Impact assessment of health and social determinants in shelter retention in San Mateo County

Context Homeless shelters facilitate interventions to reduce the health disparities faced by homeless people and help them return to permanent housing. Poor retention in shelters, however, limits their effectiveness. There is a lack of scholarship examining the relationship between health and social determinants and retention of people in homeless shelters.

Objective Evaluate the relationship between social and health determinants with shelter retention, defined as length of stay.

Study Design This study is a retrospective analysis of real-world data from LifeMoves, the largest shelter network in the Bay.

Dataset The dataset is managed and maintained by LifeMoves and includes socio-demographics and health-related data from point-in-time assessments of unhoused individuals that receive their services.

Population Studied Unhoused individuals who received services from the LifeMoves program at the Maple St. Shelter from 04/09/18 to 02/17/2023.

Instrument Two linear regression models were implemented to assess the correlation of social and health determinants with shelter retention independently. To mitigate multicollinearity, we selected features using the Variance Inflation Factor, with a standard cutoff of 5, and also employed clustering techniques.

Outcome Measures Correlation between health and social determinants with length of stay in shelters, as well as the associated effect sizes.

Results Age, veteran status, inability to stop substance consumption, and having prescribed medications had the strongest negative impact on the length of stay. An increase in age by 1 year was correlated with a 1.4-day increase in length of stay at the shelter. Veterans had an average length of stay of 40 days shorter than non-veterans, a finding that is corroborated by previous research. People with an inability to stop substance consumption had an average 23-day shorter length of stay and those with prescription medication had an average 20-day shorter length of stay.

In contrast, individuals with access to health benefits had an average 95 days longer length of stay than those without.

Conclusions Age, veteran status, and access to health benefits exhibit the strongest correlation. For future research, it is recommended to assess the generalizability of the results obtained in this study with other populations.