Many urban centres around the world have documented a high prevalence of HIV infection among injection drug users (IDUs).1 In response to this major public health problem, extensive research has been conducted to elucidate the conditions and behaviours that drive HIV epidemics within IDU populations.2 In many settings, the transmission of HIV has been attributed to specific injection practices;3 4 however, several longitudinal studies have also suggested that the sexual transmission of HIV among IDUs is common and potentially responsible for a significant proportion of infections.5 6 Compared with injection-related risk factors, the sexual transmission of HIV among IDUs has historically received much less attention within research, public health and harm reduction communities.5 Subsequently, effective interventions that address sexual risk behaviour within the context of injection drug use have also been slow to develop.6 7

It is increasingly recognised that HIV prevention and treatment programmes for IDUs should include sexual risk reduction strategies.8 In addition to the fact that IDUs are at increased risk for syphilis and other sexually transmitted infections (STIs),9 strong evidence indicates that STIs promote both HIV susceptibility and HIV transmission by increasing HIV infectiousness during acute STI episodes.10 Therefore, the prevention of STIs, including condom promotion programmes for IDUs, is a public health priority.11 Although studies of IDUs have observed increased rates of condom use following the implementation of HIV prevention interventions,12 there exists limited information about potential changes in sexual behaviour associated with the use of needle exchange programmes (NEPs) and supervised injecting facilities (SIFs).

In response to the ongoing HIV epidemic and high rates of fatal overdoses among IDUs in the Downtown Eastside (DTES) area of Vancouver, Canada, North America’s first government-sanctioned SIF opened in September 2003.13 At the SIF, known as Insite, individuals can inject pre-obtained illicit drugs, access clean injecting equipment, receive medical attention in the event of an overdose and be referred to other health services, including addiction treatment.14 Varieties of condoms (for example, lubricated and non-lubricated male and female condoms) are freely available at the facility and the provision of condoms is viewed as an important harm reduction intervention for the clients. Condoms are located adjacent to the sterile equipment and are available prior to or following injection. Condoms are also available at the front desk of the facility; therefore, registered clients do not need to use the injecting room in order to obtain condoms. Nurses who oversee injections provide safer injection education, counsel clients about condom use and safer sex practices, and provide referrals to appropriate services for sexual health and other concerns. Recent studies have documented many public health and community benefits associated with the use of Insite, including reductions in reported syringe sharing,15 positive changes in injecting practices16 and increased detoxification service use.17 We undertook this study to 1) determine the prevalence and correlates of consistent condom use and 2) to examine the longitudinal patterns of condom use among a random sample of IDUs accessing the SIF.

METHODS

The Scientific Evaluation of Supervised Injecting (SEOSI) study is a prospective cohort of IDUs randomly recruited from Insite, Vancouver’s SIF. The study has been described in detail previously.14 Briefly, everyone who used the facility at randomly selected blocks of time during the day were invited to enroll in the study. At baseline and semi-annually, subjects complete a detailed questionnaire administered by SEOSI staff and provide blood samples for HIV and hepatitis C serology. The questionnaire elicits demographic data as well as information about injection-related behaviours, sexual activity, enrollment into addiction treatment and use and satisfaction of the SIF. All participants provide informed consent and receive a stipend of $20 (CDN) for each baseline or follow-up visit.

The primary outcome of interest in this study was condom use during vaginal and/or anal intercourse with regular or casual partners in the past 6 months. All participants who reported heterosexual or same sex activity during one or more interviews were included. Regular partners were defined as “someone you have had a sexual relationship with for more than 3 months”, while casual partners were defined as “someone you have had a sexual relationship with for less than 3 months, including one night stands, but not including tricks or clients”. Participants were asked to report how often a condom was used during intercourse with all regular partners and with all casual partners in the past 6 months. Possible responses included always (100% of the time), usually (>75% of the time), sometimes (26–74% of the time), occasionally (<25% of the time) or never (0% of the time). Each dependent variable was dichotomised into “consistent” (that is, always) and “inconsistent” (that is, usually, sometimes, occasionally, never) condom use. Participants who reported inconsistent condom use during any type of sexual activity (that is, vaginal or anal intercourse with heterosexual or same sex partners) were coded as inconsistent condom users.

