and 3 prospective studies. The quality of studies was rated on a 5-point scale ranging from a low of 1 (not valid) to a high of 5 (excellent) and a mid-point of 3 (acceptable). Only two of the studies received a quality rating of 3 or higher, and two others were rated between 2 and 3. None of the studies showed increased prevalence or incidence of HIV infection among needle exchange participants.
Given the quality rating of the studies, it is not surprising that the University of California report concluded that the studies available up to the time of the report (Lurie et al., 1993) do not, and for methodological reasons probably cannot, provide clear evidence that needle exchange programs decrease HIV infection rates. However, needle exchange programs do not appear to be associated with increased rates of infection.
It is intrinsically difficult to measure effects of intervention on the incidence of new infections of rare diseases, whose victims ordinarily do not show symptoms at the time of infection. Although most of the early studies used prevalent infection as the outcome measure, the more appropriate measure is incident or new infection. However, a further complication is that incidence is low in most locations, thereby requiring larger study populations to demonstrate program effects. The University of California report noted (Lurie et al., 1993:465) appropriately:
Well-conducted, sufficiently large case-control studies offer the best combination of scientific rigor and feasibility for assessing the effect of needle exchange programs on HIV rates.
The University of California report noted that eight "acceptable" studies were identified that presented data on the issue of reported injection frequency. As Table 7.4 shows, three studies found reductions in injection associated with needle exchange programs; four found mixed or no effects; and one found an increase in injection compared with controls. This last study also found reduced needle sharing reported among needle exchange participants. This study noted that the apparent increase in injection could be attributed to several other factors, including the differential dropout of low-level injectors. The report also reviewed the methodological limitations of the studies, including the potential for socially acceptable responses by injection drug users. On balance, because of methodological problems, the report drew no strong conclusions about levels of injection drug use.