Efficacy of Continuous Passive Motion Following Total Knee Arthroplasty: A Metaanalysis
LUCIE BROSSEAU, SARAH MILNE, GEORGE WELLS, PETER TUGWELL, VIVIAN ROBINSON, LYNN CASIMIRO, LUCIE PELLAND, MARIE-JOSÉE NOEL, JENNIFER DAVIS, and HUGO DROUIN
Methods. This metaanalysis used the methodology proposed by the Cochrane Collaboration.
Results. This review of 14 studies (952 patients) found significant improvements in active knee flexion and analgesic use 2 weeks postoperatively with the use of CPM and physiotherapy (PT) compared to PT alone. In addition, length of hospital stay and need for knee manipulations were significantly decreased in the CPM group. Not enough data were available to compare the degree of knee flexion applied or number of hours of application of CPM. However, significant results were not found for other comparisons such as short term CPM application versus longterm CPM application and wide treatment range versus small treatment range for the outcomes of active knee flexion, passive knee flexion and extension, presence of a fixed flexion deformity, use of analgesic, or total knee range of motion.
Conclusion. CPM combined with PT may offer beneficial results for patients post-TKA. However, the potential benefits will need to be carefully weighed against the inconvenience and expense of CPM. More research is necessary to assess the differences in effectiveness with different characteristics of application such as total duration of treatment and intensity of CPM interventions. (J Rheumatol 2004;31:2251-64)
Key Indexing Terms:
From the School of Rehabilitation Sciences, University of Ottawa; the Institute of Population Health, University of Ottawa; and the Department of Epidemiology and Community Health, University of Ottawa, Ottawa, Ontario, Canada.
Supported by a research grant from The Arthritis Society, and three grants for salary support from the Ontario Ministry of Health and Long-Term Care (Health Research Personnel Development Program), University of Ottawa (University Research Chair), and the Ministry of Human Resources, Government of Canada (Summer Student Program).
L. Brosseau, PhD; S. Milne, BSc (Physiotherapy), MSc Candidate in Epidemiology; G. Wells, PhD; V. Robinson, MSc (Exercise Science); L. Casimiro, MA (Physical Activity); L. Pelland, PhD (Rehabilitation); P. Tugwell, MD, MSc (Epidemiology); M-J. Noel, BSc (Physiotherapy); J. Davis, BSc (Physiotherapy); H. Drouin, BSc (Physiotherapy).
Address reprint requests to Dr. L. Brosseau, Faculty of Health Sciences, University of Ottawa, 451 Smyth Road, Ottawa, Ontario K1H 8M5. E-mail: firstname.lastname@example.org
Submitted April 26, 2004; revision accepted May 21, 2004.
Knee arthroplasties (KA) are surgical procedures that have become more common in the last few decades in part due to the aging population. Rheumatoid arthritis (RA) and osteoarthritis (OA) are the 2 main reasons for KA. Historically, immobilization has been the postoperative treatment of choice for many orthopedic surgeries1. However, over the years clinical trials have shown that immobilization of a specific joint following surgery can have detrimental effects on collagen tissue healing, articular cartilage nutrition, and joint stiffness2. This insight into the disabling results of postoperative immobility has stimulated an interest in early mobilization1.
Recent studies show that early movement is beneficial for the recovery of range of motion (ROM) in an immobilized joint3. Adequate ROM of the knee, particularly in flexion, is important for mobility and activities of daily living (ADL) in patients who have undergone KA4. If patients have improved ROM, their ability to perform functional tasks increases4,5. As a result of these findings, early postsurgical mobilization has become standard practice.
Continuous passive motion (CPM) is an external motorized procedure that enables a joint to move passively through a predetermined range of motion6. It is one technique whereby patients can achieve early postsurgical mobility. CPM was first introduced in the 1970s by Salter, who initiated trials using rabbits and demonstrated enhanced healing of cartilage using CPM7. Passive exercise such as CPM is thought to be helpful in maintaining ROM and reducing edema, whereas active exercise promotes muscle strengthening8.
