What are small, medium and large effect sizes for exercise treatments of tendinopathy? A systematic review and meta-analysis

ObjectiveTo quantify and describe effect size distributions from exercise therapies across a range of tendinopathies and outcome domains to inform future research and clinical practice through conducting a systematic review with meta-analysis.DesignSystematic review with meta-analysis exploring moderating effects and context-specific small, medium and large thresholds.Eligibility criteriaRandomised and quasi-randomised controlled trials involving any persons with a diagnosis of rotator cuff, lateral elbow, patellar, Achilles or gluteal tendinopathy of any severity or duration.MethodsCommon databases, six trial registries and six grey literature databases were searched on 18 January 2021 (PROSPERO: CRD42020168187). Standardised mean difference (SMDpre) effect sizes were used with Bayesian hierarchical meta-analysis models to calculate the 0.25 (small), 0.5 (medium) and 0.75 quantiles (large) and compare pooled means across potential moderators. Risk of bias was assessed with Cochrane's Risk of Bias tool.ResultsData were obtained from 114 studies comprising 171 treatment arms 4104 participants. SMDpre effect sizes were similar across tendinopathies but varied across outcome domains. Greater threshold values were obtained for self-reported measures of pain (small=0.5, medium=0.9 and large=1.4), disability (small=0.6, medium=1.0 and large=1.5) and function (small=0.6, medium=1.1 and large=1.8) and lower threshold values obtained for quality of life (small=-0.2, medium=0.3 and large=0.7) and objective measures of physical function (small=0.2, medium=0.4 and large=0.7). Potential moderating effects of assessment duration, exercise supervision and symptom duration were also identified, with greater pooled mean effect sizes estimated for longer assessment durations, supervised therapies and studies comprising patients with shorter symptom durations.ConclusionThe effect size of exercise on tendinopathy is dependent on the type of outcome measure assessed. Threshold values presented here can be used to guide interpretation and assist with further research better establishing minimal important change.