Key Points
Question
Is a digital intervention package that includes an activity tracker, exercise app, and health coaching effective at reducing pain compared with usual care following total knee replacement in adults with osteoarthritis?
Findings
In this randomized clinical trial of 102 adults with total knee replacement, a combined digital intervention provided small but not clinically meaningful reductions in pain at 3 months. Secondary outcomes demonstrated reductions in pain intensity, pain disability, quality of life, and sedentary behavior from baseline to 3, 6, and 12 months, favoring the intervention.
Meaning
Although the reductions in pain were too small to be of clinical importance, these findings indicating that digital interventions improve additional outcomes, such as pain disability, quality of life and sedentary behavior, suggest that future studies considering digital interventions should account for participants’ abilities and preferences.
Abstract
Importance
Digital technology represents an opportunity to improve outcomes following total knee replacement (TKR). Digitally delivered interventions have been shown to be similar to face-to-face interventions and to increase participation levels in people with osteoarthritis.
Objective
To assess the effect of a digital technology package in reducing pain compared with usual care following TKR.
Design, Setting, and Participants
This randomized clinical trial recruited 102 adults after they received TKR in 3 rehabilitation hospitals in Sydney, Australia, between June 2020 and July 2021.
Interventions
All participants underwent usual care. In addition to usual care, 51 participants received a digital technology package consisting of an exercise app, fitness tracker, and online health coaching. In the usual care group, 51 participants received a fitness tracker but with all notifications turned off and goals for step count, sleep, and active hours removed. Participants were followed up for 12 months (June 2021 to July 2022).
Main Outcome and Measures
The primary outcome was mean knee pain during the past week assessed using a numerical rating scale (range, 0-10, with 10 indicating worst possible pain) at 3 months. In unadjusted analyses, considered primary and based on multiple imputations, independent t tests were used to compare means between groups. Secondary outcomes, including measures of function, activity participation, and quality of life, were analyzed using a generalized estimating equation model that accounted for repeated measurements.
Results
Of 102 participants (mean [SD] age, 67.9 [7.2] years; 68 [67%] female; and 92 [90%] White) randomly assigned to intervention or usual care groups, 47 (92%) in each group completed the 3-month follow up. At 3 months, participants in the intervention group demonstrated small but not clinically meaningful improvements in pain compared with the usual care group in the unadjusted intention-to-treat analysis (mean difference, −0.84; 95% CI, −1.59 to −0.10; P = .03). Secondary outcomes indicated a statistically significant reduction in pain intensity, (mean difference, −0.94; 95% CI, −1.82 to −0.06), pain disability (mean difference, −5.42; 95% CI, −10.00 to −0.83), and sedentary behavior (mean difference, −9.76; 95% CI, −19.17 to −0.34) favoring the intervention from baseline to 3, 6, and 12 months.
Conclusions and Relevance
In this randomized clinical trial, a combined digital technology program provided small but not clinically meaningful improvements in pain at 3 months and other longer-term favorable outcomes following TKR compared with usual care. Future studies should tailor digital interventions based on participants’ abilities and preferences to ensure that the intervention is appropriate and fosters long-term self-management.
Trial Registration
Anzctr.org.au Identifier: ACTRN12618001448235
This randomized clinical trial compares usual care alone with usual care plus a digital technology package to reduce knee pain among adults with osteoarthritis after total knee replacement.
Introduction
Osteoarthritis (OA) is a leading cause of disability worldwide.1 Following failure of conservative therapy, total knee replacement (TKR) may be considered, a cost-effective treatment of end-stage OA, improving function and quality of life.2 Following TKR, postsurgical rehabilitation is crucial for optimal recovery, allowing individuals to return to daily activities. However, outcomes following TKR vary, with 1 in 10 individuals reporting dissatisfaction at 2-year follow-up3 and physical activity levels remaining similar to preoperative levels.4
Digital technology presents an opportunity to increase engagement in rehabilitation following TKR. Digital interventions show promise in improving pain, function, and physical activity in OA.5,6,7,8,9 A systematic review by members of our team reported that digital technologies, such as telerehabilitation, educational software, and virtual reality biofeedback, ameliorate pain and improve function following TKR.10 However, the overall quality of the trials included in that review was low and duration of follow-up was short, with most studies (13 of 17 [76%]) having 12 months or less of follow-up.10 The most common method limitation in previous studies is the lack of masking of assessors or therapists.10 It is also unknown whether a combination of digital technologies may provide greater benefits. The aim of the present randomized clinical trial was to compare the effect of a combined digital technology package with usual care in reducing pain among participants following TKR.
