ralssBioMed CentBMC Public HealthOpen AcceResearch articleStructural factors associated with an increased risk of HIV and sexually transmitted infection transmission among street-involved youthBrandon DL Marshall1,2, Thomas Kerr1,3, Jean A Shoveller2, Julio SG Montaner1,3 and Evan Wood*1,3Address: 1British Columbia Centre for Excellence in HIV/AIDS, St. Paul's Hospital, 608 – 1081 Burrard Street, Vancouver, BC, V6Z 1Y6, Canada , 2School of Population and Public Health, University of British Columbia, 5804 Fairview Avenue, Vancouver, BC, V6T 1Z3, Canada and 3Department of Medicine, University of British Columbia, 608 – 1081 Burrard Street, Vancouver, BC, V6Z 1Y6, CanadaEmail: Brandon DL Marshall - bmarshall@cfenet.ubc.ca; Thomas Kerr - uhri-tk@cfenet.ubc.ca; Jean A Shoveller - jean.shoveller@ubc.ca; Julio SG Montaner - jmontaner@cfenet.ubc.ca; Evan Wood* - uhri-ew@cfenet.ubc.ca* Corresponding author AbstractBackground: The prevalence of HIV and sexually transmitted infections (STIs) among street-involved youth greatly exceed that of the general adolescent population; however, little is knownregarding the structural factors that influence disease transmission risk among this population.Methods: Between September 2005 and October 2006, 529 street-involved youth were enroledin a prospective cohort known as the At Risk Youth Study (ARYS). We examined structural factorsassociated with number of sex partners using quasi-Poisson regression and consistent condom useusing logistic regression.Results: At baseline, 415 (78.4%) were sexually active, of whom 253 (61.0%) reported multiple sexpartners and 288 (69.4%) reported inconsistent condom use in the past six months. In multivariateanalysis, self-reported barriers to health services were inversely associated with consistent condomuse (adjusted odds ratio [aOR] = 0.52, 95%CI: 0.25 – 1.07). Structural factors that were associatedwith greater numbers of sex partners included homelessness (adjusted incidence rate ratio [aIRR]= 1.54, 95%CI: 1.11 – 2.14) and having an area restriction that affects access to services (aIRR =2.32, 95%CI: 1.28 – 4.18). Being searched or detained by the police was significant for males (aIRR= 1.36, 95%CI: 1.02 – 1.81).Conclusion: Although limited by its cross-sectional design, our study found several structuralfactors amenable to policy-level interventions independently associated with sexual riskbehaviours. These findings imply that the criminalization and displacement of street-involved youthmay increase the likelihood that youth will engage in sexual risk behaviours and exacerbate thenegative impact of resultant health outcomes. Moreover, our findings indicate that environmental-structural interventions may help to reduce the burden of these diseases among street youth inurban settings.Published: 9 January 2009BMC Public Health 2009, 9:7 doi:10.1186/1471-2458-9-7Received: 28 August 2008Accepted: 9 January 2009This article is available from: http://www.biomedcentral.com/1471-2458/9/7© 2009 Marshall et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 9(page number not for citation purposes)BMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7BackgroundStructural factors, defined as the economic, social, policy,and organizational environments that "structure" the con-text in which risk production occurs [1], are increasinglyrecognised as important determinants in the acquisition,transmission, and prevalence of HIV disease [2]. In recentyears, extensive research has examined the structural fac-tors that produce and re-produce HIV risk among highprevalence populations, including injection drug users(IDU) and sex workers [3,4]. Homeless and street-involved adolescents have also been recognised as a mar-ginalised population with unique exposures to structuralenvironments that increase the likelihood of sustainedand elevated disease burden; however, these factorsremain poorly understood [5].In Canada and the United States, it is estimated thatbetween 4 and 7 percent of youth between the ages of 14and 26 are absolutely, periodically, or temporarily with-out access to safe and stable shelter [6,7]. Homeless andstreet-involved youth are known to be at a significantlyincreased risk for a wide range of adverse health outcomes[8]. Of considerable public health concern is the highprevalence of HIV and sexually transmitted infections(STIs) among these populations. In urban centres in Can-ada, the prevalence of HIV among street-involved youth isapproximately 2 percent [9,10], while the prevalence ofChlamydia has been estimated to be between 7 and 11percent [11,12]. Similar rates have been observed in theUnited States [13,14].Street-involved youth engage in a greater number of sex-ual risk behaviours than their non-homeless peers [15].The vast majority is sexually active, and among those whodo engage in sexual intercourse, inconsistent condom useis common [16,17]. Street-involved youth are also morelikely to have multiple and concurrent sex partners[18,19]. Of further concern is that approximately onequarter of street youth have engaged in survival sex (i.e.,sex in exchange for money, shelter, food or drugs) [20].Among youth who are coerced or manipulated into sur-vival sex, sexual victimization and abuse are common[21].Research that has attempted to elucidate the underlyingreasons for increased engagement in sexual risk behaviouramong