Cognitive Impairments and Health System Outcomes in Inpatient Rehabilitation: A Comparison of Patients With Stroke and Those With Type II Diabetes Mellitus

Abstract

Importance: Cognitive impairment is associated with poor outcomes in inpatient rehabilitation (IPR) but may be underrecognized among patients without neurologic diagnoses.

Objective: To compare cognitive impairment prevalence between IPR patients with ischemic stroke and patients with Type II diabetes mellitus (T2DM) without a cerebrovascular diagnosis and to examine associations with functional outcomes and readmissions.

Design: This observational, cross-sectional study used retrospective electronic medical record data collected from 2019 to 2022.

Setting: Single inpatient rehabilitation facility in an academic health system.

Participants: The sample consisted of 360 patients admitted to IPR, including 147 patients with ischemic stroke without T2DM and 213 patients with T2DM without a neurologic diagnosis.

Outcomes and Measures: Cognitive status was assessed at admission using the Montreal Cognitive Assessment (MoCA). Outcomes included discharge self-care and mobility (Section GG), within-stay readmission, and 30-day postdischarge readmission.

Results: Mild cognitive impairment was more prevalent among patients with T2DM than among patients with stroke at admission. Patients with stroke had lower mean MoCA scores at admission and discharge. Diagnosis was not associated with discharge self-care, discharge mobility, or readmission outcomes. Higher admission functional status and lower disease burden were associated with better functional outcomes, and higher admission mobility was associated with lower odds of within-stay readmission.

Conclusions and Relevance: Routine, diagnosis-agnostic cognitive screening identified high rates of cognitive impairment across diagnostic groups in IPR. Occupational therapy practitioners should incorporate systematic cognitive screening and follow-up functional cognition assessment to guide intervention planning and discharge preparation.

Plain-Language Summary: Many people receiving inpatient rehabilitation experience problems with thinking and memory that can affect daily activities and recovery. These problems are often expected after stroke but may be overlooked in people with other health conditions, such as diabetes. This study compared cognitive impairment and rehabilitation outcomes in people with stroke and people with Type II diabetes who did not have a stroke. The study found that mild cognitive impairment was common in both groups, including many people with diabetes. Whether a person had a stroke or diabetes did not explain differences in functional outcomes or hospital readmissions. Instead, a person’s functional ability at admission and overall health burden were more strongly related to outcomes. These findings suggest that occupational therapy practitioners should routinely screen cognition for all patients in inpatient rehabilitation, not only those with neurologic diagnoses. Early identification of cognitive challenges can support therapy planning, daily functioning, and safe discharge.

This study compared cognitive impairment and rehabilitation outcomes in people with stroke and people with Type II diabetes who did not have a stroke.

Cognitive impairment occurs in 50% of people by age 70 yr and is projected to rise dramatically because of aging populations and increasing rates of risk factors such as hypertension and obesity (GBD 2019 Dementia Forecasting Collaborators, 2022; Hale et al., 2020). Cognitive impairments are linked to poor outcomes, including falls, rehospitalization, and mortality (Fogg et al., 2018; Middleton et al., 2016). Early detection of cognitive impairment is crucial in guiding treatment planning and long-term care, yet the diagnosis of cognitive impairment is often delayed or missed, especially in patients without overt neurologic diagnoses (Morley et al., 2015; Sampson et al., 2009; Schimming et al., 2017). With its intensive, multidisciplinary approach, inpatient rehabilitation (IPR) is a key setting for the identification and initiation of targeted intervention to treat cognitive impairment (American Occupational Therapy Association, 2020). Additionally, compared with acute care or outpatient rehabilitation settings, therapists in IPR have more time and opportunity to conduct standardized cognitive screening as part of routine care.