Methods
Trial Design
The Participatory Health Through Behavioral Engagement and Disruptive Digital Technology for Post-Operative Rehabilitation of Lumbar Decompression and Total Knee Replacement (PATHway) study was a multicenter, assessor- and statistician-masked, superiority randomized clinical trial stratified by surgery type (TKR and lumbar decompression surgery). The results of the TKR cohort are reported in the present study. Participants who underwent TKR were randomly assigned to the intervention or usual care group with a 1:1 allocation ratio. The study lasted 12 months, with follow-up assessments completed at 3, 6, and 12 months. PATHway was prospectively registered in the Australian and New Zealand Clinical Trials Registry, and the protocol paper was published.11 The full trial protocol is provided in Supplement 1. Ethical approval was granted by the University of Sydney Human Resources Ethics Committee. All participants provided written informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.12
Participants
Participants were recruited from 3 participating rehabilitation hospitals in Sydney, Australia, between June 2020 and July 2021 and followed-up for 12 months (June 2021 to July 2022). The inclusion criteria were being at least 18 years of age; having undergone TKR for knee OA; having been admitted to rehabilitation at 1 of the participating hospitals; and having familiarity with the internet and smart devices, ownership of a smartphone, and willingness to complete study procedures for 12 months. The main exclusion criteria were comorbidities that would prevent full participation in the physical activity program, bilateral or revision TKR, and inadequate understanding of English to provide consent and to participate in health coaching.11
Potential participants were screened for eligibility by hospital site staff during their postoperative admission to the inpatient rehabilitation hospital. When eligible, the hospital site staff introduced the study to the potential participant and contacted the research team to provide further details. A research team member (V.D. or S.D.) visited the participants to further explain the study. If participants were interested, the research team member obtained written consent, randomly assigned the participant to a study group, and delivered the intervention. Cultural heritage was assessed to provide baseline characteristics of the study sample and were self-reported during the baseline survey.
Interventions
In addition to usual care, participants in the intervention group received a 6-month digital technology package aimed to optimize adherence and engagement with the rehabilitation program. The package consisted of an exercise program delivered through an Apple iPad app (PhysiApp), a Fitbit tracker (Fitbit Inc), and fortnightly health coaching for 3 months. Participants in the usual care group received a blinded Fitbit, that is, all notifications (eg, step count notifications, reminders to move) were turned off, and goals for step count, sleep, and active hours were set at 0 and removed from the participants’ Fitbit dashboard so that this information was not accessible.
Exercise Program
The exercise program was developed in consultation with physiotherapists and surgeons from the rehabilitation hospitals, and contained 2 broad categories of exercise, focusing on general knee strengthening and range of motion. The exercises were reviewed at the health coaching sessions and progressed in difficulty based on predefined criteria.
Activity Tracker
Participants received a commercially available activity tracker (Fitbit Inspire) and were instructed to wear the Fitbit 24 hours a day, removing it only for bathing or charging. The Fitbit was used to monitor step count, active hours, and sleep; the goals for each of these criteria were discussed with participants at the health coaching sessions.
Health Coaching
Fortnightly health coaching was delivered by a research team member (V.D.) who is a trained physiotherapist with experience in orthopedic settings and qualifications in health coaching through Wellness Coaching Australia via Zoom videoconferencing (Zoom Video Communications). At these sessions, progress with the exercise program and Fitbit goals were assessed, and personal goals were discussed. Each session lasted approximately 30 to 45 minutes. Participants also received motivational text messages on alternating weeks updating them about their progress. Additional information regarding the intervention can be found in the published protocol.11
Outcomes
The primary outcome was mean pain intensity during the past week, assessed using a numerical rating scale (NRS) of 0 to 10 (0 indicating no pain and 10 indicating worst possible pain) at 3 months. Prespecified secondary outcomes were mean pain intensity during the past week assessed on the NRS at 6 and 12 months; disability assessed using the Pain Disability Index (range, 0-70; higher scores indicate greater disability)13 at 3, 6, and 12 months; participation in physical activity (Active Australia Survey) at 3, 6, and 12 months14; sedentary behavior (Sedentary Behavior Questionnaire; reported as hours per week spent in sedentary behaviors)15 at 3, 6, and 12 months; health-related quality of life evaluated using the Assessment of Quality of Life (AQoL-8D) instrument (total score range, 0-45, with higher scores indicating poorer quality of life)16 at 3, 6, and 12 months; patient activation level and raw score assessed using the 13-item Patient Activation Measure17 (scores range from 0-100, with higher scores indicating higher levels of patient activation; activation levels include 1, indicating “Disengaged and overwhelmed”; 2, “Becoming aware, but still struggling”; 3, “Taking action”; and 4, “Maintaining behaviors and pushing forward”) at 3, 6, and 12 months; technology self-efficacy assessed by the Modified Computer Self-Efficacy Scale18 (range, 0-100, with higher scores indicating higher levels of technology self-efficacy) at 3 months; objectively measured participation in physical activity using activPAL (PAL Technologies Ltd) and Fitbit (Fitbit) trackers at 3, 6, and 12 months; impression of change in pain intensity assessed by the Global Rating of Change19 using a 5-point Likert scale (from much worse to much better) at 3, 6, and 12 months. The Modified Computer Self-Efficacy Scale has been shown to be a reliable and valid measure of technology self-efficacy in a clinical rehabilitation setting (Cronbach α, 0.94).18
Sample Size
A total of 102 participants were recruited to provide 80% power at a significance level of 5% to detect an effect size of 0.6 SD (ie, a difference of 0.9 points on a 10-point pain NRS) given a pooled SD of 1.5 points10 and a dropout rate of 10%. A minimal clinically important difference of 0.9 was chosen because it is the most accepted value for knee pain in people with knee OA.20
Randomization
Participants were randomly allocated to receive intervention or usual care using a computer-generated randomization schedule with random permuted block sizes of 2, 4, or 6 and stratified by study site and surgery type (TKR or lumbar decompression). The allocation was concealed by automatic assignment using the Research Electronic Data Capture (REDCap) randomization module. An independent statistician generated the sequence generation schedules and the REDCap randomization modules were set up by an independent researcher. Two research staff members (V.D. and S.D.) enrolled eligible participants and completed group allocation.
