Effectiveness of syringe exchange programs in reducing HIV risk behavior and HIV seroconversion among injecting drug users.

Abstract

Drug addiction is a major risk factor for HIV infection for an estimated 3.4–5.5 million injecting drug users (IDU) in more than 125 countries [1–3]. In the USA, injecting drug use now accounts for nearly one-third of new AIDS cases [4] and for as many as three-quarters of annual incident HIV infections [5]. Behaviors associated with drug use that are responsible for HIV transmission include shared use of injection equipment and other drug paraphernalia, and unprotected vaginal and anal intercourse with an injecting drug user. Interventions that can reduce the prevalence of these behaviors are therefore a critical component of comprehensive AIDS prevention policy. Approaches to HIV risk reduction among IDU include drug abuse treatment, HIV testing and counseling programs, street-based outreach conducted by peer educators, individual and group counseling, community-level interventions to change IDU norms concerning safer injection and safer sex, and syringe exchange programs to provide IDU with sterile injection equipment. Syringe exchanges have been established widely in some developing, and a majority of the developed countries, [3] including the USA, although a Congressional funding ban still prevents them from receiving US government support. Although evidence indicates that syringe exchange is effective in slowing the spread of HIV in IDU (see, for example [6–8]) socially conservative American politicians have embraced contrary evidence from two Canadian cities [9,10] to assert that syringe exchanges are not only immoral because they encourage drug use, but may also actively spread HIV [2]. The purpose of this review is to provide a comprehensive, critical review of published evidence of the effectiveness of syringe exchange programs in reducing HIV risk behavior and HIV seroconversion among IDU. While five previous reviews have been published, four [5–8] are now seriously out of date, and a fifth [11] did not attempt a comprehensive critical review of the published literature. We have identified and will describe and discuss 42 studies published between 1989 and the end of 1999 that evaluated syringe exchange effectiveness. Identification of studies We systematically searched the published scientific literature in the English language using Medline (National Library of Medicine, Bethesda, Maryland, USA) and PsychInfo (American Psychological Association, Washington DC, USA) data bases for the period 1989–1999 using the key words ‘needle exchange’ and ‘syringe exchange'. We then supplemented these searches by review of bibliographies in papers that came to our attention as well as inquiries among colleagues. We did not attempt to review unpublished reports, but we did examine published commentaries on research findings. Our searches did not uncover any published work on syringe exchange effectiveness in languages other than English. Almost all of the studies we found were conducted in the USA, Canada, the UK, or the Netherlands. We included all published studies of the effectiveness of syringe exchange programs (SEP) in reducing HIV risk behavior and HIV seroconversion, regardless of design. We also included the few studies we were able to find that examined effects of syringe exchange on hepatitis B (HBV) and hepatitis C (HCV) seroconversion, as hepatitis and HIV are transmitted in very similar ways and are, therefore, good biological markers for behaviors that place drug users at risk for HIV (they are also obviously important biological outcomes in their own right). The authors each independently reviewed each of the studies, and abstracted essential information concerning design, procedures, results, statistical controls, and study limitations. We then reconciled our ratings of the studies, arriving at a consensus description of each. We did not attempt a formal meta-analysis because outcome measures and/or data analysis differed markedly across studies. Criteria used in appraising the strengths and limitations of the studies included the adequacy of statistical controls, including adjustment for selection bias; the sensitivity of outcome measures used, statistical power, and where applicable, the meaningfulness of contrasts between users and non-users of SEP. In longitudinal studies, potential limitations include biases related to loss of respondents to follow-up, and in the case–control studies, failure to properly match cases and controls. In the case–control studies, recall bias is another potential limitation. Still another potential problem in longitudinal studies of syringe exchange is low baseline HIV seroprevalence and the related issue of restriction of person-time experience of incidence. Of the 42 studies reviewed, 23 were community studies in which the behavior and HIV status of users of SEP were compared with those of IDU not using SEP. Another 11 studies were conducted exclusively with SEP clients. Two studies were conducted with both community samples and with SEP clients. In addition, there were six studies that evaluated the ecological impact of SEP. The 42 studies used a variety of designs, including longitudinal or prospective cohort (10 studies), case–control (five studies), and multiple cross-sectional or observational (five studies and 16 studies, respectively). Five studies were based on evidence gathered at a single syringe exchange