This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Rehabilitation provided through home visits is part of the continuum of care after discharge from hospital following total knee arthroplasty (TKA). As demands for rehabilitation at home are growing and becoming more difficult to meet, in-home telerehabilitation has been proposed as an alternate service delivery method. However, there is a need for robust data concerning both the effectiveness and the cost of dispensing in-home telerehabilitation.
The objective of this study was to document, analyze, and compare real costs of two service delivery methods: in-home telerehabilitation and conventional home visits.
The economic analysis was conducted as part of a multicenter randomized controlled trial (RCT) on telerehabilitation for TKA, and involved data from 197 patients, post-TKA. Twice a week for 8 weeks, participants received supervised physiotherapy via two delivery methods, depending on their study group allocation: in-home telerehabilitation (TELE) and home-visit rehabilitation (VISIT). Patients were recruited from eight hospitals in the province of Quebec, Canada. The TELE group intervention was delivered by videoconferencing over high-speed Internet. The VISIT group received the same intervention at home. Costs related to the delivery of the two services (TELE and VISIT) were calculated. Student’s
The mean cost of a single session was Can $93.08 for the VISIT group (SD $35.70) and $80.99 for the TELE group (SD $26.60). When comparing both groups, real total cost analysis showed a cost differential in favor of the TELE group (TELE minus VISIT: -$263, 95% CI -$382 to -$143). However, when the patient’s home was located less than 30 km round-trip from the health care center, the difference in costs between TELE and VISIT treatments was not significant (
To our knowledge, this is the first study of the actual costs of in-home telerehabilitation covering all subcosts of telerehabilitation and distance between the health care center and the patient’s home. The cost for a single session of in-home telerehabilitation compared to conventional home-visit rehabilitation was lower or about the same, depending on the distance between the patient’s home and health care center. Under the controlled conditions of an RCT, a favorable cost differential was observed when the patient was more than 30 km from the provider. Stakeholders and program planners can use these data to guide decisions regarding introducing telerehabilitation as a new service in their clinic.
International Standard Registered Clinical Study Number (ISRCTN): 66285945; http://www.isrctn.com/ISRCTN66285945 (Archived by WebCite at http://www.webcitation.org/6WlT2nuX4).
The rapid growth in the number of total knee arthroplasty (TKA) surgeries per year and shorter hospital stays have greatly increased demand for outpatient rehabilitation services for patients returning home more quickly after hospital discharge. However, this increasing need has not been followed by an increase in accessibility to rehabilitation services after TKA, in-home care, or outpatient clinics. As part of a worldwide situation [
Teletreatment, which is defined as care and treatment delivered remotely between patients and health care professionals with the support of communication and information technology [
Our main interest is to increase access to, and efficiency of, rehabilitation services after TKA through the use of teletreatment between the health care center and patient’s home. In this context, the feasibility of providing in-home telerehabilitation following joint replacement surgery has been studied previously [
This cost analysis was embedded in the telerehabilitation for knee arthroplasty (TelAge) randomized controlled trial (RCT) on the effectiveness of in-home telerehabilitation. The TelAge project is a multicenter randomized trial, involving three research centers located in Quebec City, Sherbrooke, and Montreal, and well as eight hospitals [
The project was approved by ethics committees of each hospital and the participants’ written consent was obtained prior to their inclusion. The RCT was registered with the International Standard Registered Clinical Study Number (ISRCTN) registry (ISRCTN66285945).
Participants were recruited from the surgical waiting lists of orthopedic surgeons in each hospital. To be included in the study, they had to meet the following inclusion criteria: (1) were waiting for a primary TKA after a diagnosis of osteoarthritis, (2) returning back home after hospital discharge, (3) living in an area served by high-speed Internet services (at least 512 kbps in upload), and (4) living within a 1-hour driving distance from the treating hospital. Patients were excluded if they met the following criteria: (1) had health conditions that could interfere with tests or the rehabilitation program, including lower limb surgery in the last 9 months, (2) were planning a second lower limb surgery within 4 months, (3) had cognitive or collaboration problems, (4) had major postoperative complications, or (5) had weight-bearing restrictions for a period longer than 2 weeks after surgery.
