IJE Advance Access originally published online on November 23, 2004
International Journal of Epidemiology 2005 34(1):152-158; doi:10.1093/ije/dyh357
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
IJE vol.34 no.1 © International Epidemiological Association 2004; all rights reserved.
Article |
Baseline self-perceived risk of HIV infection independently predicts the rate of HIV seroconversion in a prospective cohort of injection drug users
1 British Columbia Centre for Excellence in HIV/AIDS, St. Paul's Hospital Vancouver, British Columbia, Canada
2 Department of Medicine, Faculty of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
3 Department of Health Care and Epidemiology, Faculty of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
* Corresponding author. Division of Epidemiology and Population Health BC Centre for Excellence in HIV/AIDS 608-1081 Burrard Street, Vancouver, B.C. V6Z 1Y6, Canada. E-mail: ewood{at}cfenet.ubc.ca
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Background The identification of individuals at the highest risk of human immunodeficiency virus (HIV) infection is critical for targeting prevention strategies. We evaluated self-perceived risk of HIV infection and rates of subsequent HIV seroconversion among a prospective cohort study of injection drug users (IDUs).
Methods We performed an analysis of the time to HIV infection among 994 baseline HIV negative IDUs enrolled in the Vancouver injection drug users study (VIDUS). IDUs were stratified based on their baseline self-perceived risk of HIV seroconversion (higher than others vs same or lower). Kaplan–Meier methods were used to estimate cumulative HIV incidence rates and Cox regression was used to determine adjusted relative hazards for HIV seroconversion.
Results At the end of 24 months after enrolment into the cohort, the cumulative HIV incidence rate was significantly elevated among the 5.9% of the sample who perceived their risk for HIV infection to be higher at baseline (26.6% vs 7.8% log-rank P < 0.001). In a Cox model that adjusted for all variables that were associated with the time to HIV infection in univariate analyses, a higher baseline self-perceived risk of acquiring HIV infection (relative hazard RH: 2.48 [95% Confidence interval (CI): 1.51, 4.10]; P = 0.004) remained independently associated with time to HIV seroconversion.
Conclusions IDUs' perception of their risk for HIV seroconversion upon enrolment into a prospective cohort study was strongly and independently associated with the subsequent rate of HIV seroconversion. Since this risk marker remained independently associated with HIV seroconversion, even after adjustment for time-updated risk behaviours, our findings have major implications that may aid outreach workers in their efforts to identify IDUs who should be targeted with prevention efforts.
Keywords HIV transmission, prevention, injection drug use, HIV/AIDS
Accepted 15 September 2004
Illicit injection drug use is associated with an array of social harms, and major health-related consequences and healthcare expenditures.1–4 For instance, bacterial infections among injection drug users (IDUs) often result in lengthy and expensive acute hospitalizations, and the prevalence of human immunodeficiency virus (HIV) and hepatitis C virus (HCV) infections are estimated to be >30% and >90% among IDUs in our setting respectively.4–6
Cohort studies of IDUs have helped to identify behavioural risk factors for HIV infection,7–10 and have been instrumental in formulation the development of HIV prevention strategies.11,12 Prevention strategies may be educational and/or service oriented, and together aim to improve IDUs' knowledge of behaviours that may place them at a risk of infection and alter risk behaviour while striving to create safer environments, such as improved condom or sterile syringe availability.13–17.
