Criteria for better detection of brucellosis in the Narok District of Kenya.

ABSTRACT

Monthly disease summary sheets from 1986-1992 of 60 dispensaries, clinics and hospitals in Narok district, Kenya were reviewed for the occurrence of brucellosis and other diseases with "flu-like symptoms". Diseases with these symptoms accounted for about 52% of the 1,037,875 cases reported for the time period. These were classified as malaria (79.3%), rheumatism (7.1%), PUO (2.4%), and brucellosis (0.8%). Brucellosis was diagnosed by a positive Rose Bengal (RB) test routinely conducted in seven out of the 60 health units. In these units, 55% of flu-like cases were classified as malaria and 21.2% as brucellosis. Individual case records of patients at four dispensaries using the RB test during 1991-92 were assessed for specific predictor symptoms. For 625 RB tested patients, a positive test result was associated with joint pain, headache, and the combinations of joint pain with headache and lameness with headache. A logistic regression model correctly predicted the RB test result in 62.3% of the time. For the 465 patients examined by the blood smear examination, identification of malaria parasites was associated with, headache, joint pain and combinations of emesis with pale mucous membranes. This regression model correctly predicted positive results 67.2% of the time. Both models indicate that selected clinical predictors represented significantly increased odds of being positive to the respective tests. However, for both diseases, clinical signs alone appear insufficient for reliable diagnosis and differentiation probably due to resemblance in symptomatology between these two and other diseases.