Blinding
Due to the nature of the intervention, participants and clinicians delivering the intervention were unable to be blinded to group allocation. The site staff, assessors, investigators, and study statistician were blinded to treatment allocation until the main results were analyzed.
Statistical Analysis
The primary outcome, pain assessed on a 0 to 10 NRS at 3 months, was analyzed using an intention-to-treat (ITT) analysis. We consider the unadjusted analysis as primary, and it was based on multiple imputations. In the unadjusted analyses, independent t tests were used to compare means between groups. A general linear model was used for the analyses, adjusted for baseline pain, age, and self-reported gender identity. Missing data were imputed using multiple imputations. Specifically, 5 sets of imputed data were created and analyzed, and the mean difference was combined to obtain a pooled effect with an associated 95% CI. The 5 multiple imputed sets for a low to moderate level of missingness are adequate to yield an efficient model effect estimator with at least 95% confidence coverage.21,22 The homoscedasticity assumption was tested using the modified Breusch-Pagan test. If the test indicated a violation of the homoscedasticity assumption, robust standard errors were used. Results were also analyzed using a complete case analysis.
The secondary outcomes were analyzed using a generalized estimating equation (GEE) model that accounted for repeated measurements. The model included the main effects of time and treatment groups and a time-by-treatment interaction and was adjusted for age, self-reported gender identity, and body mass index (calculated as weight in kilograms divided by height in meters squared). The ordinal outcome variables of change in pain intensity, Patient Activation Measure level, and physical activity level were analyzed using a panel-databased GEE model with an ordered logit link function. The effects in the GEE model were estimated and tested using an unstructured correlation structure and robust standard errors. The statistical analysis was completed in November 2022 and was performed using Stata, version 17.0 (StataCorp), with a 2-sided level of statistical significance set at P < .05 or as 95% CIs excluding 0.
Results
Participants
Among 102 participants randomly assigned (51 per group), the mean (SD) age was 67.9 (7.2) years and body mass index was 31.1 (6.3) at baseline. Approximately two-thirds of participants were female (68 [67%]), most participants were married (79 [78%]), and 1 participant (1%) self-reported their cultural heritage as African, 5 (5%) as Asian, 92 (90%) as White, and 4 as other. The intervention group had a slightly higher percentage of retired participants (30 [59%]) compared with the usual care group (21 [41%]) (Table 1).
Table 1. Demographic and Clinical Characteristics of Participants at Baseline, by Group and Total.
| Characteristic | Participants, No. (%) | ||
|---|---|---|---|
| Usual care (n = 51) | Intervention (n = 51) | Total (N = 102) | |
| Age, mean (SD), y | 66.8 (6.1) | 69.0 (8.0) | 67.9 (7.2) |
| BMI, mean (SD) | 31.8 (6.8) | 30.4 (5.7) | 31.1 (6.3) |
| Gender identity | |||
| Female | 34 (67) | 34 (67) | 68 (67) |
| Male | 17 (33) | 17 (33) | 34 (33) |
| Marital status | |||
| Single | 2 (4) | 2 (4) | 4 (4) |
| Married or de facto | 39 (77) | 40 (78) | 79 (78) |
| Divorced or separated | 7 (14) | 2 (4) | 9 (9) |
| Widowed | 3 (6) | 7 (14) | 10 (10) |
| Cultural heritage | |||
| African | 1 (2) | 0 | 1 (1) |
| Asian | 1 (2) | 4 (8) | 5 (5) |
| White | 49 (96) | 43 (84) | 92 (90) |
| Other | 0 (0) | 4 (8) | 4 (4) |
| Highest achieved educational level | |||
| <Secondary | 1 (2) | 2 (4) | 3 (3) |
| Secondary | 9 (18) | 13 (26) | 22 (22) |
| Diploma or certificate | 15 (29) | 16 (31) | 31 (30) |
| Tertiary undergraduate | 9 (18) | 8 (16) | 17 (17) |
| Tertiary postgraduate | 17 (33) | 12 (24) | 29 (28) |
| Employment status | |||
| Unemployed | 3 (6) | 2 (4) | 5 (5) |
| Casual | 3 (6) | 4 (8) | 7 (7) |
| Part-time | 14 (28) | 6 (12) | 20 (20) |
| Full-time | 10 (20) | 9 (18) | 19 (19) |
| Retired | 21 (41) | 30 (59) | 51 (50) |
| Other joints with osteoarthritis | |||
| None | 23 (45) | 21 (41) | 44 (43) |
| Hand | 13 (26) | 15 (29) | 28 (28) |
| Wrist | 5 (10) | 6 (12) | 11 (11) |
| Elbow | 2 (4) | 2 (4) | 4 (4) |
| Shoulder | 5 (10) | 7 (14) | 12 (12) |
| Ankle | 9 (18) | 5 (10) | 14 (14) |
| Foot | 6 (12) | 7 (14) | 13 (13) |
| Hip | 5 (10) | 6 (12) | 11 (11) |
| Spine | 11 (22) | 12 (24) | 23 (23) |
| Duration of symptoms, y | |||
| <1 | 4 (8) | 6 (12) | 10 (10) |
| 1 to <3 | 11 (22) | 8 (16) | 19 (19) |
| 3 to <5 | 8 (16) | 12 (24) | 20 (20) |
| 5 to <10 | 14 (28) | 11 (22) | 25 (25) |
| ≥10 | 14 (28) | 14 (28) | 28 (28) |
| Time between surgery and study enrollment, median (IQR), d | 11 (8-14) | 10 (8-14) | 10 (8-14) |
| Length of stay in operating hospital, mean (SD), d | 5 (2) | 5 (1) | 5 (2) |
| Length of stay in rehabilitation hospital, mean (SD), d | 13 (5) | 13 (5) | 13 (5) |
| Self-Administered Comorbidity Questionnaire, median (IQR), No.a | 3 (1-5) | 4 (2-6) | 4 (2-5) |
| Enrolled in day programb | 50 (98) | 47 (92) | 97 (95) |
Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared).