Explanatory variables that were considered in these analyses included sociodemographic, social, sexual and drug-related characteristics and behaviours: including age, sex (female vs male), ethnicity (Aboriginal vs other) and sexual orientation (lesbian, gay, bisexual, transgender/transsexual (LGBT) vs heterosexual). Since female-to-female sexual activity is not a risk factor for HIV transmission, women who reported lesbian or bisexual orientation were only included if they also reported a recent male sex partner. Other variables that were assessed included HIV status (positive vs negative based on serological results), homelessness (yes vs no), DTES residency (yes vs no) and sex trade work (yes vs no). Finally, drug-related variables that were considered included current enrollment in any alcohol or drug treatment programme; frequent cocaine, heroin, crack or crystal methamphetamine injection; borrowing syringes; lending syringes; binge drug use; requiring help injecting and injecting (that is, “fixing”) with a regular or casual sex partner. Participants were also asked if they had used medical services at Insite (for example, consult with nurse, HIV testing, referral to other health services) in the last 6 months (yes vs no). To be consistent with our previous work,18 19 we defined “frequent use” to be the daily injection of cocaine, heroin, crystal methamphetamine or the daily smoking of crack, and “binge drug use” to be self-reported periods when drugs were consumed more frequently than usual. All behaviours refer to activities or situations occurring within the 6 months prior to the interview. In order to assess potential changes in patterns of condom use over the study period, “time since recruitment” (per year) was included as a continuous independent variable. In addition to baseline, a total of three biannual follow-ups were used in this analysis, representing 2 years of calendar time.

Initially, we examined the bivariate associations between each independent variable and condom use using generalised estimating equations (GEE). We used GEE for binary outcomes with logit link for the analysis of correlated data since the factors potentially associated with condom use during follow-up were time-dependent measures. The GEE method permits the incorporation of data from every participant who reported a regular or casual partner at baseline or at any of the three follow-ups. For example, individuals who were not sexually active at baseline but reported a regular partner at follow-up would be properly accounted for in the regular partner model. In order to account for potential confounding, we also fit multivariate logistic GEE models adjusting for all variables that were found to be significant at p<0.10 in bivariate analyses. Separate models were constructed for regular and casual partnership data since it is recognised that sexual behaviour and condom use among IDUs are dependent on partner type.20 All statistical analyses were performed using SAS software version 9.1 (SAS, Cary, North Carolina, USA) and all reported p values are two-sided.

RESULTS

A total of 1090 participants completed at least one interview during the period from December 2003 to December 2005. The median number of interviews completed over the study period was 3 (inter-quartile range (IQR) 2–4). The median age at baseline was 38.4 (IQR 32.7–44.3), 317 (29.1%) were women, 211 (19.4%) were of Aboriginal ancestry and 99 (9.1%) were of LGBT orientation. At baseline, 650 (59.6%) reported sexual contact with a regular or casual partner during the past 6 months. Over the study period, a further 144 unique individuals reported sexual activity; therefore, 794 (72.8%) participants were included in these analyses. Of this sub-sample, 579 (72.9%) reported at least one regular partner and 485 (61.2%) reported at least one casual sex partner. At baseline, 427 (59.3%) reported having a regular sex partner; consistent condom use was reported by 108 (25.3%). Of the 333 (46.3%) sexually active participants who had a casual sex partner as baseline, 205 (61.6%) reported consistent condom use. After 2 years of follow-up, consistent condom use increased by 30.0% to 32.9% among those participants reporting regular partners and by 13.3% to 69.8% among participants reporting casual partners (table 1).

The sociodemographic, sexual and drug-related characteristics of eligible participants are presented in table 2. The majority of both groups reported living in the DTES, injecting heroin frequently (that is, more than once a day) and bingeing at least once in the past 6 months. Participants with regular partners were more likely to report injecting with their sex partner, while approximately 40% of both groups reported using the medical services at Insite in the past 6 months.

The results of bivariate GEE analyses of factors associated with consistent condom use with regular sex partners are shown in table 3. LGBT orientation, HIV positivity and time since recruitment were positively associated with consistent condom use, while enrollment in an alcohol or drug treatment programme, borrowing syringes, lending syringes and injecting with a sex partner were inversely associated with this outcome. In multivariate GEE analysis, time since recruitment remained an independent predictor of condom use with regular partners. HIV positivity also remained a significant predictor of consistent condom use, while use of SIF medical services was marginally significant. Factors that remained inversely associated with consistent condom use with regular partners included enrollment in an alcohol or drug treatment programme and injecting with a sex partner.

Factors that were associated with consistent condom use among casual partners in bivariate GEE analyses are shown in table 4. Time since recruitment attained marginal statistical significance. In multivariate analysis, factors that remained independently and significantly associated with consistent condom use with casual partners included HIV positivity, borrowing syringes and lending syringes.

