Cooking location and child health outcomes in Rwanda: a pooled analysis of two demographic and health surveys

Household air pollution from solid-fuel cooking is a leading environmental threat to children in sub-Saharan Africa. Where almost every household burns biomass, fuel type offers little contrast. This study analysed associations between household cooking location and selected health outcomes among children under five in Rwanda. Two independent, nationally representative Rwanda Demographic and Health Surveys, 2014–15 ( n = 7,552 children under five) and 2019–20 ( n = 7,796), were pooled. Self-reported cooking location was categorized as inside the house (reference), in a separate building, or outdoors. Outcomes were acute respiratory infection (ARI, DHS definition), fever, diarrhoea, stunting, anaemia and, as an exploratory outcome, low birth weight. Survey-weighted logistic regression with design-based (Taylor-series linearisation) variance adjusted for child age and sex, residence, maternal education, wealth, province, household size and survey year. The three-category exposure defined the primary analysis, with Benjamini–Hochberg correction across its twelve contrasts; a combined away-from-living-space indicator was secondary. Sensitivity analyses added water and sanitation, housing quality and household indoor smoking in a common complete-case sample and tested heterogeneity by residence and survey round. Use of solid biomass fuel for cooking was near-universal (99.8% [7,531/7,552] and 96.9% [7,555/7,796]). Compared with cooking inside the house, cooking in a separate building was associated with lower odds of diarrhoea (pooled adjusted odds ratio [aOR] 0.73, 95% CI 0.64–0.84) and stunting (0.76, 0.66–0.87); both remained statistically significant after Benjamini–Hochberg adjustment. For outdoor cooking, point estimates were below 1.0 for diarrhoea (0.84, 0.71–1.00) and stunting (0.87, 0.74–1.03), but confidence intervals included the null. With full adjustment in the same complete-case sample, these estimates were 0.76 (0.66–0.87) for diarrhoea and 0.80 (0.69–0.92) for stunting. The stunting association differed by residence (joint test p = 0.002): the outdoor estimate was 0.81 (0.68–0.98) among rural and 1.06 (0.68–1.65) among urban children. ARI, fever, anaemia and low birth weight were not associated with cooking location in the primary analysis. Cooking in a separate building was associated with lower odds of diarrhoea and stunting among children under five in Rwanda. Because the study was cross-sectional and measured no pollutant exposure, these associations need to be tested in prospective and interventional studies.