Age and Comorbidity, Not Stroke Severity, Steer Where Stroke Survivors Go for Rehabilitation

When a stroke survivor leaves the hospital, the destination they are sent to can shape the rest of their recovery. In the United States, patients who need continued care after an acute ischemic stroke typically go to one of two places: an inpatient rehabilitation facility (IRF), where they receive at least three hours of intensive, multidisciplinary therapy five to six days a week, or a skilled nursing facility (SNF), where therapy is less intensive and nursing care takes precedence. Clinical guidelines from the American Heart Association and American Stroke Association recommend IRF care over SNF care, because IRF treatment is consistently associated with greater functional improvement, higher rates of community discharge, lower readmission, and lower mortality. Yet a larger proportion of stroke survivors in the United States are discharged to SNFs rather than IRFs, and a new study suggests that the reasons may have less to do with measured rehabilitation need than with who the patient is.

Researchers led by Heather A. Hayes of Southern California University of Health Sciences, working with colleagues at the University of Utah and Intermountain Health, examined which clinical and non-clinical factors were associated with discharge to an IRF versus an SNF among people hospitalized with acute ischemic stroke. Their findings, published in Frontiers in Rehabilitation Sciences, reveal a striking pattern: after accounting for stroke severity, comorbidity, and functional status, the factors most consistently tied to placement were increasing age and greater comorbidity burden, while several social and insurance-related factors that appeared influential at first glance faded away under statistical adjustment.

The study focused on a carefully restricted population: 236 adults who had been independent and community-dwelling before their stroke. This design choice matters. Prior research on post-acute placement has often examined clinical or sociodemographic predictors in isolation, without determining whether contextual factors remain associated with IRF versus SNF placement among patients with comparable pre-stroke independence once standardized measures of stroke severity and function are considered concurrently. By limiting the cohort to previously independent individuals and incorporating objective measures alongside demographic and contextual characteristics, the researchers sought to determine whether variation in rehabilitation placement persists even when important clinical indicators of rehabilitation need are accounted for.

The cohort was drawn from a single health system, the University of Utah, with data collected between January 1, 2020 and July 31, 2024. The IRF group of 171 participants was enrolled prospectively, with informed consent, while the SNF group of 65 participants was identified retrospectively from the health system’s enterprise data warehouse, because prospective recruitment was disrupted during 2020–2022 by the COVID-19 pandemic. All participants had confirmed ischemic stroke verified by imaging, and all had been living independently in the community before their stroke, a fact documented through routine questioning by physical or occupational therapists and recorded in discrete chart fields.

The clinical variables the team examined were comprehensive. Stroke severity was measured with the National Institutes of Health Stroke Scale (NIHSS), which evaluates 11 areas and ranges from 0 to 42 points. Comorbidity burden was quantified with the Charlson Comorbidity Index (CCI). Functional status during acute hospitalization was captured with the Activity Measure for Post-Acute Care (AM-PAC)

The AM-PAC “6-Clicks” instrument deserves particular attention because of how it is used in routine care. It is a brief functional assessment that hospital therapists are required to complete, capturing three distinct domains: Basic Mobility, which reflects movements such as transfers and walking; Daily Activities, which covers self-care and everyday tasks; and Applied Cognition, which addresses thinking and communication skills in practical contexts. Because these measures are collected as part of standard workflow rather than as special research assessments, they offer a snapshot of how patients actually present during the acute admission, not just how they perform under study conditions. In this cohort, the researchers also recorded whether patients showed signs of neglect, a condition in which stroke survivors fail to attend to one side of their body or environment, drawing on both the NIHSS and clinical chart review.

The statistical approach distinguished between two levels of evidence. In univariable analyses, where each factor is considered on its own, several characteristics appeared to matter: younger age, cohabitating marital status, living with family, having assistance available, shorter hospital length of stay, private insurance, and even higher NIHSS scores were all associated with greater odds of going to an IRF. That last finding may seem counterintuitive, since more severe strokes might be expected to push patients toward less intensive settings, but it likely reflects the reality that patients with very mild strokes are sometimes discharged home directly, leaving the IRF population enriched for those with measurable rehabilitation needs.

The picture changed substantially once the researchers moved to multivariable analysis, which considers all factors simultaneously. In that adjusted model, only two variables retained independent associations with discharge destination: increasing age, with an odds ratio of 0.95 per year of age, and greater comorbidity burden on the Charlson Comorbidity Index, with an odds ratio of 0.85 per point. Both confidence intervals excluded the null value, indicating statistically reliable associations. Notably, NIHSS scores and AM-PAC Daily Activities scores were no longer associated with placement after adjustment, and marital status, which had appeared influential in univariable analysis, also lost its association. This pattern suggests that the apparent influence of social support and insurance type in simpler analyses may have been entangled with age and comorbidity rather than operating as truly independent drivers of placement decisions.