After their surgery, patients were block randomized by random generator with sealed envelopes into either the in-home telerehabilitation (TELE) experimental group or the home-visit (VISIT) control group. For both groups, the goal of the physiotherapy program was to recover function—only the delivery method differed. The duration and frequency of the supervised physiotherapy sessions were standardized for both groups and consisted of two 45-minute sessions per week for 8 weeks. Patients were assessed four times by a blind assessor—twice before and twice after the intervention—to evaluate rehabilitation efficacy, along the following timeline: (1) at baseline (ie, 1 month or less before TKA) (E1), (2) 1 to 2 days before discharge from hospital (E2), (3) 2 months postdischarge (E3), and (4) 4 months postdischarge (E4).
The telerehabilitation platform included various components to provide a user-friendly videoconferencing experience for both the clinician and the patient at home. The core was the videoconferencing system (Tandberg 550 MXP), which used an H.264 video codec and incorporated a pan-tilt-zoom (PTZ) camera with wide-angle lens and omnidirectional microphone. On the patient side, the system was mounted over a 20-inch LCD screen, which displayed the video received from the other end. Audio was played using external speakers placed on both sides of the screen, as the internal LCD screen speakers are rarely sufficient to provide a satisfactory experience. Video and audio data were encrypted and transmitted over a high-speed Internet connection—at the time of the TelAge RCT, minimum bandwidth for both upload and download was 512 kbps. On the clinician side, a computer with dedicated software (TeRA) was added to the videoconference link to enable user-friendly control and monitoring of videoconferencing sessions, near- and far-end camera controls, built-in clinical tests, photo and video recordings, and acquisition and display of external sensors and peripherals (
TelAge clinical randomized controlled trial study design including cost evaluation period.
Telerehabilitation platform.
Costs were measured during the intervention period—between E2 and E3—for both groups. All cost-related data were directly collected by the following: (1) physiotherapists after each session (ie, intervention, travel, technical problems), and (2) technicians after each installation/uninstallation (ie, travel, technical problems). Cost-related data were collected using a standardized cost sheet during the period from hospital discharge to 2 months postdischarge (ie, the end of rehabilitation intervention). Other costs, such as Internet service and technical/clinical equipment amortization, were calculated using the real cost of the service and purchase price of the equipment, respectively, at the time of the TelAge RCT.
This economic evaluation followed international guidelines for conducting a cost analysis alongside a clinical RCT [
More precisely, direct clinical time included duration of the interaction between physiotherapist and participant to deliver the intervention in both groups. Indirect clinical time was defined as the following: (1) time without direct contact with the participant, but which was required to optimize the intervention (ie, scheduling, file preparation, planning the treatment, writing the report and follow-up notes, phone follow-up with orthopedic surgeon), (2) travel time and kilometer (km) allocation for use of the physiotherapist’s car—only for the VISIT group, and (3) clinical equipment amortization, which is the theoretical time duration of use of the technology. Round-trip distance from the health care center (ie, hospital where the research center involved in the project was located) to the patient’s home was calculated for each patient in both groups.
Direct technology costs—only for the TELE group—are the costs in Canadian dollars for acquiring clinician and patient telerehabilitation platform kits as illustrated in
The standard hourly salary in effect in the public health system in Quebec at the time of the TelAge RCT was applied to the time spent by the physiotherapists and technicians—Can $45.22/hour and Can $22.09/hour, respectively. Travel expenses were calculated as distance in kilometers multiplied by Can $0.40. In addition, to determine the cost of the whole treatment (ie, 16 sessions), decision makers may be interested in the net costs of giving one TELE or one VISIT treatment, for instance, the cost per treatment. For that analysis, only costs of treatments actually received were taken into account. Moreover, the distribution of costs in each delivery-service method was analyzed.
Breakdown of costs per treatment for telerehabilitation and home visits.
Cost category | Description | Duration | Professional costacalculation | ||
|
|
|
|
TELE | VISIT |
|
|
|
|||
|
|
|
|
||
|
|
Contact with the participant | Hours | Hours x $45.22b | Hours x $45.22 |
|
|
|
|
||
|
|
Planning the session | Hours | Hours x $45.22 | Hours x $45.22 |
|
Follow-up with orthopedic surgeon | Hours | Hours x $45.22 | Hours x $45.22 | |
|
Writing the report and follow-up notes | Hours | Hours x $45.22 | Hours x $45.22 | |
|
Travel | Hours, km | N/Ac | Hours x $45.22, |
|
|
Clinical equipment amortization | 16 treatments | $0.63 per treatment |
$0.63 per treatment |
|
|
|
|
|||
|
Installation/uninstallation of technology | Travel time (hours) x $22.16d, |
N/A | ||
|
Internet costs (high speed)e | $7.63 per intervention | N/A | ||
|
Technical equipment amortizationf | $4.88 per intervention | N/A | ||
|
|
|
|||
|
Technician | Hours x $22.16 | N/A | ||
Physiotherapist | Hours x $45.22 | N/A |
aAll costs are in Canadian dollars.
bPhysiotherapist mean hourly salary in the public system in Quebec.
cNot applicable (N/A).
dThe hourly rate at the higher echelon of the salary scale of a technician.
eCost for 2 months (16 sessions): $122.08/16 sessions = $7.63 per session.
fCost for 1 clinician kit ($5760) and 1 patient kit ($7200) amortized over 3 years based on 50% of usage per week (17 hours): $5760 + $7200 = $12,960. Hours telerehabilitation: 3 years x 52 weeks/year x 17 hours (50% usage)/week = 2652 hours. Cost per treatment: $12,960/2652 hours = $4.88.
As time spent on travel is an important cost factor for the VISIT group, the distance between the health care center and patient's home is an important variable that must be taken into account when comparing costs between VISIT and TELE treatments. To consider the impact of this variable, a sensitivity analysis was performed on the basis of the distance between the patient’s home and health care center for the physiotherapist delivering the treatment (VISIT) or the technician installing/uninstalling the technology (TELE). Another way to look at this sensitivity analysis is to consider that time spent on travel may differ in metropolitan, urban, and rural areas, for instance, travelling 5 km in a metropolitan area can take more time than in urban and rural areas. Knowing that, the analyses were done separately for each site.
All analyses were performed using SAS/STAT software, version 9.3 of the SAS System for Windows (SAS Institute Inc, Cary, NC, USA). First, the two groups of patients were compared using sociodemographic characteristics. For the continuous variables, Student’s
A total of 258 patients needing TKA were recruited for the TelAge RCT [
The characteristics of the participants were comparable in each group for every sociodemographic variable measured, except for participants living alone (
Descriptive statistics at baseline of the sample included in the cost analysis (n=197).
Sociodemographic characteristic |
TELE (n=97), |
VISIT (n=100), |
|
Age (years), mean (SD) | 65 (8) | 67 (8) | .11 |
Body mass index (kg/m2), mean (SD) | 34.6 (7.6) | 33.1 (5.6) | .12 |
Comorbidity Index (%), mean (SD) | 23 (12) | 21 (9) | .16 |
Functional ability before TKA |
53 (19) | 54 (17) | .74 |
Round-trip distance from health care center to patient’s home (km), mean (SD) | 59 (67) | 34 (35) | .002 |
Sex (males), n (%) | 40 (41) | 55 (55.0) | .05 |
Operated knee (right), n (%) | 45 (46) | 51 (51.0) | .52 |
Previous lower limb surgery, n (%) | 44 (45) | 49 (49.0) | .66 |
Living alone, n (%) | 20 (21) | 10 (10.0) | .04 |
a
bWestern Ontario and McMaster Universities Arthritis Index (WOMAC).
