Fig 1.
Majete Malaria Project focal areas.
Map was made using ArcGIS Pro 2.7.0 (https://www.esri.com/en-us/arcgis/products/arcgis-pro/resources). Source settlements data: High Resolution Settlement Layer—Facebook Connectivity Lab and Center for International Earth Science Information Network—CIESIN—Columbia University.
Table 1.
Environmental and epidemiological characteristics of focal areas for the period under study (1st July 2016 to 12th May 2018).
Table 2.
MALSWOTS estimated optimal parameters and transmission summaries obtained from the top 5% of the models ranked by largest correlation.
^ confidence intervals. All parameters are calculated over the whole study period.
Fig 2.
Spreaders for each focal area obtained by MALSWOTS for the whole study period.
Each plot shows the number of new houses (from 1 to 15 houses in bins of 1) predicted to have been infected by x original houses. For example, 14 houses are likely to have infected 2 other houses each in Focal area A.
Fig 3.
Location of super-spreader houses and hotspots (stable and unstable) within each focal area.
Each colour denotes the number of ‘infected houses’ arising from the spreader. A black cross denotes the location of an uninfected house. Allocation of the hotspots has been performed manually capturing the largest number of super-spreaders within each hotspot square.
Fig 4.
MALSWOTS time and distance analyses between house connections within each focal area.
The ‘average distance connections’ refers to successful connections only. The ‘finding houses’ measure (central histogram) is the ratio between the length of the connection between two houses in which a mosquito has flown and the straight line (Euclidean) distance between the same houses (which can be shorter than the connection distance due to the zig-zag movement of the mosquito). The ‘time connections’ shows the temporal distance between infecting and infected houses in terms of RDT positive test. This time can be shorter than the days of flight (Table 2) due to the possibility that the mosquito left the infecting house prior malaria detection in the house.
Table 3.
Odds ratios (OR) associated with selected variables from generalised linear mixed model for risk of being a super-spreader house.
*If yes, the village has the same interventions as the control village but it is excluded from the trial analysis because of proximity to other villages.
Fig 5.
Successful mosquito connections (left panels) are shown as arrows departing from an infecting house (red circle) and arriving to an uninfected (and subsequently infected) house (blue circle).
Unexplained (black circle) are houses that are not analysed because they are outside the spatial and temporal scales considered by the best models. The wind rose panels on the right show the frequency of wind directions and angular sector. The line colours identify the average speed of the winds. Calm refers to winds below 0.5 m/s.
Fig 6.
Relative risk of getting infected per day (left) and risk exceedance (right) maps for each focal area.
Table 4.
Cross validation results for infected houses.
Table 5.
Cross validation results for uninfected houses.