What Do Medicaid Spending Patterns Reveal about the Impact of Supportive Housing?

Nearly 50 percent of Medicaid costs are driven by only about 4 percent of Medicaid enrollees, primarily adults who are elderly, living in unstable housing, or have a mental illness, chronic medical condition, or substance use disorder. In response to rapidly rising Medicaid expenditures, there has been an increased focus on identifying strategies to better meet the needs and reduce the health care costs of this high-use population. Permanent supportive housing, which pairs housing with services such as substance treatment is one such strategy and is thought to lead to improved health care access and decreased likelihood of high-cost medical expenses, such as emergency department visits. This study explores the association between participation in a New York City supportive housing program and Medicaid expenses among people with mental illness and chronic homelessness or with diagnoses of both mental illness and substance use.

The study used matched data from administrative records for 2,827 people and sequence analysis to trace Medicaid expenditure patterns over two years among people placed in supportive housing and people eligible for, but not placed in, supportive housing. Researchers tracked the expenditure patterns for two years before the baseline, defined as the point at which applicants were placed in supportive housing or were deemed eligible but not placed. To analyze the data, researchers grouped people into six categories based on their expenditure patterns: those who were only covered by Medicaid for the final months of the two-year period and incurred low Medicaid expenditures once covered (very low coverage); those who incurred sharp increases in Medicaid expenditures after a short period without coverage (emerging user); and those with consistent coverage but varying levels of Medicaid expenditures (low, middle, high, and very high users). The authors then used propensity score matching to determine the impact of the program on Medicaid costs for the two-year period after the baseline. The majority of the sample was non-Hispanic* Black or Hispanic (82 percent), 70 percent were male, 64 percent were ages 35 to 54 years at the time of application, and 52 percent of the applicants were diagnosed with substance abuse disorders.

Key findings

• The supportive housing program was associated with Medicaid savings. On average, those in the program had $9,526 less in Medicaid expenditures than those eligible for, but not placed in, the program.
• Reduced Medicaid expenditures were significant among supportive housing residents in the very low–coverage group (−$15,694), the emerging-users group (−$9,020), and the high-users group (−$14,450). These reduced expenditures were primarily driven by reduced costs associated with psychiatric hospitalizations.
• Among supportive housing residents, 43 percent were enrolled in managed care, compared with only 33 percent of those who were eligible but not placed in supportive housing. This disparity was statistically significant among the emerging-user and high-user groups.

Research and policy implications:

• The reduced costs associated with psychiatric hospitalizations suggest that supportive housing may be effective at preventing future inpatient hospital admissions and reducing the length of hospitalizations. Future research should examine the long-term effects and mechanisms by which supportive housing influences Medicaid cost savings.
• Enrollment in managed care likely controlled high Medicaid costs for claims not related to behavioral health, which may explain the significant reductions in Medicaid expenditures among supportive housing residents in the emerging-user and high-user groups. Future research is needed to better understand the association between housing placement and increased managed care enrollment.

* The How Housing Matters editorial team decided to use the terms “Hispanic” to refer to people of Latin American origin, in alignment with the terminology used by the authors of the study. We recognize that the term “Latinx” is more inclusive of the way this group may self-identify. How Housing Matters strives to avoid language that is exclusive and will always attempt to explain the editorial rationale behind the labeling of certain groups.

Photo by Srdjan Randjelovic/Shutterstock