numerous other plausible models that could explain their data. From an economic standpoint, it would seem that individual socioeconomic status may be causally related to both drug abuse and use of the needle exchange program, rather than to the explanation that needle exchange programs cause drug use. Nonetheless, the authors do not test any alternative plausible models to assess the relative fit of their models compared with other viable competing models. Moreover, a weak theoretical justification is provided of their postulated model (e.g., cost of needles is the underlying driving force for using the needle exchange, but that cost is minute compared with other expenses, such as the cost of the drugs themselves).
The empirical information provided on key variables is inadequate. Properties of the distributions of key variables are absent and aggregate summary statistics are used in various models without attention to the possible adverse effect of outliers. The presence of such outliers can severely distort the results and challenges the viability of the inferences drawn by these investigators. Substantial inconsistencies between data on key variables (self-report) presented in the manuscript and information extracted from the needle exchange program records raised serious concerns among panel members. Moreover, as discussed in some detail in Appendix A, the panel had serious reservations about the appropriateness of the modeling techniques as implemented by these researchers.
Although this particular study suffers from serious limitations, the conclusions reached by the authors raise interesting questions and hypotheses that should be subjected to sound empirical testing. These issues should be further studied with adequate designs, measures, and analytical methods. In the meantime, in the panel's opinion, these difficulties are serious enough to preclude making causal inferences about the effect of needle exchange programs.
The concern that having the opportunity to use a needle exchange may lead persons who are not currently injecting to begin injecting demands attention, and some information about this is available.
If the opportunity to participate in needle exchange programs were to lead to an increase in the number of new injection drug users, one would expect to see relatively large numbers of young newer injectors at the needle exchange programs. This has not been observed in any of the earlier studies (e.g., Lurie et al., 1993), or in the most recent publications (Paone et al., 1994a, 1994b; Des Jarlais et al., 1994b; Watters et al., 1994).
Investigators in Amsterdam have recently published data that permit examination of the hypothesis that "mixing" of injecting and noninjecting drug users at needle exchanges will lead noninjectors to begin injecting