Background: Modelling malaria parasitaemia as function of fever has been proposed as best alternative to estimate the attributable fraction of malaria fever and the sensitivity and specificity of different case definitions of malaria disease.
Objectives: To determine the prevalence of fever and its relation to malaria parasitaemia and to establish a pyrogenic threshold for malaria disease in the area.
Methods: We conducted two cross-sectional surveys in children of 6 months to 9 years of age (2434 during the rainy season of 1993 and 2353 during the dry season of 1994) randomly selected from 21 areas of Bandiagara district, Mali.
Results: The relationship between fever and Plasmodium falciparum parasitaemia depends strongly on the season, thus affecting the malaria-attributable fraction of fever cases and the sensitivity and specificity of malaria case definitions. The overall proportion of fever attributable to malaria parasitaemia was 33.6% during the rainy season and 23.3% during the dry season, with the highest proportion occurring among the youngest children. The cut-off value, where the sensitivity curve crosses the specificity curve, was around 3200 pf/microl for all age categories during the rainy season and 200 pf/microl during the dry season.
Conclusions: Malaria remains a main cause of fever in this area of Mali. The pyrogenic threshold of parasitaemia depends strongly on the season, and different cut-off levels of parasitaemia should be used during the two seasons to define malaria cases in this area.