Getting the data on homelessness

At a virtual event, a research team led by Lawson Health Research Institute announced details of a new project aimed at better understanding how many people in Canada are homeless and who they are. Developing more accurate sources of data related to homelessness helps ensure the proper supports and services are available.   

“The homeless experience varies significantly across Canada, especially within different rural and remote regions of the country,” says Dr. Cheryl Forchuk, Assistant Scientific Director at Lawson and Distinguished University Professor at Western University. 

With funding from the Public Health Agency of Canada, the Homelessness Counts research project builds on past work looking at how existing databases, such as provincial health data, could be better integrated and used as a more accurate, accessible and cost-effective way to determine homelessness.  

Many communities rely on shelter data, but not all regions have shelters and not all people who are experiencing homelessness use shelters even when available. There is also a standard “point in time” method that requires potentially hundreds of volunteers to go out and actually count the people in that area who are homeless.

“Generally, these methods will still miss people who are temporarily staying with others or living in rough or abandoned areas, and is limited to a single point in time – often just one day of counting. It’s often not feasible for smaller communities to do this due to the effort required and it’s a population often missed in our census data,” explains Dr. Richard Booth, Scientist at Lawson and Associate Professor at Western. 

In the initial stages of the project, the team is holding focus groups with several representative communities across Canada to learn more about how data is tracked and accessed. They will consult with groups providing services such as homeless serving agencies, mental health and addiction agencies, hospitals, emergency services and municipal governments. Individual interviews will also be held with people with lived experience of homelessness. 

“To offer effective and efficient services, we need to know how many people are experiencing homelessness, along with their characteristics, location and needs. Right now, it’s difficult to track if things are getting better or worse, or simply changing, and if efforts to reduce the homeless population are making a difference. We don’t fully know the unique challenges of the ‘invisible homeless’ or if different subgroups are emerging,” says Dr. Booth.  

More demographics could be regularly included in the data, for example gender, ethnicity, sexual orientation, involvement in the criminal justice system, psychiatric diagnoses, physical diagnoses and housing status.  

The team will bring together this information with existing datasets to improve the performance of an algorithm previously developed through a provincial research study, as well as insights gleaned from a rapid report on COVID-19 and homelessness. Machine learning will be used to generate enhanced risk and burden modeling.  

Dr. Forchuk adds that “homelessness is a complex social problem that touches on many different systems. It is made even more difficult to tackle when we have big gaps in the data. Our hope is to prototype a centralized surveillance system that can be scaled up across Canada to help make the right services available in the right places.”