Need help right now? Call 211· Text HOME to 741741· 988 Suicide & Crisis
Evidence Brief

Everyone Cites the Same Two Studies for Medical Respite. One of Them Found It Cost More.

By Common Ladder · August 20, 2026 · 11 min read

A man has surgery. The wound needs to stay clean and dry, the dressing needs changing daily, and he needs an antibiotic on a schedule for ten days. He is medically ready to leave the hospital. He has nowhere to go.

What happens next is predictable. The dressing gets dirty. The pills get lost, or stolen, or thrown out with a swept encampment. Two weeks later he is back in the emergency department with an infection costing an order of magnitude more to treat than it would have cost to prevent.

Medical respite care fills that gap: a bed to recover in, with meals, medication, nursing oversight and case management, for the few weeks between the hospital and whatever comes next. It is such an obviously good idea that it feels rude to ask for evidence. We asked anyway.

The problem is real and it is growing

Between 2016 and 2022, US hospital admissions carrying a diagnosis code for unhoused status rose from 304,750 to 473,590 — a 55% increase — during a period when total hospital admissions fell by about 8%.2 Roughly one admission in 85 now involves someone with no home to be discharged to.

The cost arithmetic is not subtle either. A day of medical respite under California's Medicaid rate schedule runs about $204, in a band from $181 to $226.9 A day in a US hospital averages $3,297 in expenses per adjusted inpatient day; in Arizona, $3,411.10 Respite is roughly six cents on the hospital dollar.

That comparison is where most of the advocacy stops. It shouldn't, because it assumes the thing that needs proving: that the respite bed prevents the hospital day.

What the studies actually found

Three studies do essentially all the work in this literature, and each is worth reading closely, because the summaries are more confident than the papers.

Chicago, 2006: fewer hospital days, no change in ED visits

Buchanan and colleagues followed 225 hospitalized homeless adults consecutively referred to a Chicago respite center over 26 months. The comparison group is the clever part: patients who met every eligibility criterion and were referred but denied admission because no bed was free. Over twelve months of follow-up, the respite group used 3.7 hospital days versus 8.3 (p = .002), with the largest reduction among people with HIV/AIDS.5

The same abstract reports "no differences in emergency department or outpatient clinic visits."5 Keep that sentence in mind — a "36% fewer ER visits" figure circulates widely and is attributed to this study. It comes from a bar chart in an advocacy brief, and the authors themselves reported no difference.6

Boston, 2009: half the adjusted odds, six thousand dollars more

Kertesz and colleagues looked at 743 homeless patients discharged from Boston Medical Center between 1998 and 2001 — 136 to respite, 174 to other planned care, 433 to their own care, meaning the street or a shelter. This is the study behind almost every "50% fewer readmissions" claim you will ever read. Here is what it found, in order.3

Unadjusted, there was no difference at all. Ninety-day readmission was 21% for respite, 22% for other planned care, 21% for own care. The paper says it plainly: "There was no difference in the proportion readmitted in comparisons not adjusted for patient characteristics."3

Adjusted, the effect appears. After propensity weighting and regression adjustment for illness burden, length of stay, and substance use, respite patients had an odds ratio of 0.54 (95% CI 0.34–0.85) for 90-day readmission relative to discharge to own care. The comparable estimate for other planned care, 0.70, was not significant.3

That is a real and defensible finding, and it is not the same claim as "half as many readmissions." It says that once you account for the fact that sicker people were sent to respite, respite looks protective. Whether you believe it depends on whether you believe the adjustment captured the sickness.

And it cost more. The mean respite stay was 31.3 days and $7,929 in charges. In adjusted analysis, a respite discharge was associated with an extra $5,994 in 90-day charges (95% CI $4,210–$7,779) compared with discharge to own care.3 The paper that anchors the cost-savings case for medical respite measured, over its own time horizon, a cost increase.

