Measurement properties of patient-reported outcome measures for rotator cuff repair surgery

Rotator cuff disease affects over 50% of individuals over age 70, with more than 250,000 rotator cuff repair (RCR) surgeries performed annually in the U.S. Despite its benefits, 20%-30% of repairs fail, and patient-reported outcome measures (PROMs) are widely used to assess recovery, though none are fully validated for RCR-specific use.

This study evaluates the psychometric properties of the PENN Shoulder Score (PSS), modified American Shoulder and Elbow Surgeons (ASES) score, and Single Assessment Numeric Evaluation (SANE) in RCR patients to determine their validity, reliability, responsiveness, and interpretability.

A retrospective analysis was conducted using the Cleveland Clinic's Outcomes Measurement and Evaluation database, including 4,909 patients who underwent primary arthroscopic RCR for superior-posterior tears between 2015 and 2024. A subset of 112 patients participated in an NIH R01 study, enabling test-retest reliability analysis. Measurement properties were assessed per COSMIN guidelines, including internal consistency (McDonald's ω ≥ 0.70), test-retest reliability (intraclass correlation coefficient, ICC ≥ 0.70), known-groups and convergent validity, structural validity, responsiveness, and interpretability (Patient Acceptable Symptom State, minimal clinically important difference, minimum detectable change).

ASES score and PSS demonstrated high internal consistency (ω = 0.86-0.97). Test-retest reliability was strong, except for PSS-Satisfaction (ICC = 0.34-0.42), SANE at baseline (ICC = 0.38), and ASES-Pain at 1 year (ICC = 0.28). Known-groups validity confirmed that larger tears were associated with worse pain and function scores preoperatively. Convergent validity showed strong correlations between ASES and PSS scores (r = 0.83-0.98), while SANE had weaker correlations at baseline (r = 0.46-0.56). Structural validity analyses indicated acceptable model fit for ASES and PSS function items. Responsiveness was excellent across all PROMs (standardized response means= 1.34-2.05). PSS outperformed other PROMs, meeting all validation criteria at baseline and 75% at 1 year.

This study provides a comprehensive psychometric evaluation of three widely used shoulder PROMs in RCR patients. While ASES and PSS were valid, reliable, and responsive, SANE's single-item format limited its effectiveness. The PSS demonstrated superior measurement properties, making it the most robust PROM for assessing RCR outcomes. These findings guide PROM selection for clinical and research applications, though unexamined factors such as patient response burden should also be considered.