Hybrid versus in-clinic rehabilitation for non-traumatic shoulder pain within a standardized digital care pathway: a prospective longitudinal observational study

Hybrid rehabilitation may increase access to structured physiotherapy for patients with longstanding non-traumatic shoulder pain who are unable to attend fully supervised in-clinic care. However, real-world comparisons are vulnerable to feasibility-based allocation and selection bias. This study described observed time to clinical discharge and functional outcomes in patients managed with hybrid or in-clinic rehabilitation within the same standardized digital care pathway. This secondary analysis used a prospectively assembled longitudinal observational cohort in a Spanish clinic. Recruitment and initiation of new rehabilitation episodes occurred from March through September 2025, and follow-up for the present analysis was defined through a common analytic cutoff of 11 November 2025. The same underlying cohort was previously reported for WMSSR implementation, within-episode clinical change, and test-battery optimization; the present analysis addressed rehabilitation delivery modality and time to documented clinical discharge. Hybrid rehabilitation comprised one supervised in-clinic and two unsupervised home-based sessions weekly; in-clinic rehabilitation comprised three supervised sessions weekly. Clinical discharge required a total deficit score < 4 points, symmetric strength, full functional range of motion, and no clinically meaningful pain during testing. The primary outcome was time to documented clinical discharge among all initiated episodes, analysed using accelerated failure time models with right censoring for episodes without documented discharge by the common analytic cutoff. Weibull and log-normal accelerated failure time models were adjusted for age, sex, baseline maximum pain, and baseline total deficit score. Secondary outcomes included discharge deficit score, Constant–Murley score, and longitudinal pain outcomes. Of 164 initiated rehabilitation episodes, 129 had documented clinical discharge by the common analytic cutoff, 24 had discontinued rehabilitation, and 11 remained under active follow-up. In the initiated-cohort analysis, the adjusted Weibull AFT model did not show a clear between-modality difference in time to documented clinical discharge (time ratio 0.92, 95% confidence interval 0.79–1.07). Among the 129 episodes with documented discharge, mean observed discharge time was 83.70 days in the hybrid group and 85.44 days in the in-clinic group. Mean deficit score decreased from 15.31 to 1.36 in the hybrid group and from 15.40 to 1.69 in the in-clinic group. Mean maximum pain decreased from 6.81 to 1.49 and from 6.37 to 1.94, respectively. Mean discharge Constant–Murley scores were 93.77 and 93.75. Longitudinal and sensitivity analyses did not show clear between-modality differences. In the initiated cohort, the adjusted time-to-discharge analysis did not show a clear between-modality difference, while secondary outcomes among completed episodes were of similar magnitude across rehabilitation formats. Because allocation was non-random and residual confounding and potentially informative censoring cannot be excluded, the findings do not establish equivalence, non-inferiority, or causal comparative effectiveness. ClinicalTrials.gov, NCT07374900. Retrospectively registered on 15 January 2026.