Unhoused people in Vancouver who had their belongings taken or were moved on by city workers were much less likely to get off the street over the following months, according to a new study that followed more than 200 people who use drugs for up to three and a half years.

The study, published in September in PLOS Medicine, is one of the first to measure how so-called encampment sweeps affect whether people find housing. Earlier research has linked sweeps to injuries, overdoses and lost medication. Whether they also keep people homeless has mostly been documented through interviews and audits rather than tracked over time.

“The confiscation of personal belongings and forced displacement might undermine pathways to housing,” the authors write.

Following people through the sweeps

The researchers, led by Kanna Hayashi of the B.C. Centre on Substance Use and Simon Fraser University, drew on two long-running cohort studies of people who use drugs in Vancouver. Participants are interviewed every six months. Most were recruited in the Downtown Eastside, where about half of the city’s unhoused population is believed to live. The co-authors include members of community groups such as the Vancouver Area Network of Drug Users, Pivot Legal Society and Police Oversight With Evidence and Research, a group largely made up of people with lived experience of drug use and housing instability.

The main analysis included 235 people who reported being homeless when they joined, followed between December 2021 and May 2025. At every interview they were asked whether city workers had taken their possessions in the past six months. City workers here included engineering crews, park rangers and police. From June 2023, a second question asked whether city workers or police had taken their tent or other shelter from a public space, or tried to, forcing them to pack up and leave.

At the start, one in four participants said their belongings had been confiscated in the past six months. In the smaller group asked about displacement, one in three said they had been forced to move.

Far less likely to find housing

Over the study period, just over half of participants (128 of 235) moved into some form of housing. After accounting for age, gender, drug use, mental and physical health, income sources, incarceration history and other factors, people who had recently had their belongings confiscated moved out of homelessness at about a third of the rate of those who hadn’t.

The link was even stronger for forced displacement. People who had been moved on were housed at less than a tenth of the rate of those who hadn’t been.

The sweeps also rarely came with somewhere to go. Participants reported being displaced 121 times during the study. In 119 of those cases, no housing was offered. Of the two people who did get an offer, one was able to move in. The other was told there was a line-up and no space.

That finding matters in B.C., the authors note. The provincial government’s position is that clearing encampments is acceptable when alternative shelter is reasonably available. They argue their data show such shelter was almost never offered to the people being moved.

Why sweeps might keep people homeless

The study did not test why sweeps would make housing harder to get, but the authors point to earlier research. Sweeps can destroy identification and other documents needed to sign a lease or join a waitlist. They can cut people off from outreach workers and services. They can worsen mental health. And they can erode trust in public workers, including the ones trying to place people in housing.

In the researchers’ own earlier work with the same cohorts, 36 per cent of people whose belongings had been confiscated said they had tried and failed to access housing services in the same six months.

Housing that may not last

Being counted as housed in this study did not necessarily mean being securely housed. Seven in ten of those who left homelessness moved into single-room occupancy hotels. These are small rooms with shared kitchens and bathrooms, and the authors note they are known for poor conditions and high rates of unlawful eviction. The study did not measure the quality of housing or whether people stayed in it. The 2025 homeless count for Greater Vancouver found that 42 per cent of respondents blamed an eviction for their current homelessness.

The authors also place the findings within the wider housing shortage. Homelessness in the city of Vancouver rose 12 per cent between 2023 and 2025, to 2,715 people counted in 2025, and point-in-time counts tend to miss people. A September 2025 report cited in the study estimated that Metro Vancouver needs $10.1 billion to $19.3 billion in investment over five years to meet its affordable rental needs. About $1.2 billion was invested in the previous five years.

Limitations

The study is observational, so it shows an association rather than proof that sweeps cause people to stay homeless. The authors also can’t rule out that the relationship runs partly the other way. When they compared sweeps in one six-month period with housing outcomes in the next, the associations pointed the same way but were no longer statistically significant. The authors suggest the effect of a sweep may be strongest in the first few months.

Participants were recruited through community outreach rather than at random. The people hit hardest by sweeps may have been the hardest to keep in the study, so the real rate of getting housed may be lower than reported. The data are also based on what participants said in interviews.

The authors call for policies that prioritize “dignified and affordable housing rather than practices that perpetuate homelessness,” echoing calls from Canada’s Federal Housing Advocate and the American Public Health Association.

Source: Hayashi, K., Sayre, E. C., McDermid, J., Marsh, S., Shane, C., Milloy, M.-J. S., & Kerr, T., “Encampment sweeps and persistent homelessness among unhoused people who use drugs in Vancouver, Canada: A prospective cohort study,” PLOS Medicine 23(9): e1005033, 2026. doi:10.1371/journal.pmed.1005033 (open access, CC BY 4.0)

Discuss this article on boreal.social