• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Evaluation of a Collaborative Model Between Managed Care and Affordable Housing on Acute Care Costs

The American Journal of Managed CareMarch 2024
Volume 30
Issue 3
Pages: 133-138

A collaborative service model between a managed care organization and an affordable housing provider reduced acute care use and costs.


Objectives: This study evaluated a collaborative service model between the largest Medicaid managed care organization (MCO) in Texas, Superior HealthPlan, and the affordable housing provider Prospera Housing Community Services.

Study Design: Using a quasi-experimental 2-groups research design, we compared health care outcomes and costs between a sample of 104 participants served by the Prospera+Superior collaborative model and a group of 104 participants who had health care coverage through the Superior HealthPlan Medicaid MCO but did not live at Prospera properties (ie, Superior-only group).

Methods: Data from medical claims were analyzed to examine change in outcomes 12 months before and after implementation of the Prospera+Superior collaborative model in 2019.

Results: The Prospera+Superior group had a 56% lower rate of emergency department/urgent care visits and spent $2061 less in prescription costs than the Superior-only group after implementation.

Conclusions: These findings provide needed evidence of the clinical and economic value of forming multisector collaborative models between MCOs and other community providers.

Am J Manag Care. 2024;30(3):133-138. https://doi.org/10.37765/ajmc.2024.89514


Takeaway Points

A collaborative service model between a managed care organization and an affordable housing provider reduced acute care use and costs.

  • Managed care organizations can partner with affordable housing providers to better serve clients.
  • Partnership between managed care organizations and housing providers may be economically viable.
  • Residents who lived on housing properties that have a managed care partner used fewer emergency and urgent care services.


There is widespread interest from the medical and health care fields in addressing social determinants of health. The housing and communities in which individuals live can be important to their health and access to health care. Well-documented links exist between income, housing, and health. For example, financial hardship is a robust predictor of health among residents of low-income housing,1 and financial strain may mediate the link between mental illness and homelessness.2 However, there is a need to examine policy levers and service models that comprehensively and effectively improve financial health, housing stability, and access to health care services.

In the health care landscape of the US, managed care organizations (MCOs) play an essential role in supporting the health of low-income and disabled populations. Many states are beginning to require MCOs to screen for and address social determinants of health among enrollees.3,4 Despite experimenting with various diverse programs to address social determinants of health, many MCOs are struggling with service integration, financing, and evaluation efforts to determine the effectiveness and sustainability of these programs.5

Many low-income individuals face challenges with obtaining and maintaining stable housing. Supportive housing programs that provide subsidized housing can offer many individuals needed affordable housing options. However, as concluded in a report by the National Academies of Sciences, Engineering, and Medicine, there is no substantial evidence as yet that supportive housing programs improve health outcomes.6 In other words, supportive housing programs improve housing outcomes7,8 but may not necessarily improve health.6,9

Collaborative service models that combine health care and homeless services have shown promise. For example, federally funded Health Care for the Homeless clinics around the country serve as primary care “health homes” for homeless and at-risk individuals and have grown to more than 200 sites over the past 3 decades.10 The US Department of Veterans Affairs (VA) has implemented the Homeless Patient Aligned Care Team (HPACT) program, which is a multidisciplinary medical care home that offers tailored primary care services to homeless veterans. There is evidence that use of HPACT is associated with reductions in emergency department (ED) use, improvements in primary care utilization, and positive patient experiences.11 However, most of the existing models have largely been hospital based instead of community based.

Unique collaborative models can offer clients multiple services on site and provide a variety of partnered services. These models can also increase communication between different providers and entities, which can enhance care and improve continuity of care. Experts have argued for more than a decade for innovative ways to pair health insurance and housing services. A well-known policy brief that was disseminated more than 10 years ago made a business case for Medicaid-financed services in supportive housing to lower costs associated with avoidable hospitalizations and other crisis services.12 There is a need to empirically study new, innovative community-based service models between housing providers and insurance payers.

Study Aims

We conducted this quasi-experimental study to examine a collaborative, community-based service model that involved an affordable housing provider and an MCO and compare outcomes with those of a control group enrolled in the MCO. Our primary outcomes were use of health care services and costs. We hypothesized that the collaborative service model would be associated with less use of acute care and lower overall medical care costs than the control group.