street-involved youth has continued to rely pre-dominantly on individual level risk factor analyses [22].However, a growing body of literature has demonstratedthat a focus on individual level characteristics (e.g., child-hood abuse, depression, knowledge) fails to acknowledgethe social structural factors that shape and determine thecontext in which sexual risk behaviour takes place [3,23].systemic discrimination, are better overall predictors ofpopulation level HIV and STI prevalence [24]. Given thesemethodological challenges and concerns, we sought todetermine whether structural factors are associated withincreased engagement in sexual risk behaviour among acommunity-recruited cohort of street-involved youth.MethodsThe At Risk Youth Study (ARYS) is a prospective cohort ofhomeless and street-involved youth in Vancouver, Can-ada that has been described in detail previously [25].Briefly, participants were recruited through snowball sam-pling and extensive street-based outreach. Persons wereeligible for the study if they were 14 to 26 years of age, hadused illicit drugs other than or in addition to marijuana inthe past 30 days, and provided informed consent. At base-line and semi-annually, participants complete an inter-viewer-administered questionnaire and provide bloodsamples for HIV and hepatitis C (HCV) serology. Thequestionnaire elicits demographic data and informationregarding injection and non-injection drug use, HIV riskbehaviours, addiction treatment experience, encounterswith police and security guards, health service utilization,and sexual activity. All participants receive a monetary sti-pend of $20 CDN after each visit. The study has beenapproved by the University of British Columbia/Provi-dence Health Care Research Ethics Board.All participants who completed a baseline survey betweenSeptember 2005 and October 2006 were included in thisstudy. Since data from just one follow-up period wasavailable at the time of study conception, only informa-tion collected at baseline was included in these analyses.We examined as our primary outcomes two sexual riskbehaviours that together play key roles in determining thesexual transmission of HIV and STIs: 1) number of sexualpartners, and 2) condom use during vaginal and analintercourse. Participants were asked to report how manydifferent male and female partners they had engaged insexual activities with in the past 6 months, excludingthose with whom they had engaged in sex for money,shelter, food, or drugs (i.e., sex trade work). Specifically,the total number of partners was obtained by addingresponses to the questions: "Could you give me a precisenumber of male/female partners you had in the past 6months?". Participants could report any set of positiveinteger values; thus, the variable was coded as continuousin bivariate and multivariate analyses. The resulting distri-bution was positively skewed, with a median of 1.0 (inter-quartile range: 0–3), a mean of 3.2 (standard deviation:5.6), and a range of 0–55. For both same and opposite sexpartnerships, participants were also asked to report howoften a condom was used during vaginal and/or analPage 2 of 9(page number not for citation purposes)Furthermore, it is increasingly recognised that structuralfactors, including economic inequities, laws, policies, andintercourse. Possible responses included: always (100%),regularly (50% to 99%), occasionally (1% to 49%), andBMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7never (0%). To be consistent with previous studies of con-dom use among street-involved youth populations [26],this variable was dichotomised into "consistent" (i.e.,always) and "inconsistent" (i.e., regularly, occasionally, ornever) condom use. Participants who reported more thanone type of sexual activity and who reported discordantcondom use patterns were coded as inconsistent condomusers.The primary variables of interest in this study were a set ofvariables addressing structural factors that were hypothe-sised to shape the context in which street youth sexual riskbehaviour is produced. We defined: "homeless" as anyparticipant who reported being homeless in the past sixmonths; "barriers to health or harm reduction services" asbeing in need of but unable to obtain health or harmreduction services (e.g., doctor, nurse, clinic, dentist,optometrist, or needle exchange); "jacked up" as beingstopped, searched or detained by the police; "warrants" ascurrently having a warrant or area restriction that affectsaccess to needle exchange programs (NEP) or other serv-ices; "unable to access treatment" as trying to access analcohol or drug treatment program but being unable to;and "assault from police/security guards" as experiencinga physical interaction with police or security guards result-ing in bruises, scratches, etc. All variables except for "war-rants" refer to behaviours and events occurring in the pastsix months since the date of the interview. Other inde-pendent variables included a broad range of sociodemo-graphic, individual level, drug-related, and social factors,chosen based on their known or a priori status as risk fac-tors for one or both sexual behaviour outcomes. Sociode-mographic variables that were examined included: age,sex (female vs. male), Aboriginal ethnicity (yes vs. no) andsexual orientation (lesbian, gay bisexual, transgendered/transsexual [LGBTT] vs. heterosexual). Other