to non-exposure to any drought. However, when excluding drought exposure during pregnancy, the association with severe drought was found to be non-significant. In addition, an increased risk of neonatal mortality was associated with severe drought (HR: 1.05; 95% CI [1.01, 1.10], p = 0.019), but not with mild drought (HR: 0.99; 95% CI [0.96, 1.02], p = 0.657).

Conclusions: Exposure to long-term severe drought was associated with increased infant mortality risk in Africa. Our findings urge more effective adaptation measures and alleviation strategies against the adverse impact of drought on child health.

Fig 1. Infant mortality rate (deaths per 1,000 live births) (A), number of drought months (B), and country-specific distributions of the number of drought months across all 10-km grids, sorted by area-weighted mean value (diamonds) and labeled with number of survey clusters (C) in Africa during 1992–2019.

The base layer of the map was obtained from ArcGIS Hub (https://hub.arcgis.com/datasets/07610d73964e4d39ab62c4245d548625/).

Fig 2. Error bar charts for the associations between risk of infant mortality and exposure to long-term drought represented by 24-month standardized precipitation and evapotranspiration index by drought severity (any, mild, or severe) and type of infant mortality (neonatal or post-neonatal mortality).

Reference group is the children who did not experience droughts. Covariates in the models included child’s sex, area of residence, mother’s education, and wealth quintile, categorical birth month, a natural cubic spline of birth year with three degrees of freedom, and a random intercept for a composite indicator for country and survey cluster.

Fig 3. Associations between risk of infant mortality and exposure to long-term drought stratified by baseline characteristics, year period of birth, and climate zone.

Statistically significant pairwise differences (p < 0.05) are marked with an asterisk.

Fig 4. Associations between infant deaths during each month within 1 year of age and long-term drought.

The exposure was measured by a binary drought indicator during each month before and after birth (the ninth and first month of drought exposure before birth represents the first and last month of pregnancy, respectively) (A) and by the number of drought months experienced before and after birth (from the first through the last month of pregnancy for prenatal exposure and from the month of birth through the relevant month examined for postnatal exposure) (B). Statistically significant results (p < 0.05) are marked with an asterisk.