Abstract
BACKGROUND: Diagnosis of an opportunistic illness (OI) in a person with HIV infection is a sentinel event, indicating opportunities for improving diagnosis of HIV infection and secondary prevention efforts. In the past, rates of OIs in the United States have been calculated in observational cohorts, which may have limited representativeness. METHODS: We used data from a 1998 population-based survey of persons in care for HIV infection to demonstrate the utility of population-based survey data for the calculation of OI rates, with inference to populations in care for HIV infection in three geographic areas: King County Washington, selected health districts in Louisiana, and the state of Michigan. RESULTS: The overall OI rate was 13.8 per 100 persons with HIV infection in care during 1998 (95% CI, 10.2-17.3). In 1998, an estimated 11.3% of all persons with HIV in care in these areas had at least one OI diagnosis (CI, 8.8-13.9). The most commonly diagnosed OIs were Pneumocystis jiroveci pneumonia (PCP) (annual incidence 2.4 per 100 persons, CI 1.0-3.8) and cytomegalovirus retinitis (annual incidence 2.4 per 100 persons, CI 1.0-3.7). OI diagnosis rates were higher in Michigan than in the other two geographic areas, and were different among patients who were white, black and of other races, but were not different by sex or history of injection drug use. CONCLUSION: Data from population-based surveys - and, in the coming years, clinical outcomes surveillance systems in the United States - can be used to calculate OI rates with improved generalizability, and such rates should be used in the future as a meaningful indicator of clinical outcomes in persons with HIV infection in care.