Secondary Logo

Journal Logo


2018 John N. Insall Award: Recovery of Knee Flexion With Unsupervised Home Exercise Is Not Inferior to Outpatient Physical Therapy After TKA: A Randomized Trial

Fleischman, Andrew N. MD; Crizer, Meredith P. BS; Tarabichi, Majd MD; Smith, Shelby BS; Rothman, Richard H. MD, PhD; Lonner, Jess H. MD; Chen, Antonia F. MD, MBA

Author Information
Clinical Orthopaedics and Related Research: January 2019 - Volume 477 - Issue 1 - p 60-69
doi: 10.1097/CORR.0000000000000561
  • Free



For many patients with end-stage knee arthritis, TKA improves function, pain, and quality of life. Most patients are able to return to their activities of daily living (ADLs), including ambulating without assistive devices, navigating stairs safely, walking without major limitations, and returning to work and some recreational sports [1]. Historically, most patients who undergo TKA participate in supervised physical therapy (PT) sessions postoperatively at rehabilitation or outpatient facilities to achieve ROM. These programs are designed to help patients achieve the flexion needed for various daily activities, including 67° for the swing phase of gait, 83° for climbing stairs, 90° for descending stairs, and 93° for standing up from a chair [6, 8, 11].

Although many orthopaedic centers still routinely refer patients for formal supervised PT after TKA [2], there has been growing interest in providing effective, home-based rehabilitation without the need for professional supervision. This likely reflects uncertainty about the benefits of costly ancillary services such as supervised PT, which accounted for USD 468 million in Medicare costs after primary TKA in 2009 as well as patient preference for performing rehabilitation in the comfort and convenience of their own homes [14]. Several randomized controlled trials (RCTs) have demonstrated the potential for similar outcomes with monitored home exercise compared with formal outpatient PT [5, 9, 10, 12, 13]. However, the effectiveness of a standalone, unsupervised home exercise program, which can be implemented through either a dedicated web-based platform or a printed paper manual, has yet to be evaluated.

The purpose of this study was to compare the efficacy of an unsupervised home exercise program, through an interactive web-based program or a printed paper manual, with the routine prescription of outpatient PT services after primary, unilateral TKA. Specifically, we asked: (1) Can unsupervised home exercise after surgery provide noninferior recovery, as indicated by return of passive knee flexion, compared with formal outpatient PT? (2) Does a web-based platform for home-based exercise provide any advantage compared with a printed PT manual?

Patients and Methods

A prospective, single-institution, three-arm, parallel-group, single-blind, noninferiority RCT was conducted with approval from our institutional review board. The study was registered at (NCT02911389) and was performed with no external funding.

Study Participants

Patients undergoing primary, unilateral TKA who were > 18 years old and who had provided written consent were eligible to participate. The following were criteria for study exclusion: patients with preoperative knee flexion < 90°; patients considering surgical intervention for pain or dysfunction in a hip or the contralateral knee; patients planning for discharge to a rehabilitation facility or other form of respite care such as a skilled nursing facility, a convalescent home, or a nursing home; and patients undergoing revision or conversion TKA with removal of previously implanted hardware. We randomly allocated 290 patients to formal outpatient PT (outpatient PT group), unsupervised home exercise using an interactive web-based platform (web PT group), or unsupervised home exercise using a printed paper manual (paper PT group). Preoperatively, we collected demographic data, including age, gender, body mass index (BMI), and Charlson Comorbidity Index (CCI).

Study Enrollment

A total of 1887 patients underwent TKA during the study period and were assessed for eligibility. Of these, 1596 patients were not randomized (85%), including 740 (39%) who declined to participate, 433 (23%) who did not meet the inclusion criteria, and 423 (22%) who could not be contacted. A total of 290 patients consented and were randomized in the study, including 97 patients in the outpatient PT group, 96 in the web PT group, and 97 in the paper PT group (Fig. 1).

Fig. 1
Fig. 1:
This is the Consolidated Standards of Reporting Trials (CONSORT) flowchart showing patient recruitment, attrition, and retention.