Studies on CPM have produced conflicting results. CPM has been shown to have positive effects on soft tissue healing, swelling, hemarthrosis, and joint function2. During the normal healing process, collagen fibers grow randomly, producing resistance to free movement7. CPM is proposed to work at the cellular level by decreasing random fiber growth and diminishing postoperative scar formation7. It has also been found that the use of CPM can decrease the incidence of postoperative deep vein thrombosis and thromboembolitic disease9. Several studies have claimed that the use of CPM can significantly increase the amount of knee flexion by the time of discharge from hospital4,10,11. Other research has reported that CPM decreases the rate of manipulation under anesthesia post-KA12. However, a retrospective study comparing CPM and physiotherapy (PT) to PT alone found no significant difference in knee ROM between the CPM and the non-CPM treatment groups13. Studying the use of CPM, some disadvantages have also been noted: (1) patients remain in bed while the machine is utilized; (2) studies show early knee ROM improvements require up to 20 hours of CPM application daily, which is time consuming and costly; (3) patients require technical support from nursing staff to operate the machines; (4) increased costs are associated with safe operation of the units and regular maintenance14,15.
Despite the recorded benefits of CPM on knee flexion post-KA, it is clear that consensus has not been attained about the longterm efficacy of the procedure16. Although controversial, CPM has been used by many surgeons as part of a standard postoperative management of patients having undergone surgical KA4,14. It has been stated that the widely conflicting findings are due to inconsistencies in the variables being studied7. Little information exists to enable the clinician to select optimal CPM parameters, such as the most appropriate number of degrees per day to advance the CPM device or the optimal daily treatment duration1,7. Coutts, et al reported on the effectiveness of CPM following KA and suggested that 20 hours of CPM daily increased ROM and decreased edema and effusion17. Similarly, Davis reported increased ROM by using the CPM following a 3-day delay in the initiation of treatment18. However, Young and Kroll concluded that the CPM did not offer additional benefits from conventional PT alone19, while others only used CPM for 6 hours a day and obtained positive results20.
The objective of this metaanalysis was to determine the effectiveness of CPM following knee arthroplasty. CPM is compared to standard PT treatments conducted on patients after a total KA. Standard PT treatment, as defined by this analysis, consists of any combination of the following interventions: ROM exercises (ROM), muscle strengthening exercises (isometric, dynamic), functional exercises, gait training, immobilization, and ice. The outcome measures of interest for this metaanalysis were active and passive knee ROM, length of hospital stay, pain, swelling, fixed flexion deformity, and quadriceps strength at end of treatment and during followup.
MATERIALS AND METHODS
This metaanalysis used the methodology proposed by the Cochrane Collaboration21.
Literature identification. The literature was searched up to and including December 2003 according to the sensitive search strategy outlined by the Cochrane Collaboration for randomized controlled trials (RCT)22, with modifications proposed by Haynes, et al23. Additional terms for study design were used to identify observational studies, including: case-control, cohort, comparative study, clinical trial. Medline, Embase, Healthstar, Sports Discus, CINAHL, the Cochrane Controlled Trials Register, the PEDro database, the specialized registry of the Cochrane musculoskeletal group, and the Cochrane field of physical and related therapies were searched using a keyword and text word search strategy (Appendix). As well, reference lists of included trials were searched and content experts were contacted for additional studies. The details of the search strategy are given in the Appendix.
Eligibility criteria. The titles and abstracts of the trials identified by the search strategy were examined by 2 independent reviewers (VR, LB) to determine whether these selected trials met the inclusion criteria. All trials classified as relevant by at least one of the reviewers were retrieved. The retrieved articles were reexamined to ensure they met the inclusion criteria.
Only trials meeting the following conditions were included: subjects 18 years of age or older and having a presurgery diagnosis of degenerative joint disease, intervention and control groups of 5 or more individuals each, and measuring rehabilitative outcomes. Both the experimental and control groups received PT. In addition to the PT intervention, the experimental group received CPM.
According to an a priori protocol, all comparative controlled trials, including RCT, controlled clinical trials without randomization (CCT), case-control, and cohort studies were included. Trials that used the same patients as their own control were not accepted. The results were graded according to the strength of the study design. Both English and French RCT were considered. Peer-reviewed abstracts were accepted.
Acceptable interventions included any form of fitness exercise. Placebo, untreated, or active interventions were all acceptable control groups.
The large number of studies in this review measured a variety of outcomes. The outcomes were as follows: active and passive knee ROM; length of hospital stay; pain; swelling; and quadriceps strength.
Data extraction. Two independent reviewers (HD, JD) examined the titles and abstracts of the trials identified by the search strategy to select trials that met the inclusion criteria. All trials classified as relevant by at least one of the reviewers were retrieved. The retrieved articles were reexamined to ensure they met the inclusion criteria.
The results of the individual trials were extracted from each of the included trials using predetermined extraction forms by 2 independent reviewers (HD, JD). The data were cross-checked by a third reviewer (LB). The extraction forms were developed and pilot-tested based on other forms used by the Cochrane musculoskeletal review group. Data of interest were grouped in either subject characteristics (age, sex, diagnosis, etc.) or CPM therapeutic application (hours/day, increments in degrees/day, etc.). The outcome measures collected were length of hospital stay, ROM (passive and active knee flexion and extension), extension lag, fixed flexion deformity, pain (on visual analog scale, VAS), pain medication intake, swelling, and quadriceps strength. These outcomes were considered pertinent to PT intervention by 3 of the authors (HD, JD, MJN). The final data values were based on consensus of the 2 reviewers.
Quality assessment. The quality of each study was assessed by 2 independent reviewers. Quality assessment examined the extent to which the RCT design, data collection, and statistical analysis minimized or avoided biases in its treatment comparisons24. The Jadad scale was used to perform the quality assessment25,26. The scale includes items pertaining to description of randomization, appropriateness of blinding, dropouts and withdrawals, and followup. Differences in scoring were resolved by consensus. A third reviewer (LB) was consulted when necessary. The quality assessment was pilot-tested on 4 unrelated articles prior to data extraction.
Statistical analysis. Results were analyzed to compare CPM combined with PT versus PT alone. Results on individual treatment techniques were analyzed separately. Data relative to the outcomes from each trial were pooled to arrive at an overall estimate to determine the effectiveness of each procedure. Where possible, the analyses were based on intention-to-treat from the individual trials. In cases where trials reported outcomes with graphs, the mean scores and standard deviations were estimated from the graphs. Subgroup analyses were attempted to determine the effects of the method administration, methodological quality, and the intervention duration on outcomes. For continuous data, results were presented as weighted mean differences (WMD), where the difference between the treated and control groups was weighted by the inverse of the variance. For dichotomous outcomes, results were presented as an odds ratio or relative risk (RR). Standardized mean differences (SMD) were used when different scales were used to measure the same concept (e.g., pain). SMD were calculated by dividing the difference between treated and control means by the pooled estimate of the baseline standard deviation. Fixed effects models were used throughout, unless statistical heterogeneity was proved by the Cochrane Q test (p < 0.05). Where heterogeneity was significant, random effects models were used. For outcomes where it is desirable to have a lower score (e.g., pain), a negative value indicates a positive effect of the intervention procedure. For outcomes where a larger value is desirable (e.g., range of motion), a positive value indicates benefits.
Summary of the trials
The literature search and hand-searching identified 178 articles. Of the 178 articles, 58 trials were screened for relevance and inclusion into this metaanalysis. Of the 58 screened, 14 were in accord with the inclusion criteria: Chen, et al 20004, Chiarello, et al 19977, Colwell and Morris 199227, Harms and Engstrom 19911, Johnson 199028, Kumar, et al 199629, MacDonald, et al 200014, May, et al 199916, McInnes, et al 19925, Montgomery and Eliasson 199630, Nielsen, et al 198831, Pope, et al 199732, Vince, et al 198733 and Walker, et al 199134. For inclusion, subjects were 18 years of age or older and were hospitalized following knee arthroplasty procedures. Presurgery diagnosis for all subjects was classified as degenerative joint disease, OA, or RA. The length of treatment in individual studies varied from 18 hours to 2 weeks. Daily CPM treatment time varied from 5 hours daily to 20 hours daily. All 14 articles included both male and female patients. A total of 952 patients were included for analysis. A summary of all trials is given in Table 1. Trials were excluded for several reasons including: no clinical outcomes of interest; not a clinical trial; subjects did not undergo knee arthroplasty; no variance reported on outcomes. Excluded trials and specific reasons for exclusion are tabulated in Table 2.
The primary diagnosis was degenerative joint disease. OA was present in over 89% of cases and RA represented no more than 8% of the subject diagnoses.
Methodological quality of the studies. The median methodological quality was 2 out of a maximum of 5 points. The quality scores ranged from 1 to 3. Of the 14 RCT, 3 scored full points for randomization, Harms and Engstrom 19911, Kumar, et al 199629, and MacDonald, et al 200014. No studies were described as double blinded and 3 failed to provide a description of dropouts: Chen, et al 20004, Harms and Engstrom 19911, and Johnson 199028. For the main comparison of CPM combined with PT versus PT alone, the median methodological quality was 2. Eight articles obtained a score of 2: Chiarello, et al 19977, Colwell and Morris 199227, Harms and Engstrom 19911, McInnes, et al 19925, Montgomery and Eliasson 199630, Nielsen, et al 198831, Vince, et al 198733, and Walker, et al 199134, while one obtained a score of 3, Kumar, et al 199629. Eight articles provided a description of withdrawals and dropouts: Chiarello, et al 19977, Colwell and Morris 199227, Kumar, et al 199629, McInnes, et al 19925, Montgomery and Eliasson 199630, Nielsen, et al 198831, Vince, et al 198733 and Walker, et al 199134. Only one article, Harms and Engstrom 1991, received full points for randomization1.
Pooled analysis. Pooled analyses were possible for the comparison of CPM combined with PT versus PT alone at the end of treatment (approximately 2 weeks). Nine trials were included in this comparison (Chiarello, et al 19977, Colwell and Morris 199227, Harms and Engstrom 19911, Kumar, et al 199629, McInnes, et al 19925, Montgomery and Eliasson 199630, Nielsen, et al 198831, Vince, et al 198733, Walker, et al 199134). Treatment was initiated on the first postoperative day for all trials except one in which CPM was started on the second postoperative day. For the outcome of active knee flexion, 4 studies were included in the pooled analysis, for a total of 286 patients (Chiarello, et al 19977, Harms and Engstrom 19911, McInnes, et al 19925, Montgomery and Eliasson 199630). Overall, CPM combined with PT significantly increased active knee flexion at 2 weeks post-KA (WMD 4.30, 95% CI 1.96 to 6.63; Figure 1). In addition, a clinically important benefit was found for active knee flexion at 3 days (relative difference 23%; Table 3), 2 weeks (relative difference 22%; Table 3), and one week of followup (relative difference 25%; Table 3). Further, patients receiving CPM achieved 90° of knee flexion on average 4.7 days faster than patients receiving PT alone (4.7 days difference, Vince, et al 198733).
Figure 1. Statistical significance determined by a weighted mean difference and confidence interval of 95% for active ROM. tx: treatment.
Statistically significant results were also obtained for length of hospital stay. Six studies were included in the analysis with a total of 382 patients (Colwell and Morris 199227, Harms and Engstrom 19911, Kumar, et al 199629, McInnes, et al 19925, Montgomery and Eliasson 199630, Walker, et al 199134). The treatment groups receiving CPM and PT were found to have a significantly shorter time to discharge (WMD –0.69 days, 95% CI –1.35 to –0.03; Figure 2) than those receiving physiotherapy alone. Discharge criteria varied among trials, and only 3 trials (Colwell and Morris 199227, Harms and Engstrom 19911, Kumar, et al 199629) actually specified their criteria. Although length of stay was found to produce a statistically significant result, no clinically important benefit was found.
Figure 2. Statistical significance determined by a weighted mean difference and confidence interval of 95% for length of stay. tx: treatment.
Positive results were obtained for the number of patients requiring manipulation post-KA. According to data pooled from 3 trials (Harms and Engstrom 19911, McInnes, et al 19925, Vince, et al 198733), subjects in the CPM group had a significantly lower incidence of post-KA manipulation (RR 0.12, 95% CI 0.03 to 0.53; Figure 3). However, no clinically important benefit was shown for the number of patients needing postoperative manipulation (5% to 18% relative difference, Table 4). All 3 trials began CPM treatment within 24 hours.
Figure 3. Statistical significance determined by a risk difference calculation for an odds ratio outcome: postoperative manipulation. tx: treatment.
Statistically significant results were also obtained for the outcomes of pain medication intake (WMD –4.18 mg, 95% CI –7.86 to –0.49) and knee swelling (WMD pooled –1.79, 95% CI –2.05 to –1.53)5,30. However, clinically important benefit was not achieved for any of the outcomes (analgesic use, 1–52 mg difference; knee swelling, 2–5% difference), and significant problems with heterogeneity of data existed for the comparison of knee swelling.
