Osteoarthritis (OA) of the hip has become an increasing burden for patients, with a high impact on the socioeconomic system [1]. The prevalence of hip OA in the population aged over 45 years is 19 % in women and 27 % in men [2]. The main factors contributing to the development and progression of hip OA, such as obesity, nutrition, sociodemographic setting, injuries, or joint shape, have been extensively studied [3,4]. Further biomechanical aspects related to overload of the cartilage [2] induced by a sum of loads have been identified. Deciding to undergo conservative exercise treatment or surgical total hip arthroplasty (THA) and identifying the best time point for THA can be difficult due to low-level evidence [5]. In Austria, a total of 18,052 primary THAs and 1290 reimplantation surgeries were performed in 2015 [6]. Despite the significant number of THAs, there is a notable deficiency in guidelines and recommendations for both pre- and postoperative management, particularly in the context of functional restoration [7,8]. Notably, Austria lacks a nationally implemented postsurgery patient pathway. However, to some extent, there is existing knowledge on the effectiveness of exercise therapy in reducing functional deficits after THA, as evidenced by previous studies [9-11]. In particular, following motor re-education approaches [12-14] and enhancing the sustainability of exercising by means of real-time feedback (RTF) have been proposed in several studies [15-17]. Regarding motor re-education, the quality of performance, compliance, and adherence to exercising are essential. However, there is no system available focusing on the relevant functional deficits for RTF-supported home exercise. These deficits are valgus thrust, which is defined as the bowing in of the knee or abrupt worsening of the dynamic valgus [18], and pelvic drop, which is increased lowering of the contralateral pelvis side during dynamic single-leg stance (SLS) phases [19]. Additionally, lateral trunk lean is also addressed in the context of such functional deficits. In contrast to the neurologically induced Duchenne limp, this dynamic functional deficit can, among other factors, be induced by pain [20]. Furthermore, there is a lack of kinematic and especially kinetic 3D data addressing functional deficits during recovery after THA [9,10,21,22]. Therefore, the aim of this trial is to investigate the effect of RTF-supported home exercise compared to written home exercise instructions on the recovery of functional performance after THA. Specifically, we want to analyze the improvement of functional deficits at 3 and 6 months postsurgery using quantifiable three-dimensional movement analysis (3DMA) of activity-of-daily-living (ADL) tasks.

This pilot cluster-randomized, 2-arm, parallel-group controlled trial aims at enhancing the understanding of the effects of RTF on functional deficits such as knee valgus thrust, pelvic drop, and lateral trunk lean after THA. Biomechanical and patient-reported outcomes will be assessed prior to the intervention, as well as at 3- and 6-month follow-ups postsurgery.

The primary research question is whether blended-care digital RTF-supported home exercise (intervention group [IG]) leads to significant higher improvements in the control of the frontal knee range of motion (RoM), pelvic obliquity, and lateral trunk lean compared to blended-care exercise supported by written instructions (control group [CG]) as a comparator. The secondary aspect will focus on potential differences between groups concerning aspects of the quality of life, function, and physical activity. Tertiary functional outcomes between groups will be compared to quantify movement quality. Consequently, the underlying primary null hypotheses were prespecified as follows:

In addition, the primary alternative hypotheses are as follows:

The 3 primary outcomes (frontal knee RoM, pelvic obliquity, and lateral trunk lean) will be tested in 3 independent hypotheses (each parameter will be tested independently of the others in 2 follow-ups).

Secondary hypotheses will be tested independently for differences between the IG and the CG in a series of metric patient-reported outcomes measures (PROMs): (1) the Harris Hip Score (HHS), (2) the Hip Osteoarthritis Outcome Score (HOOS), (3) the 12-item Short Form Health Survey (SF-12), (4) the International Physical Activity Questionnaire – Short Form (IPAQ-SF), and, if applicable, (5) the Knee Injury and Osteoarthritis Outcome Score (KOOS).

Tertiary hypotheses will be tested independently for differences between the IG and the CG in a series of metric movement analysis outcomes, including (1) kinematics, (2) kinetics, and (3) spatiotemporal results.

