Modeling Time to First Malaria Using Spatially Correlated Conditional Autoregressive Frailty Model

Abstract

The burden of malaria is a major public health concern in Ethiopia. Its dynamics is being changed by construction of dams which serve either for hydroelectric or irrigation purpose in the region. This study aimed at examining the impact of hydroelectric dam on malaria transmission in southwestern Ethiopia using Spatially Correlated Conditional Autoregressive Frailty (CAR) model. Method: A two-year weekly basis longitudinal study was conducted among children less than 10 years of age in sixteen villages, in southwest Ethiopia. CAR frailty model that accommodates the clustering effect were fitted to the malaria data set. The parameters in the model were estimated under a Bayesian framework using Markov Chain Monte Carlo (MCMC) approach. Results: Among 2040 children, 548 (26.9%) of them experienced malaria symptom in their blood samples during the study period. The minimum observed time for the first malaria infection was 4 days and the maximum was 698 days. The result reveals that the hazard of getting malaria infection is decreased by 5% for 1km distance away from the dam (HR=0.95, 95% CI: 0.88-0.99). Children aged > 3 years are more likely experienced malaria infection as compared to < 3 years of age. The result also showed that there is a marked clustering (Sigma=0.61 with 95% CI: 0.38 - 0.95) of villages in the study area. Hence the estimation of parameters with the assumption of neighborhood (Spatially Correlated CAR frailty model) was found to be parsimonious. Conclusions: Malaria control intervention program should consider the spatial variation of malaria transmission in order to get sustainable and efficient malaria control in the study area.