Low-Supervision Asynchronously Monitored Home Rehabilitation for Adhesive Capsulitis: A Pilot Randomized Study

Background: Adhesive capsulitis often requires prolonged rehabilitation, and home exercise performed without supervision is difficult to document objectively. This pilot randomized trial evaluated the preliminary clinical effects and system-recorded engagement of a low-supervision, asynchronously monitored home rehabilitation system used as an adjunct to conventional outpatient rehabilitation. Methods: Thirty adults with adhesive capsulitis were randomly assigned 1:1 at a university-affiliated outpatient rehabilitation department. Both groups received conventional outpatient rehabilitation twice weekly for 4 weeks; the intervention group additionally used a gamified, smartphone-based home program (HappyGoGo) with asynchronous system monitoring, and the control group received standard home exercise instructions. Because the outcome hierarchy was assigned at retrospective registration, after data collection and analysis had been completed, all outcomes are reported as exploratory. Fourteen clinical outcomes were analyzed by analysis of covariance adjusting for the corresponding baseline value, with Benjamini–Hochberg correction; unadjusted change scores were compared as a sensitivity analysis. Results: Substantial baseline imbalance was present, with the intervention group more restricted in range of motion and reporting higher pain. After adjustment, no outcome reached nominal significance and the 95% confidence interval included the null for all 14 outcomes. The largest estimate was for pain (adjusted mean difference −1.30 points, 95% CI −2.64 to 0.04; p = 0.057). In the unadjusted analysis, pain and internal rotation reached nominal significance but neither survived correction for multiple comparisons. All 30 participants completed the protocol, no adverse events were reported, and engagement was retrieved for 14 of 15 intervention participants (mean 61.0%, range 7.1–96.4%). Conclusions: This trial found no detectable between-group difference on any clinical outcome, with confidence intervals too wide to determine whether a clinically meaningful effect exists. The system could be implemented alongside routine care without adverse events, and participation could be documented automatically, though it varied widely between individuals.