Abstract
BACKGROUND: Rotator cuff lesions are common in the community but reproducibility of tests for shoulder assessment has not been adequately appraised and there is no uniform approach to their use. OBJECTIVE: To study interrater reproducibility of standard tests for shoulder evaluation among a rheumatology specialist, rheumatology trainee, and research nurse. METHODS: 136 patients were reviewed over 12 months at a major teaching hospital. The three assessors examined each patient in random order and were unaware of each other's evaluation. Each shoulder was examined in a standard manner by recognised tests for specific lesions and a diagnostic algorithm was used. Between-observer agreement was determined by calculating Cohen's kappa coefficients (measuring agreement beyond that expected by chance). RESULTS: Fair to substantial agreement was obtained for the observations of tenderness, painful arc, and external rotation. Tests for supraspinatus and subscapularis also showed at least fair agreement between observers. 40/55 (73%) kappa coefficient assessments were rated at >0.2, indicating at least fair concordance between observers; 21/55 (38%) were rated at >0.4, indicating at least moderate concordance between observers. CONCLUSION: The reproducibility of certain tests, employed by observers of varying experience, in the assessment of the rotator cuff and general shoulder disease was determined. This has implications for delegation of shoulder assessment to nurse specialists, the development of a simplified evaluation schedule for general practitioners, and uniformity in epidemiological research studies.