and used to model mathematically syringe exchange effectiveness. In describing the 42 studies, we group them first by study type (community study, syringe exchange study, ecological study), and second, within type by design (longitudinal/prospective, case–control, observational, modeling). The reader will note that several studies [9,12,13] used more than one design; for the sake of simplicity, we will discuss results obtained with each design separately. Community studies with longitudinal/prospective design Seven studies followed samples of IDU longitudinally, and compared HIV status and behavior of drug users who did versus those who did not use or attend syringe exchange (see Table 1). We will use the term ‘use’ and ‘attend’ interchangeably to mean that IDU themselves visited a syringe exchange for the purpose of exchanging.Table 1: Longitudinal/prospective cohort studies conducted with injecting drug user (IDU) community samples. In a study conducted in Montreal, Canada, Bruneau et al.[9] followed a cohort of respondents for a mean of 21 months, finding any SEP use in the previous 6 months to be a risk factor for seroconversion [adjusted odds ratio (OR), 1.7); the study differed from the majority of the studies described here in that IDU clients were predominantly cocaine users and had legal access to sterile syringes through pharmacies. In a recently presented abstract [15], however, Bruneau's team found no relationship between SEP use and HIV seroconversion, suggesting that the association may fade or be unstable over time. Des Jarlais et al.[16] performed a meta-analysis of three studies of SEP clients, vaccine preparedness respondents, and respondents in high-risk National AIDS Demonstration and Research cities, all of which were conducted in the USA. Controlling for potential confounders identified in a previous study, the investigators found that not using SEP continuously to be risk factor for HIV seroconversion (adjusted OR, 3.5). By contrast, Hagan et al.[17] in Seattle found no protective effect against HBV- and HCV-seroconversion of former, sporadic, or regular SEP use; their analysis adjusted for recent initiation of injection drug use and daily injection at baseline, both of which appeared to confound the effects of SEP use. In Portland, another West Coast US city, Oliver et al.[14] followed peer outreach and SEP clients over 6 months; although there was risk reduction in both groups, SEP clients were more likely to stop reusing ‘works’ without cleaning them, and were also less likely to throw away syringes (instead of returning them to then SEP); the investigators, however, did not adjust for preexisting differences between the two groups. In Vancouver, Canada, Schechter et al.[18] followed 694 treated and untreated IDU over 15 months, finding that HIV incidence was higher in frequent versus infrequent SEP attendees; however, when a lengthy list of potential confounders were controlled for, no relationship was observed between incidence and SEP use. Of added interest, the data showed that only 0.7% of the sample reported meeting new friends or other people there, suggesting that SEP use does not contribute to new sharing relationships. In New York City, Schoenbaum et al.[19] followed 329 IDU in methadone treatment over a period of 8 years. Following the introduction of SEP, they found increasing behavioral differences between SEP users and non-users, with the SEP users significantly less likely to report sharing syringes 4 years following introduction of the SEP. Interestingly, data collected from these methadone clients before SEP was implemented showed that there was less sharing of injection equipment among methadone patients who would later become SEP clients, suggesting that some of the later differences may have been due to client self-selection. van Haastrect et al.[20] also followed IDU over 8 years, examining predictors of mortality in HIV-positive and HIV-negative drug users in Amsterdam; participation in syringe exchange was not associated with a lower mortality rate. It is important to note that IDU in Amsterdam could legally purchase pharmacy syringes in addition to obtaining them from a syringe exchange. Community studies using multiple cross-sectional designs We identified four multiple cross-sectional studies (see Table 2). Bluthenthal et al.[12] interviewed untreated IDU in six waves of interviews conducted over 3 years in the San Francisco Bay Area. With each successive wave, the proportion of respondents reporting using a SEP to exchange syringes and obtain supplies increased, and there were at the same time decreases in the proportion reporting sharing syringes and supplies. Broadhead et al.[21] studied the impact of the closure of a Windham (USA) syringe exchange; rates of reusing of syringes increased from an average of 3.5 times monthly before or during closure to 7.7 times after; self-reported sharing of syringes also increased from 16% to 34%. Des Jarlais et al.[22] in New York City used a design very similar to that of Blunthenthal et al., except that the cross-sectional surveys were conducted quarterly over 3 years. Quarterly level of use of syringe exchange use correlated -0.67 and -0.44, respectively, with the proportion of respondents reporting borrowing and lending of syringes. In the fourth study, conducted by Singer et al.