The study is set up as a pilot cluster-randomized controlled trial with 2 parallel groups. Patients will be recruited prior to surgery. At the time of recruitment, the HHS will be assessed routinely in the participating hospital. Additionally, all included patients will be equipped with an activity monitor (wGT3X-BT; ActiGraph LLC) for 2 days prior to surgery in order to assess their mobility level before THA. Baseline assessment will be conducted 4-6 weeks after surgery, and randomization will follow thereafter. During the active phase of 8 weeks, supervised group exercise therapy will take place in both groups. In addition, the IG will perform a home exercise program with the support of an RTF device. The CG will perform a comparable home exercise program without the support of an RTF device (home exercise program following written instructions). Follow-up assessments will take place as close as possible to 8 weeks after baseline and 6 months postsurgery. The schedule of the study is presented in Table 1.

The number of participants assessed for eligibility, allocated to study arms, and analyzed will be recorded using the CONSORT (Consolidated Standards of Reporting Trials) extension flowchart on cluster randomized trials [23]. Figure 1 illustrates the trial design and participant flow through the trial. The protocol is according to SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) reporting guidelines [24].

As knowledge and studies concerning the effect of RTF on exercise following unilateral THA are missing, the sample size cannot be calculated based on given effect sizes. Therefore, the sample size for comparing 2 independent means was calculated using an estimated effect size (d) of 0.8. Assuming the planned 16 clusters with a cluster size of 3 participants each and an intracluster correlation of 0.05, a nominal sample size of 24 participants per group would give a power of 0.75 (α=.05). The effect size (d) of 0.8 and the intracluster correlation of 0.05 were estimated, and hence, the achieved power in this pilot cluster-randomized trial will be reported. The number of participants included in topic-specific studies range from 14 to 111 [5,11,14,25,26]. Considering a dropout of 30%, we will recruit 70 participants from the Orthopedic Hospital Speising, Vienna, Austria.

In the examination, initial status interviews including questions on personal data, such as age, gender, and insurance, and on health in terms of a health check will be conducted at t₀ (baseline 4-6 weeks postsurgery), t₁ (3 months postsurgery), and t₂ (6 months postsurgery). In particular, current pain in the surgical area, its occurrence and severity, and previous injuries and any other general medical interventions or complaints in the leg, pelvis, and spine during the past 6 months will be reviewed. Finally, participants will be asked about medications and known systemic diseases. The status interviews at t₁ will also incorporate questions on usability of the feedback prototype (IG) and on exercise adherence and subjective feelings concerning exercise execution, while status interviews at t₂ will include questions on exercise behavior and whether patients can incorporate exercising into their daily routine. All details concerning status interviews can be found in Multimedia Appendix 1. Within the status interviews, patient reported outcome measures (PROMs) will be gathered. The individual pain level on the 10-item Numeric Rating Scale (NRS-10), with focus on the surgery region, will be noted. In addition, the standardized PROM HHS (prior to surgery, t₀, t₁, and t₂) and the HOOS [28], SF-12 [29,30], and the IPAQ-SF [31] at t₀, t₁, and t₂ will be explained to the participants, and they will be asked to fill them out as part of the status interviews. If a participant reports additional problems with the knee joint, they will be asked to fill out the KOOS [32] at t₀, t₁, and t₂. At t₁, participants in the IG will also be asked to fill in their system usability score (SUS) for quantitative assessment of the usability of the real-time exercise feedback prototype.

The RTF prototype that will be provided to the IG for home exercise is a combination (Figure 2) of the Intel® RealSense Depth Camera D435i (Intel Corporation), the mini-PC Radxa Rock 5 Model B 8GB (Radxa Ltd), and a Raspberry Pi USB-C Power Supply. Using the Intel RealSense SDK 2.0, the red-green-blue (RGB) data are streamed to the mini personal computer (PC) for extraction of a 17-point kinematic model by the machine learning tool for pose estimation, MoveNet.SinglePose V4 (Google Inc). Via a self-programmed user interface operated using an infrared remote controller, the prior developed feedback modes can be selected and started.

The recorded image data, which may potentially identify a person, will be solely used for key point recognition purposes and will not be stored in any form. The RTF prototype has no wireless local area network (LAN), and the existing LAN connection is covered by the housing. Therefore, an internet connection is excluded and access from outside is not possible. The complete data processing will be carried out locally on the prototype. The following data will be stored on the SD card of the prototype: test person ID, body height, training time, achieved exercise scores, number of repetitions of the exercises, and coordinates of the key points. The SD card is located inside the housing, making it difficult to access or remove. The data from the SD card will be analyzed only by the project team and will not be passed on to third parties.

The prototype is not certified as a medical device and was therefore submitted to the Austrian Federal Office for Safety in Health Care for a safety review after a successful ethics vote. Relevant individual parts have a Conformité Européene (CE) mark of conformity.

To compare the RTF prototype with the optoelectronic reference system Vicon Nexus, both systems will be synchronized both temporally and spatially. Temporal synchronization will be achieved through a time-to-live (TTL) signal sent from the prototype to the reference system. Spatial synchronization will be ensured by calibrating the prototype’s coordinate system to match the reference coordinate system. For assessing this agreement, testing participants from the IG will perform a single set of their home exercise program at t1 after performing the set of ADL tasks.

Group exercise therapy will be guided by experienced physiotherapists in groups (as defined per cluster) of 2-5 patients. Patients will participate in 2 group exercise therapy sessions of 90 minutes each per week, with at least 48 hours of rest in between sessions during 8 consecutive weeks. Both the IG and the CG will be instructed and supervised in an equal program, which will only be individualized regarding intensity and repetitions, depending on each patient’s capacity. Group exercise therapy will take place in a suitable room at FH Campus Wien, which will be equipped with the necessary therapeutical/exercise therapy equipment and supplementary aids. At the beginning of each session, the well-being of participants will be assessed and documented.

To ensure self-paced intensity and number of repetitions for participants during the group exercise therapy sessions, time-based sets of 60 seconds will be used. Within those sets, fixed breaks of 60 seconds for recovery will be implemented. For each exercise used in the protocol, variants with decreased and increased intensity will be offered to the participants to ensure appropriate stimulus. Both group exercise therapy interventions and the home exercise program for the IG and the CG will follow the frequency, intensity, time, and type of training (FITT) principle for exercise prescription and will be reported following the Consensus on Exercise Reporting Template (CERT) [34]. The group exercise therapy intervention will consist of a 15-mintue warm-up, 60-minute exercise, and a 15-minute cool-down phase. During the group -exercise phase, patients will be instructed to complete exercises, with focus on movement quality, and if applicable, for specific exercises, they will be instructed to perform the concentric phase of each repetition fast, shortly maintain the individual end-of-range tension, and slowly perform the eccentric phase to reach an approximate time under tension of 5-6 seconds per repetition. The physiotherapists will provide verbal guidance, encouragement, and motivation during the group exercise therapy intervention. With this self-paced protocol, participants will reach approximately 8-12 repetitions per set. The intensity of each exercise will be increased or decreased following participants’ self-reported perceived rate of exertion with the help of the Borg CR10 scale [35]. Intensity will be increased if participants rate an exercise below 4 and decreased if they rate an exercise higher than 7. For all participants, the absolute exercise therapy load will be recorded. Patient-reported hip pain before and after exercise sessions will be assessed during warm-up and cool-down using the NRS-10, graded from 0 (no pain) to 10 (worst pain imaginable). Pain levels from 0 to 2 will be considered safe, from 3 to 5 acceptable, and >5 as not acceptable. On the following day after the exercise therapy intervention, pain should be reduced to the participants’ usual level, or the load will be decreased in the next session. Table 3 presents an overview of the intervention, and a detailed list concerning the planned exercise therapy intervention program, including exercise variants and steps of progression, can be found in Multimedia Appendix 2 following the CERT checklist [34].

In addition to semiweekly exercise therapy sessions, a daily 30-minute home exercise program, excluding days with the group exercise program, will be performed by participants. The home exercise program will be self-documented by the participants in an exercise diary, including the day, the exercises performed, number of repetitions, pain on the NRS-10, and perceived exertion on the Borg CR10 scale [35].

