Diagnostic accuracy of three clinical case definitions for advanced HIV disease.

ABSTRACT

PIP: The 1st case definition for AIDS was developed by the Centers for Disease Control (CDC) in 1982. WHO adopted CDC definitions for use in some countries and also developed a clinical case definition where HIV serology tests were not feasible. A multivariate analysis of data of Brazilian AIDS patients with positive HIV serology provided the basis for the Caracas definition in 1989 and it subsequent revision. The accuracy of these 3 clinical definitions was evaluated to see their predictive value in an advanced stage of AIDS. The records of 224 HIV-positive adults were reviewed in 1990. Scores were assigned to various symptoms. 80% of men and 20% of women with a median age of 33 years; 1/4 were white and 2/3 were black. 1/3 were homosexuals and 1.2 were iv drug users. 139 were asymptomatic (CDC stage I-II) and 85 were symptomatic (CDC stage IV). 58 patients had total CD4 cell counts of over 500 x 1 million/1; 91 had 200-500 x 1 million/l; and 70 had 200 x 1 million/1. 48 were taking zidovudine and Pneumocytis carinii drugs. The sensitivities of the WHO, original Caracas definition, and revised Caracas definition were 40%, 67%. and 60%, respectively, with 99-100% specificities and positive predictive values of 97-100%. The mean CD4 cell counts for the WHO, original and revised Caracas definitions were 184, 160, and 158 x 1 million/1, respectively, compared with 199 x 1 million/1 of patients with CDC stage IV disease. The predictive values of the 3 definitions for CD4 cell counts 200 x 1 million/1 reached 62%, 73%, and 71% vs. only 59% for CDC stage IV patients. The combination of stage IV symptoms or a CD4 cell count 200 x 1 million/1 produced sensitivities of 31%, 53%, and 47%, respectively, with 100% specificity and positive predictive values. The definitions were highly specific, but only moderately sensitive for advanced AIDS; the Caracas definitions were more sensitive than the WHO definition.