Classifying Atrial Fibrillation in Everyday Clinical Practice and in Clinical Trials: Caecus Caeco Dux

Editor's Note: Commentary based on Charitos EI, Pürerfellner H, Glotzer TV, Ziegler PD. Clinical classifications of atrial fibrillation poorly reflect its temporal persistence: insights from 1,195 patients continuously monitored with implantable devices. J Am Coll Cardiol. 2014;63:2840-8.


The clinical classifications of atrial fibrillation (AF) are used to individualize the choice of rate or rhythm control strategies and to select appropriate medical or interventional therapies for each AF patient. For example, although patients classified as having paroxysmal or persistent AF are generally indicated for rhythm control, patients with permanent AF are usually treated with rate control strategies.

In clinical trials, the clinical AF classifications are employed to select patients for inclusion in clinical trials with the intention to build groups of patients with similar arrhythmia magnitude in order to compare the effect of a treatment between the control and the treatment group.

We often take for granted that a clinician can accurately determine a patient's pattern of AF and classify each patient appropriately. The aim of this study1 was to assess how accurately the clinical AF classifications ("paroxysmal," "persistent") reflect the temporal persistence of AF (i.e., how much time a patient is in AF) and how homogeneous are patients classified in the same clinical AF classification.


For the purposes of the present study, patients enrolled in the OMNI2 and TRENDS3 clinical trials were included. Patients with AF-specific treatments (medical/electrical cardioversion or catheter ablation) and with permanent AF were excluded. The total population consisted of 1,195 patients with at least 180 days of documented rhythm history. As a quantitative measure of AF persistence, we used the measurement of AF burden, defined as the proportion of the monitored time that a patient was in AF.

Several statistical methods (simple descriptive test, logistic regression, receiver-operating characteristic analyses) were employed to evaluate the effectiveness of clinical AF classification to reflect the magnitude of AF temporal persistence as measured by the AF burden.


Two main findings emerged from the analysis of our data. First, there was large overlap of AF burdens within the clinical AF classifications. Some patients with very low burden of AF were classified as having persistent AF, whereas some patients with very high burden of AF were classified as having paroxysmal AF. In contrast to our expectation, increasing AF burdens did not result in an increasing probability of being classified in the persistent AF classification. Even at very high AF burdens, most patients were classified as having paroxysmal AF. Within each AF classification, patients with AF burdens of the complete spectrum could be found (from AF free to continuous AF). Second, other factors seem to influence the AF classification. Higher left ventricular ejection fraction and the presence of coronary disease were independently associated with a lower probability of a patient being classified as having persistent AF for the same underlying AF burden.

Both of the findings above signify that the correlation between time spent in AF and clinical AF classification is rather poor.

Conclusion and Perspective

This study has several implications. First, although AF classifications attempt to communicate information about the persistence and magnitude of AF recurrence as well as to denote the stage of the AF disease, the findings suggest that within the same AF class there may exist patients with vastly different degrees of temporal AF persistence. Having "paroxysmal" AF does not always imply "less amount of AF" than having "persistent" AF.

Second, AF classification strongly impacts therapy selection, whether this is a decision to pursue rhythm or rate control or the decision to perform a pulmonary vein isolation (PVI) or a more extended lesion set in patients with paroxysmal or persistent AF, respectively. The data show that because AF clinical classification is imprecise, this might influence the selection of the right patient for the appropriate therapy.

Third, in the setting of a clinical trials, we attempt to enroll patients with a similar arrhythmia magnitude in order to be able to evaluate the effect of a therapy within a specific patient population or compare the effect between patient populations, and it is of great importance that these study cohorts are as uniform as possible with respect to the magnitude of AF. The results suggest that there may be significant "blurring" of these AF classifications in clinical practice, thereby making the interpretation of such studies a challenge. For example a potential explanation for lack of complete efficacy with the PVI only approach in the ablation of patients with paroxysmal AF might be that some of the patients enrolled as having paroxysmal AF may actually have a more persistent form of the disease.

Fourth, other factors outside the AF classification criteria, such as patient-related factors (ejection fraction, presence of coronary artery disease) influence the way we classify AF. It seems that for the same magnitude of AF persistence, we classify patients with low versus high EF or patients with versus without coronary artery disease differently. This further diminishes the discrimination ability of AF clinical classifications to express the temporal persistence of AF.

On the other hand, AF classification utilizing device-based indices (time in AF) are more consistent, reproducible, and reflect the magnitude and the temporal persistence of AF much better than clinical AF classifications.1 When using device data to classify AF, the resulting classification groups were more homogenous, the borders between each class more demarcated, and, generally, the device-derived classifications more closely reflect the increases in AF burden. Patient characteristics and demographics do not influence the device-based AF classification.1 Further study is required to determine if classifying AF based on rigorous arrhythmia monitoring, and device data can improve clinical outcomes relative to traditional clinical assessments.


  1. Charitos EI, Pürerfellner H, Glotzer TV, Ziegler PD. Clinical classifications of atrial fibrillation poorly reflect its temporal persistence: Insights from 1,195 patients continuously monitored with implantable devices. J Am Coll Cardiol 2014;63:2840-48.
  2. Sweeney MO, Sakaguchi S, Simons G, et al. Response to the Center for Medicare & Medicaid Services coverage with evidence development request for primary prevention implantable cardioverter-defibrillators: data from the OMNI study. Heart Rhythm 2012;9:1058-66.
  3. Daoud EG, Glotzer TV, Wyse DG, et al. Temporal relationship of atrial tachyarrhythmias, cerebrovascular events, and systemic emboli based on stored device data: a subgroup analysis of TRENDS. Heart Rhythm 2011;8:1416-23.

Clinical Topics: Arrhythmias and Clinical EP, Atherosclerotic Disease (CAD/PAD), Implantable Devices, EP Basic Science, SCD/Ventricular Arrhythmias, Atrial Fibrillation/Supraventricular Arrhythmias

Keywords: Arrhythmias, Cardiac, Atrial Fibrillation, Catheter Ablation, Cohort Studies, Coronary Artery Disease, Coronary Disease, Cost of Illness, Demography, Electric Countershock, Intention, Pulmonary Veins, ROC Curve, Stroke Volume

< Back to Listings