Both groups will receive printed handouts describing each exercise, as well as the full home exercise program in detail. Both groups will also receive standard care treatment, including the same group exercise and home exercise programs. The IG group will perform the home exercise program supported by the digital real-time exercise feedback prototype for digitally guided training (Figure 3).

An overview of the outcome parameters related to the primary, secondary, and tertiary hypotheses is presented in Table 4. The raw data, processed data, and task parameters are described in detail next.

Data will be inspected for outliers using boxplots. Shapiro-Wilk tests and additional graphical inspections of q-q plots will be applied to test for the normality assumption. Metric baseline characteristics will be indicated as means (SDs) and categorical baseline characteristics as frequencies (n, %). Unadjusted outcome scores will be reported as means (95% CIs). People from the same cluster tend to be more similar than those from different clusters, and the analysis must allow for this. A common enhancement is to apply a linear mixed effects model with the cluster specified as a random intercept. To model changes between baseline and follow-up, we will include baseline measurements as covariates. Following suggestions by Klar and Darlington [37], we will adjust for each participant’s baseline assessment and for the baseline cluster mean. Otherwise, no adjustments for covariates are foreseen. Treatment effects will be reported independently for follow-up t₁ and follow-up t₂, including unstandardized and standardized regression coefficients as effect estimates, standard errors, t statistics, and appropriate P values via the Satterthwaite degrees-of-freedom method. Between-cluster and residual variances will be reported for all outcomes. In the case of severe outliers or major violations of the normality assumption, logarithmic data transformation approaches will be explored. Outcome parameters on the ordinal scale will be analyzed descriptively and reported as medians with ranges. All descriptive statistics and corresponding numbers of subjects (n, %) will be reported, along with losses to follow-ups. Effect sizes will be calculated and reported accordingly (per protocol). In addition, the hypotheses will be analyzed according to the intention-to-treat principle (last value carried forward), providing a more conservative effect estimate [38] when missing observations introduce bias. The α value will be set to .05, and exact 2-sided P values will be reported. Hypotheses will be tested independently of each other. The 2-armed study design does not require post hoc tests, so no adjustment for familywise α error accumulation is needed.

Additionally, the agreement between the RTF prototype and the optoelectronic reference system will be analyzed. The spatial deviation of anatomical body key points from the reference system’s key points will be determined using the root mean square error (RMSE). The measured RoM for the angle between the trunk and the thigh, as well as the angle between the thigh and the shank, will be displayed and interpreted using the Bland-Altmann plot.

This study was approved by the Ethics Committee of the City of Vienna (EK-23-103-0623). Further authorization for medical device investigation was obtained from the Austrian Federal Office for Safety in Health and Care (BASG; 10239778, 2023-27-10). The trial was registered at ClinicalTrials (NCT06161194, December 7, 2023). The final protocol approved by the Ethics Committee refers to version 3.3 from December 7, 2023. Recruitment was planned to begin on December 19, 2023, and will approximately be closed by November 6, 2024.

Participants will be assigned a continuous study code (THA01-THA70) to protect identifiable data. The assignment key will be locked in the study manager’s office, separate from identifiable study data.

RTF-supported home exercise is expected to improve exercise execution quality and therapeutic adherence compared to using paper instructions for exercise guidance. As a consequence thereof, we anticipate superior results for the selected outcome measures in the IG.

Participants might not be able to perform ramp walking or stair climbing without the use of handrails at baseline assessment. Dropouts are likely to occur, and thus, they may introduce a risk of attrition bias and, in the worst case, demand the premature termination of the study.

Concerning the use of the HHS and HOOS, subitems of these questionnaires will be interpreted cautiously in view of the limitation that they have been developed for patients with hip OA who did not undergo THA.

To the best of our knowledge, this is the first study to investigate the effect of RTF-supported home exercise following THA. Hypotheses will be tested not only using PROM data but also by means of objectively measured kinematic and kinetic data. Based on the results, the knowledge of more effective rehabilitation pathways after THA and the functional performance of patients 6 months postsurgery will be improved.