Association of physical function with hospital readmissions among older adults: A systematic review

Abstract

Background

Hospital readmissions pose significant burdens on healthcare systems, particularly among older adults. While efforts to reduce readmissions have historically focused on medical management, emerging evidence suggests physical function may also play a role in successful care transitions. However, there is a limited understanding of the relationship between functional measures and readmission risk. This systematic review aims to assess the association between physical function impairments and hospital readmissions.

Objective

This systematic review aims to assess the association between physical function impairments and hospital readmissions.

Methods

A systematic review was conducted following PRISMA guidelines, with studies identified through databases including PubMed, CINAHL, Embase, and others published January 1, 2010–December 31, 2022. Inclusion criteria encompassed observational studies of adults aged 50 and older in the United States, reporting readmissions within 90 days of discharge and assessing physical function across domains of the International Classification of Function model. Data extraction and risk of bias assessment were independently conducted by two authors using the Scottish Intercollegiate Guidelines Network (SIGN) tool.

Results

Seventeen studies, representing 80,008 participants, were included in this systematic review. Patient populations included a wide array of medical populations, including general medical inpatients and those undergoing cardiac surgery. Across various functional measures assessed before or during admission, impairments were consistently associated with increased risk for hospital readmissions up to 90 days after admission. Measures of participation, including life‐space mobility, were also associated with increased readmission risk.

Conclusions

Functional impairments are robust predictors of hospital readmissions in older adults. Routine assessment of physical function during hospitalization can improve risk stratification and may support successful care transitions, particularly in older adults.

INTRODUCTION

Readmissions, or the rehospitalization of a patient after being discharged, are a significant public health concern. While 30‐day readmissions are often captured in public reporting (e.g., Medicare Hospital Compare), some payment bundles used or tested by Medicare hold hospitals financially accountable for unplanned hospitalizations for up to 90 days. Excessive hospital readmissions can reflect a significant economic burden for hospitals and the healthcare system as a whole. In addition, prior work suggests readmission frequency is associated with functional impairment experienced by older adults after surgery—suggesting patient‐centered outcomes are also adversely affected by readmissions, and older adults may be uniquely vulnerable. ¹

Given the burdens of readmissions experienced by both patients and providers, identifying risk factors for readmission to target during care transitions is critical. Care transition models to improve readmission rates have historically focused on early discharge planning, improving provider‐to‐provider communications, improving coordination of care, and educating patients on how to self‐monitor their medical conditions. ² Models generally have not focused on physical function, ³ , ⁴ despite emerging evidence that limitations in physical function and mobility among older adults are not only a consequence of hospitalization but also a predictor of future readmissions. ⁵

Physical function is a crucial indicator of underlying health in older adults, providing insight into musculoskeletal, cardiovascular, metabolic, cognitive, and psychosocial well‐being. Functional impairments are important harbingers of a loss of independence with activities of daily living (ADLs) and may signal elevated care needs postdischarge. ⁶ Impairments in physical function before hospital admission may also make an individual more susceptible to the deleterious effects of immobility during hospitalization. And function may continue to decline during a hospital admission an unfortunate iatrogenic harm that stems in part from continued tension between efforts to promote mobility (e.g., ambulation) and preventing falls in the hospital (e.g., bed alarms that inhibit mobility). ⁷ Even among patients discharged to home, any continued decline in physical function that occurs early after hospital discharge is also highly associated with readmissions—suggesting proactive identification and management of physical function during care transitions is important. ⁸ , ⁹

Despite a growing body of evidence linking impaired physical function with readmissions, to our knowledge, no systematic reviews have evaluated the magnitude of these relationships among older adults. ⁵ , ¹⁰ , ¹¹ Notably, there is a major gap in understanding which functional measures are associated with readmissions (e.g., domains to assess), and how best to assess function (e.g., self‐report or objective measurement). ¹² Therefore, the purpose of this systematic review is to evaluate the relationship between impairments in physical function and hospital readmission risk among older adults.

METHODS

This systematic review was developed according to the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) statement. The protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) as CRD42020200264. ¹³

Inclusion criteria consisted of observational studies of adults with a mean age of 50 years and older who experienced hospitalization within the United States; accompanied by measures of hospital readmission during the 90 days following discharge from the hospital and a measure of physical impairment(s), activity limitation(s) or functioning before or at the time of hospital discharge. The design was constrained to hospitalization in the United States to provide metrics unique to that healthcare system. Qualitative studies, studies not in English, studies involving those receiving hospice or end‐of‐life care/services, and those focused on exclusively mental health diagnoses (e.g., schizophrenia) were excluded.

