Impact of osteoarthritis on the rehabilitation of patients following a stroke




To determine if functional gain for patients following stroke rehabilitation is adversely affected by osteoarthritis (OA).


A retrospective cohort analysis was conducted using data collected between 1994 and 2001 from a large national registry of US rehabilitation inpatients. Outcome variables were functional status (the Functional Independence Measure [FIM Instrument]) at discharge and followup, FIM gain during and after rehabilitation, length of stay, and discharge setting. The primary predictor variable was the presence of OA. Covariates were age, sex, race, other comorbidities, type of stroke, length of stay, and time to followup. Analysis was by multivariable regression.


Data from 3,094 patients with OA and 44,943 patients without OA admitted following a stroke to inpatient rehabilitation facilities were analyzed. In unadjusted analyses, OA was associated with significantly higher FIM ratings than patients without OA at admission (mean ± SD 65.7 ± 19.2 versus 63.1 ± 20.5; P < 0.001) and discharge (mean ± SD 89.2 ± 21.8 versus 87.7 ± 23.0; P < 0.001), but with lower FIM ratings at followup (mean ± SD 97.7 ± 24.7 versus 99.7 ± 24.9; P < 0.001). In multivariate analyses, adjusting for potential confounders, OA was associated with a 1.62-day increase in length of stay (95% confidence interval [95% CI] 1.15, 2.08) and a 1.37 smaller improvement in FIM scores between admission and followup (95% CI 0.62, 2.12). This smaller increase in FIM instrument score was caused by less improvement in FIM after discharge.


OA impairs recovery from stroke. This is compensated for by longer length of stay for inpatient medical rehabilitation. Reimbursement systems may need to consider the impact of OA as a comorbid condition in patients receiving stroke rehabilitation.


Stroke is the third leading cause of death in the US, with increasing prevalence among older adults. Many patients with previous stroke have impaired function of activities of daily living (1). When stroke occurs in people with existing comorbidity, the combined effect of stroke and comorbidity on physical disability can be additive or synergistic (2). Because individual comorbid conditions may affect stroke recovery differently, evaluating the impact of individual comorbidities on stroke recovery may be helpful. In particular, comorbidity may also increase hospitalization cost (3, 4).

We studied the impact of osteoarthritis (OA) on the change in functional status during inpatient rehabilitation and at followup. The results from a previous cross-sectional study that assessed disease prevalence with the use of self-report questionnaires suggested that the combination of stroke and arthritis negatively impacts self-care activity in older women (2). However, to our knowledge, the effect of OA on functional recovery from stroke has not been previously studied.

Because the prevalence of OA increases with age up to the ninth decade, both OA and stroke will become more prevalent as the US population ages. To evaluate the impact of OA on stroke recovery, we conducted a retrospective study using a cohort of patients admitted to hospital-based rehabilitation programs following a stroke. We hypothesized that OA was associated with delayed functional recovery from stroke causing a longer rehabilitation period.


Data source.

Data were collected from 744 hospitals in 48 states using the Uniform Data System for Medical Rehabilitation (UDSMR) registry, and National Follow-Up Services. The UDSMR, the largest national registry of standardized information on medical rehabilitation inpatients in the United States, has been used by rehabilitation facilities since 1987 (5–9). The UDSMR database includes sociodemographic variables, up to 8 diagnoses using International Classification of Diseases, ninth revision (ICD-9) codes, discharge disposition, and length of stay. In addition, scores on a standardized measure of basic daily living skills, the Functional Independence Measure (FIM Instrument, Uniform Data System for Medical Rehabilitation, UB Foundation Activities, Buffalo, NY), were recorded at admission, discharge, and followup (Table 1).