Copenhagen, 2020: the only trial, and its primary outcome was null

Bring and colleagues randomized 96 homeless patients across ten Danish hospitals to either a two-week stay at a Red Cross respite center or discharge as usual. It remains the only randomized controlled trial of medical respite anywhere.4

The pre-specified primary outcome was cost at three months. The intervention group's costs were €4,761 lower — p = 0.10. Null. The differences reach significance at six months (€8,515, p = 0.04) and at twelve (€12,603, p = 0.03), and the paper is candid that the twelve-month window was added afterward: "Post-hoc, we decided to also conduct the analyses for a 12-month period."4

Quality-of-life gains were not significant in either arm, so the cost-effectiveness ratio is not significant either — though 96% of bootstrap replications came out cost-saving, an encouraging pattern in a study too small to prove it. Crude total costs were not significantly different at three or six months, and emergency department costs were not different at any horizon.4

Three details matter. The first six participants were assigned to the intervention without randomization, to get the center running. The control group had more prior hospital admissions at baseline. And the authors flag their own external validity: "our results should be viewed as conservative estimates because Copenhagen offers many targeted services for homeless people."4 An eight-bed center staffed by one nurse and volunteers, in a country with universal coverage, is not the same intervention as a 50-bed clinical facility in Phoenix.

Where the famous numbers come from

The 2013 systematic review — still the only one — found thirteen articles, "heterogeneous in methods, study quality, inclusion of a comparison group, and outcomes examined." Its conclusion is genuinely positive: respite "reduced future hospital admissions, inpatient days, and hospital readmissions" and "also resulted in improved housing outcomes," with results for ED use and costs "mixed but promising." Its final sentence: "Future research utilizing adequate comparison groups is needed."8

That was thirteen years ago, and we could not find a peer-reviewed systematic review or meta-analysis published since. The field's own 2021 literature review is explicit about what it is not: "Articles were not reviewed or analyzed for quality or academic rigor."8

Which brings us to the numbers that actually move money. The most widely circulated respite savings figures — $3 million a year for Los Angeles hospitals, $3.5 million over three years in Portland, $6.2 million a year in Cincinnati — come from a two-page fact sheet published in 2010. Its footnotes attribute those three figures to, respectively, personal contact with a chief medical officer, a 2009 newspaper article, and personal contact with an executive director.6

None of that means the figures are wrong — program directors generally know their own numbers. But a phone call in 2010 is not a study, and sixteen years of repetition has not turned it into one. A companion brief from 2011 adds five more city totals, every one a personal communication, under the sentence: "Medical respite programs are proven to reduce future hospital readmissions by half."6 We flag this the same way we flagged two of our own numbers this month. Citation chains rot from the middle.

What has never been shown

Mortality. We found no study reporting a mortality effect of medical respite. The Danish trial recorded four deaths total, two in each arm, and was in no way powered to detect a difference. The Boston cohort excluded the eight patients who died before 90 days. Anyone telling you medical respite saves lives is making an inference, not citing a result.

Permanent housing. The 2013 review reports improved housing outcomes, but the underlying measures are inconsistent. The Danish trial found 23% of the intervention group obtained temporary housing afterward, with no comparison figure for controls.4 Circle the City reports discharging about 70% of patients to somewhere other than the street or emergency shelter — a meaningful operational number, and the organization's own, uncontrolled.14 No controlled study shows respite increases permanent housing placement. That is not a knock on respite; a respite bed cannot conjure an apartment that does not exist.

Anything past a year. The longest follow-up in this literature is twelve months.

And the best recent replication came back null. A 2024 study of The INN Between in Salt Lake City reused the Buchanan design — 29 people admitted versus 39 referred and turned away for lack of a bed — and found emergency department use down 18.9%, readmissions down 16.4% and length of stay down 10.9% for the admitted group. All three comparisons were statistically non-significant; the report's headline $47,110 saved per patient per year is an extrapolation built on those non-significant differences, published on the provider's own website.7 The authors also explain, fairly, why better designs are hard: "it is unethical to refer some individuals to medical respite while referring others that also need such services to a treatment-as-usual model."7

The strongest evidence isn't a trial. It's California.

