Outcome Assessment in Frozen Shoulder Trials: A Systematic Review and COSMIN-Informed Recommendations.
TL;DR
Future frozen shoulder RCTs should avoid relying on a single pain score, ROM alone, or non-standard global labels as the sole primary outcome, and a pragmatic minimum structure should combine one shoulder-specific PROM, one pain measure, and key directional ROM outcomes.
Abstract
Background
Outcome assessment in randomized controlled trials (RCTs) of frozen shoulder remains heterogeneous, limiting comparability across studies and reducing the usefulness of evidence synthesis. Although pain intensity and range of motion (ROM) are frequently reported, the relationship between common use and measurement-property evidence has not been sufficiently clarified.
Objective
To map outcome measures used in frozen shoulder RCTs, summarize common reporting combinations, and provide COSMIN-informed, evidence-cautious recommendations for future trial outcome selection.
Methods
PubMed, Embase, the Cochrane Library, CNKI, and VIP were searched from inception to January 31, 2026. Outcome measures were extracted from eligible frozen shoulder RCTs and classified using a taxonomy that distinguished outcome domains, standardized patient-reported outcome measures (PROMs), composite clinical scores, single-item indices, objective physical measures, and non-standard labels. Frequency and co-occurrence patterns were summarized descriptively. A separate COSMIN-informed appraisal was conducted for clearly identifiable high-frequency standardized instruments and clinically central measure families with retrievable measurement-property evidence. This appraisal used COSMIN concepts to structure interpretation but was not a full COSMIN systematic review with formal COSMIN-GRADE certainty ratings.
Results
A total of 927 frozen shoulder RCTs were included in the outcome-use mapping. Outcome assessment centered on pain, mobility/ROM, and shoulder function. ROM was reported in 448 studies (48.3%), VAS/NRS pain in 327 (35.3%), global improvement or total effective rate in 260 (28.0%), Constant-Murley Score (CMS) in 168 (18.1%), and SPADI in 80 (8.6%). ROM plus VAS/NRS pain was the most common pairing (198 studies, 21.4%). The COSMIN-informed appraisal synthesized 10 supporting evidence entries from measurement-property sources involving six instruments or measure families: SPADI, DASH/QuickDASH, CMS, ASES, VAS/NRS pain, and ROM. SPADI had comparatively more complete frozen shoulder-specific evidence among commonly used shoulder-specific PROMs, although evidence for content validity, responsiveness, interpretability, and minimal important change remains limited.
Conclusions
Future frozen shoulder RCTs should avoid relying on a single pain score, ROM alone, or non-standard global labels as the sole primary outcome. A pragmatic minimum structure should combine one shoulder-specific PROM, one pain measure, and key directional ROM outcomes, with additional instruments selected according to disease stage, follow-up timing, and study purpose. These recommendations are evidence-informed and provisional rather than a formal consensus-based core outcome set.