However, as injection drug use may often occur in hidden locations and involve criminalized behaviour, information available to AIDS outreach workers may be very limited and there is strong evidence that it may be subject to socially desirable reporting.18–21 In addition, as has been well-described,22 due to limited prevention resources, it is critical that AIDS prevention education and public health interventions are targeted at the highest risk IDUs. The present analyses were conducted to evaluate whether IDUs' self-perceived risk of HIV infection upon enrolment into a prospective cohort study predicted the subsequent rate of HIV seroconversion.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Participant recruitment
The Vancouver Injection Drug User Study (VIDUS) is a prospective study of IDUs who have been recruited through self-referral and street outreach from Vancouver's Downtown Eastside since May 1996. The cohort has been described in detail previously.6,7,23 Briefly, persons were eligible for the VIDUS study if they had injected illicit drugs at least once in the previous month and resided in the greater Vancouver region. Semi-annually and at baseline, subjects provided venous blood samples and completed an interviewer-administered questionnaire. The questionnaire elicits demographic data as well as information about drug use, HIV risk behaviour, and the use of drug treatment. All participants provided informed consent, and participants were given a stipend ($20 CDN) at each study visit. The study has been approved by the University of British Columbia's Research Ethics Board. The present analyses are restricted to those participants who were recruited between May 1996 and May 2002.
Variables of interest
The primary outcome of interest in this analysis was the time to HIV seroconversion, and the primary risk marker of interest was the participant's self-perceived risk of HIV infection at baseline. The question that assesses self-perceived HIV risk reads: ‘Compared to other drug users in Vancouver, how likely do you think you are to get infected with HIV/AIDS?’ Participants that reported that they were ‘much more likely’ or ‘a bit more likely’ to become infected with HIV were compared with participants who reported that they were ‘about the same’ or ‘less likely’ to become infected with HIV. This question was placed in the nurses questionnaire that is completely separate from questions related to unsafe sex, syringe sharing, and questionnaire items that relate to the risk of HIV, and it is noteworthy that the main questionnaire is administered by trained interviewers rather than nursing staff. As a preliminary analysis, potential associations between relevant baseline characteristics and reporting having a higher baseline perceived risk of HIV infection were explored. Pearson's chi-square test was used to compare categorical explanatory variables, and continuous variables were analyzed using the Wilcoxon rank-sum test. Odds ratios (ORs) for continuous variables were calculated by fitting unadjusted logistic regression models. Demographic variables of interest included gender, age, and ethnicity (Aboriginal vs other). Drug using characteristics considered included: current methadone use, the frequency of cocaine and heroin injection, syringe borrowing, requiring help injecting, and binge use of drugs. Sexual risk was assessed by evaluating if condoms were used exclusively. Since sexual risks may vary within the context of intimate relationships, we evaluated condom use with casual and regular partners separately.24 As done previously, regular sexual partners were defined as those partners with whom the sexual relationship lasted for more than three months.24 Casual sexual partners were defined as those partners with whom the sexual relationship lasted less than three months (including sex-trade). All behavioral variables referred to activities in the preceding six months.
Statistical analyses
For the analysis of the time to HIV infection, the date of seroconversion was estimated using the midpoint between the last negative and the first positive antibody test result. Analyses were restricted to participants who were HIV negative at enrolment, and had at least one follow-up visit. Cumulative HIV incidence rates were calculated for the cohort stratified according to participants who at baseline reported a higher self-perceived risk of HIV infection as compared with other injection drug users in the city and also those participants who perceived their risk of HIV infection to be the same or lower using Kaplan–Meier methods. Survival curves were compared using the log-rank test. In these analyses, time zero was defined as the date of enrolment into the study. Participants who consistently remained seronegative were right censored at the time of their most recent HIV antibody test result prior to May 2003.