While the research literature has been quiet, the policy world stopped waiting. California made Recuperative Care one of fourteen Community Supports under CalAIM starting January 1, 2022. In the twelve months from the last quarter of 2023 through the third quarter of 2024, 7,699 unique Medi-Cal members received it, through 429 provider contracts. Quarterly volume rose from 1,614 to 5,227 across roughly eighteen months.9

California's own analysis, conducted with its actuarial consultant, reports a 29.2% net cost reduction for recuperative care comparing six months before with six months after.9 Read that as what it is: a before-and-after comparison in a population selected because they had just had an expensive medical event. Regression to the mean will do some of that work. The independent UCLA-RAND evaluation required by CMS is not due until 2028.9

Still, this is the most consequential development in the field: in two years it should produce the first large, controlled, contemporary answer to a question the literature has not settled since 2006. Roughly a dozen states now cover medical respite through Medicaid in some form — among them Washington, New York, Massachusetts, Illinois, North Carolina, New Mexico, Kentucky and Utah — with terms that vary enormously. Kentucky reimburses 20 beds per provider for up to 45 days; Massachusetts covers as little as two days before a procedure for people without bathroom access.11

Arizona pays for housing. It does not pay for this.

Arizona's Housing and Health Opportunities (H2O) demonstration, approved by CMS in October 2022 with services live since October 1, 2024, is a real and substantial piece of policy. It covers rent and transitional housing for up to six months, utilities, housing navigation, pre-tenancy and tenancy-sustaining services, one-time moving costs, deposits and home accessibility modifications.12

Medical respite is not on that list. We checked three independent documents — the CMS special terms and conditions, AHCCCS's own provider manual chapter on H2O services, and the state Housing Program Guidebook — and "respite" and "recuperative care" appear as covered services in none of them.12 In the national tally of state waivers, Arizona is listed as covering permanent supportive housing only.11

So Phoenix has 100 medical respite beds — two 50-bed Circle the City campuses, downtown and midtown, nationally certified in March 2025 — funded by a patchwork of managed care contracts, HRSA grants, local government and philanthropy rather than a dedicated Medicaid benefit. An 85-bed East Valley center in Mesa is expected to open this year, nearly doubling that capacity, helped by a $1 million foundation gift announced this month.14

The timing here is not comfortable. In March 2025 CMS rescinded the federal framework that had guided health-related social needs waivers, while stating that "rescinding this guidance does not negate existing approvals." In June 2026 it issued new budget-neutrality rules for 1115 demonstrations, effective January 1, 2027, and in July 2026 it withdrew the fast-track review process. How health-related social needs services like medical respite will be treated under the new regime is, in the field's own summary, an open question.13

And Arizona's next waiver is being written right now. AHCCCS has published its draft application for the demonstration period running October 1, 2027 through September 30, 2032, and is accepting written public comment at waiverpublicinput@azahcccs.gov through September 6, 2026, before submitting to CMS by September 30.15 If you have an opinion about whether Arizona Medicaid should pay for a bed to recover in, this is the two-week window in which it is cheapest to have one.

Being precise about what this establishes

The evidence is thin, not negative. Every study points the same direction: a 2006 cohort, a 2009 propensity-adjusted cohort, a small 2020 trial and a 2024 replication all find fewer hospital days or lower costs. The disagreement is about statistical significance and time horizon, not sign. That is a different situation from an intervention rigorously tested and failed.

The programs being compared are not the same thing. Bed counts in the field range from 5 to 124. Median stays run from 12 days in Copenhagen to 31 in Boston to 46–60 at Circle the City. Staffing ranges from one nurse and a volunteer roster to a 24-hour interdisciplinary clinical team. "Medical respite works" is roughly as precise a claim as "surgery works."

Selection bias cuts both ways. Respite admission requires being medically appropriate, behaviorally appropriate for congregate living, and willing to participate in your own care — screening that plausibly selects people who would have done better anyway. It also selects people sicker than average, which biases the other direction. Only randomization separates them.

Costs shift as much as they fall, and "cost-effective" is a question about whose ledger. The Danish trial found the intervention group used more rehabilitation, general care and substance use treatment; the 2013 review describes the same transition from hospital services to community ones. Kertesz found respite cost nearly $6,000 more over 90 days while reducing adjusted readmission odds. Both facts hold. A hospital, a Medicaid agency and a person with a healing surgical wound are not optimizing the same thing.

What follows from this

Stop quoting the 2010 fact sheet. If you are writing a grant application, a board memo, or a legislative brief, cite Buchanan 2006, Kertesz 2009 and Bring 2020 with their actual effect sizes and their actual limitations. The case survives honest citation. It does not need the phone calls.

Fund the evaluation while you fund the beds. Every state paying for this is generating claims data with a natural comparison group already in it — people who met criteria and did not get a bed. California's 2028 evaluation should not be the only one. And measure what nobody has measured: mortality, permanent housing at 12 and 24 months, and what happens to the people who are turned away, which every program tracks and almost none publishes.

And in Arizona, say something before September 6. The state has built genuine Medicaid housing infrastructure and left recuperative care out of it, while its largest provider expands on philanthropy. Whether that is the right allocation is a live question with an open comment inbox and a deadline.

If you or someone you're helping needs medical respite in Maricopa County: Circle the City operates two 50-bed centers — Downtown, 210 S 12th Ave, Phoenix, 602-254-4282, and Midtown, 333 W Indian School Rd, Phoenix, 602-612-9781. Both are 24-hour facilities with weekday administrative hours of 8:30am–5pm. Eligibility: adults 18 and older who are homeless or imminently homeless, currently in a hospital or emergency department, with acute medical needs expected to improve within four to six weeks, or needing end-of-life care. There is no fee and no insurance requirement. Referrals usually come from a hospital social worker or case manager, but outreach workers and family can also refer — and you can call directly to ask. If you are not sure where to start, call 211. Our guide to free health clinics in Phoenix covers what to do when the need isn't a hospital-level one.14

Frequently asked questions

What is medical respite care?

Medical respite care, also called recuperative care, is short-term residential care for people experiencing homelessness who are too sick or injured to recover on the street or in a shelter, but not sick enough to occupy a hospital bed. A typical stay runs somewhere between two weeks and six weeks and includes a bed, meals, medications, nursing or clinical oversight, and case management. It is not a hospital, not a nursing home, and not permanent housing.

Does medical respite care reduce hospital readmissions?

Probably, though the evidence is thinner than the advocacy suggests. The largest US study, a 743-patient Boston cohort published in 2009, found that after adjusting for patient characteristics, people discharged to respite had about half the odds of being readmitted within 90 days compared with people discharged to their own care. But the unadjusted readmission rates were identical at 21 percent, and the same study found respite was associated with roughly $6,000 in additional 90-day charges. A 2006 Chicago cohort found respite patients used 3.7 hospital days in the following year versus 8.3 for people referred but turned away for lack of a bed, with no difference in emergency department visits.

Does medical respite care save money?

Sometimes, on some time horizons, from some points of view. The only randomized trial of medical respite anywhere in the world, conducted in Copenhagen with 96 participants, found no significant cost difference at its pre-specified three-month endpoint and significant differences only at six and twelve months. California's Medicaid agency reports a 29.2 percent net cost reduction for recuperative care, but that is a before-and-after comparison rather than a controlled trial. The dramatic city-level savings figures that circulate in advocacy materials trace back to a 2010 fact sheet whose sources are phone calls with program directors and a newspaper article.

Does Arizona Medicaid pay for medical respite care?

No. Arizona's Housing and Health Opportunities (H2O) 1115 demonstration covers rent and temporary housing for up to six months, utilities, housing navigation, tenancy support, moving costs, deposits and home accessibility modifications. Medical respite and recuperative care are not on the covered services list in the CMS special terms and conditions, the AHCCCS provider manual chapter, or the state housing program guidebook. Roughly a dozen other states cover medical respite through Medicaid; Arizona is not one of them. AHCCCS is taking public comment on its next waiver application through September 6, 2026.

Where can someone get medical respite care in Phoenix?

