Does the Massachusetts Medicaid Flexible Services Program Reduce Food and Housing Insecurity, Stress, and Health Care Use?

This study investigated whether the Massachusetts Medicaid Flexible Services Program (FSP) reduced food insecurity, housing insecurity, stress, and acute health care use within one year. Using survey, dietary recall, and claims data, researchers compared 153 FSP participants with 1,495 eligible nonparticipants enrolled between 2019 and 2020. At baseline, 72.5% of FSP participants had food insecurity and 44.4% had housing insecurity. After one year, differences in food insecurity (4.96%; 95% CI, −3.13% to 13.05%) and housing insecurity (2.75%; 95% CI, −5.39% to 10.88%) were not statistically significant. No differences were found in diet quality, stress scores, or emergency department visits.

Full Citation and Link to Article

Thorndike AN, McCurley JL, Chang Y, Cheng J, Clark CR, Vogeli C, McGovern S, Fung V, Levy DE. Food and housing insecurity, stress, and health care use after Medicaid expanded services program. JAMA Network Open. 2025;8(7):e2519507. doi:10.1001/jamanetworkopen.2025.19507

Extended Summary

Central Research Question

The study examined whether the Massachusetts Medicaid Flexible Services Program (FSP), launched under a Section 1115 waiver to address food and housing insecurity among Medicaid accountable care organization (ACO) beneficiaries, improved outcomes in the first year of implementation. Specifically, the research asked: Does participation in the FSP reduce food and housing insecurity, stress, and acute health care use, while improving diet quality, compared with similarly eligible individuals who did not participate? The study sought to measure both quantitative outcomes (survey, dietary, and claims data) and qualitative experiences reported through interviews.

Previous Literature

Health-related social needs, such as food and housing insecurity, are strongly associated with chronic disease, elevated health care costs, and health disparities. Prior research shows that food insecurity contributes to stress and adverse health behaviors, while housing instability exacerbates poor health outcomes. Policymakers and health systems have increasingly invested in addressing these social determinants of health through Medicaid demonstrations and other programs. As of late 2024, 21 states had implemented or were piloting Medicaid waivers that allowed coverage of nonmedical supports, such as nutrition or housing assistance.

Evidence from earlier studies, however, has been mixed. Many evaluations of social needs interventions relied on pre-post designs without control groups, limiting causal inference. Larger randomized and controlled studies have generally found limited or inconsistent impacts. For example, the Centers for Medicare & Medicaid Services (CMS) Accountable Health Communities Model showed fewer emergency department visits in the intervention group but no significant differences in hospitalizations or resolution of social needs. Similarly, a California Medicaid program offering one-time housing deposits did not reduce health care use over six months.

Some smaller studies suggest positive effects. Randomized trials of medically tailored meals and produce prescriptions have improved dietary quality and reduced food insecurity in select populations. Likewise, medical-legal partnerships addressing housing insecurity have reported reduced stress in some settings. Still, these findings are not universal, and broader evaluations often show muted or null effects, particularly when interventions are short in duration or when broader economic and social conditions—such as high food costs or limited affordable housing supply—constrain results.

The Massachusetts Medicaid FSP, which began in 2020, was among the most ambitious of these efforts, providing $149 million across 17 ACOs to fund nutrition and housing-related services. Earlier administrative analyses suggested that participation might reduce hospitalizations and emergency visits, but implementation barriers and low enrollment rates raised questions about program effectiveness.

Data

The study, titled LiveWell/ViveBien, was a prospective cohort evaluation of adult Medicaid beneficiaries enrolled in the Massachusetts General Brigham (MGB) ACO. Data came from surveys, dietary recalls, electronic health records (EHRs), claims, and qualitative interviews.

The analytic sample included 153 FSP participants and 1,495 nonparticipants who were eligible but not enrolled. Baseline data were collected from December 2019 through December 2020, with one-year follow-up extending into 2021. Participants provided survey responses and two 24-hour dietary recalls annually for three years. Outcomes included food insecurity, housing insecurity, stress, diet quality, depression, anxiety, and acute health care use (emergency visits and hospitalizations).

At baseline, 72.5% of FSP participants reported food insecurity and 44.4% reported housing insecurity, with 35.9% experiencing both. Participants were predominantly female (84.3%), with an average age of 43.6 years, and 65% identified as Hispanic. Nonparticipants were demographically similar after propensity score weighting, ensuring balance across measured covariates.

