[Risk approach in the diagnosis of malaria in high altitude regions].

Sixteen cases of malaria are reported per thousand inhabitants per year in Kayanza, a province in the High Plateaux of Burundi. A case control study was conducted between October 1990 and March 1991 to better document this phenomenon. The majority of confirmed cases presented in persons who had recently sojourned in an endemic area (relative risk = 18). All except one of the other patients were carriers of plasmodium species that can cause late recurrences, which is compatible with the absence of local transmission. The presumptive diagnosis of malaria, as it is routinely made in the health centres, is not very specific (60%) and it has a positive predictive value of only 10%. This implies that an anti-malaria treatment is useless for 90% of the patients who receive it. It also means that the true malaria incidence rate in Kayanza can be estimated at less than 1 per thousand per year. We propose an alternative diagnostic strategy based on the risk differences between population sub-groups. It should permit an adequate patient management and a satisfactory surveillance of malaria in non-endemic epidemiological strata.