Range, 0 to 36, with higher scores indicating greater number of comorbid conditions.
Day program includes outpatient physiotherapy program at the rehabilitation hospital that patients attended after they were discharged as an inpatient.
Of 548 participants screened, 446 were excluded. At 3 months, data from all randomized participants were assessed in an ITT analysis. For the complete case analysis, 47 of 51 participants (92%) in both groups were analyzed. The participant flowchart is presented in Figure 1.
Figure 1. Participant Flowchart.
TKR indicates total knee replacement.
Five participants (10%) in each group reported complications due to their TKR surgery (eTable 1 in Supplement 2). There were a total of 3 adverse events experienced by participants that were related to the study intervention. Participants experienced skin irritation due to the bandages used to affix the activPAL devices to their thigh for the 5-day activity tracking. These events were assessed as mild and expected. Treatment adherence for the intervention group and outpatient health care use are presented in eTable 2 and eTable 3 in Supplement 2.
Primary Outcome
At the primary time point of 3 months, the mean difference between groups was statistically significant only in the unadjusted ITT model (mean difference, −0.84; 95% CI, −1.59 to −0.10; P = .03). Following adjustment for baseline variables (pain, self-reported gender identity, and age), the mean difference in the adjusted ITT analysis was −0.75 (95% CI, −1.53 to 0.02; P = .06) (Table 2). Participants in the intervention group reported lower pain levels compared with the usual care group, but this did not meet the minimal clinically important threshold in both adjusted and unadjusted ITT analyses. The between-group differences in pain across all time points assessed are displayed in Figure 2. In the complete case analysis, the mean difference between groups resulted in clinically and statistically significant reductions in pain as assessed with the NRS of 0.94 (95% CI, −1.70 to −0.20; P = .01) in the unadjusted model. Based on robust standard errors, the adjusted mean difference was −0.81 (95% CI, −1.60 to −0.01; P = .047; partial eta squared, 0.04, considered a small effect).
Table 2. Primary Outcome.
| Time | NRS pain score, mean (SD) [No.] | Mean difference between groups (95% CI) | ||||
|---|---|---|---|---|---|---|
| Usual care | Intervention | Unadjusted | P value | Adjusted | P value | |
| Baseline | 6.4 (2.1) [51] | 5.5 (2.6) [51] | NA | NA | NA | NA |
| At 3 mo | 2.6 (2.1) [47] | 1.7 (1.4) [47] | −0.84 (−1.59 to −0.10) | .03a | −0.75 (−1.53 to 0.02) | .06 |
Abbreviations: NA, not applicable; NRS, numerical rating scale.
Statistically significant at P < .05.
Figure 2. Mean Pain Intensity Between Groups.
Error bars represent SDs.
Secondary Outcomes
The results of the GEE model–based effects for secondary outcomes are presented in Table 3. The interaction between time and group was not significant for any secondary outcome variable, indicating that the treatment effect was equal across each period.
Table 3. Longitudinal Analysis of Secondary Outcomes Based on GEE Model Results.