We hypothesized that having a higher self-perceived risk of HIV infection would be a marker for elevated HIV risk behavior and potentially the rate of HIV seroconversion. We recognized that if the measure were only a crude marker of other risk factors, it would nonetheless be of value to AIDS outreach workers as a means of evaluating if an individual was at an increased risk of infection. However, in light of studies that have demonstrated that IDUs' self-reports may be subject to socially desirable responding,19,20 we further hypothesized that IDUs may be more willing to report believing their HIV risk was higher, than to report undesirable behaviours, such as syringe sharing. These hypotheses were tested by performing unadjusted and multivariate analyses to examine how the significance of having a high self-perceived risk of HIV infection at baseline, was modified by adjustment for traditional risk factors including those that have and have not been previously associated with HIV seroconversion in the VIDUS cohort. Here, we used Cox regression to examine how the unadjusted relative hazard (RH) for HIV seroconversion was modified by adjustment for known risk factors. We applied an a priori defined statistical protocol that examined the independent effect of having a high self-perceived HIV risk by fitting a statistical model that included all variables that were found to be significantly associated with time to HIV seroconversion at the P 
0.1 level in univariate analyses. In the multivariate analysis, all behavioural variables were treated as time-updated covariates, so that changes in behaviour that were observed at each semi-annual follow-up visit could be incorporated into the statistical model. Alternatively, since we were interested in baseline self-perceived risk, this covariate was retained as a baseline measure only. All P-values were two-sided.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Between May 1996 and May 2003, 1548 IDUs were recruited into the VIDUS cohort, of whom 1332 (86.0%) had at least one follow-up visit. Overall, 319 (23.9%) participants were HIV positive at baseline. Notably, among the 310 (23.3%) participants who were unaware of their HIV status at baseline (263 HIV negative and 47 HIV positive), there was a positive statistical association between baseline HIV positivity and having a higher self-perceived HIV risk (OR: 2.81 [95% CI: 1.01–7.82]; P = 0.039). There were 19 HIV negative individuals who reported being HIV infected at baseline and who did not complete the questions on self-perceived risk, leaving 994 baseline HIV-negative participants to be considered in the analyses of time to HIV seroconversion. Overall among this population the median age was 34.6 years, 342 (34.4%) were women, and 246 (24.8%) were Aboriginal. The median number of follow-up visits was eight. At baseline, 59 (5.9%) individuals reported having a higher self-perceived risk of becoming HIV infected.
The univariate analysis of baseline characteristics of study participants stratified by their self-perceived HIV risk is shown in Table 1. As shown here, having a higher self-perceived HIV risk was associated with being more likely to report daily cocaine injection (OR = 1.8 [95% CI = 1.0–3.0]; P = 0.035) and to report borrowing syringes (OR = 2.4 [95% CI = 1.4–4.0]; P = 0.001). With the exception of female gender (P = 0.058) and Aboriginal ethnicity (P = 0.093) which tended to be associated with a higher self-perceived risk of acquiring HIV, none of the other variables tested, including age, methadone use, frequency of heroin injection, requiring help injection, binge drug use, and condom use, were statistically associated with a higher self-perceived risk of acquiring HIV infection. As noted above, all of the behavioural data referred to risk behaviours in the six-month period preceding the interview.
|
As of May 2003, 123 of the 994 participants had become HIV-infected, yielding a cumulative incidence rate of HIV infection of 3.2% (95% CI: 2.6–3.8) per 100 person-years. As shown in Figure 1, in Kaplan–Meier analyses the cumulative HIV-incidence rate was significantly elevated among those who reported a higher self-perceived risk of HIV infection (log-rank P < 0.001). Overall, among the 935 individuals who perceived their risk of HIV infection to be the same or lower 105 (11.2%) participants seroconverted, whereas among the 59 individuals who reported that their perceived risk of HIV infection was higher, 18 (30.5%) participants seroconverted during the study period. As shown in Figure 1, at the end of 24 months after enrolment into the cohort, the cumulative HIV-incidence rate was 26.6% among those who perceived their HIV risk to be higher at baseline and was 7.8% among those who perceived their risk to be the same or lower at baseline. We noted that a substantial proportion of the seroconversions occurred in the first six months of follow-up and we performed a sub-analysis where we excluded all individuals who seroconverted in the first six months and used the six month follow-up visit as baseline. Here, the results were identical with a statistically significant difference in the time to seroconversion persisting (log-rank P < 0.001) among the 931 participants who were eligible for this sub-analysis.
