How environmental, geographic, socio-demographic, and epidemiological indicators influence malaria prevalence

Abstract

Background: The interactions of environmental, geographic, socio-demographic, and epidemiological factors in shaping mosquito-borne disease transmission dynamics are complex and changeable, influencing the abundance and distribution of vectors and the pathogens they transmit. In this study, 27 years of prevalence data (1990-2017) were used to examine the effects of environmental (temperature, precipitation, normalized difference vegetation index, isothermality), geographic (elevation), socio-economic (population density, gross domestic product per capita, human development index), and epidemiological (basic reproductive number) factors on Plasmodium falciparum and Plasmodium vivax malaria in Africa and Asia. Methods: Monthly long-term, open-source data for each factor were compiled and analyzed using two approaches: generalized linear models (GLM) and classification and regression trees (CART). Both approaches were used to identify associations between environmental, geographic, socioeconomic, and epidemiological factors with malaria prevalence (the proportion of people infected with malaria in a given time period). Plasmodium. falciparum and P. vivax malaria prevalence were examined in Asia, and only P. falciparum in Africa. Model fit was assessed, estimating the accuracy, precision, recall, and the f1-score of the models. Results: Both CART and GLMs provided robust predictions of malaria prevalence, but the CART model exhibited slightly better accuracy and precision. An optimum temperature for malaria prevalence around 25 ◦C for P. falciparum prevalence was observed in Africa but not for P. falciparum and P. vivax in Asia which was below 25 ◦C. The decline of malaria prevalence from 2000 to 2012 was well capture by the models as well as the resurgence of P. falciparum prevalence in Africa and Asia. Precipitation showed a unimodal response to malaria prevalence with a positive effect up to a point of 200 - 400 mm of rainfall; then a negative effect was observed as precipitation increased. Higher predicted malaria prevalence in regions with lower GDPPC and lower HDI and a positive correlation with the basic reproductive number. Conclusion: This study showed that there was a combination of environmental, geographic, socioeconomic, and epidemiological indicators driving malaria prevalence. Identifying these, and describing their associations provides key information to inform planning strategies and interventions to reduce malaria burden in Africa and Asia.