Program Description

Prospera Housing Community Services is a nonprofit organization that has been building and operating affordable housing with resident support services in Texas for nearly 30 years. It serves individuals and families by providing safe, high-quality, affordable housing and support services at more than 50 sites across 19 cities throughout south and central Texas. At each site, on-site service managers help offer resident services and facilitate care and billing of services with insurance payers such as Superior HealthPlan. In this study, 104 residents across 11 sites in the San Antonio, Corpus Christi, and Rio Grande Valley regions of Texas were included; these sites are listed in the eAppendix (available at ajmc.com).

Superior HealthPlan was founded in 1999 in El Paso, Texas, and has become the largest Medicaid MCO in Texas. It operates under parent company Centene Corporation, the largest Medicaid provider in the US. Superior HealthPlan provides enrollees with access to all Medicaid programs in Texas including STAR, STAR+PLUS, STAR Kids, and STAR Health. Additionally, Superior HealthPlan provides coverage through the Children’s Health Insurance Program, Medicare, and the health insurance marketplace.

A collaborative service model involving the affordable housing provider Prospera Housing Community Services and the Superior HealthPlan Medicaid MCO was implemented in 2019. This partnership allowed for facilitated care between the 2 organizations and joint programs that aimed to improve social determinants of health, including access to nutritious foods, transportation, affordable health care, and secure housing.13

Research Design

This study used a matched-groups design (matching on age, gender, service delivery area, and having 6 months of coverage) combined with regression-based adjusted analyses to compare a group of 104 residents (treatment group) living at Prospera Housing Community Services properties with coverage by Superior HealthPlan (herein referred to as Prospera+Superior) with a matched group of 104 residents (control group) who did not live at Prospera Housing Community Services properties and only had coverage by Superior HealthPlan (herein referred to as Superior only). Prior to the regression analyses, participants in the Prospera+Superior group were matched with similar individuals from the Superior-only group to develop a comparable control group. These groups were matched on demographics (age, gender, service delivery area), any dual coverage (eg, Medicaid-Medicare coverage), and insurance product (eg, STAR+PLUS). To test our study hypothesis, we examined the Prospera+Superior and Superior-only groups in the 12 months before implementation of the Prospera+Superior collaboration (ie, before 2019) and the 12 months after implementation at each site. The main outcomes were health care utilization and costs. All study procedures were approved by the institutional review board at the University of Texas Health Science Center at Houston (UTHealth) (project #HSC-SPH-21-0841).

Sample Size Calculations

Based on our sample size calculations, we found that to detect a significant mean difference of 0.2—at an α of 0.05 and a power of 0.80—for the number of outpatient, inpatient, or ED visits, with an SD of 0.5 for both groups, we needed 99 participants in each group. Similarly, to detect a significant mean difference of $1000 at an α of 0.05 and a power of 0.80 for the medical and pharmacy costs, with an SD of $2000 for both groups, we needed 63 participants in each group.

Data Sources

Data from enrollment, medical, and pharmacy files for years 2018-2020 from Superior HealthPlan for both Prospera+Superior and Superior-only groups were transferred to the research team at the UTHealth School of Public Health. Among dual-eligible participants, some were enrolled in the Medicare-Medicaid Plan (MMP) and some were not. The MMP is a state demonstration program that tests integrative care delivery of both Medicare and Medicaid benefits for dual-eligible individuals with the goal of improving quality and coordination of care.14 Full claims history data were received for those in the MMP program; for dual-eligible participants not in the MMP, data were pulled from Medicare claims files available through the UTHealth School of Public Health, which is a CMS Qualified Entity. This linkage allowed for a more complete review of health care utilization and costs for the study samples. Data on the insurance product were also extracted to determine whether participants were in STAR+PLUS, a Texas Medicaid managed care program for adults who have disabilities or are 65 years or older that provides Medicaid health care and long-term services through a health plan that members choose. All aspects of the Medicaid insurance plans and the housing programs remained constant during the study period.

To ensure there was no overcounting of services and costs, data from randomly selected participants were manually reviewed and compared across different data sources. The analytic data set created for this study contained information on participants’ registration in Prospera housing (for the Prospera+Superior group), demographic characteristics, Medicaid enrollment information, health care utilization (ie, number of outpatient visits, inpatient admissions, ED visits), and overall spending for health care services. The first day the Prospera+Superior collaboration was implemented at each housing site in 2019 served as the index date, and the analytic data set included the 12 months before and after the index date at each site.