| Outcome | Baselinea | Intervention time | |||||
|---|---|---|---|---|---|---|---|
| 3 mo | 6 mo | 12 mo | |||||
| Mean difference (95% CI) | P value | Mean difference (95% CI) | P value | Mean difference (95% CI) | P value | ||
| Pain intensity (NRS) | Reference | −3.77 (−4.57 to −2.98) | <.001 | −4.42 (−5.18 to −3.67) | <.001 | −4.91 (−5.63 to −4.19) | <.001 |
| Pain Disability Index | Reference | −10.67 (−15.06 to −6.28) | <.001 | −16.41 (−20.68 to −12.14) | <.001 | −15.98 (−20.43 to −11.52) | <.001 |
| Global Rating of Change | NA | Reference | NA | 1.21 (0.37 to 3.98) | .75 | 2.59 (0.70 to 9.60) | .16 |
| AQoL | |||||||
| Total (AQoL-8D) | Reference | −6.78 (−9.93 to −3.63) | <.001 | −6.72 (−9.69 to −3.75) | <.001 | −6.68 (−10.07 to −3.28) | <.001 |
| PCS Social | NA | Reference | NA | −0.36 (−2.47 to 1.75) | .74 | −0.94 (−3.88 to 2.01) | .53 |
| Physical | NA | Reference | NA | 0.88 (−2.02 to 3.77) | .55 | 2.26 (−1.05 to 5.57) | .18 |
| Sedentary Behavior Questionnaire | Reference | −12.16 (−20.59 to −3.73) | .005 | −13.32 (−21.42 to 5.22) | .001 | −11.25 (−19.27 to −3.24) | .006 |
| Fitbit | |||||||
| Sedentary time, min/d | NA | Reference | NA | −14.46 (−43.30 to 14.39) | .33 | −1.37 (−31.16 to 28.41) | .93 |
| Sleep duration, min/d | NA | Reference | NA | −0.34 (−27.45 to 26.57) | .98 | −28.90 (−51.42 to −6.38) | .01 |
| Steps, count/d | NA | Reference | NA | 612.68 (−285.17 to 1510.53) | .18 | 716.18 (−246.21 to 1678.57) | .14 |
| Time spent in bed, min/d | NA | Reference | NA | −1.01 (−29.04 to 27.02) | .94 | −34.03 (−55.88 to −12.18) | .002 |
| ActivPAL | |||||||
| Time spent sitting or lying, min/d | NA | Reference | NA | −21.44 (−56.69 to 13.81) | .23 | −12.87 (−45.36 to 19.61) | .44 |
| Standing time, min/d | NA | Reference | NA | 16.47 (−13.73 to 46.68) | .28 | −11.38 (−14.67 to 37.42) | .39 |
| Steps, counts/d | NA | Reference | NA | 346.77 (−354.6 to 1048.1) | .33 | 114.39 (−644.17 to 872.96) | .77 |
| Stepping time, min/d | NA | Reference | NA | 5.35 (−3.71 to 14.42) | .25 | 1.69 (−7.87 to 11.25) | .73 |
| Transitions, No./d | NA | Reference | NA | 0.74 (−2.71 to 4.19) | .67 | 3.29 (−0.07 to 6.66) | .06 |
| mCSES score (95% CI)b | Reference | 2.36 (−2.62 to 7.34) | .35 | NA | NA | NA | NA |
| Patient Activation Measure | |||||||
| Score (95% CI)c | NA | Reference | NA | 1.79 (−1.75 to 5.34) | .32 | −0.04 (−4.67 to 4.59) | .99 |
| Level (95% CI)d | NA | Reference | NA | 0.11 (−0.09 to 0.30) | .28 | 0.03 (−0.26 to 0.32) | .83 |
| Physical activity AAS, OR (95% CI)e | Reference | 13.17 (4.51 to 38.45) | <.001 | 9.11 (3.33 to 24.98) | <.001 | 20.79 (6.36 to 67.96) | <.001 |
Abbreviations: AAS, Active Australia Survey; AQoL, Assessment of Quality of Life; GEE, generalized estimating equation; mCSES, Modified Computer Self-Efficacy Scale; NA, not applicable; NRS, numerical rating scale; OR, odds ratio; PCS, physical component summary.
Reference indicates reference category used in the estimation of model effects for categorical variables.
Scores range from 0 to 100, with higher scores indicating higher levels of technology self-efficacy.
Activation scores range from 0 to 100, with higher scores indicating higher levels of patient activation.
Activation levels include 1, indicating “Disengaged and overwhelmed”; 2, “Becoming aware, but still struggling”; 3, “Taking action”; and 4, “Maintaining behaviors and pushing forward.”
Effects are odds ratios based on the GEE model with ordered logit link function.
The results of the GEE model indicated a significant reduction in pain intensity (0-10 NRS level), Pain Disability Index score, quality of life (AQoL-8D total score), and Sedentary Behavior Questionnaire scores from baseline to 3, 6, and 12 months, favoring the intervention group. Over time, the reduction in mean pain intensity increased, with the largest reduction observed at 12 months (mean difference, −3.77 [95% CI, −4.57 to −2.98] at 3 months; MD, −4.42 [95% CI, −5.18 to −3.67] at 6 months, and MD, −4.91 [95% CI, −5.63 to −4.19] at 12 months).
In addition, the study found significant decreases in sleep duration (mean difference, −28.90 [95% CI, −51.42 to −6.38] minutes) and time in bed (mean difference, −34.03 [95% CI, −55.88 to −12.18] minutes) at 12 months compared with 3 months favoring the intervention group. The results also suggested a significantly higher likelihood of increased physical activity (from sedentary to insufficiently active, or insufficiently active to sufficiently active) at each period, as measured by the Active Australia Survey, with odds ratio estimates ranging from 9.11 to 20.79.
