Faculty of Computing, Health and Science
School of Engineering (SOE) / Natural Resources Modelling and Simulation Research Group
Malaria is a major challenge to both the public health and the socio-economic development of Ghana. Major factors which account for this situation include poor environmental conditions and the lack of prevention services. In spite of the numerous intervention measures, the disease continues to be the most prevalent health problem in the country. The risk assessment reports for Ghana were based on household surveys which provide inadequate data for accurate analysis of incidence cases. This poses a serious threat to planning and management for the health care delivery system in Ghana. Malaria transmission varies with geographical location and time (or season). Spatio-temporal modelling coupled with adequate data has shown to better define the public burden of the disease, providing risk maps to describe the incidence variation in space and time and also identifying high risk areas for health policy implementation. Geostatistics contributes immensely to the prediction of the random processes distributed in space or time in epidemiological studies. In this study, we conduct spatial statistical analysis of malaria incidence to produce evidence-based monthly maps of Ghana illustrating the patterns of malaria risk over space and time. This is achieved using monthly morbidity cases reported on the disease from public health facilities at district level and population data over the period 1998-2010 to compute the malaria incidence rates, being the number of reported cases per unit resident population of 10,000. Lognormal ordinary kriging is used to model the spatial and temporal correlations, and then back-transformed to estimate the monthly malaria risk at local level. The space-time experimental variogram describing the correlations structure is modelled with nested spherical and exponential-cosine functions coupled with nugget effect. The modelled variogram indicate both short and long spatial and temporal dependence of the malaria incidence rates at local level with the temporal component exhibiting an increasing seasonal pattern of period of 12 months. The results also indicate varied spatial distribution of malaria incidence across the country, the highest risk being observed in the northern most and several locations in central and western parts of the country, and lowest in some areas in the north and south along the coast. This statistical-based model approach of malaria epidemiology will be useful for short-term prediction and also provide a basis for resource allocation for the disease’s control in the country.