UBC Faculty Research and Publications

Predictors of homelessness among vulnerably housed adults in 3 Canadian cities: a prospective cohort… To, Matthew J; Palepu, Anita; Aubry, Tim; Nisenbaum, Rosane; Gogosis, Evie; Gadermann, Anne; Cherner, Rebecca; Farrell, Susan; Misir, Vachan; Hwang, Stephen W Oct 3, 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-12889_2016_Article_3711.pdf [ 603.21kB ]
Metadata
JSON: 52383-1.0362014.json
JSON-LD: 52383-1.0362014-ld.json
RDF/XML (Pretty): 52383-1.0362014-rdf.xml
RDF/JSON: 52383-1.0362014-rdf.json
Turtle: 52383-1.0362014-turtle.txt
N-Triples: 52383-1.0362014-rdf-ntriples.txt
Original Record: 52383-1.0362014-source.json
Full Text
52383-1.0362014-fulltext.txt
Citation
52383-1.0362014.ris

Full Text

RESEARCH ARTICLE Open AccessPredictors of homelessness amongvulnerably housed adults in 3 Canadiancities: a prospective cohort studyMatthew J. To1*, Anita Palepu2, Tim Aubry3, Rosane Nisenbaum1, Evie Gogosis1, Anne Gadermann2,Rebecca Cherner3, Susan Farrell4, Vachan Misir1 and Stephen W. Hwang1,5AbstractBackground: Homelessness is a major concern in many urban communities across North America. Since vulnerablyhoused individuals are at risk of experiencing homelessness, it is important to identify predictive factors linked tosubsequent homelessness in this population. The objectives of this study were to determine the probability ofexperiencing homelessness among vulnerably housed adults over three years and factors associated with higherrisk of homelessness.Methods: Vulnerably housed adults were recruited in three Canadian cities. Data on demographic characteristics,chronic health conditions, and drug use problems were collected through structured interviews. Housing historywas obtained at baseline and annual follow-up interviews. Generalized estimating equations were used tocharacterize associations between candidate predictors and subsequent experiences of homelessness during eachfollow-up year.Results: Among 561 participants, the prevalence of homelessness was 29.2 % over three years. Male gender (AOR= 1.59, 95 % CI: 1.14–2.21) and severe drug use problems (AOR = 1.98, 95 % CI: 1.22–3.20) were independentlyassociated with experiencing homelessness during the follow-up period. Having ≥3 chronic conditions (AOR = 0.55,95 % CI: 0.33–0.94) and reporting higher housing quality (AOR = 0.99, 95 % CI: 0.97–1.00) were protective againsthomelessness.Conclusions: Vulnerably housed individuals are at high risk for experiencing homelessness. The study has publichealth implications, highlighting the need for enhanced access to addiction treatment and improved housingquality for this population.Keywords: Homeless persons, Housing, Urban health, Substance-related disorders, Public healthBackgroundHomelessness is a major public health concern in manycommunities across North America. Recent reports sug-gest an estimated 650,000 individuals across the UnitedStates and Canada are homeless on any given night [1, 2].Compared with the general population, homeless individ-uals have poorer health status and a high prevalence ofphysical and mental health problems [3–5]. As a result,they experience high rates of healthcare utilization, mor-bidity, and mortality [5].Numerous studies have examined risk factors for onsetof homelessness and identified several risk factor cat-egories such as demographic characteristics, physicaland mental health status, substance use, involvementwith the criminal justice system, and housing conditions[6–13]. With regards to demographic factors, youngerage has been associated with a higher likelihood of be-coming homeless and shorter duration of homelessness[6, 7]. Male gender and African American ethnicity havebeen identified as independent predictors of homeless-ness [8, 9]. Obtaining less than a high school education* Correspondence: mto3@alumni.uwo.ca1Centre for Urban Health Solutions, Li Ka Shing Knowledge Institute, St.Michael’s Hospital, 30 Bond Street, Toronto, ON M5B 1W8, CanadaFull list of author information is available at the end of the article© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.To et al. BMC Public Health  (2016) 16:1041 DOI 10.1186/s12889-016-3711-8has also been associated with homelessness [6, 10],while being a college graduate has been recognized asa protective factor [11]. Having no income, lower in-come, or financial difficulties are risk factors forhomelessness [10, 12, 13]. Unemployment has beenassociated with homelessness [11], while employmentand earned income are predictors of a shorter dur-ation of homelessness [7].A variety of physical and mental health conditionshave been linked to homelessness [4]. Physical healthproblems and worsening of general health have beenassociated with homelessness [6, 11, 14]. Mental illnessand family history of mental illness have been identifiedas predictors of homelessness [12, 13]. Specifically,homelessness has been linked to schizophrenia [8], bipo-lar disorder [8], anxiety disorder [11], post-traumaticstress disorder [11], and personality disorder [15].Moreover, substance use and addictions are importantrisk factors for homelessness [8, 13, 16]. Illicit drug useand having an alcohol use disorder are both predictorsof homelessness [6, 15]. Previous research has identifiedcrack cocaine use as a risk factor for becoming andremaining homeless [17]. In addition, recent drug injec-tion use is associated with homelessness [14].Previous research has also linked housing status andhousing conditions such as crowding with homelessness[9]. Living in unstable housing is also a predictor ofhomelessness [12]. Protective factors against homeless-ness include receipt of subsidized housing and havingone’s own place [9]. Preliminary research has also foundan association between social support and homelessness,suggesting that lower social support is linked to chronichomelessness [18].Although previous research has identified risk factorsfor onset of homelessness, few studies have examinedrisk factors for homelessness among individuals who arevulnerably housed, which has been defined as experien-cing prior homelessness or having frequent housingtransitions [19]. Emerging research suggests that vulner-ably housed individuals have similar health and socialoutcomes to homeless individuals and are at an in-creased risk of experiencing homelessness [9, 19].Recent studies have found that a substantial proportionof homeless individuals who obtain housing subsequentlyexperience a recurrence of homelessness [20, 21]. In onestudy of 344 single adults in emergency shelters in NewYork City who were newly homeless, 24 % of those whoobtained housing (81 %) over an 18-month period hadanother episode of homelessness. Recurrent homelessnesswas more common among those who were initiallyrehoused with family and those with a high school educa-tion [20]. Compared to housed individuals, those experi-encing recurrent homelessness were more likely to have a30-day and lifetime history of alcohol and substance usedisorders [20]. Another U.S. study examined predictors ofreturning to homelessness after attaining housing in asample of 392 formerly homeless veterans who partici-pated in a trial of case management and rent subsidies,case management only, or standard care [21]. Over thecourse of a five year period, 44 % of all participants experi-enced another homeless episode after being housed. Par-ticipants who received case management and rentsubsidies had significantly longer periods of continuoushousing compared with participants in the two othergroups [21].Given that vulnerably housed individuals are at risk ofexperiencing episodes of homelessness, it is important toidentify potentially predictive factors linked to subse-quent homelessness in this population. The presentpaper reports findings from the Health and Housing inTransition (HHiT) study, a prospective cohort study thattracked the health and housing status of homeless andvulnerably housed adults in three major Canadian cities[19]. The objectives of this paper are to examine partici-pants who were vulnerably housed at the baseline inter-view and determine the probability of experiencinghomelessness over a three-year follow-up period and theindividual characteristics associated with higher risk ofhomelessness.MethodsParticipantsHomeless and vulnerably housed persons aged 18 orolder who were single (i.e. not living with a partner ordependent child) were recruited in Ottawa, Toronto, andVancouver from January to December 2009. Homeless-ness was defined as living within the last seven days at ashelter, public space, vehicle, abandoned building, orsomeone else’s home, and not having a home of one’sown. Vulnerably housed was defined as currently livingin one’s own room or apartment, but having been home-less or had two or more moves in the past twelvemonths. Full-time students and individuals who werevisiting the city for three months or less were excluded.RecruitmentThe sampling procedure for recruiting homeless partici-pants was based on the design suggested by Ardilly andLe Blanc (2001) [22]. Study participants were recruitedat homeless shelters and meal programs. Homeless par-ticipants who did not use shelters were recruited at mealprograms proportionally to the estimated number ofhomeless persons that slept on the street in eachrespective city. Vulnerably housed participants wererecruited from randomly selected rooming houses inOttawa and Toronto, and from Single Room Occupancy(SRO) hotels in Vancouver. Due to difficulties in gainingaccess to some of these locations, the recruitmentTo et al. BMC Public Health  (2016) 16:1041 Page 2 of 12strategy for vulnerably housed individuals was modifiedto include meal programs, drop-in centers, and commu-nity health centers. Data were collected from partici-pants between January 2009 and February 2013. Allstudy participants provided written informed consentand received $20 CDN upon interview completion. TheResearch Ethics Boards at the University of Ottawa; St.Michael’s Hospital, Toronto; and the University of BritishColumbia, Vancouver approved this study.Survey instrumentFull details of all survey instruments used in the studyhave been reported elsewhere [19]. Data on demographiccharacteristics, health conditions and health status, alco-hol and drug use, housing history and quality, socialsupport, legal incidents, and victimization were collectedusing structured, in-person interviews conducted bytrained research personnel immediately following re-cruitment. Interviews took approximately 60 to 90 mi-nutes to complete. Participants reported their ethnicbackground based on categories adapted from the Statis-tics Canada Ethnic Diversity Survey [23].Chronic health conditions listed in the survey toolwere adapted from the Canadian Community HealthSurvey [24], and participants were asked to report anychronic health conditions that had lasted or wereexpected to last six months or more and had been diag-nosed by a healthcare professional. Lifetime prevalenceof mental health diagnoses was determined by self-report. Lifetime prevalence of traumatic brain injury(TBI) was determined using a question from a previousstudy on prison inmates [25]. Participants were askedwhether they had ever had “an injury to the head whichknocked you out or at least left you dazed, confused, ordisoriented?” Health status was determined using the12-item Short Form Health Survey (SF-12) to generatePhysical Component Summary (PCS) and MentalComponent Summary (MCS) subscale scores [26].Alcohol use was assessed using the Alcohol Use Disor-ders Identification Test (AUDIT), with a score of eight ormore resulting in a positive screen, with scores of 8–15indicating hazardous, 16–19 harmful, and 20–40 indicat-ing high levels of risk related to alcohol [27]. Drug useproblems experienced by participants were assessed usingthe 10-item version of the Drug Abuse Screening Test(DAST-10) [28]. Scores of three or higher on the DAST-10 resulted in a positive screen, with higher scores indicat-ing moderate (scores 3–5), substantial (scores 6–8), orsevere (scores 9–10) drug use problems. The HousingQuality Score was used to determine self-reported qualityof the current living environment in 6 domains: comfort,safety, spaciousness, privacy, friendliness, and overall qual-ity [29]. Each item was ranked on a 7-point Likert scalewith a maximum total score of 42. Social support wasassessed using the Social Support Network Inventory(SSNI), a questionnaire that measured the size of a per-son’s social network and perceived social support [30].Housing history data were categorized based onmethods adapted from Tsemberis et al. [31]. Each resi-dence in a participant’s housing history was classifiedinto one of 25 types of residence, which were then clas-sified into one of three mutually exclusive residence cat-egories: housed, institution, and homeless. Periods oftime spent in institutions were considered periods ofbeing homeless or housed based on a functional classifi-cation [31]. Further details are available from the authorsupon request.Participants provided contact information duringadministration of the baseline survey so that they couldbe located for follow-up surveys administered approxi-mately one year, two years, and three years after thebaseline survey. The follow-up survey included questionsof a similar nature to the baseline survey on health sta-tus and health conditions, alcohol and drug use, housingstatus and quality, and social support.Data analysisVulnerably housed participants originally recruited intothe study who did not complete any follow-up inter-views were excluded from the analyses. Descriptive sta-tistics were used to summarize all quantitative variables.The percentage of vulnerably housed adults who experi-enced homelessness anytime over the three-year follow-up period was calculated. The main outcome of interestwas whether a vulnerably housed participant ever experi-enced homelessness during any of the three one-yearperiods between the baseline and follow-up 1 interviews,the follow-up 1 and follow-up 2 interviews, and thefollow-up 2 and follow-up 3 interviews.Baseline characteristics were summarized using means,standard deviations, medians, interquartile ranges, andproportions, wherever appropriate. Comparisons be-tween vulnerably housed participants who did and didnot experience homelessness during the three-yearfollow-up period were performed for baseline character-istics. P-values were calculated from t-test or Wilcoxonrank-sum test for continuous variables. Chi-square testor Fisher’s exact test were used for categorical variables.Various demographic, health, and housing variableswere assessed for an association with a higher probabil-ity of becoming homeless over a three-year follow-upperiod. These characteristics included fixed covariates(determined at the baseline interview) and time-varyingcovariates (determined at baseline, follow-up 1, follow-up 2 interviews).The list of candidate predictors of homelessness wasdeveloped based on a literature review and consultationwith experts. Characteristics assessed for associationTo et al. BMC Public Health  (2016) 16:1041 Page 3 of 12with experiencing homelessness in the follow-up periodwere city, time interval, and 1) fixed predictors includ-ing: age, gender, ethnicity, highest level of education,percentage of time spent homeless two years prior tobaseline divided by 10 for ease of interpretation, numberof chronic health conditions (≥3 versus <3), history of amental health diagnosis, history of TBI, and 2) time-varying predictors evaluated at the beginning of eachone-year interval including: employment in the past12 months, total income in the past 12 months, SF-12PCS, SF-12 MCS, AUDIT risk level, DAST risk level,housing quality score, and social support network size.For example, SF-12 PCS at baseline, follow-up 1, andfollow-up 2 interviews was a time-varying predictor forthe main outcome of homelessness during the periodsbetween the baseline and follow-up 1 interviews, thefollow-up 1 and follow-up 2 interviews, and the follow-up 2 and follow-up 3 interviews.Generalized estimating equations (GEE) with the logitlink were used to determine the association betweenpredictors and experiencing homelessness, accountingfor the correlations between repeated measurements(SAS PROC GENMOD). For fixed predictors, the quasi-likelihood information criteria (QIC) were used to findthe correlation structure. For time-varying predictors,we applied the Rotnitzky and Jewell approach [32, 33],and chose the correlation structure for which its associ-ated empirical covariance matrix was closer to themodel-based covariance matrix. The GEE model was de-veloped in two steps. Step 1 included city, time intervaland fixed predictors, which were retained in the model ifsignificantly associated with or clinically relevant for theoutcome (Core Model); Step 2 added time-varying pre-dictors to the Core Model, one at a time. Time-varyingpredictors significantly associated with or clinically rele-vant for the outcome were retained in the final model.Analyses for the 2-step process were performed usingthe exchangeable working correlation structure and cod-ing of time as a continuous variable (time interval years1, 2, 3) because these settings yielded slightly bettergoodness of fit statistics. All statistical tests were two-tailed and statistical significance was set at a P-value of0.05 or less. SAS 9.4 (SAS Institute, Inc., Cary, NC) wasused for all analyses.ResultsOf 594 vulnerably housed individuals interviewed atbaseline, 561 (94.4 %) completed at least one of threefollow-up interviews and were included in the analyses.Attrition was due to inability to locate participants,refusal to participate, and death. Baseline characteristicsfor the whole sample and stratified by city are providedin Table 1.As per design, equal numbers of vulnerably housedparticipants were recruited in each city (34 %, 33 %,33 % in Ottawa, Toronto, and Vancouver, respectively).The mean age was 42.6 (standard deviation [SD] 9.8)years, and the majority of participants were male, White,born in Canada, had completed some high school, notpartnered, and currently unemployed. Participants had amedian monthly income of $900 CDN (interquartilerange [IQR] 600–1320) with a median percentage ofincome spent on rent equal to 41.4 % (IQR 25–62 %).More than half of vulnerably housed participants(54 %) were living with 3 or more chronic health condi-tions. Fifty-five percent of participants had ever receiveda mental health diagnosis and 64 % reported having aprior TBI. Mean SF-12 PCS score was 43.6 (SD 10.79)and mean SF-12 MCS score was 39.93 (SD 13.02). Twohundred and twenty-two (40 %) participants had a posi-tive AUDIT screen and 322 (58 %) had a positive DASTscreen. Almost all participants had their own place(95.7 %) or stayed in a place belonging to friends or fam-ily (4.3 %). Of this group, 85 % had experienced home-lessness in the one-year period before entering thestudy. Median age at first homelessness was 24 (IQR16–38) and median lifetime years of homelessness was3.15 (IQR 1–6.73) among participants. Less than half ofparticipants (41 %) were residing in subsidized housing.Mean housing quality score was 27.53 (SD 8.21).Over the three-year period, 269 of 561 (48 %) partici-pants experienced at least one episode of homelessness,while 292 (52 %) never reported homelessness. Amongthose who had experienced homelessness, the medianduration was 202 (IQR 92–456) days. Those who experi-enced homelessness during the three-year follow-upperiod were significantly more likely at baseline to havebeen younger, born in Canada, completed high school/equivalent, employed in the past 12 months, smokedcigarettes daily, have positive AUDIT and DAST screens,and experienced arrests or incarceration in the past12 months (p < 0.05) (Table 2).Across 1618 residential records of the 561 participants,the prevalence of homelessness was 29 %. Specifically,the prevalence of homelessness among residentialrecords of participants was 30.7 %, 28.5 % and 28.3 %between baseline and follow-up at 1 year, betweenfollow-up at 1 year and follow-up at 2 years, betweenfollow-up at 2 years and follow-up at 3 years, respect-ively. Participants experienced a variety of residentialstate trajectories, where many individuals experiencedtransitions into and out of homelessness (Fig. 1).Table 3 presents GEE results. In the final multivariablemodel, residing in Ottawa (Adjusted Odds Ratio [AOR]= 1.88, 95 % Confidence Interval [CI]: 1.31–2.70), malegender (AOR = 1.59, 95 % CI: 1.14–2.21), percentage oftime homeless prior to the baseline interview (AOR =To et al. BMC Public Health  (2016) 16:1041 Page 4 of 12Table 1 Baseline characteristics of 561 vulnerably housed adults in 3 Canadian citiesCharacteristic Total(N = 561)an (%)Ottawa(N = 190)an (%)Toronto(N = 186)an (%)Vancouver(N = 185)an (%)Age Group< 30 years 63 (11.2) 30 (15.8) 12 (6.5) 21 (11.4)30–39 years 137 (24.4) 50 (26.3) 40 (21.5) 47 (25.4)40–49 years 221 (39.4) 62 (32.6) 87 (46.8) 72 (38.9)≥ 50 years 140 (25) 48 (25.3) 47 (25.3) 45 (24.3)Mean age (SD) 42.6 (9.8) 41.5 (10.5) 43.8 (9.4) 42.5 (9.4)GenderMale 391 (69.7) 149 (78.4) 125 (67.2) 117 (63.2)Female 162 (28.9) 41 (21.6) 57 (30.7) 64 (34.6)Transgendered 8 (1.4) 0 4 (2.2) 4 (2.2)EthnicityWhite 344 (63.4) 149 (79.3) 90 (51.1) 105 (58.7)Black/African-Canadian 36 (6.6) 3 (1.6) 27 (15.3) 6 (3.4)First Nations/Aboriginal 122 (22.5) 32 (17) 37 (21) 53 (29.6)Mixed/other 41 (7.6) 4 (2.1) 22 (12.5) 15 (8.4)Born in Canada 496 (89.4) 183 (96.8) 147 (79) 166 (92.2)Highest level of educationSome high school 279 (50.4) 100 (53.2) 86 (46.7) 93 (51.1)Completed high school/equivalent 121 (21.8) 36 (19.2) 42 (22.8) 43 (23.6)Some post-secondary or higher 154 (27.8) 52 (27.7) 56 (30.4) 46 (25.3)Partnered 134 (24.3) 48 (25.7) 37 (20.1) 49 (27.1)Employed in past 12 months 213 (38) 84 (44.4) 61 (32.8) 68 (36.8)Monthly income, median (IQR) 900 (600–1320) 912.33 (591–1340) 750 (550–1200) 966 (698–1480)Percent of income spent on rent, median (IQR) 41.4 (25.0–62.0) 43.0 (27.3–63.6) 38.9 (22.8–65.0) 41.1 (27.8–57.9)Chronic health conditions0 55 (9.8) 10 (5.3) 36 (19.4) 9 (4.9)1 106 (18.9) 31 (16.3) 38 (20.4) 37 (20)2 97 (17.3) 32 (16.8) 36 (19.4) 29 (15.7)≥ 3 303 (54) 117 (61.6) 76 (40.9) 110 (59.5)History of a mental health diagnosis 303 (54.9) 120 (64.9) 78 (42.2) 105 (57.7)History of traumatic brain injury 358 (64) 136 (72) 97 (52.4) 125 (67.6)SF-12 PCS, mean (SD) 43.6 (10.8) 43.01 (11.3) 44.42 (10.26) 43.35 (10.8)SF-12 MCS, mean (SD) 39.93 (13) 39.04 (13.5) 40.77 (12.79) 39.97 (12.76)Pregnancy in past 12 months 12 (7.5) 4 (10) 3 (5.3) 5 (7.8)Currently smokingDaily 451 (80.8) 161 (85.2) 138 (74.6) 152 (82.6)Occasionally 46 (8.2) 11 (5.8) 22 (11.9) 13 (7.1)Not at all 61 (10.9) 17 (9) 25 (13.5) 19 (10.3)Positive AUDIT screen 222 (39.7) 82 (43.4) 76 (40.9) 64 (34.8)To et al. BMC Public Health  (2016) 16:1041 Page 5 of 121.06, 95 % CI: 1.01–1.11), moderate DAST risk level(AOR = 1.4, 95 % CI: 1.00–1.97) substantial DAST risklevel (AOR = 1.71, 95 % CI: 1.20–2.44), and severe DASTrisk level (AOR = 1.98, 95 % CI: 1.22–3.20) were inde-pendently associated with experiencing homelessnessover the three-year follow-up period. Factors that wereindependently associated with a decreased likelihood ofexperiencing homelessness over the three-year follow-upperiod were having less than a high school education(AOR = 0.68, 95 % CI: 0.50–0.91), having 3 or morechronic health conditions (AOR = 0.55, 95 % CI: 0.33–0.94), and higher per unit housing quality (AOR = 0.99,95 % CI: 0.97–1.00).DiscussionAmong residential records of vulnerably housed partici-pants, the prevalence of homelessness was 29 % over thethree-year follow-up period, suggesting that experiencesof homelessness are relatively common in this popula-tion. This finding is similar to other studies that haveTable 1 Baseline characteristics of 561 vulnerably housed adults in 3 Canadian cities (Continued)AUDIT riskLow 337 (60.3) 107 (56.6) 110 (59.1) 120 (65.2)Hazardous 88 (15.7) 31 (16.4) 30 (16.1) 27 (14.7)Harmful 31 (5.6) 12 (6.4) 11 (5.9) 8 (4.4)High 103 (18.4) 39 (20.6) 35 (18.8) 29 (15.8)Positive DAST screen 322 (57.6) 109 (57.7) 85 (45.7) 128 (69.6)DAST riskNo drug use in past 12 months 136 (24.3) 35 (18.5) 72 (38.7) 29 (15.8)Low 101 (18.1) 45 (23.8) 29 (15.6) 27 (14.7)Moderate 141 (25.2) 50 (26.5) 34 (18.3) 57 (31)Substantial 134 (24) 37 (19.6) 38 (20.4) 59 (32.1)Severe 47 (8.4) 22 (11.6) 13 (7) 12 (6.5)Arrests and/or incarceration in past 12 months 190 (34) 68 (36.2) 61 (32.8) 61 (33)Physical assault victim in past 12 months 206 (37) 74 (39.6) 63 (34.1) 69 (37.3)Sexual assault victim in past 12 months 46 (8.3) 14 (7.5) 11 (5.9) 21 (11.5)Age at first homelessness, median (IQR) 24 (16–38) 22 (16–38) 27 (16–39) 23 (15–38)Lifetime years of homelessness, median (IQR) 3.15 (1–6.73) 2.38 (0.93–6) 3.5 (1–8.56) 3.56 (1.38–6.74)Own place 511 (95.7) 182 (97.3) 155 (91.2) 174 (98.3)Subsidized housing 213 (40.8) 42 (23.2) 97 (57.1) 74 (43.3)Residence TypeOwn house, apartment 180 (32.1) 66 (34.7) 99 (53.2) 15 (8.1)Stay with friends and/or relatives 13 (2.3) 2 (1.1) 9 (4.8) 2 (1.1)Rooming house 197 (35.1) 118 (62.1) 75 (40.3) 4 (2.2)SRO 160 (28.5) 0 0 160 (86.5)Substance abuse treatment facility 4 (0.7) 0 0 4 (2.2)Halfway house 2 (0.4) 2 (1.1) 0 0Supportive housing 1 (0.2) 1 (0.5) 0 0Alternative housing 3 (0.5) 0 3 (1.6) 0Other 1 (0.2) 1 (0.5) 0 0Housing quality, mean (SD) 27.53 (8.2) 27.44 (8.6) 28.48 (8.1) 26.67 (7.85)Social support network size0 95 (17.3) 33 (17.8) 36 (19.4) 26 (14.5)1 61 (11.1) 21 (11.4) 16 (8.6) 24 (13.4)2–3 104 (18.9) 29 (15.7) 38 (20.4) 37 (20.7)≥ 4 290 (52.7) 102 (55.1) 96 (51.6) 92 (51.4)aPercentages based on complete data; percentages provided by columnTo et al. BMC Public Health  (2016) 16:1041 Page 6 of 12Table 2 Baseline characteristics for 561 vulnerably housed participants across 3 Canadian cities who did and did not experiencehomelessness during the 3-year follow-up periodCharacteristic Total(N = 561)an (%)Ever homeless(N = 269)an (%)Never homeless(N = 292)an (%)P-valuebAge Group 0.0226< 30 years 63 (11.2) 31 (11.5) 32 (11.0)30–39 years 137 (24.4) 77 (28.6) 60 (20.6)40–49 years 221 (39.4) 108 (40.2) 113 (38.7)≥50 years 140 (25) 53 (19.7) 87 (29.8)Mean age (SD) 42.6 (9.8) 41.3 (9.2) 43.8 (10.2) 0.0033Gender 0.0527Male 391 (69.7) 65 (24.2) 97 (33.2)Female 162 (28.9) 200 (74.4) 191 (65.4)Transgendered 8 (1.4) 4 (1.5) 4 (1.4)Ethnicity 0.7361White 344 (63.4) 167 (64.5) 177 (62.3)Black/African-Canadian 36 (6.6) 14 (5.4) 22 (7.8)First Nations/Aboriginal 122 (22.5) 59 (22.8) 63 (22.2)Mixed/other 41 (7.6) 19 (7.3) 22 (7.8)Born in Canada 496 (89.4) 246 (92.5) 250 (86.5) 0.0225Highest level of education 0.0009Some high school 279 (50.4) 119 (44.7) 160 (55.6)Completed high school/equivalent 121 (21.8) 76 (28.6) 45 (15.6)Some post-secondary or higher 154 (27.8) 71 (26.7) 83 (28.8)Partnered 134 (24.3) 64 (24.2) 70 (24.4) 0.9478Employed in past 12 months 213 (38.0) 114 (42.4) 99 (34.0) 0.0418Monthly income, median (IQR) 900 (600–1320) 920.0 (630–1540) 889 (586–1200) 0.0906Percent of income spent on rent, median (IQR) 41.4 (25.0–62.0) 41.0 (22.5–62.3) 41.7 (27.3–60.9) 0.4403Chronic health conditions 0.06090 55 (9.8) 32 (11.9) 23 (7.9)1 106 (18.9) 52 (19.3) 54 (18.5)2 97 (17.3) 54 (20.1) 43 (14.7)≥3 303 (54) 131 (48.7) 172 (58.9)History of a mental health diagnosis 303 (54.9) 153 (57.3) 150 (52.6) 0.2703History of traumatic brain injury 358 (64) 169 (62.8) 189 (65.2) 0.5634SF-12 PCS, mean (SD) 43.6 (10.8) 44.3 (11.2) 43.0 (10.4) 0.1507SF-12 MCS, mean (SD) 39.9 (13.0) 40.3 (12.4) 39.6 (13.6) 0.5231Pregnancy in past 12 months 12 (7.5) 6 (9.2) 6 (6.3) 0.4798Currently smoking 0.0002Daily 451 (80.8) 231 (86.2) 220 (75.9)Occasionally 46 (8.2) 23 (8.6) 23 (7.9)Not at all 61 (10.9) 14 (5.2) 47 (16.2)To et al. BMC Public Health  (2016) 16:1041 Page 7 of 12found that 24–44 % of individuals with prior homeless ep-isodes experienced a recurrence of homelessness [20, 21].The current study also revealed a variety of housingtrajectories among participants, with many individualsexperiencing multiple episodes of homelessness andbeing housed over the follow-up period. These find-ings suggest that vulnerably housed individuals fre-quently experience housing instability and housingtransitions, similar to what has been previously re-ported for homeless individuals [19].The study also identified several risk factors of sub-sequent homelessness. Vulnerably housed individualswho were male, residing in Ottawa, had spent ahigher percentage of time homeless prior to the base-line interview, or had moderate to severe drug useproblems were significantly more likely to experiencehomelessness over the three-year follow-up perioddespite adjustment for potential confounders.Gender was a significant predictor of homelessnessamong vulnerably housed individuals, with men being1.6 times more likely to experience homelessness com-pared to women during the follow-up period. This isconsistent with previous research that has identifiedmale gender as a risk factor for homelessness [8]. Ottawaparticipants were also more likely to experience home-lessness during the follow-up period compared toTable 2 Baseline characteristics for 561 vulnerably housed participants across 3 Canadian cities who did and did not experiencehomelessness during the 3-year follow-up period (Continued)Positive AUDIT screen 222 (39.7) 119 (44.2) 103 (35.5) 0.0353AUDIT risk 0.1388Low 337 (60.3) 150 (55.8) 187 (64.5)Hazardous 88 (15.7) 44 (16.4) 44 (15.2)Harmful 31 (5.6) 16 (6.0) 15 (5.2)High 103 (18.4) 59 (21.9) 44 (15.2)Positive DAST screen 322 (57.6) 169 (62.8) 153 (52.8) 0.0161DAST risk 0.0994No drug use in past 12 months 136 (24.3) 57 (21.2) 79 (27.2)Low 101 (18.1) 43 (16.0) 58 (20.0)Moderate 141 (25.2) 68 (25.3) 73 (25.2)Substantial 134 (24) 76 (28.3) 58 (20.0)Severe 47 (8.4) 25 (9.3) 22 (7.6)Arrests and/or incarceration in past 12 months 190 (34) 108 (40.3) 82 (28.2) 0.0025Physical assault victim in past 12 months 206 (37) 99 (37.1) 107 (36.9) 0.9645Sexual assault victim in past 12 months 46 (8.3) 26 (9.7) 20 (6.9) 0.2335Age at first homelessness, median (IQR) 24 (16–38) 22.0 (16–36) 25 (16–40) 0.1910Lifetime years of homelessness, median (IQR) 3.2 (1.0–6.7) 3.0 (1.0–6.7) 3.3 (1.0–6.7) 0.6692Own place 511 (95.7) 240 (94.5) 271 (96.8) 0.1915Subsidized housing 213 (40.8) 91 (36.6) 122 (44.7) 0.0587Residence Type 0.2207Own house, apartment 180 (32.1) 89 (33.1) 91 (31.2)Stay with friends and/or relatives 13 (2.3) 8 (3.0) 5 (1.7)Rooming house/SRO 357 (63.6) 164 (61.0) 193 (66.1)Institution/Other 11 (2.0) 8 (3.0) 3 (1.0)Housing quality, mean (SD) 27.53 (8.2) 27.4 (8.2) 27.7 (8.2) 0.6849Social support network size 0.32890 95 (17.3) 45 (17.0) 50 (17.5)1 61 (11.1) 23 (8.7) 38 (13.3)2–3 104 (18.9) 50 (18.9) 54 (19.0)≥4 290 (52.7) 147 (55.5) 143 (50.2)aPercentages based on complete data; percentages provided by columnbP-value calculated from t-test or Wilcoxon rank-sum test for continuous variables, chi-square test or Fisher’s exact test for categorical variablesTo et al. BMC Public Health  (2016) 16:1041 Page 8 of 12individuals living in Vancouver. The reasons for thisobservation are unclear, but these findings may be attrib-uted to differences that could not be captured betweenparticipants at different study sites and the availability ofhousing and social services in each respective city. Simi-lar to prior studies [20, 21], those who reported a higherproportion of time spent homeless before the baselineinterview were also more likely to experience homeless-ness during the follow-up period.Moderate to severe drug use problems were inde-pendently associated with experiencing homelessnessin the follow-up period, with severe drug use prob-lems significantly associated with the greatest likeli-hood of homelessness. Those who reported severedrug use problems were almost two times more likelyto experience homelessness during the follow-upperiod compared with participants who reported nodrug use. These findings are consistent with previousresearch that identified drug use as the greatest riskfactor for housing instability [21]. Findings from thecurrent study additionally suggest a dose-dependentrelationship of drug use and homelessness, with moresevere drug use problems being linked to an in-creased likelihood of subsequent homelessness amongvulnerably housed adults.Despite adjustment, participants who had less than ahigh school education, had 3 or more chronic healthconditions, or reported higher housing quality were sig-nificantly less likely to experience homelessness over thefollow-up period. While the association between attain-ing less than a high school education and decreased like-lihood of homelessness was unexpected [6], it is similarto a previous study which found that high school com-pletion was associated with recurrent homelessnessamong adults who had experienced prior homelessness[20]. The underlying explanation for this finding is un-clear, but may in part, be due to the higher likelihood ofindividuals with lower education levels to have a learn-ing disorder or experience unemployment and subse-quently qualify for social assistance programs.Paradoxically, these vulnerably housed individuals maybe less likely to experience homelessness while thosewith a high school education may be eligible for moreemployment opportunities, but may also be more likelyto experience recurrent homelessness [20].Vulnerably housed adults who had 3 or more chronichealth conditions were less likely to experience home-lessness over the follow-up period. This finding couldalso be attributed to the fact that individuals living withmultiple chronic conditions may be more likely to beFig. 1 Housing transitions for 561 vulnerably housed participants in 3 Canadian cities over a 3-year follow-up periodTo et al. BMC Public Health  (2016) 16:1041 Page 9 of 12eligible for financial assistance programs and access tohealth and social services. They may also be given prior-ity for subsidized and supportive housing because oftheir chronic medical conditions.Those who reported higher housing quality were alsoless likely to experience homelessness over the follow-upperiod. This is an important finding that has not been pre-viously reported and has implications for efforts to pre-vent homelessness. It appears that individuals living inhigher quality housing characterized by aspects such ascomfort, safety, privacy, and spaciousness were more likelyto remain in that housing. Previous research suggests thathousing quality concerns are common among vulnerablyhoused populations, with up to 85 % of affordable housingproperties having at least one health-related housing qual-ity issue [34]. Our findings suggest that low housing qual-ity may be a modifiable protective factor for homelessness.Thus, improving the quality of low-cost housing maydecrease the likelihood that vulnerably housed individualsbecome homeless in the future.The study has service provision and public health im-plications, highlighting the prevalence of homelessnessamong vulnerably housed individuals and the need forscreening and treating modifiable risk factors among thispopulation. Specifically, the findings highlight the im-portance of connecting individuals with addictions treat-ment to potentially reduce the risk of subsequenthomelessness. The study also found that housing qualitymay be a protective factor against homelessness amongvulnerably housed individuals, suggesting that this popu-lation may benefit from targeted efforts to improvehousing quality in domains such as comfort, privacy,spaciousness, and safety which in turn, may help preventsubsequent homelessness. This can include assistingindividuals to access subsidized housing and rent supple-ments to achieve housing stability [9, 35].LimitationsThe study has several limitations which may restrict theinterpretation of its findings. The study sample wasTable 3 Multivariable GEE logistic regression model of characteristics associated with experiencing homelessness during a 3-yearfollow-up period among vulnerably housed participants in 3 Canadian citiesCharacteristic Core Model Adjusted Odds Ratio (95 % CI) Final Model Adjusted Odds Ratio (95 % CI) P valueCityOttawa 1.66 (1.16–2.37) 1.88 (1.31–2.70) <0.001Toronto 1.04 (0.71–1.54) 1.15 (0.77–1.71) 0.497Vancouver 1 1 –For every year of age 0.98 (0.97–1.00) 0.99 (0.97–1.00) 0.126GenderMale 1.64 (1.19–2.28) 1.59 (1.14–2.21) 0.007Female 1 1 –Highest level of educationLess than high school 0.65 (0.48–0.87) 0.68 (0.50–0.91) 0.009More than high school 1 1 –Chronic conditions1 0.66 (0.38–1.15) 0.63 (0.35–1.11) 0.1102 0.81 (0.47–1.41) 0.83 (0.47–1.45) 0.505≥3 0.60 (0.36–1.01) 0.55 (0.33–0.94) 0.029None 1 1 –Per 10 % of time spent homeless prior to baseline 1.07 (1.02–1.12) 1.06 (1.01–1.11) 0.018Per interval year of follow-up 0.96 (0.85–1.08) 0.96 (0.85–1.09) 0.515Per unit housing quality – 0.99 (0.97–1.00) 0.048DAST risk levelLow – 1.29 (0.90–1.84) 0.164Moderate – 1.40 (1.00–1.97) 0.049Substantial – 1.71 (1.20–2.44) 0.006Severe – 1.98 (1.22–3.20) 0.003No drug use – 1 –To et al. BMC Public Health  (2016) 16:1041 Page 10 of 12limited to single adults and the sampling strategy maynot have been fully representative of the entire vulner-ably housed population. Since data on demographiccharacteristics, health conditions, and housing were col-lected by self-report, accuracy may have been affected byrecall and other sources of reporting bias. The study didnot examine personality traits of participants whichcould contribute to both increased risk of substance useand likelihood of experiencing homelessness. In addition,participants who had severe drug use problems may nothave completed the survey. Thus, the relationship be-tween drug use problems and subsequent homelessnessmay have revealed an even stronger association.Future studies should investigate risk factors for recurrenthomelessness and examine predictors of homelessness overa longer follow-up period. In addition, interventions to pre-vent homelessness among vulnerably housed individualsshould be explored given the potential public healthimplications.ConclusionsThe study followed housing trajectories of vulnerablyhoused adults and found that 29 % of the study sampleexperienced homelessness over a 3-year period. Thestudy also identified risk factors for homelessness amongvulnerably housed individuals such as male gender,higher percentage of time spent homeless prior to base-line, and moderate to severe drug use problems. Protect-ive factors included having 3 or more chronic conditionsand higher housing quality. The study has importantpublic health implications, highlighting the need foraddictions treatment and efforts to improve housingquality among vulnerably housed individuals, which mayhelp prevent subsequent experiences of homelessness inthis population.AbbreviationsAOR: Adjusted odds ratio; AUDIT: Alcohol Use Disorders Identification Test;DAST-10: 10-item Drug Abuse Screening Test; GEE: Generalized estimatingequations; IQR: Interquartile range; MCS: Mental Component Summary;PCS: Physical Component Summary; SD: Standard deviation; SF-12: 12-itemShort Form Health Survey; SRO: Single room occupancy; SSNI: Social SupportNetwork Inventory; TBI: Traumatic brain injuryAcknowledgementsWe would like to acknowledge the following individuals from ourcommunity partner organizations: Laura Cowan, Liz Evans, Sarah Evans,Stephanie Gee, Clare Haskel, and Erika Khandor. The authors also thank thestudy coordinators and interviewers in each of the three cities as well as theshelter, drop-in, and municipal and provincial staff for their assistance withparticipant recruitment and follow-up. The Centre for Urban Health Solutionsin the Li Ka Shing Knowledge Institute at St. Michael’s Hospital gratefullyacknowledges the support of the Ontario Ministry of Health and Long-TermCare. The views expressed here are the views of the authors and do notnecessarily reflect the views of any of the above named organizations.FundingFunding for the Health and Housing in Transition (HHiT) study was receivedfrom an operating grant (MOP-86765) and an Interdisciplinary CapacityEnhancement Grant on Homelessness, Housing and Health (HOA-80066)from the Canadian Institutes of Health Research. The funding body had norole in the design of the study and collection, analysis, and interpretation ofdata and in writing the manuscript.Availability of data and materialsThe datasets generated and/or analysed during the current study areavailable from the corresponding author on reasonable request.Authors’ contributionsMJT helped conceived of the study and its design, coordinated andparticipated in data collection, analysis and interpretation of the data, anddrafted the manuscript. AP helped conceive of the study and its design andinterpretation of the data. TA helped conceive of the study and its designand interpretation of the data. RN carried out statistical analyses andinterpretation of the data. EG coordinated and participated in data collectionand interpretation of the data. AG participated in interpretation of the data.RC participated in interpretation of the data. SF participated in interpretationof the data. VM carried out statistical analyses and interpretation of the data.SWH helped conceive of the study and its design, interpretation of the data,and supervised the study. All authors read, revised, and approved the finalmanuscript.Competing interestsThe authors declare that they have no competing interests.Ethics approval and consent to participateThe Research Ethics Boards at the University of Ottawa; St. Michael’s Hospital,Toronto; and the University of British Columbia, Vancouver approved thisstudy. All study participants provided written informed consent.Author details1Centre for Urban Health Solutions, Li Ka Shing Knowledge Institute, St.Michael’s Hospital, 30 Bond Street, Toronto, ON M5B 1W8, Canada. 2Centrefor Health Evaluation and Outcome Sciences, Division of General InternalMedicine, University of British Columbia, Vancouver, BC, Canada. 3Universityof Ottawa, Ottawa, ON, Canada. 4Royal Ottawa Health Care Group, Ottawa,ON, Canada. 5Division of General Internal Medicine, Department of Medicine,University of Toronto, Toronto, ON, Canada.Received: 10 April 2016 Accepted: 23 September 2016References1. The 2012 Point-in-Time Estimates of Homelessness. Annual HomelessnessAssessment Report. Washington, D.C.: The U.S. Department of Housing andUrban Development; 2012. p. 2012.2. The state of homelessness in Canada. 2013. Available at: http://homelesshub.ca/sites/default/files/SOHC2103.pdf. Accessed 16 Jan 2016.3. Frankish CJ, Hwang SW, Quantz D. Homelessness and health in Canada:research lessons and priorities. Can J Public Health. 2005;96 Suppl 2:S23–9.4. Hwang SW. Homelessness and health. CMAJ. 2001;164(2):229–33.5. Aubry T, Klodawsky F, Coulombe D. Comparing the housing trajectories ofdifferent classes within a diverse homeless population. Am J Commun Psychol.2012;49(1–2):142–55.6. Phinney R, Danziger S, Pollack HA, Seefeldt K. Housing instability amongcurrent and former welfare recipients. Am J Public Health. 2007;97(5):832–7.7. Caton CL, Dominguez B, Schanzer B, Hasin DS, Shrout PE, Felix A, et al. Riskfactors for long-term homelessness: findings from a longitudinal study offirst-time homeless single adults. Am J Public Health. 2005;95(10):1753–9.8. Folsom DP, Hawthorne W, Lindamer L, Gilmer T, Bailey A, Golshan S, et al.Prevalence and risk factors for homelessness and utilization of mentalhealth services among 10,340 patients with serious mental illness in a largepublic mental health system. Am J Psychiatry. 2005;162(2):370–6.9. Shinn M, Weitzman BC, Stojanovic D, Knickman JR, Jiménez L, Duchon L, etal. Predictors of homelessness among families in New York City: fromshelter request to housing stability. Am J Public Health. 1998;88(11):1651–7.10. Caton CL, Hasin D, Shrout PE, Opler LA, Hirshfield S, Dominguez B, etal. Risk factors for homelessness among indigent urban adults with nohistory of psychotic illness: a case–control study. Am J Public Health. 2000;90(2):258–63.To et al. BMC Public Health  (2016) 16:1041 Page 11 of 1211. Washington DL, Yano EM, McGuire J, Hines V, Lee M, Gelberg L. Risk factorsfor homelessness among women veterans. J Health Care Poor Underserved.2010;21(1):82–91.12. Ran MS, Chan CL, Chen EY, Xiang MZ, Caine ED, Conwell Y. Homelessnessamong patients with schizophrenia in rural China: a 10-year cohort study.Acta Psychiatr Scand. 2006;114(2):118–23.13. Shelton KH, Taylor PJ, Bonner A, van den Bree M. Risk factors forhomelessness: evidence from a population-based study. Psychiatr Serv.2009;60(4):465–72.14. Kemp PA, Neale J, Robertson M. Homelessness among problem drug users:prevalence, risk factors and trigger events. Health Soc Care Community.2006;14(4):319–28.15. Edens EL, Kasprow W, Tsai J, Rosenheck RA. Association of substance useand VA service-connected disability benefits with risk of homelessnessamong veterans. Am J Addict. 2011;20(5):412–9.16. Odell SM, Commander MJ. Risks factors for homelessness amongpeople with psychotic disorders. Soc Psychiatry Psychiatr Epidemiol.2000;35(9):396–401.17. Orwin RG, Scott CK, Arieira C. Transitions through homelessness and factorsthat predict them: three-year treatment outcomes. J Subst Abuse Treat.2005;28 Suppl 1:S23–39.18. van den Berk-Clark C, McGuire J. Elderly Homeless Veterans in Los Angeles:Chronicity and Precipitants of Homelessness. Am J Public Health. 2013;103(Suppl 2):S232–38.19. Hwang SW, Aubry T, Palepu A, Farrell S, Nisenbaum R, Hubley AM, et al. Thehealth and housing in transition study: a longitudinal study of the health ofhomeless and vulnerably housed adults in three Canadian cities. Int J PublicHealth. 2011;56(6):609–23.20. McQuistion HL, Gorroochurn P, Hsu E, Caton CL. Risk factors associated withrecurrent homelessness after a first homeless episode. Community MentHealth J. 2014;50(5):505–13.21. O’Connell MJ, Kasprow W, Rosenheck RA. Rates and risk factors forhomelessness after successful housing in a sample of formerly homelessveterans. Psychiatr Serv. 2008;59(3):268–75.22. Ardilly P, Le Blanc D. Sampling and weighting a survey of homelesspersons: a French example. Survey Methodology. 2001;27(1):109–18.23. Housing, Family and Social Statistics Division. Ethnic diversity surveyquestionnaire. Ottawa, ON. Statistics Canada. 2002; Available at: http://www5.statcan.gc.ca/bsolc/olc-cel/olc-cel?catno=89M0019XCB&lang=eng.Accessed 16 Jan 2016.24. Statistics Canada. Canadian community health survey (CCHS). 2012.Available at: http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3226&lang=en&db=imdb&adm=8&dis=2.Accessed Jan 16, 2016.25. Slaughter B, Fann JR, Ehde D. Traumatic brain injury in a county jailpopulation: prevalence, neuropsychological functioning and psychiatricdisorders. Brain Inj. 2003;17(9):731–41.26. Ware JE, Kosinski M, Keller SD. SF-12: How to score the SF-12 physical andmental health summary scales. 2nd ed. Boston (MA): The Health Institute,New England Medical Center; 1995.27. Babor TF, Higgins-Biddle JC, Saunders JB. AUDIT: the alcohol use disordersidentification test: guidelines for use in primary care. Geneva: World HealthOrganization Department of Mental Health and Substance Dependence; 2001.28. French MT, Roebuck MC, McGeary KA, Chitwood DD, McCoy CB. Using thedrug abuse screening test (DAST-10) to analyze health services utilizationand cost for substance users in a community-based setting. Substance UseMisuse. 2001;36(6):925–46.29. Toro PA, Bellavia CW, Daeschler CV, Owens BJ, Wall DD, Passero JM, et al.Distinguishing homelessness from poverty: a comparative study.J Consult Clin Psychol. 1995;63(2):280–9.30. Flaherty JA, Gaviria FM, Pathak DS. The measurement of social support: theSocial Support Network Inventory. Compr Psychiatry. 1983;24(6):521–9.31. Tsemberis S, McHugo G, Williams V, Hanrahan P, Stefancic A. Measuringhomelessness and residential stability: the residential time-line follow-backinventory. J Community Psychol. 2007;35(1):29–42.32. Rotnitzky A, Jewell NP. Hypothesis testing of regression parameters insemiparametric generalized linear models for cluster correlated data.Biometrika. 1990;77:485–97.33. Shults J, Sun W, Tu X, Kim H, Amsterdam J, Hilbe JM, et al. A comparison ofseveral approaches for choosing between working correlation structures ingeneralized estimating equation analysis of longitudinal binary data.Stat Med. 2009;28(18):2338–55.34. Klein EG, Keller B, Hood N, Holtzen H. Affordable housing and health: ahealth impact assessment on physical inspection frequency. J Public HealthManag Pract. 2015;21(4):368–74.35. Rog DJ, Marshall T, Dougherty RH, George P, Daniels AS, Ghose SS, et al.Permanent supportive housing: assessing the evidence. Psychiatr Serv.2014;65(3):287–94.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:To et al. BMC Public Health  (2016) 16:1041 Page 12 of 12

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0362014/manifest

Comment

Related Items