|
We then investigated if there was a gradient of risk that increased between participants who reported that they thought they were ‘a bit more likely’ and participants who reported that they were ‘much more likely’ to become infected with HIV. Here, we performed a Cox regression analysis where we stratified responses to this question and created indicator variables such that reporting that the self-perceived risk of HIV was about the same or less, was the reference category. Here we found that the risk of HIV for participants who reported a ‘much higher’ risk had the highest risk of HIV (RH: 4.72 [95% CI: 2.65–8.41]; P < 0.001), where as reporting a ‘bit higher’ risk remained associated with elevated risk of HIV seroconversion (RH: 2.00 [95% CI: 0.88–4.56]; P = 0.098), in comparison with the reference category.
Table 2 presents the results of unadjusted and adjusted Cox proportional hazards regression analyses of sociodemographic variables, self-perceived HIV risk, substance use and sexual risk behaviours, and their association with time to HIV seroconversion. As shown here, in unadjusted analyses, female gender (RH: 1.51 [95% CI: 1.05–2.15]), Aboriginal ethnicity (RH: 1.89 [95% CI: 1.31–2.72]), daily cocaine injection (RH: 3.28 [95% CI: 2.28–4.72]), daily heroin injection (RH: 1.38 [95% CI: 0.97–1.98]), syringe borrowing (RH: 1.69 [95% CI: 1.15–2.50]), requiring help injecting (RH: 2.21 [95% CI: 1.53–3.20]), binge drug use (RH: 1.98 [95% CI: 1.38–2.85]) and a higher baseline self-perceived risk of acquiring HIV infection (RH: 3.31 [95% CI: 2.02–5.39]) were positively associated with the time to HIV infection at the P 0.1 level. Use of methadone maintenance therapy was associated with a decreased risk of HIV seroconversion (RH: 0.67 [95% CI: 0.42–1.10]). We tested the coefficients for time-dependence to determine if the proportional hazards assumption of the Cox model was violated and found no evidence of this concern for the self-perceived risk variable (P = 0.838).
|
In a multivariate Cox model that adjusted for all variables that were associated with the time to HIV infection in univariate analyses, Aboriginal ethnicity (RH: 1.69 [95% CI: 1.14–2.50]; P = 0.009), daily cocaine injection (RH: 2.52 [95% CI: 1.72–3.70]), requiring help injecting (RH: 1.57 [95% CI: 1.06–2.33]), binge drug use (RH: 1.52 [1.05–2.20]) and a higher baseline self-perceived risk of acquiring HIV infection (RH: 2.48 [95% CI: 1.51–4.10]) remained independently associated with time to HIV seroconversion. These findings remained unchanged when we adjusted for additional non-significant variables, such as age (data not shown). It is important to note that the low prevalence of high self-perceived risk at baseline (5.9%) and the attributable risk percentage (defined as RH-1/RH) indicate that the population attributable risk of acquiring HIV in this sample is 3.5% due to the low prevalence of this risk marker.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
In the present study, we found that baseline HIV prevalence was associated with higher self-perceived baseline HIV risk among persons who were unaware of their HIV status. More importantly, in prospective analyses we found markedly elevated HIV incidence rates among persons who reported having a higher self-perceived risk of HIV infection at baseline. Interestingly this risk marker remained independently associated with an elevated rate of HIV incidence in multivariate analyses even after adjustment for time updated HIV risk behaviours and drug using characteristics.