Table 1. Domains and subscales included in Functional Independence Measure Instrument*
  • *


  • Independence: 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%)

   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
   N. Comprehension
   O. Expression
 Social cognition
   P. Social interaction
   Q. Problem solving
   R. Memory

The FIM Instrument was administered within 72 hours of inpatient rehabilitation admission and discharge. Followup data were collected by telephone interview 80–180 days after discharge by nurses trained in administering and interpreting the FIM Instrument (10, 11). Independent researchers have confirmed the reliability and validity of the data collection process (12, 13). The FIM Instrument has demonstrated excellent sensitivity and responsiveness (14).

Variables of interest.

The predictor variable was the presence of OA (ICD-9 codes 715.0–715.99, 721.0–721.99, or 724.0–724.99). Outcome variables were functional status (FIM) at discharge and followup; rehabilitation gain; changes in FIM between admission, discharge, and followup; length of stay; and rate of discharge home. These variables are described below.

The FIM Instrument serves as an indicator of the severity of disability as defined by the World Health Organization. The FIM Instrument is an 18-item assessment measuring self-care, sphincter control, transfers, mobility, communication, and social cognition. The score range is 1–7, with 7 indicating “complete independence” and 1 indicating “complete dependence.” These areas are divided into 2 domains: motor (self-care, sphincter control, transfers, and locomotion subscales) and cognitive (communication and social cognition subscales). Total possible FIM scores range from 18 to 126. The interrater reliability of subscriber facilities has been demonstrated (intraclass correlation coefficient 0.96 for total FIM scores with κ values from 0.53 to 0.69 for individual items) (15–17).

Rehabilitation gain is the average weekly gain in FIM ratings while in rehabilitation: average of individual gain = ([discharge FIM − admission FIM]/length of stay) × 7. Length of stay is calculated as the total number of medical rehabilitation days. When a patient is transferred to an acute-care hospital and returns to the initial rehabilitation service within 30 days, we count only those days the patient was on the rehabilitation service. Post discharge information regarding living setting is dichotomized into discharge home versus not home in this study.

Covariates included patients' age on admission, sex, ethnicity (white versus nonwhite), type of stroke (bilateral, left-sided, right-sided, and nonparesis), time between discharge and followup, and comorbid conditions. To define the independent impact of OA, other comorbidities were classified into one of 3 categories: none, 1–3, or >3 (18).

Statistical analyses.

Patients' characteristics were summarized using descriptive statistics. Associations between OA and outcomes were identified by regression analyses. All analyses were performed using Stata version 7.0 (Stata, College Station, TX).


Admission, discharge, and followup information were available for 48,037 patients (age ≥16 years) receiving inpatient rehabilitation services after a stroke from 1994 through 2001. Univariate analyses of baseline data and outcomes are presented in Tables 2 and 3. Table 2 summarizes the characteristics of patients with and without OA. Patients with OA were older than those without OA and more likely to be female (65.6% versus 51.2%). Table 3 presents the FIM ratings and subscale ratings. The mean FIM ratings for those with OA was 2.6 points higher than for those without OA on admission. At discharge the FIM ratings were also higher among patients with OA; however, at followup assessment, they were 2.0 FIM points lower among patients with OA than among those without OA.

Table 2. Characteristics of patients with and without osteoarthritis (OA) admitted to rehabilitation facilities following a stroke*
 OA (n = 3,094)Non-OA (n = 44,943)
  • *

    P values < 0.001. Unless otherwise indicated, values are %.

  • P < 0.001.

Age, mean ± SD75.2 ± 9.869.5 ± 12.8
 Non-Hispanic white78.076.9
 Non-Hispanic black13.413.6
No. of comorbidities  
Length of stay, median (range)20 (1–123)19 (1–391)
Days to followup, median (range)90 (80–178)89 (79–180)
Table 3. Unadjusted admitting, discharge, and followup FIM ratings for stroke patients with and without osteoarthritis (OA)*
 OA (n = 3,094)Non-OA (n = 44,943)
  • *

    Values are mean ± SD. All P values < 0.001. FIM = Functional Independence Measure Instrument.