In the randomized trial, some participants from the TELE group and VISIT group did not receive all 16 scheduled treatments due to medical, technical, or personal reasons. Average treatment frequency was 16.0 (SD 0.2) for the VISIT group and 15.3 (SD 2.1) for the TELE group (
Costs of telerehabilitation and home visits.
Cost | TELE ($) (n=97), |
VISIT ($) (n=100), |
TELE-VISIT ($), |
(TELE-VISIT)/VISIT (%), |
|
Total costa | 1224 (241) | 1487 (553) | -263 (-382 to -143) | -18 (-26 to -10) | <.001 |
Cost per treatmentb | 80.99 (26.60) | 93.08 (35.70) | -12.09 (-20.91 to -3.27) | -13 (-23 to -4) | .008 |
aTotal cost is for the total intervention, which includes all received and cancelled treatments.
bCost per treatment is for one treatment and is based only on treatments received.
Cost per treatment shows that the mean cost of treatment for the VISIT group was Can $93.08 (SD $35.70) versus Can $80.99 (SD $26.60) for TELE group. The differential cost between the two groups was 13% in favor of the TELE group (TELE minus VISIT: Can -$12.09, 95% CI -$20.91 to -$3.27) (
As expected, the travel cost for the VISIT group is an important cost-driven variable—49% of the total cost. Interestingly, costs related to technology support during teletreatment, outside of the expected time for equipment installation/uninstallation, did not significantly affect the cost—1.5% of total cost.
At a micro level, it is somewhat surprising to see discrepancies between some subcosts. For example, therapists spent the same proportion of time with the patient in both groups—TELE 44%, VISIT 44%—but a larger proportion of time was spent in the TELE group preparing for the treatment—TELE 5%, VISIT 2%—and on follow-up—TELE 9%, VISIT 4%.
There was a difference in costs per treatment between the two groups, and this difference was not the same in the five distance strata (
For urban and rural areas, TELE treatments were less expensive than VISIT treatments when the round-trip distance was 50 km or more (Can $97<$144,
Breakdown of costs per treatment (telerehabilitation and home visits).
Cost category | TELE | VISIT | ||||
|
|
|
Can $ | % | Can $ | % |
|
|
|
|
|
||
|
|
|
|
|
|
|
|
|
Treatment | 35.85 | 44 | 41.02 | 44 |
Clinical equipment amortizationa | 0.74 | 1 | 0.63 | 1 | ||
Total | 36.59 | 45 | 41.65 | 45 | ||
|
|
|
|
|
|
|
|
|
Time related to preparation for the treatment | 4.48 | 5 | 2.26 | 2 |
Follow-up with patients | 7.03 | 9 | 3.55 | 4 | ||
Travel to participant’s home (time + km allocation) | N/Ab | N/A | 45.62 | 49 | ||
Total | 11.51 | 14 | 51.43 | 55 | ||
Total clinical costs | 48.10 | 59 | 93.08 | 100 | ||
|
|
|
|
|
||
|
|
|
|
|
|
|
|
|
Technical equipment amortizationa | 5.81 | 7.2 | N/A | N/A |
Installation/uninstallation of technology | 18.12 | 22.4 | N/A | N/A | ||
Internet costs | 7.63 | 9.4 | N/A | N/A | ||
Technical problem resolution time and travel time (technician) | 1.25 | 1.5 | N/A | N/A | ||
Technical problems after sessions by the physiotherapist | 0.08 | 0.1 | N/A | N/A | ||
Total technological costs | 32.88 | 41 | N/A | N/A | ||
|
80.99 | 100 | 93.08 | 100 |
aThe clinical and technical equipment amortization costs are slightly different from those in
bNot applicable (N/A).
Cost sensitivity analysis based on round-trip distance from health care center to patient’s home in the total sample (n=197).
