Abstract
Introduction
Successful post-operative telerehabilitation following total knee replacement (TKR) has been documented using synchronous (real-time) video. Bandwidth and the need for expensive hardware are cited as barriers to implementation. Web-based asynchronous visual platforms promise to address these problems but have not been evaluated.
We performed a randomized control study comparing an asynchronous video-based software platform to in-person outpatient physical therapy visits following TKR.
Materials and methods
Fifty-one patients were randomized to either the intervention group, using an asynchronous video application on a mobile device, or the traditional group undergoing outpatient physical therapy. Outcome data were collected using validated instruments prior to surgery and at a minimum three-month follow-up.
Results
Twenty-nine patients completed the study. There were no statistically significant differences in any clinical outcome between groups. The satisfaction with care was equivalent between groups. Overall utilization of hospital-based resources was 60% less than for the traditional group.
Discussion
We report that clinical outcomes following asynchronous telerehabilitation administered over the web and through a hand-held device were not inferior to those achieved with traditional care. Outpatient resource utilization was lower. Patient satisfaction was high for both groups. The results suggest that asynchronous telerehabilitation may be a more practical alternative to real-time video visits and are clinically equivalent to the in-person care model.
Introduction
Telemedicine has the promise to improve quality, increase access, and lower cost in health care. The potential to apply live or asynchronous video to the physical rehabilitation of patients following injury, stroke, or surgery has been studied by several authors.1–5 This research has documented clinical results that are generally equal to those obtained using traditional in-person care models.6–9
Studies looking at clinical outcomes when using telerehabilitation following total knee arthroplasty (TKA) have also been promising. Synchronous real-time video telerehabilitation provided favourable results in two randomized trials.8,10 A meta-analysis of the literature relative to telerehabilitation following total knee replacement surgery showed it to provide measurably superior clinical results following total knee surgery for patients treated with telerehabilitation. 11
However synchronous (i.e. live-video), models of telerehabilitation requiring real-time interactions have limitations.8,12,13 Both the patient and the provider need to arrange to be in the same place at the same time and have access to a high-quality screen with a reliable high-speed internet connection. In some case, training is required to use the technology platform.3,14
Asynchronous visual platforms on mobile devices that use either short videos or images to exchange information without requiring simultaneous access aim to address many of these concerns and create a very appealing alternative to synchronous telerehabilitation.15–17 However, there have been no published reports on the use of an asynchronous platform for the telerehabilitation of TKA patients.
Therefore, we set out to compare physical therapy delivered through an asynchronous video-based tool to traditional in-person outpatient physical therapy (PT) following routine TKR.
Our primary outcome was the comparison of patient-reported clinical outcomes obtained using validated instruments. Secondary outcomes included measuring differences in resource utilization between groups, patient satisfaction with the digital tool, and barriers to implementation.
Materials and methods
This was a process improvement project using Standards for Quality Improvement Reporting Excellence (SQUIRE) guidelines. 18 Prior to the initiation of the study, the senior author met with all stakeholders associated with patient care pathways. Internal funding was obtained to assist with software licenses (CaptureProof, San Francisco, CA) and to make the hardware platform available (iPod touch. Apple, Cupertino CA). The urban medical centre where the study took place (Kaiser Permanente Oakland Medical Center) performs approximately 800 primary total knee replacements annually. All patients are managed through a single, standardized total joint protocol.
Fifty-one patients undergoing unilateral, uncomplicated total knee arthroplasty between 1 April 2014 and 31 April 2015 were recruited to prospectively participate in the study. Inclusion criteria included age less than or equal to 75 years, home access to the internet, fluency in English, the ability to send and receive email, and the intent to pursue post-operative care locally at our institution. Patients were informed that they would be randomized to either the intervention arm or the traditional care arm of the study at their first (two-week) visit after surgery. All patients signed a research study consent form and completed the baseline intake survey and patient reported outcomes (PRO) questionnaire prior to surgery. The PRO questionnaire included three validated clinical outcome scores: the 10-point Visual Analog Scale (VAS), the Veterans-RAND 12 item health survey (VR-12), and the Knee Injury and Osteoarthritis Outcome Score Physical Function Short Form (KOOS-PS).19–23
Perioperative protocol
All patients in the study followed our institution’s protocol, which included a pre-operative education class and a single post-operative care protocol. A home health nurse visit occurred within 48 hours of discharge and the patients received “in-home” physical therapy for two weeks.