Self-perceived risk of HIV infection is a concept that has been considered for a number of years.25 One concern raised by the present study is that, although self-perceived risk of HIV infection predicted subsequent HIV seroconversion, since behaviours that cause HIV were under-reported, education workers may face challenges with identifying which behaviours put IDUs at risk. One study by Celentano and colleagues reported that HIV knowledge was not associated with HIV status, although the study was cross-sectional and it was not known how long HIV positive participants had been infected.26 Other studies have noted the challenges faced by HIV educators,27 although it is generally accepted that HIV prevention education is highly effective.11 Nevertheless, the degree of under-reporting of risk behaviours observed in the present study supports the notion that individual-level educational messaging should be coupled with network-level interventions.28
Interpreting our findings requires an understanding of both residual confounding and socially desirable responding to questionnaire items in observational HIV incidence studies of IDUs. With regard to residual confounding, it is important to note that measurement of risk factors is imprecise, although it is noteworthy that the apparent effect of a high self-perceived HIV risk was not substantially reduced after adjustment for other measured covariates. These covariates included all variables that have ever previously been demonstrated to be associated with HIV infection in any of the previous risk factor analyses that have been conducted using the VIDUS cohort,5,7,23,24,29 and the relative hazard for having a higher self-perceived risk was modified by only 
25%.
This point brings up the issue of socially desirable reporting among IDUs enrolled in prospective cohort studies such as VIDUS. Specifically, previous studies have demonstrated that IDUs may under-report socially undesirable behaviours, such as syringe sharing,19,20 and may over-report socially desirable behaviours, such as needle exchange attendance.30 Since only a fraction of the participants who seroconverted in the present study reported high self-perceived risk of HIV, it is likely that this variable also suffers from this concern. Imprecise variable measurement and socially desirable reporting are evident in all studies of IDUs, including the present study, since if syringe sharing with HIV infected individuals was perfectly measured (and blood transfusions and other HIV risk behaviours were not relevant) traditional risk factors for HIV among IDUs, such as shooting gallery attendance and cocaine injection, would not be independently associated with HIV seroconversion after statistical adjustment for the direct mode of HIV transmission (ie sexual risk or in our setting predominantly syringe sharing),7 and if the HIV status of the distributive sharing partner were known. Nevertheless, the present study indicates that there is substantial measurement error in the cohort, which has implications for previous findings.23
In light of the above, our findings have major implications for HIV prevention service providers and for AIDS outreach workers. First, since knowledgeable experts in this area have stressed the importance of targeting interventions towards the highest risk IDUs,22 the present study demonstrates that IDUs' perceptions of their HIV risk may be strongly associated with their present HIV status and their future risk of acquiring HIV. As such, IDUs who report that they believe their risk of acquiring HIV is higher than other IDUs should be considered for targeted HIV prevention efforts. Second, since IDUs may be reluctant to accurately report HIV risk behaviours, such as syringe sharing,19 the present study demonstrates that eliciting IDUs' self-perceived risk of HIV infection may aid public health outreach workers' efforts to gauge an IDUs' HIV risk and service needs.
This study has several limitations. First, although previous studies have indicated that the study sample is highly representative of Vancouver IDUs, VIDUS is not a random sample. Second, as outlined above, it has been shown that audio-computer assisted self-interviewing (ACASI) may be superior to interviewer administered questionnaires with regard to eliciting HIV risk behaviours, such as syringe sharing.19,20 However, with regard to the present study's implications for informing HIV risk assessment strategies for outreach workers, the use of trained interviewers and research nurses may actually be superior to ACASI since this interface may more closely replicate the interactions between IDUs and outreach workers. While incorporating ACASI into outreach efforts would probably improve the assessment of risk behaviours, it is unlikely that these methods will be used in shooting galleries and other hidden environments where AIDS outreach workers commonly interface with IDUs.
In the present study, we found that IDUs' self-perceived risk of HIV infection was strongly associated with baseline HIV prevalence and subsequent HIV incidence. At the end of 2 years after enrolment into the study, IDUs who perceived their risk of HIV to be higher than other IDUs in the city, had a rate of HIV infection that was over three times higher than IDUs who perceived their risk of HIV infection to be the same or lower. In adjusted analyses that considered time-updated drug use characteristics and risk behaviours, the relative hazard for HIV infection remained approximately 2.5 times higher among IDUs who had a higher self-perceived risk of HIV infection at baseline. Since the measure provided prognostic information that was independent of self-reported risk behaviours, it may be that evaluating IDUs' perceptions of their own risk of HIV infection may prove to be highly valuable with regard to targeting IDUs at highest risk and providing them with educational and HIV prevention services.
