Interobserver and intraobserver reliability of the Goutallier classification using magnetic resonance imaging: proposal of a simplified classification system to increase reliability.

Background: The Goutallier classification of fatty infiltration of the rotator cuff was developed for use in axial computed tomography arthrography. Now the Goutallier classification is being used with magnetic resonance imaging (MRI). Not only is there debate on the validity of the use of this system in MRI, but current literature is unclear as to the clinical use of the Goutallier classification.

Hypothesis: There will be significant inter- and intraobserver variability of the Goutallier classification grading system for patients with chronic rotator cuff tears.

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

Methods: An online database consisting of 35 single MRI images from 35 patients with chronic rotator cuff tears was designed and sent to members of the American Shoulder and Elbow Society. Surgeons were asked to identify the stage of rotator cuff fatty infiltration using the Goutallier classification system. Thirty surgeons responded and completed the survey in its entirety. At a minimum of 2 months later, 28 of the 30 initial respondents completed evaluations of the same online database. The responding surgeons were divided dichotomously according to their demographics and the interobserver reliability of the groups compared. A kappa analysis was performed to determine inter- and intraobserver reliability using 95% confidence intervals (95% CIs). A simplified 3-tiered classification was proposed combining Goutallier grades 0 and 1 as well as grades 2 and 3.

Results: Statistical analysis of all respondent data demonstrated moderate intraobserver variability with a κ value of 0.56 (95% CI, 0.53-0.60). In addition, moderate interobserver variability was shown with a κ value of 0.43 (range, 0.16-0.74). With the simplified classification, intraobserver reliability was 0.70 (95% CI, 0.66-0.74) and interobserver reliability was 0.61 (range, 0.21-0.87). Correlation analysis showed no correlation with the presence or absence of fellowship training or board certification with either the Goutallier classification or the proposed modification (P > .05). Sports versus shoulder/elbow fellows had statistically better intraobserver variability (κ = 0.63 vs 0.50) with the Goutallier classification. Years in practice was negatively correlated with the level of agreement for both classifications (-r value, P < .05). The number of rotator cuff repairs performed per year negatively correlated with the level of agreement in the proposed modification only (r = -0.44, P = .022). Percent of practice dedicated to the shoulder did not correlate significantly with either classification (P > .05).

Conclusion: There is significant inter- and intraobserver variability observed among experienced shoulder surgeons using the Goutallier classification for assessing fatty infiltration of the rotator cuff muscles after chronic rotator cuff tears. Respondents were more likely to agree with themselves than with other respondents. A simplification of the MRI classification system is proposed that takes into consideration the variability determined by this study.