Pooled analysis revealed that CPM did not significantly improve passive knee flexion at end of treatment or at 6 weeks, 3 months, or 6 months of followup (Figure 4). For the outcome of passive knee extension, 3 trials were pooled (Chiarello, et al 19977, Kumar, et al 199629, McInnes, et al 19925) and were homogeneous for comparison. Two trials (Chiarello, et al 19977, McInnes, et al 19925) measured fixed flexion deformity, while the third measured passive knee extension (Kumar, et al 199629). Passive extension and fixed flexion deformity were considered to be the same outcome as they both represent the limit of available knee extension. Results for passive knee extension at end of treatment were not found to be statistically significant (WMD 0.49°, 95% CI –0.99 to 1.97). For the outcome of active knee extension (Chiarello, et al 19977, McInnes, et al 19925) only 2 trials could be pooled for analysis with 113 patients included. Neither result was found to be statistically significant (WMD –1.06°, 95% CI –7.53 to 5.40; Figure 5). However, a clinically important benefit was shown for passive knee extension (18% to 95% relative difference27,30) and flexion deformity (23% relative difference, Pope, et al 199732) despite the statistical insignificance.
Figure 4. Statistical significance determined by a weighted mean difference and confidence interval of 95% for passive knee flexion. tx: treatment.
Figure 5. Statistical significance determined by a weighted mean difference and confidence interval of 95% for passive knee extension. tx: treatment.
Results from individual studies. Pooled analyses were not possible for several aspects of this review. CPM combined with PT versus PT alone produced statistically insignificant results for the outcome of pain as measured by a VAS at mid-treatment one week (Montgomery and Eliasson 199630) and end of treatment 2 weeks (McInnes, et al 19925) or as measured by the proportion of patients with pain at the end of treatment. In addition, no clinically important benefit was shown for pain at end of treatment as measured by a VAS (2% difference, McInnes, et al 19925).
Additional comparisons of CPM combined with PT versus PT alone found varying results. At the end of treatment, no statistically significant differences were found for the outcomes of number of patients with ROM improvement (Nielsen, et al 198831), presence of an extension lag (degrees) (Nielsen, et al 198831), knee circumference (Chen, et al 20004), or quadriceps strength (WMD 1.60, 95% CI –1.88 to 5.08) (McInnes, et al 19925). For the outcome of extension/flexion deformity, statistically significant results in favor of CPM combined with PT were found at mid-treatment (WMD –1.42, 95% CI –2.69 to –0.15) and at the end of treatment (WMD –3.80, 95% CI –6.04 to –1.56) (Harms and Engstrom 19911, Figure 6). However, these results were not significant after one week or one year of followup (Pope, et al 199732, Figure 6). In addition, no clinically important benefit was shown for global extension/flexion deformity (5% relative difference, Harms and Engstrom 19911).
Figure 6. Statistical significance determined by a weighted mean difference and confidence interval of 95% for flexion and extension deformity. tx: treatment, FU: followup.
A statistically significant benefit was also not demonstrated for the outcome of function as measured by a 0 to 70 scale (Pope, et al 199732), or using the Knee Society Score (MacDonald, et al 200014) at one year followup, or by the Health Assessment Questionnaire at 6 months followup (McInnes, et al 19925). Meanwhile, statistically significant results in favor of the treatment group were found on time to achieve 90° flexion (days) at the end of 2 weeks of treatment (WMD –4.70, 95% CI –7.37 to –2.03, Vince, et al 198733).
For the comparison of CPM combined with PT versus splinting combined with PT, one trial (Johnson 199028) was included; 102 patients were included for comparison. Outcomes were assessed at end of treatment (one week), and followup (2 weeks, 6 weeks, 3 months, 6 months, and one year). Measuring ROM into knee flexion (Figure 7), statistically significant results favoring the CPM group were found at end of treatment (WMD –16.00°, 95% CI 10.52 to 21.48) and followup at 2 weeks (WMD –10.00°, 95% CI 4.35 to 15.65), 6 weeks (WMD –8.00°, 95% CI 1.32 to 14.68), 6 months (WMD –7.00°, 95% CI 0.60 to 13.40), and one year (WMD –9.00°, 95% CI 7.63 to 10.37). At 3 months, results favored the CPM group (WMD –5.00°, 95% CI –0.64 to 10.64), but were found not to be statistically significant. Comparing CPM and PT versus splinting for flexion deformity and extension lag deformity, no statistically significant results were found.
Figure 7. Statistical significance determined by a weighted mean difference and confidence interval of 95% for knee flexion. tx: treatment.
For the comparison of short-time CPM application versus long-time CPM application (results not shown), no statistically significant results were found for the outcomes of active knee flexion, presence of a fixed flexion deformity, analgesic use, or total knee ROM at end of treatment (Chiarello, et al 19977). However, only 20 patients were included in this comparison. As well, no clinically important benefits were found for flexion ROM in comparing short-time versus long-time CPM application.