However, current rehabilitation cognitive screening practices tend to prioritize patients with neurologic diagnoses such as stroke, in which cognitive impairment is well recognized. In contrast, patients with other chronic conditions may have cognitive risks that are comparable yet often overlooked. Type 2 diabetes mellitus (T2DM) is a prime example. One in four inpatient stays (26.4%) in the United States involved diabetes as a comorbid condition, making it one of the most common chronic conditions encountered in hospital and rehabilitation settings (Owens et al., 2022). However, cognitive impairment in patients with T2DM is frequently missed, even though it is associated with deficits in executive function, attention, and processing speed that can occur independently of a neurologic diagnosis (e.g., stroke) because of the systemic effects of chronic inflammation, insulin resistance, and vascular injury (Kodl & Seaquist, 2008; Schimming et al., 2017). Failure to identify these cognitive impairments in patients with T2DM represents a missed opportunity in rehabilitation to provide therapeutic interventions and connect patients with resources that could support community engagement and long-term self-management (Aronow, 2017; Rouch et al., 2023; Saedi et al., 2016; Verdelho et al., 2021).

Mandated by the Centers for Medicare and Medicaid Services (CMS) across all postacute settings, including IPR, the Brief Interview for Mental Status (BIMS) is widely used but lacks sensitivity to the milder cognitive impairments and domains that are most affected by T2DM, such as executive function (Harmon & Gillen, 2023; Li et al., 2022; Marks et al., 2021; Saliba et al., 2012). Reliance on the BIMS may contribute to underrecognition of cognitive deficits in patients with T2DM, further limiting opportunities for timely intervention. In contrast, the Montreal Cognitive Assessment (MoCA) assesses multiple cognitive domains, including attention and concentration, executive function, orientation, memory, language, and conceptual thinking; compared with the BIMS, it has demonstrated greater sensitivity for detecting mild cognitive impairments (Alagiakrishnan et al., 2013; Cumming et al., 2011). At our study site, a clinician-led quality improvement initiative established the MoCA as a routine admission screen for all inpatients, regardless of admitting diagnosis. This site-level policy goes beyond the CMS-mandated BIMS and was designed to improve detection of milder cognitive deficits (e.g., executive function) that affect rehabilitation participation and self-management, particularly in conditions such as T2DM. This context provides a naturalistic test of what broader, diagnosis-agnostic screening reveals in everyday practice.

This study had two aims. The first aim was to compare the prevalence of cognitive impairment in patients admitted with ischemic stroke, who are routinely screened for cognitive impairments caused by neurologic injury, with that in patients who have T2DM without cerebrovascular diagnoses, a non-neurologic group in whom screening is less common. The second aim was to examine how diagnosis (stroke versus T2DM), admission cognitive status (MoCA), and baseline functional status are associated with health system outcomes, including self-care, mobility, and hospital readmissions. Conducting this comparison within an IPR setting that uses routine MoCA screening for all admissions provides a unique opportunity to evaluate the practice implications of diagnosis-agnostic cognitive screening beyond the BIMS.

Method

Data Source and Cohort Selection

We performed an observational, cross-sectional study using retrospective, deidentified, patient-level electronic medical records (EMRs) from patients who were admitted to an IPR facility within a large health system between December 2019 and December 2022. Patients were included if they met criteria for one of two mutually exclusive groups: (1) a primary diagnosis of ischemic stroke without T2DM or (2) a diagnosis of T2DM without ischemic stroke. The primary diagnosis of stroke was identified using the rehabilitation impairment category, a standardized classification system that defines the patient’s main reason for IPR. Because T2DM is not represented as a primary rehabilitation diagnosis, we identified patients with T2DM using the International Classification of Diseases, 10th revision (ICD-10) codes; specifically, the E08, E09, E11, and E13 root codes. We excluded patients with underlying neurological conditions from the T2DM group, which included 40 patients with brain dysfunction and 38 patients with other neurologic conditions on the basis of their rehabilitation impairment category (for the rehabilitation impairment categories for patients in the T2DM group, see Table 1). We did not restrict other rehabilitation diagnoses within the T2DM group, because our aim was to evaluate cognitive impairment among patients with diabetes broadly, regardless of their primary admitting condition. The final analytic sample included 360 patients: 213 in the T2DM group and 147 in the stroke group. MoCA data were missing for 3 of 360 patients (0.8%). These patients were retained in the overall sample but were excluded from analyses through listwise deletion. The study received approval from the University of Pittsburgh institutional review board.

Table 1.

