A new report from the University of Houston, released in partnership with the Fort Bend County Department of Health and Human Services and PolicyMap, finds that housing instability is a powerful and consistent predictor of poor mental health outcomes among Fort Bend County residents. Drawing on a representative countywide survey and combining traditional regression with machine-learning analysis, the study shows that affordability pressures, unstable occupancy, and unsafe or substandard conditions jointly raise residents’ risk of stress, depression, and anxiety.
The report, titled Housing Instability and Mental Health, was authored by a research team led by Dr. Jeronimo Cortina, the Senator Don Henderson Endowed Chair Professor in the Department of Political Science at the University of Houston. The research was funded in part by the National Institutes of Health through the AIM-AHEAD South-Central Hub.
Key Findings
- 33% of residents experienced moderate perceived stress.
- 17% experienced mild to moderate depression.
- 60% experienced mild anxiety.
- Across all three mental health outcomes, housing instability and health-related quality of life ranked as the two most influential predictors—often outweighing income, education, and demographic characteristics.
- The machine-learning models explained 88% of the variation in perceived stress, 93% in depression, and 92% in anxiety.
- Risks are not distributed evenly. Mapping reveals specific ZIP Codes where housing instability and poor mental health are concentrated together, signaling clear targets for community-level intervention.
Housing as a Determinant of Health
The report reframes housing as a core non-medical determinant of health, alongside factors such as income, education, and neighborhood conditions. When housing is affordable, stable, and safe, it supports daily routines, financial security, and social connection. When it is not, the consequences ripple into mental and physical well-being.
“Housing isn’t just shelter—it’s the platform on which the rest of life is built. When that platform is shaky, everything else gets harder: holding a job, raising children, managing a chronic illness, staying connected to neighbors. What we’re seeing in Fort Bend County is that instability shows up in residents’ mental health long before it shows up in any single eviction or foreclosure number.”
— Dr. Jeronimo Cortina, the report’s lead author
How the Study Was Conducted
The team fielded a representative survey of adults living in Fort Bend County using a dual-frame design that combined random-digit-dialed landline numbers with cellphone numbers selected from local call areas and billing ZIP Codes. Responses were weighted to U.S. Census American Community Survey benchmarks for age, race, ethnicity, and language spoken at home.
Housing conditions were measured through a Housing Instability Index built using structural equation modeling, combining three domains: lack of affordability, lack of stable occupancy, and lack of safety and decency. Mental health was assessed using three widely validated instruments—the Perceived Stress Scale (PSS-4), the Patient Health Questionnaire (PHQ-9) for depression, and the Generalized Anxiety Disorder scale (GAD-7). The team also incorporated the EQ-5D health-related quality of life instrument and a self-rated quality of life measure.
Analyses combined linear regression with Gradient Boosting Regression implemented through the H2O machine-learning platform, allowing the team to capture non-linear and interaction effects.
“What stood out across every model we ran was the interaction between housing instability and health-related quality of life. Each one is harmful on its own. Together, they compound—residents experiencing both at the same time face a sharper rise in predicted stress, depression, and anxiety than either factor would suggest in isolation. That tells us we can’t treat housing and well-being as separate policy lanes.”
— Dr. Jeronimo Cortina
A Map for Targeted Action
The report includes maps grouping ZIP Code areas into four categories based on the co-location of housing instability and mental health symptoms. The most concerning category—high housing instability paired with high mental health symptoms—is concentrated in specific parts of the county, where residents also tend to face greater limits on health-related quality of life.
“Risk in Fort Bend County isn’t spread evenly, and that’s actually good news for policymakers. It means resources don’t have to be diluted across the entire map. We can identify the ZIP Codes where multiple challenges are stacking on top of each other and direct stabilization, repair, and quality-of-life supports there first. Place-based, data-driven action is exactly what these findings call for.”
— Dr. Jeronimo Cortina
Policy Implications
The authors argue that improving mental health in Fort Bend County will require coordinated attention to both housing stability and the conditions that make stability meaningful—safe and decent housing, predictable occupancy, affordability that does not force trade-offs against food or healthcare, and the broader supports that allow residents to live healthy, connected lives.
The findings reinforce the case for treating housing as a frontline health investment, particularly in communities where instability and mental health burdens overlap. Demographic characteristics such as sex, race/ethnicity, and language spoken at home had limited predictive value once differences in housing conditions and quality of life were accounted for, suggesting that targeting housing and well-being directly may yield broad benefits across populations.
About the Project
The report is a partnership between the University of Houston, the Fort Bend County Department of Health and Human Services, and PolicyMap, with support from the AIM-AHEAD South-Central Hub. The full report is available online and an interactive data exploration tool are available at: https://www.policymap.com/newmaps/e/uhfbcunstablehousing
This research was, in part, funded by the National Institutes of Health (NIH) Agreement No. 1OT2OD032581. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the NIH.
###