Finally, in an attempt to control for potential selection bias due to the self-selection of the 412 repeaters, standardized seroconversion rates accounting for differences in age, sex, race, and participation in methadone maintenance between repeaters and nonrepeaters were derived. The estimated standardized rates were found to be 1.12 percent per person years for the entire group, 0.51 percent per person years for those who had never used needle exchange, and 2.97 percent per person years for those who had ever used the needle exchange.

These findings were interpreted by the authors as indicating that needle exchange programs attract a high-risk, chaotic, and indigent population of injection drug users. The authors conclude that needle exchange programs are well suited for implementing HIV preventive interventions, in view of their findings that such programs provide direct access to high-risk injectors. Moreover, the authors state that it is unlikely that the extremely high HIV seroconversion rate among study participants who had ever used the needle exchange compared with participants who had never used the needle exchange, or that the large hazards ratio for seroconversion associated with needle exchange participation, is the result of needle exchange use itself. Several possible explanations are offered by the investigators: (1) most participants had a very brief exposure period to the needle exchange; (2) the highest seroconversion rates were observed in the subgroup with the lowest amount of time exposed to the needle exchange, which is less than 3 months, less than the typical detection window for the HIV antibody; and (3) seroconversion rates were not based on a fixed schedule (i.e., voluntary re-enrollment).


The panel's main concern about this research is the potential misinterpretation of the reported findings. Although the authors do not attribute the observed disparities in seroconversion rates between needle exchange program participants and nonparticipants to exposure to these programs, their readers may misinterpret these study findings as reflecting the causal effect of needle exchange programs on seroconversion. As indicated by one of the investigators, the study design and analytical methods used by the researchers preclude such an assessment (Moss, 1995).

Indeed, for a number of reasons, the existing data are inadequate to assess a needle exchange effect by comparing the seroconversion rates of needle exchange users and nonusers. One possible strategy for attempting to make such an assessment would have been to estimate the seroconversion rate per injection drug user per unit of time at risk while exposed to the needle exchange with the seroconversion rate per injection drug user per unit of time at risk while not exposed to the needle exchange. This would have allowed a direct inferential test of the null hypothesis that the seroconversion

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement