Active treatments outperform minimal intervention for adults with rotator cuff tendinopathy: a systematic review with predictive and network meta-analyses of complex interventions

OBJECTIVE To guide treatment of adults with rotator cuff tendinopathy (RoCuTe) by evaluating the relative efficacy of treatments, benchmarked against minimal intervention, for the co-primary outcomes of pain, function and quality-of-life (QoL) across short, medium, and long-term follow-up. DESIGN Systematic review with Bayesian predictive and network meta-analyses for synthesising complex interventions, guided by stakeholder involvement. FUNDING Private Physiotherapy Education Foundation (UK) Silver Jubilee Award. DATA SOURCES PubMed, Embase, Web of Science, CINAHL, and SPORTDiscus, searched to 22/8/2025. ELIGIBILITY CRITERIA FOR SELECTING STUDIES High-quality (PEDro score equal or above 7) randomised controlled trials comparing any intervention with another active or minimal intervention for patients clinically diagnosed with RoCuTe of either traumatic or insidious presentation; and reporting outcomes for pain, function and/or QoL. METHODS Title and abstract screening, full-text screening, and quality assessments were completed by two reviewers. Data extraction used the Elicit AI tool and was manually checked. Interventions were classified by treatment focus. Guided by patient and public involvement, pooled results from active interventions at short (1 to 12 weeks included), mid (>12 weeks to <12 months) and long-term (12 months included or more) were calculated for the primary analysis using Bayesian predictive meta-analysis models of within group change scores. Outcomes were benchmarked against an empirically derived minimal-intervention comparator (wait-and-see or sham). As a secondary analysis, network meta-analyses were conducted to synthesise relative effects and provide comparative rankings of active interventions. Risk of bias was assessed using the Cochrane Risk of Bias 2 tool, and certainty of evidence evaluated using GRADE. RESULTS We retained and analysed 140 high-quality studies that included 10,260 patients, 55.9% female, with a mean age of 48 (SD 8) years. Minimal interventions were associated with small short-term improvements, modest medium-term improvements and some regression in the long-term; in pain (0 to 100 scale: short=2.6; mid=23.3; long=21.1), function (standardised mean change (SMC): short=0.13; mid=0.87; long=0.76), and QoL (SMC: short=0.05; mid=0.33). At all timepoints, all active interventions with sufficient data were superior to minimal intervention for pain (0 to 100 scale: short = 18.1 to 37.9 [14 categories]; mid = 25.8 to 34.8 [8 categories]; long = 30.8 to 45.0 [6 categories]), function (SMC: short = 1.1 to 2.4 [14 categories]; mid = 1.1 to 2.0 [11 categories]; long = 1.0 to 1.8 [10 categories]), and QoL (short = 0.8 to 1.7 [7 categories]; mid = 0.9 to 1.8 [6 categories]). Certainty varied widely. Accordingly, three recommendation groups were defined based on the availability of comparative evidence and presence of higher-certainty findings. The strongest recommendation group included strengthening, range-of-motion exercises, complex interventions and movement pattern retraining. CONCLUSIONS A range of active treatments were superior to minimal intervention at each time point, so a wait-and-see approach should not be used, even in in the short-term. The most credible evidence was for interventions with a focus on strengthening, range-of-motion exercises, movement pattern retraining, and complex interventions. Clinicians should prioritise active management and deploy personalised clinical reasoning to tailor treatment to patient preferences and the available resources. SYSTEMATIC REVIEW REGISTRATION PROSPERO CRD42024584126