individuallevel factors that were examined included: engaging inanal intercourse in the past six months, depression(defined using the Centre for Epidemiologic StudiesDepression [CES-D] scale), and the self-efficacy for limit-ing HIV risk behaviours (LHRB) scale. The CES-D hasbeen shown to have high levels of internal consistencyand reliability among groups of adolescents [27]. Thepresence of depressive symptoms was evaluated using awell-defined cut-off (CES-D ≥ 16 [yes] versus CES-D < 16[no]). The self-efficacy for LHRB scale is a validated instru-ment found to have high levels of consistency among at-risk youth [28]. Responses were dichotomised into "high"versus "low" self-efficacy for LHRB based on the samplemedian. Social and drug-related factors that were exam-ined included: relationship status (single or casually dat-ing vs. regular partner or married), childhood sexualabuse, drug dealing, alcohol dependence, crack use,injection drug use, syringe sharing, and binge drug use(yes vs. no). All drug use variables refer to behavioursoccurring in the past six months and include both injec-tion and non-injection routes of consumption. To be con-sistent with our previous work, "syringe sharing" wasdefined as lending or borrowing a syringe that had beenused by someone else, and "binge drug use" was definedas the self-reported consumption of drugs (injection ornon-injection) more often than usual [29]. Finally, alco-hol dependence was measured using the Perceived-Bene-fit-of-Drinking Scale (PBDS), a validated true/falseinstrument that assesses drinking behaviours among ado-lescents [30].Initially, we examined bivariate associations betweeneach independent variable and each sexual risk behaviouroutcome. Given that the precise number of recent sexualpartners was obtained for each participant, we used a Pois-son-type regression to estimate the unadjusted incidencerate ratio (IRR) and 95% confidence interval (95% CI)associated with each explanatory variable. Since the distri-bution of recent sex partners was highly skewed, we useda log-linear quasi-Poisson regression to account for over-dispersion in the data. To examine the bivariate associa-tions between each independent variable and consistentcondom use, we used the Pearson χ2 test. Fisher's exact testwas used when one or more of the cells contained valuesless than or equal to 5. Since sexual risk behaviour profilesamong street-involved youth are observed to be moder-ated by gender [31,32], we also assessed each structuralvariable for possible interaction with sex. If a statisticallysignificant interaction effect was observed, the coefficientscorresponding to the main and interaction terms werecombined to construct IRR estimates corresponding toeach sex. The overall significance of the main and interac-tion effect was assessed using the likelihood ratio test.Since research among populations of IDU and street-based sex workers has demonstrated that policies andlaws promoting the displacement and criminalization ofmarginalised persons are associated with sexual- andinjection-related HIV risk production [3,4,33], we choseto focus our analysis on structural variables that addressthese issues and thus may potentially shape the produc-tion of sexual risk-taking behaviour among street-involved youth. In order to account for potential con-founding, we used an a priori defined bivariate cut-off of p< 0.10 as the criterion for inclusion of variables into mul-tivariate analyses. Each independent variable wasincluded as a potential explanatory factor when not usedas the primary outcome of interest. All statistical analyseswere conducted using S-PLUS software version 8.0. Allreported p-values are two-sided.Page 3 of 9(page number not for citation purposes)cocaine use, heroin use, crystal methamphetamine use,BMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7ResultsA total of 529 participants completed an interviewbetween September 1, 2005 and October 31, 2006, ofwhom 159 (30.1%) were female, 127 (24.0%) were ofAboriginal ethnicity, and 69 (13.0%) self-identified asLGBTT. The majority, 415 (78.4%), reported engaging involuntary sexual activity in the past six months. Of theseparticipants, 288 (69.4%) reported inconsistent condomuse and 253 (61.0%) reported multiple sex partners. Ofthe entire sample, the median number of sex partners inthe past six months was 1 (interquartile range [IQR]: 1 –3; range: 0 – 55).The results of the bivariate quasi-Poisson analyses areshown in Table 1. Structural variables that were positivelyassociated with number of recent sex partners includedhomelessness (incidence rate ratio [IRR] = 1.87, 95% con-fidence interval [95%CI]: 1.24 – 2.82) and having a war-rant or area restriction that affects access to services (IRR =2.51, 95%CI: 1.21 – 5.18). Statistically significant interac-tion effects were observed for both "jacked up" and "bar-riers to health or harm reduction services" variables. Theformer was positively associated with number of recentsex partners for males (IRR = 1.53, 95%CI: 1.07 – 2.18),while the latter was marginally significant for females(IRR = 1.92, 95%CI: 0.97 – 3.79). Barriers to accessinghealth or harm reduction services (odds ratio [OR] = 0.53,95% CI: 0.28 – 1.00) was the only structural factor associ-ated with consistent condom use in bivariate analysis (seeTable 2). The results of the multivariate analyses model-ling