Diagnostic accuracy of three clinical case definitions for advanced HIV disease

Objective: To assess the accuracy of three clinical case definitions for advanced HIV disease: the World Health Organization (WHO) case definition, and the original and revised Caracas case definitions. Design: Retrospective chart review. Setting: A clinic for patients with all stages of HIV infection at the Johns Hopkins Hospital, Bethesda, Maryland, USA, a tertiary care university hospital. Patients, participants: Two hundred and twenty-four HIV-positive adults who underwent initial evaluation between 1 January 1990 and 31 December 1990. Main outcome measures: A score for each definition was assigned based on initial evaluation. The sensitivity, specificity, and predictive values were calculated using the Centers for Disease Control (CDC) staging criteria, and results were correlated with total CD4 cell counts. Results: The sensitivities of the WHO, and the original and revised Caracas definitions were 40, 67, and 60%, respectively, using CDC disease stage IV as a positive standard. Specificities were between 99 and 100%, using CDC stage II-III disease as a negative standard. Mean CD4 cell counts for patients with positive scores were 184, 160, and 158 × 10⁶/I, respectively, compared to 191 × 10⁶/I for CDC stage IV patients. Sensitivity was lower when the positive standard was expanded to include all patients with CD4 cell counts < 200 × 10⁶/I. Conclusions: In our study population, case definitions were specific, but only moderately sensitive for advanced HIV disease. Prospective studies should be conducted in diverse geographic regions, using lymphocyte or CD4 cell counts when possible.