[23] in Hartford, successive decreases in reuse of syringes and supplies followed legalization of pharmacy sale of syringes, and later, the introduction of an SEP.Table 2: Multiple cross-sectional studies conducted with injecting drug user (IDU) community samples. Case control studies with community samples We identified four case–control studies which evaluated syringe exchange effectiveness (see Table 3). In a case–control study embedded in the longitudinal study already described, Bruneau et al.[9] compared 88 HIV seroconverters with a matched sample of 320 seronegative IDU. Surprisingly, cases were more likely to use syringe exchange [adjusted odds ratio (AOR), 4.2], when a host of other variables were controlled. As noted, this was a largely cocaine-using sample obtained in a city where pharmacy sale of syringes is legal. A very different result was obtained by Hagan et al.[13,24] in Tacoma USA. Matching cases and controls on demographic characteristics and injection frequency, they found ‘never use’ of SEP to be associated with a six- and sevenfold increased risk, respectively, of acquiring HBV and HCV. In Vancouver, Canada, Patrick et al.[25] conducted a case–control study to identify predictors of HIV seroconversion. No association was found between ever use of SEP and HIV seropositivity; 98% of the cases and 89% of the controls used an SEP. It is also worth noting that syringes may be purchased legally in pharmacies in Vancouver. Finally, van Ameijden et al.[26] in Amsterdam found no protective effect related to the proportion of syringes exchanged at an SEP; in Amsterdam as in Vancouver, IDU can obtain syringes legally from pharmacies.Table 3: Case–control studies conducted with injecting drug user (IDU) community samples. Observational studies conducted with IDU community samples Thirteen observational studies of IDU were identified, the bulk of which compared users and non-users of syringe exchange; several studies, however, compared IDU according to their frequency of SEP use (see Table 4).Table 4: Observational studies conducted with injecting drug user (IDU) community samples. Bluthenthal et al.[12] found a > 40% protective effect (AOR, 0.57) associated with SEP use, controlling for previously identified confounders. Bruneau et al.[9], obtained very different results in Montreal, finding an AOR of 2.2 for respondents reporting recent use of the SEP. As noted, this was a study conducted primarily with cocaine injectors in a setting where syringes could be purchased legally. In London, where pharmacy purchase of syringes was also an option, Donoghoe et al.[27] found users and non-users of SEP did not differ with regard to the extent to which they shared used injection equipment. The same was not true of a sample studied by Frischer et al. in Glasgow [28,29]; there, SEP attenders were less likely (mean 4.5 times/month) than non-attenders (mean 9.2 times/month) to borrow used syringes, despite legal pharmacy sale of syringes. Two studies conducted by Hartgers et al. in Amsterdam [30,31] found contradictory results. In the first study, exchangers were less likely than non-exchangers to have borrowed syringes during both the previous month (10% and 23% respectively) and the previous 2 years (33% and 57% respectively). In the second study, regular exchanging was not associated with increased or decreased borrowing. The second study controlled for previously identified confounders. As noted, IDU in Amsterdam could legally purchase syringes in pharmacies. The same was true of IDU in a study conducted by Keene et al.[32] in Wales. The investigators, however, found use of the SEP to be strongly protective against borrowing of syringes; only 9% and 10% of attenders reported borrowing in 2 consecutive years, compared with 41% and 39% of non-attenders. The investigators did not control for potentially confounding variables. Klee et al. conducted several studies of SEP in the northwest of England [33,34]. In these studies, SEP use was not consistently associated with accepting used syringes; in one, use of SEP was associated with passing used syringes to others. In Vancouver, Canada, Strathdee et al.[10] found frequent attenders (more than once a week) of the SEP were somewhat more likely (81%) than non-attenders (71%) to be HIV positive; a wide variety of potential confounders was controlled, and the AOR for HIV seropositivity of SEP use was small but significant (AOR, 1.68). However, a later study already referred to [18], which was conducted in the same setting, found that when additional confounders were controlled for, no relationship was observed between incidence and SEP use. In Amsterdam, a study conducted by van Ameijden et al.[35] found bivariate negative associations between SEP use and borrowing and lending of syringes; these associations, however, did not survive multivariate analyses. In a later study, also conducted in Amsterdam, van Ameijden and Coutinho [36] examined the impact of regular versus irregular versus no use of an SEP, adjusting for a number of possible confounders. Irregular use of the syringe exchange was associated with an 11.6% excess risk compared with regular users, but non-users were at 2.9% decreased risk, suggesting that non-users had other stable sources of syringes, possibly through pharmacies. In San Francisco, Watters et al.[37] found an almost twofold protective effect of SEP use (AOR, 0.54) with regard to reported needle sharing during the previous 30 days. Longitudinal/prospective cohort studies conducted with SEP clients Four studies followed SEP clients prospectively, comparing behavior reported at the point clients began to use syringe exchange with that at a follow-up interval (see Table 5). Of the four, only Donoghoe et al.