FIGURE 6.1 HIV seroconversion (per 100 person years) among out-of-treatment injection drug users in Chicago 1988-1992. SOURCE: Prevention of New HIV Infections Among Out-of-Treatment Injection Drug Users: A Four Year Prospective Study (Wiebel et al., 1994).

during that period. However, two independent data sets do provide support for the potential causal program effect on the observed findings.

During the sixth reinterview wave, the researchers recruited from nearby neighborhoods 248 previously unsurveyed street injection drug users. The prevalence of drug-use risk behavior among those who were seronegative was 50 percent for those with no reported outreach program contact. In addition, NIDA's surveys of drug treatment entrants in Chicago over a number of years include questions about needle sharing. The prevalence of self-reported needle sharing among 500 injection drug user treatment entrants surveyed in 1992 was 67 percent. The higher rates of risky injections found in these independent samples of injection drug users in the same communities lend some credence to the potential causal effect of the intervention on the observed decline in injection drug-use risk behaviors and HIV seroconversion. Nonetheless, it should be noted that these two separate data sets reflect substantially lower levels of drug-use risk behaviors than were observed at baseline among members of the intervention group (i.e., 100 percent). This implies that one cannot completely rule out some