I have read with interest the report recently published by Yarmohammadi et al evaluating the role of the CHADS ₂ score in the risk stratification of patients with nonvalvular atrial fibrillation (AF) who undergo direct-current cardioversion. The investigators stated that “the value of CHADS ₂ scoring system in the risk stratification of patients with AF who undergo direct-current cardioversion has not yet been specifically investigated.” That is not accurate.

My group recently published in Europace a study whose aim was to identify possible ways of refining the CHADS ₂ score’s diagnostic accuracy for this purpose, namely, through the addition of echocardiographic parameters (left atrial area and the left ventricular ejection fraction) and the CHA ₂ DS ₂ -VASc risk score in this setting. Our sample was composed of 376 patients who underwent transesophageal echocardiography during AF episodes, with an average CHADS ₂ score of 2.2 ± 1.2. In 93.6% (n = 352), the indication for transesophageal echocardiography was the exclusion of left atrial appendage (LAA) thrombus before direct-current cardioversion. Thrombi were found in 9.8% (n = 37), and the prevalence of thrombus increased along with CHADS ₂ score: 4.2% (1 of 24) in patients with CHADS ₂ scores of 0, 7.8% (17 of 230) in those with CHADS ₂ scores of 0 to 2, and 13.7% (20 of 146) in those with CHADS ₂ scores of 3 to 6 (chi-square for trend: γ = 0.31, p = 0.018).

In the Assessment of Cardioversion Using Transesophageal Echocardiography (ACUTE) trial substudy, a higher prevalence of LAA thrombus in patients with CHADS ₂ scores of 0 was found: 10% (14 of 138). Moreover, unlike in our study, the prevalence of LAA thrombus was not significantly different among patients with different CHADS ₂ scores (p = 0.52). In our sample, patients with CHADS ₂ scores of 0 or 1 accounted for 27.4%, much lower than in the study of Yarmohammadi et al, in which these patients represented 63% of the sample. Despite this fact, our population did not have a very high thromboembolic risk. It was similar to the risk observed in the Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) and Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trials and lower than in the Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF). Therefore, it is possible that this underrepresentation of higher risk patients in the ACUTE trial substudy may have led to the observed lack of an association between increasing CHADS ₂ score and higher prevalence of LAA thrombus. Another possible explanation may be an underestimation of risk in some patients with CHADS ₂ scores of 0, who may have undiagnosed arterial hypertension.

In the ACUTE trial, patients anticoagulated for >7 days were not included in the analysis. In our sample, only a minority of patients were effectively anticoagulated (22.6% of all patients and 58.2% of those receiving oral anticoagulation). However, this is unlikely to have had an influence on our results, because a similar prevalence of LAA thrombus was observed in the 2 studies (9.8% vs 12.4%). Furthermore, in our sample, only 38.0% of subjects had enoxaparin before transesophageal echocardiography. In the ACUTE trial substudy, this figure was 66.8%.

A careful analysis of patients with CHADS ₂ scores of 0 and LAA thrombi was performed by Yarmohammadi et al. Lower values of the left ventricular ejection fraction were found in these patients. This supports our findings, in which the addition of the left ventricular ejection fraction and left atrial area to the CHADS ₂ score led to an overall improvement in the detection of LAA thrombus (with the area under the receiver-operating characteristic curve increasing from 0.62 to 0.73, p = 0.003) and no patients with LAA thrombus being classified as having scores of 0.

In our sample, the CHA ₂ DS ₂ -VASc score displayed higher sensitivity compared to the CHADS ₂ score in classifying all patients with LAA thrombi with scores ≥1 at expense of lower specificity, confirming its overinclusive behavior (only 2.9% of patients were classified at low risk, score = 0). Still, in light of current knowledge, we should be cautious and avoid selecting patients for transesophageal echocardiography before direct-current cardioversion on the basis of the values of these scores, until the results of these preliminary studies are further reproduced and validated.