At the first clinic visit, generally two weeks after surgery, the patients were randomized to either the traditional care group (TG) or the intervention group (IG) using a sealed envelope protocol. Patients who were randomized to receive the standard protocols joined other patients in the outpatient physical therapy and rehabilitation clinics who were not in the study. Patients who were randomized into the IG were provided with an iPod touch for the period of the study, unless they already owned a similar device, in which case the CaptureProof app (www.captureproof.com) was downloaded onto their personal device.
Patients in the IG were taught how to use the application. The therapists had created a total of 23 videos illustrating the same exercises taught in the outpatient clinic. The videos were each less than three minutes in length, narrated by a therapist in English and had on-screen text-based instructions. One of two physical therapists (PT) involved with the study would send instructional videos to the patients (the platform is Health Insurance Portability and Accountability Act (HIPAA) compliant) and the patients would respond with recordings of themselves completing their exercises. The PT was notified, reviewed the video, compared the patient’s progress to previous videos, and made any necessary comments. The PT then uploaded more advanced exercise videos for the patient based on the progress seen. The cycle was repeated until the patient or therapist chose to end the intervention. Other members of the care team could be brought into the conversation and patients could review and compare prior videos to monitor their own progress. Phone or in-person access to technical support staff was made available. During the course of the research, patients would interface with one or both of the therapists involved with the intervention arm of the study, while for the traditional arm of the study, occasionally other therapists were involved but utilized the same protocols. While the types of exercises recommended to each group of patients were the same, the type, number, and frequency of the exercises were always customized to the needs of each patient based on their recovery.
There was no prescribed end point for the therapy sessions or a prescribed minimum number of visits, virtual or in person, for either group. All patients were encouraged to access an exercise gym to perform self-directed therapy. Patients in the IG could request additional in-person visits at any time or to return to using traditional protocols for none, part, or all of their care.
Patients were asked to complete the follow-up PRO questionnaires 90 days following surgery. For patients in the intervention arm, an extended questionnaire using a 5-point Likert scale was added to evaluate their experience with the software platform. 24 Metrics were obtained relative to the use of the web-based interface, such as the total number of sign-ins, photos, and videos uploaded, and total messages sent were collected. Mean and average values are reported. Range of motion (ROM) data were abstracted from the clinical record and the most recent ROM recorded at a minimum of three months was utilized for comparison.
Data were summarized overall and by the IG. Comparisons of demographic and baseline variables were calculated using Student’s t-test for continuous variables and chi-squared statistics for categorical variables. Within the IG, mean changes of VAS, KOOS, physical component scores (PCS), and mental component scores (MCS) pre- and post-operation were evaluated using paired t-tests. Both parametric (one-way analysis of variance) and non-parametric (independent samples median test) were used to compare pre-op differences between groups and pre- to post-op changes between groups. Improvement was scored as 0 or 1 based on published criteria for minimal improvement. 25 Differences between IGs were tested using the chi-squared statistic.
Results
Fifty-one patients were recruited for participation and completed the intake survey – 17 declined to participate prior to randomization at the first follow-up appointment, and five patients were lost to follow-up or declined to complete the follow-up questionnaire. Of the 29 patients who completed the final survey, 14 randomized to the intervention arm and 15 randomized to the traditional arm. The mean age was approximately 62 in both groups, and there were slightly more males in the TG. The majority of patients in both groups lived at home with a partner (Table 1). There were no statistical differences between groups relative to age (p = 0.82), gender (p = 0.43), or mean and median pre-operative KOOS, VAS, and VR-12 (PCS12, MCS12) (Table 2). Box plots of representation illustrate the data distribution and outliers (Figures 1–4).
Box plot of VAS score change between pre- and post-operative data in the intervention group and the traditional group. Demographic characteristics. One-way analysis of variance test of means comparing the pre-operative patient reported scores for the VAS, KOOS, and VR-12 outcome tools between the intervention group (IG) and traditional care group (TG). PCS: Physical component scores; MCS: Mental component scores.
Analysis of statistical significance of change from pre-operative to post-operative score for each of the patient-reported outcome scores, stratified by intervention group (IG). The VR-12 is presented as physical component score (PCS) and mental (MCS) component score.
The TG reported exercising for a mean of 60 min per day compared to the IG with a mean of 47 min per day. The mean time the TG reported door-to-door time spent travelling to physical therapy was 75 min. Both groups found it easy or very easy to contact and communicate with their PT (intervention: mean = 4.5; traditional: mean = 5.0) and reported receiving clear explanations on how to do their exercises (intervention: mean = 4.5; traditional: mean = 5.0). Three patients in the TG (29% of patients) and four patients in the IG (20%) requested to be seen in person by their PT during the three months following their surgery. All four of the patients in the IG that were seen in person continued to access the CaptureProof platform following the visit. In the TG, 13 (87%) patients participated in the group therapy sessions compared to four (29%) patients in the IG. The total number of visits to the PT for “one-on-one” visits (including the initial, mandatory PT visit for all patients) and for “group” visits were 31 and 66 respectively for the TG, and 17 and three respectively for the IG. The last recorded ROM in each group was similar (TG: median = 0–120; IG: median = 1–120). ROM data were missing in four (29%) patients in the IG that were neither seen by their PT nor in clinic at three-month follow-up, electing instead for a telephone visit.
