An exploration of the protective effect of rodent species richness on the geographical expansion of Lassa fever in West Africa

Abstract

Background: Lassa fever (LF) is one of the most devastating rodent-borne diseases in West Africa, causing thousands of deaths annually. The geographical expansion of LF is also a concern; cases were recently identified in Ghana and Benin. Previous ecological studies have suggested that high natural-host biodiversity reduces the likelihood of spillover transmission of rodent-borne diseases, by suppressing the activities of reservoir species. However, the association of biodiversity with the geographical expansion of LF has not been the subject of epidemiological studies.

Methodology/principal findings: We conducted a spatial analysis based on sociodemographic, geographical, and ecological data, and found that higher rodent species richness was significantly associated with a lower risk of LF emergence in West Africa from 2008 to 2017 (Odds Ratio = 0.852, 95% Credible Interval = 0.745-0.971).

Conclusions/significance: The results reinforce the importance of the 'One Health' approach by demonstrating that a high level of biodiversity could benefit human health.

Fig 1. Study design.

The outcome of interest was Lassa fever (LF) emergence events over a 10-year period beginning in 2008, and the main explanatory variable was rodent species richness. Regions with LF cases reported before 2008 were excluded from the analysis; we compared rodent species richness between regions with LF cases from 2008 to 2017 and those with no reported LF cases up to 2017. Regions A and D had LF cases before 2008, so were excluded from the analysis. In Region C, LF emerged after 2008. We compared rodent species richness between Regions C and B.

Fig 2. Study area and study units.

Based on web-based surveillance data and a prior review, historical Lassa fever (LF) cases (up to 2017) were analyzed. Provinces in West Africa were categorized as follows: grey, outbreak reported before 2007; red, LF emergence events over a 10-year period beginning in 2008; and white, no human LF cases up to 2017. We created a 1 × 1° grid map, and categorized the grids based on their intersection with the provinces. For example, using a 50% threshold, if a grid consisted of 60% province A and 40% province B, it was considered to be province A. We created two datasets using thresholds of 40% and 60% to confirm the robustness of the results. Made with Natural Earth.