INTERPRETING HIV SEROPREVALENCE DATA FROM A STREET-BASED OUTREACH PROGRAM

A total of 4,375 subjects were recruited through continuous street out reach over 31 months for interviews on HIV-related risk behavior and H IV-antibody testing. Changes over time among sampled subjects with res pect to HIV infection and HIV-risk behavior have been examined retrosp ectively, and significant and consistent trends noted across successiv e cohorts. Although overall 42% of the sample tested HIV antibody-posi tive, HIV infection exhibited a consistent downward trend from 60% in the first quarter year of interviewing to 22% in the final quarter yea r. Scores on a multivariate index of HIV risk also declined significan tly. Mean age, proportion of Black subjects, mean length of drug injec tion career, frequency of drug injection, and the use of shooting gall eries all declined significantly across quarters as well. We argue tha t these observed differences result from a systematic sampling bias in herent in our outreach-driven sampling procedures, which initially fav ored recruitment of IDUs with greater behavioral and demographic risk for HIV.