The bibliographic databases searched were Allied and Complementary Medicine (via Ovid, 1985 to search date), CINAHL (Cumulative Index to Nursing and Allied Health Literature via EBSCOhost, 1981 to search date), Embase (via Elsevier, Embase.com, 1947 to search date), OTseeker (Occupational Therapy Systematic Evaluation of Evidence, otseeker.com), and PubMed (1946 to search date). The search strategy was customized to each database with search terms relating to disability, frailty, weakness, falls, ADL, instrumental activities of daily living (IADL), physical function, ambulation, mobility, hospitalization, rehospitalization, or readmission. Frailty was included as it is often conflated with functional status. Frailty is sometimes categorized using functional measures (eg, walking performance) as opposed to a multi‐dimensional construct of vulnerability. Papers labeled with a risk factor as “frailty” were carefully reviewed and included if that designation was made exclusive from measures of impairment, activity limitations, or participation restrictions. Language was restricted to English, and databases were searched by a medical librarian from the date of their inception to December 20, 2022. We included studies published in 2010 or later to maximize the policy relevance of the findings. Details and database search strategies can be found in Supporting Information S1: Appendix A.

The first phase for identifying studies for inclusion involved screening the titles and abstracts of retrieved studies independently by two authors to identify those studies that potentially met inclusion criteria. Any eligibility disagreement between reviewers was resolved by review from a third author. In the second phase of the identification of studies for inclusion, the full text of the potentially eligible studies was retrieved and independently assessed for eligibility by two authors. Any disagreement between the two reviewers regarding the eligibility of particular studies was resolved by review of a third author. Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org). was used for the title and abstract screening and full‐text review.

Risk of bias (quality) assessment was measured for all included studies with the Scottish Intercollegiate Guidelines Network (SIGN) Methodology Checklist 3: Cohort Studies ¹⁴ as that tool has been endorsed by the American Physical Therapy Association for the review of cohort studies. ¹⁵ Two study authors independently scored the risk of bias and discrepancies were resolved by review from a third author (Supporting Information S2: Table 1).

A predetermined, standardized form was used to record data extracted from the included studies. Extracted information included demographic characteristics of subjects, data on the type and length of hospitalization, measures of physical impairment(s), measures of activity limitation(s), measures of functioning, and hospital readmission rate. Two review authors extracted data independently and discrepancies were resolved by review from a third author. The extracted data were organized in a table for evidence synthesis by the authors.

RESULTS

The initial literature search identified a total of 5485 studies. After removing 2271 duplicates, 3214 article titles and abstracts were screened based on study inclusion criteria. Five hundred and ninety articles were excluded due to not meeting study inclusion criteria, resulting in 607 advancing to full‐text review (Figure 1). Data were aggregated from the final 17 uniquely selected articles meeting study inclusion criteria. Methodological quality for the included studies was overall acceptable with three studies rated as high quality, 12 with acceptable quality, and two were rated as unacceptable.

Figure 1.

PRISMA flow diagram. This flowchart illustrates the selection process for studies included in the review. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

Number of participants

The 17 studies selected within the United States comprised 80,008 participants.

Age of participants

The mean age of subjects for studies included in this systematic review was 50 years and older.

Clinical populations

Studies included in this review evaluated readmission risk across a broad range of high‐risk populations, including older adults and individuals with chronic and acute conditions. Key groups included patients hospitalized for COPD, heart failure, cancer, myocardial infarction, and cirrhosis, as well as those undergoing major surgeries such as coronary artery bypass grafting, valve procedures, and vascular surgeries (Table 1).

Table 1.

Summary of included studies.