Rehabilitation Admission Diagnoses for Patients With Type 2 Diabetes Mellitus Without Neurological Conditions

Rehabilitation Diagnosis	n (%)
Amputation	23 (10.8)
Cardiac conditions	4 (1.9)
Debility	98 (46.0)
Major multiple trauma	1 (0.5)
Orthopedic disorders	66 (31.0)
Other disabling impairments	3 (1.4)
Pneumonia	1 (0.5)
Pulmonary disorders	6 (2.8)
Rehabilitation	1 (0.5)
Spinal cord dysfunction	10 (4.7)
Total	213 (100)

Note. This table includes only patients in the diabetes-only group. Diagnoses are based on the primary rehabilitation impairment category recorded at admission.

MoCA Implementation Context

During the study window, at the study site, we operated a clinician-initiated quality improvement program that embedded the MoCA as a standard intake screen for all admissions. Licensed speech-language pathologists administered the MoCA within the first 3 days of admission as part of the routine intake workflow. Screening was not contingent on suspected impairment or specific admitting diagnoses; possible exceptions occurred only when patients were medically unstable or declined testing. This operational model minimizes selection bias and supports near-universal capture of MoCA scores at admission.

Measures

Cognitive Impairment

We used the Montreal Cognitive Assessment (MoCA) as the universal cognitive screening tool at the study site. The MoCA evaluates multiple domains—its scores range from 0 to 30, with scores lower than 26 suggesting impairment—and it has demonstrated validity in patients with both T2DM and stroke (Alagiakrishnan et al., 2013; Gupta et al., 2024) and in patients with stroke only (Feng et al., 2021; Koski, 2013). At our study site, speech-language pathologists who are MoCA certified were designated to administer the routine admission cognitive screen for all patients, and they administered the MoCA within the first 3 days of admission during our study timeframe.

Health System Outcomes

We examined four health system outcomes (self-care status at discharge, mobility status at discharge, within-stay readmission, and 30-day readmission) between the two study groups. All four health system outcomes are part of the Inpatient Rehabilitation Facility (IRF) Quality Reporting Program, a CMS initiative that requires IRFs to collect and report data on patient outcomes and care quality to improve patient care and ensure transparency (Centers for Medicare and Medicaid Services, 2023). Discharge self-care and mobility were defined according to Section GG in the IRF Patient Assessment Instrument, which assesses a patient’s functional abilities in self-care and mobility. Section GG must be completed for patients within the first 3 days of admission and the last 3 days before discharge. Each Section GG item is scored on the basis of patients’ usual performance rather than on their best or worst performance. The scoring scale ranges from 1 (dependent) to 6 (independent). Section GG has seven self-care items (e.g., eating, oral hygiene) and 15 mobility items (e.g., sit to stand, toilet transfer). If an item was coded as “activity not attempted” (Items 7, 9, 10, and 88), we recoded the score as a 1 (Deutsch et al., 2022). There were no missing data. To determine discharge scores, we summed all seven self-care items, and all 15 mobility items were summed separately (Deutsch et al., 2022; Vaughan et al., 2022). Self-care scores ranged from 7 to 42, and mobility scores ranged from 7 to 90, with higher scores indicating greater levels of independence.

Within-stay readmissions and 30-day postdischarge readmissions were also included as health system outcomes. Within-stay readmissions are defined as those occurring after a patient has been admitted to IPR but before discharge. We only included within-stay readmissions that were 3 days or less in the acute hospital, because shorter readmissions are often scrutinized for the potential to reflect preventable issues. A 30-day readmission refers to a patient’s readmission to an acute hospital that occurred within 30 days after discharge from the IRF. We were unable to determine whether either type of readmission was potentially preventable, and only readmissions occurring within the same hospital system were identified from the available data.

Covariates

The following covariates were included in our logistic and regression models: race (White or non-White), gender (male or female), self-care score at admission (defined by Section GG items), mobility score at admission (defined by Section GG items), MoCA score at admission, and disease burden. The disease burden variable was the total number of a patient’s ICD-10 codes during hospitalization.

