OUP user menu

Undetected paroxysmal atrial fibrillation in chronic heart failure patients: is it clinically relevant to catch the atrial phantom?

Ron Pisters, Cees B. de Vos, Robert Dennert, Harry J. Crijns
DOI: http://dx.doi.org/10.1093/europace/eup268 1257-1259 First published online: 24 September 2009

Chronic heart failure (CHF) is frequently accompanied by atrial fibrillation (AF). How often AF occurs in CHF patients is difficult to know since many episodes come and go undetected because of lacking symptoms. Therefore, assessing the total burden of AF calls for continuous rhythm monitoring. At present, we do not know whether assessing burden of AF is clinically relevant. Furthermore, we do not know whether AF has any impact on survival in CHF patients, since previous clinical studies show conflicting results.

Caldwell et al.1 shed light on this issue by retrospectively studying cardiac resynchronization therapy device downloads in 162 patients with CHF in order to address two questions. First, what is the prevalence of asymptomatic paroxysmal AF (PAF) in CHF patients? Secondly, what is the subsequent effect on mortality, thrombo-embolism and all-cause hospitalizations? They defined AF as a mode-switch event with an atrial rate >200 bpm for a minimum of 30 s. Using medical records, patients were divided into four groups: (i) patients who remained in sinus rhythm (SR), (ii) patients in chronic AF, (iii) patients who were known to have paroxysmal AF (known PAF), and (iv) patients thought to be persistently in SR in whom episodes of AF were newly identified (new PAF). One quarter of patients developed new PAF but AF burden was low. During the mean follow-up period of 424 days, no statistically significant differences in outcome between the SR, new PAF, and known PAF patients were found. There was, however, a trend towards higher mortality in patients with new PAF which matches with an important observation in the Euro Heart Survey on AF in which patients with first documented AF had higher mortality than patients with known PAF or persistent AF.2 Similarly, Pozzoli et al.3 showed that new onset AF in a previously stable CHF population often marked the onset of haemodynamic instability and death. Therefore, we believe that in clinical practice, newly detected AF should always trigger search for new correctable cardiovascular abnormalities leading to further fine-tuning of existing treatments as needed.

Whether AF has an impact on survival in CHF patients still is a matter of debate and speculation with some studies indicating an impact whereas others do not. The recently published landmark trial ‘Rhythm control versus rate control for atrial fibrillation and heart failure’ (AF-CHF) did not show any difference in cardiovascular mortality between the two treatment strategies, despite the majority of patients maintaining SR.4 This suggests lack of an impact of AF on survival in CHF patients. On the other hand, several post hoc analyses showed that patients who remained in SR had a better survival, although SR (and vice versa AF) may have served as a marker for survival rather than being instrumental in causing survival.

The discrepancies between studies relate to the type of study: an epidemiological cohort, survey, observational clinical study, or clinical trial. Obviously, patient selection with variation in inclusion of all comers or not, and differences in severity of CHF as well as in its duration until onset of AF (frequently not reported) play a large role here. In addition, variations in the extent of guidelines adherent treatment during follow-up are important. As an example, the time interval between the event causing heart failure and the onset of AF may be extremely important because if short, AF certainly may increase mortality mainly because the potential non-survivors are not missed, and the AF marks severe CHF.5 Unfortunately, the report by Caldwell et al.1 does not add too much to this discussion if only for small numbers and lack of information on duration of CHF vs. duration of AF. Most probably, the duration of CHF before onset of AF was long in their patients. In addition, undetected and unrecognized AF did not elicit haemodynamic instability. Both features suggest lack of an impact of AF on subsequent survival or hospitalization. There was, however, a trend towards increased mortality in new PAF compared with SR patients (19 vs. 10%) which may have been caused by omitting search for correctable CHF causes and insufficient guidelines adherence in PAF patients. A recent report from the Euro Heart Survey on AF showed a significant non-adherence to combination of the CHF and AF guidelines.6 Considering the relatively low use of ACE-inhibitors and angiotensin receptor blockers (76%), beta-blockers (67%), and oral anticoagulation (15%), this may have also been the case in the study by Caldwell et al.1 Therefore, we believe that new PAF in CHF patients should trigger further diagnostic actions and fine-tuning of treatment including anticoagulation, beta-blockade, and renin–angiotensin system blockers. Also invasive treatment of CHF may be contemplated.

