Associations between landscape fires and child morbidity in southern Mozambique: a time-series study

Background Epidemiological evidence linking exposure to landscape fires to child health remains scarce. We assessed the association between daily landscape fire smoke and child hospital visits and admissions in the Manhi & ccedil;a district, Mozambique, an area characterised by frequent forest and cropland fires. Methods In this time-series analysis (2012-20), our primary metric for exposure to landscape fires was fire-originated PM25 from smoke dispersion hindcasts. We also assessed total and upwind fire exposure using daily satellite-derived fire density data. Daily numbers of hospital visits and admissions were extracted from an ongoing paediatric morbidity surveillance system (children aged <= 15 years). We applied quasi-Poisson regression models controlling for season, long-term trend, day of the week, temperature, and rainfall, and offsetting by annual population-time at risk to examine lag-specific association of fires on morbidity. Findings A 10 mu g/m(3) increase in fire-originated PM25 was associated with a 612% (95% CI 037-1221) increase in all-cause and a 1243% (507-2031) increase in respiratory-linked hospital visits on the following day. Positive associations were also observed for lag 0 and the cumulative lag of 0-1 days. Null associations were observed for hospital admissions. Landscape fires mostly occurred in forested areas; however, associations with child morbidity were stronger for cropland than for forest fires.