The results of the main effect of the treatment group in the GEE model suggested that the intervention significantly reduced pain intensity, improved the Pain Disability Index score, and increased physical activity (step count and stepping time). There was a marginal but statistically significant decrease in pain intensity (mean difference, −0.94; 95% CI, −1.82 to −0.06). The mean Pain Disability Index was 5.42 points lower with the intervention (95% CI, −10.00 to −0.83), and the Sedentary Behavior Questionnaire score was 9.76 points lower (mean difference, −9.76, 95% CI, −19.17 to −0.34). The step count per day was 1572 steps higher (95% CI, 345 to 2798 steps) with the intervention as measured by the Fitbit (predicted mean steps, 8576 for intervention and 7004 for usual care), and 1606 steps higher (95% CI, 510-2703) as measured by activPAL (predicted mean steps, 8474 for intervention and 6867 for the usual care), and the stepping time per day was a mean of 19.85 minutes (95% CI, 6.31-33.38 minutes) higher with the intervention (predicted mean stepping time, 111.5 minutes for the intervention and 91.7 minutes for the usual care).
Discussion
The number of TKR procedures continues to steadily rise each year.23 The use of digitally delivered interventions may improve the delivery of rehabilitation, delivering cost savings to the health care system and decreasing the burden on health workers. This randomized clinical trial assessed a combination of interventions, which were selected based on their previous efficacy and availability. A digital technology package consisting of a Fitbit tracker, exercise app, health coaching, and motivational text messages demonstrated small improvements in pain 3 months following TKR compared with usual care in the unadjusted models only. There were also improvements in secondary outcomes of pain intensity, pain disability, health-related quality of life, and sedentary behavior from baseline to 3, 6, and 12 months. A total of 3 adverse events were reported related to the study intervention, suggesting that the intervention was safe and there were no increased harms. A formal cost-effectiveness analysis between the groups is pending.
The difference in pain reduction between groups at 3 months was observed only in the unadjusted ITT and complete case analyses. The pain reduction reached clinical significance (>0.9 points on an 11-point NRS20) only in the unadjusted complete case analysis. This finding may be associated with the nature of the intervention, which required adequate technology self-efficacy to complete. Older participants may have found this more challenging, leading to decreased adherence to the intervention. A cohort study based on the intervention group participants found that those who were younger and possessed higher levels of patient activation and technology self-efficacy were more adherent to the step count component of the intervention.24 Several methods were used to enhance technology self-efficacy throughout the study. Participants in the intervention group were taught how to use the devices at the baseline visit and provided a user manual, instructional videos, and the contact details of the trial coordinators (V.D. and S.D.) for further support. Study staff attempted to resolve technical difficulties through telephone support or Zoom videoconferencing when possible, during or outside of the fortnightly health coaching sessions. Additional face-to-face support may have been beneficial for some participants; however, due to restrictions on traveling during the COVID-19 pandemic, this support was not possible.
Pain disability was reduced and health-related quality of life was improved in the intervention group compared with usual care. These findings are in contrast to a systematic review that found similar improvements only following a digital program compared with face-to-face delivery,25 or greater reductions in pain for face-to-face care of hip and knee OA.5 Our positive results may be explained by ongoing support from the health coaching video calls as well as from the Fitbit trackers used to self-monitor activity following the main intervention period.
The findings of our study are in accordance with previous studies that investigated digital interventions or health coaching following TKR; however, most studies only investigated interventions in isolation or did not included pain as an outcome measure.10,25,26 A feasibility study found similar improvements in step count following a physiotherapist-led intervention that included a Fitbit, step goals, and monthly phone calls.27 A randomized clinical trial investigating health-coaching and financial incentives following TKR reported mean (SD) participant step counts of 5619 (381) and 7152 (4070) steps at 6 months in the health coaching and combined groups, respectively, as measured by Fitbit.26 The cohort of participants in the present study completed slightly more steps, and both groups in the present study exceeded the number of steps recommended to reduce functional limitation in knee OA (6000 steps/d)28 but not all-cause mortality (7500 steps/d).29
Digital interventions may play an important role in postoperative rehabilitation; however, shared decision-making between patients and health professionals can ensure that the interventions and mode of delivery are appropriate and suited to patients’ preferences and abilities.30 Adequate training sessions delivered face-to-face may also help to improve confidence and foster self-efficacy using new digital technologies. Further implementation of a digital intervention program in a larger clinical setting will need to consider important concerns such as data ownership, data monitoring, and resources required (eg, additional staff, training).
Strengths and Limitations
A major strength of our study was the design and intervention. We conducted a high-quality randomized clinical trial that was adequately powered to detect an effect size of 0.6 SD. Adequate follow-up measures were completed at 6 and 12 months following the primary outcome at 3 months, ensuring that any effects of the intervention were measured long-term. The semi-individualized intervention provided to participants following TKR included behavior change techniques, such as self-monitoring and goal setting in collaboration with a health coach for 3 months. The exercises provided to participants were tailored based on their progress, accounting for differences in individual performance and physical activity as measured objectively using Fitbit and activPAL devices. Additionally, the entire intervention was remotely delivered, and all components of the intervention are commercially available, allowing for rapid implementation.
