Denver Pay for Success Initiative: Supportive Housing and Health Outcomes

Project Summary

The project team estimated the impact of a supportive housing intervention on homeless individuals who cycle in and out of jail, detoxification centers, and emergency medical services. The study’s treatment group received supportive housing services in project-based and scattered-site supportive housing units. The team also evaluated inmate health records, filling an existing gap in the evidence base on health interventions for a frequent jail reentry population.

Research Question(s)

  • Does supportive housing impact the target population’s patterns of primary care, emergency and hospital care, and mental health and substance abuse treatment? Does supportive housing decrease substance use disorder relapse rate? Does supportive housing increase continuity of care?
  • Does supportive housing decrease avoidable emergency department and hospital services for the target population? Does supportive housing decrease other emergency department and hospital services?
  • Does supportive housing impact the target population’s severity or burden of illness as measured by a clinical risk group?


  • Inform stakeholder, policymaker, and practitioner decision-making about making investments in supportive housing, and whether and how cross-sector investments impact health and wellbeing.


Housing First increased office-based care for psychiatric diagnoses, decreased emergency department visits, and increased access to prescription medication.


Health: access to health care services (mental and physical health care, access to substance abuse treatment, continuity of care); use of health care services (relapse rate of substance use disorder, avoidable ER and hospital visits); overall health (severity of illness, mental and physical health)

Other: housing stability (homelessness); criminal justice involvement (arrests, jail days)


Randomized control trial and Intent to Treat (ITT) and Treatment on the Treated (TOT) framework for analysis. The ITT estimate applied a regression-based method to control for measured characteristics including age, race, gender, arrest history, referral type, and risk category. An instrumental variables (IV) approach was employed to calculate the TOT, in which randomization into the treatment group was used as an instrument for actual treatment, defined as housing lease-up.

Homeless person leaning agains a wall
Grantee and Partner organizations

Urban Institute
City and County of Denver
Corporation for Supportive Housing
Colorado Access

Grant status
Principal investigators
Sarah Gillespie, MPA
Devlin Hanson, PhD
Start date
Award amount
36 months

Stay Connected