Homelessness and opioid use disorder are two widespread public health problems in the United States. Providing housing and supportive services, without requiring drug treatment, is a surprisingly cost-effective approach to helping unhoused people with opioid use disorder, Stanford researchers found in a new study in JAMA Network Open.
Worsened by the increasing prevalence of dangerous substances like fentanyl, overdoses are the leading cause of death among unhoused people. "If you're living on the streets, you're not going to be successfully treated for your opioid use disorder or for your other health conditions," said senior author Margaret Brandeau, the Coleman F. Fung professor of engineering in the School of Engineering and a professor of health policy, by courtesy, in the Freeman Spogli Institute for International Studies and the Stanford School of Medicine. Building on that fact, she wanted to study the impact of providing housing for this population.
Brandeau and her then-graduate student Isabelle Rao, now an assistant professor in industrial engineering at the University of Toronto, focused on the "housing first" approach in their study. This is one of the two general schools of thought in providing housing for people with substance use problems. The other, called "treatment first," requires that individuals seek treatment before receiving housing. But that policy has faced challenges, said Brandeau. "It's really, really hard for people on the street to get into treatment and to stay in treatment," she said. "The treatment-first approach has not been particularly helpful in many populations."
Simulating supportive housing
To study the impacts of a "housing first" intervention, Rao and Brandeau built a mathematical model simulating the treatment and health outcomes for 1,000 unhoused people with opioid use disorder. In the "status quo" model output, these individuals remained unhoused. In the "housing first" output, the same people were given housing, health care, and supportive services, with no requirement for sobriety or treatment.
From previous research, the researchers already had a model of opioid treatment that reflected the dynamic process of recovery, complete with ups and downs as people go in and out of treatment. They built on that treatment model, adding in additional equations that estimated the health outcomes and treatment trajectory for unhoused people.
Rao and Brandeau derived these equations from the research literature. Studies have found that people with stable housing are more likely to enter treatment for opioid use and have a higher likelihood of successful treatment. So, in the "housing first" model output, people were assigned a higher probability of recovery.
The researchers also wanted to quantify the costs and benefits of the housing intervention compared to the status quo. Their analysis considered the costs of housing, supportive services such as a case worker, health care, and drug treatment.
A cost-effective solution
With all the variables inputted, Rao and Brandeau ran the model 25,000 times to capture a wide range of outcomes. The simulation found that, over five years, an average of 191 out of the 1,000 unhoused people with opioid use disorder died in the status quo scenario. In the supportive housing intervention, 140 people died over the same time period.
The researchers also used the model to analyze lifetime outcomes for the 1,000 simulated individuals. First, they found out how many years people lived. Then they multiplied those years by a quality-of-life value between 0 and 1, where 1 means a person is in perfect health and 0 means they are dead. By multiplying the years they lived by the quality-of-life value, they calculated quality-adjusted life years.
Compared to the status quo, the housing first intervention added 3.59 quality-adjusted life years. Essentially, that's like giving each person an extra three and a half healthy years, on average.
How much would these extra years cost? Adding up housing, treatment, and health care costs, the researchers found that the housing intervention would cost $96,000 per person over their lifetime. They divided this number by the quality-adjusted life years gained (3.59) to determine the increased cost for each of those years gained over the status quo: $26,200.
In other words, each healthy year gained would cost those paying the bill an average of $26,200. By health economics standards, that extra cost is a great value for the health benefits it provides, said Brandeau. "These programs are highly cost-effective," she said. "You're investing money wisely to help improve outcomes for these marginalized individuals."
"Housed people have a higher likelihood of getting into treatment, which means that they have a higher likelihood of becoming abstinent, and that is going to save costs on the health care system," said Rao. "You also save a bunch of lives, first from having fewer people who are addicted, and then also because people who are homeless have a much higher mortality rate." Rao added that the model didn't include the criminal justice costs associated with homelessness, which would have made the housing intervention even more cost-effective.
The researchers are planning to work with Santa Clara County officials to inform policies around homelessness. Rao is also planning to conduct outreach in Toronto, where homelessness and opioid use are also challenges.
Brandeau adds that this research demonstrates how engineering know-how can be applied to solving societal problems. Sophisticated modelling is not just for designing efficient engines and sturdy structures. "Engineers are always trying to make things better," she said. "We really want our work to make a difference. And homelessness is a significant humanitarian crisis in our country."