Patient Demographics

For the study cohort, mean age was 65 years (range, 43–87 years), BMI was 31 kg/m2 (range, 17–41 kg/m2), CCI was 3 (range, 0–11), and length of stay was 1 day (range, 0–5 days). The cohort consisted of 142 men (49.6%) and 148 women (50.4%). There was no difference in age, gender, BMI, CCI, or length of stay among the three groups (Table 1).

Table 1.
Table 1.:
Patient characteristics

Perioperative Procedures

Eight fellowship-trained adult joint reconstruction surgeons performed primary TKA at seven hospitals or surgical centers during a 25-month period between March 2016 and April 2018. The surgeons utilized standardized perioperative protocols for pain management and thromboprophylaxis. All patients received a cemented posterior-stabilized or cruciate-retaining TKA from various manufacturers using a medial parapatellar approach performed under spinal anesthesia with 0.5% bupivacaine. In addition, all patients were mobilized on the day of surgery and instructed to use a walker for 1 to 2 weeks postoperatively and a cane for 2 to 4 weeks, irrespective of treatment allocation.


All patients received daily inpatient PT and occupational therapy until hospital discharge. Patients in the outpatient PT group received formal outpatient PT supervised by a licensed therapist with two to three weekly sessions for 4 to 8 weeks after surgery. Additionally, patients were provided a list of suggested supplemental exercises to be performed at home. The web PT and paper PT groups followed an 8-week unsupervised home exercise program using either an interactive web-based platform (FORCE Therapeutics, New York, NY, USA) or a printed PT manual that was provided to patients before discharge. Both unsupervised home-based PT programs recommended the same weekly exercises, which were to be performed three times daily and graduated from week to week. The printed PT manual provided images and written explanations of exercises, and the web-based platform delivered online video demonstrations and an interactive patient monitoring and communication portal. Additionally, exercises were demonstrated to all patients in each of the groups before hospital discharge. Although no formal rehabilitation program was provided to patients beyond 8 weeks, patients were encouraged to continue with exercises beyond the formal program.

To ensure optimal outcomes for all patients, a delayed recovery intervention protocol was implemented as an integral part of both home exercise programs. Patients in the home exercise groups were evaluated 2 weeks postoperatively, and those with < 70° of knee flexion were required to attend outpatient PT sessions. Patients were again evaluated during the first routine office visit at 4 to 6 weeks, and patients with < 90° of knee flexion were also required to undergo outpatient PT. There was no difference in the proportion of patients in the web PT and paper PT groups who required outpatient PT per the “delayed recovery intervention” protocol (p = 0.175). For the web PT group, 9.4% (nine of 96) and 12.5% (12 of 96) of patients were required to begin outpatient therapy after 2 weeks and 4 to 6 weeks postoperatively, respectively, because they did not meet predefined recovery goals. Similarly, 10.3% (10 of 97) and 20.6% (20 of 97) of patients in the paper PT group required outpatient therapy after 2 weeks and 4 to 6 weeks, respectively.

Outcome Measures

The primary outcome for assessing noninferiority was change in passive knee flexion from baseline assessed longitudinally at both 4 to 6 weeks and 6 months postoperatively. Passive knee flexion, as measured by a goniometer, was recorded by the surgeon who was blinded to patient allocation. Secondary exploratory outcomes included evaluation of the Knee Injury and Osteoarthritis Outcome Score (KOOS), time for return to ADLs, and time until discontinuation of opiate pain medications. KOOS is a validated patient-reported outcome tool that records pain, other symptoms, function in daily activities, function in sports and recreation, and knee-related quality of life; scores range from 0 to 100 with a higher score indicating a better outcome. Functional outcome instruments were administered preoperatively and after routine postoperative office visits at 4 to 6 weeks and 6 months. Time back to ADLs, including work, driving, and independent walking without an assistive device, and time until discontinuation of opiate pain medications were evaluated at 3 months postoperatively. Finally, adverse events, including hospital readmissions, reoperations, and manipulations under anesthesia (MUA) within the 90-day postoperative period, were recorded. There were no universal criteria provided to surgeons for performing MUA in the postoperative period, and surgeons were allowed to use their own clinical judgment.

Sample Size

This study was powered as a noninferiority trial contingent on a two-sided p < 0.05 for significance and a power level of 0.80. Based on a SD of 14° and a noninferiority margin of 5° for change in knee flexion, which provided an effect size of 0.36, the necessary sample size was determined to be 97 patients per group.


Randomization was performed with stratification of age (three strata: < 60 years [25%], 60-70 years [50%], and > 70 years [25%]) and gender (two strata: men [50%] and women [50%]) to ensure balance among the treatment groups [7]. An Excel random number generator (Excel 2013; Microsoft, Redmond, WA, USA) was used to determine the sequential allocation order for randomization. After recruitment, randomization was performed and patient identifiers were added to an internal HIPAA-compliant protected electronic database with group assignments. Patients were provided their assignment by a study coordinator with access to the locked database just before the date of surgery; patients using the web PT platform were enrolled electronically at this time and patients allocated to outpatient PT were informed to allow time for scheduling PT sessions.

Statistical Analysis

The primary analysis was conducted on an intention-to-treat basis in that patients were analyzed based on their group allocation and protocol adherence was ignored. A linear mixed-effect model that adjusted for baseline covariates and considered longitudinal repeated measures was used to assess differences among the groups. Participants with missing data were retained and available data were used to compute maximum likelihood estimates. Statistical significance was set at p < 0.05 for two-sided tests and all analyses were performed using SPSS, Version 23 (IBM Corp, Armonk, NY, USA).


Passive Knee Flexion

After adjusting for baseline flexion, the difference in total change in knee flexion relative to the standard outpatient PT group was +3° (95% confidence interval [CI], -1° to 6°) for web PT and +5° (95% CI, 1°-9°) for paper PT with both investigational groups demonstrating noninferiority to standard outpatient PT based on the predefined noninferiority margin of 5° (Fig. 2). There was a small difference in mean passive knee flexion at preoperative baseline between groups, which were 111°, 114°, and 108° for the outpatient PT, web PT, and paper PT groups, respectively (p = 0.002; Fig. 3). After adjusting for baseline covariates, change in knee flexion from baseline was 0°, -2°, and -1°, respectively, after 4 to 6 weeks (Fig. 4A) and 8°, 8°, and 12°, respectively, after 6 months (Fig. 4B).

Fig. 2
Fig. 2:
This histogram demonstrates the change in knee flexion from preoperative baseline at 4 to 6 weeks to 6 months.
Fig. 3
Fig. 3:
These are the box plots of passive knee flexion by group.
Fig. 4 A-B
Fig. 4 A-B:
These figures demonstrate (A) a histogram of change in knee flexion from preoperative baseline after 4 to 6 weeks postoperatively; and (B) a histogram of change in knee flexion from preoperative baseline to after 6 months postoperatively.

KOOS, ADLs, and Complications

There was no difference in KOOS between groups across multiple repeated measures during the study period (Fig. 5). At preoperative baseline, KOOS scores were 47 points, 46 points, and 43 points for the outpatient PT, web PT, and paper PT groups, respectively, with no difference between groups (p = 0.206; Fig. 6). At 4 to 6 weeks postoperatively, KOOS scores had improved from baseline by an adjusted 17 points (95% CI, 14-20), 18 points (95% CI, 15-21), and 18 points (95% CI, 15-21), respectively, for the outpatient PT, web PT, and paper PT groups. At 6 months, KOOS scores had improved from baseline by an adjusted 34 points (95% CI, 31-37), 32 points (95% CI, 29-35), and 33 points (95% CI, 30-36), respectively, for the outpatient PT, web PT, and paper PT groups (Fig. 5).

Fig. 5
Fig. 5:
This figure demonstrates the change in KOOS scores from preoperative baseline.
Fig. 6
Fig. 6:
This figure demonstrates the box plots of KOOS scores by group.

There also were no differences between the groups with regard to return to ADLs or time to discontinuation of opiate medications. Mean time back to work was 6 weeks (95% CI, 5-7), 6 weeks (95% CI, 5-7), and 7 weeks (95% CI, 6-8) for outpatient PT, web PT, and paper PT groups, respectively (p = 0.218). Mean time back to driving was 4 weeks (95% CI, 3-4), 4 weeks (95% CI, 3-4), and 4 weeks (95% CI, 3-4) for the outpatient PT, web PT, and paper PT groups, respectively (p = 0.814). Mean time back to walking without an assistive device was 4 weeks (95% CI, 3-4), 4 weeks (95% CI, 3-4), and 4 weeks (95% CI, 3-4) for the outpatient PT, web PT, and paper PT groups, respectively (p = 0.953). Mean time back to baseline social activity level was 6 weeks (95% CI, 5-7), 7 weeks (95% CI, 6-8), and 7 weeks (95% CI, 7-8) for the outpatient PT, web PT, and paper PT groups, respectively. Although the outpatient PT group returned to baseline social activity level more quickly than the web PT groups (p = 0.025), there was no difference between the web PT and paper PT groups (p = 0.711). Finally, patients in the home exercise groups discontinued opiates earlier than patients in the outpatient PT group, and mean time off of opiate pain medications was 4 weeks (95% CI, 3-4), 3 weeks (95% CI, 2-4), and 4 weeks (95% CI, 3-4) for the outpatient PT, web PT, and paper PT groups, respectively (p = 0.020).

The risk of readmission (for reasons other than MUA) were not different among the groups. Six patients in the outpatient PT group and four patients each in the web PT and paper PT groups were readmitted to the hospital (p = 0.484), and no patients underwent reoperation. There was also no difference in the proportion of patients in each of the three groups who underwent MUA (3% [three of 97], outpatient PT; 2% [two of 96], web PT; 7% [seven of 97], paper PT patients; p = 0.232).


Supervised PT services, either at an outpatient facility or an inpatient rehabilitation center, historically have been widely perceived as required for all patients after primary TKA subsequent to hospital discharge. However, there is limited investigational evidence to support the presumed benefits of formal PT and its routine use [2]. In fact, a recent RCT found that patients who had undergone TKA receiving an 8-week monitored home exercise program had equivalent mobility as those who first transitioned to an inpatient rehabilitation facility for 10 days [3]. Because a home-based exercise program does not require in-person supervision, it provides a more convenient option for patients eager to return to their normal daily routine; it also reduces postdischarge costs. An increasing interest in unsupervised, home-based rehabilitation has also been fueled by the availability of interactive web-based PT platforms, which offer patients a comprehensive home program that may not be fully achieved with a printed home PT exercise manual. The purpose of this RCT was to compare the efficacy of an unsupervised home exercise program, either on a web-based PT platform or in printed matter, with the routine prescription of outpatient PT services after primary, unilateral TKA based on a primary noninferiority analysis of recovery of knee flexion.

This study was not without limitations. Like with most RCTs, there was some unavoidable deviation from the assigned group allocation. However, an intention-to-treat analysis is a widely accepted approach for overcoming issues with subject compliance without systematically excluding noncompliant patients and maintaining the balance of the initial randomization [4]. Stratified randomization was also implemented to prevent imbalances in age and gender among the treatment groups, which could have influenced treatment responsiveness and expectations for functional recovery. This is a well-accepted practice for noninferiority trials with fewer than 400 participants and can reduce the likelihood of committing type I error [7]. Additionally, patients were only followed for a study period of 6 months. Although there may be additional improvement in knee flexion and patient-reported outcomes beyond the study period for up to 2 years postoperatively, there is no reason to expect any improvement in outcomes would differ between groups so far removed from the study intervention. Moreover, these improvements would be very unlikely to alter the study conclusions. Although a recruitment rate of 5% to 30% is not atypical for interventional RCTs, difficulties with enrollment can lead to selection bias that may influence the true randomness of our cohort. We may have been more likely to recruit motivated patients who could typically follow through in performing assigned exercises without supervision, whereas patients with less discipline or who perceived themselves to be at risk for a poor outcome may have been more likely to opt out of study enrollment. However, it was our experience that patients were more likely to decline enrollment in this study to partake in the web-based home program, which became the standard approach for many participating surgeons for their patients who were not in this trial. This web-based home program was also feasible given the clinical support staff of the participating surgeons at this specific institution. Patients appeared to embrace the opportunity to avoid the burdens imposed by regular outpatient PT sessions. Despite our findings, it is important not to lose sight of the reality that although home-based exercise may provide noninferior outcomes for appropriately selected patients, not all patients are fit to participate in an unsupervised program. It is important that we preserve the opportunity for formal care from a certified physical therapist for those patients who need it the most.

