Poor Agreement on Classification and Treatment of Subscapularis Tendon Tears

Purpose: To assess the inter- and intraobserver agreement for classification and management of subscapularis tendon pathology based on arthroscopy and magnetic resonance imaging (MRI).

Methods: Twenty-two orthopaedic surgeons from the Multicenter Orthopaedic Outcomes Network (MOON) shoulder group reviewed still arthroscopic and MRI images of the subscapularis tendon from patients with a random assortment of subscapularis morphology. The surgeons were asked to classify the pathology based on 2 systems (Lafosse and Lyons) and choose whether they would repair the tendon and, if so, the method of repair (open or arthroscopic). The survey was administered 3 times to each surgeon. Inter- and intraobserver reliability between testing rounds was determined by kappa analysis.

Results: Interobserver reliability on classification of tears was poor based on MRI (k = 0.18 to 0.19) and fair based on arthroscopy (k = 0.26 to 0.29). Interobserver agreement on whether surgical treatment was indicated was fair for both MRI (k = 0.28) and arthroscopy (k = 0.38), while the agreement for type of surgery was poor based on MRI (k = 0.18) and fair based on arthroscopy (k = 0.28). Interobserver agreement did not improve when both MRI and arthroscopy were provided simultaneously (k = 0.24 to 0.30). Intraobserver reliability for classification and treatment was fair to moderate for both MRI (k = 0.32 to 0.50) and arthroscopic imaging (k = 0.39 to 0.56). When considering just those patients with normal tendons, surgeon agreement improved. For all questions, the arthroscopic images had a higher level of agreement among surgeons than the MRI (P < .001).

Conclusions: Although surgeons tended to have higher reliability when presented with arthroscopic images compared with MRI, there was very little agreement on the classification and management of subscapularis tendon tears.