For small-range versus big-range, no statistically significant results were found for active ROM into knee flexion, total ROM, flexion deformity, or function at one week and one year followup (Pope, et al 199732). However, statistically significant results were obtained for analgesic use at one week followup favoring a larger range (WMD –8.90, 95% CI –15.36 to –2.44, Pope, et al 199732, Figure 8).
Figure 8. Statistical significance determined by a weighted mean difference and confidence interval of 95% for knee pain. tx: treatment.
In addition, no statistically significant results were found comparing low-range CPM application to high-range CPM application for the outcomes of analgesic use, length of hospital stay, or knee range of motion at end of treatment or followup at 6 weeks and one year (MacDonald, et al 200014, Pope, et al 199732). In addition, results for function as measured using the Knee Society Score (0–200: WMD 1.0, 95% CI –8.05 to 10.05, MacDonald, et al 200014) were not significant. Despite the lack of statistical significance, a clinically important benefit was demonstrated for pain medication intake (16 mg difference, Table 5) favoring the high-range CPM group. However, no clinically important benefit was shown for knee flexion ROM (relative difference 2%) or length of hospital stay.
For the comparison of CPM versus lower limb mobility training combined with PT, no statistically significant results were found on the outcomes of pain, active knee flexion, active knee extension, passive knee extension, and gait speed (results not shown)16. However, only 19 patients were included for comparison. Results were measured at the end of treatment (one month).
The results from this metaanalysis suggest that CPM combined with PT interventions is effective at increasing active knee flexion 2 weeks post-knee arthroplasty relative to physiotherapy intervention alone1,5,7,30. However, the clinical significance of an additional 4° of knee flexion can be questioned. Adequate ROM of the knee, particularly in flexion, is important for performing mobility tasks such as walking, transfers, and activities of daily living. A minimum of 65° of knee flexion is required in the swing phase of normal gait, 90° of flexion is required to descend stairs, and at least 105° is required to rise from a toilet or low chair12. Due to its functional importance, knee ROM was a primary outcome. Results from this metaanalysis suggest, however, that although CPM may produce small changes in active knee flexion range in the short term it does not result in additional range over the long term, one or 2 years post-surgery.
Statistically significant results were also found for the outcome of length of hospital stay. This metaanalysis suggests that patients who receive CPM in addition to PT are discharged home from the hospital earlier than those who receive PT treatment alone. In the current age of hospital cutbacks and limited resources, even small reductions in length of hospital stay after a surgical procedure may be important. Length of hospital stay was also a primary outcome for this metaanalysis.
CPM in addition to physiotherapy intervention also reduced the number of postoperative knee manipulations required relative to PT alone. It has been suggested in related research that the greatest benefit of CPM appears to be its ability to decrease the number of knee manipulations5. Manipulation is used to facilitate the postoperative rehabilitation program for patients with painful, limited ROM of the knee12. However, manipulation is a painful process and an added complication to the initial surgery. Therefore, any reduction in the number of procedures required is beneficial to both the surgeon and the patient; however, the absolute reduction in risk will depend on the baseline risk.
No statistically significant difference was found for the outcome of active or passive knee extension. However, this is not surprising, as CPM was designed to improve knee flexion7. It has been suggested that an active hold at the point of maximum extension, activating the quadriceps, would be necessary while utilizing the CPM machine in order to enhance active extension7.
Information on the outcome of pain is limited. It has been suggested that rhythmic joint movement inhibits the pain-spasm reflex48. However, these results were not supported by the limited data available in this metaanalysis.
Information biases were identified in several trials28,30. Heterogeneity of results was also a problem for several outcomes. In several trials the ROM measurements were not specified as being active or passive. These measures need to be performed and reported in a standardized manner to allow appropriate comparisons. Heterogeneity or variability may have been introduced in the outcomes measured, the type of implants used (cemented vs uncemented), and the patient diagnosis. Under ideal circumstances, interventions are to be delivered in a blinded fashion. However, in many instances it is impossible to blind patients or clinicians when using physical interventions. This, however, could introduce bias into the study. Due to the limited sample size, subgroup analyses could not be performed based on low methodological quality (2/5). This could introduce bias into the results.
