Measurement Properties of the Oxford Shoulder Score and Minimal Clinically Important Changes After Primary Total Shoulder Replacement Surgery

Abstract

ObjectiveTo evaluate the measurement properties of the Oxford Shoulder Score (OSS) and estimate the minimal clinically important change (MCIC) in patients undergoing primary total shoulder replacement in Australia.MethodsDe‐identified data from the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR) were used for this analysis. Pre‐ and 6‐month post‐operative OSS scores were used, with EQ‐5D‐5L and shoulder pain scores used as comparators. Floor and ceiling effects, internal consistency reliability, construct validity and responsiveness to change were evaluated using standard psychometric methods. Mean change and predictive modelling approaches (with and without adjustment for the proportion of improved patients) were used to calculate MCIC thresholds, with patient‐perceived improvement after surgery as the anchor.ResultsPre‐operative OSS data were available for 1,117 patients (59% female; 90% aged ≥ 60 years) undergoing primary total shoulder replacement. No floor or ceiling effects were observed pre‐ or post‐operatively. The OSS showed high internal consistency reliability (Cronbach alpha >0.89), good construct validity and high responsiveness to change (effect size 1.88). The MCIC derived from the mean change method was 6.50 points (95% confidence interval [CI] 4.41, 8.61). The predictive modelling approach produced an MCIC estimate of 8.42 points (95% CI 5.68, 12.23) after adjustment.ConclusionThe OSS has good measurement properties to capture pain and function outcomes following shoulder replacement procedures, and is highly responsive to change. Based on robust methods, an increase in OSS scores of at least 8 points can be considered as meaningful improvement after surgery from the patient's perspective.