KEY MESSAGES
|
|
Acknowledgments |
|---|
We would particularly like to thank the VIDUS participants for their willingness to participate in the study. We also thank Dr David Patrick, Dr Will Small, Dr Patricia Spittal, and Dr Steffanie Strathdee, for their research assistance, and Bonnie Devlin, John Charette, Caitlin Johnston, Vanessa Volkommer, Steve Kain, Dave Isham, and Peter Vann for administrative assistance. This study would not have been possible without the financial support of Status of Women Canada. The study was further supported by the US National Institutes of Health (R01 DA011591–04A1) and CIHR grant (MOP-67262). Evan Wood is supported by the Canadian Institutes of Health Research and the Michael Smith Foundation for Health Research.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
1 Public policy statement on national drug policy. American Society of Addiction Medicine. J Addict Dis 1994; 13:231–34.[CrossRef][ISI][Medline]
2 Wall R, Rehm J, Fischer B et al. Social costs of untreated opioid dependence. J Urban Health 2000; 77:688–722.[CrossRef][ISI][Medline]
3 Palepu A, Tyndall MW, Leon H et al. Hospital utilization and costs in a cohort of injection drug users. CMAJ 2001; 165:415–20.
4 Wood E, Kerr T, Spittal PM, Tyndall MW, O'Shaughnessy MV, Schechter MT. The healthcare and fiscal costs of the illicit drug use epidemic: the impact of conventional drug control strategies and the impact of a comprehensive approach. BC Med J 2003; 45:130–36.
5 Craib KJ, Spittal PM, Wood E et al. Risk factors for elevated HIV incidence among Aboriginal injection drug users in Vancouver. CMAJ 2003; 168:19–24.
6 Miller CL, Johnston C, Spittal PM et al. Opportunities for prevention: hepatitis C prevalence and incidence in a cohort of young injection drug users. Hepatology 2002; 36:737–42.[CrossRef][ISI][Medline]
7 Tyndall MW, Currie S, Spittal P et al. Intensive injection cocaine use as the primary risk factor in the Vancouver HIV-1 epidemic. AIDS 2003; 17:887–93.[CrossRef][ISI][Medline]
8 Wood E, Tyndall MW, Spittal PM et al. Unsafe injection practices in a cohort of injection drug users in Vancouver: could safer injecting rooms help? CMAJ 2001; 165:405–10.
9 Broers B, Junet C, Bourquin M, Deglon JJ, Perrin L, Hirschel B. Prevalence and incidence rate of HIV, hepatitis B and C among drug users on methadone maintenance treatment in Geneva between 1988 and 1995. AIDS 1998; 12:2059–66.[ISI][Medline]
10 Strathdee SA, Galai N, Safaeian M et al. Sex differences in risk factors for HIV seroconversion among injection drug users: a ten year perspective. Arch Intern Med 2001; 161:1281–88.
11 Des Jarlais DC, Hagan H, Friedman SR et al. Maintaining low HIV seroprevalence in populations of injecting drug users. JAMA 1995; 274:1226–31.[Abstract]
12 Des Jarlais DC, Friedman SR. Fifteen years of research on preventing HIV infection among injecting drug users: what we have learned, what we have not learned, what we have done, what we have not done. Public Health Rep 1998; 113:(Suppl.1):182–88.[ISI][Medline]
13 Wood E, Tyndall MW, Spittal P et al. Needle exchange and difficulty with needle access during an ongoing HIV epidemic. Int J Drug Policy 2002; 13:95–102.
