Effect of preoperative exercise on measures of functional status in men and women undergoing total hip and knee arthroplasty




To evaluate the effect of a short preoperative exercise intervention on the functional status, pain, and muscle strength of patients before and after total joint arthroplasty.


A total of 108 men and women scheduled for total hip arthroplasty (THA) or total knee arthroplasty (TKA) were randomized to a 6-week exercise or education (control) intervention immediately prior to surgery. We assessed outcomes through questionnaires and performance measures. Analyses examined differences between groups over the preoperative and immediate postoperative periods and at 8 and 26 weeks postsurgery.


Among THA patients, the exercise intervention was associated with improvements in preoperative Western Ontario and McMaster Universities Osteoarthritis Index function score (improvement of 2.2 in exercisers versus decline of 3.9 in controls; P = 0.02) and Short Form 36 physical function score (decline of 0.4 in exercisers versus decline of 14.3 in controls; P = 0.003). No significant differences were seen in TKA patients. Exercise participation increased muscle strength preoperatively (18% in THA patients and 20% in TKA patients), whereas the control patients had essentially no change in strength (P > 0.05 for exercise versus education in both THA and TKA groups). Exercise participation prior to total joint arthroplasty substantially reduced the risk of discharge to a rehabilitation facility in THA and TKA patients (adjusted odds ratio 0.27, 95% confidence interval 0.074–0.998). The intervention had no effects on outcomes 8 and 26 weeks postoperatively.


A 6-week presurgical exercise program can safely improve preoperative functional status and muscle strength levels in persons undergoing THA. Additionally, exercise participation prior to total joint arthroplasty dramatically reduces the odds of inpatient rehabilitation.


Osteoarthritis is an increasingly prevalent chronic disease of aging Americans and the leading cause of disability in the US (1). Progression of symptoms often leads to decreased mobility, deconditioning, and reduced functional status and quality of life. Total joint arthroplasty improves pain and its sequelae in patients with end-stage arthritis (2–4). In 2002, 484,484 primary total joint arthroplasties (320,174 total knee arthroplasties [TKAs] and 164,310 total hip arthroplasties [THAs]) were performed in the US, mostly for osteoarthritis (5–7). This represents a 43% increase from the 339,792 such procedures performed in 1993 (7).

Appropriate exercise offers many benefits in treating the patient with osteoarthritis (8–10). Stronger, better conditioned periarticular muscles, tendons, and ligaments have advantageous biomechanical effects on attenuating joint forces during movement (1). In more severe disease, which often leads to reduced mobility and disuse atrophy, exercise can improve pain, muscle strength, cardiovascular fitness, self-efficacy, and function (9–11). Aerobic exercise has been the most widely studied type of exercise in the osteoarthritis population (12, 13), with increasing attention being paid to the effects of strength training (9, 10, 14, 15). Little is known about the effects of exercise on patients with end-stage osteoarthritis.

Exercise is a cornerstone of rehabilitation following total joint arthroplasty and other surgical procedures (16). Little attention, however, has been placed on the potential role exercise might play in preparation for surgery. Arthur and colleagues (17) examined the effects of an exercise and education intervention prior to coronary artery bypass graft surgery. They reported that patients who participated in a presurgical fitness program experienced an increase in preoperative functional status and quality of life and a reduction in postoperative length of hospital stay and time in the intensive care unit. This study may have particular relevance to total joint arthroplasty, because preoperative functional status has been shown to be strongly related to postoperative status in hip and knee arthroplasty patients (18). Data on the direct effects of a prearthroplasty exercise intervention are scarce. In the past decade, 3 studies have examined the effects of preoperative exercise interventions on recovery from TKA (19–21). Methodologic limitations, including less potent exercise interventions and small sample sizes, render the results of these studies inconclusive. Recently, Gilbey and colleagues (22) reported the positive effects of a more sophisticated exercise intervention on improved postoperative functional status in hip replacement patients. The purpose of this prospective, randomized, controlled study was to examine the effectiveness of a 6-week preoperative program of cardiovascular, strength-training, and flexibility exercise on the level of pre- and postoperative function, pain, and muscle strength in a large sample of men and women undergoing primary THA or TKA.



