Interobserver agreement in the classification of rotator cuff tears.

Abstract

Six classification systems have been proposed for describing rotator cuff tears designed to help understand their natural history and make treatment decisions. To assess the interobserver variation for these classification systems and identify the method with the best interobserver agreement. Cohort study (diagnosis); Level of evidence, 2. Six rotator cuff tear classification systems were identified in a literature search. The components of these systems included partial-thickness rotator cuff tears and classification by size, shape, configuration, number of tendons involved, and by extent, topography, and nature of the biceps. Twelve fellowship-trained orthopaedic surgeons who each perform at least 30 rotator cuff repairs per year reviewed arthroscopy videos from 30 patients with a random assortment of rotator cuff tears and classified them by the 6 classification systems. Interobserver variation was determined by a kappa analysis. Interobserver agreement was high when distinguishing between full-thickness and partial-thickness tears (0.95, kappa = 0.85). The investigators agreed on the side (articular vs bursal) of involvement for partial-thickness tears (observed agreement 0.92, kappa = 0.85) but could not agree when classifying the depth of the partial-thickness tear (observed agreement 0.49, kappa = 0.19). The best agreement for full-thickness tears was seen when the tear was classified by topography (degree of retraction) in the frontal plane (observed agreement 0.70, kappa = 0.54). With the exception of distinguishing partial-thickness from full-thickness rotator cuff tears and identifying the side (articular vs bursal) of involvement with partial-thickness tears, currently described rotator cuff classification systems have little interobserver agreement among experienced shoulder surgeons. Researchers should consider describing full-thickness rotator cuff tears by topography (degree of retraction) in the frontal plane.