do housing first trials show better housing but not better symptoms or substance use?
“We have an necessarily seen improvements in all outcomes with housing first such as consistent improvement in mental health symptoms or substance use health issues”
a guest, a psychiatrist and senior scientist at CAMH said on PsychEd: Educational Psychiatry Podcast, 2026-09-01, at 30:53. hear it in context →
the episode: PsychEd Episode 78: Mental Health and Homelessness with Dr. Vicky Stergiopoulos (at 30:53). we link the show's own page; we do not re-host audio.
first, the part that holds
The speaker's claim holds up. In the randomized trials, Housing First keeps people housed for far more days than usual care. Mental health symptoms and substance use are not consistently better than in usual care, and in the largest Canadian trial they were no better on any measure.
what we found when we went and looked
We read the 2019 meta-analysis of randomized Housing First trials and three reports from Canada's At Home/Chez Soi trial. Housing outcomes were large: in the meta-analysis people were 2.46 times as likely to be stably housed (95% CI 1.58-3.84), and in the four-city trial the share of days stably housed rose from 24-39% under usual care to 63-77%. Self-rated mental health barely moved (SMD 0.07, CI -0.19 to 0.33), and substance use showed no clear difference (ratio of rate ratios 0.96, CI 0.72-1.28). The Canadian trial found no significant group difference in symptom severity or substance use problems at 24 months. One thing the short version leaves out: people in both groups did improve on symptoms over time, so the finding is that Housing First did not do better than usual care, not that nobody got better.
assertion by assertion
| assertion | verdict | what the record says |
|---|---|---|
| housing first improves housing stability | confirmed | In the 2019 meta-analysis of randomized trials, the risk ratio for being stably housed was 2.46 (95% CI 1.58-3.84) and days housed had an SMD of 1.24. In the four-city Canadian trial with intensive case management, the share of days stably housed was 62.7-77.2% versus 23.6-38.8% under usual care, depending on the city. In the Toronto high-needs arm with assertive community treatment, it was 73.6% versus 27.8%. All four trials in the meta-analysis were rated at high risk of bias. [1][2] |
| housing first trials have not shown consistent improvement in mental health symptoms | confirmed | Meta-analysis: self-rated mental health SMD 0.07 (CI -0.19 to 0.33), with no meaningful difference in symptom severity. In the four-city trial, the change in symptom severity did not differ from usual care at 24 months (0.57, CI -0.88 to 2.01). In the Toronto assertive community treatment arm, symptom index scores did not differ between groups. The evidence comes from a few trials, mostly Canadian, in adults with mental illness, with follow-up of about two years. [1][2] |
| housing first trials have not shown consistent improvement in substance use | confirmed | Meta-analysis: no clear difference in substance use, with both groups reducing (ratio of rate ratios 0.96, CI 0.72-1.28; 2 trials). Four-city trial: substance use problems 0.94 (CI 0.79-1.12), not significant. Toronto assistive community treatment arm: no difference in substance use severity. The flip side, also from the record, is that Housing First did not increase problematic substance use. [1][2] |
| people receiving housing first still see their symptoms improve over time | confirmed | The trial investigators' own review reports that Housing First participants improved in mental health symptoms and substance problems, but the usual-care group improved just as much. In the five-city analysis of suicidal behaviour, suicidal ideation fell from 37.3% to 21.3% across both groups, and Housing First was no better than usual care. So the null result is about the difference between groups, not about whether anyone got better. [3][4] |
| popular framing: beyond housing, housing first does nothing measurable for health | overstated | Meta-analysis: fewer emergency department visits (IRR 0.63, CI 0.48-0.82) and fewer hospitalisation episodes (IRR 0.76, CI 0.70-0.83). In the Toronto assertive community treatment arm, psychiatric hospital days were 65.8 versus 168.4, and arrests fell. But the four-city intensive case management trial found no significant drop in emergency department visits (0.73, CI 0.49-1.07) and no difference in generic quality of life (EQ-5D difference 0.10, CI -2.92 to 3.13). The benefits beyond housing are real for some measures but inconsistent. [1][2] |
both directions
for the speaker: Every randomized source we read fits the claim. The meta-analysis found a large housing effect but no clear difference in substance use and almost none in self-rated mental health. The four-city Canadian trial raised days stably housed by 33-50 percentage points but found no significant difference from usual care in symptom severity, substance use problems or generic quality of life at 24 months. The Toronto arm for people with higher needs found the same pattern for symptoms and substance use. The Canadian trial investigators' own review says the usual-care group made equivalent gains on symptoms and substance problems.
