Suppression of malaria vector densities and human infection prevalence associated with scale-up of mosquito-proofed housing in Dar es Salaam, Tanzania: re-analysis of an observational series of parasitological and entomological surveys

Summary: Background: In the city of Dar es Salaam, Tanzania, rapid and spontaneous scale-up of window screening occurred through purely horizontal commercial distribution systems without any public subsidies or promotion. Scale-up of window screening coincided with a planned evaluation of programmatic, vertically managed scale-up of regular larvicide application as an intervention against malaria vectors and transmission. We aimed to establish whether scale-up of window screening was associated with suppression of mosquito populations, especially for malaria vectors that strongly prefer humans as their source of blood. Methods: This study was a re-analysis of a previous observational series of epidemiological data plus new analyses of previously partly reported complementary entomological data, from Dar es Salaam. Between 2004 and 2008, six rounds of cluster-sampled, rolling, cross-sectional parasitological and questionnaire surveys were done in urban Dar es Salaam to assess the effect of larviciding and other determinants of malaria risk, such as use of bed nets and antimalarial drugs, socioeconomic status, age, sex, travel history, mosquito-proofed housing, and spending time outdoors. The effects of scaled-up larvicide application and window screening were estimated by fitting generalised linear mixed models that allowed for both spatial variation between survey locations and temporal autocorrelation within locations. We also conducted continuous longitudinal entomological surveys of outdoor human biting rates by mosquitoes and experimental measurements of mosquito host preferences. Findings: Best-fit models of Plasmodium falciparum malaria infection prevalence among humans were largely consistent with the results of the previous analyses. Re-analysis of previously reported epidemiological data revealed that most of the empirically fitted downward time trend in P falciparum malaria prevalence over the course of the study (odds ratio [OR] 0·04; 95% CI 0·03–0·06; p<0·0001), which was not previously reported numerically or attributed to any explanatory factor, could be plausibly explained by association with an upward trend in city-wide window screening coverage (OR 0·07; 0·05–0·09; p<0·0001) and progressive rollout of larviciding (OR 0·50; 0·41–0·60; p<0·0001). Increasing coverage of complete window screening was also associated with reduced biting densities of all taxonomic groups of mosquitoes (all p<0·0001), especially the Anopheles gambiae complex (relative rate [RR] 0·23; 95% CI 0·16–0·33) and Anopheles funestus group (RR 0·08; 0·04–0·16), which were confirmed as the most efficient vectors of malaria with strong preferences for humans over cattle. Larviciding was also associated with reduced biting densities of all mosquito taxa (p