Impact of osteoarthritis on rehabilitation for persons with hip fracture


  • FIM Instrument is a trademark of the Uniform Data System for Medical Rehabilitation, a division of the UB Foundation Activities, Inc.



To determine the impact of osteoarthritis (OA) on length of rehabilitation stay, Functional Independence Measure (FIM Instrument) ratings at discharge and followup, functional gain, and percentage of patients discharged home.


We conducted a retrospective cohort analysis using a national registry of US medical rehabilitation inpatients. We obtained standardized data for all patients admitted after a hip fracture between 1994 and 2001. Our primary analytical method was multiple regression analysis. Outcome variables were length of stay, FIM Instrument ratings at discharge and followup, functional gain, and percentage of patients discharged home. The predictor variable was the presence of OA. Covariates were age, sex, race/ethnicity, other comorbidity, admission FIM ratings, total hip replacement, and time to followup.


We studied 1,953 patients with OA and 11,441 patients without OA admitted to inpatient rehabilitation facilities after hip fracture. Mean ± SD length of stay for patients with OA was 18.1 ± 10.0 days versus 16.5 ± 8.9 days for those without OA (P < 0.01). After adjusting for age, sex, race/ethnicity, comorbidity, admission FIM ratings, and total hip replacement, OA was associated with a longer rehabilitation stay (1.4 days; P < 0.01) and slightly higher discharge FIM ratings; however, OA was not associated with lower weekly rehabilitation gain, followup FIM ratings, and percentage discharged home.


Persons with hip fracture and OA had longer inpatient rehabilitation length of stay than persons without OA, but there were similarities in weekly rehabilitation gain and percentage discharged home.


Osteoarthritis (OA) is a common disorder among older adults; up to 80% of all individuals over the age of 60 have radiographic findings of OA (1), even though the proportion with symptomatic disease is lower. When symptomatic, OA can cause significant joint pain or dysfunction and interfere with activities of daily living (2, 3).

The impact of OA as a comorbid condition is not well established. OA has been shown to both adversely impact functional recovery among patients who experience stroke and prolong inpatient rehabilitation stay (4). Because delayed functional recovery increases rehabilitation length of stay and associated costs, more investigations are needed to understand the impact of OA as a comorbid condition in recovery from acute illnesses.

In 2002, the Centers for Medicare and Medicaid Services changed reimbursement for rehabilitation hospitalization from fee-for-service to a prospective payment system (PPS). PPS reimburses rehabilitation services based on case-mix groups. Although PPS allows for higher payment based on comorbidities, OA is not one of the comorbidities that result in increased reimbursement. The Centers for Medicare and Medicaid Services currently encourage research on the impact of arthritis among persons with disabilities (5).

The purpose of this study was to examine the impact of OA as a comorbid condition during and after rehabilitation for persons who experience a hip fracture. Among elderly patients undergoing rehabilitation, hip fracture is an important cause of morbidity and mortality. In 1997, the attributable cost of hip fracture was approximated at $81,000; the estimated cost of all hip fracture in the US for that year was $20 billion (6). Hip fracture prevalence increases with age (7), as does OA prevalence (2). OA can interfere with rehabilitation activities, such as range of motion and strengthening exercises, potentially causing delay in functional recovery and longer length of stay among patients with hip fracture. We hypothesized that OA would be associated with longer inpatient rehabilitation stay and impaired functional recovery after a hip fracture. Examining the impact of OA on length of rehabilitation stay and functional recovery (Functional Independence Measure [FIM Instrument,8] ratings at discharge and followup, functional gain, and percentage of patients discharged home) will provide information necessary to evaluate and refine PPS for rehabilitation.


