Spatial Modelling of Malaria Risk in Bayesian Setting: A Case Study of Wolaita and Dawuro Zones in SNNPR, Ethiopia

Abstract

Background: Malaria is a mosquito-borne infectious disease affecting humans and other animals caused by parasitic protozoans. The main objective of the study was to develop Bayesian spatial model for malaria risk in Wolaita and Dawuro zone of Southren Regional State, Ethiopia.

Methods:

In this study, malaria data obtained from seven woreda Health Centers of Wolaita and Dawuro zones at 345 spatial locations were used. At these locations, about three hundred twelve respondents had malaria in their blood samples out of 5, 062 respondents were tested for malaria infection. In the analysis, Generalized Linear Mixed Model was fitted to estimate Generalized Linear Mixed Model parameters to identify significant explanatory variables analyzed by using  statistical softwares (STATAversion 12 and SPSS version 16).

Result: The overall malaria prevalence in the study area was about 19.87%. Results indicate that malaria incidence follows spatial pattern because the test result indicates that there is statistically significant local clustering of malaria incidence at 5% level of significance. Statistically significant local clustering of malaria incidence is detected in all the woredas except in the two woredas (Kido Didaye and Loma woredas). In Kido Didaye and Loma woredas, the spatial correlation is negative that means the observed is less than that of expected value. The rest of the woredas exhibit positive spatial correlation since the observed value is greater than expected value. For Bayesian spatial models; the environmental factor elevation was negatively associated with malaria risk. This is to mean that as elevation above sea level of the study area increases, the chance of being a candidate of malaria decreases. A negative relation of maximum temperature with malaria risk reveals that the lower the maximum temperature the higher malaria risk.

Conclusions: Spatial modelling of malaria risk was the basis for differentiation of predicted malaria prevalence from high level to low on a map. The differentiation may allow effective use of limited financial and human resources. It also helps to identify priority areas to control malaria in case of change of climatic variables.

Keywords: Spatial autocorrelation, GLMM, Local risk factor, Global Local risk factor