Background: For severe mitral valve (MV) degenerative disease, repair is recommended. Prediction of repair complexity and referral to centers of excellence can increase rates of successful repair. This study sought to demonstrate that TEE is a feasible imaging modality to predict the surgical MV complexity score previously developed by Anyanwu et al. Methods: Two hundred TEE examinations of patients who underwent MV repair (2009 – 2011) were retrospectively reviewed and scored by two cardiac anesthesiologists. TEE scores were compared to surgical complexity scores of same subset of patients. Kappa values were reported for the agreement of TEE and surgical scores. McNemar’s tests were used to test the homogeneity of the marginal probabilities of different scoring categories. Results: TEE scores were slightly lower (2[1,3]) than surgical scores (3[1,4]). Agreement was 66% between the scoring methods, with a moderate kappa (0.46). Using surgical scores as the gold standard, 70%, 71% and 46% of simple, intermediate and complex surgical scores, respectively, were correctly scored by TEE. P1, P2, P3, and A2 prolapse were easiest to identify with TEE and had the highest agreement with surgical scoring (P1 agreement 79% with kappa 0.55, P2 96% (kappa 0.8), P3 77% (kappa 0.51), A2 88% (kappa 0.6)). The lowest agreement between the two scores occurred with A1 prolapse (kappa 0.05) and posteromedial commissure prolapse (kappa 0.14) (Figure 3). In the presence of significant disagreement, TEE scores were more likely to be of higher complexity than surgical. McNemar’s test was significant for prolapse of P1 (p=0.005), A1 (p=0.025), A2 (p=0.041), and the posteromedial commissure (p<.0001).