The limitations of our study included participants in the usual care group also receiving a Fitbit device. The inclusion criteria may limit the generalizability of our results. For example, we required participants to own a smartphone and be familiar with using the internet and smart devices. Our results may not be applicable to individuals who are unfamiliar with the use of these technologies. However, the use of the internet and smart devices is becoming increasingly more prevalent in Australians aged 65 years or older.31 Participants were recruited from rehabilitation hospitals, and it is unclear whether a similar intervention would be effective if provided to participants who did not attend rehabilitation.
Conclusions
This randomized clinical trial found that a digitally delivered intervention following TKR was associated with pain reduction at 3 months as well as improvements in pain-related disability, health-related quality of life, and sedentary behavior for 12 months. Although reductions in pain were too small to be of clinical importance, digital interventions may play an important role in other nonclinical outcomes, such as improving accessibility to health care for individuals in rural or remote locations, improving continuity of care by remotely monitoring patients, and reducing health care costs. Future studies should tailor digital interventions based on participants’ abilities and preferences to ensure that the intervention is appropriate and fosters long-term self-management.
Trial Protocol
eTable 1. Complications Reported Due to Total Knee Replacement Surgery
eTable 2. Intervention Treatment Adherence at 3 Months
eTable 3. Fortnightly and Monthly Healthcare Usage Surveys
Data Sharing Statement
References
- 1.Safiri S, Kolahi A-A, Smith E, et al. Global, regional and national burden of osteoarthritis 1990-2017: a systematic analysis of the Global Burden of Disease Study 2017. Ann Rheum Dis. 2020;79(6):819-828. doi: 10.1136/annrheumdis-2019-216515 [DOI] [PubMed] [Google Scholar]
- 2.Daigle ME, Weinstein AM, Katz JN, Losina E. The cost-effectiveness of total joint arthroplasty: a systematic review of published literature. Best Pract Res Clin Rheumatol. 2012;26(5):649-658. doi: 10.1016/j.berh.2012.07.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Heck DA, Robinson RL, Partridge CM, Lubitz RM, Freund DA. Patient outcomes after knee replacement. Clin Orthop Relat Res. 1998;356(356):93-110. doi: 10.1097/00003086-199811000-00015 [DOI] [PubMed] [Google Scholar]
- 4.Smith TO, Mansfield M, Dainty J, Hilton G, Mann CJV, Sackley CM. Does physical activity change following hip and knee replacement? matched case-control study evaluating Physical Activity Scale for the Elderly data from the Osteoarthritis Initiative. Physiotherapy. 2018;104(1):80-90. doi: 10.1016/j.physio.2017.02.001 [DOI] [PubMed] [Google Scholar]
- 5.Jönsson T, Dell’Isola A, Lohmander LS, Wagner P, Cronström A. Comparison of face-to-face vs digital delivery of an osteoarthritis treatment program for hip or knee osteoarthritis. JAMA Netw Open. 2022;5(11):e2240126. doi: 10.1001/jamanetworkopen.2022.40126 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Skrepnik N, Spitzer A, Altman R, Hoekstra J, Stewart J, Toselli R. Assessing the impact of a novel smartphone application compared with standard follow-up on mobility of patients with knee osteoarthritis following treatment with Hylan GF 20: a randomized controlled trial. JMIR Mhealth Uhealth. 2017;5(5):e64. doi: 10.2196/mhealth.7179 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kwasnicki RM, Ali R, Jordan SJ, et al. A wearable mobility assessment device for total knee replacement: a longitudinal feasibility study. Int J Surg. 2015;18:14-20. doi: 10.1016/j.ijsu.2015.04.032 [DOI] [PubMed] [Google Scholar]
- 8.Schoeppe S, Alley S, Van Lippevelde W, et al. Efficacy of interventions that use apps to improve diet, physical activity and sedentary behaviour: a systematic review. Int J Behav Nutr Phys Act. 2016;13(1):127. doi: 10.1186/s12966-016-0454-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gal R, May AM, van Overmeeren EJ, Simons M, Monninkhof EM. The effect of physical activity interventions comprising wearables and smartphone applications on physical activity: a systematic review and meta-analysis. Sports Med Open. 2018;4(1):42. doi: 10.1186/s40798-018-0157-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wang X, Hunter DJ, Vesentini G, Pozzobon D, Ferreira ML. Technology-assisted rehabilitation following total knee or hip replacement for people with osteoarthritis: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2019;20(1):506. doi: 10.1186/s12891-019-2900-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wang X, Hunter DJ, Robbins S, et al. Participatory health through behavioural engagement and disruptive digital technology for postoperative rehabilitation: protocol of the PATHway trial. BMJ Open. 2021;11(1):e041328. doi: 10.1136/bmjopen-2020-041328 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Schulz KF, Altman DG, Moher D; CONSORT Group . CONSORT 2010 statement: updated guidelines for reporting parallel group randomized trials. Ann Intern Med. 2010;152(11):726-732. doi: 10.7326/0003-4819-152-11-201006010-00232 [DOI] [PubMed] [Google Scholar]