References	Number of participants (age)	Inclusion criteria	Exclusion criteria
Witt et al. 17	n = 87 (median age 63.5, IQR 58.1–71.3)	Admission to a general medicine service for acute exacerbation of COPD	Current admission to the intensive care unit and inability to give informed consent
Chavan et al. 22	n = 17,715 (mean age 75)	Defined cancer survivors as those having survived cancer for at least 1 year after cancer diagnosis	Developed new cancer during the study period, or died during the first 6 months since the enrollment into MCBS
Afilalo et al. 19	n = 8287 (median age 74 years)	Age 65 or older; undergoing CABG surgery, aortic valve surgery, mitral valve surgery, or CABG combined with aortic or mitral valve surgery; preoperative gait speed recorded	Unable to safely walk or had a critical preoperative status before surgery, defined as emergency or emergency salvage surgery, cardiogenic shock, or requirement for inotropic support or intra‐aortic balloon pump
Courtwright et al. 26	n = 90 (mean age 53.5)	Lung transplant recipients	Non‐lung transplant or multiorgan transplant patients
Greysen et al. 23	n = 22,289 (mean age 78.5)	Admitted to an eligible hospital at least once during the study; sensitivity analysis, restricted to those admitted for heart failure, myocardial infarction, and pneumonia	Admissions for the following reasons: transition to health maintenance organization plan within 30 days of discharge, death in hospital, or within 30 days of discharge
Dodson et al. 25	n = 3151 (mean age 81.5)	Hospitalized with AMI. SILVER‐AMI; assessment of non‐disease‐specific impairments in important functional domains at the time of hospitalization	Development of AMI secondary to another cause; transferred after >24 h; admission at an outside hospital; incarceration; inability to provide informed consent
Kapoor et al. 29	n = 403 (mean age 72)	Undergoing any surgery with a risk of a serious complication of 5% or greater as defined using the ACS calculator	Positive Mini Cognitive Test (failed 5‐min recall of three items or combination of partial recall with failed clock drawing)
Keeney et al. 27	n = 1053 (mean age not reported, range 65–90+)	History of heart failure claim	Missing data for frailty and/or function measures
Fathi et al. 28	n = 478 (mean age 59)	English‐speaking admitted to the hospital from home with CHF or COPD had an estimated life expectancy of longer than 6 months, had a telephone, and were expected to be discharged to their home	Being considered for a heart transplant or placement of a ventricular assist device, were undergoing dialysis, or were receiving intensive monitoring services for cystic fibrosis
Fischer et al. 20	n = 110 (mean age 77.35)	Inpatients 65 years and older at time of admission; ability to speak/comprehend spoken/written English; capable of giving informed consent; ambulate 150' with/without a device	Patients unable to speak/comprehend spoken/written English; required more than supervision to perform transfers and to ambulate; patients whose medical condition precluded participation, such as severe heart or lung disease; those refusing to sign informed consent; patients with neurodegenerative disease and patients with terminal disease, expected to worsen within 12 months
Joseph et al. 18	n = 101 (mean age 79)	Aged 65 years and older, ability to understand study instructions for performing repetitive elbow flexion, and having had at least one ground‐level fall causing an injury within the prior 2 weeks that led to hospital admission	Significant upper‐extremity disorders in both arms (e.g., bilateral fractures or rheumatoid arthritis with elbow or shoulder involvement), or if they were in a physical condition (e.g., such as severe head injury and unconsciousness)
Kronzer et al. 31	n = 7982 (mean age 59 for nonfallers and 60 for fallers)	Adults undergoing elective surgery who attended the preoperative assessment clinic and provided signed informed consent	Patients with >1 surgery; records with 30‐day or 1‐year survey completion times outside the study's prespecified time range
Sinvani et al. 32	n = 2383 (mean age 84.7)	Hospitalized adults 75 years and older who were admitted to the medicine service	None reported
Tapper et al. 30	n = 734 (mean age 57.3 years)	Admission to transplant unit with a diagnosis of decompensated cirrhosis or medically complicated liver transplant	If the medical record lacked complete data regarding frailty assessments
Yanquez et al. 16	n = 37 (mean age 62.0 for nonfrail, 65.6 for prefrail, 68.0 for frail)	Aged 50 years of age and older; possessing the ability to understand the study instructions for performing the function test and answering the questionnaire; diagnosed with either an acute or chronic vascular disease, which required elective or urgent but nonemergent operation	Significant upper extremity disorders in both arms (e.g., bilateral fractures or severe rheumatoid arthritis with elbow or shoulder involvement; scheduled operation that would not allow sufficient time for performing UEF test and other clinical measures; surgery was canceled
Sanchez et al. 21	n = 10,345 (mean age 81.2)	Surgery for self‐expanding transcatheter aortic valve replacement (database registry involving 350 centers)	Patients treated via direct aortic access, patients undergoing transcatheter aortic valve replacement for a failed transcatheter or surgical valve. Patients with index hospitalization of >14 days and patients who were not discharged were excluded
Schiltz et al. 24	n = 6617 (mean age 77)	All hospitalizations among those 65 and older	Hospitalizations where the person died in‐hospital or within 30 days of discharge, transferred, or left against medical advice, were excluded

Abbreviations: ACS, acute coronary syndrome; AMI, acute myocardial infarction; CABG, coronary artery bypass graft; COPD, chronic obstructive pulmonary disease; MCBS, Medicare Current Beneficiary Survey; UEF, upper extremity function.