Distance (km) | Costs per treatment (Can $) |
|
|||
|
TELE | VISIT |
|
||
|
n (%) | Mean (SD) | n (%) | Mean (SD) |
|
<10 | 12 (6.1) | 71.0 (9.8) | 21 (10.7) | 63.9 (7.4) | .248 |
10-19 | 24 (12.2) | 69.5 (8.5) | 27 (13.7) | 74.8 (10.5) | .260 |
20-29 | 14 (7.1) | 73.6 (10.1) | 19 (9.6) | 83.1 (8.3) | .112 |
30-49 | 12 (6.1) | 81.3 (13.1) | 13 (6.6) | 102.7 (19.5) | .002 |
≥50 | 35 (17.8) | 90.2 (24.4) | 20 (10.2) | 151.6 (30.6) | <.001 |
aTwo-way ANOVA with interaction was conducted to compare costs per treatment between the two groups and between the five distance strata (
Cost sensitivity analysis based on round-trip distance from health care center to patient’s home, calculated separately for each site.
Site |
|
Costs per treatment (Can $) | Interaction |
Post hoc |
||||
|
|
|
TELE | VISIT |
|
|
||
|
|
Distance (km) | n (%) | Mean (SD) | n (%) | Mean (SD) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
|
< 50 | 34 (31.2) | 69.8 (8.3) | 46 (42.2) | 71.4 (10.4) | <.001 | .684 |
≥ 50 | 20 (18.3) | 97.3 (28.8) | 9 (8.3) | 143.5 (26.1) |
|
<.001 | ||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
< 30 | 4 (11) | 74.6 (10.4) | 8 (22) | 92.4 (12.3) | .002 | .192 |
≥ 30 | 13 (35) | 80.2 (9.0) | 12 (32) | 152.0 (34.9) |
|
<.001 | ||
|
|
|
|
|
|
|
|
|
|
< 30 | 14 (28) | 73.7 (10.6) | 16 (32) | 75.0 (9.0) | .001 | .815 | |
≥ 30 | 11 (22) | 81.7 (15.7) | 9 (18) | 107.7 (28.6) |
|
.001 |
aFor each site, a two-way ANOVA with interaction was conducted to compare costs per treatment between the two groups and between the distance strata. Results shown in the table are
The purpose of this study was to document costs for in-home teletreatment versus conventional home-visit rehabilitation following TKA based on an RCT. To our knowledge, this is the first cost analysis of in-home telerehabilitation involving an assessment of the cost structure for this type of service delivery versus traditional in-home, face-to-face rehabilitation. One unique facet of our study is the inclusion of all subcosts related to telerehabilitation (ie, equipment amortization, installation/uninstallation, technical problems related to teletreatment). Our results show that delivering physiotherapy by telerehabilitation post-TKA is less expensive than providing the same service in home visits if the distance between the health care center and patient’s home is more than 30 km. Overall, when we take into account all direct and indirect costs of the two approaches, teletreatment between the health care center and patient’s home is less costly than home visits by at least 18%.
In the context of cost constraints in the health care system, is the cost savings of 18% sufficient to introduce in-home telerehabilitation? This result did not specifically take into account one major cost-driven factor related to home visits, which is proximity of the participant’s home to the health care center (ie, sensitivity analyses). The inherent hypothesis that comes to mind is that proximity of health care centers to patients’ homes can significantly influence the cost of the VISIT group—the farther away a participant lives, the higher the costs related to home-visit rehabilitation. For example, in the VISIT group, 16 round-trips were needed to provide treatment in the home by the physiotherapist, but only two round-trips were needed for the TELE group (ie, installation/uninstallation) by the technician. To take into consideration the proximity of the participant’s home to the health care center, a sensitivity analysis clearly showed that if travel distance is less than 30 km, round-trip, differential costs between the two alternatives decline. On the other hand, if round-trip travel distance is longer than 30 km, the cost differential increases dramatically, reaching 123% for 50 km or more. This is an important result for those who want to implement telerehabilitation—to benefit from cost savings, telerehabilitation must involve patients who live at least a 30 km drive, round-trip, from the clinic if the resources are available. In addition, it would be more efficient to use specialized health care providers to administer treatments than to spend their time travelling, as was done in the VISIT group.