With respect to the utilization of the digital platform, the average user logged in 49 times (range 15–123, median = 37), posted nine videos (range 0–17, median = 9.5) and five photographs (range 0–13, median = 1.5), and sent 10 messages (range 0–30, median = 6). Of the two therapists in the study, the first prescribed an average of 14 exercise videos, logged in 84 times during the course of the study, and spent on average 5 min in a patient session. The second therapist prescribed an average of 10 exercise videos, logged in 121 times, and spent on average 4 min in a patient session. Combined, the PTs spent a total of 907 min (15.1 h) logged in during the three-month course of treatment (approximately 1 h per patient on average) and averaged 14.6 sessions per patient.
When it came to the patient experience of the digital platform, the IG’s average rating for the clarity of instruction was 4.6 (mean = 5.0), ease of taking a video 3.9 (mean = 4.0), ease of sharing the video 4.4 (mean = 4.5), satisfaction with the experience 4.2 (mean = 4.5), and ease of seeing their progress over time 4.4 (mean = 4.0). When patients in the IG were asked, in a free-form format, what they liked most about using the video application, the ability to not travel to the hospital was cited by eight (57%), lack of parking/travel fees associated with travel by three (21%), overall convenience by three (21%), the ability to stay at home by three (21%), ease of access by three (21%), and one patient (7%) stated that the interface was motivational. When asked what could be improved, patients mentioned issues related to taking a video (such as ambient lighting, camera positioning, and screen size) four times (29%), desire for more pro-active outreach by the therapists three times (23%), and software-related aspects two times (16%).
Discussion
We found evidence that patients undergoing telerehabilitation following primary TKA using an asynchronous web-based digital platform may have equivalent clinical outcomes to patients treated with traditional physical therapy. Furthermore, there was a large decrease in outpatient-based physical therapy utilization, and high patient acceptance of the digital platform and the telerehabilitation protocol. Also of note, we have documented that patients are willing and able to actively participate in their care by taking their own videos and submitting them for review – a relatively new model for telerehabilitation.
With respect to clinical outcomes, we found no statistically significant differences between groups, thus proving our hypothesis. Looking at the pre-operative data, the groups were not statistically different demographically. We also found that there were no differences in their pre-operative baseline scores on the outcomes questionnaires, thus validating any subsequent comparison between groups. At last follow-up, we found no statistical difference in the VAS, KOOS, and VR-12 scores between the IG and the TG. Registered ROM values were also similar.
Studies looking at clinical outcomes using synchronous, real-time video telerehabilitation following TKA found similar results. Moffet et al. evaluated results in 198 patients and found that clinical results as well as patient satisfaction were not different between the in-person therapy and the telerehabilitation group. 8 Piqueras et al. reported on 142 patients enrolled in a study where, for two weeks following surgery, patients used a synchronous, interactive, video-based virtual therapy model. At three months the patients showed no difference in clinical outcomes. 10 Agostini et al, in a meta-analysis of the literature relative to telerehabilitation following total knee replacement surgery, found measurably superior clinical results following total knee surgery for patients treated with telerehabilitation. 11
Resource utilization was favourably impacted in our study. We demonstrated a large drop in the number of in-person visits to the outpatient physical therapy clinic from patients in the IG. Unlike in prior studies, which required that all patients complete the same number of visits for the same amount of time to isolate the methodology of the intervention as the primary variable being evaluated,8,10 we allowed patients to access care as they deemed necessary. Patients and therapists thus titrated utilization to what was actually needed. A similar drop in clinic visits, if extended across all total knee patients, would effectively increase the capacity of the PT department by two thirds.
Patient satisfaction overall with both the traditional patient care pathway and the digital interface was high, and there was no major difference in the average scores reported in each group. This was surprising particularly for a group of patients whose average age was over 60 and who were not “digital natives”. Patient satisfaction with the web-based interface was corroborated by the number of “sign-ins”, which was remarkably high, as were the number of videos uploaded and notes written by the patients in the first 90 days following surgery.
