Quantifying the impact of human mobility on malaria

Abstract

Human movements contribute to the transmission of malaria on spatial scales that exceed the limits of mosquito dispersal. Identifying the sources and sinks of imported infections due to human travel and locating high-risk sites of parasite importation could greatly improve malaria control programs. Here, we use spatially explicit mobile phone data and malaria prevalence information from Kenya to identify the dynamics of human carriers that drive parasite importation between regions. Our analysis identifies importation routes that contribute to malaria epidemiology on regional spatial scales.

Figure 1. The distribution of settlements, cell towers, and malaria risk in Kenya

A) Malaria prevalence in Kenya in 2009 (from PfPR_2–10<0.1% in yellow to PfPR_2–10>40% in red) and the locations of settlements used in the analysis (settlement centers are shown in black, and mapped with a 10km extent around the perimeter of the settlement in grey). B) Mobile phone towers (black or blue dots) are overlaid over the extended settlement boundaries. Towers that fall within a settlement are shown in black and those excluded from the analysis are shown in blue. C) Regions used for visual mapping of transmission routes. Each settlement was allocated to one of twenty regions by a clustering algorithm (see Methods in the SI) based on homogenous malaria risk and geography, as shown. Regions near Lake Victoria (LV), in Nairobi (Nairobi), the central areas (Cen), and along the coast (C) are labeled accordingly.

Figure 2. Travel networks of people and parasites between settlements and regions

A) Average monthly travel between regions (nodes), with edges weighted by volume of traffic. For clarity, the top 50% of routes are shown with arrows indicating the direction of movement (humans or parasites) from a primary settlement to a visited settlement. B) Average monthly parasite importation by returning residents, by region. C) Average monthly parasite importation by visitors, where importation is not considered if the destination has extremely low transmission (see SI). The labeling of nodes is as described in Figure 1. For each network, node layout is roughly based on geographic location.

Figure 3. Sources and sinks of human and parasite dispersal

Kernel density maps showing ranked sources and sinks of human travel and total parasite movement in Kenya. A) Settlements were designated as primarily sources or sinks based on yearly estimates of human travel. The direction movement is shown highlighting principal sources (in red) versus principal sinks (in blue). B) The directional parasite movements, major parasite sources (in red) and sinks (in blue) are shown.

Figure 4. Local analysis of source/sink anomalies

A) Source outliers and B) sink outliers. Settlements are colored by their outlier rank (from low values in blue to high values in red) and sized according to R_can indicator of receptivity (see Methods in SI). C) The localized importation into Nairobi compared to clinical cases. A topographic map of Nairobi was provided from National Geographic and ESRI highlighting the national park, commercial, and residential areas of the city.