number of recent sex partners and consistent condomuse are shown in Tables 1 and 3, respectively. Homeless-ness (adjusted incidence rate ratio [aIRR] = 1.54, 95% CI:1.11 – 2.14) and having a warrant or area restriction thataffects access to services (aIRR = 2.32, 95% CI: 1.28 –4.18) were positively and independently associated withnumber of recent sex partners. Furthermore, the overallcontributions (main and interaction effect) of both"jacked up" and "barriers to health or harm reductionservices" to the final model were highly significant (p <0.001 for both variables). For males, being jacked up bythe police was positively associated with number of recentsex partners (aIRR = 1.36, 95% CI: 1.02 – 1.81), while bar-riers to health or harm reduction services was marginallyTable 1: Factors associated with number of sex partners among a cohort of street-involved youth (n = 529).Unadjusted Incidence Rate Ratio (IRR) Adjusted Incidence Rate Ratio (aIRR)Characteristic IRR (95% CI) p – value aIRR (95% CI) p – valueAge (per year older) 0.98 (0.93 – 1.04) 0.566Sex (female vs. male) 0.80 (0.57 – 1.13) 0.198Aboriginal ethnicity (yes vs. no) 0.84 (0.58 – 1.22) 0.366Sexual orientation (LGBTTa vs. heterosexual) 1.90 (1.37 – 2.63) < 0.001 1.58 (1.16 – 2.16) 0.004Relationship (single vs. partner) 1.79 (1.19 – 2.69) 0.005 1.44 (1.04 – 2.00) 0.028Depressionb (yes vs. no) 1.14 (0.84 – 1.56) 0.402Self-Efficacy LHRBc (low vs. high) 1.55 (1.14 – 2.13) 0.006 1.41 (1.10 – 1.81) 0.007Condom use† (consistent vs. inconsistent) 0.86 (0.62 – 1.20) 0.380Anal intercourse† (yes vs. no) 2.52 (1.83 – 3.48) < 0.001 2.01 (1.51 – 2.69) < 0.001Sexual abuse‡ (yes vs. no) 1.67 (1.25 – 2.24) < 0.001 1.40 (1.08 – 1.83) 0.011Drug dealing† (yes vs. no) 1.30 (0.95 – 1.77) 0.104Alcohol dependence (yes vs. no) 1.32 (0.97 – 1.79) 0.073 1.05 (0.82 – 1.35) 0.711Crack use† (yes vs. no) 1.45 (1.07 – 1.98) 0.018 1.20 (0.88 – 1.64) 0.249Cocaine use† (yes vs. no) 1.62 (1.20 – 2.19) 0.002 1.63 (1.28 – 2.08) < 0.001Heroin use† (yes vs. no) 0.99 (0.72 – 1.37) 0.952Crystal meth use† (yes vs. no) 1.07 (0.79 – 1.44) 0.671Injection drug use† (yes vs. no) 1.01 (0.73 – 1.40) 0.944Sharing syringes† (yes vs. no) 1.17 (0.72 – 1.91) 0.521Binge drug use† (yes vs. no) 1.35 (1.00 – 1.83) 0.047 0.94 (0.71 – 1.26) 0.681Homelessness† (yes vs. no) 1.87 (1.24 – 2.82) 0.003 1.54 (1.11 – 2.14) 0.011Barriers to health/HRd services† (yes vs. no) < 0.001* < 0.001*Male 0.97 (0.62 – 1.51) 0.889 0.82 (0.57 – 1.16) 0.259Female 1.92 (0.97 – 3.79) 0.061 1.76 (0.98 – 3.15) 0.058Jacked up† (yes vs. no) < 0.001* < 0.001*Male 1.53 (1.07 – 2.18) 0.020 1.36 (1.02 – 1.81) 0.034Female 1.15 (0.62 – 2.10) 0.661 0.85 (0.51 – 1.41) 0.526Warrants (yes vs. no) 2.51 (1.21 – 5.18) 0.007 2.32 (1.28 – 4.18) 0.005Unable to access treatment† (yes vs. no) 1.14 (0.74 – 1.78) 0.545Assault from police/guards† (yes vs. no) 1.12 (0.79 – 1.61) 0.500Page 4 of 9(page number not for citation purposes)Note: a – LGBTT denotes lesbian, gay, bisexual or transgendered/transsexual; b – CES-D standard cut-off score of 16 or greater;c – denotes self-efficacy for limiting HIV risk behaviours scale; d – HR denotes harm reduction; † – refers to activities in the past 6 months; ‡ – refers to lifetime history; * – overall p-value for main and interaction effectBMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7Table 2: Factors associated with consistent condom use among a cohort of street-involved youth (n = 415).Characteristic Consistentn (%)n = 127Inconsistentn (%)n = 288Odds Ratio(95% CI)p – valueAge¶< 22 61 (48.0) 162 (56.3) 0.72 (0.47 – 1.09) 0.150≥ 22 66 (52.0) 126 (43.7)SexFemale 36 (28.6) 95 (33.1) 0.81 (0.51 – 1.28) 0.426Male 90 (71.4) 192 (66.9)Aboriginal ethnicityYes 37 (29.1) 67 (23.3) 1.36 (0.85 – 2.17) 0.251No 90 (70.9) 221 (76.7)Sexual orientationLGBTTa 9 (7.1) 44 (15.3) 0.42 (0.20 – 0.90) 0.033Heterosexual 117 (92.9) 243 (84.7)Relationship statusSingle/Dating 105 (84.0) 193 (67.7) 2.50 (1.46 – 4.30) 0.001Regular Partner 20 (16.0) 92 (32.3)DepressionbYes 63 (51.2) 153 (55.2) 0.98 (0.64 – 1.49) 0.526No 60 (48.8) 124 (44.8)Self Efficacy LHRBcLow 46 (36.8) 129 (46.4) 0.67 (0.44 – 1.04) 0.091High 79 (63.2) 149 (53.6)Number of sex partners†> 1 79 (62.2) 165 (57.3) 1.23 (0.80 – 1.88) 0.407≤ 1 48 (37.8) 123 (42.7)Anal intercourse†Yes 12 (9.8) 59 (20.9) 0.41 (0.21 – 0.79) 0.010No 111 (90.2) 223 (79.1)Sexual abuse‡Yes 33 (26.6) 79 (27.7) 0.95 (0.59 – 1.52) 0.912No 91 (73.4) 206 (72.3)Drug dealing†Yes 71 (55.9) 175 (60.8) 0.82 (0.54 – 1.25) 0.412No 56 (44.1) 113 (39.2)Alcohol dependenceYes 56 (47.1) 154 (55.8) 0.70 (0.46 – 1.09) 0.137No 63 (52.9) 122 (44.2)Crack use†Yes 67 (52.8) 174 (60.4) 0.73 (0.48 – 1.12) 0.177No 60 (47.2) 114 (39.6)Cocaine use†Yes 58 (45.7) 137 (47.6) 0.93 (0.61 – 1.41) 0.802No 69 (54.3) 151 (52.4)Heroin use†Yes 43 (33.9) 93 (32.3) 1.07 (0.69 – 1.67) 0.842No 84 (66.1) 195 (67.7)Crystal meth use†Yes 52 (40.9) 146 (50.7) 0.67 (0.44 – 1.03) 0.084No 75 (59.1) 142 (49.3)Injection drug use†Yes 35 (27.6) 84 (29.2) 0.92 (0.58 – 1.47) 0.829No 92 (72.4) 204 (70.8)Syringe sharing†Yes 7 (5.5) 32 (11.1) 0.47 (0.20 – 1.09) 0.105No 120 (94.5) 256 (88.9)Binge drug use†Yes 49 (39.8) 139 (49.5) 0.68 (0.44 – 1.04) 0.094Page 5 of 9(page number not for citation purposes)No 74 (60.2) 142 (50.5)BMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7significant for females (aIRR = 1.76, 95% CI: 0.98 – 3.15).In multivariate