Explore the vast range of titles available.

Background Because clustering of Plasmodium falciparum infection had been noted previously, the clustering of infection was examined at four field sites in West Africa: Dangassa and Dioro in Mali, Gambissara in The Gambia and Madina Fall in Senegal. Methods Clustering of infection was defined by the percent of persons with positive slides for asexual P. falciparum sleeping in a house which had been geopositioned. Data from each site were then tested for spatial, temporal and spatio-temporal clustering in relation to the prevalence of infection from smear surveys. Results These studies suggest that clustering of P. falciparum infection also affects the effectiveness of control interventions. For example, the clustering of infection in Madina Fall disappeared in 2014–2016 after vector control eliminated the only breeding site in 2013. In contrast, the temporal clustering of infection in Dioro (rainy season of 2014, dry season of 2015) was consistent with the loss of funding for Dioro in the second quarter of 2014 and disappeared when funds again became available in late 2015. The clustering of infection in rural (western) areas of Gambissara was consistent with known rural–urban differences in the prevalence of infection and with the thatched roofs, open eaves and mud walls of houses in rural Gambissara. In contrast, the most intense transmission was in Dangassa, where the only encouraging observation was a lower prevalence of infection in the dry season. Taken together, these results suggest: (a) the transmission of infection was stopped in Madina Fall by eliminating the only known breeding site, (b) the prevalence of infection was reduced in Dioro after financial support became available again for malaria control in the second half of 2015, (c) improvements in housing should improve malaria control by reducing the number of vectors in rural communities such as western Gambissara, and (d) beginning malaria control during the dry season may reduce transmission in hyperendemic areas such as Dangassa. Conclusions From a conceptual perspective, testing for spatial, temporal and spatio-temporal clustering based on epidemiologic data permits the generation of hypotheses for the clustering observed and the testing of candidate interventions to confirm or refute those hypotheses.