Admitting FIM65.7 ± 19.263.1 ± 20.5
 Motor subscale42.1 ± 14.241.1 ± 14.9
 Cognitive subscale23.6 ± 8.022.0 ± 8.5
Discharge FIM89.2 ± 21.887.7 ± 23.0
 Motor subscale62.3 ± 17.162.0 ± 17.7
 Cognitive subscale26.9 ± 7.025.7 ± 7.6
Followup FIM97.7 ± 24.799.7 ± 24.9
 Motor subscale66.7 ± 20.968.9 ± 20.7
 Cognitive subscale31.0 ± 5.530.7 ± 5.9

The relationship of OA to change in FIM ratings from admission to discharge and admission to followup was assessed using multivariable linear regression analysis, controlling for confounders (Table 4). OA was associated with a 1.37 smaller improvement in FIM rating from admission to followup. However, that decrement was entirely accounted for by delayed functional recovery between post discharge and followup. There was no significant effect of OA on change in FIM during the rehabilitation stay.

Table 4. Multivariate analysis of factors affecting change in FIM rating*
PredictorAdmission to followupAdmission to dischargeDischarge to followup
  • *

    Values are change in FIM (95% confidence interval). Analyses have been adjusted for type of stroke. FIM = Functional Independence Measure Instrument.

Osteoarthritis−1.37 (−2.12, −0.62)0.38 (−0.10, 0.86)−1.55 (−2.21, −0.89)
Age (each year)−0.33 (−0.35, −0.32)−0.20 (−0.21, −0.19)−0.19 (−0.20, −0.17)
Sex (female versus male)−1.45 (−1.82, −1.08)−0.33 (−0.57, −0.09)−1.10 (−1.42, −0.77)
Ethnicity (non-white versus white)−1.42 (−1.87, −0.89)−1.41 (−1.70, −1.13)−0.30 (−0.69, 0.09)
No. of comorbidities   
 1–3 versus 0−0.54 (−1.77, 0.69)0.31 (−0.47, 1.09)−0.53 (−1.60, 0.54)
 >3 versus 0−1.15 (−2.33, 0.04)−0.06 (−0.81, 0.70)−0.84 (−1.88, 0.19)
Admission FIM−0.33 (−0.34, −0.32)−0.10 (−0.10, −0.09)
Discharge FIM−0.24 (−0.25, −0.23)
Time to followup−0.05 (−0.06, −0.03)−0.04 (−0.05, −0.02)

Further multivariate analyses of the motor and cognitive domains of followup FIM revealed that only the motor domain was negatively influenced by OA. Under the motor domain, subscale ratings for self-care, transfer, and locomotion were impacted by OA, while the subscale rating for sphincter control was not influenced by OA (data not shown).

As shown in Table 5, multivariate analyses revealed that OA was associated with a 1.6-day increase in length of stay. With equivalent change in FIM during this time, the average increase in FIM per week during inpatient rehabilitation was somewhat lower in those with OA than in those without (mean ± SD 9.4 ± 7.4 versus 10.3 ± 9.4; P < 0.001). OA was not associated with lower rate of returning home (81% for both groups).

Table 5. Multivariate analysis of effect of osteoarthritis diagnosis on length of stay*
 Coefficient95% confidence interval
  • *

    Coefficients have been adjusted for sex, ethnicity, and type of stroke. All P values < 0.001.

  • Continuous variable. FIM = Functional Independence Measure Instrument.

Osteoarthritis1.62(1.15, 2.08)
Age at admission−0.08(−0.09, −0.07)
1–3 comorbidities1.49(0.75, 2.24)
>3 comorbidities2.27(1.54, 2.98)
Admission FIM−0.29(−0.30, −0.28)

To determine if our results were robust, we repeated the analyses using 2 additional regression models. In the first one, we computed the analyses for all outcomes adjusting for hemorrhagic stroke, in addition to the previously stated covariates, and in the second model we restricted the analyses to patients ages 50 years or older (∼93% of the total number of patients). The repeated analyses confirmed the impact of OA on FIM ratings and length of stay.