We conducted a retrospective cohort analysis of the Uniform Data System for Medical Rehabilitation (UDSMR) database. The UDSMR is the largest national registry of standardized information on medical rehabilitation inpatients in the US (8, 9). The UDSMR database includes demographic variables; up to 8 diagnoses (International Classification of Diseases, Ninth Revision [ICD-9] codes); length of stay; discharge setting; and performance on a standardized measure of basic daily living skills, the FIM Instrument. Detailed information on the UDSMR has been published elsewhere (9) and is also available at the Web site (

The FIM Instrument (8) serves as an interdisciplinary indicator of the severity of disability as defined by the World Health Organization. The FIM Instrument includes 18 items in 2 domains: motor and cognitive (Table 1). The motor domain includes 6 items assessing self care, 2 items assessing sphincter control, 3 items assessing transfers, and 2 items assessing locomotion. The cognitive domain includes 2 items assessing communication and 3 items assessing social cognition. The rating of individual items focuses on the amount of assistance needed by the person to complete the activity being evaluated, with a rating of 1 indicating complete dependence and a rating of 7 indicating complete independence. Total possible FIM ratings range from 18 to 126.

Table 1. Items included in the Functional Independence Measure (FIM Instrument): domains and subscales (8)
Motor domain
 Self care
  A. Eating
  B. Grooming
  C. Bathing
  D. Dressing upper body
  E. Dressing lower body
  F. Toileting
 Sphincter control
  G. Bladder management
  H. Bowel management
  I. Bed, chair, wheelchair
  J. Toilet
  K. Tub, shower
  L. Walk/wheelchair
  M. Stairs
Cognitive domain
  N. Comprehension
  O. Expression
 Social cognition
  P. Social interaction
  Q. Problem solving
  R. Memory
  7 = complete independence (timely, safely)
  6 = modified independence (device)
 Modified dependence
  5 = supervision
  4 = minimal assistance (subject 75%+)
  3 = moderate assistance (subject 50%+)
  2 = maximal assistance (subject 25%+)
 Complete dependence
  1 = total assistance (subject 0%+)

Facilities contributing to the UDSMR followed a standard protocol for collecting and submitting information. Using the UDSMR protocol, the FIM Instrument was administered to persons receiving inpatient medical rehabilitation within 72 hours of both admission and discharge. Followup data were collected by telephone interview conducted by the National Follow-Up Services 80–180 days after discharge. The interrater reliability and stability of the FIM information collected at discharge and followup ranged from 0.79 to 0.99 using the intraclass correlation approach (10, 11).

Overview of the analysis.

Data from the UDSMR were analyzed for patients admitted to inpatient rehabilitation facilities between 1994 and 2001 following a hip fracture. Each patient had a primary diagnosis for admission and up to 7 other diagnoses. Using all diagnoses, we identified patients who received inpatient rehabilitation after a hip fracture with and without comorbid OA (ICD-9 codes 715.0–715.99, 721.0–721.99, and 724.0–724.99). Our predictor variable was the presence of OA. Confounding variables were age, sex, race/ethnicity (non-Hispanic white versus nonwhite), admission FIM ratings, total hip arthroplasty, other comorbidities (none, 1–3, and >3) (12), and time to followup.

Outcome measures.

Primary outcomes were FIM ratings at discharge and followup, rehabilitation gain, length of rehabilitation stay, and percentage of patients discharged home. Weekly rehabilitation gain, the average weekly gain in FIM ratings while in rehabilitation, was calculated as follows: average of individual gain {[(discharge FIM − admission FIM)/length of stay] × 7} (12). A positive change in ratings indicated better functional status, and a negative change indicated worse functional status. Living setting, collected at discharge from medical rehabilitation, was dichotomized into discharge home versus not home. Length of stay was calculated as the total number of medical rehabilitation days. When a patient was transferred to an acute-care hospital and returned to the initial rehabilitation service within 30 days, we counted only those days on which the patient was at the rehabilitation service. Patients were excluded if the following variables were missing: age at admission, length of stay, admission FIM ratings, discharge FIM ratings, or followup interview.

Statistical analyses.

Descriptive and summary statistics were performed for demographic characteristics and outcomes. Kruskal-Wallis equality of population rank tests were performed to compare mean FIM ratings (admission, discharge, and followup), rehabilitation gain, and length of rehabilitation stay between patients with and without OA. Multiple linear regression analyses were performed to examine associations between OA and FIM ratings, rehabilitation gain, and length of rehabilitation stay, adjusted for appropriate covariates. Multiple logistic regression analysis was performed to examine associations between OA and rate of being discharged home, adjusted for appropriate covariates. Analyses were performed using Stata, version 7.0 (Stata, College Station, TX).


We studied 13,394 older patients (age ≥65 years) with followup data who were receiving inpatient rehabilitation services after a hip fracture; of these, 1,953 had a diagnosis of OA. Patients with OA were 0.6 years older than patients without OA (likely not clinically older). They were also more likely to be women and from an ethnic minority.

Baseline characteristics are presented in Table 2, and bivariate analyses of outcomes are presented in Table 3. The duration of rehabilitation stay was longer in patients with OA than patients without OA. Patients with OA had mean admission FIM ratings similar to patients without OA; however, patients with OA had higher mean FIM ratings at discharge. At followup, FIM ratings were similar between patients with and without OA. Rehabilitation gain (mean weekly gain in FIM ratings) was 11.8 for patients with OA and 11.9 for those without OA. The percentage discharged home was not different between the 2 groups.

Table 2. Characteristics of patients with and without osteoarthritis (OA) admitted to rehabilitation facilities after a hip fracture*
CharacteristicOA (n = 1,953)Non-OA (n = 11,441)
  • *

    Values are the percentage unless otherwise indicated.

  • Number of comorbidities does not include OA.

Age, mean ± SD years80.7 ± 7.280.1 ± 7.1
Female sex81.976.1
 Non-Hispanic white84.591.1
 Non-Hispanic black4.03.2
Table 3. Bivariate analyses of outcomes*
 OA (n = 1,953)Non-OA (n = 11,441)P
  • *

    Values are the mean ± SD unless otherwise indicated. Comparisons were performed using Kruskal-Wallis equality of populations rank test. OA = osteoarthritis; FIM = Functional Independence Measure.

FIM ratings on admission73.9 ± 14.773.5 ± 14.90.32
Weekly rehabilitation gain10.9 ± 10.810.8 ± 10.80.15
FIM ratings at discharge98.2 ± 17.696.4 ± 18.3< 0.01
FIM ratings at followup107.0 ± 18.4106.5 ± 18.70.39
Percentage discharged home81.382.20.34
Length of stay   
 Median (range)16 (1–107)15 (1–101) 
 Mean ± SD18.1 ± 10.016.5 ± 8.9< 0.01
Days to followup   
 Median (range)89 (80–178)89 (78–180) 
 Mean ± SD92.1 ± 12.391.2 ± 12.1 

The results of the multiple regression analyses are presented in Table 4. The adjusted mean length of stay was 1.4 days longer in patients with OA than in those without OA. The FIM ratings at discharge in patients with OA were ∼2.4 points higher on average than for patients without OA. Because the length of stay was also longer in patients with OA, the rehabilitation gain per week was similar in the 2 groups. The mean FIM ratings for both groups were similar at followup assessment. The percentage of patients discharged home was not different in the adjusted analyses (data not shown).

Table 4. Multiple linear regression coefficients associated with OA*
 Coefficients95% CIP
  • *

    OA = osteoarthritis; 95% CI = 95% confidence interval; FIM = Functional Independence Measure.

  • Adjusted for age, sex, race (white versus nonwhite), comorbidity, and total hip replacement (n = 13,256).

  • Adjusted for age, sex, race, comorbidity, total hip replacement, and admission FIM (n = 13,256).

  • §

    Adjusted for age, sex, race/ethnicity, comorbidity, total hip replacement, discharged FIM ratings, and time to followup (n = 12,919).

Admission FIM0.90.3, 1.6< 0.01
 Motor domain0.2−0.2, 0.70.3
 Cognitive domain0.70.4, 1.1< 0.01
Discharge FIM2.41.6, 3.3< 0.01
 Motor domain1.61.0, 2.3< 0.01
 Cognitive domain0.80.5, 1.1< 0.01
Weekly rehabilitation gain0.1−0.3, 0.50.6
Followup FIM§−0.1−0.8, 0.60.7
 Motor domain−0.1−0.7, 0.50.8
 Cognitive domain0.1−0.1, 0.30.2
Length of stay1.41.0, 1.8< 0.01


Our results indicate that OA was associated with a longer length of stay (mean 1.4 days) and higher FIM ratings at discharge in patients admitted for rehabilitation after a hip fracture after controlling for age, sex, race/ethnicity, total hip replacement, and comorbidity. Although patients with OA had statistically higher FIM ratings at discharge than patients without OA, the difference might not be clinically meaningful.

The small increase in FIM rating was presumably due to longer mean length of stay among patients with OA. When rehabilitation gain was calculated as the rate of FIM change per week, there was no difference between patients with and without OA. Any gain in functional skills resulting from the higher FIM ratings at discharge and the increased length of stay for patients with OA appears to have been lost at followup. Therefore, the longer length of stay probably resulted in equitable functional recovery at followup.

The data reported in this study were from 1994 to 2001 when rehabilitation was under a fee-for-service payment system prior to PPS implementation. In 2002, the Centers for Medicare and Medicaid Service began providing rehabilitation payment via PPS. PPS is adjusted for some comorbidities, but not for arthritis. A recent study examining the impact of PPS on rehabilitative care for persons with stroke reported that PPS reimbursements were an average of 37% lower than actual pre-PPS costs (13). The authors suggest that the difference in PPS reimbursement and actual cost will be made up by changes in service delivery, including decreased length of stay.

Our findings show that, prior to the introduction of PPS, persons with OA who sustained a hip fracture experienced a longer, probably justifiable inpatient rehabilitation stay than persons without OA. Under PPS, these individuals are more likely to be discharged before they are fully functional because PPS will pay for the same length of stay for patients with or without OA. From 1994 to 2001, in all groups undergoing inpatient rehabilitation, a trend towards shorter length of stay was observed by Ottenbacher and colleagues (12). The trend in length of stay under PPS and its impact on functional recovery among patients with and without comorbid OA is currently under investigation. The impact of PPS on functional recovery is an area of active investigation and has implications for future health policy.

Past studies on the impact of different comorbid conditions on functional status, living setting, and survival have yielded varying results (14–17). In our investigation of the impact of OA after stroke rehabilitation, we hypothesized that pain may contribute to delayed functional recovery. In this study, after hip fracture, the longer length of stay positively impacted functional recovery at discharge, but was justified for preserving equitable functional recovery at followup.

A strength of our study was the large sample size, which allowed us to perform multiple regression analyses adjusting for various covariates. However, this large sample size can also be a limitation: the statistically significant difference in FIM ratings at discharge may not have meaningful clinical significance. Another constraint was the use of ICD-9-CM codes, which tend to underreport comorbidity (18), lower the power of the study, and allow misclassification of patients into either a cohort with OA or a cohort without OA; however, systematic bias was not suspected (19). Also, the current database does not indicate site affected by OA; therefore, we could not compare the impact of upper- versus lower-extremity OA.

Other limitations included those associated with analyzing a large database (20). The sociodemographic information in the UDSMR database is obtained from medical records, performance-based observation, and self reports. Although the consistency of the information collection process has been examined (10, 11), the possibility of coding and reporting errors exists. Lastly, we did not have data from the acute-care portion of a patient's hospital stay. A priority for our future research is linking the UDSMR database with Medicare files containing information on acute care. This process will help define the sample and allow comparisons with cases not in the UDSMR database.

Among patients receiving rehabilitation services after hip fracture, a diagnosis of OA is associated with longer lengths of stay but similar rates of rehabilitation gain and discharge disposition compared with no OA diagnosis. The PPS implemented after the data for this study were collected should be monitored carefully to determine if the anticipated reductions in length of stay will have a differential impact on functional outcomes for persons with OA and other comorbidities.