14 Wood E, Tyndall MW, Spittal PM et al. Factors associated with persistent high-risk syringe sharing in the presence of an established needle exchange programme. AIDS 2002; 16:941–43.[CrossRef][ISI][Medline]
15 Des Jarlais DC, Marmor M, Paone D et al. HIV incidence among injecting drug users in New York City syringe-exchange programmes. Lancet 1996; 348:987–91.[CrossRef][ISI][Medline]
16 Des Jarlais DC, Friedman SR, Friedmann P et al. HIV/AIDS-related behavior change among injecting drug users in different national settings. AIDS 1995; 9:611–17.[ISI][Medline]
17 Des Jarlais DC, Choopanya K, Vanichseni S et al. AIDS risk reduction and reduced HIV seroconversion among injection drug users in Bangkok. Am J Public Health 1994; 84:452–5.
18 Grund JP, Blanken P, Adriaans NF, Kaplan CD, Barendregt C, Meeuwsen M. Reaching the unreached: targeting hidden IDU populations with clean needles via known user groups. J Psychoactive Drugs 1992; 24:41–47.[ISI][Medline]
19 Des Jarlais DC, Paone D, Milliken J et al. Audio-computer interviewing to measure risk behaviour for HIV among injecting drug users: a quasi-randomised trial. Lancet 1999; 353:1657–61.[CrossRef][ISI][Medline]
20 Macalino GE, Celentano DD, Latkin C, Strathdee SA, Vlahov D. Risk behaviors by audio computer-assisted self-interviews among HIV-seropositive and HIV-seronegative injection drug users. AIDS Educ Prev 2002; 14:367–78.[CrossRef][ISI][Medline]
21 van Griensven F, Supawitkul S, Kilmarx PH et al. Rapid assessment of sexual behavior, drug use, human immunodeficiency virus, and sexually transmitted diseases in northern thai youth using audio-computer-assisted self-interviewing and noninvasive specimen collection. Pediatrics 2001; 108:E13.[CrossRef][Medline]
22 Des Jarlais DC, Padian NS, Winkelstein W Jr. Targeted HIV-prevention programs. N Engl J Med 1994; 331:1451–53.
23 Strathdee SA, Patrick DM, Currie SL et al. Needle exchange is not enough: lessons from the Vancouver injecting drug use study. AIDS 1997; 11:F59–65.[CrossRef][ISI][Medline]
24 Spittal PM, Craib KJ, Wood E et al. Risk factors for elevated HIV incidence rates among female injection drug users in Vancouver. CMAJ 2002; 166:894–99.
25 Latkin CA, Vlahov D. Socially desirable response tendency as a correlate of accuracy of self-reported HIV serostatus for HIV seropositive injection drug users. Addiction 1998; 93:1191–97.[CrossRef][ISI][Medline]
26 Celentano D, Vlahov D, Menon A, Polk F. HIV knowledge and attitudes among intravenous drug users: comparisons to the U.S. population and by drug use behaviors. J Drug Issues 1991; 21:635.
27 Booth RE, Watters JK. How effective are risk-reduction interventions targeting injecting drug users? AIDS 1994; 8:1515–24.[ISI][Medline]
28 Latkin CA, Mandell W, Vlahov D, Oziemkowska M, Celentano DD. The long-term outcome of a personal network-oriented HIV prevention intervention for injection drug users: the SAFE Study. Am J Community Psychol 1996; 24:341–64.[CrossRef][ISI][Medline]
29 Miller CL, Spittal PM, LaLiberte N et al. Females experiencing sexual and drug vulnerabilities are at elevated risk for HIV infection among youth who use injection drugs. J Acquir Immune Defic Syndr 2002; 30:335–41.[ISI][Medline]
30 Safaeian M, Brookmeyer R, Vlahov D, Latkin C, Marx M, Strathdee SA. Validity of self-reported needle exchange attendance among injection drug users: implications for program evaluation. Am J Epidemiol 2002; 155:169–75.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. A Strathdee Commentary: Targeting HIV interventions to the most at risk--hindsight is 20/20 Int. J. Epidemiol., February 1, 2005; 34(1): 158 - 159. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||