“Street Sweeps”: The Municipal Government-Enforced Confiscation of Personal Belongings Among Unstably Housed People Who Use Drugs in Vancouver, Canada

Abstract

Objectives:

Street sweeps, involving the state-enforced removal of makeshift residences and confiscation of personal belongings from people in public spaces, are a common tool employed by urban governments to address public health and safety concerns. Amidst the ongoing housing and toxic drug poisoning crises in Vancouver, Canada, we sought to characterize experiences of confiscation of personal belongings by city workers among unstably-housed people who use drugs people who use drugs.

Study Design:

Cohort study.

Methods:

We used multivariable generalized estimating equations models to longitudinal data derived from unstably-housed people who use drugs (including those reporting homelessness and living in single room occupancy hotels) participating in three harmonized cohort studies of community-recruited people who use drugs in Vancouver in 2021–2023.

Results:

In total, 13.6% of 691 eligible participants (and 23.6% of 233 reporting homelessness) reported municipal government-enforced belonging confiscation in the past six months at least once. In multivariable analyses, violent victimization (adjusted odds ratio [AOR]= 2.14; 95% confidence interval [CI]: 1.27, 3.60) and inability to access health/social services (AOR=2.19, 95% CI: 1.32, 3.65) were significantly and positively associated with belonging confiscation, and so was non-fatal overdose (AOR=1.94, 95% CI: 1.01, 3.74) among those reporting homelessness.

Conclusion:

Findings underscore that confiscation of belongings was relatively widespread among our cohort of people who use drugs and concentrated among individuals who are structurally marginalized, emphasizing the need to end street sweeps and prioritize development of dignified housing and harm reduction policies involving affected communities in decision-making.

1. INTRODUCTION

In recent years, many urban areas worldwide have seen a substantial rise in people sheltering outdoors and in otherwise precarious locales, a reflection of deeper systemic issues such as social-housing policies and profit-driven housing practices^{1, 2, 3}. This situation has been further exacerbated by the COVID-19 pandemic and related regulations as market housing has become untenable for an increased number of people^{4, 5}. In British Columbia (BC), the number of people experiencing homelessness has steadily increased over the past decade. According to the 2023 Homeless Count, the number of individuals sheltering outdoors in Vancouver increased by 51% since 2013⁶, with over 4,800 people experiencing homelessness in Metro Vancouver alone⁶. Of these, approximately one-third were unsheltered⁶. This increase coincides with a long-standing shortage of affordable housing in the region. The compounding effects of the COVID-19 pandemic and the policy response to it as well as a deepening affordable housing shortage have not only worsened existing housing problems but also highlighted ineffective governance strategies and a lack of appropriate housing support for marginalized populations.

Since 2008, street sweeps have been part of the City of Vancouver’s efforts to remove visible manifestations of homelessness^7,8. These actions—distinct from routine sanitation—typically involve city engineering workers and park rangers, often accompanied by police officers, clearing sidewalks and parks of makeshift residences and personal belongings using tools like pitchforks⁸.In recent years, the City of Vancouver relied on a number of intersecting municipal bylaws and policies, including Street and Traffic By-law, to justify clearing encampments and items deemed hazardous or obstructive by the city workers and the law enforcement officers that accompany them.^{8, 10, 11, 12}. The practice and nature of street sweeps have evolved in recent years but they have been typically carried out with few, if any, alternatives offered to affected individuals^{8, 9}. Despite community-driven measures to end police presence at street sweeps, these practices persist.

Street sweeps have provoked human right concerns across Canada^{8, 9}. Street sweeps are commonly justified by state actors as efforts to maintain public order in many jurisdictions beyond Vancouver as well^{13, 14}. While state actors often cite largely undefined “public safety” concerns as the rationale behind these actions, people impacted and other critics argue that such actions disproportionately and negatively impact already marginalized populations, including those sheltering outdoors and using unregulated drugs^{14, 15}. The removal of personal belongings, including essential items, medications and irreplaceable personal artifacts, further destabilizes individuals who are already struggling to meet their basic needs^{8, 15}.

Confiscation of personal belongings during street sweeps can have profound impacts on people, including those who use drugs. The confiscation of personal items—such as drugs, harm reduction supplies (e.g., sterile needles, naloxone kits), medications, and identification documents—not only increases health outcomes and disrupts treatment adherence but also exacerbates vulnerability to overdose, infection, and other harms^{15, 16, 17, 18}. These losses can severely undermine the mental health of affected individuals, further marginalizing those who are already precariously housed¹⁵.

Displaced individuals often have limited options for relocation, with many forced to seek alternative shelter in already overcrowded or inaccessible shelters^{9, 15}. Others are left to set up encampments in the same or nearby locations where someone was previously forced to move, or in less visible or accessible areas, thus exacerbating overdose risk and perpetuating cycles of instability and marginalization^{8, 15, 19, 20}. In February 2024, Canada’s Federal Housing Advocate labeled forced encampment evictions as human rights violations, urging all levels of government to immediately cease forced decampment²¹.

Given that unstably housed populations are disproportionately impacted by harms associated with the increasingly toxic drug supply²², there is a growing recognition of the need to examine the impacts of street sweeps on the health and well-being of unstably housed people who use drugs^{14, 15, 23}. The ongoing toxic drug crisis in both Canada and the US carries severe consequences for communities, as toxic drug deaths became the leading cause of death among individuals aged 10–59 in BC in 2023^{24, 25}. While unstable housing is a risk factor for overdose^{15, 26} and emerging qualitative research has documented some harms associated with street sweeps in the US ^{13, 14, 15}, there is a paucity of quantitative empirical studies that examine the scale and nature of street sweeps among unstably housed individuals in Canada during the post-pandemic period. For example, Barocas et al. estimated substantial increases in drug-related morbidity and mortality associated with involuntary displacement among people who inject drugs and experience homelessness in the US; however, it was a simulation study¹⁶. A cross-sectional study of people experiencing homelessness in Denver, Colorado in 2018–2019 found that involuntary displacement was linked to self-reported infectious disease acquisition, substance use, and declines in mental health²⁷. However, this study was conducted in the US prior to the COVID-19 pandemic emergence and not specific to people who use drugs. Another U.S.-based cross-sectional survey of people who inject drugs in 2018–2019 found that residential relocation was associated with increased odds of violence, non-fatal overdose, syringe sharing and severe food insecurity¹⁸, but it did not specifically examine the impacts of street sweeps. Our study builds on these prior studies and focus on unstably housed people who use drugs and the confiscation of personal belongings – a common and understudied part of experiencing involuntary displacement. Further, the focus on the post-pandemic period is of significance given the worsened housing instability, the greater visibility and scale of outdoor encampments ^{4, 5, 6}, and the corresponding municipal government responses in Canada—including the frequency and intensity of street sweeps^{8, 14, 28}. Moreover, public and political discourse around homelessness and public space use has intensified, prompting evolving enforcement strategies that may disproportionately affect marginalized populations, including people who use drugs^{8, 15, 27}. Therefore, we sought to estimate the prevalence and correlates of having personal belongings confiscated by city workers among unstably housed people who use drugs in Vancouver in 2021–2023.

2. METHODS

2.1. Study Setting, Design, and Participants

Data were derived from three ongoing open prospective cohort studies of people who use drugs in Vancouver: the Vancouver Injection Drug Users Study (VIDUS), the AIDS Care Cohort to evaluate Exposure to Survival Services (ACCESS), and the At-Risk Youth Study (ARYS)²⁹. Previous publications have provided descriptions of these cohorts^{29, 30, 31}. In brief, VIDUS enrolls adults (≥18 years old) who are HIV at-risk and have injected drugs within the month prior to enrollment. ACCESS recruits adults living with HIV, while ARYS enrolls street-involved youth aged 14–26 who have engaged in unregulated drug use in the month prior to enrollment²⁹. Recruitment occurs through word-of-mouth and street outreach efforts, primarily in two Vancouver neighborhoods: Downtown Eastside (DTES) for VIDUS and ACCESS, and Downtown South for ARYS²⁹. The DTES is a neighbourhood in Vancouver severely impacted by the ongoing housing and toxic drug poisoning crises⁸. In recent years, particularly since 2022, the DTES has been subject to significant enforcement measures, driven by cited concerns of fire risk and public safety^{8, 21}. However, using fire safety as a rationale for evictions has been contested, with local advocacy groups and residents challenging a fire order injunction as unjust, arguing that it unfairly targets encampments without addressing underlying issues^{10, 21}.

VIDUS was the first of the three cohorts to launch (in 1996), followed by ACCESS and ARYS in 2005^{29, 30, 31}. Participants in each cohort are followed at six-month intervals, and there is no pre-determined study end date. All three studies use harmonized data collection instruments and procedures (e.g., using the same questionnaires, following participants at the same six-month intervals) to enable combined analyses^{29, 30, 31}. Participants receive a CAD $50 honorarium upon completing each study visit. Baseline and semi-annual follow-up assessments involve interviewer-administered questionnaires, covering various domains such as demographics, substance use, healthcare access, and interactions with law enforcement²⁹. Additionally, at each study visit, blood samples are drawn for HIV and HCV serology.

For the present analyses, an analytic sample was derived from VIDUS/ACCESS/ARYS participants who completed at least one interview between December 2021 and May 2023 and who reported current unstable housing (i.e., residing in a single-room occupancy hotel [SRO], shelter, other transitional housing, or on the street), used any drugs in the past six months (excluding exclusive cannabis use), and provided valid responses to the survey question for the outcome. A valid response was defined as answering either ‘yes’ or ‘no’ to the outcome question; responses marked as ‘refused’ (n=4) or left blank (n=12) were considered invalid and excluded. Participants are enrolled in only one cohort at a time. The study period was determined because the question for the outcome was added to the questionnaire in December 2021 and the most recent data available for the analysis at the time of the study extended through May 2023.

2.2. Measures

The primary outcome was municipal government-enforced confiscation of belongings within the past six months (yes vs. no). This variable was ascertained by asking the question, ‘Have you had your possessions confiscated by city workers in the last six months?’ City workers included city engineering workers and park rangers, both with and without police officers. Although participants were not explicitly asked whether they experienced a ‘street sweep,’ street sweeps are virtually the only circumstance in which these city workers would confiscate someone’s personal belongings. Based on the previous literature^{13, 14, 15} and the lived experiences of some co-authors with street sweeps, we considered a range of variables that might be associated with confiscation of belongings²⁸. Demographic variables included: age (continuous, per year increase); self-identified gender, which refers to a person’s deeply felt internal sense of gender and may differ from their assigned sex at birth or physical traits³² (man vs. woman vs. transgender or other non-binary); self-identified ethnicity/ancestry (white vs. Indigenous or other person of colour), and residence in the city of Vancouver. Substance use-related variables included: ≥daily use of unregulated opioids (i.e. heroin, fentanyl, or down [a colloquial term locally used to refer to unregulated opioids]) and unregulated stimulants (i.e. crystal methamphetamine or powder/crack cocaine), respectively; and non-fatal overdose (yes vs. no, with a follow-up question asking the number of overdoses in the past six months). Social and structural exposures included: homelessness (i.e., shelter, or on the street), and self-reported inability to access health and/or social services. Health-related variables included HIV antibody status (positive vs. negative), experiences of physical or sexual violence (as in our previously published studies^{33, 34}, assessed through non-standardized survey items asking whether participants experienced physical or sexual violence, with follow-up questions asking about the perpetrator), as well as levels of depression and anxiety symptoms measured by the Patient-Reported Outcomes Measurement Information System (PROMIS) short forms for depression (8b) and anxiety (7a)³⁵ (moderate to severe vs. mild to none). All variables except for ethnicity/ancestry referred to the six months prior to the interviews and were time-varying and dichotomized as yes vs. no unless otherwise stated.

2.3. Statistical analyses

Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). All p-values were two-sided, and the significance level was set at 5%. Descriptive statistics of baseline sample characteristics were generated, stratified by confiscation of belongings at least once during the study period and using exact chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables.

We used generalized estimating equations (GEE) binary logistic models to identify factors associated with confiscation of belongings to account for the correlated data that may result from multiple visits per participant. Given that those reporting homelessness might have experienced differential impacts of street sweeps compared to those unstably housed but not reporting homelessness (e.g., living in an SRO), we fit three multivariable models: Model 1 using the full sample, Model 2 using a sub-sample of those reporting homelessness and Model 3 using a sub-sample of those not reporting homelessness. The working correlation structure in GEE models was selected based on the lowest quasi-information criterion, with options including independence, exchangeable, or autoregressive structures (the latter two penalized by 2.0). As our primary objective was to assess associations rather than make predictions, we selected explanatory variables that were theoretically or substantively important based on prior research and domain expertise. To enhance model parsimony, we initially included variables associated with the outcome at p < 0.10 in bivariable analyses and subsequently refined the model using a stepwise process guided by the Akaike’s Information Criterion (AIC). Given that independence was identified as the optimal working correlation structure, variables were systematically removed based on largest negative impact on the AIC in models fitted via maximum likelihood estimation (MLE). The process continued until no additional variables could be dropped without increasing AIC. The final selected model was fitted via MLE. Further, as sensitivity analyses, we re-ran the three multivariable models by forwarding all the explanatory variables to the subsequent 6-month cohort follow-up period. This is an exploratory analysis to assess whether confiscation of belongings may be associated with subsequent experiences such as non-fatal overdose, service access, or housing status six or more months later.

To explore the impacts of missing data on our analyses, we performed several procedures. First, to examine whether differential lost to follow-up rates impacted the outcome variable (i.e., confiscation of belongings), we used the Kruskal Wallis test to test for differences in the number of interviews contributing to the analysis vs. confiscation of belongings. Next, to investigate potential impacts of missing analytic variables in each multivariable GEE model, we performed imputation using per-subject carry-forward where available for a subject, otherwise using the sample mean, for each variable.

In a sub-analysis, further univariate descriptive statistics were computed among the last records for each participant where confiscation of belongings was reported. These analyses involved variables including self-reported inability to access housing services, primary care clinics, supervised consumption sites and addiction treatment, respectively, instances of being compelled into addiction treatment or other health services by the police, and experiencing physical violence perpetrated by the police.

3. RESULTS

In total, 691 eligible participants contributed 1264 observations. Of these, 188 were from ACCESS, 178 from ARYS, and 325 from VIDUS. The median number of study visits completed per participant during the study period was 2.0 (1^st and 3^rd quartiles: 1.0–2.0). Among them, 94 (13.6%) reported municipal government-enforced confiscation of belongings in the past six months at least once. Among these participants, the median number of reports of belongings confiscation during the study period was 1.0 (1^st and 3^rd quartiles: 1.0–1.0), with a mean (standard deviation) of 1.1 (0.3). Further, 55 (23.6%) of 233 reporting homelessness ever during the study period reported belongings confiscation at least once. As shown in Table 1 (baseline sample characteristics), the median age was 43 years (1st and 3rd quartiles: 34–54). Additionally, 383 (55.9% of 685 valid responses) individuals self-identified themselves as white, 278 (40.6%) as Indigenous, and 24 (3.5%) as persons of colour. Further, 380 (55.0% of 691 valid responses) participants identified as men, 286 (41.4%) as women, and 25 (3.6%) as transgender/other non-binary. In total, 193 (28.0%) reported homelessness in the past six months at baseline, and the baseline prevalence of confiscation of belongings was higher among individuals reporting homelessness compared to non-homeless individuals (24.5% vs. 9.3%, p=<0.001). Additionally, 151 (22.0%) reported having experienced non-fatal overdose in the past six months at baseline. Among 146 out of those 151 individuals who reported the number of overdose events in the past six months, the median number of overdose events reported was 1 (1^st and 3^rd quartiles: 1–3).

Table 1:

Baseline characteristics of participants who reported municipal government-enforced confiscation of belongings in the past six months among unstably housed people who use drugs in Vancouver, Canada, 2021–2023 (n=691).

Characteristic	Total n (%) 691 (100%)	Reporting confiscation of belongings by city workersa during the study period		p - value
		Yes 94 (13.6%)	No 597 (86.4%)
Age (median, 1st, and 3rd quartiles)
	43 (34 – 54)	35 (29 – 43)	45 (35 – 55)	<0.001
Self-identified ethnicity/ancestry
White	383 (55.9%)	50 (13.1%)	333 (86.9%)	0.961
Indigenous	275 (40.1%)	38 (13.8%)	237 (86.2%)
Person of colour	27 (3.9%)	4 (14.8%)	23 (85.2%)
Self-identified gender a
Man	380 (55.0%)	58 (15.3%)	322 (84.7%)	0.191
Woman	286 (41.4%)	35 (12.2%)	251 (87.8%)
Transgender/other non-binary	25 (3.6%)	1 (4.0%)	24 (96.0%)
Homeless a
yes	193 (28.0%)	48 (24.9%)	145 (75.1%)	<0.001
no	497 (72.0%)	46 (9.3%)	451 (90.7%)
Living in the city of Vancouver a
yes	650 (94.2%)	87 (13.4%)	563 (86.6%)	0.475
no	40 (5.8%)	7 (17.5%)	33 (82.5%)
Inability to access health and/or social services a
yes	194 (28.8%)	41 (21.1%)	153 (78.9%)	<0.001
no	480 (71.2%)	48 (10.0%)	432 (90.0%)
Daily unregulated opioid use a, b
yes	425 (61.6%)	71 (16.7%)	354 (83.3%)	0.003
no	265 (38.4%)	23 (8.7%)	242 (91.3%)
Daily unregulated stimulant use a, c
yes	278 (40.3%)	53 (19.1%)	225 (80.9%)	0.001
no	411 (59.7%)	41 (10.0%)	370 (90.0%)
Using drugs alone (injection or non-injection) a
yes	340 (49.4%)	47 (13.8%)	293 (86.2%)	0.912
no	348 (50.6%)	47 (13.5%)	301 (86.5%)
Sharing syringe (borrowed or lent) a
yes	21 (3.1%)	3 (14.3%)	18 (85.7%)	1.000
no	667 (96.9%)	90 (13.5%)	577 (86.5%)
Non-fatal overdose a
yes	151 (22.0%)	29 (19.2%)	122 (80.8%)	0.032
no	536 (78.0%)	65 (12.1%)	471 (87.9%)
HIV seropositive
yes	188 (27.2%)	15 (8.0%)	173 (92.0%)	0.009
no	503 (72.8%)	79 (15.7%)	424 (84.3%)
Experienced physical or sexual violence a
yes	176 (26.5%)	46 (26.1%)	130 (73.9%)	<0.001
no	489 (73.5%)	44 (9.0%)	445 (91.0%)
Moderate to severe depressive symptoms d
yes	231 (38.8%)	46 (19.9%)	185 (80.1%)	<0.001
no	365 (61.2%)	27 (7.4%)	338 (92.6%)
Moderate to severe anxiety symptoms d
yes	261 (43.9%)	43 (16.5%)	218 (83.5%)	0.012
no	333 (56.1%)	31 (9.3%)	302 (90.7%)

Percentages are calculated based on the valid responses.

^aDenotes activities and events in the past six months.

^bUnregulated opioids include heroin, fentanyl, down, speedball or goofball.

^cUnregulated stimulants include crystal methamphetamine, powder/crack cocaine, speedball or goofball.

^dAscertained by the PROMIS short forms (depression 8b and anxiety 7a).

Among the 94 participants who reported having their belongings confiscated, many indicated various forms of inability to access essential services during their most recent interview, as shown in Table 2. Specifically, 22 (36.1%) reported having tried but been unable to access housing services, 23 (25.6%) primary care clinics, 19 (20.4%) supervised consumption sites, and 16.1% addiction treatment. Additionally, 25 (26.9%) reported experiencing physical violence from the police.

Table 2:

Descriptive statistics of reports of barriers to services and the nature of police encounters at the most recent observations of reporting confiscation of belongings in Vancouver, Canada (n=94).

Characteristic	N (%)
Tried but being unable to access housing servicesa	22 (36.1%)
Tried but being unable to access primary care clinicsa	23 (25.6%)
Tried but being unable to access supervised consumption sitesa	19 (20.4%)
Tried but being unable to access addiction treatmenta	15 (16.1%)
Being forced into addiction treatment or other health services by policea	7 (7.4%)
Experienced physical violence from the policea	25 (26.9%)

Percentages are calculated based on the valid responses.

^adenotes activities and events in the previous six months.

Table 3 presents the results of GEE analyses. In the multivariable analysis of the full sample, inability to access health and social services (adjusted odds ratio [AOR] = 2.19, 95% confidence interval [CI]: 1.32–3.65), experiencing violence (AOR = 2.14, 95% CI: 1.27–3.60) and younger age (AOR = 0.96, 95% CI: 0.93–0.98) were statistically significantly associated with confiscation of belongings. Among the subgroup reporting homelessness, age (AOR = 0.97, 95% CI: 0.94–1.00), ≥daily unregulated stimulant use (AOR = 2.20, 95% CI: 1.20–4.04) and non-fatal overdose (AOR = 1.94, 95% CI: 1.01–3.74) remained statistically significantly associated with confiscation of belongings, and so did experiencing violence (AOR = 2.56, 95% CI: 1.20–5.47) among non-homeless individuals. In sensitivity analyses (Table 4), experiencing violence in the subsequent 6-month cohort follow-up period remained statistically significantly associated with the outcome in all three models.

Table 3:

Bivariable and multivariable generalized estimating equations analyses of factors associated with municipal government-enforced confiscation of belongings among 691 unstably housed people who use drugs in Vancouver, Canada, 2021–2023 (n= 1264 total visits; n=1263 visits with known homelessness status).

Variable	All Sample (n=1264 visits from 691 participants)		Homeless Sample (n=329 visits from 233 participants)		Non-Homeless Sample (n=934 visits from 528 participants)
	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)
Age (per year increase)	0.94 (0.92, 0.96)**	0.96 (0.93, 0.98)**	0.97 (0.94, 1,00)**	0.97 (0.94, 1.00)**	0.95 (0.92, 0.97)**	0.97 (0.93, 1.00)*
Self-identified as white	1.03 (0.69, 1.55)	-	1.06 (0.60, 1.88)	-	0.93 (0.51, 1.70)	-
Self-identified as a man a	1.29 (0.86, 1.94)	-	1.63 (0.87, 3.03)	-	0.70 (0.38, 1.28)	-
Living in the city of Vancouver a	0.61 (0.27, 1.38)	-	0.89 (0.35,2.27)	-	1.22 (0.16, 9.23)	-
Inability to access health and/or social services a	2.81 (1.85, 4.25)**	2.19 (1.32, 3.65)**	1.33 (0.72, 2.44)	-	2.51 (1.34, 4.69)**	2.07 (0.96, 4.45)*
Daily unregulated opioid use a, b	1.79 (1.13, 2.84)**	-	1.34 (0.70, 2.58)	-	1.77 (0.90, 3.48)*	-
Daily unregulated stimulant use a, c	1.41 (0.94, 2.10)*	-	2.20 (1.24, 3.90)**	2.20 (1.20, 4.04)**	0.75 (0.40, 1.43)	-
Non-fatal overdose a	1.67 (1.07, 2.61)**	-	1.72 (0.94, 3.13)*	1.94 (1.01, 3.74)**	1.23 (0.60, 2.53)	-
HIV seropositive	0.57 (0.34, 0.96)**	-	1.07 (0.49, 2.36)	-	0.57 (0.27, 1.20)	-
Experienced physical or sexual violence a	3.35 (2.20, 5.09)**	2.14 (1.27, 3.60)**	1.93 (1.07, 3.48)**	1.59 (0.86, 2.95)	4.06 (2.18, 7.55)**	2.56 (1.20, 5.47)**
Moderate to severe depressive symptoms d	2.44 (1.52, 3.93)**	1.45 (0.87, 2.42)	1.67 (0.85, 3.27)	-	2.85 (1.42, 5.73)**	1.68 (0.79, 3.58)
Moderate to severe anxiety symptoms d	2.20 (1.37, 3.54)**	-	1.51 (0.77, 2.96)	-	2.60 (1.29, 5.23)**	-

AOR: adjusted odds ratio. CI: confidence interval. OR: odds ratio.

^aDenotes activities and events in the past six months.

^bUnregulated opioids include heroin, fentanyl, down, speedball or goofball.

^cUnregulated stimulants include crystal methamphetamine, powder/crack cocaine, speedball or goofball.

^dAscertained by the PROMIS short forms (depression 8b and anxiety 7a).

^*denotes p<0.10

^**denotes p<0.05

Table 4:

Bivariable and multivariable generalized estimating equations analyses of factors associated with municipal government-enforced confiscation of belongings among 691 unstably housed people who use drugs in Vancouver, Canada, 2021–2023 (n= 1264 total visits; n=1263 visits with known homelessness status).

Variable (forwarded by one 6-month cohort follow-up period)	All Sample (n=1264 visits from 691 participants)		Homeless Sample (n=329 visits from 233 participants)		Non-Homeless Sample (n=934 visits from 528 participants)
	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)
Age (per year increase)	0.93 (0.91, 0.96)**	0.94 (0.91, 0.97)**	0.96 (0.92, 0.99)**	0.96 (0.92, 1.00)*	0.93 (0.90, 0.97)**	0.95 (0.91, 1.00)**
Self-identified as white	1.00 (0.59, 1.69)	-	0.94 (0.43, 2.05)	-	0.92 (0.44, 1.91)	-
Self-identified as a man a	1.01 (0.60, 1.70)	-	1.58 (0.67, 3.75)	-	0.49 (0.23, 1.07)*	-
Living in the city of Vancouver a	2.18 (0.29, 16.22)	-	2.39 (0.30, 18.86)	-	e	-
Inability to access health and/or social services a	2.10 (1.22, 3.64)**	1.76 (0.89, 3.48)	1.33 (0.57, 3.08)	-	1.90 (0.89, 4.06)*	1.40 (0.56, 3.46)
Daily unregulated opioid use a, b	1.38 (0.79, 2.43)	-	0.95 (0.43, 2.12)	-	1.80 (0.79, 4.12)	-
Daily unregulated stimulant use a, c	1.14 (0.67, 1.94)	-	1.49 (0.68, 3.27)	1.17 (0.48, 2.81)	1.09 (0.51, 2.29)	-
Non-fatal overdose a	1.39 (0.76, 2.56)	-	1.33 (0.55, 3.22)	1.56 (0.59, 4.16)	1.30 (0.55, 3.09)	-
HIV seropositive	0.36 (0.16, 0.80)**	-	0.65 (0.19, 2.31)	-	0.36 (0.12, 1.03)*	-
Experienced physical or sexual violence a	3.33 (1.91, 5.81)**	3.79 (1.89, 7.57)**	3.32 (1.43, 7.70)**	2.81 (1.18, 6.74)**	2.83 (1.32, 6.08)**	2.71 (1.09, 6.74)**
Moderate to severe depressive symptoms d	2.54 (1.34, 4.78)**	1.32 (0.66, 2.64)	1.01 (0.38, 2.69)	-	4.19 (1.75, 10.06)**	2.51 (0.98, 6.41)*
Moderate to severe anxiety symptoms d	2.77 (1.43, 5.36)**	-	1.11 (0.41, 2.96)	-	4.62 (1.80, 11.89)**	-

AOR: adjusted odds ratio. CI: confidence interval. OR: odds ratio.

In the sensitivity analyses, all explanatory variables were derived from the subsequent cohort follow-up interviews.

^aDenotes activities and events in the past six months.

^bUnregulated opioids include heroin, fentanyl, down, speedball or goofball.

^cUnregulated stimulants include crystal methamphetamine, powder/crack cocaine, speedball or goofball.

^dAscertained by the PROMIS short forms (depression 8b and anxiety 7a).

^eThe model did not converge.

^*denotes p<0.10

^**denotes p<0.05

The Kruskal Wallis test showed that there was no difference in the number of interviews contributing to the analysis vs. confiscation of belongings (p=0.217). Further, the three primary multivariable GEE models (for the full, homeless, and non-homeless samples) had 16.8%, 7.0%, and 13.6% missing observations, respectively, while the models used in the sensitivity analyses had 38.1%, 38.3% and 34% missing observations, respectively. The models using the imputed datasets yielded comparable effect estimates in all models, with highly consistent direction, magnitude and significance of effects.

4. DISCUSSION

Approximately one in seven unstably housed people who use drugs in our study (and almost a quarter of people who use drugs reporting homelessness) experienced municipal government-enforced confiscation of belongings at least once in 2021–2023. Our results indicate that nonfatal overdose (among participants reporting homelessness), younger age and experiencing physical and sexual violence were statistically significantly associated with municipal government-enforced confiscation of belongings.

The statistically significant association between non-fatal overdose and the confiscation of belongings among those reporting homelessness is of significant concern. While our model treated non-fatal overdose as an independent variable, the cross-sectional nature of the data means this relationship may be bidirectional, and the findings should be interpreted as associations rather than evidence of a specific causal direction. There are a couple of possible interpretations of this association. This finding may suggest that individuals facing confiscations may be forced into withdrawal and/or pushed into different drug markets where they do not have access to their regular drug suppliers to replenish confiscated drugs^{14, 15}. With their belongings taken away, individuals may also feel an increased sense of desperation, distress and instability, potentially leading them to seek comfort in cheaper but more toxic drugs obtained from unfamiliar sources, thereby heightening their susceptibility to overdose^{14, 15}. The loss of personal belongings not only disrupts individuals’ lives but also limits their ability to access safer alternatives and potentially naloxone, intensifying their vulnerability to substance-related harms^{14, 15}. Alternatively, those who experienced non-fatal overdose might have been more likely to shelter in visibly crowded tent cities where people who use drugs intervene to reverse each other’s overdoses, and in turn, more likely to be subject to street sweeps. Indeed, anecdotal reports indicate that the Park Board uses the presence of harm reduction materials or drug paraphernalia as a reason to confiscate individuals’ tents^{8, 36}. In either scenario, our study results indicate that confiscation of belongings is likely implicated in the ongoing toxic drug crisis among those experiencing homelessness.

We also found a significant association between experiences of violence and confiscation of belongings. The findings may indicate that the act of confiscation of belongings may exacerbate tensions between law enforcement actors and unstably-housed people who use drugs¹⁵. Indeed, we also found that among those who had their belongings confiscated, one in four reported experiencing physical violence from the police. Additionally, the loss of valuables may increase the risk of violence in drug markets, especially if debts to workers in the unregulated drug market cannot be paid¹⁵. The results of sensitivity analyses also suggest that such heightened risk of violence may persist at least six months after experiencing confiscation of belongings. Furthermore, experiences of violence might force individuals into more remote or hidden settings, making them less visible and reducing the likelihood that someone will respond in the event of an overdose.

It is also concerning that those who experienced confiscation of belongings faced substantial barriers to accessing essential services, including housing services, primary care, and harm reduction services. Street sweeps may interfere with service access as people may be physically displaced and face more difficulty accessing services due to geographic distance, mobility issues, or unfamiliarity with nearby services when an individual is moved to a new location. Additionally, mistrust in the system, originating from past experiences of structural neglect or mistreatment, can prevent individuals from seeking assistance (sometimes referred to as care avoidance³⁷), contributing to people experiencing continued vulnerability to violence^{14, 15, 17}. This mistrust can also interfere with housing outreach services, which may lose track of individuals when they are displaced, further complicating efforts to provide support and thus exacerbating their vulnerability^{28, 38}. Our findings also suggest a vicious cycle, wherein the inability to access housing services heightens vulnerability to the confiscation of belongings, which in turn leads to additional barriers to accessing services, including housing^{14, 39}. Addressing these underlying structural factors and improving access to essential services is critical for promoting the safety and well-being of unstably housed individuals in Vancouver.

Our study has several limitations. First, the not-randomly-selected sample affects the generalizability of our findings. In particular, unstably-housed people who use drugs who were the most severely impacted by street sweeps might have also experienced barriers to participating in research. Therefore, the prevalence of experiencing confiscation of belongings is likely underestimated in our study. Second, the reliance on self-reported data introduces potential reporting bias. Third, the observational nature of the study limits our ability to infer causation and ascertain the temporal relationships between belonging confiscation and health outcomes or service access. Although the sensitivity analyses partially addressed this concern, this part of the analyses are exploratory in nature. Future research should employ more rigorous approaches (e.g., time-to-event analyses) to identify the impacts of belonging confiscation on each of the subsequent health outcomes and service access.

This study highlights the need to address the multifaceted needs of unstably housed people who use drugs in urban settings. First, our study findings echo previous calls for ending street sweeps as made by affected communities and human rights advocates^{8, 14, 40}. While dignified and adequate housing and Housing First initiatives are alternative approaches to mitigate the adverse effects of homelessness^{14, 41}, it is important to prioritize the needs and preferences of those affected when developing new housing initiatives. This is especially important because many housing strategies, including existing shelters and modular housing, often exacerbate the very challenges they aim to resolve, displacing marginally housed individuals further or leaving them without adequate support²⁸. Further, considering the ongoing occurrence of street sweeps, implementing more immediate risk mitigation measures is also of significance. For example, it is notable that street sweeps typically do not involve the issuance of receipts or documentation, limiting the legal or material recourse that affected people can access^{8, 40}. Therefore, as long as street sweeps continue, mandatory information should be provided on how to retrieve belongings, including receipts and documentation, if items are taken. Secondly, provision of accessible and safe storage services for unstably housed individuals is an intervention proposed by local affected communities in our study setting and would alleviate some of the safety concerns in public spaces.

In summary, municipal government-enforced confiscation of belongings appeared to be targeting some of the most structurally marginalized populations and implicated in the exacerbation of violence in these populations. Our findings also highlight the inadequate reach of essential services to these populations—potentially exacerbated by street sweeps themselves—and indicate the need for more comprehensive support systems and harm reduction strategies. Future interventions should prioritize the agency of unstably housed individuals^{42, 43, 44}.