Mapping the local effectiveness of mass drug administration for malaria using transportability methods

Abstract Numerous trials have evaluated the effectiveness of mass drug administration (MDA) in rapidly reducing malaria transmission, but it is unknown whether the estimated effects generalize to other populations eligible for MDA. A recent cluster randomized trial in Senegal found that MDA reduced malaria incidence by 55% in areas that routinely deploy seasonal malaria chemoprevention (SMC). Here, we used transportability models with machine learning to generalize trial effects to 116 non-trial communes where SMC is the standard of care. Accounting for differences in weather, vegetation and population density between trial and non-trial areas, we estimated considerable reductions in incidence (ranging from 36% to 65%) in 74 non-trial communes, with larger decreases in areas having higher precipitation, denser vegetation and lower temperatures. We found that MDA was not effective in the postintervention year in non-trial communes, supporting the notion that MDA’s effects are short-lived. Our approach offers a scalable framework for generalizing trial findings to target environmentally mediated infectious disease interventions.