An important part of the home exercise program was the delayed recovery intervention protocol. Understandably, not all patients may regain full ROM in the early postoperative period after TKA. Therefore, we established clear guidelines for determining when patients initially allocated to receive home-based exercise should undergo outpatient PT because of limited progress in attaining minimum early flexion goals, which were 70° of flexion at 2 weeks and 90° of flexion at 4 to 6 weeks. However, it is not clear that the method of engaging patients in postoperative rehabilitation, whether PT is supervised by a licensed therapist or unsupervised at home, plays any role in determining which patients will be slow to progress. Furthermore, it is unclear if a delay in seeking formal outpatient PT services for those in greatest need would have any effect on their final outcome. Future studies should focus on identifying patients for whom a home-based program may not be appropriate and for whom early formal outpatient PT services may be of greatest benefit.

The results of this study demonstrated that an unsupervised home exercise program can provide noninferior recovery after primary, unilateral TKA compared with formal outpatient PT. Specifically, recovery of passive knee flexion was found to be noninferior with unsupervised home exercise, and there was also no significant difference in patient-reported functional outcomes or return to ADLs. There were also few hospital readmissions in each of the groups, although our study was not powered to effectively evaluate differences in safety outcomes such as hospital readmission or the subsequent need for reoperation within the first 90 days. Although several clinical trials have found similar ROM and self-reported function with a home rehabilitation program monitored by a physical therapist [5, 9, 10, 13], we are aware of only one other trial that evaluated an independent home exercise program without monitored supervision [12]. These trials collectively affirm the effectiveness of home exercise after TKA.

Although we did not observe any fundamental advantages to the web-based rehabilitation platform, this interface may provide patients with a sense that they are, in fact, receiving organized, professional PT services. In turn, this may give patients more confidence in their decision to forego a traditional outpatient PT strategy at a modest cost of USD 50 to USD 100 per patient. It is not yet clear how the emergence of online PT platforms will affect reimbursement within current and emerging bundled care arrangements. The platform used in this study provides instructional education videos both before and after surgery and allows patients to stay connected to providers through a web-based communication portal.

With an increasing emphasis on value-based care, it is important to consider the clinical efficacy and cost-effectiveness of routine interventions such as postoperative outpatient PT. With a reported cost of USD 2500 per patient, an 8-week formal outpatient PT regimen for 97 patients would have had an estimated cost to healthcare payers of USD 242,500 as compared with a cost of less than USD 75,000 for 30 patients (of 193 total) requiring PT services per the delayed intervention protocol in the two investigational home exercise groups (mean cost USD 389 per patient) [14]. A portion of the additional PT costs may be passed along to patients by way of copays, which can be a considerable burden. For 693,400 patients who underwent TKA in the United States in 2010, even a 30% reduction in formal outpatient PT could save over USD 500 million [15]. The benefits of a home-based exercise program are not limited to cost savings alone. Patients may find it difficult to return to a normal daily schedule when they must attend two to three weekly outpatient PT sessions, and transportation to PT sessions can be a challenge in the first few weeks after surgery.

In conclusion, the results of this study suggest that unsupervised home exercise is an effective and adequate rehabilitation strategy for selected patients undergoing primary, unilateral TKA. Most patients were able to follow the home-based exercise program to completion and avoid costly physiotherapy services while achieving noninferior functional results. Importantly, a delayed recovery protocol is necessary and can provide a means for early intervention for those struggling to meet minimum recovery milestones. With this limitation in mind, it may be time to reconsider the current practice of routinely prescribing outpatient PT after discharge subsequent to primary, unilateral TKA.


