Across Canada, there are several disparate databases that track occurrence and utilization of emergency shelters and housing services for people who experience homelessness. Population-level understanding of homelessness across Canada
rely on these individual data repositories. While these data sources are useful in providing policy makers and service providers with situational and community level awareness to homelessness issues, separate data sources may diminish
the representativeness of issues related to homelessness on a population level. Specifically, these databases may not capture the experiences of homeless individuals who are part of a marginalized or isolated group. This makes it difficult
to determine the true extent of homelessness across the country.
The aim of the Homelessness Counts study is to access and integrate existing sources of information to better determine who, and how many people in Canada are homeless. Given the variations in the homeless experience across the country,
especially within rural and remote regions of the nation, it is important to obtain a Canada wide sample to better understand which data sources could be helpful. Thus, our team will collect primary data (e.g., focus groups with community
stakeholders, and individual interviews with people experiencing homelessness) in at least one community in each province or territory, including sites with a potentially higher level of rural or 'invisible' homeless population.
Our past work has included seeing how existing databases (such as provincial health data) could be used as an alternative to determine homelessness. Previously, our team created an algorithm to determine if a person is experiencing
homelessness based on other information about them in various healthcare databases. Data from interviews conducted in our cross-Canada study on homelessness will be used to validate and refine our algorithm. The algorithm will then
be applied to determine who and how many people are homeless across the nation. Using elements of machine learning, risk and burden modelling will be conducted. The findings from these analyses will be used to develop a prototype for
a nation-wide integrated homelessness surveillance system that will better capture the extent of homelessness and the nature of homelessness issues that occur across Canada.
1. What are the societal and cultural determinants of homelessness?
2. What are risk factors/risk profiles for homelessness?
3. What trajectories are followed by people experiencing homelessness?
The Homelessness Counts study uses a mixed-methods research design. Primary data collection is currently on-going. Individual interviews with participants who are experiencing homelessness and focus groups with stakeholders (i.e., municipal
contacts and service providers) are being conducted across Canada. Data will be collected from at least one community in each province or territory across Canada, with a focus on individuals who may be described as the "invisible homeless"
(e.g., couch-surfers). Our team aims to collect data from at least 13 individuals who are experiencing homelessness in each community that we visit.
Instruments used in our individual interviews with participants capture information on their demographics, housing status and history, mental and physical health conditions, community integration, health and social service utilization,
perceived barriers and facilitators to service access, quality of life and their housing needs.
Through focus groups conducted with community stakeholders (e.g., service providers), we hope to discuss and identify individuals in each community who experience homelessness, but are often 'missing' in existing databases (e.g., individuals
who are couch-surfing).
Primary, mixed-methods data collection in at least one community in each province and territory across Canada. Interviews with individuals experiencing homelessness. Focus groups with stakeholders.
Refine homelessness algorithm by integrating existing data sources and primary data. Apply algorithm to better determine who, and how many are homeless in Canada.
Generate enhanced risk and burden modelling using elements of machine learning.
Begin development of a prototype homelessness surveillance system that can use both manual and explainable AI (XAI).
Share key learnings and offer opportunities for knowledge translation. Community forums at each site to share project outcomes.
Through this study, we hope to gain a better understanding of the extent and nature of the homeless phenomena across Canada. Fine-tuning an algorithm, to identify homelessness by linking primary data collected from this study to provincial health data,
will generate a deeper understanding of the health, social, and economic impacts of homelessness on communities and marginalized groups nationwide. The use of machine learning approaches on data collected, will offer future potential for
the development of a pan-Canadian surveillance system. The development and implementation of such a system will provide policy makers, researchers, and practitioners more accurate population-level insights into groups who experience homelessness
and the issues associated with the homeless experience across Canada.
This new arm to the original Homelessness Counts project will explore the intersection of homelessness and dementia.
One emergent finding from our multi-method Homelessness Counts project is our recognition of the prevalence of dementia within certain homeless populations. This finding was uncovered and triangulated through our interviews with people
experiencing homelessness (PEH); and, via our homeless case ascertainment algorithm in administrative data (i.e., ICES).
Currently, population-level insights of dementia in homeless populations from a Canadian context does not exist. With the changing nature of homelessness due to the COVID-19 pandemic, we believe deeper analysis into the prevalence, burden,
and outcomes of individuals experiencing both homelessness with dementia is important and timely.
We will conduct an analysis using healthcare administrative data from Ontario to compare the one-year prevalence of dementia between people experiencing homelessness, low-income housed residents, and general-population housed residents
of Ontario aged 45 and older in 2019.
We anticipate that our findings will afford practitioners and policy makers with population-level insights into the realities of this highly vulnerable group, underscoring the urgent need for dementia supports and effective supportive
housing solutions for this population.