Statistical Analyses

We examined whether diagnosis (T2DM versus stroke), cognitive status at admission (MoCA score), and baseline functional status (Section GG self-care or mobility item, depending on outcome) are associated with discharge functional outcomes and readmissions. We assessed group differences (T2DM versus stroke) in length of stay, MoCA score at admission or discharge, self-care, and mobility using independent t tests, and we assessed categorical variables using χ² tests. We estimated two linear regression models for self-care and mobility outcomes at discharge and two logistic regression models for within-stay and 30-day readmission outcomes. All models were adjusted for age, gender, race, disease burden, MoCA score at admission, and the relevant functional score (self-care or mobility) at admission. For modeling, disease burden was categorized as 26 or fewer, 27 to 34, 35 to 42, and 43 or more ICD-10 codes. Missing data were handled by means of listwise deletion (SPSS, Inc.) so that patients without MoCA scores (0.8%) were excluded from the analyses. We used multiple statistics to evaluate the goodness-of-fit statistics for both the linear and logistic regression models. For the linear models, we assessed the R² and adjusted R² values, whereas for the logistic regression models, we assessed the pseudo r² values and used the Hosmer-Lemeshow test. Before running the regression models, we assessed all the independent variables for multicollinearity to ensure that there were no strong correlations that could affect the stability of the estimates.

Results

We analyzed cognitive and functional outcomes for 213 patients with T2DM without ischemic stroke and 147 patients with ischemic stroke without T2DM. At admission, a higher proportion of patients in the T2DM group had mild cognitive impairments (57.5%) compared with those in the stroke group (37.4%; Figure 1). Nearly 90% of participants in both groups exhibited some form of cognitive impairment at both admission and discharge, with minimal transition across categories during their IPR stay. MoCA completion at admission was 99.2% (357/360), reflecting the routine intake workflow.

Figure 1.

Comparison of cognitive impairment at admission and discharge between patients with stroke and patients with diabetes.

Note. Bar graphs depict the percentage of patients classified into four cognitive status categories—severe cognitive impairment (Montreal Cognitive Assessment [MoCA] scores ≤10), moderate impairment (MoCA scores 11–17), mild impairment (MoCA scores 18–25), and normal cognitive function (MoCA scores ≥26)—at admission (left) and at discharge (right) during inpatient rehabilitation. Orange bars represent patients with stroke, and black bars represent patients with diabetes. At both time points, a significantly greater proportion of patients with diabetes had mild cognitive impairment, whereas a greater proportion of patients with stroke had moderate or severe impairment. MoCA scores at discharge were missing in 16% of patients, which may limit interpretation of these descriptive comparisons.*p < .05.

MoCA scores were significantly lower among patients with stroke at admission (M = 17.52, SD = 6.25) and discharge (M = 18.02, SD = 6.31) than among patients with T2DM (admission: M = 19.94, SD = 5.11; discharge: M = 20.18, SD = 5.07); p < .001 and p = .001, respectively (Table 2). Although patients with stroke had higher mobility scores at admission (M = 35.54, SD = 12.57) than those with T2DM (M = 32.91, SD = 12.84), the difference did not reach statistical significance (p = .056). There were no significant group differences in self-care scores at admission or discharge, mobility scores at discharge, or length of stay. In terms of demographics, the T2DM group had a significantly higher proportion of male participants (59.6%) than the stroke group (44.2%, p = .004). Although a greater percentage of patients with stroke identified as a racial minority (17.0%) compared with that in the T2DM group (10.3%), this difference did not reach significance (p = .065). Discharge to home with home health services was the most common destination in both groups, although patients with stroke were more often discharged home with no services (11.6% versus 4.2%, p = .043). Disease burden differed significantly between groups, with a higher proportion of patients with T2DM categorized as having high disease burden (≥43 ICD-10 codes; 29.6% vs. 11.3%; p < .001).

Table 2.

