infection, the sharing of drug injection equipment, including needle sharing and the use of shooting galleries, was the most significant predictor of HIV (Marmor et al., 1987; Vlahov et al., 1990; Solomon et al., 1993; Schoenbaum et al., 1989; Nicolosi et al., 1992). The type of drug administered intravenously is also associated with HIV infection. After controlling for other known risk factors, intravenous drug users who inject cocaine are more likely to be infected than intravenous drug users who inject heroin (Anthony et al., 1991; Chaisson et al., 1989; Novick et al., 1989). This may be partly because cocaine injectors are known to inject more frequently than other injectors in any given drug session.
Certain demographic factors are also associated with infection. As mentioned above, there are dramatic geographic differences in the distribution of AIDS in general, and HIV infection among intravenous drug users in particular (Des Jarlais et al., 1991; Lange et al., 1988). In general, rates are also higher among minority intravenous drug users (Friedman et al., 1987; Schoenbaum et al., 1989), women (Nelson et al., 1991, in press), and younger users (Nicolosi et al., 1992; Solomon et al., 1993; Nelson et al., 1993). These factors highlight the importance of directing prevention activities to these groups. In addition to minority women, such at-risk individuals include young, recent initiates into drug use, many of whom may be excluded from programs in which evidence of chronic use is a criterion for participation.
As mentioned above, gay and bisexual intravenous drug users are not rare, and bisexual intravenous drug users may act as a conduit for transmission from both the homosexual and the intravenous drug user communities to heterosexuals, particularly women. In a survey of intravenous drug users recruited in San Francisco from street outreach and drug treatment programs (Lewis and Watters, 1991), 12 percent (49/396) of intravenous drug-using men were bisexual, based on reports of female partners prior to entry into the study. However, fewer than half of these men identified themselves as being bisexual, highlighting the difficulties that female partners have in assessing their male sexual partner's risk history. This may be particularly important because, in at least one study (Mandell et al., 1994), bisexual men (along with homosexual men) were more likely to report needle sharing than their heterosexual counterparts. Bisexual men may also have been important by acting as a conduit for HIV infection between homosexual men and the injection drug user community, at least in some areas. For example, in Chicago (Lampinen, 1992), the first reported AIDS case from injection drug use occurred in a bisexual man, and large numbers of