Eligible candidates were men or women scheduled to undergo unilateral, primary THA or TKA for advanced osteoarthritis. Additional eligibility criteria included the ability to answer questions in English and an interval of 8–12 weeks between enrollment and surgery to permit time for the intervention. Eligible patients were those from 7 collaborating orthopedic surgeon practices scheduled for THA or TKA between November 2001 and November 2003 at the New England Baptist Hospital (Boston, MA). A research assistant administered a telephone screen to determine a patient's interest and eligibility. Patients with inflammatory arthritis (i.e., rheumatoid arthritis, systemic lupus erythematosus), Parkinson's disease, or any medical condition in which a moderate level of exercise was contraindicated (i.e., uncontrolled diabetes or hypertension) were excluded. Patients were also excluded if they were scheduled to have bilateral joint replacements or an extended out-of-town vacation during the 6 weeks prior to surgery.

Prior to the baseline testing visit, medical clearance was obtained from the patient's primary care physician and a packet was mailed to the patient that included a welcome letter, directions to the test location, a copy of the informed consent document, and a self-assessment questionnaire. Participants' questions were answered and they gave written informed consent at the baseline visit before testing began. The Institutional Review Boards of the New England Baptist Hospital and Beth Israel Deaconess Medical Center approved the study.

Patient assessment.

Data were collected at 4 testing time points (preintervention, postintervention [presurgery], and 8 and 26 weeks postsurgery) and during the hospitalization. Pre- and postintervention testing occurred within 7 days of starting and completing the exercise program. Postintervention testing was performed within 1 week of surgery. A research assistant and 3 graduate students trained in the measurement techniques administered the performance measures. Test administrators used written scripts to standardize verbal instruction and intertrial rest periods.

Outcome measures.

This study was designed to evaluate the intervention and its effect on a range of clinically related patient self-assessment and performance-based outcome measures. The function subscale of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (23) was our primary outcome measure.

Patient self-reported measures.

The WOMAC (23) and Short Form 36 (SF-36) (24) were used to provide disease-specific and generic assessments of functional status, respectively. The WOMAC was designed and validated specifically to assess activities integral to functional independence in persons with osteoarthritis of the lower extremity (25), and consists of a 17-item functional status subscale, a 5-item pain subscale, and a 2-item stiffness subscale. The WOMAC function scale is scored from 0 (best) to 68 (worst), whereas the WOMAC pain scale is scored from 0 (best) to 20 (worst). The SF-36 is a valid and reliable instrument for assessing the general health and function of men and women who have undergone THA (24, 26) and is a core component of suggested outcome measures for this procedure (25). The 36-item scale covers 8 dimensions of health. The SF-36 physical function and bodily pain scales served as secondary outcome measures in this study. Both are scored from 0 (worst) to 100 (best).

Performance measures.

We used the 1-repetition maximum leg press test to determine lower-extremity strength (27, 28). The leg press was selected as a measure of integrated lower-extremity function and strength because the movement requires hip and knee extension that involves synchronized activation of muscles around the 2 joints, similar to the common activity of standing up from a chair. The participant performed repeated single repetitions of the leg press movement (Life Fitness, Franklin Park, IL) separated by 60-second rest periods. The tester demonstrated proper technique of the movement and instructed how to perform the test. Patients performed 6–8 repetitions with low resistance to warm up and to become familiar with the movement. The tester systematically increased the resistance after each repetition until the participant's maximum voluntary muscle force could not move the resistance through the full range of motion (i.e., momentary failure). The heaviest weight a person could move through a full range of motion was recorded as the 1-repetition maximum. Our goal was to identify the 1-repetition maximum within 5 trials.

Balance was assessed by the functional reach test (29) using a yardstick mounted to a wall at the height of the patient's acromion. Patients stood parallel to the wall and reached forward along the yardstick as far as possible without taking a step or twisting. The average of 2 trials was used as the score.

Mobility was assessed using the timed up and go test (30), which requires a person to rise from a chair of standard height, walk 10 feet, turn 180°, return to the chair, and sit down. The average of 2 trials was used as the score.

Salient outcomes from the inpatient admission were documented with medical record review. Data included surgeon, dates of admission and discharge, length of stay, discharge location (home, rehabilitation facility), amount and type of pain medication, distance walked on postoperative day 3 (<50 feet, >50 feet), and perioperative complications.


The exercise protocol was adapted from previous studies by our group of persons with osteoarthritis and fibromyalgia (27, 28). Because this was our first exercise study of patients undergoing total joint replacement, we selected a regimen of exercises and progression of intensity of the intervention to accommodate the expected level of deconditioning seen in patients with osteoarthritis (31). The exercise group performed water and land-based exercise 3 times weekly over a 6-week period immediately prior to surgery. During the first 3 weeks, participants performed 1–2 sets of 8–12 repetitions of single-joint movements while standing in chest-deep, 93°F water. Pool exercises focused on single planar motion of the cervical spine, shoulders, elbows, wrists, hands, hips, knees, and ankles. During weeks 4–6, exercise sessions involved a total body fitness program of cardiovascular, strength, and flexibility training. Participants used the recumbent stationary bicycle or elliptical device for 10 minutes of cardiovascular exercise at moderate intensity. Strength training activities, the focus of the intervention, were performed isotonically and included the seated row, chest press, and leg press movements using resistance machines (Life Fitness); biceps curl and triceps kickback exercises using hand weights; and movements for the shoulders and abdomen using bodyweight. Two sets of 8–12 repetitions of each resistance activity were performed at each session. The intervention was individually tailored to each person's fitness level and comfort with performing the movements. At the end of each session, participants performed flexibility exercises involving hip, knee, and ankle flexors and extensors and hip adductors, holding each position for 20 seconds and repeating twice. A physical therapist (SAB) administered the exercise sessions, which lasted 30–60 minutes and took place at a community fitness facility.

We used information from the preoperative education booklet provided to all patients undergoing total joint arthroplasty at New England Baptist Hospital as our control intervention. One handout included information on modifying a home to improve accessibility and to reduce the risk of falling and injury for a person recovering from total joint replacement surgery. The second handout provided information on preparing for surgery and things to bring for the hospital stay. Patients in the education control group received the first handout via mail during week 1 of the 6-week intervention and a followup telephone call during week 2. During week 3, participants received the second handout, and during week 4 they received a telephone call to answer any questions and schedule the postintervention testing appointment. A final telephone call during week 6 confirmed the postintervention testing date and time. The 2 mailings and 3 telephone calls were designed to provide some degree of attention control.

Statistical power and randomization procedures.

Power calculations using data from an unpublished pilot study documented that a sample size of 30 intervention and 30 control patients in each surgical group (60 THA and 60 TKA, for a total of 120 patients) would provide sufficient power to detect a change in the primary outcome variable (WOMAC function score) of 0.74 SDs (effect size of 0.74). This corresponds to an 8-point difference in score, which has been shown to be clinically meaningful (32). Patients were randomized using a block design for age, joint, and surgeon.

Statistical analysis.

All analyses were performed using SAS software, version 8.1 (SAS Institute, Cary, NC). The analyses of WOMAC and SF-36 function and pain scores compared the exercise and control groups at several points, including 1) measures at baseline, measures at the preoperative (postintervention) evaluation, and changes from baseline to the preoperative evaluation, and 2) measures at week 8 and week 26 postsurgery and changes from baseline. Values and changes from baseline of continuous outcomes between the intervention and control groups at each time point were compared using t-tests. As the primary analysis, we used t-tests to analyze results at specific time points, because these results are most readily interpretable.

The primary objective of this study was to evaluate the effect of the intervention on functional status, pain, and muscle strength in patients undergoing THA and TKA. Therefore, we defined intent-to-treat to include all study completers regardless of their level of adherence to the intervention and only reported results from completers. We also performed a longitudinal analysis with baseline, treatment, time, and time–treatment interactions. The results of the longitudinal analysis confirmed the more straightforward t-test results, which are presented here. The analyses of secondary outcomes were exploratory in nature; therefore, we did not adjust for the multiple comparisons. We used 0.05 as the significance level for all tests.

Additional analyses evaluated the effects of the intervention on discharge location (home versus inpatient rehabilitation) and ability to walk >50 feet independently. Logistic regression was used to examine the association between discharge location (rehabilitation versus home) or ambulation distance at discharge (>50 feet versus <50 feet) and covariates of interest: treatment (exercise versus control), joint (hip versus knee), age, sex (female versus male), and baseline WOMAC pain and function scores.


Of the 942 patients who met the inclusion criteria, 108 consented and enrolled in the study, resulting in a recruitment rate of 12% (Figure 1). Patients gave 2 main reasons for not participating: problems with transportation to the intervention site and the time commitment needed to participate. Throughout the 32-week study period, 30 participants (28%) dropped out at various times. Dropouts included 14 THA patients (7 exercisers, 7 controls) and 16 TKA patients (8 exercisers, 8 controls). Data presented here are from the 49 THA and 29 TKA patients (Figure 1) who completed the study. Seventeen participants (16%; 4 THA exercise, 5 THA control, 4 TKA exercise, and 4 TKA control) dropped out prior to the end of the intervention period (preoperatively). Three participants withdrew because of medical issues unrelated to the study, 5 did not like the intervention to which they were randomized, 1 decided to have bilateral hip arthroplasty, 2 could not fulfill the time commitment necessary to participate in the exercise group, and 6 cancelled their surgeries. Postoperatively, 2 participants dropped out because of medical issues unrelated to the surgery and 1 was lost to followup by week 8. Ten participants were lost to followup between weeks 8 and 26. Adherence to the exercise intervention was high; participants attended an average of 16 (89%) of 18 class sessions.

Figure 1.

Enrollment, randomization, and retention results.

Randomization created similar groups at baseline (Table 1). Overall, patients in the THA group were 57% women, overweight to obese by body mass index (BMI) calculations, ∼60 years of age, and had 1–2 comorbidities. Patients in the TKA group were 53% women, obese by BMI, slightly older than THA patients at ∼67 years of age, and had 2–3 comorbidities. Baseline data of completers and dropouts within each of the 4 randomization groups were similar (Table 2). The only difference in baseline measures was a higher WOMAC function score in the THA exercise dropout group. The score suggests that dropouts experienced a higher level of function than completers. SF-36 physical function scores were not different between the groups.

Table 1. Characteristics of study participants having primary hip and knee arthroplasty*
 Hip (n = 63)Knee (n = 45)
Exercise (n = 32)Control (n = 31)Exercise (n = 22)Control (n = 23)
  • *

    Values are the mean ± SD unless otherwise indicated. BMI = body mass index.

Female, no (%)20 (63)16 (52)11 (50)13 (57)
Age at surgery, years65 ± 1159 ± 765 ± 869 ± 8
No. of comorbidities, median (range)1.0 (0–7)1.0 (0–6)2.0 (0–8)1.0 (0–6)
Medications, no.    
Height, cm168.7 ± 10.5167.4 ± 10.4170.2 ± 8.7166.0 ± 9.2
Weight, kg81.1 ± 18.485.3 ± 27.3102.1 ± 20.592.7 ± 14.5
BMI, kg/m228.4 ± 5.330.3 ± 9.135.7 ± 9.233.9 ± 6.5
Table 2. Characteristics of total hip and total knee arthroplasty patients who completed the study and who dropped out*
Completed (n = 25)Dropped (n = 7)Completed (n = 24)Dropped (n = 7)
  • *

    Values are the mean ± SD unless otherwise indicated. BMI = body mass index; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 = Short Form 36.

Total hip arthroplasty    
 No. of patients257247
 Age, years65 ± 1166 ± 1559 ± 765 ± 9
 Height, cm170.2 ± 10.2175.3 ± 2.5166.9 ± 11.7162.1 ± 4.1
 Weight, kg79.4 ± 19.186.2 ± 7.788.9 ± 29.570.3 ± 18.1
 BMI, kg/m227.0 ± 4.428.0 ± 2.831.6 ± 8.526.5 ± 5.6
 WOMAC pain7.5 ± 3.010.5 ± 4.58.5 ± 3.010.0 ± 4.5
 WOMAC function25.5 ± 11.940.8 ± 11.928.9 ± 10.230.6 ± 15.3
 SF-36 physical function41.4 ± 19.538.3 ± 16.144.6 ± 14.144.3 ± 28.1
Total knee arthroplasty    
 No. of patients148158
 Age, years67 ± 963 ± 368 ± 970 ± 9
 Height, cm170.9 ± 9.9168.9 ± 7.6165.6 ± 7.4160.2 ± 0
 Weight, kg99.8 ± 20.4104.8 ± 2489.4 ± 15.490.3 ± 18.2
 BMI, kg/m234.7 ± 9.536.9 ± 8.632.7 ± 6.535.2 ± 7.8
 WOMAC pain7.0 ± 2.08.0 ± 3.06.5 ± 4.57.5 ± 3.0
 WOMAC function25.5 ± 8.527.2 ± 11.922.1 ± 10.223.8 ± 15.3
 SF-36 physical function48.3 ± 15.840.1 ± 2345.8 ± 2239.9 ± 11.7

Postintervention, presurgical period.

In THA patients, exercisers improved or stabilized their WOMAC function and pain scores, respectively, from baseline to the 6-week presurgical evaluation. Scores of controls worsened, resulting in significant differences in the change in function score from baseline to preoperation between the exercise and control groups prior to surgery (mean difference 6.1 points, P = 0.02). Comparison of changes in pain scores (mean difference 1.3 points, P > 0.05) was not statistically significant (Table 3). TKA patients showed no significant difference in either WOMAC score between the 2 groups (Table 4).

Table 3. Scores for all outcome measures of total hip arthroplasty participants (25 exercisers and 24 controls)*
 BaselinePreoperative8 Weeks26 Weeks
  • *

    Values are the mean ± SD. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 = Short Form 36.

  • P < 0.05 for between-group differences and for between-group changes from baseline determined by t-test.

  • P < 0.01 for between-group changes from baseline determined by t-test.

  • §

    P < 0.05 for between-group changes from baseline determined by t-test.

WOMAC function    
 Exercise29.1 ± 12.926.9 ± 11.912.8 ± 9.05.4 ± 5.8
 Control29.8 ± 11.233.7 ± 10.912.9 ± 8.05.3 ± 5.4
WOMAC pain    
 Exercise8.0 ± 3.77.8 ± 4.12.6 ± 2.61.1 ± 1.7
 Control8.8 ± 3.29.9 ± 2.92.7 ± 2.01.0 ± 1.2
SF-36 physical function    
 Exercise40.8 ± 18.640.4 ± 23.457.6 ± 22.081.7 ± 18.1
 Control44.6 ± 17.630.3 ± 17.155.1 ± 22.476.6 ± 18.6
SF-36 pain    
 Exercise47.9 ± 20.349.5 ± 19.471.4 ± 20.179.6 ± 21.2
 Control43.8 ± 14.437.7 ± 17.970.8 ± 21.277.4 ± 16.3
SF-36 role limitation physical    
 Exercise49.2 ± 41.244.6 ± 37.548.2 ± 41.983.0 ± 35.2
 Control50.3 ± 40.232.1 ± 39.047.2 ± 42.486.5 ± 24.4
1-repetition maximum, kg    
 Exercise87 ± 47102 ± 4980 ± 3299 ± 37
 Control104 ± 50103 ± 43106 ± 51117 ± 51
Functional reach, cm    
 Exercise28.2 ± 8.130.6 ± 6.630.7 ± 6.933.5 ± 5.2§
 Control32.0 ± 7.731.5 ± 7.132.8 ± 6.131.4 ± 7.1§
Timed up and go, seconds    
 Exercise12.39 ± 3.2811.35 ± 2.3511.53 ± 2.429.76 ± 1.29
 Control11.00 ± 2.1911.3 ± 2.2510.9 ± 2.839.41 ± 1.46
Table 4. Scores for all outcome measures of total knee arthroplasty participants (14 exercisers and 15 controls)*
 BaselinePreoperative8 Weeks26 Weeks
  • *

    Values are the mean ± SD. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 = Short Form 36.

  • P < 0.05 for between-group changes from baseline determined by t-test.

  • P < 0.05 for between-group differences and P < 0.01 for between-group changes from baseline determined by t-test.

WOMAC function    
 Exercise26.2 ± 9.227.7 ± 11.616.3 ± 7.19.9 ± 9.0
 Control23.1 ± 11.925.0 ± 11.915.3 ± 11.41.4 ± 11.9
WOMAC pain    
 Exercise7.4 ± 2.37.3 ± 0.74.7 ± 2.42.4 ± 2.7
 Control6.8 ± 4.07.5 ± 5.05.0 ± 3.42.3 ± 2.0
SF-36 physical function    
 Exercise45.5 ± 18.634.0 ± 21.549.9 ± 15.068.0 ± 19.8
 Control43.7 ± 18.840.2 ± 19.453.1 ± 26.366.1 ± 26.6
SF-36 pain    
 Exercise47.5 ± 17.842.1 ± 16.659.8 ± 16.471.2 ± 19.3
 Control55.9 ± 22.156.7 ± 21.468.1 ± 16.668.1 ± 25.1
SF-36 role limitation physical    
 Exercise50.0 ± 39.920.8 ± 33.448.7 ± 38.671.1 ± 36.4
 Control52.5 ± 38.158.9 ± 41.249.4 ± 39.858.3 ± 46.9
1-repetition maximum, kg    
 Exercise97 ± 37116 ± 3799 ± 47129 ± 42
 Control103 ± 52102 ± 4686 ± 43111 ± 49
Functional reach, cm    
 Exercise26.7 ± 7.626.7 ± 7.627.9 ± 6.429.2 ± 7.6
 Control28.0 ± 5.128.0 ± 6.426.7 ± 7.627.9 ± 7.6
Timed up and go, seconds    
 Exercise11.85 ± 2.3812.15 ± 2.6512.03 ± 2.0510.08 ± 1.76
 Control12.55 ± 6.0110.38 ± 1.7712.40 ± 3.2211.66 ± 4.47

In THA patients in the exercise group, the SF-36 physical function subscale scores improved whereas bodily pain scores remained constant over the 6-week intervention period; these scores worsened in the control group (Table 3). These changes resulted in different mean scores at the presurgical evaluation that were significant in physical function (mean difference 13.9 points, P = 0.003) and role limitation physical (mean difference 13.6 points, P = 0.007), but not in bodily pain (mean difference 7.7 points, P = 0.11). TKA patients in the exercise group worsened, whereas controls remained constant during the intervention period. This resulted in a significant difference between the 2 groups at the presurgical evaluation (Table 4).

The THA exercise group had lower leg press strength at baseline compared with the control group. The mean strength scores of the 2 groups were almost the same at the postintervention, presurgical evaluation. The exercisers improved their scores by an average of 18%, whereas controls had almost no change over the 6-week intervention period (P = 0.16). Baseline leg press scores in TKA patients were closer between the exercise and control groups than in the THA patients. TKA exercisers demonstrated a 20% increase in strength, whereas controls showed no change (P = 0.08).

Perioperative period.

There were 12 postoperative complications in 11 of the 87 participants with complete perioperative data: 4 with THA data and 7 with TKA data. THA patients who exercised had 0 complications as compared with 4 complications in the education group (P = 0.04). Complications in TKA patients were seen in 3 exercisers and 4 nonexercisers. There were no serious adverse events (those requiring readmission to the hospital) in any of the patients.

Sixty-five percent of the exercise group were discharged directly home after the inpatient stay and 35% were discharged to an inpatient rehabilitation facility. In contrast, 44% of the education control patients were discharged home and 56% were discharged to an inpatient rehabilitation facility following surgery (Table 5). Similarly, 76% of exercisers were able to walk 50 feet on day 3 of the inpatient hospitalization compared with 61% of the education control group.

Table 5. Discharge location and ability to walk >50 feet at discharge
 Discharged to homeDischarged to rehabilitationWalk >50 feetWalk <50 feet
  • *

    P < 0.05 for chi-square test of between-group differences.

Combined sample    
Total hip replacement    
Total knee replacement    

In a stepwise logistic regression model, participation in the exercise program, higher baseline WOMAC function score, younger age, and male sex were all associated with a lower likelihood of discharge to home than to a rehabilitation facility (P < 0.05 for each). After adjustment for sex, baseline WOMAC function score, and age, patients in the exercise group were much less likely to be discharged to a rehabilitation facility than patients in the education group (adjusted odds ratio [OR] 0.27, 95% confidence interval [95% CI] 0.07–0.99).

Similarly, in a stepwise logistic regression model, exercise treatment group and hip replacement (versus knee replacement) were identified as significant predictors of being able to ambulate >50 feet on day 3. Adjusting for joint, the exercise patients had 3.2-fold greater odds of ambulating >50 feet (adjusted OR 3.2, 95% CI 1.2–8.9).

Postoperative period.

No significant effect of the intervention was seen at weeks 8 or 26 postoperatively for self-assessed function and pain measured by WOMAC and SF-36 (Tables 3 and 4). Postoperative SF-36 scores improved more from baseline in the exercise group than in the control group, especially for TKA patients.

For the THA patients, 1-repetition maximum leg press scores at the 8-week postsurgery evaluation were similar to the baseline values. However, the exercise group demonstrated a 21% reduction in leg press scores from the presurgery evaluation to the 8-week postsurgery evaluation, whereas there was almost no change in the control group (Table 3). The leg press scores improved in both groups from week 8 to week 26 postsurgery and were not significantly different at week 26. Compared with baseline, THA exercise and control groups had modest improvements of 14% and 12% at week 26, respectively. For TKA patients, both the exercise and the control groups lost a similar percentage of strength (16%) at week 8 compared with presurgery evaluations. Compared with baseline, however, exercisers had a 2% increase in leg press scores, whereas controls had a 17% decrease at week 8 postsurgery. Both groups showed improvement in muscle strength from baseline to week 26 postsurgery (32% in exercisers versus 8% in controls).


We evaluated the effect of a 6-week preoperative exercise intervention on self-assessed and performance-based measures of functional status before and after THA and TKA using a randomized controlled trial design. We found that hip and knee arthroplasty patients who completed the 6-week intervention responded differently in the preoperative and immediate postoperative period. Although both patient groups that exercised increased lower-extremity muscle strength prior to surgery, only those scheduled for hip replacement surgery showed concurrent improvements in function levels. Of particular note, THA and TKA patients who participated in the exercise intervention went home more frequently at discharge than nonexercisers, who more often went to an inpatient rehabilitation facility. The exercisers were also more likely to walk >50 feet at discharge.

Our findings show that an appropriately designed program of water and land-based exercise involving cardiovascular, strength training, and flexibility activities can be a safe, tolerable, and effective approach to improving muscle strength in middle-aged and older adults with severe osteoarthritis of the hip and knee. These findings support similar results found in samples reported to have moderate (10, 27) and end-stage (22) disease.