logistic regression analysis, barriers toaccessing health or harm reduction services was margin-ally and inversely associated with consistent condom use(adjusted odds ratio [aOR] = 0.52, 95% CI: 0.25 – 1.07);no significant interaction with sex was observed.DiscussionThese findings reveal high rates of inconsistent condomuse and multiple sexual partnerships among a cohort ofstreet-involved youth in Vancouver, Canada. Given thatthese behaviours describe two parameters which partiallydetermine the population level transmission dynamics ofHIV and STIs, we conclude that the continued propaga-tion of these diseases among this population is likely. Ourresults also suggest that structural factors may play a rolein driving risk behaviours that increase the likelihood ofHIV and STI transmission. Further, the impact of struc-tural factors on the sexual risk behaviours of street-involved youth appear to be moderated by gender, lead-ing us to conclude that the intersection of structural deter-minants with gender and sexual inequities may promotethe production of HIV risk within this population.Having a warrant or area restriction that affects access toNEPs or other services was the strongest correlate ofHomeless†Yes 91 (71.7) 227 (78.8) 0.68 (0.42 – 1.10) 0.143No 36 (28.3) 61 (21.2)Barriers to health/HRd services†Yes 14 (11.0) 54 (18.9) 0.53 (0.28 – 1.00) 0.065No 113 (89.0) 232 (81.1)Jacked up†Yes 55 (44.4) 135 (47.5) 0.88 (0.58 – 1.35) 0.628No 69 (55.6) 149 (52.5)WarrantsYes 4 (3.2) 5 (1.8) 1.82 (0.48 – 6.92) 0.492No 121 (96.8) 275 (98.2)Unable to access treatment†Yes 13 (10.2) 37 (12.9) 0.77 (0.39 – 1.51) 0.548No 114 (89.8) 250 (87.1)Assault from police/guards†Yes 27 (21.8) 67 (23.7) 0.90 (0.54 – 1.49) 0.771No 97 (78.2) 216 (76.3)Note: a – LGBTT denotes lesbian, gay, bisexual or transgendered/transsexual; b – CES-D standard cut-off score of 16 or greater; c – denotes self-efficacy for limiting HIV risk behaviours scale;d – HR denotes harm reduction; ¶ – dichotomisation based on sample median; † – refers to activities in the past 6 months; ‡ – refers to lifetime history.Table 2: Factors associated with consistent condom use among a cohort of street-involved youth (n = 415). (Continued)Table 3: Logistic regression analysis of factors associated with consistent condom use among a cohort of street-involved youth (n = 415).Variable Adjusted Odds Ratio (AOR) 95% Confidence Interval (95% CI) p – valueSexual orientation(LGBTTa vs. heterosexual) 0.38 (0.15 – 0.97) 0.044Relationship status(single/dating vs. regular) 2.82 (1.59 – 5.01) < 0.001Self-Efficacy LHRBb(low vs. high) 0.66 (0.41 – 1.07) 0.091Anal intercourse†(yes vs. no) 0.61 (0.30 – 1.24) 0.173Crystal meth use†(yes vs. no) 0.74 (0.47 – 1.19) 0.217Binge drug use†(yes vs. no) 0.67 (0.42 – 1.08) 0.098Barriers to health/HRc services†(yes vs. no) 0.52 (0.25 – 1.07) 0.074Page 6 of 9(page number not for citation purposes)Note: a – LGBTT denotes lesbian, gay, bisexual or transgendered/transsexual; b – denotes self-efficacy for limiting HIV risk behaviours scale; c – HR denotes harm reduction; † – refers to activities in the past 6 months.BMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7number of recent sex partners, even after adjustment forpotential confounders such as homelessness. Further-more, being "jacked up" by the police was independentlyassociated with number of recent sex partners amongmales in our sample. These findings suggest that enforce-ment-based policies and practices which result in thecriminalization of street youth activity may be a contrib-uting factor in the production of HIV and STI risk amongthese populations. While few studies have characterisedthe potential impact of policing and enforcement policieson HIV and STI transmission among street youth, severalauthors have argued that street-level law enforcement pro-motes HIV risk behaviour among older populationsincluding IDU who consume drugs in public spaces. Eth-nographic research among IDU has shown that havingoutstanding warrants exacerbates the health and safetyconcerns associated with public injection due to fears ofbeing arrested by police [34]. Specific enforcement prac-tices may also impact the spread of HIV and STIs throughmore direct mechanisms. For example, the separation ofsex partners due to the removal or displacement of indi-viduals from normative structural environments has beentheorised to increase the likelihood of new discordant sex-ual partnerships and riskier sexual behaviours [24]. Ourresults provide quantitative evidence to support thesehypotheses and also indicate that more research isrequired to examine how police and other authority fig-ures interact with street-involved youth in such a way thataugments the production of HIV risk.Our finding that individuals who have experienced barri-ers to health and harm reduction services were half aslikely to report consistent condom use is worrisome. Judg-mental policies and procedures, a failure to adhere to sex-positive principles, and a lack of systems that discourageheterosexist cultures have all been recognised as structuralbarriers that prevent street-involved youth from accessingservices that sell or distribute condoms [35,36]. It isimportant to note that our findings regarding barriers tohealth or harm reduction services must be interpreted cau-tiously, as the associations between service barriers andboth sexual