Atrial fibrillation (AF) is a common arrhythmia that represents an independent risk factor for strokes and is associated with marked morbidity and mortality (1). Recently, a newly developed scoring system, the CHA₂DS₂-VASc score, which extends the CHADS₂ scheme by considering additional stroke risk factors, was recommended to guide antithrombotic therapy in patients with AF (2- 3). However, these 2 scoring systems were developed and validated in AF patients who did not receive catheter ablations. This study aimed to investigate whether the CHADS₂ or CHA₂DS₂-VASc score is useful for predicting thromboembolic events and mortality after catheter ablation for AF.

A total of 565 consecutive patients with symptomatic drug-refractory AF who received radiofrequency catheter ablation were enrolled in the study. The CHADS₂ score was calculated for each patient (4). The CHA₂DS₂-VASc score was calculated for every patient based on a point system in which 2 points were assigned for a history of stroke or transient ischemic attack (TIA) or age ≥75 years. One point was assigned for age between 65 and 74 years; history of hypertension, diabetes, recent cardiac failure, and vascular disease (myocardial infarction, complex aortic plaque, or peripheral artery disease); and female sex (2).

Before catheter ablation, every patient underwent transesophageal echocardiography to confirm the absence of a left atrial thrombus. Details of ablation procedures are provided in the Online Appendix.

Before catheter ablation, the use of oral anticoagulation therapy with an adjusted dose of warfarin was based on the patients' CHADS₂ scores. Warfarin was intended to be prescribed for patients whose CHADS₂ scores were ≥2 to maintain the international normalized ratio (INR) of the prothrombin time between 2 and 3 (3). After catheter ablation, warfarin was continued for 3 months for patients who were already receiving warfarin treatment before ablation. After 3 months, discontinuation of warfarin was generally not considered for patients with CHADS₂ scores of ≥2. However, if patients remained symptom free, and there was no evidence of AF recurrences, replacement of warfarin with antiplatelet agents was considered. This was decided by the physicians responsible for treatment according to the individual characteristics of each patient.

The clinical endpoint was the combined occurrence of clinical events, including thromboembolic events (ischemic stroke, TIA, peripheral embolism, or pulmonary embolism) and death. Patients were followed-up every 1 to 3 months at our cardiology clinic or by referring physicians after catheter ablation. During each follow-up, patients were carefully examined, and 24-h Holter monitoring and/or cardiac event recording for 1 week were performed. AF recurrence was defined as an episode lasting longer than 1 min and confirmed by electrocardiograms 2 months after ablation. The status of patients who did not receive regular follow-up was assessed by chart reviews and telephone consultations to determine whether adverse events occurred. Our database was linked with the National Death Registry, and mortalities were further validated through a unique, lifelong personal identification number given to every Taiwan citizen.

Differences between continuous values were assessed using an unpaired 2-tailed t test for normally distributed continuous variables, the Mann-Whitney U test for skewed variables, and the chi-square test for nominal variables. Cox regression analysis was used to identify factors associated with adverse events. Details of the statistical analysis are provided in the Online Appendix.

Baseline characteristics of patients with and without events are shown in (Table 1). During the follow-up period of 39.2 ± 22.6 months, 27 patients (4.8%) experienced adverse events, including death in 9 patients, ischemic stroke in 9, TIA in 6, pulmonary embolisms in 2, and peripheral embolism in 1. Three of the 27 patients experienced events within the blanking period after catheter ablation. Significant predictors of adverse events based on the univariate Cox regression analysis are shown in (Table 2). The CHADS₂ and CHA₂DS₂-VASc scores remained independent predictors of adverse events in separate multivariate models (Tables 3, 4). Besides the CHADS₂ and CHA₂DS₂-VASc scores, AF recurrence after multiple procedures was a significant predictor of adverse events. The event rate was higher in patients with recurrences than in patients without recurrences (9.6% vs. 2.8%, p = 0.001). When events were further divided into ischemic stroke/TIA, other embolic events, and death, the CHADS₂ and CHA₂DS₂-VASc scores remained significant predictors. Hazard ratios of each increment of the CHADS₂ scores to predict ischemic stroke/TIA, other embolic events, and death were 1.893 (95% confidence interval [CI]: 1.364 to 2.627, p < 0.001), 2.306 (95% CI: 1.116 to 4.764, p = 0.024), and 1.786 (95% CI: 1.159 to 2.754, p = 0.009), respectively. Similarly, hazard ratios of each increment of the CHA₂DS₂-VASc scores to predict ischemic stroke/TIA, other embolic events, and death were 1.694 (95% CI: 1.321 to 2.173, p < 0.001), 2.088 (95% CI: 1.1865 to 3.6789, p = 0.011), and 1.551 (95% CI: 1.108 to 2.171, p = 0.010), respectively. The usefulness of these 2 scoring systems in predicting strokes/TIAs was consistent in patients with and without recurrences.

The event rates of patients continuously increased with increases in the CHADS₂ and CHA₂DS₂-VASc scores (Figure 1).(Figure 2) shows the receiver-operator characteristic (ROC) curves for predicting events after catheter ablation based on the CHADS₂ and CHA₂DS₂-VASc scores. At a cutoff point of 2 identified by the ROC curve, Kaplan-Meier survival analysis showed that patients with CHADS₂ scores of ≥2 (sensitivity, 59.3%; specificity, 83.5%) were associated with a higher event rate than the patients with CHADS₂ scores of <2 (15.2% vs. 2.4%, p < 0.001) during the follow-up period (Figure 3A). Furthermore, a CHA₂DS₂-VASc score of ≥2 identified by the ROC curve (sensitivity, 85.2%; specificity, 66.5%) also significantly predicted occurrences of events (11.3% vs. 1.1%, p < 0.001) (Figure 3B).