- 13.Tait RC, Chibnall JT, Krause S. The Pain Disability Index: psychometric properties. Pain. 1990;40(2):171-182. doi: 10.1016/0304-3959(90)90068-O [DOI] [PubMed] [Google Scholar]
- 14.Brown WJ, Trost SG, Bauman A, Mummery K, Owen N. Test-retest reliability of four physical activity measures used in population surveys. J Sci Med Sport. 2004;7(2):205-215. doi: 10.1016/S1440-2440(04)80010-0 [DOI] [PubMed] [Google Scholar]
- 15.Marshall AL, Miller YD, Burton NW, Brown WJ. Measuring total and domain-specific sitting: a study of reliability and validity. Med Sci Sports Exerc. 2010;42(6):1094-1102. doi: 10.1249/MSS.0b013e3181c5ec18 [DOI] [PubMed] [Google Scholar]
- 16.Hawthorne G. Assessing utility where short measures are required: development of the short Assessment of Quality of Life-8 (AQoL-8) instrument. Value Health. 2009;12(6):948-957. doi: 10.1111/j.1524-4733.2009.00526.x [DOI] [PubMed] [Google Scholar]
- 17.Hibbard JH, Mahoney ER, Stockard J, Tusler M. Development and testing of a short form of the Patient Activation Measure. Health Serv Res. 2005;40(6 Pt 1):1918-1930. doi: 10.1111/j.1475-6773.2005.00438.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Laver K, George S, Ratcliffe J, Crotty M. Measuring technology self efficacy: reliability and construct validity of a modified computer self efficacy scale in a clinical rehabilitation setting. Disabil Rehabil. 2012;34(3):220-227. doi: 10.3109/09638288.2011.593682 [DOI] [PubMed] [Google Scholar]
- 19.Kamper SJ, Maher CG, Mackay G. Global rating of change scales: a review of strengths and weaknesses and considerations for design. J Man Manip Ther. 2009;17(3):163-170. doi: 10.1179/jmt.2009.17.3.163 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bellamy N, Hochberg M, Tubach F, et al. Development of multinational definitions of minimal clinically important improvement and patient acceptable symptomatic state in osteoarthritis. Arthritis Care Res (Hoboken). 2015;67(7):972-980. doi: 10.1002/acr.22538 [DOI] [PubMed] [Google Scholar]
- 21.Rubin DB. Multiple Imputation for Survey Nonresponse. John Wiley & Sons; 1987. doi: 10.1002/9780470316696 [DOI] [Google Scholar]
- 22.Rubin DB. Multiple imputation after 18+ years. J Am Stat Assoc. 1996;91(434):473-489. doi: 10.1080/01621459.1996.10476908 [DOI] [Google Scholar]
- 23.Ackerman IN, Bohensky MA, de Steiger R, et al. Substantial rise in the lifetime risk of primary total knee replacement surgery for osteoarthritis from 2003 to 2013: an international, population-level analysis. Osteoarthritis Cartilage. 2017;25(4):455-461. doi: 10.1016/j.joca.2016.11.005 [DOI] [PubMed] [Google Scholar]
- 24.Duong V, Dennis S, Ferreira ML, et al. Predictors of adherence to a step count intervention following total knee replacement: an exploratory cohort study. J Orthop Sports Phys Ther. 2022;52(9):620-629. doi: 10.2519/jospt.2022.11133 [DOI] [PubMed] [Google Scholar]
- 25.Pietrzak E, Cotea C, Pullman S, Nasveld P. Self-management and rehabilitation in osteoarthritis: is there a place for internet-based interventions? Telemed J E Health. 2013;19(10):800-805. doi: 10.1089/tmj.2012.0288 [DOI] [PubMed] [Google Scholar]
- 26.Losina E, Collins JE, Deshpande BR, et al. Financial incentives and health coaching to improve physical activity following total knee replacement: a randomized controlled trial. Arthritis Care Res (Hoboken). 2018;70(5):732-740. doi: 10.1002/acr.23324 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Christiansen MB, Thoma LM, Master H, et al. Feasibility and preliminary outcomes of a physical therapist-administered physical activity intervention after total knee replacement. Arthritis Care Res (Hoboken). 2020;72(5):661-668. doi: 10.1002/acr.23882 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.White DK, Tudor-Locke C, Zhang Y, et al. Daily walking and the risk of incident functional limitation in knee osteoarthritis: an observational study. Arthritis Care Res (Hoboken). 2014;66(9):1328-1336. doi: 10.1002/acr.22362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lee I-M, Shiroma EJ, Kamada M, Bassett DR, Matthews CE, Buring JE. Association of step volume and intensity with all-cause mortality in older women. JAMA Intern Med. 2019;179(8):1105-1112. doi: 10.1001/jamainternmed.2019.0899 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Elwyn G, Frosch DL, Kobrin S. Implementing shared decision-making: consider all the consequences. Implement Sci. 2016;11(1):114. doi: 10.1186/s13012-016-0480-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Use of information technology by people with disability, older people and primary carers. Australian Bureau of Statistics. July 21, 2020. Accessed August 7, 2023. https://www.abs.gov.au/articles/use-information-technology-people-disability-older-people-and-primary-carers
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
eTable 1. Complications Reported Due to Total Knee Replacement Surgery
eTable 2. Intervention Treatment Adherence at 3 Months
eTable 3. Fortnightly and Monthly Healthcare Usage Surveys
Data Sharing Statement
