Reproducibility and reliability of the Snyder classification of superior labral anterior posterior lesions among shoulder surgeons.

BACKGROUND Previous studies of the reliability of classifications for superior labral anterior and posterior (SLAP) lesions suggest that intraobserver/interobserver reliability is poor. PURPOSE The goals were to (1) evaluate intraobserver/interobserver reliability of the Snyder classification of labrum tears among experienced surgeons, (2) determine the effect of simplifying that classification into normal versus abnormal labrums, (3) determine the reliability of subdividing type II SLAP lesions, and (4) evaluate the effect of videotape quality on diagnostic confidence. STUDY DESIGN Cohort study (diagnosis); Level of evidence, 3. METHODS At 2 separate times, 5 experienced shoulder surgeons (all >10 years of practice) evaluated 90 videotapes of shoulder arthroscopy performed for SLAP lesions to identify lesion type, to grade video quality, and to determine degree of confidence in diagnosis. The results were compared for intraobserver reliability (κ analysis) and interobserver reliability (intraclass correlation coefficients) and effect of video quality on diagnostic confidence. Significance was set at P < .05. RESULTS When the choices were normal labrum or the 4 types of SLAP lesions, the intraobserver agreements (mean κ, 0.670) and interobserver agreements (mean correlation, 0.804) were substantial. When the labrums were divided into normal (normal and type I) and abnormal (types II-IV), the intraobserver agreements (mean κ, 0.792) and interobserver agreements (mean correlation, 0.648) were substantial. When the 3 type II SLAP lesion subvariants were evaluated, the intraobserver agreement (mean κ, 0.598) was moderate, and the interobserver agreement (mean correlation, 0.804) was substantial. The confidence of the diagnosis was highly correlated with the perceived video quality (average Pearson correlation, 0.718; P < .01). Repeated measures analysis showed a significant relationship between confidence and quality (parameter estimate, 0.732; standard error, 0.021; P < .01) adjusting for rater and review. CONCLUSION For experienced shoulder surgeons, the Snyder classification is a reliable system for identifying SLAP lesions.