Our study design allows a comparison of the responses of 2 patient populations, THA and TKA, to a single preoperative exercise intervention. The data show that leading up to surgery, THA patients who exercised stabilized their levels of function and improved muscle strength (18%), whereas nonexercisers experienced worse function and pain and had no change in strength. Interestingly, both TKA groups responded similarly, showing stability or decline in self-reported levels of function during the preoperative period despite the exercise group demonstrating a 20% increase in muscle strength. This lack of translation of muscle strength to improvement in function or pain level in TKA patients suggests that the effect of this amount of exercise on patients with advanced osteoarthritis may depend on the location of the involved joint. Our findings support the current literature reporting positive findings in THA patients who used a strength-training intervention (22) and null findings in TKA patients using a physical therapy or exercise regimen prior to surgery (19–21). Our intervention used strength training for only 3 weeks (9 exercise sessions), which is well below the recommended duration required to bring about significant strength gains (33). Therefore, the observed increase in strength is most likely due to a combination of increased neuromuscular coordination (34) and a reduction in fear of anticipated pain associated with increased muscular effort. An improvement in preoperative function or pain in TKA patients may require greater changes in lower-extremity muscle strength, which would require an intervention of longer duration. During the postoperative period, all 4 patient groups showed expected improvement in function, pain, and muscle strength.

Our most striking finding was that regardless of affected joint, participating in the exercise intervention reduced the odds of discharge to a rehabilitation facility by 73%. A greater proportion of nonexercisers (54%) went to inpatient rehabilitation facilities compared with exercisers (33%) (Table 5). Our findings conflict with those of one prior study reporting that participants in a preoperative physical therapy group went to rehabilitation facilities more often than controls (19). The potential economic implication of this finding is noteworthy and should be examined in future studies, particularly with the rise in inpatient rehabilitation use (35).

Our study has several limitations. We were only able to recruit 12% of eligible patients. The low recruitment rate was primarily a result of our single-site, community-based design, which suggests that a single intervention site may not be sufficient for a surgical program that attracts patients from a wide geographic area. Requiring participants to travel 3 times per week in all types of weather to a single community location appears to have been an obstacle for many potential participants and several participants who enrolled. Overall, 60% of the participants who dropped out did so for nonmedical reasons. The numbers of dropouts were similar across intervention groups. Individuals who are sedentary and deconditioned often experience greater gains in fitness and emotional status from an exercise program compared with more fit individuals (36). Recruitment rate and attrition could have reduced the efficacy of the exercise by eliminating those patients who could have benefited most from the intervention. The low rate of recruitment from the single hospital location and the level of attrition limit the generalizability of the trial.

Strengths of this study include its randomized, controlled design; blinding of testers to patient group assignment; standardized measures of outcomes; a well-defined community-based exercise intervention; and use of a single hospital to standardize pre- and postoperative procedures. Additionally, participants in the exercise groups had a high attendance rate, which was explained in part by a reported high degree of enjoyment and perceived value of the intervention (unpublished data from program evaluations).

Our findings suggest that men and women with severe osteoarthritis can safely increase lower-extremity muscle strength through participation in a program of cardiovascular fitness, strength training, and flexibility exercise prior to THA or TKA. This improvement in strength is accompanied by improved levels of function in men and women scheduled for THA. Participation in preoperative exercise is strongly associated with a reduced use of inpatient rehabilitation services in both patient groups. Although both THA and TKA populations participated in the same intervention, they did not benefit equally. This finding suggests the need for different approaches for persons with end-stage osteoarthritis in the hips and knees. Additional attention should be placed on testing postoperative interventions for building on preoperative gains in function and fitness (22), adapting the intervention more successfully for the TKA population, and examining the cost effectiveness of exercise for patients undergoing total joint replacement. In addition, future studies of this population should consider the location and convenience of the intervention.