[38] had a non-SEP attending comparison group. Among SEP attenders, 20% who reported sharing at baseline had ceased sharing by a 2 month follow-up, although 4% who reported no sharing at baseline reported it at follow-up; no changes in sharing were observed in the comparison group. In a study conducted by Hagen et al.[13] in Tacoma USA, the portion of SEP clients who reported borrowing and lending of syringes in the previous month declined from 58% and 72%, respectively, at baseline to 33% and 46% at follow-up. In nearby Portland USA, Oliver et al.[14] found that the proportion of clients who reported borrowing a syringe declined from 20% at initiation of SEP use to 7% at a 6 month follow-up, and the proportion of clients who reported renting a syringe decreased from 9% to 3%. Using a similar follow-up interval, Vlahov et al.[39] in Baltimore USA, found a 22% to 8% decrease in borrowing a used syringe, a 27% to 12% reduction in lending a used syringe, and additional reductions in indirect sharing (use of ‘cottons’ or ‘cookers,’ or using a syringe to divide drugs).Table 5: Longitudinal/prospective cohort studies conducted with syringe exchange program (SEP) clients. Observational studies conducted with SEP clients We located four observational (one-shot) studies conducted with SEP clients (see Table 6). In two studies from San Francisco USA [40,41], Guydish et al. found that recent users of syringe exchange had fewer sharing partners; in the first study, number of sharing partners was correlated −0.39 with number of exchange visits in the past 30 days, but was not correlated with number of sharing occasions. In the second study, a finding of decreased numbers of partners was only a trend, but respondents who did not report sharing obtained a somewhat higher proportion of their syringes from SEP (92% versus 81%). Paone et al.[42], in New York City, interviewed IDU concerning their behavior during the previous 30 days and the 30 days preceding the initiation of SEP use. The proportion who reported borrowing used syringes decreased from 29% at initiation of SEP use to 12% at follow-up, and renting or buying used syringe decreased from 22% to 6%.Table 6: Observational studies conducted with sryringe exchange pogram (SEP) clients. Multiple cross-sectional study with SEP clients Peak et al.[43] repeatedly surveyed syringe exchange participants over 4 years, but did not attempt to measure the pre- to post-differences of individuals (the study can thus be properly classified as multiple cross-sectional). Over 4 years, the investigators found that the median number of times participants shared needles decreased from 14 to two, and the median number of people they shared with decreased from two to one (see Table 7).Table 7: Multiple cross-sectional study conducted with sryringe exchange pogram (SEP) clients. Ecological studies We identified five ecological studies of SEP effectiveness. Des Jarlais et al.[44] described the characteristics of averted HIV epidemics in five cities internationally. Three similarities among the five cities were found: implementation of prevention activities when HIV seropositivity was still low; provision of sterile injection equipment; and community-based outreach. Groseclose et al.[45] found that syringe sharing decreased after partial repeal of needle prescription and drug paraphernalia laws; there was also evidence that pharmacy sale of syringes increased following repeal. Hurley et al.[46] found that seropositivity increased by 5.9% annually in 52 cities without SEP, and decreased by 5.8% in cities with SEP. Lamden et al.[47] found that seroprevalence of HBV and HCV was not significantly different for IDU who began injecting before versus after implementation of an SEP, after taking into account differences in duration of injection. Ljungberg et al.[48] found a seroprevalence of zero in a city in south Sweden with an SEP, compared with prevalences up to 60% in other regions without them. Finally Stimson [49] suggested that there is ‘prima facie ’ evidence that rapid implementation of syringe distribution and expansion of methadone and outreach programs prevented an HIV epidemic from occurring in all but a few UK cities. Modeling studies Kaplan and his colleagues in New Haven USA, conducted a series of studies [50–54] to model the effectiveness an SEP in reducing the circulation time of syringes exchanged at the SEP. They reasoned that with decreased circulation time resulting from increased exchanging, the HIV infection rate in returned syringes would also decrease [54]. This proved to be the case. Within 3 months of initiating the syringe exchange, the circulation time decreased from 1 week to 2–3 days, and the percentage of syringes that were infected decreased by about one-third, from 67% to 44%[50,54]. They also estimated the circulation time before introduction of syringe exchange to be 23.5 days. [54] Further, incidence among exchange clients was estimated first as zero new infections per 100 drug injectors with a 95% confidence interval (CI) of 0–10.2 [50], and later as 1.63 new infections per 100 drug users per year, with a 95% CI of 0–7.2 [52]. In another study they determined the decrease in proportion of HIV-infected needles returned to the SEP was not due to a shift in population of clients served [53]. Although