The odds ratios, while modest in magnitude, translate into meaningful differences across the age spectrum typical of stroke populations. A difference of twenty years in age, for example, corresponds to roughly a threefold difference in the odds of IRF placement when the per-year odds ratio of 0.95 is compounded. Similarly, each additional point on the Charlson Comorbidity Index, which weights conditions such as diabetes, heart failure, and prior malignancy, steadily reduces the likelihood of IRF discharge. Because the study population was restricted to people who had been independent before their stroke, these associations cannot be attributed to pre-existing disability, strengthening the conclusion that age and comorbidity themselves are shaping placement decisions.

The finding that functional status measured during the acute hospitalization did not differ significantly between the two groups, despite their different destinations, is among the most thought-provoking results. If patients going to SNFs and IRFs had comparable AM-PAC scores and similar stroke severity, then the divergence in placement suggests that factors beyond measured rehabilitation need are at work. The authors’ framing of these as non-clinical factors is deliberate: age and comorbidity are clinical in a broad sense, but they are not direct measures of current functional capacity or rehabilitation potential. They may instead serve as proxies for how clinicians, families, and payers perceive a patient’s ability to tolerate intensive therapy, or for expectations about recovery trajectory that are not captured by standardized assessments.

The policy context helps explain why this matters. Medicare regulations require that IRF patients be able to participate in at least three hours of therapy per day, five to six days per week, and demonstrate capacity for measurable functional improvement. These criteria are inherently judgment-laden: predicting who can tolerate intensive therapy and who will improve is not an exact science, and clinicians may reasonably weigh age and comorbidity when making such predictions. But if those judgments systematically channel older patients and those with chronic conditions toward less intensive settings, even when their measured function and stroke severity are comparable to peers who go to IRFs, then some patients may be receiving less rehabilitation than guidelines would suggest is optimal during the critical early recovery window.

The timing of the study window adds an important layer of interpretation. Data collection spanned January 2020 through July 2024, a period encompassing the acute disruptions of the COVID-19 pandemic and its aftermath on hospital operations, rehabilitation capacity, and discharge planning norms. The retrospective identification of the SNF cohort was itself a consequence of pandemic-related disruption to prospective recruitment. While the researchers applied consistent variable definitions across both cohorts, differences in how prospective and retrospective data were ascertained represent a methodological consideration when interpreting the results.

Several limitations frame the findings appropriately. The single health system sample, drawn from an academic medical center in Utah, may not generalize to regions with different rehabilitation capacity, payer mixes, or discharge planning cultures. The sample size of 236, while adequate for the adjusted analyses performed, limits the precision with which less common characteristics could be evaluated. Additionally, the study captured placement at discharge from acute care but did not follow patients forward to compare functional outcomes between those who went to IRFs versus SNFs, so the findings speak to placement patterns rather than to the consequences of those patterns for recovery.

Nevertheless, the study’s central contribution is clear: among previously independent stroke survivors with comparable measured rehabilitation need, who ends up where is shaped substantially by age and comorbidity. This raises questions worth pursuing, including whether these factors reflect appropriate clinical judgment about therapy tolerance, implicit biases in discharge planning, structural constraints such as insurance coverage and bed availability, or some combination of all three. Understanding which of these mechanisms drives the observed associations will be essential if the goal is to ensure that stroke survivors’ access to intensive rehabilitation during the first ninety days after stroke, the period most critical for optimizing functional recovery, is determined by their rehabilitation potential rather than by characteristics that merely travel alongside it.

Subject of Research: Non-clinical and clinical factors associated with post-acute rehabilitation placement after acute ischemic stroke

Article Title: Different rehabilitation destinations: non-clinical factors associated with inpatient rehabilitation versus skilled nursing placement

Article References: Hayes, H. A., Zhang, C., Xiang, S., Smith, B., Williams, P., Presson, A. P., & French, M. (2026). Different rehabilitation destinations: non-clinical factors associated with inpatient rehabilitation versus skilled nursing placement. Frontiers in Rehabilitation Sciences, 7, Article 1984193. https://doi.org/10.3389/fresc.2026.1984193

Image Credits: AI Generated

DOI: 10.3389/fresc.2026.1984193

Keywords: stroke rehabilitation, inpatient rehabilitation facility, skilled nursing facility, discharge planning, post-acute care, NIHSS, Charlson Comorbidity Index, AM-PAC, health disparities, age, comorbidity, ischemic stroke