Patient Characteristics and Outcome by Diagnosis Group: Stroke Only and Diabetes Only

Variable	n (%)		p
	Stroke Only (n = 147)	Diabetes Only (n = 213)
Score, M (SD)
MoCA at admission	17.52 (6.25)	19.94 (5.11)	<.001**
MoCA at discharge	18.02 (6.31)	20.18 (5.07)	.001*
Self-care at admission	23.16 (6.06)	22.79 (5.28)	.539
Mobility at admission	35.54 (12.57)	32.91 (12.84)	.056
Self-care at discharge	34.79 (7.85)	34.32 (7.66)	.574
Mobility at discharge	67.47 (19.95)	63.22 (21.91)	.062
Within-stay readmission
No	135 (91.8)	185 (86.9)	.139
Yes	12 (8.1)	28 (13.1)
30-day readmission
No	135 (91.8)	190 (89.2)	.407
Yes	12 (8.1)	23 (10.8)
Age, M (SD)	73.41 (12.64)	72.08 (11.63)	.303
Length of stay, day, M (SD)	17.43 (9.50)	16.03 (7.70)	.13
Discharge destination			.043*
Home with no services	17 (11.6)	9 (4.2)
Home with home health	95 (64.6)	147 (68.8)
Skilled nursing facility	20 (13.6)	22 (10.3)
Other	15 (10.2)	35 (16.4)
Gender			.004*
Male	65 (44.2)	127 (59.6)
Female	82 (55.8)	86 (40.4)
Race			.065
Non-White	25 (17.0)	22 (10.3)
White	122 (83.0)	191 (89.7)
Disease burden			<.001**
≤26	58 (41.1)	34 (16.0)
27–34	44 (31.2)	60 (28.2)
35–42	23 (16.3)	56 (26.3)
≥43	16 (11.3)	63 (29.6)

Note. Values are presented as M (SD) for continuous variables and n (%) for categorical variables. The p values reflect comparisons between patients in the stroke-only group and patients in the diabetes-only group using independent t tests for continuous variables and χ² tests for categorical variables. Montreal Cognitive Assessment (MoCA) scores reflect cognitive performance at admission and discharge, with higher scores indicating better function. Self-care and mobility scores were assessed at admission and discharge using the Section GG functional outcome scales from the Inpatient Rehabilitation Facility Patient Assessment Instrument. Disease burden reflects the number of diagnostic codes from the International Classification of Diseases (10th revision), grouped into quartiles.

*p < .01. **p < .001.

Linear Regression Results for Self-Care and Mobility at Discharge

Diagnosis (T2DM versus stroke) was not a significant predictor of scores for self-care or mobility at discharge (Table 3). However, higher self-care scores (p < .001; 95% confidence interval [CI] [0.520, 0.766]) and MoCA scores (p = .002; 95% CI [0.072, 0.321]) at admission were significantly associated with higher self-care scores at discharge, whereas older age was associated with lower scores (p = .005; 95% CI = [−0.140, −0.025]). Lower disease burden was significantly associated with higher self-care scores at discharge, with patients who had ≤26 ICD–10 diagnosis codes (p = .019; 95% CI [0.407, 4.503]) and those with 27 to 34 codes (p = .006; 95% CI [0.730, 4.418]) performing better than those with the highest disease burden (≥43 codes). In the mobility model, higher mobility scores at admission (p < .001; 95% CI = [0.819, 1.084]) and lower disease burden were significantly associated with better mobility at discharge. Specifically, compared with patients with the highest disease burden (≥43 ICD–10 diagnosis codes), those with 26 or fewer codes (p = .001; 95% CI [4.093, 14.514]) and those with 27 to 34 codes (p < .001; 95% CI [4.901, 14.100]) had significantly better mobility outcomes. Older age was also associated with lower mobility scores at discharge (p = .019; 95% CI [−0.315, −6]; p = .001); and MoCA scores were marginally associated with mobility at discharge (p=.066; 95% CI [−0.019, 0.589]).

Table 3.