Protocols were another area in which bias may have been introduced. Protocols differed from trial to trial and in some cases, treatment parameters were not reported adequately. For the main comparison of CPM combined with PT versus PT alone, 5 studies (Chiarello, et al 19977, Harms and Engstrom 19911, McInnes, et al 19925, Nielsen, et al 198831, Vince, et al 198733) provided identical PT treatment to the experimental and control groups, while 4 studies (Colwell and Morris 199227, Kumar, et al 199629, Montgomery and Eliasson 199630, Walker, et al 199134) were found to have provided one group additional PT. In addition, there is no consensus on the clinical application characteristics such as selected ROM for treatment, treatment duration, or intensity of application. Several studies (Chiarello, et al 19977, MacDonald, et al 200014, Pope, et al 199732) attempted to compare CPM duration and treatment ROM. However, data could not be pooled and the sample size was low for individual trials. Finally, PT interventions were not uniform, since all studies provided a different PT intervention. It is not clear what effect these differences may have on the reported efficacy of CPM.
Another important factor that may influence results is the use of preoperative exercises as part of the rehabilitation protocol for KA. There is no consensus regarding the effect of preoperative PT in total KA; however, it has been suggested that the decrease in muscle strength observed post-surgery may be reduced through implementation of a preoperative PT regime49.
The failure to use validated outcome measures is also a limitation of this analysis. No functional activities (sit to stand, supine to sit, ambulation, stair climbing, ambulation velocity, functional status) were assessed using validated outcome measure scales in any of the analyzed studies. Since the focus of PT treatment is aimed increasingly at functional activities, the outcome measures used to assess CPM should reflect this situation.
It has been stated by many orthopedic specialists that good surgical management and postoperative outcome of a patient are dependent upon proper rehabilitative input8. In other words, exercise is a key element in the success rate of hip and knee arthroplasty8. Whether it is the use of a CPM machine or a physical therapist, passive ROM exercises are suggested to begin the third postoperative day8. This act of passive ROM maintains extension/flexion activity in the knee and helps the muscles reeducate themselves8. Since full knee extension is one of the most difficult tasks for the patients to achieve, initiating exercise in extension and assisting knee flexion is more easily accomplished passively than actively attempting to achieve full extension8. The use of a CPM machine to achieve passive ROM is appropriate. During the second postoperative week, isometric exercises begin to strengthen both quadriceps8. These exercises are followed by gait training8. Here the benefits of CPM may come into play, since the importance of early adequate knee ROM determines the rate at which the patient achieves normal gait. It is clear that postoperative exercise plays a vital role in the rehabilitation rate of the patient. As long as the benefits of CPM outweigh the costs, it should be considered a viable rehabilitative intervention.
CPM combined with conventional PT may be utilized to produce small increases in active knee flexion ROM, to decrease length of hospital stay, and to reduce the risk of manipulation following total knee arthroplasty. These potential benefits will need to be carefully weighed against the inconvenience and expense of CPM. Further studies are required to assess the effectiveness of CPM by altering treatment variables. For example, modifying the total duration of treatment and the intensity of CPM interventions, and using different types of patients at various disease states would aid in defining the most efficacious CPM treatment regime. In addition, the effect of CPM combined with and compared to various other physiotherapy interventions should be studied further.
The authors are indebted to Catherine Lamothe for her technical support and her help in extraction of data. Special thanks to Jessie McGowan (MLIS), Director of the Ottawa Hospital Library, for her consultation on the search strategy and Maria Judd for her helpful feedback on the final draft.
Search strategy for identification of studies
1 exp arthritis, rheumatoid/
2 arthritis, juvenile rheumatoid/
3 1 not 2
4 (rheumat$ adj arthrit$).tw.
5 3 or 4
6 osteoarthritis, knee/
10 7 or 8
11 9 and 10
12 6 or 11
13 5 or 12
14 arthroplasty, replacement, knee/
15 knee prosthesis/
16 total knee.tw.
18 exp physical therapy/
19 motion therapy, continuous passive/
20 continuous passive motion.tw.
21 gait therapy.tw.
22 exercise therapy.tw.
23 (ice or cold).tw.
24 therapeutic exercise/
25 "heat/cold application"/
27 17 and 26
30 (compare or comparative).tw.
32 exp clinical trials/
33 comparative studies/
34 exp prospective studies/
38 cross sectional studies/
39 exp case control studies/
41 27 and 40
42 27 not 41
2. McCarthy MR, O'Donoghue PC, Yates CK, Yates-McCarthy JL. The clinical use of continuous passive motion in physical therapy. J Sport Phys Ther 1992;15:132-40.
