We performed a multicenter, randomized controlled trial comparing the clinical outcome measures of patients who had been randomly assigned to perform ergometer cycling beginning two weeks after total hip arthroplasty or total knee arthroplasty with those of patients who had not. The local ethics committee approved the study protocol. A data and safety monitoring board monitored the study. The study was registered at ClinicalTrials.gov (NCT00951990).
All patients who received a primary unilateral total hip or knee replacement on an elective basis after a diagnosis of osteoarthritis or osteonecrosis were candidates for inclusion in the study. Exclusion criteria were (1) a history of septic arthritis, (2) a hip or knee fracture, (3) an intraoperative complication, (4) revision arthroplasty, (5) rheumatoid arthritis, (6) lower-extremity amputation, (7) a malignant tumor, and (8) an inability to complete the questionnaires because of cognitive or language difficulties. Information about the study was provided to participants on the day of admission to the hospital. Eligible patients were identified by the admitting physicians and were approached about participating in the trial. Patients providing written informed consent were enrolled in the trial. The participating centers are all located in Germany.
Interventions
Patients were randomized into two groups. Those in the ergometer cycling group were introduced to a standard bicycle ergometer (Fig. 1) under the guidance of a physical therapist after the second postoperative week. The ergometer cycling sessions were scheduled to be performed three times a week for at least three weeks. Special attention was given to setting the resistance of the ergometer to a minimum (for example, 30 W), since the aim was not to perform cardiac exercise but to improve muscular coordination, proprioception, and the range of motion. The physical therapists were instructed to set the height of the saddle so that the forefoot reached the pedal with the knee in extension14. The patients in the control group did not perform any ergometer cycling.
All patients participated in a standard postoperative program of daily physiotherapy. This program consisted of range-of-motion activities; exercises to improve muscle strength, venous return, balance, coordination, and gait; and instruction in activities of daily living, including transfers, walking, and negotiation of stairs and uneven surfaces. The patients who had had knee replacement surgery used a continuous passive motion machine on a daily basis after removal of suction drains. All patients were given analgesics according to a standard scheme.
To ensure that the intervention was continued after discharge, written information about the intervention was forwarded to the medical staff involved in the further treatment of the participant.
Randomization
All participants had an equal probability of assignment to the groups. External randomization was achieved by means of computer-generated random-numbers tables in blocks of twenty subjects, stratified by the participating hospital. The randomization tables were stored at the coordinating center. At the time of the preoperative enrollment of the participant, the recruiting hospital faxed the participant's information to the coordinating center. There, a study nurse added the participant to the randomization table in sequential order. After surgery, the result of the randomization was faxed back to the participating hospital. Thus, neither the study participant nor the surgeon knew the randomization result beforehand (allocation concealment)15. The method of generation of the random-numbers tables was unknown to the participating hospitals. As a result of the nature of the intervention, blinding of the study participants and physiotherapists was not possible.
Outcomes
The primary outcome was self-reported physical function three, six, twelve, and twenty-four months postoperatively. Physical function was measured with a validated translated version of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)16,17.
Secondary outcomes consisted of limb-specific stiffness and pain, both measured with the WOMAC; the Physical Component Summary (PCS) score of the thirty-six-item Short-Form General Health Survey (SF-36)18,19; the Lequesne hip and knee score20; and a question about patient satisfaction ("How satisfied are you with the results of your joint replacement surgery?") with very satisfied, somewhat satisfied, somewhat dissatisfied, and very dissatisfied as the possible responses21.
The responses to the WOMAC were recorded on a visual analogue scale with terminal descriptors. The scores for the separate categories were added together and standardized to a score of 0 to 100 points, with higher scores indicating more pain, stiffness, or dysfunction. All patients were asked to answer the questionnaire at the time of hospital admission. During the hospital stay, the study nurse visited the patient to ensure that the questionnaire was filled in completely.
At three, six, twelve, and twenty-four months postoperatively, the participants were mailed a questionnaire and a prepaid return envelope. Participants who did not respond were reminded by mail up to three times at intervals of two weeks. Participants who still did not respond were contacted by telephone to determine the reason. Data were entered into a database at the coordinating center.
To assess whether the significant differences observed in this study represent clinically meaningful change, we compared the observed differences in the primary outcome to previously published thresholds for the "minimal clinically important difference" that were stratified by the severity of the disease22. The minimal clinically important difference has been defined as the "smallest treatment effect that would result in a change in patient management, given its side effects, costs and inconveniences."23 These disease-specific thresholds are 5.3 WOMAC function units for knee osteoarthritis, stratified for a WOMAC function score of =35.3, and 2.6 WOMAC function units for hip osteoarthritis, stratified for a WOMAC function score of =38.222.
