Inter-rater agreement of the Goutallier, Patte, and Warner classification scores using preoperative magnetic resonance imaging in patients with rotator cuff tears.

PURPOSE

The purpose of this study was to determine the interobserver reliability of 3 commonly used classification systems in describing preoperative magnetic resonance imaging (MRI) studies of patients undergoing surgery for full-thickness rotator cuff tears.

METHODS

Thirty-one patients who underwent arthroscopic rotator cuff repair and had preoperative MRI studies available were selected over a 2-year period. Three board-certified shoulder surgeons independently reviewed these images. Each was instructed in the published method for determining the Patte score on the T2 coronal images, supraspinatus and infraspinatus atrophy on the T1 sagittal images as described by Warner et al., and the Goutallier score of fatty infiltration of the supraspinatus on the T1 coronal/sagittal images. Statistical analysis was then performed to determine the interobserver agreement using the κ statistic, with the level of significance set a priori at P < .01.

RESULTS

None of the classification systems studied yielded excellent or high interobserver reliability. The strongest agreement was found with the Patte classification assessing tendon retraction in the frontal plane (κ = 0.58). The Goutallier classification, which grades fatty infiltration of the supraspinatus, showed moderate interobserver agreement (κ = 0.53) when dichotomized into none to mild (grades 0, 1, and 2) and moderate to severe (grades 3 and 4). Muscle atrophy of both the supraspinatus and infraspinatus yielded the worst interobserver reliability, with only 28% agreement.

CONCLUSIONS

The Goutallier, Patte, and Warner MRI classification systems for describing rotator cuff tears did not have high interobserver reliability among 3 experienced orthopaedic surgeons. Fatty infiltration of the supraspinatus and tendon retraction in the frontal planes showed only moderate reliability and moderate to high reliability, respectively. These findings have potential implications in the evaluation of the literature regarding the preoperative classification of rotator cuff tears and subsequent treatment algorithms.

LEVEL OF EVIDENCE

Level III, diagnostic agreement study with nonconsecutive patients.