Comparison of human versus digital instructions for exercise in patients with hip osteoarthritis: results of a non-inferiority randomized cross-over trial

Abstract

Purpose: The global burden of hip and knee osteoarthritis (OA) was ranked as the 11th highest contributor to global disability. Exercise has been shown to be effective to decrease pain and increase physical function. In this regard it is particularly important to enable patients to perform physical exercises in the best possible health-promoting and low-risk manner, taking into account their own physical condition. Face-to-face supervision is the gold standard for exercise guidance at the initiation of an exercise regime. However the age of digital media provides new opportunities to guide patients. It is crucial to evaluate if this type of intervention is not inferior to a personal contact with a trainer.The aim of this study is to evaluate whether instruction and guidance via a digital application, which is based on an evidence-based exercise regime, is not inferior to supervision by a physiotherapist with regard to movement quality, control competence for physical training and exercise-specific self-efficacy. Methods: Patients with clinically hip OA were recruited via print advertisements. The intervention was a training session consisting of four exercises with different focuses: mobility, strength and balance. One session was guided by a physiotherapist (P). The other session was guided by a digital application (A), including a pedagogical agent to foster training knowledge, videos for exercise instructions, and feedback mechanisms allowing adaptation of training intensity. Outcomes of interest were movement quality (MQ), exercise specific self-efficacy (ESE) and control competence for physical training (CCPT). MQ was assessed by rating the video recorded exercises. ESE and CCPT were assessed prior and after each training session via self-administered questionnaires. Participants were randomly assigned to one of two exercise treatment sequences. One sequence started with the digital application, the other sequence started with the intervention supervised by a physiotherapist. The wash-out phase between sessions was set to a minimum of three weeks. Participants, assessors and statisticians were not blinded to the treatment sequence and the type of intervention. Participants did not know which start sequence they were assigned to before the start of session 1. The digital’s application non-inferiority is given if the upper limit of the 95%-Confidence Interval (CI) of the effect size Hedges GIG for independent groups does not exceed the non-inferiority margin of G IG <0.2. Results: 54 participants started the first training session (32 female, 22 male; mean age 62.4, SD 8.2 years). Both start sequences were similar in size (PA: n=26; AP: n=28). 7 subjects did not attend the second training session (PA: n=3; AP: n=4). 47 people completed both training sessions. The individual time period between T1 and T2 was 27 to 42 days with an average time of 34.7 days. The app is inferior to the physiotherapist in all outcomes considered, except MQ of the mobility exercise (GIG=-0.13, CI -0.16, -0.09). In contrast to the strengthening exercises in two positions (supine GIG=0.75, CI 0.70, 0.81; table GIG=1.19, CI 1.13, 1.25)), MQ of balance exercise is close to non-inferiority (GIG= 0.15, CI 0.13, 0.20). ESE shows a strong effect on the side of the physiotherapist (GIG = 0.84, 95% CI 0.78, 0.89). In terms of CCPT the app is only slightly inferior to the physiotherapist (GIG=0.18, 95% 0.13, 0.22). Conclusions: Despite the absence of non-inferiority in almost any measure of interest, ESE and CCPT do also improve by the use of the app and MQ is acceptable for exercises that are easy to conduct. However relevant differences in movement quality are present in challenging tasks. The digital app therefore opens up possibilities for a supplementary tool to support the patient in independent home training for less complex exercises; however it cannot replace a physiotherapist equivalently. The trial was registered in the German Clinical Trial Register (DRKS00015759) and funded by the Leibniz-WissenschaftsCampus Tuebingen "Cognitive Interface" with funds from the Ministry of Science, Research and the Arts Baden-Wuerttemberg.