Unlike in prior studies, the use of a web-based platform on mobile telecommunication devices meant that there were few, if any, hardware-related barriers to implementation. Our protocol minimizes barriers to implementation by using only standard data speeds to upload and download images and video through a handheld device that the majority of the patients already own. Technical support staff were made available by phone or email, but did not have to visit the patient at home to install or check hardware. This is in contrast to other studies that require a highly complex technological platform, dedicated software, and clinician-controlled multidirectional cameras connected to the physiotherapist at the rehabilitation centre and the patient at home. 8 In the report by Piqueras et al., patients used a sophisticated, synchronous, interactive video-based virtual therapy model that included the application of ROM sensors. 10
While this paper is not a direct comparison to live (synchronous) telerehabilitation following TKA, some considerations can be made. We argue that attempts to recreate the “in-person” visit digitally must inevitably fall short. In the study by Tousignant et al., therapists noted spending up to an additional hour either waiting for patients to log on for their visit or to follow-up with patients by email or telephone after the video visit.26,27 Further, synchronous video visits have several down sides: generally, the video of the visit is not captured, there is only a limited record of the visit kept in the medical record, and usually the patient has no access to that record. The asynchronous model differs because it creates an accessible digital record that can be easily accessed. It thus creates an experience that leverages the strengths of the digital medium. For example, all “visits” become available for later review and comparison to both the patient and any other member of the care team who was not present at the visit. The files can be viewed to review the patient’s progress. Further, the clinic waiting time is eliminated as is the time needed to arrange those appointments. In fact, these aspects of asynchronous care are an improvement over the existing “analogue” in-person care model or the synchronous telemedicine models that attempt to recreate it. In today’s world, an asynchronous solution is thus likely to be easier to implement, less costly, and more effective than synchronous solutions. A direct comparison of the two options might clarify this thesis.
We acknowledge that this study has several limitations, starting with the fact that the study was designed as a process-improvement study with a limited number of patients. Further, our study included only patients 75 years of age or younger and who have some familiarity with internet connectivity. Our data may not be applicable to an older population or one that is fundamentally not interested in using an internet-based tool. The results may be biased in favour of the IG by the selection process used to recruit patients, although we attempted to minimize this bias by randomizing patients between study arms. Further, all the patients in our study belong to a health care plan that has embraced web-based patient access, has adopted standardized care models with excellent published outcomes, even with short hospital stays, 28 and which is based in close proximity to Silicon Valley, with a high penetration of smartphones and high-speed internet connectivity. The results thus obtained might not be immediately applicable to patients who do not share similar demographics or live in a similar environment. However, we believe that with higher technological penetration these differences may be less relevant. We also note that there were significant differences in the time frame for follow-up between patients due to the challenges in contacting some patients and their ability to complete the questionnaire. However, Williams and others have noted that after three months, clinical improvement plateaus following TKA, with only minor differences thereafter.29–31 We do not think that the delay in response by some patients affected the aggregate results, particularly as the average delay was similar in both groups.
In summary, the results of this study suggest that the use of asynchronous telerehabilitation following primary TKA delivers clinical outcomes comparable to those achieved with in-person physical therapy. We also document high patient acceptance and adoption of the mobile application used in this study, and lower barriers to implementation than with published telerehabilitation models of care. Asynchronous models disrupt the current paradigm by allowing patients to access care when they want it, how they want it, and from where they want it. Further, asynchronous interactions create digital information trails that can be reviewed, compared, and shared by care teams, irrespective of time and place, thus playing to the strengths of digital media rather than its weaknesses. We believe that the results of this study, if borne out in a larger study, may represent a significant step forward in how telerehabilitation is used in post-operative rehabilitation protocols.
Box plot of KOOS score change between pre- and post-operative data in the intervention group and the traditional group. Box plot of the change in the mental component score of the VR-12 (MCS-12) in the intervention group and the traditional group. Box plot of physical component score (PCS) change between pre- and post-operative data in the intervention group and the traditional group.
Footnotes
Acknowledgements
The authors would like to acknowledge Matt Stark PT and Jan Richardson PT for embracing this study, and Winnie Steele LVN for her help with patient care and coordination.
The VR-12 was originally developed in the Veterans Administration and builds on the MOS SF-36 version 1.0 and the VR-36 (Veterans Rand 36 Item Health Survey). The scoring algorithms for the VR-12 are copyright by the Trustees of Boston University.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: SA Bini (MD) serves on the Medical Advisory Board of CaptureProof.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Permanente Medical Group provided funds for the purchase of iPods and staff support. CaptureProof provided technical support to patients and researchers relative to the use of the application and development of exercise videos.