Circle the City operates two 50-bed medical respite centers in Phoenix: a downtown campus at 210 S 12th Avenue (602-254-4282) and a midtown campus at 333 W Indian School Road (602-612-9781). Both serve adults 18 and older who are homeless or imminently homeless, are currently in a hospital or emergency department, and have acute medical needs expected to improve within four to six weeks, or who need end-of-life care. There is no fee and no insurance requirement. Referrals usually come from a hospital social worker or case manager, though outreach workers and family members can also refer. An 85-bed East Valley center in Mesa is expected to open in 2026.

Sources & footnotes

  1. Program counts over time. 78 programs: National Health Care for the Homeless Council, "Updated 2016 Medical Respite Program Directory" (July 2016). "Over 120 programs" in 35 states and DC: NHCHC/NIMRC, "Standards for Medical Respite Care Programs" (2021 edition) — eight standards covering accommodations, environmental services, transitions into and out of care, post-acute clinical care, care coordination, personnel, and quality improvement. 152 programs in 40 states and DC (2023 data): HHS ASPE, "Medical Respite Programs: A Critical Service for Individuals Experiencing Homelessness" (December 2024) — this figure is from search indexing of the brief; the PDF did not return extractable text to our fetcher and it should be re-verified before republication. "At least 17 new medical respite care programs opened" 2025–July 2026: NHCHC/NIMRC, "Insights from the Field" (published July 14, 2026). Caveat on any current total: NHCHC launched a new directory in March 2025 and did not migrate prior entries, so the live directory undercounts operating programs. NIMRC itself launched July 15, 2020; its voluntary certification opened to all applicants April 1, 2025 after a 15-program testing cohort, with a second application window June 16–25, 2026 and a 2026 update to the Standards explained in an August 26, 2026 webinar.
  2. Fernandez, Duval, McIntyre et al., "Increasing trends of hospital admissions for unhoused patients in the United States," Scientific Reports 16:15446 (2026). Admissions carrying an unhoused-status diagnosis code rose from 304,750 (2016) to 473,590 (2022), a 55.4% relative increase, while total US admissions fell 7.8% (35,675,421 to 32,891,849). Across the period, 2,818,983 admissions — 1.2%, or about one in 85 — carried an unhoused code.
  3. Kertesz SG, Posner MA, O'Connell JJ, Swain S, Mullins AN, Shwartz M, Ash AS, "Post-Hospital Medical Respite Care and Hospital Readmission of Homeless Persons," Journal of Prevention & Intervention in the Community 37(2):129–142 (2009), PMCID PMC2702998. Retrospective cohort, Boston Medical Center 1998–2001; 743 homeless individuals discharged (Respite 136, Other Planned Care 174, Own Care 433); 735 analyzed after 8 deaths. Propensity-score weighting plus multivariable logistic regression (propensity model c = .76). Unadjusted 90-day readmission: 21% overall, and per the paper, "There was no difference in the proportion readmitted in comparisons not adjusted for patient characteristics." Adjusted: Respite OR 0.54 (95% CI 0.34–0.85); Other Planned Care OR 0.70 (95% CI 0.46–1.06, non-significant). Costs: mean respite stay 31.3 days and $7,929 in charges; "a Respite disposition was associated with +$5994 (95% CI, $4,210–$7,779) in excess charges, relative to Own Care" (2002 dollars, referring-hospital charges only). Stated limitation: "The principal limitation to this study is reliance on observational data."
  4. Bring C, Kruse M, Ankarfeldt MZ, Brünés N, Pedersen M, Petersen J, Andersen O, "Post-hospital medical respite care for homeless people in Denmark: a randomized controlled trial and cost-utility analysis," BMC Health Services Research 20:508 (2020), trial registration NCT02649595. Pragmatic RCT plus societal-perspective cost-utility analysis, April 2014–March 2016, ten hospitals in the Capital Region of Denmark; n = 96 (53 intervention, 43 control); intervention was a two-week stay at a Red Cross respite center (initially 4 beds, later 8; one paid RN, "primarily staffed with volunteers"; median stay 12 days). Sample size was powered to detect a 25% cost difference at three months. Cost differences favoring the intervention: €4,761 at 3 months (p = 0.10), €8,515 at 6 months (p = 0.04), €12,603 at 12 months (p = 0.03); the paper reports p-values only, no confidence intervals, and states "Post-hoc, we decided to also conduct the analyses for a 12-month period." Crude total costs were not significant at 3 months (€8,448 vs €13,553, p = 0.07) or 6 months (p = 0.1); ED costs were non-significant at every horizon. "Both groups achieved minor QALY gains, which were slightly higher in the intervention group but not significant"; 96% of bootstrap replications indicated the intervention was cost saving, but "the QALY difference was not significant, rendering the ICER as not significant." "The first six patients included in the study were all assigned to the intervention group without randomization to get the medical respite care program up and running." "The control group had more admissions 2 years prior to admission." 23% of the intervention group obtained temporary housing after their stay. Authors' external-validity caveat: "our results should be viewed as conservative estimates because Copenhagen offers many targeted services for homeless people."
  5. Buchanan D, Doblin B, Sai T, Garcia P, "The effects of respite care for homeless patients: a cohort study," American Journal of Public Health 96(7):1278–1281 (2006); free abstract via UCSF SIREN. 225 hospitalized homeless adults consecutively referred from an urban public hospital (Stroger Hospital of Cook County, Chicago) over 26 months, split into those accepted into the respite center and those "referred but denied admission because beds were unavailable"; all met the center's predefined eligibility criteria. From the abstract: "During 12 months of follow-up, the respite care group required fewer hospital days than the usual care group (3.7 vs 8.3 days; P=.002), with no differences in emergency department or outpatient clinic visits. Individuals with HIV/AIDS experienced the greatest reduction in hospital days." Per-arm sample sizes and confidence intervals could not be verified: the AJPH full text and PMC1483848 were not retrievable to our fetcher. Figures here are from the published abstract only.
  6. National Health Care for the Homeless Council, "Medical Respite Care: Demonstrated Cost Savings" (fact sheet, February 2010) and "Medical Respite Care: Reducing Costs and Improving Care" (policy brief, April 2011). The 2010 sheet lists "Los Angeles, CA $3 million total annual savings for hospitals," "Portland, OR $3.5 million total savings over three years for one hospital," and "Cincinnati, OH $6.2 million total annual savings for three hospitals and the community," footnoted respectively to personal contact with JWCH Institute's chief medical officer, a 2009 article in The Oregonian, and personal contact with the Center for Respite Care's executive director. The 2011 brief adds five more city figures (San Diego, Atlanta, Sacramento, Richmond, Salt Lake City), each sourced to a personal communication with a program director, and states that "Medical respite programs are proven to reduce future hospital readmissions by half." The "50 percent fewer readmissions" callout is footnoted to Kertesz 2009, whose actual finding is an adjusted odds ratio of 0.54 with identical crude rates. The 2011 brief's Figure 1, attributed to Buchanan 2006, plots inpatient days 3.4 vs 8.1 and ED visits 1.4 vs 2.2 — figures that do not match the published AJPH abstract (3.7 vs 8.3 days; no ED difference) and that appear to be the origin of the circulating "36% fewer ER visits" claim.
  7. Weller A, Panos A, "Medical Respite as an Intervention for Reducing Acute Healthcare Utilization for Individuals Experiencing Homelessness: A collaborative study with The INN Between and local hospitals," Utah Valley University (2024). Retrospective analysis of 2023 referrals from University of Utah Hospital and Huntsman Cancer Institute: 68 referred, 29 admitted, 39 not admitted solely for lack of bed space. Reported differences for the admitted group versus those referred and not admitted: emergency department utilization −18.88%, all-cause readmissions −16.36%, length of stay −10.89%. The authors state: "Two-tailed t-tests on the emergency department utilization, readmission rates, and lengths of stay of patients admitted to The INN Between and those of patients not admitted showed no statistical significance." The "$47,110 in savings per patient per year" is an extrapolation applying national cost averages to those non-significant differences. Not peer-reviewed; hosted on the operating provider's own website. Ethics quote: "it is unethical to refer some individuals to medical respite while referring others that also need such services to a treatment-as-usual model. As such, researchers have been limited in their abilities to make direct comparisons between groups."
