Altitude-associated gradients in child health in rwanda and modelled scenarios for under-five mortality: An analysis of two demographic and health surveys

Rwanda’s steep altitude gradient produces large spatial variation in temperature and rainfall over short distances. This analysis describes altitude-associated gradients in child health in two national surveys conducted before and after a large change in malaria, and estimates the share of under-five mortality attributable to household poverty, short birth intervals and the cluster malaria environment. The 2014–15 and 2019–20 Rwanda Demographic and Health Surveys were analysed (492 and 500 clusters; 7,558 and 7,796 children under five). Cluster altitude, available in both rounds, was analysed as an altitude-associated gradient, together with modelled temperature and precipitation for both rounds. Outcomes were malaria (rapid diagnostic test), diarrhoea, acute respiratory infection, fever, low birthweight, and neonatal, infant and under-five mortality. Weighted generalized estimating equations gave adjusted odds ratios (aOR) per SD, and round×altitude interactions tested differences between surveys. Mortality scenarios used model-based standardisation. Malaria positivity fell from 7.9% to 2.7%. In 2014–15 malaria was higher in warmer clusters (aOR 2.67 per SD, 95% CI 2.22–3.21) and lower at higher altitude (0.30, 0.23–0.39), and fever, low birthweight and infant and under-five mortality were also higher in warmer clusters. The malaria–altitude odds ratio was similar in 2019–20 (0.38, 0.26–0.55; interaction p = 0.36), but the absolute gradient fell from 8.1 to 2.4% points per SD, and the fever–altitude association was no longer seen ( p < 0.001). A birth interval under 24 months (aOR 1.91, 1.52–2.39) and household poverty (1.34, 1.09–1.64) were associated with under-five mortality. A cluster malaria indicator from the survey’s own tests was not (1.10, 0.89–1.36), but a time-matched modelled malaria prevalence surface was (1.54, 1.26–1.88). Modelled scenarios lowered under-five mortality by 12% (8–17) for birth spacing, 13% (4–22) for poverty, and 3% or 22% for the malaria environment, depending on the measure. Child health in Rwanda follows a marked altitude-associated gradient, with malaria concentrated in the warm eastern and southern lowlands. The two-survey comparison is descriptive. Short birth intervals and poverty were consistently associated with under-five mortality, and the contribution of the malaria environment cannot be ruled out. The warm lowlands remain a priority for integrated malaria, newborn and child-survival programming.