Developing models for the prediction of Malaria Epidemics

The Southern African countries encompass the southern fringe of malaria in Africa and malaria transmission is characterised by being distinctly seasonal and epidemic (le Sueur & Sharp, 1996). The severity of these epidemics can be such that the health systems capacity to cope are stretched resulting in pronounced detrimental effects on the communities at risk.

This situation further highlights the need for an "Early Warning System" which will enable Health authorities to respond timeously to such epidemics and thus offset the "exponential" pressure they exert on the health system and its logistical ability to cope (le Sueur & Sharp, 1996).

The following models are currently being used to predict malaria epidemics, but a more robust basis for prediction of malaria epidemics based on a multilevel approach is necessary, as opposed to the use of just one technique.

Clinic Cases

This study investigated the use of early season malaria cases in predicting the severity of an epidemic. Since malaria is a notifiable disease in South Africa, data was obtained from the three Malaria Provinces in South Africa for the period 1981-1996.

In South Africa malaria epidemics are currently defined as an increase in the annual malaria case total above the annual mean, calculated from 1981. The weekly total number of cases reported in the three respective provinces were analysed from October through to February to discern any association between these weekly totals and severity of epidemic.

Statistical models

As part of the project that outlined the process model for malaria prediction, Dr. D le Sueur and Dr. M Jury developed the first statistical malaria prediction model for South Africa (le Sueur & Sharp, 1996).

Collaborators

Medical Research Council, University of Zululand and Provincial Malaria Control Programmes

Funding

Medical Research Council and National Oceanic and Atmospheric Administration (USA)