Members of highly vulnerable populations, such as injection drug users, are known to frequently come in contact with various social institutions. These include public medical facilities, such as drug treatment facilities, clinics for treating sexually transmitted diseases, emergency rooms, and correction facilities. For example, more injection drug users can be found in prisons than in drug treatment programs, hospitals, and social services (Brewer and Derrickson, 1992). It is plausible that preimprisonment drug-use behavior among incarcerated injection drug users will be more heterogeneous than that of users in treatment facilities or needle exchange programs (Vlahov and Polk, 1988). Moreover, Siegal et al. (1994), in a study that examined the relationship between level of HIV risk behavior and history of exposure to jail or prison, report that active injection drug users with the highest HIV risk behaviors were those most likely to spend time in jail. These findings, combined with the fact that more than 4 million people are incarcerated annually in the United States, argue for devoting more prevention research efforts to this subpopulation of injection drug users.
The emphasis of this chapter has been on identifying research issues that need to be vigorously pursued in order to improve the current knowledge base. In addition, we feel that the issue of randomized trials of needle exchange programs should be addressed, because some researchers have argued for their implementation to adequately answer issues concerning the effectiveness of such programs.
As was stressed in a previous National Research Council evaluation report on AIDS (Coyle et al., 1991), the randomized experiment might be the ideal scheme to adopt when attempting to assess the effect of AIDS prevention strategies. That report acknowledged, however, that because of practical constraints (e.g., complying with true random assignments of individuals or communities, inability to provide ''blinding" of behavioral interventions, treatment attrition, cost), one may choose to focus instead on well-conceived observational epidemiologic designs. These practical constraints are severe in the case of needle exchange and bleach distribution programs.
The more important problems associated with randomized trials (i.e., randomized field experiments) are practical. There are currently over 55 cities in the United States that have implemented needle exchange programs. This places serious constraints on finding comparable communities that do not have needle exchange programs and would be willing to be