Factors associated with persistent high-risk syringe sharing in the presence of an established needle exchange programme

Vancouver has experienced an explosive HIV epidemic despite the presence of a needle exchange programme (NEP). We sought possible explanations for high-risk syringe sharing among Vancouver injection drug users over the period January 1999 to October 2000. Overall, 14% of participants reported high-risk sharing. Although acquiring needles exclusively from the NEP was independently associated with less sharing, we identified several risk factors for persistent sharing, including difficulty accessing sterile needles, bingeing, and frequent cocaine injection. Needle exchange programmes (NEP) have been shown to be effective at reducing HIV risk behaviour [1] and HIV transmission among injection drug users (IDU) [2]. Nevertheless, the explosive HIV epidemic observed in Vancouver [3], despite the presence of a well-established NEP, has led to persistent questions about the efficacy of NEP as a public health intervention [4–6]. We have therefore sought possible explanations for high-risk needle sharing in an environment in which much effort has been made to make sterile needles accessible. Data for these analyses were collected through the Vancouver Injection Drug User Study, a prospective cohort study described previously [7]. For the purposes of this study, high-risk needle sharing was defined as borrowing a used needle from someone other than the participant's intimate sexual partner in the previous 6 months. This definition was employed because we sought to examine weaker network linkages that were likely responsible for the rapid spread of HIV observed in the late 1990s [4]. Data from each participant's most recent follow-up during the period January 1999 to October 2000 were evaluated. Univariate and multivariate statistical techniques were applied to determine factors associated with high-risk needle sharing in the previous 6 months. Categorical explanatory variables were analysed using Pearson's chi-square test, and continuous variables were analysed using the Wilcoxon rank sum test. Variables were considered in logistic regression analyses. Drug using characteristics considered in the analyses included: whether participants reported that they currently find it hard to get clean needles, bingeing, the frequency of cocaine and heroin injection, the source of sterile needles, HIV serostatus, and being on methadone maintenance treatment. Individuals who reported injecting cocaine or heroin once or more per day were defined as frequent cocaine and frequent heroin users, respectively. Bingeing was defined as binges or runs when drugs were injected more frequently than usual. In order to evaluate the effect of the NEP, we compared participants whose exclusive needle source was the needle exchange to participants who acquired their needles exclusively from other sources or from several sources, including the NEP, pharmacies, the street, dealers, or friends. In an attempt to account for socially desirable reporting that could potentially increase with the number of encounters with the study research nurses, the multivariate model was adjusted for the number of previous visits of each participant. A total of 997 individuals responded to at least one questionnaire during the period January 1999 to October 2000, and were eligible. Of the 997 participants eligible for the study, 776 (78%) reported injecting drugs in the past 6 months and were therefore included in these analyses. Of these, 671 subjects were defined as non-high-risk needle users and 105 (14%) were defined as high-risk. Univariate analyses found that being married [odds ratio (OR) 0.5], being older (OR 0.97 per year), and living in the Downtown Eastside HIV epicentre (OR 0.5) were inversely associated with high-risk needle sharing. Conversely, male sex (OR 1.7) was positively associated with high-risk sharing. We found no evidence that ethnicity, education, gay or bisexual sexual identity, unstable housing, or employment were associated with high-risk needle sharing. Univariate analyses of drug using characteristics indicated that finding it hard to get clean needles (OR 3.6), bingeing (OR 2.0), and frequent cocaine injection (OR 1.8) were all positively associated with high-risk needle sharing. Conversely, we found that acquiring needles exclusively from needle exchange was associated with less syringe sharing (OR 0.4). We found no evidence that the frequency of heroin use, HIV positivity, or methadone were associated with sharing. Logistic regression analyses, shown in Table 1, found that difficulty accessing clean needles [adjusted odds ratio (AOR) 3.36], bingeing (AOR 1.82), frequent cocaine injection (AOR 1.76), and male sex (AOR 1.89) were all independently associated with high-risk sharing. Variables negatively associated with sharing in the regression analyses included being married (AOR 0.49), and acquiring needles exclusively from needle exchange (AOR 0.46). The model was adjusted for the number of previous visits of each participant, HIV serostatus, and residence in proximity to the needle exchange.Table 1: Logistic regression analysisa of factors associated with high-risk needle sharing during the past 6 months. Given the evidence supporting the effectiveness of NEP as a public health intervention, it is not surprising that we found that individuals who acquire needles exclusively from the NEP were significantly less likely to report syringe sharing [1,2,8–11]. However, we identified several independent risk factors for high-risk needle sharing, the strongest of which was difficulty accessing clean needles. Together, these findings suggest that the needle exchange program has had a strong protective effect against high-risk sharing, but that expanding needle access would probably have a substantial harm-reducing impact. Furthermore, it is noteworthy that bingeing and frequent cocaine injection remained independently associated with high-risk sharing, even after adjustment for residence in proximity to NEP and difficulty accessing needles. These findings suggest that more comprehensive harm reduction strategies may be required to prevent and reduce HIV epidemics in some settings [12,13]. There are several limitations to our analyses. First, in comparison with the participants who were included in these analyses, those participants who were not followed up had a number of characteristics such as being single, younger, and HIV negative, which our analyses suggest make them more likely to be involved in high-risk needle sharing. Second, whereas some studies have shown self-reported HIV risk behaviour to be valid [14], other investigations have suggested that IDU may under-report HIV risk behaviour [15]. Our analyses indicate that Vancouver's NEP has had a strong protective effect against high-risk needle sharing, a finding that was consistent when we considered other definitions of sharing (data not shown). Nevertheless, difficulty accessing sufficient sterile needles, bingeing, and frequent cocaine injection remained strongly associated with this risk behaviour. Whereas improving needle access by expanding the NEP's services may have substantial harm-reducing potential, our findings also suggest that improved needle provision alone, in the absence of additional support for IDU, may not be sufficient to prevent continued HIV transmission. This may be especially important in settings in which cocaine injection is prevalent, and where comprehensive harm reduction measures have yet to be implemented. Evan Wooda,b Mark W. Tyndalla,b Patricia M. Spittala,b Kathy Lia Robert S. Hogga,b Julio S. G. Montanera,b,c Michael V. O'Shaughnessya,d Martin T. Schechtera,b Acknowledgements The authors would like to thank Bonnie Devlin, Caitlin Johnston, Robin Brooks, Suzy Coulter, Steve Kain, Guillermo Fernandez, John Charette, Will Small and Nancy Laliberte for their research and administrative assistance, and all the participants in the VIDUS study.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.