In the subgroup analysis of 460 patients with CHADS₂ scores of 0 or 1, 11 (2.4%) patients experienced adverse events. The event rate progressively increased from 0.6% in patients with CHA₂DS₂-VASc scores of 0 to 11.8% in patients with CHA₂DS₂-VASc scores of 3 (Figure 4). Using a CHA₂DS₂-VASc score of 2 as the cutoff point, patients with CHA₂DS₂-VASc scores of ≥2 were associated with a higher event rate as compared with patients whose CHA₂DS₂-VASc scores were <2 (7.1% vs. 1.1%, p = 0.003) (Figure 5).

The main findings were as follows: CHADS₂ and CHA₂DS₂-VASc scores were useful parameters for predicting adverse events after catheter ablation of AF; and the CHA₂DS₂-VASc score was helpful in further risk stratifications among patients with CHADS₂ scores of 0 or 1.

In the current study, both scores proved to be useful predictors of events after catheter ablation of AF in separate multivariate models. These results provide evidence for extending the usefulness of CHADS₂ and CHA₂DS₂-VASc scores to predict adverse events in AF patients who have received catheter ablation. In view of the ROC curves based on the CHADS₂ and CHA₂DS₂-VASc scores in predicting events, the differences between the areas under the curves did not reach statistical significance (p = 0.116). This may be because of the low event rate of the current study. Therefore, further trials with a larger sample size may be able to demonstrate a significant statistical difference with sufficient power.

Although the predictive accuracies of the CHADS₂ and CHA₂DS₂-VASc scores assessed by the ROC curves did not differ significantly, the CHA₂DS₂-VASc scores could further stratify the patients with CHADS₂ scores of 0 or 1 into 2 groups with different risks of events at the cutoff value of 2. According to the flowchart of the current guidelines of the European Society of Cardiology on the use of oral anticoagulants after catheter ablations, the CHADS₂ scheme should be used as a simple initial means of assessing stroke risk, and chronic oral anticoagulant therapy is recommended for patients with CHADS₂ scores of ≥2. For patients with a CHADS₂ score of 0 or 1, it is recommended that the CHA₂DS₂-VASc scoring system be used for a more comprehensive assessment of the risks (3). However, this recommendation was mainly based on consensus among experts, and the data in patients who received catheter ablation of AF were insufficient. The results of the present study may provide the evidence for the basis of the previously mentioned flowchart.

Oral et al. (5) reported that discontinuation of anticoagulant therapy appears to be safe after successful ablation of AF. Recently, Bunch et al. (6) reported that among patients with AF, the stroke risk was lower in patients who underwent catheter ablation than in those who did not. The 3-year stroke rate (2% to 3%) was similar to that of our study. However, these previous studies did not provide a useful method to identify patients with high risk of events. According to the results of the present study, the CHADS₂ scoring scheme can be used to screen patients with CHADS₂ scores of ≥2 who are predisposed to adverse events after catheter ablation. Furthermore, a more detailed CHA₂DS₂-VASc scoring scheme can further identify patients with risk of events among patients with CHADS₂ scores of <2. The clinical use of CHA₂DS₂-VASc scores should be emphasized because among patients with CHADS₂ scores of 0 or 1 who were traditionally assumed to be at low risk, those with CHA₂DS₂-VASc scores of 3 can have an event rate as high as 11.8%. Among the 9 patients who experienced stroke, 2 patients were receiving long-term medication with warfarin and had an INR within the therapeutic range at the time the events occurred. Five and 8 patients had CHADS₂ and CHA₂DS₂-VASc scores of ≥2, respectively. Of the 4 patients with CHADS₂ scores of 0 or 1, 3 had CHA₂DS₂-VASc scores of ≥2. This may further demonstrate the important role of the CHA₂DS₂-VASc score in identifying low-risk patients who are at risk of stroke; an anticoagulant should be prescribed to lower this risk. Furthermore, for patients with scores indicating a high risk of stroke and who are receiving warfarin treatment, it may be helpful to monitor the INR level more closely to avoid suboptimal treatment. However, a further prospective and large-scale trial is necessary to prove this supposition.

First, although the follow-up was not uniform for every patient, more than 70% of the study population underwent follow-up regularly at the cardiology clinics, and the remaining patients were completely assessed through telephone consultations. The accuracy of mortality events was further confirmed through the National Death Registry of Taiwan. Second, the strategy about the use of warfarin was determined by the physicians responsible for treatment according to the individual characteristics of each patient. However, the management was based on the same principle for individuals with similar risk. Moreover, long-term use of warfarin after catheter ablation was not a significant predictor of adverse events based on the analysis and therefore may not confound the results of the present study. Last, although only 18.6% of the patients in the study had CHADS₂ scores ≥2, the mean person-years of follow-up was >1,500, which was sufficient to show statistical significance in a population with low event rates.

CHADS₂ and CHA₂DS₂-VASc scores can provide an estimate of the risk of adverse events in patients who undergo catheter ablation of AF. Among patients with CHADS₂ scores of 0 or 1, CHA₂DS₂-VASc scores were useful to further identify which patients were predisposed to adverse events.

For an expanded Methods section, please see the online version of this article.