Ultrasound detection of rotator cuff tears: observer agreement related to increasing experience.

OBJECTIVE Ultrasound of the rotator cuff is considered to be operator-dependent with its accuracy being related to the operator's level of experience. This study was conducted to test the hypothesis that ultrasound performed by operators with different levels of experience will give nonreproducible results. SUBJECTS AND METHODS Two radiologists, one general radiologist with no experience in musculoskeletal ultrasound and one experienced musculoskeletal radiologist, independently performed ultrasound on 200 shoulders in 183 consecutive patients. Agreement was assessed. Cohen's kappa values with standard errors were calculated. In 71 patients the ultrasound findings could be related to surgical findings. RESULTS The diagnoses of full-thickness and partial-thickness rotator cuff tears were made with agreements of 98% (κ-value: 0.95 [standard error, 0.03]) and 90% (κ-value: 0.79 [0.05]), respectively. Agreement for full-thickness tears was constant; the agreement for partial-thickness tears improved from 80% to 98% in the last quarter of the study period. Based on the 71 patients who underwent shoulder surgery, sensitivity, specificity, and accuracy for detecting full-thickness tears by the experienced and general radiologists were 94%, 94%, and 94% and 89%, 91%, and 90%, respectively. Sensitivity, specificity, and accuracy for detecting partial-thickness tears by the experienced and general radiologists were 100%, 32%, and 57% and 84%, 35%, and 53%, respectively. CONCLUSION The hypothesis that ultrasound of the shoulder is operator-dependent and related to experience was refuted. In this study, there was excellent agreement for the detection of rotator cuff tears, which only slightly improved with the increasing experience of the general radiologist. Accuracy of rotator cuff tear detection was high and in accordance with the results in the literature.