Capture-recapture to estimate the size of the population with human immunodeficiency virus type 1 infection.

Estimating and monitoring the total number of people infected with human immunodeficiency virus type 1 (HIV-1) is a critical public health objective. No single epidemiologic methodology yields completely reliable estimates. We used techniques originally designed to estimate the size of wildlife populations to complement estimates of the size of the population with HIV-1 infection in Lazio, Italy, during 1990 obtained from surveillance, surveys, and dynamic mathematical models. We used reports from four large testing sites to generate incomplete, partially overlapping lists of HIV-positive subjects. Log-linear models yielded estimated prevalences of 5.65 per 1,000 among males (95% confidence interval = 4.52-6.78) and 1.84 per 1,000 (95% confidence interval = 1.34-2.33) among females in the population age 15-64 years. This method provides a simple and inexpensive means of obtaining accurate estimates of the total number of HIV seropositives. It could be applied easily in all situations in which data from multiple sources are available.