The 3 main independent variables of interest were the binary indicator for the group to which each patient belonged (treatment or control), the binary indicator for the pre-/postintervention period of the observation in the panel data, and the interaction between the treatment/control groups and pre-/postintervention period variable, which provides the adjusted effect of the intervention. Information on participant sociodemographic, health care coverage, and clinical characteristics was extracted from eligibility and enrollment files and also included in the regressions. The sociodemographic characteristics in the regressions included age measured as a continuous variable in years, 2-category gender variable (male, female), and 4-category race-ethnicity variable (non-Hispanic White, non-Hispanic Black, Hispanic, other). Health care coverage variables in the regressions included 2-category dual insurance status variable (yes, no) and 2-category insurance product variable (MMP, STAR+PLUS). Medical claims were reviewed to examine medical diagnoses and to measure each participant’s Charlson Comorbidity Index (CCI) score as a continuous variable to capture participants’ clinical characteristics.15 An 11-category variable measuring the Prospera property a participant stayed in was also adjusted for in the regressions. All sociodemographic, health care coverage, and clinical characteristics were measured at baseline and were not time varying.

Four health care utilization measures and 2 types of health care costs were examined as dependent variables before and after implementation of the Prospera+Superior collaboration to study the effect of the collaborative service model on participants. These measures were number of outpatient visits, number of ED/urgent care visits, number of inpatient admissions, inpatient length of stay, total medical care costs, and total pharmaceutical costs.

Data Analysis

First, the Prospera+Superior and Superior-only groups were compared on demographics, health care coverage, and clinical diagnoses using bivariate tests with independent t tests and χ2 tests. Second, the groups were compared descriptively on health care utilization and costs before and after implementation of the Prospera+Superior collaboration. The Wilcoxon signed rank test was used to compare the mean between groups unadjusted for preimplementation differences. Third, the groups were compared on health care utilization and costs, controlling for differences in health care coverage and clinical diagnoses before implementation of the Prospera+Superior collaboration. Because the data have a panel structure in which the patients are repeat sampled before and after implementation, panel data regressions were used based on various specification tests to help pick the right regressions. The fixed-effect panel data regression analysis was used, which is the most conservative regression method in this context; participant fixed effects were adjusted for. We tested different types of mixed-effects panel data regressions as well as the fixed-effect regressions and picked the most conservative fixed-effects regression. Because all covariates are time invariant, the fixed-effects regression absorbs all the variations in addition to unobserved time-invariant variations. For health care costs, which had large ranges, a fixed-effects linear regression was used. For health care utilization, such as ED/urgent care visits and number of inpatient admissions, which had a limited range, fixed-effects Poisson regression was used to account for the count data–like nature of the utilization measures. We tried other mixed-effects models, and our results were robust to the type of regression used.


Table 1 shows the demographic characteristics and health care coverage of participants. The majority of participants were female and Hispanic and had dual health care coverage. There were no significant differences between the Prospera+Superior and Superior-only groups on demographic characteristics and rates of dual coverage, which was expected given that the 2 groups were matched on these characteristics. The Prospera+Superior group was more likely to be in the STAR+PLUS program and less likely to be in the MMP program than the Superior-only group.

Table 2 shows medical conditions from the CCI among participants. The most common medical conditions were diabetes (with and without complications) and renal disease. The Superior-only group was significantly more likely to have 6 medical conditions from the CCI and had higher CCI scores than the Prospera+Superior group.

Table 3 shows the health care utilization of inpatient, outpatient, and ED/urgent care services 12 months before and after implementation of the Prospera+Superior collaborative model among the Prospera+Superior and Superior-only groups. The Superior-only group had significantly higher pharmacy costs after the implementation of the intervention compared with the period before the implementation of the intervention. This was the only significant pre-/postintervention difference in this group. The Prospera+Superior group had a significant reduction in the number of ED/urgent care visits and pharmacy costs after the implementation of the intervention compared with the period before the implementation of the intervention. When we compared the Superior-only and Prospera+Superior groups, we found that the former group had higher numbers of outpatient and inpatient visits, longer lengths of stay, and higher medical costs across both periods, consistent with the higher number of medical conditions in the Superior-only group.

Due to group differences in sociodemographic and clinical characteristics before implementation, adjusted fixed-effects panel data regressions were conducted controlling for participant-level effects to examine differences between groups on health care utilization. As shown in Table 4, the Prospera+Superior group had a significant 56% lower rate of ED/urgent care visits than the Superior-only group after controlling for differences in baseline utilization and clinical characteristics. The Prospera+Superior group also had significantly lower pharmaceutical costs than the Superior-only group. On average, the Prospera+Superior group spent $2061 less on prescription drugs than the Superior-only group after implementation of the Prospera+Superior collaborative model and controlling for preimplementation differences.


