Utility and Responsiveness of Objective and Subjective Outcomes Following Shoulder Surgery: A Systematic Review and Meta-analysis

Abstract

Postoperative recovery following shoulder surgery is assessed using a wide array of patient-reported and objective outcome measures, yet no standardized framework guides their selection across clinical contexts. Outcome instruments are frequently applied interchangeably despite being developed for distinct pathologies, procedures, and populations, limiting cross-study comparability and clinical interpretation. To quantify the responsiveness of commonly used shoulder outcome measures, both patient-reported outcomes and the objective metric of range of motion, and to examine the consistency of that responsiveness across age groups, injury types, surgical procedures, and follow-up intervals. Systematic review and meta-analysis; Level of evidence, 2. PubMed, Embase, and Scopus were searched for level 1 to 2 studies published between 2014 and 2024 reporting preoperative and postoperative patient-reported and objective outcome measures following shoulder surgery with a minimum 1-year follow-up. For each measure, responsiveness was quantified as the standardized pre-to-post change and pooled with a random-effects model, and the mean change in native units was reported secondarily. Heterogeneity was summarized with the I 2 statistic, and a 95% prediction interval described the range of responsiveness expected in a future study. Subgroup analyses by age, pathology, procedure, and follow-up interval were exploratory and not adjusted for multiple comparisons. A total of 137 studies were included, and 13 outcome measures met criteria for pooling. Every measure was responsive to surgery, with the pooled standardized change ranging from 0.9 for the Disabilities of the Arm, Shoulder and Hand score to 8.5 for the Rowe score, and all confidence intervals excluded zero. Between-study heterogeneity remained high throughout, with I 2 values above 95% for most measures, and all 13 prediction intervals included zero, indicating that responsiveness, although positive on average, cannot be assumed in any individual future study. Only 1 subgroup contrast, shoulder abduction by follow-up interval, reached nominal significance ( P = .03), and no contrast was robust to the number of comparisons performed. Commonly used shoulder outcome measures were responsive to surgery, but the magnitude of that responsiveness varied widely across studies, and no single measure performed uniformly across clinical contexts. These findings corroborate and extend prior psychometric work across a broader data set and caution against using outcome measures interchangeably across distinct clinical contexts.