Use of generalized additives to study the association of incidence of malaria and climate variables in epidemic prone sites in Tanzania between 1998 and 2005

Abstract

A Generalized additive poisson model was used to study the association of climatic variables and incidence of malaria. GAMs allow non-linear relationships via smooth functions between the predictors and the dependent variable . The smoothing parameters are used to regulate the overall flexibility and the degree of non-linearity of the model. Smoothing parameters are directly estimated from the data using the penalized likelihood introduced by Wood (2000). Malaria incidence was calculated by deviding the number of new cases during period by population at risk at that particular period. Outpatients hospital data collected over eight years in four districts located in epidemic prone areas was used to study the association of climate variables with malaria incidence in highlands areas. Our findings showed that there is a non linear relationship between malaria incidence and climatic variables. Malaria incidence increases with rainfall and peaks at 250mm and starts declining when rainfall exceeds 300mm. Adjusting for other factors, mean temperature was linear related with malaria incidence of which the highest incidence was observed at 26 0 C The findings of this study confirmed that; climate variables playa role on faciliating the transmision of malaria. Adjusted for other climate variables we observed a linear relationship between incidence of malaria and mean monthly temperature(df=l).