Measurement of function by International Classification of Function (ICF) domain

Study results were organized by individual domains of the ICF Framework—this framework helps delineate functional measures commonly used for evaluation into those used to evaluate body function and structure (e.g., strength), activities (e.g., performance of bathing or dressing tasks), or participation (involvement in life situations in real‐world settings). Across the domains of impairment, activity limitation, and participation, there was a generally consistent association between lower physical function and readmissions (Table 2).

Table 2.

Relationship between physical function and hospital readmissions.

References	Population	Functional measurements	Association of function with readmissions
Chavan et al. 22	Medicare current beneficiary claims data for patients with cancer	Physical limitations: difficulty in stooping/crouching/kneeling, walking ¼ miles, reaching or extending arms above the shoulder, lifting/carrying 10 lbs., and writing/handling objects. ADL limitations include difficulty in bathing, dressing, walking in the house, eating, getting in and out of chair or bed, and using the toilet. IADL limitations include difficulty in using the telephone, shopping, doing light housework, preparing meals, and paying bills	2 or more ADL/IADL disabilities or 2 or more physical limitations, as compared with those with ≤1, had a higher count of 30‐day hospital readmissions Physical limitations: RR 1.14, 95% CI 1.00–1.29 ADL impairments: RR 1.15, 95% CI 0.99–1.33 IADL Impairments: RR 1.41, 95% CI 1.25–1.58
Afilalo et al. 19	CABG, aortic valve, mitral valve surgery, or combination of CABG and valve surgery	5‐m gait speed measured presurgically. Gait was categorized in tertiles as slow (<0.83 m/s), middle (0.83–1.00 m/s), and fast (>1.00 m/s)	Slower walking speed associated with a higher risk of readmission hazards for 30‐day readmission as compared with fast gait tertile Middle tertile: HR 1.17, 95% CI 1.06–1.29 Slow tertile: HR 1.43, 95% CI 1.29–1.59 Hazards for 30‐day readmission using continuous gait speed For each 0.1 m/s lower speed: HR 1.71, 95% CI 1.45–2.02
Courtwright et al. 26	Lung transplant recipients who survived initial discharge	Modified short physical performance battery score (0–8) including balance and chair rise assessments	SPPB score less than 6/8, associated with unplanned readmissions within 30 days, OR 3.5, 95% CI 1.1–11.8 as compared with those with >6/8 scores.
Greysen et al. 23	Health and Retirement Study Medicare beneficiaries.	Number of ADLs impairments, including self‐care activities (e.g., bathing and dressing) IADL impairment (e.g., medication management, meal preparation)	Progressive increase in the odds of readmission within 30 days with increasing impairment adjusted odds for 30‐day hospital readmission as compared with those with no functional impairment Difficulty with ≥1 IADLs: OR 1.06, 95% CI 0.94–1.20 Difficulty with ≥1 ADL: OR 1.08, 95% CI 0.96–1.21 Dependency with 1–2 ADLs: OR 1.26, 95% CI 1.11–1.44 Dependency in ≥3 ADLs: OR 1.42, 95% CI 1.20–1.69
Dodson et al. 25	Older adults (>75 years) recovering from an AMI	TUG time, categorized as <15 s, 15–25 s, >25 s, or unable to safely complete	Worse performance on the TUG test was associated with a higher risk of readmissions Odds for readmission as compared with a TUG of <15 s TUG 15–25 s: OR 1.46, 95% CI 0.98–2.17 TUG > 25 s: OR 1.86, 95% CI 1.32–2.61 Unable to complete TUG: OR 1.49, 95% CI 1.01–2.19
Kapoor et al. 29	Older adults undergoing surgery	LLFDI, standardized score categorized as <50, 50–60, or >60 points	Unadjusted, inverse monontonic relationship observed between LLFDI scores and readmission risk at 30 days (LLFDI < 50: 20.4%; LLFDI 50–60: 12.6%; LLFDI > 60: 10.7%). Including the LLFDI score variable in the same model as the American College of Surgeons Universal Risk Score improved the prediction of adverse events including readmissions at 30 days (C‐statistic 0.65–0.70).
Keeney et al. 27	Older adults with heart failure admitted to inpatient settings	SPPB score, range 0–12	Among all subjects, 9.6% of persons with heart failure experienced a 30‐day hospital readmission. Persons not readmitted had a significantly higher mean SPPB score (score of 4.44) compared with those who were readmitted (score of 3.02).
Fathi et al. 28	Admitted with the diagnosis of CHF or COPD	Life‐Space Assessment, measuring community mobility based on the frequency of travel to various locations, and the need for assistance to reach those levels in the prior 4 weeks, categorized as restricted (<60/120), or not (≥60/120)	Baseline restricted life space was associated with a greater risk of hospital readmission within 90 days, OR 1.7, 95% CI 1.00–2.87.