risk behaviour outcomes achieved only mar-ginal statistical significance. However, these results doprovide further evidence for the hypothesised associationbetween barriers to health care and harm reduction serv-ices and increased HIV and STI rates within street youthcommunities [37]. Future studies should seek to examinehow specific mechanisms or barriers (e.g., stigma, inade-quate coverage, inappropriateness of services) influencethe accessibility and use of HIV/STI programs andresources for this population.The other two factors that were associated with both sex-fied individuals are more likely to have multiple sex part-ners and less likely to report consistent condom use isworrisome and suggests that the development of interven-tions sensitive to diverse sexualities and orientations isurgently required. Furthermore, these results corroborateother research in the United States demonstrating thatLGBTT homeless youth are at increased risk for a host ofnegative sexual health outcomes [35,38]. Our finding thathomeless youth who are single or casually dating havemore sexual partners but higher rates of condom use isalso consistent with other studies demonstrating that rela-tionship status is a strong determinant of sexual riskbehaviour among youth at high risk for HIV [39,40]. Sex-ual abuse and drug use including cocaine consumption,both significantly associated with number of sex partnersin our study, have also been associated with increasednumbers of sex partners in other studies of homeless ado-lescents [41,42].This study has a number of important implications forpolicies, programmes, and interventions that attempt toreduce population level burden of HIV and STI amongyoung street-involved communities. We have shown thatthe displacement of street youth and the regulation oftheir behaviour through law enforcement strategies andother legal practices are independently associated withbehaviours that increase the likelihood of HIV and STItransmission. Therefore, socio-legal reforms that de-emphasise enforcement-based policies and incorporatehealth or harm reduction frameworks may be effective atreducing HIV and STI incidence in the future. For exam-ple, policy and legal reforms that promote the health andsafety of street-based sex workers have been shown to beeffective at reducing HIV vulnerability among these popu-lations [43]. Consistent with other studies [42,44], ourresults also indicate that homelessness may be an impor-tant driver of HIV and STI transmission. Interventions andpublic health programmes may seek to target youth whoare homeless and deeply entrenched within the street cul-ture and economy, and may benefit from incorporatingyouth-friendly, sex-positive policies and practices thatreduce social-structural barriers to traditional health careenvironments. For example, street-based STI testing that isincorporated within pre-existing outreach services hasbeen shown to be highly effective at reducing the struc-tural barriers associated with traditional hospital or clinicsettings [37].This study has a number of sampling and methodologicallimitations. It is important to note that, due to the cross-sectional nature of the study design, these results are cor-relational and therefore no inferences can be made withrespect to causation. It is possible that the constellation ofPage 7 of 9(page number not for citation purposes)ual risk behaviour outcomes included sexual orientationand relationship status. Our finding that LGBTT-identi-structural factors observed in this study may simply clus-ter among youth who are more likely to engage in sexualBMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/7risk behaviours for other unmeasured reasons. However,it is noteworthy that since all of the structural variablessignificant in bivariate analyses remained significant inthe multivariate models, they do not appear to be con-founded by each other. Future research using longitudinalstudy designs are required to corroborate these findings.Secondly, although extensive snowball and street-basedoutreach was used in an attempt to maximise the repre-sentativeness of our sample, we are unable to generaliseour findings to other settings with different structuralenvironments. Thirdly, the low sample size across severalcovariates resulted in wide confidence intervals that mayhave reduced our ability to observe small but significanteffects. Therefore, we encourage the cautious interpreta-tion of marginally significant results. Lastly, it is also pos-sible that socially desirable reporting resulted in an under-estimate of stigmatised behaviours such as anal inter-course and inconsistent condom use, particularly withcasual sex partners. However, we have no reason to sus-pect that differential reporting of these behavioursoccurred between those who reported structural barriersand those that did not.ConclusionWe have shown that structural factors, in particular thosethat correspond to the displacement, regulation, andcriminalization of street youth activity, are correlated withbehaviours which increase the likelihood for HIV and STItransmission. Furthermore, street-involved youth whoreport barriers to traditional health or harm reductionservices are more likely to engage in sexual risk behavioursthat put them at an increased risk for the acquisition andtransmission of these diseases. Structural factors remainedassociated with the drivers of HIV and STI transmissionindependently of individual, social, and drug-relatedcharacteristics; therefore, structural interventions thatincorporate youth-friendly, accessible, health-based poli-cies and practices may be effective at improving popula-tion level sexual health outcomes. These findings supportthe need for innovative interventions including legalreforms, non-coercive policing practices, and street-basedoutreach and sexual health services to reduce the preva-lence of HIV and other STIs among marginalised youthpopulations in the future.