Towards empirical description of malaria seasonality in southern Africa: the example of Zimbabwe

BACKGROUND Quantitative description and mapping of malaria seasonality is important for timely spatial targeting of interventions and for modelling malaria risk. There is a need for seasonality models that predict quantitative variation in transmission between months. METHODS We use Zimbabwe as an example for developing an empirical map of malaria seasonality. We describe the relationship between seasonality in malaria and environmental covariates for the period 1988-1999, by fitting a spatial-temporal regression model within a Bayesian framework to provide smoothed maps of the seasonal trend. We adapt a seasonality concentration index used previously for rainfall to quantify malaria case load during the peak transmission season based on monthly values. RESULTS Combinations of mean monthly temperature (range 28-32 degrees C), maximum temperature (24-28 degrees C) and high rainfall provide suitable conditions for seasonal transmission. High monthly maximum and mean monthly minimum temperatures limit months of high transmission. The intensity of seasonal transmission was highest in the north western part of the country from February to May with the peak in April and lowest in the whole country from July to December. The north western lowlands had the highest concentration of malaria cases (> 25%) followed by some districts in the north central and eastern part with a moderate concentration of cases (20-25%). The central highlands and south eastern part of the country had the lowest concentration of malaria cases (< 20%). This pattern was closely associated to the geographic variation in the seasonality of climatic covariates particularly rainfall and temperature. CONCLUSIONS Our modelling approach quantifies the geographical variation in seasonal trend and the concentration of cases during the peak transmission season and therefore has potential application in malaria control. The use of a covariate adjusted empirical model may prove useful for predicting the seasonal risk pattern across southern Africa.