Fischer et al. 20	Admission to the medicine unit of the acute care hospital	5x sit to stand: time taken to complete 5 chair rises Gait speed: meters/second over 10 m Reach test: Distance reached outside base of support Fall history: Any falls within the past 12 months, self‐reported	17.8% of participants were readmitted within 30 days. Fall history was associated with 30‐day readmission risk in unadjusted (OR 4.6, 95% CI 1.6–13.6) and adjusted analysis (OR 14.831, 95% CI 1.587–138.57) among female participants in sex‐stratified analysis. No significant relationships were observed for other measures.
Joseph et al. 18	Older adults with ground‐level fall injuries	UEF measured by a wearable sensor technology measuring speed and variability of movement (measured continuously from 0 to 1 in 0.01 increments)	Cumulative impairment in the UEF measure was associated monotonically with increasing readmission risk. Every 0.01 increase in UEF was associated with 1.05 times higher odds (95% CI 1.01–1.09) of 30‐day readmission.a Every 0.01 increase in UEF was associated with 1.04 times higher odds (95% CI 1.01–1.08) of 60‐day readmission.a
Kronzer et al. 31	Subset of patients in the systematic assessment and targeted improvement of services following yearlong surgical outcomes surveys study	Patient survey of reported falls in the 6 months before surgery	Preoperative falls were not associated with 30‐day readmission in unadjusted (OR 1.28, 99% CI 0.97–1.70) or adjusted (OR for 1 fall vs. 0 falls: 1.2, 95% CI 0.8, 1.6; 2 falls vs. 0: OR 1.2, 95% CI 0.7–2.0; and 3+ falls vs. 0 falls: OR = 1.1, 95% CI 0.6–1.9.
Sinvani et al. 32	Cohort study of records for adults older than 75 during hospital admission	Mobility in the hospital as documented using a validated tool from nurse observations as recorded in the medical record Mobility was categorized as low (bedrest or chair), intermediate (ambulation in a room with physical assistance), and high mobility (in room mobility or greater with supervision or independently)	Patients in the low mobility group had higher rates of 30‐day hospital readmission (24.40%) as compared with 16.67% for intermediate and 10.93% among the high mobility group. Relationships remained significant in adjusted models (studies reported no point estimates, only p values).
Tapper et al. 30	Diagnosis of cirrhosis	ADLs (self‐reported ability to feed, toilet, dress, bathe, and transfer; scores from 5 to 15)	None of the frailty measures (including ADLs) were associated with 30‐day readmission.
Yanquez et al. 16	Undergoing vascular surgery, acute or chronic vascular disease requiring elective or urgent operation	Levels of frailty determined by UEF score 0–100, categorized as nonfrail (0–30), prefrail (30–60), and frail (60–100)	UEF status was not significantly associated with readmission, with 50% of nonfrail, 27% of prefrail, and 57% of frail patients experiencing readmissions (p = .32). UEF status was associated with repeat surgical admission within 90 days, with 75% of nonfrail, 23% of prefrail, and 86% of frail patients requiring reintervention (p < .001).
Sanchez et al. 21	Patients undergoing transcatheter aortic valve replacement discharged from acute care	Presurgical 5‐m gait speed, dichotomized as a time >6 vs. ≤6 s	Gait speed was one of 10 identified factors associated with 30‐day readmission risk in prediction model development. Unadjusted 5 m gait speed time >6 s, as compared with ≤6 s, associated with 1.47 (95% CI 1.17–1.84) times higher odds of readmission. Covariate‐adjusted models estimate 5 m gait speed time >6 s, as compared with ≤6 s, associated with 1.26 (95% CI 0.98–1.62) times higher odds of readmission.
Schiltz et al. 24	Health and retirement study respondents with fee for service Medicare claims	Presence of any IADL limitations (yes/no), or limitations in one or more of the following activities: managing money, shopping for groceries or other necessities, preparing meals, using the telephone or other communication, or managing medications	Having any IADL limitation, as compared with no IADL limitations, was associated with 30‐day hospital readmission risk, OR = 1.17 (95% CI 1.06–1.29).
Witt et al. 17	Patients hospitalized for acute exacerbation of COPD	Handgrip strength, assessed using a handheld dynamometer using the average of 3 isometric grip attempts. Weak grip strength was defined as below published cut‐off points for the lowest 20th percentile for sex and BMI	Weak grip strength, as compared with those without weak grip strength, was associated with higher 30‐day readmissions OR 11.2, 95% CI 1.3– 93.2.

