Lagged effect of temperature and rainfall on malaria incidence in Colombia (2013–2023): An approach with Bayesian spatiotemporal adjustment

Malaria remains a major public health challenge in Colombia, with a significant increase in cases in recent years. Climate variables-particularly temperature and rainfall-are key drivers of malaria transmission, yet their lagged, non-linear effects across space and time are poorly characterized in the Colombian context. We conducted an ecological, spatiotemporal analysis using weekly malaria case data from 970 municipalities in Colombia (2013-2023) combined with satellite-derived climate data. We applied distributed lag non-linear models (DLNMs) embedded within a Bayesian hierarchical framework using integrated nested Laplace approximation (INLA) to estimate the delayed, non-linear associations between weekly temperature and rainfall and malaria incidence, while accounting for spatial and temporal autocorrelation, forest cover, multidimensional poverty, altitude, population size, and prior case counts. Our results show that malaria risk increases non-linearly with temperature, peaking around 28 °C, with a global exposure of minimum risk (EMR) at 16.43 °C, and significant effects observed at lags of 0-6 weeks. In contrast, lower weekly rainfall was associated with higher malaria risk, with an EMR at 0.85 mm. Sensitivity analyses confirmed the robustness of these findings. These results challenge previous studies about climate-driven malaria risk and highlight accelerated transmission dynamics in Colombia's endemic zones. The identification of specific climate thresholds linked to elevated malaria incidence provides actionable evidence for climate-informed early warning systems and targeted interventions to support malaria elimination efforts in Colombia.