Reliable interobserver and intraobserver agreement of the International Society of Arthroscopy, Knee Surgery and Orthopaedic Sports Medicine (ISAKOS) classification system of rotator cuff tears.

OBJECTIVE

The ISAKOS Shoulder Committee developed a new comprehensive classification system aimed to describe all rotator cuff tears. The five characteristics of the tears included are pattern (P), extension (E), fatty atrophy (A), retraction (R), and location (L), conforming to the acronym "PEARL." The objective of this study is to assess if the ISAKOS Rotator Cuff Tear Classification System is reliable by measuring the intraobserver and interobserver multirater reliability.

METHODS

Arthroscopic videos of 36 rotator cuff tears, including tears of varying sizes and configurations, were evaluated twice by four surgeons from different continents blinded to tear type. Intraobserver and interobserver reliability was measured using multirater and intraobserver Kappa coefficients.

RESULTS

Intraobserver reliability: An almost perfect agreement for "location" (κ = 0.98), substantial agreement for "extension" (κ = 0.73) were obtained for mean intrarater kappa, lower κ agreement for "pattern" (κ = 0.58), and relatively high agreement of 0.79 for "retraction." Every characteristic of "location" had an almost perfect agreement among the surgeons (κ = 0.91). Intrarrater reliability: In the partial-thickness posterosuperior tears "location," there was good agreement in tears involving less than 50% of the tendon (κ = 0.74) and moderate in those deeper than 50% of the tendon thickness (κ = 0.58). "Extension" in full-thickness posterosuperior RCT achieved moderate agreement. Within the anterior subscapularis tears, we have a substantial agreement in Lafosse´s Type 1 (κ = 0.73), moderate in Types 2 and 3 (κ = 0.45 and κ = 0.46) and slight agreement in Type 4 (κ = 0.06). The overall kappa summarizing all categories indicated moderate agreement (κ = 0.52). The articular pattern in the partial-thickness posterosuperior RCT "location" had a perfect kappa of 1, while the bursal pattern showed an almost complete agreement (κ = 0.87). In the full-thickness posterosuperior RCTs, we observed fair agreement in C, U, and reverse L configurations and slight agreement in L pattern (κ = 0.18). The overall kappa is 0.44 with a confidence interval of 0.41-0.47. There was substantial agreement in every category and the overall kappa for "retraction" (k = 0.70).

CONCLUSION

ISAKOS rotator cuff tear classification system provides sufficient interobserver reliability for communicating among surgeons and for pooling of data from clinical studies.

STUDY DESIGN

Cohort study (diagnosis); Level of evidence, 2.