High reproducibility in the interpretation of intraoperative transesophageal echocardiographic evaluation of aortic atheromatous disease.

Intraoperative decisions are often based on interpretation of results from transesophageal echocardiography (TEE). One such area is the intraoperative evaluation of atheromatous disease of the thoracic aorta and subsequent classification or grading. These grading schemes are predictive of stroke after cardiac surgery. Since intraoperative strategies may be modified based on this TEE aortic atheroma grading, assessment of the interobserver variability of TEE findings between observers is essential. Forty TEE videotape segments imaging three portions of the thoracic aorta (ascending, arch, descending) were selected from 189 reports of a larger cohort. Three independent, blinded observers, experienced in TEE, evaluated these examinations for atheroma severity. If a TEE segment had insufficient data, "uninterpretable" was recorded. Weighted kappa coefficients of agreement were calculated on the three data sets. Mean weighted kappa coefficients for the three observers A, B, and C were 0.69, 0.74, and 0.72, for the ascending, arch, and descending aorta segments, respectively, representing excellent agreement. We have demonstrated uniformly high agreement for interpretation of TEE, which indicates the excellent reproducibility of TEE grading and stratification of aortic atheroma. Reproducibility within and across specialties and institutions is essential for widespread application of TEE for evaluation of the thoracic aorta.