6. Sheppard, MS, Westlake SM, McQuarrie A. Continuous passive motion where are we now? Physiotherapy Canada 1995;47:36-9.
7. Chiarello CM, Gundersen MS, O'Halloran T. The effect of continuous passive motion duration and increment on range of motion in total knee arthroplasty patients. J Sport Phys Ther 1997;25:119-27.
9. Lynch JA, Baker PL, Polly RE, McCoy MT, Sund RN, Roundybush D. Continuous passive motion: A prophylaxis for deep vein thrombosis following total knee replacement. Stanmore, UK: British Association for Knee Surgery, Royal Orthopaedic Hospital; 1986.
11. Schnebel BE, Evans JP, Flinn D. The use of a passive motion machine. Am J Knee Surg 1989;2:131-6.
12. Fox JL, Poss R. The role of manipulation following total knee replacement. J Bone Joint Surg Am 1981;63 Suppl:357-62.
14. MacDonald SJ, Bourne RB, Rorabeck CH, McCalden RW, Kramer J, Vaz M. Prospective randomized clinical trial of continuous passive motion after total knee arthroplasty. Clin Orthop 2000;380:30-5. [MEDLINE]
15. Beaupre LA, Davies DM, Jones CA, Cinats JG. Exercise combined with continuous passive motion or slider board therapy compared with exercise only: a randomized controlled trial of patients following total knee arthroplasty. Phys Ther 2001;81:1029-37. [MEDLINE]
16. May AL, Busse W, Zayac D, Withridge MR. Comparison of continuous passive motion machines and lower limb mobility boards in the rehabilitation of patients with total knee arthroplasty. Can J Rehabil 1999;12:257-63.
17. Coutts RD, Toth C, Kaita JH. The role of continuous passive motion in the rehabilitation of the total knee patient. In: Hungerford DS, Krackow K, Kenna RV, editors. Total knee arthroplasty: a comprehensive approach. Baltimore: Williams & Wilkins; 1984:126-32.
18. Davis D. Continuous passive motion for total knee arthroplasty [abstract]. Phys Ther 1984;64:709.
19. Young JS, Kroll MA. Continuous passive motion compared to active assisted range of motion. Phys Ther 1984;64:721.
21. Mulrow CD, Oxman A, editors. How to conduct a Cochrane Systematic Review. San Antonio: Cochrane Collaboration, version 3.0.2., 1997.
23. Haynes RB, Wilczynski N, McKibbon KA, Walker CJ, Sinclair JC. Developing optimal search strategies for detecting clinically sound studies in Medline. J Am Med Informat Assoc 1994;1:447-58. [MEDLINE]
24. Moher D, Jadad A, Nichol G, Penman M, Tugwell P, Walsh S. Assessing the quality of randomized controlled trials: an annotated bibliography of scales and checklists. Control Clin Trials 1995;16:62-73. [MEDLINE]
30. Montgomery F, Eliasson M. Continuous passive motion compared to active physical therapy after total knee arthroplasty: Similar hospitalization times in a randomized study of 68 patients. Acta Orthop Scand 1996;67:7-9. [MEDLINE]
34. Walker RH, Morris BA, Angulo DL, Schneider J, Colwell CW. Postoperative use of continuous passive motion, transcutaneous electrical nerve stimulation, and continuous cooling pad following total knee arthroplasty (Phase 1). J Arthroplasty 1991;6:151-6. [MEDLINE]
35. Aubriot JH, Guincestre JY, Grandbastien B. Rehabilitation following total knee arthroplasty. Role of passive motion. A randomized study about 120 subjects [French]. Rev Chir Orthop Reparatrice Appar Mot 1993;79:586-90. [MEDLINE]
43. Rasti Z, Olsen O. Continuous passive motion for rehabilitation after total knee arthroplasty in patients with osteoarthritis. In: The Cochrane Library, Issue 4, 2001.
46. Worland RL, Arredondo J, Angles F, Jimenez FL, Jessup DE. Home continuous passive motion machine versus professional physical therapy following total knee replacement. J Arthroplasty 1998;13:784-7. [MEDLINE]
50. Leonard GM, Tremblay LE, Chabot M, Larivière J, Papadopoulos P. The effects of early continuous passive motion after total knee arthroplasty. Physiother Canada 2004; (in press).