Qualitative data came from in-depth interviews with 27 FSP participants, sampled to reflect the broader cohort’s diversity in demographics and service type.

Methods

The study design was a prospective cohort comparison with propensity score weighting to approximate randomization. FSP participants were compared to eligible nonparticipants with similar demographic, clinical, and social characteristics. Propensity scores accounted for age, sex, race/ethnicity, income, health diagnoses, baseline social needs, and timing of enrollment.

Quantitative outcomes included:

• Food insecurity, measured by the USDA 10-item Food Security Scale (score ≥3 indicating insecurity).

• Housing insecurity, measured by three questions about lack of own housing, frequent moves, or risk of losing housing.

• Stress, measured by the 10-item Perceived Stress Scale (score ≥14 indicating moderate to severe stress).

• Depression and anxiety, measured by the PHQ-8 and GAD-7 respectively (scores ≥10 indicating moderate to severe symptoms).

• Diet quality, measured by the Healthy Eating Index (HEI-2020, range 0–100, higher scores indicating better diet quality).

• Acute health care use, defined as emergency visits and unplanned hospitalizations in the prior year.

Regression models estimated one-year changes in these outcomes, testing time-by-group interactions. Sensitivity analyses focused on subgroups actively receiving nutrition services, those in the program for six months or longer, and those still enrolled at one-year follow-up.

Qualitative interviews followed a structured thematic analysis using Dedoose software, with coding conducted by bilingual research assistants. Themes addressed perceptions of food and housing services, dietary changes, stress, and health outcomes.

Findings/Size Effects

Quantitative analysis showed that FSP participation was not associated with statistically significant improvements at one year.

• Food insecurity: In the FSP group, prevalence fell by 1.33%, compared with a 6.29% decline among nonparticipants. The difference in changes was 4.96% (95% CI, −3.13% to 13.05%), not significant.

• Housing insecurity: The FSP group declined by 3.92%, compared with a 6.67% decline among nonparticipants. The difference in changes was 2.75% (95% CI, −5.39% to 10.88%), also not significant.

• Diet quality: The HEI-2020 score increased by 0.50 in the FSP group and by 0.02 in the non-FSP group. The difference was 0.48 (95% CI, −2.04 to 3.00), nonsignificant.

• Stress: PSS scores decreased slightly in both groups. The between-group difference was 0.34 (95% CI, −0.75 to 1.43), nonsignificant.

• Acute health care use: Emergency and hospitalization visits increased slightly among FSP participants (0.12 to 0.25) compared to a small increase among nonparticipants (0.12 to 0.13). The difference in change was 0.11 (95% CI, −0.02 to 0.24), not significant.

• Depression and anxiety scores also showed no significant differences between groups.

Sensitivity analyses confirmed these null results, whether restricting to participants with longer exposure, those actively receiving nutrition services, or excluding housing-only participants.

Qualitative findings revealed more nuance. Some participants reported positive outcomes, such as improved access to fruits and vegetables, reduced stress from grocery vouchers, or better control of chronic conditions like diabetes and high cholesterol. Others reported little to no change, citing the limited duration of benefits, high food prices, and lack of affordable housing. Some participants described increased stress due to poor communication with housing service providers or unmet expectations.

Overall, while quantitative data suggested no aggregate improvements, individual experiences varied widely, highlighting both the promise and limitations of targeted Medicaid social needs interventions.

Conclusion

The study concludes that participation in Massachusetts’ Medicaid Flexible Services Program was not associated with significant one-year improvements in food or housing insecurity, stress, dietary quality, or acute health care use. These findings suggest that the FSP, as implemented, did not achieve its policy goals in the short term.

However, qualitative interviews show that some participants did benefit, particularly in terms of diet and reduced stress, while others experienced disappointment or unchanged circumstances. This heterogeneity underscores the complexity of addressing entrenched social needs through time-limited, health-system-based interventions.

The authors argue that such programs may have limited effectiveness because they target downstream consequences of structural inequities, such as unaffordable housing and rising food costs, rather than the root causes. They call for continued rigorous evaluation of Medicaid social needs interventions, as well as broader policy solutions that address systemic barriers. The results highlight the need for multi-sector collaborations, sustained funding, and tailored approaches that extend beyond short-term service delivery if health equity is to be advanced meaningfully.