against: The claim is about outcomes compared with usual care. It should not be heard as "housing first doesn't help people's health." Both groups improved on symptoms over two years, so people in Housing First did get better, just not more than the comparison group. Housing First did not increase problematic substance use, which was a common worry about programs that don't require sobriety. Some service-use outcomes favored Housing First: the meta-analysis found fewer emergency visits and hospitalisations, and the Toronto high-needs arm had far fewer psychiatric hospital days and fewer arrests. The four-city intensive case management trial showed a short-lived gain in community functioning at 18 months. The claim's word "consistent" is the right qualifier. It does not mean there were no benefits beyond housing.
what this cannot say
The meta-analysis included only four randomized trials, all rated at high risk of bias. Participants could not be blinded, and most outcomes are self-reported. Most of the evidence comes from one Canadian program, which enrolled adults who were homeless and had mental illness, with follow-up of about two years. A null result partly reflects what the comparison group got: in Canada, usual care already included housing and support services, and that group's symptoms also improved, which makes a difference between groups harder to detect. Symptom and substance use scales may not pick up changes that matter to the person. Our own check covers the reviews and trial reports we could open in this session. Some of our numbers come from a page-reading tool's summary rather than the full text. We did not read the US or French trials directly, and we did not find a Cochrane review of Housing First.
sources, each one fetched and read
- Effects of Housing First approaches on health and well-being of adults who are homeless or at risk of homelessness: systematic review and meta-analysis of randomised controlled trials. Journal of Epidemiology and Community Health, 2019; 73(5):379-387 What we read there: Four randomized trials, all at high risk of bias. Stably housed risk ratio 2.46 (95% CI 1.58-3.84). Self-rated mental health SMD 0.07 (CI -0.19 to 0.33). Substance use ratio of rate ratios 0.96 (CI 0.72-1.28), no clear difference. Emergency department visits IRR 0.63, hospitalisation episodes IRR 0.76. Conclusion: Housing First approaches 'successfully improve housing stability and may improve some aspects of health' without an increase in problematic substance use. Fetched 2026-10-22 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC6581117/
- How did a Housing First intervention improve health and social outcomes among homeless adults with mental illness in Toronto? Two-year outcomes from a randomised trial. BMJ Open, 2016 What we read there: 197 high-needs participants. Days stably housed 73.6% versus 27.8% (difference 45.8%). No significant group differences in Colorado Symptom Index scores or substance use severity. Fewer psychiatric hospital days (65.8 versus 168.4) and a greater drop in arrests with Housing First plus assertive community treatment. Fetched 2026-10-22 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC5030577
- Putting Housing First: the evidence and impact. Canadian Journal of Psychiatry, 2015 What we read there: Review by the Canadian trial's investigators. Housing First was very successful at the primary outcome of stable housing. Participants improved in mental health symptoms and substance problems, but the standard-treatment group showed equivalent gains. Fetched 2026-10-22 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC4679126
- Effect of Housing First on suicidal behaviour: a randomised controlled trial of homeless adults with mental disorders. Canadian Journal of Psychiatry, 2017 What we read there: 2,255 participants in 5 Canadian cities. Suicidal ideation fell from 37.3% at baseline to 21.3% at two years across both groups. The trial 'failed to find evidence that HF is superior to TAU' for suicidal behaviour. Fetched 2026-10-22 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/5528985/
speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam