Effect of Variation in Early Rehabilitation on Hospital Readmission After Hip Fracture

Abstract

Objective

Provision of early rehabilitation services during acute hospitalization after a hip fracture is vital for improving patient outcomes. The purpose of this study was to examine the association between the amount of rehabilitation services received during the acute care stay and hospital readmission in older patients after a hip fracture.

Methods

Medicare claims data (2016–2017) for older adults admitted to acute hospitals for a hip fracture (n = 131,127) were used. Hospital-based rehabilitation (physical therapy, occupational therapy, or both) was categorized into tertiles by minutes per day as low (median = 17.5), middle (median = 30.0), and high (median = 48.8). The study outcome was risk-adjusted 7-day and 30-day all-cause hospital readmission.

Results

The median hospital stay was 5 days (interquartile range [IQR] = 4–6 days). The median rehabilitation minutes per day was 30 (IQR = 21–42.5 minutes), with 17 (IQR = 12.6–20.6 minutes) in the low tertile, 30 (IQR = 12.6–20.6 minutes) in the middle tertile, and 48.8 (IQR = 42.8–60.0 minutes) in the high tertile. Compared with high therapy minutes groups, those in the low and middle tertiles had higher odds of a 30-day readmission (low tertile: odds ratio [OR] = 1.11, 95% CI = 1.06–1.17; middle tertile: OR = 1.07, 95% CI = 1.02–1.12). In addition, patients who received low rehabilitation volume had higher odds of a 7-day readmission (OR = 1.20; 95% CI = 1.10–1.30) compared with high volume.

Conclusion

Elderly patients with hip fractures who received less rehabilitation were at higher risk of readmission within 7 and 30 days.

Impact

These findings confirm the need to update clinical guidelines in the provision of early rehabilitation services to improve patient outcomes during acute hospital stays for individuals with hip fracture.

Lay summary

There is significant individual- and hospital-level variation in the amount of hospital-based rehabilitation delivered to older adults during hip fracture hospitalization. Higher intensity of hospital-based rehabilitation care was associated with a lower risk of hospital readmission within 7 and 30 days.

Introduction

Hip fractures have a devastating impact on the lives of older people, including high rates of 1-year mortality (36%–47%),^1,2 new disability (27%–33%),^3,4 and long-term institutionalization (25%).⁵ With over 323,000 hip fractures occurring annually in the United States and aggregate costs of approximately $5.6 billion,⁶ it is critical to identify modifiable risk factors to avoid negative outcomes, such as hospital readmission, that further increase the risk of disability.⁷

Early and intense physical rehabilitation is critical to the recovery of function and reduction of morbidity after a hip fracture. It is recommended that physical therapist interventions start the day after surgery with physical therapists facilitating the patient’s bedside mobility, including transfers and gait training, at least once per day.^8,9 Clinical practice guidelines from several organizations including the American Academy of Orthopedic Surgeons issued guidelines supporting supervised physical therapy and occupational therapy across the continuum of care, including intensive physical therapy after hospital discharge.^9,10 Most patients receive rehabilitation while in the acute care hospital,¹¹ and early rehabilitation is promoted.¹² However, there is no consensus on the optimal doses of rehabilitation during acute hospitalization.¹³

Increased rehabilitation is associated with better patient-centered outcomes, including reducing risk of hospital readmissions in patients with stroke, heart failure, hip fracture, pneumonia, and acute myocardial infarction.^14–18 However, these results categorize rehabilitation by total hospital charges instead of physical therapy and occupational therapy minutes. Therefore, they lack precision in quantifying the rehabilitation services provided during acute hospitalization and the ability to determine if rehabilitation dosage impacts outcomes.^18,20

The additional value-based payment models being introduced by the Centers for Medicare and Medicaid Services all emphasize reducing hospital readmissions. By reexamining the roles of various clinical service lines during acute hospitalization, rehabilitation services could be considered a modifiable factor to improve patient outcomes and reduce risk of hospital readmission.^19,20 However, additional evidence proving the role of rehabilitation services during acute hospitalization is needed. Prior studies examining factors associated with readmissions after a hip fracture have only included post-acute rehabilitation^21–23 and not hospital-based rehabilitation. Data from the Healthcare Cost and Utilization Project found that in patients with hip fractures and joint replacement, 37.1% to 38.9% of the readmissions occurred within the first 15 days after discharge.^24,25 This high proportion of readmission could be attributable to the quality and quantity of hospital-based care, including rehabilitation services. However, to what extent these services are associated with hospital readmissions is unclear.

This study fills an important gap in knowledge regarding older adults after a hip fracture by using nationally representative Medicare claims data for 2016 to 2017 to evaluate (1) variation in rehabilitation services delivered during acute hospitalization, and (2) the association between rehabilitation dose with 7-day and 30-day hospital readmissions. We hypothesize that patients receiving a lower amount of rehabilitation during acute hospitalization will have a higher risk of hospital readmission.

Methods

Data Source

The data for this study came from the 2016 and 2017 Medicare inpatient claims using standard analytic file (SAF) format. The SAF comprise claims for all inpatient stays, including acute hospital stays, and have information on diagnoses, surgical procedures, and admission and discharge dates. There are also separate revenue center codes and dates for evaluations for all rehabilitation services as well as for the number of rehabilitation therapy units delivered. We linked SAF to the Master Beneficiary Summary File, which contains Medicare enrollment indicators and beneficiary demographic characteristics. Hospital-level factors (eg, profit status, teaching status) were retrieved from the provider of services file. The Chronic Condition Data Warehouse was used to determine the number of chronic conditions for the study cohort. The study was approved by the Northern Arizona University institutional review board, and data use agreement was approved by the Centers for Medicare and Medicaid Services.

Study Population

The study sample consisted of Medicare beneficiaries aged 66 years and older with fee-for-service plans who were admitted to acute or critical access hospitals from July 1, 2016, to September 30, 2017, with a primary diagnosis of hip fracture using the Medicare Severity-Diagnosis Related Group 480, 481, 482, 533, 534, or ICD-10 diagnosis codes S72.0, S72.1, and S72.2. The data extended back to January 1, 2016, which allowed us to use 6 month of data to search for previous hospitalizations, long-term nursing home stays, and Medicare fee-for-service enrollment. Figure 1 presents the number and percentage of excluded patients in the deriving study cohort. The sample included beneficiaries who (1) had 6 months of continuous Medicare fee-for-service enrollment prior to index hospitalization to capture a full account of health services used, and (2) were living in community settings prior to the index hospitalization. Patients who were excluded (1) were enrolled in Medicare based on disability, end-stage renal disease, or living in a US territory (n = 68,066); (2) died during the index hospitalization or within 30 days after discharge (n = 2310); (3) had recurrent hip fracture (n = 9916); (4) were discharged against medical advice, hospice, or psychiatric facilities (n = 124,279); (5) had an index admission of multiple traumas, including serious head trauma/brain damage (ie, subdural hematoma, subarachnoid hemorrhage, stroke), cervical trauma (concurrent vertebral fractures), or abdominal injury (n = 4520); and (6) had a long-term nursing home stay 90 days prior to hospitalization (n = 7434).

Figure 1.

Derivation of study cohort.

Primary Exposure Variable

Hospital-based rehabilitation service was categorized using the revenue center codes for evaluation and number of units for physical therapy (physical therapist evaluation, consultation, visit, hourly, and group = 0424, 0420, 0421, 0422, 0423, 0429) and occupational therapy (occupational therapist evaluation, consultation, visit, hourly, and group  = 0434, 0430, 0431, 0432, 0433, 0439). We assigned 15 minutes for each therapy unit and 30 minutes for each evaluation unit at the time of admission.^17,26 We used these cutoffs based on feedback from clinicians and a previously validated approach.^17,26 To estimate total rehabilitation services provided, physical therapy and occupational therapy minutes were summed for the entire hospital stay and for the amount of rehabilitation per day; then the total amount was divided by the hospital length of stay. For analysis, we categorized the amount of rehabilitation per day into tertiles: low, medium, and high based on previous literature.^17,26

Outcome Variables

The outcome variable was all-cause readmission within 7 days or 30 days after the discharge date from the index hospitalization, which was risk adjusted. Thirty-day readmission includes individuals who were also readmitted within 7 days. This measure was dichotomized into yes or no for both time frames.

Covariates

Patient-level variables were gathered from the Master Beneficiary Summary File (eg, age, sex, and race/ethnicity). Race/ethnicity was categorized into non-Hispanic White, Black, Hispanic, and other. Dual eligibility in Medicare and Medicaid (yes vs no) was used as a proxy for socioeconomic status.²⁷ Case-mix variables included comorbidity index, frailty, length of stay, prior hospitalization, intensive care unit (ICU) stay, hospital-related procedures and complications from the inpatient SAF, and chronic conditions from the Chronic Condition Data Warehouse files. A critical barrier in health services research using claims data is the lack of physical functional measures and frailty. Missing information on frailty can have a residual confounding effect in studies examining the efficacy of treatment on health outcomes.²⁸ We overcame this limitation by using a validated Functional Comorbidity Index and claim-based frailty risk score to reflect some level of severity. We accounted for Functional Comorbidity Index, which was developed to predict physical function in the community-dwelling population²⁹ and has been validated in Medicare data.³⁰ The Functional Comorbidity Index includes 18 comorbid conditions.³¹ Because the Functional Comorbidity Index is not strongly associated with functional status among Medicare beneficiaries, we further computed the claim-based “frailty risk score” using the algorithm published by Gilbert and colleagues,³² which categorizes frailty in an individual in 1 of the 3 levels: high, moderate, and low. The length of hospital stay was calculated as the number of days from hospital admission to discharge and was used as a continuous variable. Prior hospitalization in the last 6 months, an ICU stay, and hospital-related complications (deep vein thrombosis, pulmonary embolism, and urinary tract infection) were dichotomized to yes or no. Finally, hospital procedures concerning the hip fracture management were categorized to open or internal fixation, closed reduction, and partial or total hip replacement. Discharge disposition was categorized into skilled nursing facility (SNF), inpatient rehabilitation facility (IRF), home health, and home/self-care based on the variable in SAF.

Hospital-level Variables

Hospital-level variables were obtained from the Provider of Service file, which includes Medicare-certified hospitals. These variables included location (urban/rural), profit status (for profit/not for profit/other), teaching status (teaching/nonteaching/unknown), volume of hip fractures per facility, safety-net status (yes/no), and hospital beds (quartiles). Prior studies reported that hospital characteristics, including volume,^33,34 safety-net status,³⁵ and teaching status,³⁶ were hospital processes that are associated with outcomes after a hip fracture.

Analysis

Descriptive statistics by the amount of rehabilitation (low, medium, or high) were calculated with counts (percentages) for categorical variables and mean (SD) or median (IQR) for continuous variables. To assess the association of hospital rehabilitation with the likelihood of 7-day and 30-day readmission, we used hierarchical generalized linear mixed effect models to account for the clustering of patients within hospitals. The outcomes were readmissions within 7 days and 30 days, with tertile of hospital rehabilitation as the exposure. The models were adjusted for age, sex, race, dual Medicare and Medicaid eligibility, comorbidities, frailty, prior hospitalization in the 6 months before index hospitalization, length of hospital stay, ICU stay, hospital-related complications, discharge disposition (for 30-day readmission), and hospital characteristics (urban/rural, profit status, safety net status, hospital volume, teaching status). All statistical analyses were performed using SAS 9.4 (SAS Inc., Cary, NC, USA), and α was set at .05.

Sensitivity Analysis

We conducted an additional sensitivity analysis to examine how quality of post-acute care settings affect risk of readmission. For this analysis we focused only on individuals discharged to SNFs after acute hospitalization because quality reporting for the time frame we used (2016 to 2017) was available only for SNFs as a star rating system. We linked data from nursing home to compare with our analytical cohort and categorized the SNFs into low-rated (1–2), average-rated (3), and high-rated (4–5) facilities and included those in the outcomes model, controlling for all the previously mentioned covariates.

Role of the Funding Source

Funders and sponsors had no role in the study design, analysis, or interpretation of the data. Sponsors had no role in writing the manuscript or in submitting the manuscript to a journal.

Results

Our study included 131,127 Medicare beneficiaries who were admitted to the hospital for hip fracture. Over 99% of these patients received rehabilitation while in the hospital. Only 979 patients (<1%) did not receive any rehabilitation during their hospital stay. Patients were on average 82.9 (SD = 7.9) years old, with a higher proportion being women (75.3%) and non-Hispanic White (91.3%) patients. A substantial number of patients (50.4%) had a moderate frailty score (5–15), and 14% had a high frailty score (>15). In our study cohort, the average number of functional-related comorbidities was 2.2 (1.6) years. A large proportion (54.0%) had surgical management through open reduction and internal fixation. On average, patients stayed in the hospital 5.0 days ([IQR] = 4.0–7.0 days), with most (85.5%) discharged to institutional-based post–acute care facilities (SNF and IRF).

Figure 2 illustrates the variation in rehabilitation per day across all study patients and average rehabilitation per day across the 2043 hospitals. After adjusting for patient- and hospital-level characteristics, the amount of rehabilitation per day that the patients received varied substantially. Across all patients, the median rehabilitation in minutes per day was 30.0 (IQR = 21.0–42.5), but the range spanned from 8 to more than 98.6. Rehabilitation minutes per day were categorized into tertiles: low was defined as a median of 17.5 (IQR = 12.7–20.6 minutes), medium was defined as a median of 30.0 (IQR = 27.0–33.8 minutes), and high was defined as a median of 48.8 (IQR = 42.9–60.0 minutes). We found similar and wide variation across the hospital level: 25% provided >23 min/d of rehabilitation services.

Figure 2.

Variation in hospital-based rehabilitation minutes per day.

As shown in Table 1, a larger proportion of non-Hispanic White patients received a high amount of rehabilitation (33%) compared with Black patients (26.6%) and Hispanic patients (28.5%). For dual-eligible patients, 42% received low amounts of rehabilitation compared with 30% for non-duals. The presence of comorbid conditions, frailty, other hospital-acquired conditions, or ICU stays also resulted in lower amounts of rehabilitation care. For patients in safety-net hospitals, 39.6% received a low amount of therapy compared with patients in non-safety net (31%) or critical access hospitals (22%). More patients in large hospitals (>435 beds) received a low amount of rehabilitation at the hospital (39%) compared with patients who were treated at smaller facilities (28%).

Table 1.

Patient and Hospital Characteristics by Amount of Rehabilitation Received at the Hospital for Patients Admitted for Hip Fracture^a

Characteristic	Overall	Amount of Therapy (min/d)			P
		Low Median (IQR) 17.5 (12.7–20.6)	Medium Median (IQR) 30.0 (27.0–33.8)	High Median (IQR) 48.8 (42.9–60.0)
	N = 131,127	N = 42,475	N = 45,376	N = 43,276
Rehabilitation, median (IQR)	30.0 (21.0–42.5)	17.5 (12.6–20.6)	30.0 (27.0–33.7)	48.8 (42.8–60.0)	<.0001
Patient characteristics
Age, mean (SD), y	82.9 (7.9)	83.7 (7.8)	82.9 (7.9)	82.1 (7.9)	<.0001
Length of hospital stay, median (IQR), days	5.0 (4–7)	6.0 (5–7)	5.0 (4–7)	5.0 (4–6)	<.0001
Sex (%)
Male	24.7	31.0	35.3	33.7	<.0001
Female	75.3	32.8	34.4	32.8
Race (%)
Non-Hispanic White	91.3	31.9	34.7	33.4	<.0001
Black	2.8	39.2	34.2	26.6
Hispanic	3.1	38.6	32.9	28.5
Other	2.8	33.9	33.2	32.9
Dual eligibility in Medicare and Medicaid, %
Yes	18.5	41.9	32.8	25.3	<.0001
No	81.5	30.2	35.0	34.8
Comorbid conditions
Functional Comorbidity Index, mean (SD)	2.2 (1.6)	2.3 (1.6)	2.2 (1.6)	2.0 (1.5)	<.0001
Frailty score group, %
Low (≤ 5)	35.3	28.8	34.2	37.0	<.0001
Moderate (>5 to ≤15)	50.4	32.9	35.0	32.1
Severe (>15)	14.3	39.5	34.2	26.3
Hospital-acquired complications
Urinary tract infection, %
Yes	14.3	37.4	34.2	28.4	<.0001
No	85.7	31.6	34.7	33.7
Deep vein thrombosis, %
Yes	0.4	45.2	30.4	24.4	<.0001
No	99.6	32.3	34.6	33.1
Pulmonary embolism, %
Yes	0.5	41.3	33.4	25.3	<.0001
No	99.5	32.4	34.6	33.0
ICU admission, %
Yes	17.8	36.0	35.2	28.8	<.0001
No	82.2	31.6	34.5	33.9
Fracture management, %
Open or internal	54.0	33.9	35.1	31.0	<.0001
Closed reduction	7.0	32.8	35.7	31.5
Partial or total hip replacement	30.7	27.9	34.5	37.6
Nonsurgical/other	8.3	39.1	31.2	29.7
Discharge disposition, %
Home/self-care	7.3	29.1	33.0	37.9	<.0001
Home/home health	7.2	22.7	31.9	45.4
Skilled nursing facility	66.9	35.5	34.6	29.9
Inpatient rehabilitation facility	18.6	26.3	36.4	37.3
Discharged to community, %
Yes	14.5	25.9	32.5	41.6	<.0001
No	85.5	33.5	35.0	31.5
Prior hospitalization within 6 mo of index hospitalization, %
Yes	16.5	35.9	34.4	29.7	<.0001
No	83.5	31.7	34.6	33.7
Readmitted within 7 d of discharge from hospital, %
Yes	3.1	37.0	33.6	29.4	<.0001
No	96.9	32.2	34.6	33.2
Readmitted within 30 d of discharge from hospital, %
Yes	9.5	35.0	35.0	30.0	<.0001
No	90.5	32.1	34.6	33.3
Hospital characteristics
Location, %
Urban	88.8	32.5	34.7	32.8	<.0001
Rural	11.2	31.6	33.8	34.6
Profit status, %
Nonprofit	65.0	34.2	34.4	31.4	<.0001
For profit	11.0	29.2	38.0	32.8
Other/unknown	24.0	29.1	33.6	37.3

(Continued)

This cohort’s average 7-day and 30-day readmission rates were 3.1% and 9.5%, respectively. These rates decreased with a high amount of rehabilitation. Readmitted patients were older men or were dually enrolled in Medicare and Medicaid (Suppl. Tab. 1). A higher proportion of those readmitted within 30 days had hospital-related complications, an ICU stay, or a surgical intervention during the index hospitalization. Those discharged to institutions (SNF or IRF) also had a higher proportion of being readmitted within 30 days. There was a monotonic relationship where increasing the amount of rehabilitation was associated with a lower proportion of patients being readmitted within 7 and 30 days. After adjusting for patient and hospital characteristics, those who received a medium and low amount of rehabilitation had 11% and 7% increased odds of 30-day hospital readmission, respectively, compared with those who received a high amount of rehabilitation (Tab. 2). Similarly, those who received low rehabilitation had 20% increased odds of 7-day hospital readmission compared with those who received high rehabilitation. In the supplemental analysis (Suppl. Tab. 2) evaluating those patients discharged to SNF, we found associations between lower-quality ratings of SNF and higher risk of readmission. Compared with patients discharged to a high-quality SNF, the odds of 30-day hospital readmission were 13% higher in patients discharged to a lower-quality SNF after acute hospitalization (OR = 1.13; 95% CI = 1.07– 1.19).

Table 2.

Association of Minutes of Rehabilitation Per Day With 7-Day and 30-Day Hospital Readmission^a

Outcome	7-Day Readmission		30-Day Readmission
Min of Rehabilitation Per Day Tertiles, Median (IQR)	Unadjusted OR (95% CI) N = 131,127	Adjusted  b  OR (95% CI) N = 130,234	Unadjusted OR (95% CI) N = 131,127	Adjusted  b  OR (95% CI) N = 130,234
High: 48.8 (42.9–60.0) min/d	Reference group	Reference group	Reference group	Reference group
Medium: 30.0 (27.0–33.8) min/d	1.10 (1.02–1.19)	1.04 (0.96–1.13)	1.14 (1.09–1.19)	1.07 (1.02–1.12)
Low: 17.5 (12.9–20.6) min/d	1.31 (1.21–1.42)	1.20 (1.10–1.30)	1.23 (1.18–1.30)	1.11 (1.06–1.17)

^a IQR = interquartile range; OR = odds ratio. ^bAdjusted for age, sex, race, dual eligibility, comorbid conditions (Functional Comorbidity Index), frailty, prior hospitalization within 6 months of the index hospitalization, length of index hospital stay, intensive care unit stay, hospital-related complications, type of fracture management, post-acute discharge disposition, and hospital characteristics (urban/rural, safety net status, number of hospital beds, volume of hip fracture cases in hospital).

Discussion

This study explores variations in the amount of hospital-based rehabilitation for patients with Medicare after a hip fracture to determine the association with 7-day and 30-day readmissions. We found substantial variability in the use of rehabilitation services across acute care hospitals, even after accounting for patient- and hospital-level characteristics. For example, approximately 25% of hospitals provided <23 min/d of rehabilitation services, which may not be enough to optimally improve outcomes. Furthermore, our results demonstrate that high rehabilitation was significantly associated with lower odds of hospital readmission compared with the low amount of rehabilitation, complementing prior work that suggests that physical therapists and occupational therapists play an essential role in improving safety during care transitions.²⁰ Improving rehabilitation care and preventing hospital readmissions are important for patients, payers, and health care providers. These findings have important implications for clinical practice, suggesting a need to prioritize the receipt of more rehabilitation for older hip fracture survivors in the acute care setting to improve patient-centered outcomes.

Table 1.

Continued

Characteristic	Overall	Amount of Therapy (min/d)			P
		Low Median (IQR) 17.5 (12.7–20.6)	Medium Median (IQR) 30.0 (27.0–33.8)	High Median (IQR) 48.8 (42.9–60.0)
	N = 131,127	N = 42,475	N = 45,376	N = 43,276
Teaching status, %
Nonteaching	43.2	30.2	34.4	35.4	<.0001
Teaching	41.0	36.5	35.8	27.7
Unknown	15.8	27.9	32.0	40.1
Safety net status, %
Yes	17.6	39.6	31.9	28.5	<.0001
No	81.7	30.9	35.2	33.9
Critical access hospitals	0.7	21.9	31.9	46.2
Hospital beds, quartiles, %
<172	25.3	28.1	34.0	37.9	<.0001
172–275	24.7	30.4	34.7	34.8
275–435	25.0	32.4	35.0	32.6
>435	25.0	38.9	34.8	26.3
Volume of hip fractures in hospital, %
Low	15.1	35.4	34.6	30.0	<.0001
Medium	43.4	33.8	34.3	31.9
High	41.5	29.8	35.0	35.2

^a ICU = intensive care unit; IQR = interquartile range.

Unfortunately, there are barriers to implementing early rehabilitation in the acute setting. Patients readmitted within 7 days may have different clinical pathways and rehabilitation needs. During the acute stay, medical stabilization is prioritized over rehabilitation. More importantly, current hospital payment is based on diagnosis-related groups and includes rehabilitation care along with surgical procedures and medical interventions.^37,38 We found a significant level of variation in the provision of rehabilitation services during acute hospitalization after hip fracture at both the patient and hospital levels. Similar findings were previously reported for other medical conditions, and most patients received rehabilitation in the acute hospital,^11,15 but the amount varied from almost none to >4 h/d. Recently, Freburger et al reported variation in acute hospital-based rehabilitation in 30,746 patients with pneumonia, with 26% receiving 1 to 3 therapist (physical therapist and/or occupational therapist) visits, 19% receiving 4 to 6 therapist visits, and 24% receiving more than 6 therapist visits.¹⁵ A similar variation in hospital-based rehabilitation has been reported in patients with stroke,²⁶ but this is the first study, to our knowledge, to look at the variability of rehabilitation in the acute hospital for patients after a hip fracture. These findings can aid in supporting a standardization of the provision of rehabilitation services during an acute stay.

Unlike rehabilitation utilization during acute stays, evidence on rehabilitation variability in post-acute care settings exists.^3,16,39–41 Comprehensive hospital-based evaluations by occupational and physical therapists and the timely provision of an appropriate amount of rehabilitation services maximize functional recovery and minimize length of stay and health care costs.^42,43 In addition, their involvement during interdisciplinary discharge planning also reduces fragmentation of care and enhances a smooth transition to home using home health care without increasing readmission rates.^33,44,45 Johns Hopkins Hospital has implemented an early rehabilitation and multidisciplinary mobility promotion quality-improvement program during acute care stays, which has demonstrated that early mobilization during acute care improves physical functioning and reduces complications, hospital length of stay, and risk of hospital readmission.^46,47 In agreement with other conditions,^18–21 the amount of rehabilitation was inversely associated with the odds of hospital readmission independent of patient and hospital characteristics.^{14,15,17,18,48} Systematic reviews of 26 studies in 2020⁴⁹ and 22 studies in 2017⁵⁰ examined predictive factors for readmissions after hip arthroplasty and fractures and found that postoperative functional status was a strong predictor of 30-day readmissions. This is important because functional status can be ameliorated with rehabilitation interventions, and physical therapy services delivered to hospitalized patients can improve care transitions through the reduction of unmet functional needs.