Moreover, if we consider traffic in metropolitan, urban, and rural areas as a confounding variable to proximity of the participant’s home to the health care center, we observed the same tendency, with the difference being that in the urban and rural area (Sherbrooke), a difference of 50 km was needed to keep the cost savings of the TELE versus the VISIT group. This study shows that costs were not higher for the home-visit group, which could be of importance to stakeholders and managers.
As a novel piece of the puzzle to understand costs of telerehabilitation compared to home visits, examination of the subcosts of the two approaches produced some interesting results. For that analysis, only costs of the treatment actually delivered were considered. Direct treatment time was not the same in both groups—it was less in the TELE group. Even though there was a standardization of the treatment in both groups in the research protocol, this difference can be attributed to taking less time to discuss with the patient or less time between exercises because of the virtual mode of communication. However, we saw that the proportion of time spent on preparing for the treatment and following up after therapy was more than double in the TELE group (
As expected, travel drove the costs of home visits (ie, 49% of total costs). The technology of the TELE group was also costly (ie, 41% of total), but not enough to counterbalance travel costs. Looking at the subcategories of the 41% technology costs, 7.2% of the costs were attributable to equipment amortization and 9.4% to telecommunication (ie, high-speed Internet). At the time of the TelAge RCT (2008-2013), the videoconferencing equipment purchase costs and telecommunication costs were much higher than for comparable equipment and services today. Comparable videoconferencing equipment is available for less than half the purchase price used in our analyses. Telecommunication costs for activation fees and bandwidth are significantly lower as accessibility and use of high-speed Internet have now reached unprecedented levels, even with older adults, and bandwidth costs have declined. As the cost per treatment for the technological factors of in-home telerehabilitation was only 16% of the total costs, this would not appear to be a determining variable in our results. However, the exponential evolution of these technological factors in today’s technological context (ie, lower cost of videoconferencing, ubiquity, and performance of telecommunication networks) could impact the cost differential between the TELE group and the face-to-face VISIT group services delivery by lowering the barrier to entry of the TELE option when scaling up in-home visit rehabilitation.
Regarding internal validity, information bias was minimized by using specific procedures. First, data collection was standardized with cost forms and information was collected immediately after each intervention (ie, physiotherapy session). Second, the standardized cost forms were pretested. This assured us of the specific details required to calculate all subcosts. For each subcost, the person involved in service delivery (ie, technician, engineer, physiotherapist, study coordinator, etc) completed the appropriate form the same day that the intervention occurred. Following this rigorous process, memory bias should have been minimized. Finally, to decrease missing data, the site coordinator checked the data for completeness. With all these precautions, we are confident in the validity of the cost data.
Selection bias was controlled by the randomization procedure in the TelAge RCT leading to two equivalent groups regarding sociodemographic variables [
We already know that teletreatment for TKA is efficient [
In the context of increasing access to rehabilitation, our study demonstrated the cost savings of teletreatment compared to home visits for patients post-TKA. Managers can now decide whether to implement telerehabilitation based on robust clinical and economic data. Given the recent attention that telerehabilitation is drawing on an international level, it would be interesting in future research to carry out a large-scale study about one of the major practical issues related to the provision of telemedicine services to a specific patient population.
analysis of variance
Canadian Institutes of Health Research
assessment time, at baseline
assessment time, 1 to 2 days before discharge from hospital
assessment time, 2 months postdischarge
assessment time, 4 months postdischarge
general linear model
International Standard Registered Clinical Study Number
kilometer
least squares means
not applicable
pan-tilt-zoom
randomized controlled trial
telerehabilitation for knee arthroplasty
in-home telerehabilitation (experimental) treatment
total knee arthroplasty
home-visit (control) treatment
Western Ontario and McMaster Universities Arthritis Index
The authors would like to thank all participants, study personnel, and orthopedic surgeons for their contributions to this project. This research was supported by funding from the Canadian Institutes of Health Research (CIHR).
None declared.