We also examined whether participants reported exchanging sex for money or drugs in the past 6 months and whether a condom was used with all clients. Among 241 participants who reported exchanging sex at baseline, 189 (79.4%) reported consistent condom use during these encounters. There was no significant change in the proportion reporting consistent condom use with clients over 2 years of follow-up (p = 0.213). Finally, we conducted subanalyses to determine whether the subpopulation of participants who reported not using the facility in the past 6 months had similar increases in consistent condom use over time. Among this subsample of SIF non-users who reported a regular sex partner, time since recruitment was found to be non-significant (OR 1.22; 95% CI 0.56 to 2.70 per year; p = 0.615). There was also no significant association between consistent condom use and time since recruitment (OR 0.63; 95% CI 0.11 to 3.77; p = 0.612) among SIF non-users who reported a casual sex partner in the past 6 months.

DISCUSSION

Our results show a high prevalence of inconsistent condom use among participants accessing a SIF. Over 2 years of follow-up, we observed that consistent condom use increased by 30% and 13% among individuals reporting regular or casual sex partners, respectively. In multivariate longitudinal analysis, we found that time since recruitment from within the SIF remained positively and independently associated with consistent condom use among regular partners, even after extensive adjustment for other sex and drug-related predictors. Therefore, we conclude that the SIF has a possible beneficial effect in terms of the consistent use of condoms by IDUs accessing this facility. Since the SIF has been shown to attract a higher risk subpopulation of IDUs compared with the larger IDU population in Vancouver,21 22 it is possible that targeted sexual risk reduction strategies, such as condom distribution at the SIF, may have a greater impact on sexual behaviour compared with interventions within the greater IDU community. This research also supports data from European SIFs that indicate increased condom use reported by clients.23 To our knowledge, this study is the first to examine quantitatively condom use patterns among a prospective cohort of SIF users and provides further evidence that harm reduction strategies for IDUs, including SIFs, should continue to incorporate both sexual and drug-related HIV risk reduction policies and programming.

We also observed several strong associations between injection-related risk behaviour and inconsistent condom use. For example, borrowing and lending syringes were inversely associated with condom use in the casual partner model. This finding is consistent with other studies that have observed significant correlations between syringe sharing and sexual risk behaviours, including not using a condom24^–26 Our finding, that individuals who inject with a regular sex partner were half as likely to report consistent condom use even after adjustment for HIV status, provides further evidence that sexual and injection HIV risks co-occur, particularly within longer-term relationships.

It is concerning that among this representative sample of SIF users, current enrollment in an alcohol or drug treatment programme was inversely associated with consistent condom use with regular partners. Several studies have observed a significant increase in condom use following entry into drug treatment programmes,27 28 while others have failed to observe a beneficial effect.29 30 Although additional research is required to elucidate the potential benefits of drug and alcohol treatment on condom use, these findings suggest that facilities should incorporate or continue to emphasise sexual risk reduction strategies within their programming infrastructures. It is also possible that the observed association between enrollment in addiction treatment and not using a condom is due to a selection effect whereby IDU with higher risk drug-using patterns, and therefore higher risk sexual profiles, are more likely to be enrolled in some form of treatment.

This study has several limitations that should be noted. Since we limited our analysis of condom use to voluntary regular and casual partners, we are unable to generalise our results to other sexual situations in which HIV transmission is possible, including forced sexual encounters. However, since our sample consists of randomly selected SIF users,14 we are confident that our results generalise to all SIF clients who report voluntary sexual activity. A second limitation is that of reporting bias—a tendency that if it exists may result in the under-reporting of socially undesirably activities such as syringe sharing or not using condom. It is possible that individuals who may be less likely to report injection-related HIV risk behaviour may also be less likely to report inconsistent condom use. We have attempted to mitigate this bias by reassuring confidentiality at several stages of the interview. It is also possible that loss to follow-up biased our findings; however, we expect the magnitude of this effect to be small since the follow-up rate for this cohort exceeds 85%. Finally, we did not conduct routine STI testing and thus were unable to validate the derived condom use variable.

In summary, our results demonstrate that among IDUs accessing Vancouver’s SIF, consistent condom use was more frequent with casual sex partners and among those with known HIV positive status. Importantly, while self-reported condom use was low at baseline, it increased over time among study participants. Our results indicate that SIFs and other IDU-targeted programmes, such as NEPs, should continue to distribute condoms and other sex supplies such as lubrication. Furthermore, nurses who currently distribute condoms and offer condom education to clients at the Vancouver SIF may wish to target IDUs who report syringe sharing, injecting with a sex partner and other injection-related risk behaviours. Since statistically significant increases in consistent condom use with casual partners and sex trade clients were not observed, future condom promotion and sexual risk reduction programmes for IDUs should focus on condom use within such relationships. These findings also suggest a need for continued support and expansion of targeted interventions that address both sexual and injection-related risk reduction strategies.