There are some limitations with the use of Medicare claims, including errors in billing and coding and potential missing data, although high levels of agreement have been reported in administrative data and medical records.^51,52 In addition, Medicare claims do not include measures of functional status, which has been associated with readmissions in that population.⁵³ The lack of information on functional status may lead to unmeasured confounding. A final limitation of these administrative data is that they included only people with Medicare fee-for-service; caution should be taken when generalizing these results to the Medicare Advantage, a growing segment of the overall Medicare and non-Medicare population.⁵⁴ Quality information for other post-acute care settings (eg, IRF) for 2016–2017 was not available. Therefore, although very important, we could not use that information in our main analysis, presenting that as sensitivity analysis for only those patients discharged to SNF. Based on our findings, future studies must account for the quality rating of post-acute care settings and quality measure comprising those ratings. Cumulatively these are significant indicators of quality of services provided during post-acute stay and how that affects patient-centered outcomes, such as hospital readmission. Despite the limitations for the use of administrative data, this study had many strengths. First, the use of Medicare inpatient claims in the SAF includes revenue center codes, signifying billing for physical therapist and/or occupational therapist, allowing quantification of rehabilitation rather than just receipt of care or not. Second, by linking Medicare beneficiary-level files with provider level files, both patient and hospital characteristics can be accounted for, minimizing the risk of bias. Finally, using national representative data, we were able to demonstrate variation in rehabilitation and association with readmissions in older adults with hip fractures.

The evidence from our study is significant to clinicians, hospital administrators, policymakers, and payers, especially in the value-based payment system era. As a result of the prospective payment system, lengths of stay during acute hospitalization were already shortened. As several value-based care initiatives are implemented, health care providers are under pressure to improve the discharge planning process and rapid transition to appropriate and less resource-intense post–acute settings, such as discharge to the community with home health care. This study lays the groundwork for further research using Medicare claims data that can examine the impact of hospital-based rehabilitation care on post–acute care transitions, cost savings, and long-term functional recovery after accounting for patient-level case severity. Future work should expand the timing of initiation and intensity of rehabilitation to fully describe the benefits of early rehabilitation.

Author Contributions

Concept/idea/research design: A. Kumar, A. Karmarkar, I. Roy, P. Sood, J. Falvey, J. Rudolph

Writing: A. Kumar, I. Roy, J. Falvey, J. Rudolph, S. Shaibi, M. Rivera-Hernandez, P. Sood, C. Childers, A. Karmarkar

Data acquisition, analysis, or interpretation: A. Kumar, A. Karmarkar, I. Roy, C. Childers, M. Rivera-Hernandez

Critical revision of the manuscript for important intellectual content: A. Kumar, I. Roy, J. Falvey, J. Rudolph, S. Shaibi, M. Rivera-Hernandez, P. Sood, C. Childers, A. Karmarkar

Statistical analysis: A. Kumar, I. Roy, P. Sood, A. Karmarkar

Administrative, technical, or material support: A. Kumar, A. Karmarkar

Fund procurement: A. Kumar, A. Karmarkar,

Funding

This work was supported in part by the National Institutes of Health (grant nos. R03AG060345 and U54MD012388). J. Falvey was supported during the conduct of this research by a Paul B. Beeson Emerging Leader Career Development Award (K76AG074926).

Ethics Approval

This study was reviewed and approved by the Institutional Review Board of Northern Arizona University, and a Data Use Agreement was established with the Centers for Medicare & Medicaid Services (DUA RSCH-2019-52,868).

Data Availability Statement

Please note that the person-level data for this study (ie, inpatient claims) are covered under the strict terms of a Data Use Agreement (DUA) with the Centers for Medicare and Medicaid Services (CMS). We are prohibited from making any person-level data file, no matter how de-identified, available. However, researchers interested in replicating the results of these analyses may enter into their own DUA with CMS. See the Research Data Assistance Center (ResDAC) at www.resdac.org for assistance.

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest. J. Falvey and A. Karmarkar are editorial board members of PTJ. J. Falvey receives royalties from Medbridge Inc. for courses related to hospital readmissions. No other authors have conflicts of interest or financial disclosures to report, and no authors have commercial interests relevant to the research described.