Regression Models Identifying Factors Associated With Functional Outcomes at Discharge: Self-Care and Mobility

Variable	Self-Care Discharge Score				Mobility Discharge Score
	B	SE	p	95% CI	B	SE	p	95% CI
Stroke diagnosis (ref.: diabetes)	0.03	0.725	.966	[−1.395, 1.456]	−0.38	1.817	.835	[−3.954, 3.194]
MoCA admission score	0.196	0.063	.002**	[0.072, 0.321]	0.285	0.154	.066	[−0.019, 0.589]
Admission function score	0.643	0.063	<.001***	[0.520, 0.766]	0.951	0.067	<.001***	[0.819, 1.084]
Age, yr	−0.083	0.029	.005**	[−0.140, −0.025]	−0.171	0.073	.019*	[−0.315, −0.028]
Racial minority (ref.: White)	0.564	1.01	.577	[−1.422, 2.551]	2.592	2.525	.305	[−2.376, 7.559]
Male gender (ref.: female)	0.174	0.664	.793	[−1.132, 1.480]	0.269	1.659	.871	[−2.995, 3.533]
Disease burden (ref.: ≥43)
≤26	2.455	1.041	.019*	[0.407, 4.503]	9.303	2.649	.001**	[4.093, 14.514]
27–34	2.574	0.937	.006**	[0.730, 4.418]	9.5	2.338	.000***	[4.901, 14.100]
35–42	−0.176	0.973	.856	[−2.090, 1.738]	3.146	2.435	.197	[−1.644, 7.936]

Note. For the admission function score, the self-care admission score was entered for the self-care discharge model and the mobility admission score was entered for the mobility discharge model. The Montreal Cognitive Assessment (MoCA) admission score refers to the initial cognitive function score measured with the MoCA. MoCA admission scores were missing for 3 patients (0.8%) and were excluded from analyses through listwise deletion. CI = confidence interval; ref. = reference category.

*p < .05. **p < .01. ***p < .001.

Logistic Regression Results for Readmission Outcomes

Logistic regression results similarly indicated that diagnosis was not associated with within-stay or 30-day readmission (Table 4). However, compared with female patients, male patients had significantly higher odds of within-stay readmission (odds ratio [OR] = 2.593; 95% CI = [1.172, 5.736]; p = .019), and higher mobility scores at admission were associated with lower odds of within-stay readmission (OR = 0.933; 95% CI = [0.903, 0.965]; p < .001). A lower disease burden (codes ≤26) also reduced the odds of within-stay readmission (OR = 0.105; 95% CI = [0.012, 0.906]; p = .040). No variables were significantly associated with 30-day readmission.

Table 4.

Logistic Regression Models Predicting Within-Stay and 30-Day Readmission

Variable	Within-Stay Readmission			30-Day Readmission
	OR	95% CI	p	OR	95% CI	p
Stroke diagnosis (ref.: diabetes)	0.944	[0.396, 2.251]	.896	0.744	[0.326, 1.696]	.482
MoCA admission score	1.022	[0.952, 1.097]	.554	0.98	[0.914, 1.050]	.559
Mobility admission score	0.933	[0.903, 0.965]	.000**	1.014	[0.984, 1.045]	.367
Racial minority (ref.: White)	1.537	[0.455, 5.184]	.489	1.427	[0.395, 5.154]	.587
Male gender (ref.: female)	2.593	[1.172, 5.736]	.019*	2.279	[1.067, 4.870]	.033*
Disease burden (ref.: ≥43)
≤26	0.105	[0.012, 0.906]	.04*	0.496	[0.153, 1.605]	.242
27–34	0.715	[0.270, 1.889]	.498	0.672	[0.258, 1.746]	.414
35–42	1.261	[0.510, 3.120]	.616	0.637	[0.229, 1.775]	.389

Note. The mobility admission score refers to functional mobility score at admission according to the Inpatient Rehabilitation Facility Patient Assessment Instrument. The Montreal Cognitive Assessment (MoCA) admission score refers to the initial cognitive function score measured with the MoCA. MoCA admission scores were missing for 3 patients (0.8%) and were excluded from analyses through listwise deletion. CI = confidence interval; ref = reference category.

*p < .05. **p < .001.

Discussion

This study shows what an IRF staff learns when cognitive screening is routine and diagnosis agnostic: High rates of mild cognitive impairment are identified across populations, including people with T2DM without overt neurologic diagnoses. In this context, rehabilitation decision-making that prioritizes only the primary diagnosis risks missing actionable cognitive needs. Our findings indicate that baseline function and disease burden, not diagnosis alone, most strongly relate to outcomes, which supports routine screening beyond the BIMS and timely rehabilitation follow-up with occupation-based functional cognition assessment. Although T2DM is not typically an admitting diagnosis of focus in an IRF, it is pervasive, with 26.4% of U.S. inpatient stays including diabetes as a comorbidity in 2019 (Owens et al., 2022). These rates highlight the importance of systematic screening across diagnostic categories.

