Does catheter ablation reduce mortality and stroke in patients with atrial fibrillation? Patient selection or causality?

This editorial refers to ‘Catheter ablation for atrial fibrillation is associated with lower incidence of stroke and death: data from Swedish health registries’ † , by L. Friberg et al . on page 2478 . Catheter ablation is as an effective treatment for reducing arrhythmia burden and improving quality of life in patients with symptomatic atrial fibrillation (AF), 1 , 2 and most effective in patients with paroxysmal AF. Many patients will experience symptomatic or asymptomatic episodes of AF after catheter ablation, and, in most patients with AF, catheter ablation cannot be considered curative. 3 , 4 Population studies indicate that AF is closely associated with a higher risk of stroke and increased mortality. 5 Recent data support paroxysmal AF as being associated with lower stroke risk than permanent AF. 6 Theoretically, performing catheter ablation and reducing AF burden therefore may improve patients' outcome with respect to stroke and death. Some observational studies support this, even though it was never supported by any randomized controlled trial. In contrast, it is well documented that a rhythm control strategy by antiarrhythmic drug therapy is not superior to rate control therapy with respect to survival. 7 In this issue of the journal, Friberg et al . report their findings when analysing data from the Swedish Health Registries on the diagnosis of AF, catheter ablation for AF, and the outcomes ischaemic stroke and death during medium term follow-up. 8 They identified >360 000 patients with a diagnosis of AF during the period 2006–2012, and 4278 patients who had undergone ablation for AF. The authors used propensity score matching to create two ‘comparable’ groups of AF patients, with and without catheter ablation, respectively. They identified a wealth of baseline characteristics from different health registries to include in the analysis. Catheter ablation for AF was associated with a 31% lower risk of stroke and a 50% lower risk of death within a mean 4.4 years of follow-up, both highly statistically significant. The authors extend their findings to suggest that prognostic considerations may be relevant when deciding whether to perform catheter ablation or not. The authors should be commended for performing this huge data management and data analysis. The results are intriguing and attractive. With respect to the endpoint of ischaemic stroke, a recent study by Karazoy et al . reported similar findings from Danish Health Registries. 9 The remaining question is whether these associations indicate any causality. Do these data support that catheter ablation for AF saves lives and reduces stroke risk? The patients who underwent catheter ablation for AF differed profoundly from the remaining AF population with respect to almost any baseline parameter. Mean ages were 59 years in ablated patients vs. 75 years in those not ablated, 62% of patients in the ablated group were working vs. 15% in those not ablated, mean CHA 2 DS 2 -VASc score was 1.5 vs. 3.6 in the two groups, and every co-morbid condition recorded was less common in the ablated group. From this population, can matching create two groups that can be considered comparable? It is important that parameters included in a propensity score matching are measured accurately and recorded uniformly in the two groups. We do not know the accuracy of the health registry data included in the analysis. In addition, it is unknown whether recording of co-morbidity was done equally in the two populations. Patients referred for catheter ablation are seen in tertiary centres by arrhythmia specialists, familiar with the AF-associated co-morbid conditions, and their diagnosis and treatment. Do other doctors, who encounter AF in their patient's ECG and report this as a diagnosis, record and treat these co-morbid conditions in a similar way? Non-diagnosed co-morbidity in the not ablated group would tend to cause findings such as those reported here. Catheter ablation is most effective for patients with paroxysmal AF and less useful for those with long-standing persistent AF. The prognosis may differ markedly between these two patient categories. 10 The authors do not account for type of AF in their analysis, which is a major limitation; nor is left ventricular function or left atrial size taken into account. Furthermore, time at risk is different between the two groups compared. For catheter-ablated patients, time at risk started from the day of first catheter ablation. At that point in time, patients were months or years after AF debut, in a clinically stable state, and fully anticoagulated. In contrast, for patients not ablated, time at risk was counted from 30 days after first contact with an AF diagnosis during the study period. Many of these patients may have had their first AF diagnosis 30 days before time at risk started. Recent data indicate a markedly increased risk of death beyond 30 days and at least 90 days after an AF diagnosis. 11 To perform analysis of complex observational data sets, multifactorial methods such as, for example, propensity score matching are commonly used, making adjustment for known and observed prognostic factors possible. Such analyses, however, cannot adjust for unknown or unmeasured variables. To study an effect of any treatment intervention on patient outcome, a randomized controlled trial is the standard method. Only by using randomization is it possible to create groups, where prognostic factors—known as well as unknown—are in mean distributed equally between comparison groups. Only randomized controlled trials allow for concluding on causality. Therefore, the present associations observed between catheter ablation for AF and lower risks of stroke and death must be interpreted extremely cautiously. An uneven distribution of prognostic factors not accounted for in the analysis may alone explain these findings. The group of patients ablated for AF probaby represents a selected population with a better prognosis. There is a close correlation between risk of death and patient selection, even in AF populations ( Figure 1 ). Catheter ablation is effective for reducing AF and improving quality of life. The present results should not lead us to recommend catheter ablation to our AF patients because catheter ablation reduces stroke and death! The currently ongoing randomized trials )CABANA (NCT00911508) and EAST (NCT01288352)] may provide some of these important answers. Mortality in atrial fibrillation (AF) patients depends on patient selection. The two curves in blue and red illustrate the findings from the Swedish Health Registries in patients who underwent catheter ablation for AF (red) and matched patients not ablated (blue). The green line illustrates mortality in patients with AF adapted from the Framingham data. 5 Conflicts of interest: none declared.