ABSTRACT
A clinical AIDS case definition is needed for surveillance in countries where the CDC case definition is not practical. To derive such a definition, we compared 110 HIV-seropositive and 135 randomly selected HIV-seronegative adult medical-ward inpatients in Brazil. Multivariate analysis of clinical signs and symptoms and simple diagnoses resulted in a discriminant function with sensitivity of 89% and specificity of 96% in predicting for AIDS. These data were the empirical basis for a clinical definition of AIDS in adults drafted in a Caracas, Venezuela, workshop sponsored by the Pan American Health Organization. The revised "Caracas" definition presented here requires a positive HIV serology, the absence of cancer or other cause of immunosuppression, plus > or = 10 cumulative points, as follows: Kaposi's sarcoma (10 points); extrapulmonary/noncavitary pulmonary tuberculosis (10); oral candidiasis or hairy leukoplakia (5); cavitary pulmonary/unspecified tuberculosis (5); herpes zoster < 60 years of age (5); CNS dysfunction (5); diarrhea > or = 1 month (2); fever > or = 1 month (2); cachexia or > 10% weight loss (2); asthenia > or = 1 month (2); persistent dermatitis (2); anemia, lymphopenia, or thrombocytopenia (2); persistent cough or any pneumonia except TB (2); and lymphadenopathy > or = 1 cm at > or = 2 noninguinal sites for > or = 1 month (2). This definition has a sensitivity of 95% and a specificity of 100% (91% without HIV serology) when applied to the Brazilian patients in this study. The Caracas definition has been adopted by Brazil, Honduras, and Surinam, and is in validation elsewhere. The use of a reasonably sensitive and specific case definition commensurate with available diagnostic resources should facilitate AIDS surveillance in developing countries.