the insights that these studies provide are probably quite good, they are subject to the same limitations of all modeling studies, being only an approximation to the real world. For example, Kaplan and his colleagues assumed that needles are shared among all IDU, not just among small subsets of them; he also made assumptions about the timing of needle exchange and the efficacy of bleaching, as well as the representativeness of tested syringes of all syringes available in the community. Discussion To summarize, there is substantial evidence that syringe exchange programs are effective in preventing HIV risk behavior and HIV seroconversion among IDU. Of the 42 studies reviewed here, 28 found positive effects associated with use of syringe exchange, two found negative associations, and 14 found either no association or a mix of positive and negative effects. The distribution of the results differed according to the research design in question. The negative or null findings were concentrated in the studies conducted with IDU community samples, which typically compared users with non-users of syringe exchange. Here positive associations were noted in 12 studies, negative associations in two, and null effects in 11. By contrast, all of the studies conducted with SEP clients (eight out of eight) found positive associations. Similarly, five of the six ecological studies suggested beneficial effects associated with use of syringe exchange. The five modeling studies by Kaplan and his colleagues, conducted with syringes as opposed to clients, also provided evidence of syringe exchange effectiveness. The negative and null findings are both counterintuitive and difficult to interpret. A straightforward explanation offered by Strathdee et al.[10] for their own negative findings is that syringe exchange alone may not be sufficient to prevent spread of HIV in IDU. The investigators studied what they describe as being the largest and most successful syringe exchange in North America, both in terms of the volume of syringes exchanged and the proportion of IDU reached. Nevertheless, despite ready access to sterile syringes, exchange clients continued practicing behavior that put themselves and others at risk of infection, and were more likely than non-clients to seroconvert. Of the 24 seroconverters identified, only two reported difficulty gaining access to syringes, whereas 19 reported no difficulty with regard to access. While we concur that syringe exchange may not alone be sufficient to prevent epidemic spread of HIV among IDU, we speculate that selection and dilution biases may have contributed to the negative and null findings reviewed here. Regarding selection bias, in a retrospective study that tracked IDU in San Francisco over time from before until after syringe exchange was implemented, Hahn et al.[55] found a seroconversion rate of 0.38% per person year among subjects who would never use syringe exchange, compared with 9.34% among those who did later. Schoenbaum et al.[19] obtained similar findings. In their study, begun before implementation of an SEP, IDU who would later use syringe exchange were more likely to report sharing syringes. Additionally, Schechter et al.[18] demonstrated that selection bias alone could account for the higher seroincidence among frequent as opposed to infrequent attenders of SEP, and Hagan et al.[56] found not only that new exchangers were at higher risk than non-exchangers, but also that higher risk IDU were less likely to stop using an SEP over the course of a year. Bourgois and Bruneau [57], seeking to understand reasons for a higher HIV seroconversion rate among Montreal exchange clients than in non-clients, confirmed what other studies [58–62] have shown, namely that cocaine users (who formed the bulk of the sample in the Montreal study [9]) are much more likely than opiate users to engage in high-risk behavior and contract HIV. They also obtained vivid ethnographic evidence that helps to explain these repeated findings. They found that cocaine users inject much more frequently, engage more often in the practice of ‘booting and jacking’ (drawing blood into a syringe to mix with the drug), use larger syringes that hold more blood and drug, attend shooting galleries more frequently, and more frequently engage in sex work and/or maintain close relationships with sex workers [56]. They also speculated that heroin users in Montreal have a more stable lifestyle and are more likely than cocaine users to purchase syringes in pharmacies during daylight hours rather than attend the late-night syringe exchange. In a recent published forum on policy and ethical issues related to syringe exchange [63], Vlahov not only commented on how access to pharmacy syringes may blur differences between users and non-users of syringe exchange, but also cited three studies suggesting that IDU who obtain their syringes from pharmacies are more socially integrated that those who attend syringe exchange; one study provided evidence of the pharmacist's decision to sell syringes or not based on the customer's appearance [64]. Regarding dilution bias, we suspect that an additional factor producing the negative and null findings was availability of other sources of syringes. If this is were indeed the case, contrasts between SEP clients and non-clients with ready access to other sources of syringes would not be very meaningful, and, in fact, could be an artifact of where particular subgroups of IDU happen