The Embolism in Left Atrial Thrombi (ELAT) [1] and Stroke Prevention in Atrial Fibrillation (SPAF-III) [2] study groups recently published prospective studies that identified relations between transesophageal echocardiography (TEE) findings and risk for embolism in patients with nonvalvular atrial fibrillation. In the accompanying editorial [3], Manning and Douglas suggest that risk shown by TEE may supersede clinical risk with regard to clinical management and prognosis in patients with atrial fibrillation. However, the findings of the two original studies do not seem to justify this hypothesis.

The ELAT investigators found that the clinical risk factors of hypertension, previous stroke, and age were the strongest predictors of future embolism. Although they also found that many TEE findings were associated with embolism, these associations were not statistically significant after adjustment for differences in clinical risk factors. The SPAF-III investigators also found an association between TEE findings and embolism. However, although multiple analyses were done, only one of these associations (that with complex aortic plaque) achieved a conventional level of statistical significance (P < 0.05). Analyses were not performed to determine whether TEE findings were associated with embolism independent of the nature, number, and severity of individual clinical risk factors. Consequently, it is unknown from the SPAF-III study whether TEE findings or clinical factors were most strongly predictive of embolism. Although absence of a constellation of TEE abnormalities identified a subgroup of this clinically high-risk population who had a low incidence of embolism, it is not clear that the TEE criteria included in this constellation were assembled prospectively.

Patients with atrial fibrillation who are younger than 65 years of age and who have no clinical risk factors have an annual risk for embolism of only 1% in the absence of antithrombotic therapy [4]; it is unlikely that TEE will improve risk stratifications of such patients. Although it may be established that TEE usefully supplements clinical and transthoracic echocardiographic assessment of embolic risk in subgroups of patients with nonvalvular atrial fibrillation, it is unlikely to routinely supersede these evaluations. We should guard against more widespread use of TEE to evaluate and manage patients with nonvalvular atrial fibrillation until we have good evidence that this will result in improved clinical outcomes [5].