We agree with Dr. Kearon that in the younger patient with atrial fibrillation who lacks clinical risk factors, TEE is unlikely to improve on risk stratification. Our comments on the potential role of TEE were restricted to a population at high clinical risk. The SPAF-III study specifically looked at this population (women >75 years of age or persons with hypertension, thromboembolism, left ventricular dysfunction, or congestive heart failure). In this population, dense spontaneous echocardiographic contrast and complex atherosclerotic plaque in the thoracic aorta were independent predictors of clinical thromboembolism [1]. Patients at high clinical risk for thromboembolism who lacked TEE evidence of both left atrial abnormalities and complex aortic plaque had an embolic rate of only 1.3% while receiving combination therapy. This value is similar to that expected in a population at low clinical risk. We therefore hypothesized that if these data are validated in clinical trials, patients with high clinical risk who lack adverse TEE findings may benefit from treatment with aspirin alone (rather than adjusted-dose warfarin).

Drs. Stollberger and Finsterer argue that we drew common conclusions from the SPAF-III [1] and ELAT [2] reports. We specifically highlighted the differences between the two studies with respect to clinical risk, the low incidences of thrombi (2.5% and 10%) and spontaneous echocardiographic contrast (12% and 63%) in the ELAT study compared with the SPAF-III study, and the surprising lack of association between left atrial thrombi and spontaneous echocardiographic contrast in the ELAT study. The latter included patients who were primarily at low clinical risk for whom anticoagulation was not controlled, whereas the SPAF-III study included participants at high clinical risk for whom anticoagulation was carefully controlled. We agree with Stollberger and Finsterer that patients in both studies who had atrial thrombi experienced a high stroke rate and that adjusted-dose warfarin was not of proven benefit. But only a minority (10%) of even the patients with high clinical risk have atrial thrombi. Far more patients have dense spontaneous echocardiographic contrast (20%) and complex aortic plaque (35%) [2]. These are groups for whom adjusted-dose warfarin was of proven benefit. Even more important, the absence of "high-risk" TEE indexes in the SPAF-III study sample identified a population at very low risk, for whom aspirin may be sufficient.

We do not advocate disregard for the clinical characteristics of hypertension, female sex, advanced age, previous thromboembolism, systolic dysfunction, and congestive heart failure. For patients with these risk factors, however, TEE seems to further discriminate between high- and low-risk subgroups.