We found that patients in the stroke group demonstrated significantly lower cognitive function at both admission and discharge compared with those in the T2DM group. These findings were unsurprising, because mild to severe cognitive impairments poststroke affect up to 60% of patients after a stroke, with higher rates occurring immediately after the stroke (Ihle-Hansen et al., 2011; Lo et al., 2019). The difference between the MoCA scores for the study groups was ∼2 points for both admission and discharge. Interpretation of these findings is difficult. The 2-point difference does not meet the minimal detectable change (MDC) of 5.1 points (Lindvall et al., 2024). However, it does meet the minimal clinically important difference (MCID), which, depending on the anchor or method, ranges between 1.0 to 2.0 points. Despite the reported values for the MCID, warnings have been issued when using a MCID of 1.0 to 2.0 points for the MoCA. Although a difference of 1.0 to 2.0 points on the MoCA may be seen as clinically meaningful by patients and clinicians, inherent variability exists with scoring the MoCA, and sometimes a difference of 1.0 to 2.0 points may just be due to scoring variability, not to a clinically meaningful change in cognition. Therefore, our results should be interpreted with caution. Although there is a statistically significant difference between the two groups that is based on their MoCA scores at admission and discharge, the difference does not meet the MDC threshold, which indicates a true change in cognitive function that is not attributed to measurement error. As a result, the cognitive differences between the two groups may be minimal in the clinical setting.

The primary purpose of the MoCA is to detect mild cognitive impairments (Nasreddine et al., 2005). At admission, most patients in our study sample were classified as either having moderate (stroke: 42.9%, T2DM: 26.3%) or mild (stroke: 37.4%, T2DM: 59.9%) cognitive impairments according to the MoCA, with very few patients transitioning to different categories by discharge. This highlights the need to administer outcome measures such as the BIMS, which are sensitive to mild cognitive impairments in IPR, not just severe cognitive impairments. Up to 92% of patients with mild cognitive impairments are undiagnosed and, therefore, untreated (Mattke et al., 2023). Our findings confirm that mild cognitive impairments are prevalent in IPR, making it an ideal setting for identifying mild cognitive impairment and ensuring that patients are receiving effective treatment and useful resources.

It is important to acknowledge that paper-and-pencil tools such as the MoCA have known limitations, including potential biases related to education, language, cultural background, and motor or sensory impairments (Caporusso et al., 2025; Sahoo & Grover, 2022). They capture performance in isolated cognitive domains rather than how cognition manifests in everyday occupations. For this reason, MoCA results should be viewed as an initial screening step that prompts more comprehensive, occupation-based assessments to evaluate functional cognition. Such assessments can guide interventions that both remediate cognitive deficits and support participation through compensatory strategies, aligning with rehabilitation’s ultimate goal of optimizing function and independence.

Our regression analyses showed that diagnosis (T2DM versus stroke) was not a significant predictor of any of the health system outcomes: self-care discharge score, mobility discharge score, within-stay readmission, or 30-day readmission. Instead, higher admission functional status, self-care or mobility depending on the model, was consistently associated with better functional outcomes and a reduced likelihood of within-stay readmission. Admission functional status is a strong predictor of patient outcomes at discharge from IPR and long term (Chang et al., 2013; Chu et al., 2023; Middleton et al., 2017; Sands et al., 2003). These findings highlight the importance of focusing on baseline functional status as a key predictor of patient progress and needs during and after their IPR stay. Early identification of patient needs can guide tailored interventions and support multidisciplinary collaboration, ultimately facilitating smoother transitions back to the community for patients after IPR. Future studies may consider creating subgroups of patients on the basis of their admission functional status and examining variation in health system outcomes across the subgroups.