Statistical Analysis
For the power analysis, we chose an effect size d of 0.3 and a significance level of 0.05. On the basis of a power of 0.8 to detect a significant difference (p = 0.05, two-sided), 176 patients were required for each study group. Since we expected a loss to follow-up of approximately 25% to 30%, we increased the number of recruited patients accordingly. We prespecified separate analyses for knee and hip arthroplasties.
All analyses were conducted according to the intention-to-treat principle. We compared the two treatment groups at baseline to ensure that they were equivalent with regard to all measured variables. This was done to ensure that our randomization strategy was successful. Continuous follow-up data were initially tested for normal distribution with the Kolmogorov-Smirnov test. Because many of them deviated from such a pattern, the nonparametric Mann-Whitney U test was used to determine differences between the two treatment groups. The chi-square test was used to compare categorical data (patient satisfaction). All p values are two-tailed; no corrections were made for multiple comparisons. Effect sizes24 d as the standardized differences between two groups were calculated as described by Cohen25.
Source of Funding
The study was supported by the Society for Support of Research in and Fighting of Rheumatic Diseases, Bad Bramstedt, registered society; Society for Support of Rehabilitation Research, Schleswig-Holstein, registered society; State Insurance Agency of the Free and Hanseatic City of Hamburg; and German Arthrosis Society, registered society. The funding sources had no involvement in the study design; in the collection, analysis, or interpretation of the data; in the writing of the report; or in the decision to submit the paper for publication.
Participants
In total, 457 patients were candidates for participation in the study from January 1, 2005, to April 30, 2006, and 362 of them underwent randomization. The recruitment process and participant flow are outlined in Figure 2.
There were no significant pretreatment differences between the study groups (Table I), suggesting that the randomization procedure produced well-balanced and comparable groups at baseline.
Overall, 318 patients completed the postal questionnaire at three months, resulting in a follow-up rate of 88%. The follow-up rate remained at 88% at six months and dropped to 85% at twelve months and to 77% at twenty-four months (Fig. 2). There was no significant difference between the patients who did and those who did not respond to the follow-up questionnaire in terms of baseline characteristics.
Ergometer Cycling After Hip Arthroplasty
After the total hip arthroplasties, the ergometer cycling group reported better physical function and less pain and stiffness as measured with the WOMAC (Fig. 3). In addition, there were more "very satisfied" patients in the ergometer cycling group (Fig. 4) than in the control group, and the patients in the ergometer cycling group had better PCS scores on the SF-36 and better Lequesne hip scores (Tables II and III).
There was a significant difference between the groups with respect to the primary outcome, physical function as measured with the WOMAC, at both three months (mean difference = 5.2 points, effect size = 0.33, p = 0.046) and twenty-four months (mean difference = 5.7 points, effect size = 0.37, p = 0.019); the WOMAC pain score at three months (mean difference = 4.8 points, effect size = 0.30, p = 0.049); the WOMAC score for stiffness at twenty-four months (mean difference = 5.2 points, effect size = 0.27, p = 0.047); patient satisfaction at three months (92% compared with 80%, relative risk = 2.75 [95% confidence interval = 1.09 to 6.90], p = 0.027); the PCS score of the SF-36 at both six months (effect size = 0.34, p = 0.011) and twenty-four months (effect size = 0.47, p = 0.004); and the Lequesne hip score at twenty-four months (effect size = 0.32, p = 0.043).
The difference in the primary outcome in the ergometer cycling group exceeded the absolute threshold of 2.6 points for the minimal clinically important difference by a factor of 2.0 at three months and by a factor of 2.2 at twenty-four months.
Ergometer Cycling After Knee Arthroplasty
No significant difference was observed between the study groups with respect to any outcome examined after knee replacement surgery (Tables II and III). In addition, the mean differences did not reach the absolute threshold of 5.3 points for the minimal clinically important difference in the primary outcome.
Adverse Effects
The prevalence of postoperative complications was similar in all treatment groups. Of the patients treated with hip arthroplasty, five who had performed ergometer cycling were readmitted to the hospital within three months (because of dislocation of the hip, other hip problems, fracture of lumbar vertebrae, cardiovascular problems, and overall pain in one patient each) and five who had not performed ergometer cycling were readmitted to the hospital within three months (because of dislocation of the hip, acute coronary syndrome, hematoma revision, appendicitis, and an unknown reason in one each). Of the patients treated with knee arthroplasty, five who had performed ergometer cycling were readmitted to the hospital within three months (two because of liver problems due to analgesics and one each because of a limited range of motion, for an aspiration of the knee joint, and for an unknown reason) and three who had not performed ergometer cycling were readmitted to the hospital within three months (because of a burn injury to the leg, hyperthyroidism, and an unknown reason in one each).
To our knowledge, this is the first study of the effect of ergometer cycling after hip or knee replacement surgery on physical function, pain, joint stiffness, quality of life, and patient satisfaction. These dimensions of health-related quality of life are recommended for use as a rationale for the implementation of the standard of care26.
This randomized study provides strong evidence that ergometer cycling leads to a superior health-related quality of life after hip replacement surgery. The clinical importance of this finding is emphasized by two findings. First, the observed significant changes exceed the published specific thresholds for the minimal clinically important difference22 in the primary outcome measure by a factor of 2.2. Second, the effect sizes for the primary outcome in the present study range from 0.33 at three months postoperatively, to 0.19 at six months, 0.16 at twelve months, and 0.37 at twenty-four months. Therefore, at three and twenty-four months, the observed effect size exceeded the pooled effect size of 0.20 that was obtained from a meta-analysis of randomized placebo-controlled trials evaluating the reduction in functional disability by nonsteroidal anti-inflammatory drugs in patients with osteoarthritis of the knee27.
The percentage of patients who were not very satisfied in the present study compares well with the ranges in other reports28,29. More importantly, ergometer cycling strongly affected patient satisfaction, which, at three months after the hip replacements, was 80% in the group that had not performed ergometer cycling and 92% in the group that had.
According to a review26, age, sex, the operatively treated joint, whether the operation was a primary or revision procedure, comorbidities, and baseline characteristics influence the health-related quality of life after total joint arthroplasty. None of these factors can be influenced by the physician. On the other hand, studies in which patients were randomized to be treated with different types of prostheses failed to demonstrate a notable effect of the prosthetic type on the health-related quality of life26. Also, there appears to be no difference between the effects of inpatient and home-based rehabilitation30. Recently, an association between hospital and surgeon procedure volume and patient-centered outcomes was described31. Ours is the first study of which we are aware to demonstrate a significant and clinically important effect on the health-related quality of life after total joint arthroplasty by a factor that can be influenced by the physician, apart from procedure volume.
We chose three and six months as the appropriate study intervals because the most improvement in postoperative physical health takes place during that time26. We added twelve and twenty-four-month study intervals to standardize our research with that of other authors who have analyzed health-related quality of life after total joint arthroplasty26. Like other studies of health-related quality of life26, the current study was not designed to analyze long-term implant failure. This issue has been extensively addressed previously32.
As this trial was conducted in a multicenter setting, performed at two university hospitals, two rural hospitals, and one municipal hospital, a high degree of external validity is ensured. Furthermore, this setting provided a broader coverage of surgical experience levels than would have been possible with a single-center study. We also used the WOMAC as the primary outcome score, as is recommended in this setting5,33. Our overall postoperative results compare well with those in other studies, and we can also confirm that total hip arthroplasty restores function to a greater extent than does total knee arthroplasty26.
We hypothesize that the beneficial effect of ergometer cycling after total hip replacement is due to improved muscular coordination, proprioception, and range of motion. However, the question of why ergometer cycling did not produce a positive effect after total knee replacement surgery remains. We assume that ergometer cycling results in an improvement in muscular coordination and proprioception after knee replacement as well. However, the cyclic loading of the knee could lead to increased edema of its periarticular tissues, which, in turn, could lead to joint effusion and pain. This observation was reported by several physiotherapists involved in our study. There are three reasons why ergometer cycling, unlike continuous passive motion, could lead to increased edema. First, ergometer cycling is much faster (on the average, 0.5 to one cycle per second) than continuous passive motion (sixty to ninety seconds for one cycle). Second, ergometer cycling is active whereas continuous passive motion is passive. Finally, the knee is above heart level during continuous passive motion, whereas it is below heart level during ergometer cycling. Early after knee replacement surgery, this mechanism could offset the beneficial effects of muscular coordination and proprioception attributable to ergometer cycling.
Although this study had several strengths, it also had limitations. Although the follow-up rate was 88% at both three and six months and 85% at twelve months, it dropped to 77% at twenty-four months. Since the results of the twenty-four-month follow-up are quite similar to the results at the three earlier times, it appears unlikely that a more complete twenty-four-month follow-up would have altered the study results. It should also be noted that, although the eligibility criteria were fairly broad, the trial was restricted to patients undergoing unilateral primary total joint replacement. Therefore, our results should not be generalized to patients undergoing revision or bilateral total joint replacement.
Using an intention-to-treat analysis underestimates the real benefit of the treatment24 if patients are not fully compliant. Therefore, the positive findings of this study should not have been compromised by noncompliance.
Our results suggest that ergometer cycling after total hip replacement results in superior health-related quality of life and patient satisfaction. This effect is clinically meaningful because the significant changes exceed the thresholds for the minimal clinically important improvement in the primary outcome22 and the observed effect size is greater than the published effect sizes of treatment of osteoarthritis of the knee with nonsteroidal anti-inflammatory drugs27. Therefore, the results of this study could change clinical practice, especially since the intervention has not been previously studied, is easy to administer, and improves quality of life at minimal costs.