There has been widespread interest in multisector collaborative models of care, particularly between MCOs and community providers. However, limited empirical data exist to support the effectiveness of particular models. This study contributes to the literature by a controlled comparison between a collaborative service model involving an MCO-affordable housing partnership and a control group. Our main finding showed this model was associated with decreased use of ED/urgent care services compared with a group that was insured under the same Medicaid MCO but did not have the opportunity to receive collaborative services with a housing provider. Although participants were not randomly assigned, so we cannot infer causality, the finding does suggest that the collaborative model improved access to care and facilitated greater participation in health care prevention activities upstream that resulted in fewer acute care needs.

A secondary and important finding was that the collaborative service model was associated with lower overall pharmaceutical costs among its participants than the control group. This finding provides further data supporting a business case for these types of collaborative models.12 Furthermore, given the wide reliance on medications and efforts to reduce inappropriate polypharmacy16,17 as well as concerns about rising medication costs in the US,18 the observed decrease in pharmaceutical costs may have broad program and policy implications. For example, one study of the top 150 medications administered and prescribed in EDs in the US found that their costs increased by 28% to 125% over the past decade.19 It may be important to note that although our findings support other collaborative efforts to integrate care in different settings, the Prospera+Superior collaborative model is different from more medically based models such as the national Health Care for the Homeless clinics10 and the VA HPACT programs.11 The Prospera+Superior model is embedded in where people live and works to address multiple social determinants of health instead of specific health conditions.


Several limitations of the study are worth noting. First, as mentioned earlier, we did not randomly assign participants into groups, so a randomized clinical trial is needed to confirm these findings. Second, this was a unique partnership between Prospera and Superior HealthPlan, so the generalizability of these findings to partnerships between other similar agencies has yet to be determined. Third, 4 of the 11 study sites were followed to April 30, 2020, and Texas began ordering social distancing and statewide lockdowns on March 19, which could have affected our findings marginally during the last 40 days of follow-up for some individuals. Fourth, given our sample size, we were limited to examining broad categories of health care utilization and costs and did not have adequate cell sizes to examine more specific categories as originally intended (eg, use of specialized preventive services or costs). We also did not have access to data on Prospera’s operating costs per participant, so a cost-effectiveness analysis could not be conducted. Future multisite studies with full cost data are needed to examine these issues more directly. These limitations were counterbalanced by the strengths of the study, which included a research design with a control group, a statistically rigorous approach that included matched groups with regression-based adjusted analyses, and examination of both health care utilization and costs.


There is evidence that collaborative service models between MCOs and housing providers, such as between Superior HealthPlan and Prospera Housing Community Services, can reduce use of costly ED and urgent care services and overall pharmaceutical costs. These findings may have policy implications as MCOs focus on social determinants of health and consider new models of care to effectively address them.


The authors thank Shao-Chee Sim and C. J. Eisenbarth Hager at Episcopal Health Foundation; Scott Ackerson, Carmen Hancock, Diane Warren, and the staff and residents at Prospera Housing Community Services; and Michelle Murdock, Alex Goldson, and analysts at Superior HealthPlan for their support.

Author Affiliations: Department of Management, Policy, and Community Health, The University of Texas Health Science Center at Houston (JT, SR, CT, VS, CGC), Houston, TX; National Center on Homelessness Among Veterans, US Department of Veterans Affairs (JT), Washington, DC; Department of Psychiatry, Yale School of Medicine (JT), New Haven, CT.

Source of Funding: This evaluation was funded by the Episcopal Health Foundation and Prospera Housing Community Services.

Author Disclosures: The authors report that this study received grant funding from Prospera, which was a study site.

Authorship Information: Concept and design (SR, VS, CGC); acquisition of data (CGC); analysis and interpretation of data (SR, CT, VS, CGC); drafting of the manuscript (JT, SR, CT, CGC); critical revision of the manuscript for important intellectual content (SR, VS, CGC); statistical analysis (SR, CT, CGC); obtaining funding (JT); administrative, technical, or logistic support (JT, VS); and supervision (JT, CGC).

Address Correspondence to: Jack Tsai, PhD, MSCP, The University of Texas Health Science Center at Houston, 1200 Pressler St, Houston, TX 77030. Email: jack.tsai@uth.tmc.edu.


