Fig. 4

Summary results for the four evaluation criteria in unadjusted data and all adjusted malaria datasets. Each dot represents the metric of interest in one set of adjusted data, and colors represent the pooling strategy (e.g. 2 nn = pooling of Fokontany with its 2 nearest neighbors; 3 km = pooling with neighbors within 3 km). The dashed line shows values for the unadjusted dataset. Shaded green areas show target ranges of each evaluation criteria as described in Table 1. a Overall malaria incidence: ratio of malaria in adjusted dataset to malaria in optimal access areas. Values closer to 1 mean better performance. b Distance decay: ratio of incidence in Fokontany less than 5 km from a health center to incidence in Fokontany more than 5 km from a health center. Values closer to 1 mean better performance. c Trends in financial access to care: ratio of average monthly incidence in fee-exempt to non-fee-exempt populations in each adjusted dataset. Values closer to 1 mean better performance. d Geographic consistency with IHOPE cohort data: percent of overlap between hotspots of fever identified in the IHOPE cohort study data and malaria incidence in each adjusted dataset. Greater values mean better performance. Equivalent plots including analyses for children under 5 years can be found Additional file 1: Figure S2