We thank Mikayla McGrath for her incredible contributions with data collection and Carol Foltz for her extraordinary statistical analysis.


1. Anouchi YS, McShane M, Kelly F Jr, Elting J, Stiehl J. Range of motion in total knee replacement. Clin Orthop Relat Res. 1996;331:87-92.
2. Artz N, Dixon S, Wylde V, Beswick A, Blom A, Gooberman-Hill R. Physiotherapy provision following discharge after total hip and total knee replacement: a survey of current practice at high-volume NHS hospitals in England and Wales. Musculoskeletal Care. 2013;11:31-38.
3. Buhagiar MA, Naylor JM, Harris IA, Xuan W, Kohler F, Wright R, Fortunato R. Effect of inpatient rehabilitation vs. a monitored home-based program on mobility in patients with total knee arthroplasty: the HIHO randomized clinical trial. JAMA. 2017;317:1037-1046.
4. Gupta SK. Intention-to-treat concept: a review. Perspect Clin Res. 2011;2:109-112.
5. Han AS, Nairn L, Harmer AR, Crosbie J, March L, Parker D, Crawford R, Fransen M. Early rehabilitation after total knee replacement surgery: a multicenter, noninferiority, randomized clinical trial comparing a home exercise program with usual outpatient care. Arthritis Care Res (Hoboken). 2015;67:196-202.
6. Huddleston J, Alaiti A, Goldvasser D, Scarborough D, Freiberg A, Rubash H, Malchau H, Harris W, Krebs D. Ambulatory measurement of knee motion and physical activity: preliminary evaluation of a smart activity monitor. J Neuroeng Rehabil. 2006;3:21.
7. Kernan WN, Viscoli CM, Makuch RW, Brass LM, Horwitz RI. Stratified randomization for clinical trials. J Clin Epidemiol. 1999;52:19-26.
8. Kettelkamp DB, Johnson RJ, Smidt GL, Chao EY, Walker M. An electrogoniometric study of knee motion in normal gait. J Bone Joint Surg Am. 1970;52:775-790.
9. Ko V, Naylor J, Harris I, Crosbie J, Yeo A, Mittal R. One-to-one therapy is not superior to group or home-based therapy after total knee arthroplasty: a randomized, superiority trial. J Bone Joint Surg Am. 2013;95:1942-1949.
10. Kramer JF, Speechley M, Bourne R, Rorabeck C, Vaz M. Comparison of clinic- and home-based rehabilitation programs after total knee arthroplasty. Clin Orthop Relat Res. 2003;410:225-234.
11. Laubenthal KN, Smidt GL, Kettelkamp DB. A quantitative analysis of knee motion during activities of daily living. Phys Ther. 1972;52:34-43.
12. Mockford BJ, Thompson NW, Humphreys P, Beverland DE. Does a standard outpatient physiotherapy regime improve the range of knee motion after primary total knee arthroplasty? J Arthroplasty. 2008;23:1110-1114.
13. Moffet H, Tousignant M, Nadeau S, Merette C, Boissy P, Corriveau H, Marquis F, Cabana F, Ranger P, Belzile ÉL, Dimentberg R. In-home telerehabilitation compared with face-to-face rehabilitation after total knee arthroplasty: a noninferiority randomized controlled trial. J Bone Joint Surg Am. 2015;97:1129-1141.
14. Ong KL, Lotke PA, Lau E, Manley MT, Kurtz SM. Prevalence and costs of rehabilitation and physical therapy after primary TJA. J Arthroplasty. 2015;30:1121-1126.
15. Williams SN, Wolford ML, Bercovitz A. Hospitalization for total knee replacement among inpatients aged 45 and over: United States, 2000-2010. NCHS Data Brief. 2015;210:1-8.
© 2018 by the Association of Bone and Joint Surgeons