In patients after stroke admitted to rehabilitation hospitals, OA was associated with a longer length of stay. However, overall change in FIM ratings during rehabilitation was similar in patients with and those without OA. After discharge from rehabilitation facilities, OA was associated with less continued progress of rate of recovery.

During this study, Medicare reimbursement for rehabilitation hospitalization was under a fee-for-service payment system. The longer length of stay for stroke patients with OA was presumably made by providers to compensate for the somewhat slower rate of recovery of these patients. Starting in 2002, Medicare reimbursement for rehabilitation hospitalization changed to a prospective payment system (PPS) based on Case-Mix Groups (9). It is possible that under the new PPS, pressure to reduce the length of stay may negatively impact recovery among post-stroke patients with OA. This is a question that requires continued monitoring and additional research.

The PPS allows for higher payment based on comorbidities, but OA is not one of the comorbidities that triggers increased reimbursement (19). Interestingly, OA had a more noticeable effect on stroke outcome than did a number of other comorbidities. For example, in Table 4, OA was associated with a significant change in FIM rating between admission and followup, whereas both categories of other comorbidities (1–3 or >3) had no significant effect. One explanation might be that many of these conditions were asymptomatic, or produced symptoms that would not interfere with rehabilitation activities. Whatever the reason, our findings suggest that OA is indeed a relevant comorbidity in the PPS for rehabilitation after stroke.

After discharge, patients with OA continue to lag in improvement compared with patients without OA. Although a 1.37-point lower FIM rating at followup is of uncertain clinical significance, the delayed functional recovery may have other implications. Dunlop et al have reported that functional limitations may be a mediator of arthritis-related depression risk (20). Moreover, the disparity in improvement might continue to increase after the followup assessment. This speculation suggests the potential need for continued outpatient rehabilitation services for patients with OA after stroke.

Our findings are consistent with some previous studies reporting a negative impact of comorbid conditions on rehabilitation (21, 22), but not all (4, 23). The most important difference between our investigation and earlier studies is that OA has not previously been examined as a single comorbidity. Also, a larger sample size enabled us to use multivariate procedures for a more detailed analysis. Our outcome variables were also different and included followup FIM ratings.

FIM rating was higher at admission for patients with OA. We are unsure of the reason, but it may be related to an expected higher use of aspirin by these patients, which might then lessen stroke severity.

There are several reasons why OA may impair recovery from stroke; OA pain might have interfered with rehabilitation gain and functional recovery. Patients with rheumatic disease rarely experience complete pain relief (24); such pain would interfere with activities such as standing, walking, sitting, and sleeping (25), the very activities that are the focus of rehabilitation efforts during and after discharge from rehabilitation facilities. Also, pain is associated with depression, anxiety, and fatigue (25), all of which can interfere with stroke rehabilitation and recovery. Supervised exercise programs among patients with OA have been shown to improve pain and self-reported disability (26). Further studies may be indicated for outpatient intervention programs to improve functional outcome for patients with OA recovering from stroke.

Our database has several limitations. One constraint is the use of ICD-9 Clinical Modification (CM) codes, which tend to under report comorbidity (27), perhaps resulting in lower power of the study. The ICD-9-CM codes could also allow misclassification among both groups, but systematic bias is unlikely is this situation (28).

Other limitations include those associated with analyzing a large database (29). 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. Another limitation is the lack of information regarding acute-care hospitalization. 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 admitted to a rehabilitation facility after a stroke, a concurrent diagnosis of OA is associated with longer length of stay, but equal gains in functional status. After discharge from rehabilitation, OA is associated with a slower continuing recovery in functional status. Our findings should be verified through additional studies, but we suggest that the new inpatient medical rehabilitation prospective reimbursement system should be monitored for patients with stroke to determine if cost adjustments are indicated for OA.