to obtain syringes. It is probably not a coincidence that the 13 negative and null findings were all from studies conducted in settings, mostly in Canada (including those conducted in Vancouver and Montreal), the UK and the Netherlands, where IDU have access not only to syringe exchange but also to pharmacies where they can purchase syringes legally at low cost. Five of the 28 studies [13,23,24,28–30,32], however, found positive effects of syringe exchange in communities with pharmacy access, showing that such access is not necessarily a study limitation. Nevertheless, among the IDU community studies a cross-tabulation of legal pharmacy access (yes/no) by findings of positive versus negative and null findings of SEP effectiveness shows this relationship to be statistically significant at P = 0.002 (first author's calculation). This would appear to call into question the recent assertion of Moss [65] that the Canadian [9,10,18] and Seattle [17] studies ‘have clearly put a dent in the idea that needle exchanges provide an easy solution to the problem of infectious disease transmission among drug users'. Given the uncertainty about how pharmacy access affected the outcomes of these studies, that assertion may not be warranted. It should be noted, however, that pharmacists may not sell syringes to people they believe to be drug users even if it is legal to do so; in Canada, the sale of syringes is left to the discretion of individual pharmacists [66]. The fact that some SEP may attract higher risk IDU or that IDU may have access to other sources of syringes, however, at least in principle, should not pose insuperable obstacles to empiric investigations of syringe exchange effectiveness. With longitudinal data analyzed properly, differences between IDU who seek versus those who do not seek to use syringe exchange (i.e., selection effects) could be controlled in evaluating the impact of syringe exchange per se. In the absence of randomized trials of syringe exchange (none have yet been published), an ideal study would involve a prospective cohort recruited and interviewed before implementation of syringe exchange, such that IDU who would later go on to use an SEP could be clearly differentiated in terms of risk from those wouldn't. Baseline behavioral data from such a study would provide better data for controlling confounding than was obtained in nearly all the studies reviewed here, most of which used proxies (e.g., gender, age, frequency of injection) for high risk behavior rather than behavior itself. Taking into account other sources of syringes available to IDU is also critical to properly evaluating the impact of SEP. Only one [9] of the 42 studies reviewed here did this. It is likely that failure to obtain measures of the extent of these other sources and/or use them in data analysis contributed to failure to find differences between IDU who did versus IDU who did not use syringe exchange. Whether IDU obtain the syringes from others who attend syringe exchange for them (so-called ‘secondary’ exchange) is obviously another factor which has been neglected, and needs to be measured and analyzed in assessing SEP effectiveness. In closing, it is worth considering several other, less serious problems with research on syringe exchange effectiveness. Biological outcomes (e.g., HIV, HBV, HCV seroconversion) were used in only five studies, despite being generally more valid, if sometimes less sensitive, than self-reported behavior, which is subject to response bias. Interestingly, of the three studies [13,17,24,47] using HBV or HCV seroconversion as outcomes, one found a decreased risk of seroconversion associated with use of SEP, while the other found no association, despite the fact that HBV and HCV are transmitted more readily than HIV, and thus are more sensitive markers of infection risk. However, if IDU do not share syringes but share dissolved drug doses and drug paraphernalia such as cottons and cookers, it would be more difficult to detect differences between users and non-users of an SEP [67]. An additional shortcoming of research on SEP has to do with adjustment for confounders. A close inspection of Tables 1 –7 (see columns labeled ‘statistical controls’ and ‘matching variables') reveals a generally haphazard approach to controlling confounders. In many, if not most, of the studies reviewed here (including longitudinal, observational, and case–control studies) no systematic attempt was made to identify and control for confounders that might account for differences between IDU who did versus those who did not use syringe exchange. By systematic, we mean a careful review of literature on confounders identified in previous studies, and inclusion of these variables, where they are available, in data analysis. In the absence of such an approach, findings of differences (or the lack thereof) could be spurious due to unmeasured or unanalyzed ‘third’ variables. Still another problem was the frequently crude measurement of SEP use. We have noted that secondary use of SEP was not measured in any of these studies. In addition, however, measures such as ‘used SEP once or more in the last 6 months’ (e.g., [9]) were common. In our view, such measures are not epidemiologically sensitive enough to detect the effects of SEP. In conclusion, we believe that the methodologic rigor of evaluations of SEP needs to be improved upon if we are to come to a better understanding of the true impact of syringe exchange.