Abbreviations: ACS, acute coronary syndrome; AD, assistive device; ADL, activity of daily living; aHR, adjusted hazard ratio; AMI, acute myocardial infarction; ANOVA, analysis of variance; aRR, adjusted risk ratio; CABG, coronary artery bypass graft; CHF, congestive heart failure; CI, confidence interval; COPD, chronic obstructive pulmonary disease; C statistic, concordance statistic; ER, emergency room; IADL, instrumental activities of daily living; ICU, intensive care unit; LLFDI, late life function and disability instrument; MCBS, Medicare Current Beneficiary Survey; NR, not reported; OR, odds ratio; RR, rate ratio; SILVER‐AMI, comprehensive evaluation of risk factors in older patients with AMI; SPPB, short physical performance battery; TAVR, transcatheter aortic valve replacement; TUG, timed up and go; UEF, upper extremity function; 5TSTS, five time sit to stand.

^a Effects per increment adapted from logistic regression model estimating the impact of a change from 0 (no frailty) to 1 (extreme frailty) on the continuous scale.

Measurements related to body structure and function included handgrip strength and upper extremity function, including speed, flexibility, and moment at the elbow as measured by using wearable sensor technology. ¹⁶ , ¹⁷ , ¹⁸ Measurements of activity limitations included gait speed, ADL such as the self‐reported capacity dress or bathe, timed up and go (TUG), 5 times sit‐to‐stand (5xSTS), and short physical performance battery (SPPB). ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Study measures pertaining to participation restriction included composite measures spanning multiple domains such as IADL (e.g., managing money or meal preparation), Life‐Space Assessment, and the Late Life Function and Disability Instrument. ²² , ²³ , ²⁴ , ²⁸ , ²⁹

Associations of body function and structure measures with hospital readmissions

Measures of muscle strength, movement quality, and balance were evaluated in three studies as a potential risk factor for readmissions. ¹⁶ , ¹⁷ , ¹⁸ Simple measures of handgrip strength, categorized as above or below to 20th percentile by established sex and body‐mass index normalized values, were associated with readmission risk among a population of adults (median age 63.5 years) admitted with acute exacerbations of COPD. ¹⁷ After accounting for clinical covariates, those with weak grip strength had more than 10‐fold odds of 30‐day readmission as compared with those without weak grip strength (Table 2). ¹⁷ Other less common, instrumented composite measures of upper extremity movement speed and quality, used in two studies of adults with fall‐related injuries and undergoing vascular surgery respectively, were weakly associated with readmission rates at 30, 60, and 90 days. ¹⁶ , ¹⁸ Poor scores (6/8 or lower) on a modified SPPB test (chair rises and static balance) were associated with a more than threefold higher readmission rate as compared with those with scores >6/8. ²⁶