  8. Doran KM, Ragins KT, Gross CP, Zerger S, "Medical respite programs for homeless patients: a systematic review," Journal of Health Care for the Poor and Underserved 24(2):499–524 (2013); abstract also at CRD/DARE. Thirteen articles met inclusion criteria; the review reports they "were heterogeneous in methods, study quality, inclusion of a comparison group, and outcomes examined," that "available evidence showed that medical respite programs reduced future hospital admissions, inpatient days, and hospital readmissions" and "improved housing outcomes," that "results for emergency department use and costs were mixed but promising," and that "future research utilizing adequate comparison groups is needed." No pooled effect estimate and no formal strength-of-evidence grade. We found no peer-reviewed systematic review or meta-analysis of medical respite outcomes published since. The field's own update — NIMRC, "Medical Respite Literature Review: An Update on the Evidence for Medical Respite Care" (March 2021) — covers 45 articles from 2010–2020 and states plainly: "Articles were not reviewed or analyzed for quality or academic rigor," recommending that "an updated systematic review is recommended to further evaluate the level and quality of evidence." See also Daaleman TP et al., "Protocol of a mixed method research design to study medical respite programs for persons experiencing homelessness," PLoS ONE 19(1):e0295543 (2024): "prior studies of MRP programs and outcomes have been limited to individual sites and may not generalize."
  9. California. Rate: California Department of Health Care Services, "Community Supports Pricing Resource" (December 2025) — Recuperative Care $204 per day, range $181–$226; non-binding guidance. Utilization: DHCS, "Community Supports (ILOS) Annual Report" (2025), Table 2 — 7,699 unique members received Recuperative Care from Q4 2023 through Q3 2024 (quarterly: 1,614 / 2,677 / 3,416 / 3,746), and Table 9 — 429 provider contracts. Later quarters (3,320 → 5,227 for Q2 2024 through Q1 2025): California Health Policy Strategies, "CalAIM Community Supports: Utilization by County and MCP" (December 2025), from DHCS data. Cost effect: DHCS, "Medi-Cal Community Supports Are Delivering on Their Promise" (October 2025 update) — 29.2% net cost reduction for Recuperative Care, six months pre versus six months post, DHCS/Mercer analysis; this is a before-and-after comparison, not a controlled study. Independent evaluation: DHCS/UCLA-RAND, Community Supports HRSN evaluation design (August 4, 2025), results due 2028. Community Supports launched January 1, 2022; Recuperative Care and Short-Term Post-Hospitalization Housing are the two authorized under California's 1115 demonstration.
  10. Kaiser Family Foundation, "Hospital Adjusted Expenses per Inpatient Day", State Health Facts, 2024 data from the AHA Annual Survey (page modified January 16, 2026): US average $3,297.26; Arizona $3,410.51. Note this is an expenses-per-day measure across all hospital types and payers, not a charge or a homeless-specific figure.
  11. National Health Care for the Homeless Council, "Summary of CMS State Medicaid Director Letter on Medicaid 1115 Waivers" (July 2026), table of states with 1115 waivers for supportive housing or medical respite care, status as of June 30, 2026. States listed as covering medical respite (alone or alongside permanent supportive housing) include CA, NY, MA, VT, WA, IL, NC, HI, KY, NM, UT, with ME and NE pending; Arizona is listed as permanent supportive housing only. NHCHC cautions that "pending" reflects states "in various stages of planning, implementation, or negotiations with CMS" and does not necessarily mean CMS is blocking a state. State-specific detail from NHCHC's state status pages (May 2025): Kentucky (20 reimbursed beds per eligible provider, up to 45 days), Massachusetts (up to two days pre-procedure for members lacking bathroom access), Washington (provider reimbursement began July 1, 2025), New Mexico, New York (billing eligibility January 1, 2025, though as of May 2025 "most programs have not yet been able to obtain Medicaid reimbursement"), Utah, North Carolina. Terms vary enough that a single national count is not meaningful; we say "roughly a dozen" for that reason.