Although all-cause death did not differ significantly, AF-related deaths—i.e. fatal ventricular arrhythmia and fatal strokes—might have. Sudden death occurring more often in the ischaemic AF population, suggesting myocardial ischaemia and scar formation as key players, may account for worse outcome with AF. This would be in line with the study by Pedersen et al.5 and might explain why in particular hospitalized CHF patients with acute coronary syndromes developing new onset AF have increased in-hospital mortality.7 Unfortunately, perhaps due to the limited number of patients, no aetiology-specific data are presented. Also, the authors do not report the occurrence of AF or other atrial arrhythmias prior to documented episodes of ventricular tachycardia/ventricular fibrillation, nor the occurrence of (inappropriate) shocks which could influence differences in hospitalizations.

Atrial fibrillation is clearly related with thrombo-embolism. Currently, an important question is whether or not the duration of AF influences risk of stroke. Until now, no differences in stroke risk between PAF and persistent AF have been found when correcting for CHADS2 score. This means that the stroke risk score rather than the type of AF should drive the decision to anticoagulate or not.8 Recently, Botto et al.9 studying a paced AF population found that AF burden is important since the longer episodes of AF last, the higher is the risk of stroke, irrespective of risk assessed by the CHADS2 score. In addition, Capucci et al.10 reported data suggesting that duration of AF adds to risk of stroke if AF lasts longer than 24 h. Caldwell et al.1 did not find an impact of new PAF on the subsequent stroke rates, despite a significant under-use of oral anticoagulation. The follow-up in this study may have been too short and the number of patients was limited. Even more important is the fact that data on risk of stroke are not systematically reported. If risk of stroke is low, even without anticoagulation the stroke rate will be low. Trying to calculate risk of stroke using the CHADS2 score (which the authors could have done easily), one has to consider that all patients had CHF (1 point), that average age was 66 (0.5 point), and that only relatively few patients had either diabetes or hypertension (0 point in a significant proportion of patients). Assuming that none of the patients had suffered from previous stroke or transient ischaemic attack, most patients may have had low risk of stroke, with a CHADS2 score not higher than 1.5, which in turn may explain why the stroke rate was so low even without anticoagulation. This analysis may still be done but should include the approach like the one suggested by the data from Botto et al., 9 especially since the duration of AF episodes (represented by duration of mode switches) is available. In this respect, indeed, the outcomes of ASSERT and TRENDS studies are eagerly awaited.

In addition, inherent drawbacks because of the retrospective nature of the study are important. Beside suboptimal retrieval of thrombo-embolic events, crucial information on cause of death is frequently missing. Perhaps, fatal strokes were simply classified as a fatality rather than specified as fatal stroke.

Beside differences in the medication use, baseline demographic and clinical between the known and new PAF patients were similar. Assuming undetected and as a result untreated, AF in CHF patients has a profound negative impact on outcome, one would try to find a tool to identify CHF patients at risk of developing AF. It would, therefore, be interesting to know the risk scores of all patients according to the recently published scheme by Schnabel et al.11 Since echocardiography is routinely applied in this population, it would be worthwhile to implement electroechocardiographic parameters (Pa-TDI) in future research because of its potential.12

We challenge the use of the phrase ‘asymptomatic’ PAF, because classic asymptomatic AF refers to electrocardiographic documentation of the arrhythmia without complaints at that time. Because of the retrospective nature of the study, this cannot be proven and ‘clinically undetected PAF’ seems more appropriate. Moreover, apart from palpitations, AF patients frequently complain of dyspnoea. Considering that all CHF patients were in NYHA functional class III or IV, dyspnoea is very common among these patients. Therefore, it is likely that patients explained AF provoked dyspnoea as an exacerbation of their heart failure. This might seem trivial to discuss, but optimizing patient education could trigger more appropriate read outs, thereby early detection of AF and appropriate guideline adherent treatment resulting in improved outcomes. Truly asymptomatic episodes, which cannot be linked to any physical complaint, would thus require constant monitoring.

While the latter might seem undoable at present, upcoming home-linked monitoring systems might be the ultimate ghost busters. These systems allow read outs on a daily basis, immediate detection of the arrhythmia and subsequent adequate reaction. Whether this will actually save the lives of those who already have both CHF and AF is very doubtful, especially when severe left ventricular dysfunction is already present. Perhaps that installing home-linked care in an early (pre-disease) state might prove to be more efficient in the long term.

Conflict of interest: none declared.


  • The opinions expressed in this article are not necessarily those of the Editors of Europace or of the European Society of Cardiology.


View Abstract