Associations of activity measures with hospital readmissions

Measures integrating walking performance, either as gait speed or TUG performance, were commonly evaluated in five studies as a potential risk factor for readmissions. ¹⁹ , ²⁰ , ²¹ , ²⁵ , ²⁷ Three used gait speed over 5 m (two studies) or 10 m (one study). ¹⁹ , ²⁰ , ²¹ Among populations undergoing cardiac surgeries, two separate studies found associations between gait speed <0.83 m/s and elevated risk for 30‐day readmission—notably, gait speed was included as 1 of the 10 strongest predictors of readmission risk among patients undergoing transcatheter aortic valve replacement. ¹⁹ , ²¹ One other low‐quality study did not find associations between 10‐m gait speed and 30‐day readmission risk. ²⁰ Another cohort study using gait speed as part of an overall SPPB score found that among older adults admitted with chronic heart failure, SPPB scores were significantly lower among those readmitted but offered minimal improvement in predicting readmissions when included in models along with other medical complexity measures. ²⁷ Gait performance in one additional study of older myocardial infarction survivors (>75 years old) was assessed with the TUG score, a simple test of mobility performance over 3 m. ²⁵ A TUG test score of >25 s was the single strongest predictor in this study, with 86% higher odds of readmission as compared with those completing in less than 15 s. ²⁵

Other activity domains included ADL performance and physical limitations—commonly, involving activities that are commonly associated with independence among older adults. Among older Medicare beneficiaries, one study found a progressive increase in the risk of 30‐day readmission as the degree of functional impairment increased—with notably increasing risk when older adults had dependency in 1–2 ADLs (26% increase in odds for readmission) or 3+ ADLs (42% increase in odds for readmission) as compared with those with no impairments. ²³ ADL impairments were also associated with a higher count of 30‐day hospital readmissions among Medicare beneficiaries with a cancer diagnosis. ²² However, one study among persons with liver cirrhosis found that ADL impairment was not predictive of hospital readmissions at 30 days. ³⁰

Other activity domain measures included self‐reported measures of prehospitalization falls, and maximal mobility performance during admission as extracted from nursing documentation. ²⁰ , ³¹ , ³² Preadmissions falls were inconsistently associated with hospital readmissions at 30 and 60 days, with one study finding fall history was associated with upward of a 14‐fold increase in odds for readmission, and a second finding no associations between falls and readmissions among those with 1, 2, or 3+ falls (compared with none). ²⁰ , ³¹ Among older adults >75 years old admitted to medicine units, those with low mobility (limited to bed or a chair in the room) have more than twofold higher rates of 30‐day readmission (24.4% vs. 10.9%) as compared to those with high mobility (walked in hospital independently or with supervision only). ³²

Associations of participation measures with hospital readmissions

Measures of IADL impairment, or the ability to participate in key life roles, were strongly associated with readmissions in two studies. ²² , ²⁴ One study estimated 17% higher odds of readmission in one large cohort study for those with any IADL limitations as compared with those with no limitations, and another study observed a significantly higher count of readmissions for those with 2+ IADL impairments as compared with those with 1 or fewer. ²² , ²³ Other studies relied on standardized measures of participation level impairments, the Late Life Function and Disability Instrument. ²⁹ This instrument was used to measure the association of participation in life roles using a standardized score comparing performance to normative values. ²⁹ The LLFDI scores had an inverse monotonic relationship with readmission rates, and when added to commonly used surgical risk models, the prediction accuracy improved significantly. Participation was also assessed with the Life‐Space Assessment (score: 0–120) in one study, categorizing older adults with CHF or COPD as having restricted (<60/120) or unrestricted life space (60 or higher). ²⁸ Restricted life space was associated with 70% higher odds for readmission at 90 days. ²⁸

DISCUSSION

The results of this review of 17 studies with over 80,000 hospitalized participants suggest strong associations between functional impairments, activity limitations, or participation restrictions and hospital readmissions (up to 90 days in some studies) among older adults across a wide array of health conditions. Our findings highlight the continued importance of assessing function among older inpatients during their admission. While myriad assessments of function were used across studies, relationships were largely observed between worse function and higher readmission rates across functional domains. This suggests that clinicians can align measures of function with clinical needs and patient priorities, and still obtain valuable prognostic information to guide care transition risk stratification. Overall, inclusion of functional assessments may improve identification of vulnerable subpopulations and also provide a roadmap for targeting interventions. ³³