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsEW had full access to all of the data and takes responsibil-ity for the integrity of the results and the accuracy of thestatistical analysis. BM, TK, and JS conceived the studyconcept and design and BM was responsible for the com-performed by BM, TK, JS, JM, and EW. The manuscript wasedited and revised by BM, TK, JS, JM and TK. All authorsread and approved the final manuscriptFundingThe study was supported by the US National Institutes ofHealth (R01 DA011591) and Canadian Institutes ofHealth Research (HHP-67262 and RRA-79918). ThomasKerr is supported by a New Investigator Award from CIHRand a Scholar Award from the Michael Smith Foundationfor Health Research (MSFHR). Jean Shoveller is supportedby a Senior Scholar Award from MSFHR and a PublicHealth Chair in Improving Youth Sexual Health fromCIHR. Brandon Marshall is supported by a Canada Grad-uate Scholarship from CIHR and a Junior GraduateTrainee Award from MSFHR.AcknowledgementsWe would particularly like to thank the At Risk Youth Study (ARYS) par-ticipants for volunteering their time to participate in the study. We thank Deborah Graham, Tricia Collingham, Leslie Rae, Caitlin Johnston, Steve Kain, and Calvin Lai for their research and administrative assistance. We would also like to acknowledge Drs. Thomas Patterson and Jane Buxton for their advice and assistance regarding the analysis.References1. Rhodes T: The 'risk environment': a framework for under-standing and reducing drug-related harm. Int J Drug Policy 2002,13:85-94.2. Sumartojo E: Structural factors in HIV prevention: concepts,examples, and implications for research. AIDS 2000, 14(Suppl1):S3-S10.3. Rhodes T, Singer M, Bourgois P, Friedman SR, Strathdee SA: Thesocial structural production of HIV risk among injecting drugusers. Soc Sci Med 2005, 61:1026-1044.4. Shannon K, Rusch M, Shoveller J, Alexson D, Gibson K, Tyndall MW:Mapping violence and policing as an environmental-struc-tural barrier to health service and syringe availability amongsubstance-using women in street-level sex work. Int J Drug Pol-icy 2008, 19:140-147.5. Rotheram-Borus MJ: Expanding the range of interventions toreduce HIV among adolescents. AIDS 2000, 14(Suppl1):S33-S40.6. Public Health Agency of Canada: Street Youth in Canada: Find-ings from Enhanced Surveillance of Canadian Street Youth,1999–2003. [http://www.phac-aspc.gc.ca/std-mts/reports_06/youth_e.html].7. Ringwalt CL, Greene JM, Robertson M, McPheeters M: The preva-lence of homelessness among adolescents in the UnitedStates. Am J Public Health 1998, 88:1325-1329.8. Roy E, Haley N, Leclerc P, Sochanski B, Boudreau JF, Boivin JF: Mor-tality in a cohort of street youth in Montreal. JAMA 2004,292:569-574.9. DeMatteo D, Major C, Block B, Coates R, Fearon M, Goldberg E, KingSM, Millson M, O'Shaughnessy M, Read SE: Toronto street youthand HIV/AIDS: Prevalence, demographics, and risks. J AdolescHealth 1999, 25:358-366.10. Roy E, Haley N, Leclerc P, Lemire N, Boivin JF, Frappier JY, ClaessensC: Prevalence of HIV infection and risk behaviours amongMontreal street youth. Int J STD AIDS 2000, 11:241-247.11. Public Health Agency of Canada: Sexually Transmitted Infectionsin Canadian Street Youth: Findings from Enhanced Surveil-lance of Canadian Street Youth, 1999–2003. [http://www.phac-aspc.gc.ca/std-mts/reports_06/sti-youth_e.html].12. Haley N, Roy E, Leclerc P, Lambert G, Boivin JF, Cedras L, VincelettePage 8 of 9(page number not for citation purposes)position of the manuscript. The statistical analysis wasconducted by BM, and the interpretation of the results wasJ: Risk behaviours and prevalence of Chlamydia trachomatisand Neisseria gonorrhoeae genital infections among Mon-treal street youth. Int J STD AIDS 2002, 13:238-245.Publish with BioMed Central and every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Public Health 2009, 9:7 http://www.biomedcentral.com/1471-2458/9/713. Noell J, Rohde P, Ochs L, Yovanoff P, Alter MJ, Schmid S, Bullard J,Black C: Incidence and prevalence of chlamydia, herpes, andviral hepatitis in a homeless adolescent population. SexTransm Dis 2001, 28:4-10.14. Milburn NG, Rotheram-Borus MJ, Rice E, Mallet S, Rosenthal D:Cross-national variations in behavioral profiles amonghomeless youth. Am J Commun Psychol 2006, 37:63-76.15. Kral AH, Molnar BE, Booth RE, Watters JK: Prevalence of sexualrisk behaviour and substance use among runaway and home-less adolescents in San Francisco, Denver and New YorkCity. Int J STD AIDS 1997, 8:109-117.16. Clements K, Gleghorn A, Garcia D, Katz M, Marx R: A risk profileof street youth in northern California: implications for gen-der-specific human immunodeficiency virus prevention. JAdolesc Health 1997, 20:343-353.17. Halcon LL, Lifson AR: Prevalence and predictors of sexual risksamong homeless