their drug habit, it also could facilitate the much wider dissemination of sterile syringes, particularly in areas in which limited program hours allow only a small number of injection drug users to be served directly by the program. More research on secondary exchange is warranted.

To properly measure sources of sterile syringes, asking injection drug users only if they use or do not use the needle exchange program is insufficient. Other extant sources of sterile syringes can impact enrollment and measurement of outcomes from the program.

Prior studies in Baltimore (Gleghorn, in press) show that about half of injection drug users had sterile sources of needles prior to the opening of the needle exchange program, which were primarily from diabetics and pharmacies in a city with a paraphernalia—but no prescription —law. If needle exchange programs differentially attract people with nonsterile street sources but not those who already have sources of sterile needles, then, when comparing participants and nonparticipants, an evaluation study might show no relative reduction in HIV incidence resulting from program participation. This may partially explain the lack of significant findings in the Amsterdam case-control study (in which nonparticipants had access to unrestricted pharmacy sales) and appears to contribute, in part, to the findings of the Montréal needle exchange program (see Appendix A). Evaluations of needle exchange programs should collect and analyze information about sources of needles and syringes for all study participants, regardless of their program participation status.

Some studies to date have collected data from programs on needle exchange participants and then sought data on a group of nonparticipants to generate comparisons. However, the different settings in which data are collected are likely to raise concerns about differential bias of interviewers and recall among participants.

The panel recommends the use of prospective community-based studies of injection drug users—independent of the operation of needle exchange programs—while gathering information on program use. Comparison groups of nonparticipants might nevertheless have contact with program users and even obtain needles from the program, although they are not directly enrolled. Therefore, information on contacts with persons who use needle exchange programs and are the recipients of secondary exchange should be collected and incorporated into the analysis.