Limitations

Our study has several limitations. We used EMRs from a single health system, which may introduce data accuracy issues and limit generalizability in terms of patient demographics and diagnoses. Only patients who accessed IPR services within this system were included, introducing potential selection bias, and the lack of all-payer claims data may further restrict representation. We did not have information on diabetes control status (e.g., controlled versus uncontrolled, with versus without complications); extracting such detail would require advanced methods such as natural language processing of clinical notes, which are beyond the scope of this project. Similarly, data on stroke lesion location were not available, which is a limitation common to most large rehabilitation datasets, despite its known impact on cognitive outcomes. To address heterogeneity across admitting diagnoses, we controlled for primary diagnoses of stroke and T2DM, presented the distribution of admitting diagnoses in Table 1, and included MoCA scores to capture cognitive status, although we did not adjust for every admitting diagnosis. Further adjustment for each diagnosis would have substantially expanded the number of categories in the regression models, increasing the risk of sparse cells, unstable estimates, and reduced interpretability, given our sample size. Instead, we excluded neurological diagnoses from the T2DM group to reduce overlap and enhance internal validity. This analytic approach is consistent with prior rehabilitation outcomes research and allowed us to focus on the unique contributions of T2DM and stroke to cognitive impairment and health system outcomes (Andrews et al., 2015; Freburger et al., 2012; Kumar et al., 2019). Future studies should integrate more granular clinical data (e.g., diabetes control, stroke lesion characteristics), link EMRs with all-payer claims, and test these associations across multiple rehabilitation settings to strengthen generalizability.

Implications for Occupational Therapy Practice

In this single-site inpatient rehabilitation study, routine cognitive screening identified high rates of cognitive impairment in both study diagnostic groups. Diagnosis (stroke vs. T2DM) did not explain differences in self-care at discharge, mobility at discharge, or readmission outcomes; instead, admission function and disease burden showed the clearest relationships with outcomes. These findings suggest practical ways occupational therapy practitioners can strengthen assessment and care planning when cognition may be present but not obvious. Practitioners should

• ▪ screen cognition for all IPR admissions—not only neurologic diagnoses. Many patients without stroke can still have cognitive impairment that may affect participation, learning, and safety.

• ▪ treat cognitive screen results as a starting point, then assess functional cognition. Use occupation-based observation and follow-up assessment to understand how cognition shows up during activities of daily living and instrumental activities of daily living (e.g., sequencing, safety awareness, problem solving, self-management).

• ▪ use self-care and mobility at admission to guide early care planning. Because baseline function is strongly related to discharge function (and admission mobility is related to within-stay readmission), build intervention plans and supports around what the person can do at admission and where they need cues, setup, or supervision.

• ▪ factor in overall medical complexity. Higher disease burden is linked to poorer functional outcomes (and lower disease burden to lower within-stay readmission odds), so disease burden should be considered when anticipating barriers to progress and when planning discharge supports and follow-up needs.

• ▪ communicate cognition-related needs in plain, actionable terms. Document and share specific performance issues (e.g., needs cueing for sequencing, difficulty with new learning, reduced insight) to guide team decisions about caregiver training, supervision level, and discharge recommendations.

A feasible way to extend these findings to routine practice is to pair diagnosis-agnostic cognitive screening with a consistent occupational therapy follow-up process: (1) Interpret screening results in terms of everyday performance, (2) prioritize early functional cognition assessment when risk is flagged, and (3) tailor intervention and discharge planning using admission function and medical complexity rather than diagnosis alone.

Conclusion

This study reinforces that functional status and cognitive screening—rather than diagnosis alone—are central to understanding patient outcomes in IPR. Despite differences in underlying conditions, both groups demonstrated comparable rehabilitation potential, with higher baseline functional status being a strong predictor of better discharge functional outcomes and lower within-stay readmission rates. These findings emphasize the critical role of early and thorough cognitive and functional assessments in IPR that are not based solely on primary diagnosis, which can guide tailored interventions and enhance patient outcomes. Future research should focus on exploring subgroup variations that are based on admission functional status and integrating broader datasets, such as all-payer claims data, to improve the generalizability and robustness of findings across diverse patient populations. These findings underscore the importance of a multidisciplinary approach in IPR that prioritizes early identification of patient needs, ensuring that all patients, regardless of diagnosis, receive the tailored care and resources necessary for successful reintegration into the community.