Risk stratification of hospitalized older adults preparing for discharge is complex, and conventional medical markers like lab values and chronic condition counts may not fully reflect the risk of readmissions. ³ A key finding from this systematic review is that functional impairment across several measurement domains was consistently associated with hospital readmission risk. While many studies used performance assessments either during or clinic‐based measures before hospital admission, other studies that used even more pragmatic and potentially more ecologically valid measures, such as patient self‐reports, automatic extractions of function from electronic medical records, or from wearable sensor data also found strong association with readmission risk. ¹⁸ , ²⁰ , ²⁵ , ³⁰ , ³¹ , ³² This suggests that recording functional status comprehensively during hospital admission has a high value and should be prioritized in future quality improvement efforts. These measures also are valuable, because they directly help identify specific targets for intervention. As an example, impaired chair rise performance can indicate a need for lower extremity strength training, equipment, and/or environmental modifications during care transitions. Similarly, identification of basic ADLs impairment could spur additional recommendations or prescriptions for durable medical equipment before discharge that could be addressed by staff at the next level of care. ³ Taking an earlier and proactive approach to functional assessment in the hospital may help ensure care is aligned with key patient and care partner priorities for independence.

Our findings also have a significant public health impact relevant to readmission reductions and, more broadly, to patient‐centered outcomes following acute hospital discharge. Functional deficits, beyond their immediate impact on mobility and safety, can hint at declines in broader resilience and physiological reserve (e.g., frailty) in older adults. ³⁴ Additionally, disability that persists after hospital discharge often leads to downstream complications, including social isolation, decreased quality of life, delayed return to work, and increased risk for nursing home admission. ³⁵ , ³⁶ From a population health standpoint, hospitalization offers a window of opportunity to identify and intervene on functional impairments that may not otherwise have been identified—prioritizing functional assessments in workflow for all clinicians and developing informatics infrastructure to make this information widely available to the treatment team and clinicians at the receiving end of care transitions could pay dividends for individual patients, and the older adults population as a whole.

For clinical staff and hospital administrators, systematically measuring function appears essential for creating clinical operations that can better risk stratify patients and target interventions such as in‐hospital rehabilitation therapy. ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ Current work illustrates that interdisciplinary measures of function and mobility performance are feasible and can help identify patients most likely to require in‐hospital rehabilitation resources, monitor a patient's care progression, and inform discharge planning. ⁶ , ³³ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ Notably, an American Geriatrics Society White Paper advocating for routine mobility assessments of hospitalized older adults recommends tools including the Activity Measure for Post‐Acute Care 6‐Clicks and Johns Hopkins Highest Level of Mobility both of which can be completed by nursing or rehabilitation staff and used to guide treatment decision making on a daily basis. ⁴⁴

Clinical decision tools utilizing functional measures can likely improve risk identification and targeting of effective in‐hospital interventions and also support reducing unmet needs during care transitions. Examining the effect of intervention dosage (type, timing, frequency, duration, intensity) on physical function and readmissions could support readmission efforts, discharge planning, and successful care transitions.

Our study had some modest limitations, and several key strengths. We acknowledge that function is only one factor in a complex system of what leads to patient readmissions, and the independent contribution of function is often difficult to untangle from other comorbid conditions. Additionally, our studies largely had modest methodological quality, which suggests more high‐quality and well‐designed cohort data is needed to further refine our understanding of these relationships. Our data also had significant heterogeneity in terms of measures used, reporting of results, and underlying patient populations precluding opportunities to pool estimates in a meta‐analysis. Major strengths include a focus on older populations who are uniquely vulnerable, and a comprehensive focus on function that offers a uniquely rich estimation of how function across ICF domains may be best measured as a risk factor for readmissions.

CONCLUSION

The ability of an older adult to recover from an acute illness and withstand physiological stressors is often mirrored in their functional capabilities. Routine measurement of function may help identify patients at high risk for readmission before hospital discharge as well as the need for specific interventions to support successful care transitions and optimize postdischarge functional status.

CONFLICT OF INTEREST STATEMENT

Drs. Jason R. Falvey and Kyle Ridgeway receive royalties from MedBridge Inc., for continuing education courses for physical therapists about hospital readmissions while Dr. James Smith also receives royalties from MedBridge Inc., for other continuing education courses.

Supporting information

Supporting information.

Supporting information.

Supporting information.

Thomas EM, Smith J, Curry A, et al. Association of physical function with hospital readmissions among older ads: A systematic review. J Hosp Med. 2025;20:277‐287. 10.1002/jhm.13538