  12. Arizona H2O covered services. CMS, Approved Special Terms and Conditions, Arizona Section 1115 Demonstration (August 30, 2024) — STC 33 lists the approved health-related social needs housing supports: rent/temporary housing up to six months, utilities, pre-tenancy and tenancy-sustaining services, housing transition navigation, one-time transition and moving costs, housing deposits, and medically necessary home accessibility modifications; STC 35 excludes room and board except as provided in STC 33. No respite or recuperative care service appears. AHCCCS, Fee-for-Service Provider Manual, Chapter 29 — H2O Services (revised July 14, 2026), and AHCCCS, Housing Program Guidebook, 3rd edition (September 2025) — same covered-services list; neither document mentions medical respite or recuperative care. Timeline and enrollment: CMS approved H2O October 14, 2022; services implementation was effective October 1, 2024; AHCCCS reported more than 250 members determined eligible as of December 31, 2024 (FFY2025 Q1 quarterly report) and over 500 determined eligible with roughly 30 housed as of the Q2 FY2025 report (May 29, 2025). Arizona separately funds the state-only AHCCCS Housing Program (permanent supportive housing for members with serious mental illness), which also does not cover medical respite.
  13. Federal policy. CMS, CMCS Informational Bulletin, March 4, 2025, rescinding the November 2023 and December 2024 health-related social needs bulletins and the 2023 HRSN Framework; CMS states "Rescinding this guidance does not negate existing approvals" and that requests will be evaluated case by case. Eighteen states had approved HRSN waivers under the rescinded framework: KFF, "Section 1115 Waiver Watch" (May 2, 2025, updated August 11, 2025), which also covers the April 2025 decision not to approve or extend federal match for Designated State Health Programs. The June 11, 2026 CMS State Medicaid Director letter overhauling 1115 budget neutrality effective January 1, 2027, the July 7, 2026 rescission of fast-track review, and the unresolved question of "how will health-related social needs activities like tenancy supports and medical respite care be treated?" are summarized in the NHCHC fact sheet cited at note 11.
  14. Circle the City. Capacity, certification and funding mix: National Health Care for the Homeless Council medical respite directory profile, "Circle the City Medical Respite" (profile modified July 28, 2026) — two 50-bed programs, average length of stay 46–60 days, NIMRC-certified March 10, 2025 through March 10, 2028, funded through Medicaid reimbursement, managed care contracts, HRSA grants, local and state government, foundations and private donations; no dedicated state respite benefit is listed. Addresses, phone numbers, eligibility, hours, fees and referral process: 211 Arizona listing for the Downtown Medical Respite Center (210 S 12th Ave, Phoenix, AZ 85007; 602-254-4282 main; last assured February 1, 2026) and 211 Arizona listing for the Midtown Medical Respite Center (333 W Indian School Rd, Phoenix, AZ 85013; 602-612-9781; last assured February 1, 2026). The 70% discharge figure is the organization's own, from Circle the City's medical respite page: "discharging approximately 70% of those individuals to living situations other than the streets or emergency shelter systems." East Valley center: Circle the City, "Circle the City Growing East"; groundbreaking and 85-bed capacity reported by the Mesa Chamber of Commerce and KJZZ; the $1 million Bob & Renee Parsons Foundation gift was announced in August 2026. Opening date is the organization's stated expectation and has moved before — call before relying on it.
  15. AHCCCS, "Arizona's Section 1115 Waiver Renewal Request" — draft application for the demonstration period October 1, 2027 through September 30, 2032; written public comment accepted at waiverpublicinput@azahcccs.gov through September 6, 2026, with submission to CMS due by September 30, 2026. Summary and public-comment context: Arizona Public Health Association, "Arizona's Medicaid Waiver Up for Renewal: Your Opportunity to Comment is Now" (August 9, 2026). H2O is listed as continuing in the draft; no cuts to it are proposed.

Phone numbers, addresses and eligibility criteria on this page were checked against 211 Arizona's provider listings (last assured February 1, 2026) and Circle the City's own materials in August 2026. No number was dialed. Programs change — call before you go, or call 211. Effect sizes are reported as the source papers state them, including their null results; where we could not reach a primary document, the footnote says so. Have a correction or an updated figure? Email corrections.commonladder@gmail.com.