youth. J Youth Adolescence 2004, 33:71-80.18. Zimet GD, Sobo EJ, Zimmerman T, Jackson J, Mortimer J, Yanda CP,Lazebnik R: Sexual behavior, drug use, and AIDS knowledgeamong Midwestern runaways. Youth Soc 1995, 26:450-462.19. Montgomery SB, Hyde J, De Rosa CJ, Rohrbach LA, Ennett S, HarveySM, Clatts M, Iverson E, Kipke MD: Gender differences in HIVrisk behaviors among young injectors and their social net-work members. Am J Drug Alcohol Abuse 2002, 28:453-475.20. Greene JM, Ennett ST, Ringwalt CL: Prevalence and correlates ofsurvival sex among runaway and homeless youth. Am J PublicHealth 1999, 89:1406-1409.21. Tyler KA, Whitbeck LB, Hoyt DR, Cauce AM: Risk factors for sex-ual victimization among male and female homeless and run-away youth. J Interpers Violence 2004, 19:503-520.22. Zerger S, Strehlow AJ, Gundlapalli AV: Homeless young adultsand behavioral health – An overview. Am Behav Sci 2008,51:824-841.23. Shoveller JA, Johnson JL: Risky groups, risky behaviour, and riskypersons: Dominating discourses on youth sexual health. CritPublic Health 2006, 16:47-60.24. Aral SO, Lipshutz J, Blanchard J: Drivers of STD/HIV epidemiol-ogy and the timing and targets of STD/HIV prevention. SexTransm Infect 2007, 83(Suppl 1):i1-i4.25. Wood E, Stoltz JA, Montaner JS, Kerr T: Evaluating methamphet-amine use and risks of injection initiation among streetyouth: The ARYS study. Harm Reduct J 2006, 3:18.26. de Carvalho FT, Neiva-Silva L, Ramos MC, Evans J, Koller SH, PiccininiCA, Page-Shafer K: Sexual and drug use risk behaviors amongchildren and youth in street circumstances in Porto Alegre,Brazil. AIDS Behav 2006, 10:S57-S66.27. Rushton JL, Forcier M, Schectman RM: Epidemiology of depres-sive symptoms in the National Longitudinal Study of Adoles-cent Health. J Am Acad Child Adolesc Psychiatry 2002, 41:199-205.28. Smith KW, McGraw SA, Costa LA, McKinlay JB: A self-efficacyscale for HIV risk behaviors: Development and evaluation.AIDS Educ Prev 1996, 8:97-105.29. Kerr T, Tyndall M, Li K, Montaner J, Wood E: Safer injection facil-ity use and syringe sharing in injection drug users. Lancet2005, 366:316-318.30. Petchers MK, Singer MI: Perceived-Benefit-of-Drinking Scale:Approach to screening for adolescent alcohol abuse. J Pediatr1987, 110:977-981.31. MacKellar DA, Valleroy LA, Hoffmann JP, Glebatis D, Lalota M,McFarland W, Westerholm J, Janssen RS: Gender differences insexual behaviors and factors associated with nonuse of con-doms among homeless and runaway youths. AIDS Educ Prev2000, 12:477-491.32. Evans JL, Hahn JA, Page-Shafer K, Lum PJ, Stein ES, Davidson PJ, MossAR: Gender differences in sexual and injection risk behavioramong active young injection drug users in San Francisco(the UFO Study). J Urban Health 2003, 80:137-146.33. Kerr T, Small W, Wood E: The public health and social impactsof drug market enforcement: A review of the evidence. Int JDrug Policy 2005, 16:210-220.34. Kerr T, Small W, Moore D, Wood E: A micro-environmentalintervention to reduce the harms associated with drug-related overdose: Evidence from the evaluation of Vancou-ver's safer injection facility. Int J Drug Policy 2007, 18:37-45.bisexual, and transgender homeless adolescents with theirheterosexual counterparts. Am J Public Health 2002, 92:773-777.36. Howard J: Taking a chance on love: Risk behaviour of Sydneystreet youth. J Paediatr Child Health 1993, 29(Suppl 1):S60-S65.37. Auerswald CL, Sugano E, Ellen JM, Klausner JD: Street-based STDtesting and treatment of homeless youth are feasible,acceptable and effective. J Adolesc Health 2006, 38:208-212.38. Gangamma R, Slesnick N, Toviessi P, Serovich J: Comparison ofHIV risks among gay, lesbian, bisexual and heterosexualhomeless youth. J Youth Adolescence 2008, 37:456-464.39. Fortenberry JD, Tu W, Harezlak J, Katz BP, Orr DP: Condom useas a function of time in new and established adolescent sex-ual relationships. Am J Public Health 2002, 92:211-213.40. Friedman SR, Flom PL, Kottiri BJ, Neaigus A, Sandoval M, Curtis R,Des Jarlais DC, Zenilman JM: Consistent condom use in the het-erosexual relationships of young adults who live in a high-HIV-risk neighbourhood and do not use "hard drugs". AIDSCare 2001, 13:285-296.41. Rotheram-Borus MJ, Mahler KA, Koopman C, Langabeer K: Sexualabuse history and associated multiple risk behavior in ado-lescent runaways. Am J Orthopsychiat 1996, 66:390-400.42. Solorio MR, Rosenthal D, Milburn NG, Weiss RE, Batterham PJ, Gan-dara M, Rotheram-Borus MJ: Predictors of sexual risk behaviorsamong newly homeless youth: A longitudinal study. J AdolescHealth 2008, 42:401-409.43. Kerrigan D, Moreno L, Rosario S, Gomez B, Jerez H, Barrington C,Weiss E, Sweat M: Environmental-structural interventions toreduce HIV/STI risk among female sex workers in theDominican Republic. Am J Public Health 2006, 96:120-125.44. Coady MH, Latka MH, Thiede H, Golub ET, Ouellet L, Hudson SM,Kapadia F, Garfein RS: Housing status and associated differ-ences in HIV risk behaviors among young injection drugusers (IDUs). AIDS Behav 2007, 11:854-863.Pre-publication historyThe pre-publication history for this paper can be accessedhere:http://www.biomedcentral.com/1471-2458/9/7/prepubyours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 9 of 9(page number not for citation purposes)35. Cochran BN, Stewart AJ, Ginzler JA, Cauce AM: Challenges facedby homeless sexual minorities: Comparison of gay, lesbian,