1. Tucker-Seeley RD, Harley AE, Stoddard AM, Sorensen GG. Financial hardship and self-rated health among low-income housing residents. Health Educ Behav. 2013;40(4):442-448. doi:10.1177/1090198112463021

2. Elbogen EB, Lanier M, Wagner HR, Tsai J. Financial strain, mental illness, and homelessness: results from a national longitudinal study. Med Care. 2021;59(suppl 2):S132-S138. doi:10.1097/MLR.0000000000001453

3. Shrank WH, Keyser DJ, Lovelace JG. Redistributing investment in health and social services—the evolving role of managed care. JAMA. 2018;320(21):2197-2198. doi:10.1001/jama.2018.14987

4. Medicaid managed care provides opportunities for states to address social determinants of health and health equity. Health Management Associates blog. January 6, 2022. Accessed October 27, 2022. https://www.healthmanagement.com/blog/medicaid-managed-care-provides-opportunities-for-states-to-address-social-determinants-of-health-and-health-equity/

5. Gottlieb L, Ackerman S, Wing H, Manchanda R. Understanding Medicaid managed care investments in members’ social determinants of health. Popul Health Manag. 2017;20(4):302-308. doi:10.1089/pop.2016.0092

6. National Academies of Sciences, Engineering, and Medicine. Permanent Supportive Housing: Evaluating the Evidence for Improving Health Outcomes Among People Experiencing Chronic Homelessness. The National Academies Press; 2018.

7. Tsai J, Mares AS, Rosenheck RA. A multi-site comparison of supported housing for chronically homeless adults: “housing first” versus “residential treatment first.” Psychol Serv. 2010;7(4):219-232. doi:10.1037/a0020460

8. Leff HS, Chow CM, Pepin R, Conley J, Allen IE, Seaman CA. Does one size fit all? what we can and can’t learn from a meta-analysis of housing models for persons with mental illness. Psychiatr Serv. 2009;60(4):473-482. doi:10.1176/ps.2009.60.4.473

9. Tsai J, Gelberg L, Rosenheck RA. Changes in physical health after supported housing: results from the collaborative initiative to end chronic homelessness. J Gen Intern Med. 2019;34(9):1703-1708. doi:10.1007/s11606-019-05070-y

10. Zlotnick C, Zerger S, Wolfe PB. Health care for the homeless: what we have learned in the past 30 years and what’s next. Am J Public Health. 2013;103(suppl 2):S199-S205. doi:10.2105/AJPH.2013.301586

11. Tsai J, Havlik J, Howell BA, Johnson E, Rosenthal D. Primary care for veterans experiencing homelessness: a narrative review of the Homeless Patient Aligned Care Team (HPACT) model. J Gen Intern Med. 2023;38(3):765-783. doi:10.1007/s11606-022-07970-y

12. Nardone M, Cho R, Moses K. Medicaid-financed services in supportive housing for high-need homeless beneficiaries: the business case. Center for Health Care Strategies Inc. June 2012. Accessed June 23, 2022. https://www.chcs.org/media/SH_Medicaid_Bz_Case_081712_brief.pdf

13. About Superior HealthPlan. Prospera Housing and Community Services. Accessed December 4, 2022. https://prosperahcs.org/partners/superior-healthplan/

14. Dual Eligible Integrated Care Demonstration Project. Texas Health and Human Services. Accessed June 23, 2022. https://www.hhs.texas.gov/services/health/medicaid-chip/medicaid-chip-programs-services/dual-eligible-integrated-care-demonstration-project

15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8

16. Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the process of deprescribing. JAMA Intern Med. 2015;175(5):827-834. doi:10.1001/jamainternmed.2015.0324

17. Reeve E, Shakib S, Hendrix I, Roberts MS, Wiese MD. Review of deprescribing processes and development of an evidence-based, patient-centred deprescribing process. Br J Clin Pharmacol. 2014;78(4):738-747. doi:10.1111/bcp.12386

18. Leighl NB, Nirmalakumar S, Ezeife DA, Gyawali B. An arm and a leg: the rising cost of cancer drugs and impact on access. Am Soc Clin Oncol Educ Book. 2021;41:1-12. doi:10.1200/EDBK_100028

19. Tebo C, Mazer-Amirshahi M, Zocchi MS, et al. The rising cost of commonly used emergency department medications (2006-15). Am J Emerg Med. 2021;42:137-142. doi:10.1016/j.ajem.2020.02.010

Related Videos
Leslie Fish, PharmD.
Ronesh Sinha, MD
Pat Van Burkleo
Pat Van Burkleo
dr mitzi joi williams
Kathy Oubre, MS, Pontchartrain Cancer Center
Jonathan E. Levitt, Esq, Frier Levitt, LLC
Judy Alberto, MHA, RPh, BCOP, Community Oncology Alliance
Sandra Stein, MD
Pat Van Burkleo
Related Content
© 2024 MJH Life Sciences
All rights reserved.