Author + information
- Received October 11, 2001
- Revision received March 27, 2002
- Accepted April 5, 2002
- Published online July 3, 2002.
- David Schwartzman, MD FACC*,* (, )
- Shailesh Kumar Musley, PhD†,
- Charles Swerdlow, MD, FACC‡,
- Robert H Hoyt, MD, FACC§ and
- Eduardo N Warman, PhD†
- ↵*Reprint requests and correspondence:
Dr. David Schwartzman, Cardiovascular Institute, Presbyterian University Hospital, B535, 200 Lothrop Street, Pittsburgh, Pennsylvania 15213-2582, USA.
Objectives We sought to gain insights into the early recurrence of atrial fibrillation (ERAF) after cardioversion shocks delivered by permanently implanted rhythm management systems.
Background Several reports have characterized ERAF, but these reports used a limited definition and did not evaluate an association between clinical or device variables and ERAF.
Methods A total of 144 patients with recurrent, drug-resistant, symptomatic atrial fibrillation (AF) underwent implantation of an atrial rhythm management system (Medtronic Jewel AF, Model 7250, Minneapolis, Minnesota). The device was programmed to deliver cardioversion shocks automatically and/or on patient command. The incidence of ERAF was evaluated after 1,092 successful shocks among 97 patients. Three different ERAF definitions were used: recurrence within 1 min, 1 h or 1 day. Multiple clinical and device variables were assessed for their relationship with ERAF.
Results The per-patient incidences of ERAF were 44%, 61% and 70% for ERAF within 1 min, 1 h and 1 day, respectively. The per-episode incidences of ERAF were 17%, 30% and 43% for ERAF within 1 min, 1 h and 1 day, respectively. Variables that were independently associated with ERAF included AF duration <3 h before termination, more than one shock required to cardiovert and the absence of a previous myocardial infarction. The most potent variable was AF duration <3 h, associated with a threefold increase in the incidence of ERAF.
Conclusions Recurrence of AF early after ambulatory shock cardioversion is common. In this retrospective study, both clinical and device variables were predictive.
Early recurrence of atrial fibrillation (ERAF) has been observed after successful atrial cardioversion mediated by direct-current shock, most recently those delivered via permanently implanted systems (1–9). Early recurrence of atrial fibrillation represents a serious obstacle in the path to broadened use of “device-based” strategies for atrial rhythm control. Although previous reports have chronicled the incidence of ERAF, the definitions used were limited to those episodes occurring within minutes of a successful shock. From the patient’s perspective, such a definition may be inadequate, as atrial fibrillation (AF) recurrence within hours or days of a successful shock, if occurring regularly, would not be tolerable. In addition, no report has addressed the predictors of ERAF. The characterization of risk factors may permit the immediate refinement of patient management and provide insights for the future development of improved therapeutic strategies.
In the present study, we investigated the incidence of post-shock ERAF in a cohort of patients who underwent implantation of an atrial rhythm management system. We extend previous reports by characterizing ERAF using intervals ranging from 1 min to 1 day, assessing individual incidences and evaluating multiple variables for significant association.
The data were compiled from 146 consecutive patients from multiple centers who were enrolled in a study evaluating the Jewel AF device (model 7250, Medtronic, Inc.; Minneapolis, Minnesota) and lead system. Participating centers and investigators have been acknowledged previously (10). Each patient fulfilled the following enrollment criteria: 1) ≥2 symptomatic atrial tachyarrhythmia events within three months of system implantation, with electrocardiographic documentation of at least one episode; 2) inefficacy and/or intolerance of at least one type I or III anti-arrhythmic drug; and 3) no history of clinically important ventricular arrhythmia. Exclusion criteria included New York Heart Association functional class IV heart failure, cardiac surgery within the previous month, a left atrial thrombus documented within the previous 6 months, a cerebrovascular accident within the previous 12 months, a mechanical tricuspid valve and life expectancy <1 year. Each patient gave written, informed consent, according to a protocol approved by the Human Subjects Committee of the institution at which the devices were implanted. Patient characteristics are summarized in Table 1.
The Jewel AF system is an implantable dual-chamber cardioverter-defibrillator system capable of sensing, pacing and defibrillation in both the atrium and ventricle. Atrial tachyarrhythmia detection is programmable in two zones—atrial tachycardia and AF—distinguished by the rate and regularity on the atrial electrogram. For the purposes of the present study, all atrial tachyarrhythmias have been termed AF. The positive predictive accuracy for spontaneous atrial tachyarrhythmia detection by the 7250 device was previously shown to be 99% (10). Undersensing was minimized by an auto-adjusting sensitivity and a 30-ms cross-chamber blanking period.
Therapies for AF included antitachycardia or 50-Hz burst pacing and shocks. Atrial arrhythmia prevention features included a switchback delay (high-rate DDI pacing continued for a programmable duration immediately after an atrial tachyarrhythmic event) and atrial rate stabilization (which prevents long pauses after premature atrial complexes). Specific device programming for individual patients was at the discretion of the investigator (Table 2).
In the present study, a coil-bearing lead located in the coronary sinus comprised a portion of this circuit in 85 patients. Among the patients without a coronary sinus lead, the most common circuit incorporated two coils (right ventricle and superior vena cava) on a single lead and the device can (n = 47). Shocks could be delivered automatically or on patient command. The timing of commanded shocks relative to AF onset was not controlled. Automatic shocks were programmed to occur after a sustained AF duration (range 1 min to 24 h) and, in some cases, only at selected nighttime hours.
After each shock, the atrial interelectrogram interval was monitored. As defined by the device, a shock was successful in terminating an AF event if five consecutive ventricular beats occurred during sinus or atrial paced rhythm. The ERAF was defined as a newly detected AF event occurring after a successful shock. In the present study, ERAF was defined by using three nonexclusive terms: 1) ERAF1 min= successful shock followed by AF within 1 min; 2) ERAF1 h= successful shock followed by AF within 1 h; and 3) ERAF1 day= successful shock followed by AF within 24 h.
For each treated AF episode, the device stored 5 s of electrographic recordings and 60 atrial and ventricular marker channel events preceding the initial therapy, as well as 60 events preceding AF termination. These data were reviewed by the investigators, and only appropriately detected episodes were included. The date and time of detection, type of arrhythmia (e.g., fibrillation or tachycardia), median interatrial electrogram interval, total arrhythmia duration and device-defined outcome of therapy were also recorded. No information was stored after a failed shock. Therefore, we were unable to quantify the incidence of successful shocks followed by new AF recurrence before device confirmation of shock success (Fig. 1).
The present study focuses on events after successful shock only. Data are reported as the mean value ± SD, unless otherwise specified. The Wilcoxon signed rank test was used for paired comparisons. To adjust for multiple episodes in each patient, the generalized estimating equation (GEE) statistical method with exchangeable correlation was used (11). The GEE model yielded an estimate of ERAF incidence, along with 95% confidence intervals (CIs). The univariate associations between prospectively identified demographic, clinical and device variables and ERAF (Table 3) were also evaluated using the GEE model. A multivariate model was then developed using a backwards elimination method with significant univariate predictors. A p value <0.05 was considered significant. All analyses were performed using SAS version 6.12 (SAS Institute Inc., Cary, North Carolina).
Data from 146 patients were compiled. The system could not be implanted in two patients (inability to implant the atrial lead in 1; unacceptably high ventricular defibrillation threshold in 1). At the time of this analysis, the device had been implanted for 12.7 ± 6.1 months (range 0.1 to 25.9 months). A total of 1,427 shocks were delivered to 1,200 spontaneous atrial episodes in 107 patients. Of these 1,200 episodes, 1,092 were successfully converted in 97 patients. These 97 patients formed the cohort of the present study.
Unsuccessful pacing therapies preceded shocks in 726 AF events (66%) among 82 patients (85%). The median number of successful cardioversion shocks per patient was six (13 patients with 1 shock, 35 patients with 2 to 5 shocks and 18 patients with 6 to 10 shocks). There were 519 commanded shocks (47.5%) among 61 patients (63%).
The per-patient incidences of ERAF (patients with one or more ERAF event[s]) were 44% for ERAF1 min, 61% for ERAF1 hand 70% for ERAF1 day. The per-episode incidences of ERAF (successful cardioversions events followed by ERAF) were 17% for ERAF1 min, 30% for ERAF1 hand 43% for ERAF1 day. A disproportionate number of ERAF events occurred within minutes of cardioversion (Fig. 2). Individual patient incidences of ERAF were highly variable. Tables 3 and 4⇓summarize the results of univariable and multivariable analyses, respectively, assessing for a relationship between ERAF and clinical and device variables. There was no significant relationship between ERAF1 min, ERAF1 hor ERAF1 dayand the number of successful shocks. Factors emerging in the multivariable analysis as significantly associated with each ERAF classification included AF duration <3 h before cardioversion, failed shocks delivered before the successful one and the absence of a previous myocardial infarction. Of these, AF duration <3 h before shock was a very powerful factor associated with a threefold increase in the incidence of ERAF. Analysis of the median AF duration showed that the incidence of ERAF decreased as the pre-cardioversion AF duration increased (Fig. 3).
To exclude a bias due to event clustering, a subset of 79 patients (444 shocks) in whom no AF was observed for at least 24 h before successful cardioversion was assessed. The per-patient incidences of ERAF were 31% for ERAF1 min, 45% for ERAF1 hand 56% for ERAF1 day. The per-episode incidences were 16% for ERAF1 min, 25% for ERAF1 hand 36% for ERAF1 day. The ERAF incidence (any definition) for pre-cardioversion AF duration >3 h was 11% (95% CI 6% to 18%), significantly lower than the 25% incidence for pre-cardioversion AF duration <3 h (95% CI 17% to 35%; p < 0.001). Some individuals had multiple successful cardioversions, some of which were followed by ERAF. In these individuals, a paired analysis of the median AF duration demonstrated that successful cardioversions succeeded by ERAF were performed significantly earlier, relative to atrial arrhythmia onset, than those not succeeded by ERAF (Fig. 4).
Post-shock ERAF can occur after transthoracic and transvenous cardioversion (1–6). Regarding transthoracic cardioversion, Yu et al. (2), who defined ERAF as AF recurrence <1 min after conversion, reported an incidence of 26% among 50 patients. Daoud et al. (3), who defined ERAF as AF recurrence <5 min after conversion, reported an incidence of 9% among 337 patients. Regarding transvenous cardioversion, utilizing temporary defibrillation lead systems and with ERAF defined as AF recurrence within 5 min of conversion, several studies have reported per-patient incidences of ERAF ranging from 13% to 36% (4–7). No clinical or echocardiographic predictors were reported. More recent reports have also addressed the phenomenon of ERAF after transvenous shock cardioversion by permanent lead systems. Wellens et al. (8)demonstrated 27% per-episode and 51% per-patient incidences of ERAF (defined as AF recurrence within 1 min) among 41 patients implanted with the Metrix system (Model 3000 or 3020, Guidant Inc., St. Paul, Minnesota) who underwent cardioversion under direct physician observation. Daoud et al. (9)reported a per-episode incidence of ∼25% in patients with this system who underwent ambulatory conversion. However, due to limitations in event logging by this device, the exact incidence and timing of ERAF events could not be determined. Neither of these studies reported factors significantly associated with ERAF.
Our data confirm and extend those of the aforementioned studies. First, we present a broader and potentially more clinically relevant analysis of ERAF. From the patient’s perspective, ERAF is not limited to minutes after cardioversion. For example, the need to deliver additional cardioversion shocks for new AF events within hours or days of a previous successful shock might be viewed as equally objectionable as the need to deliver additional shocks within minutes. Although a disproportionate number of ERAF events occurred within minutes of conversion, a substantial number occurred later. Among recurrent AF episodes, those that were early versus new (e.g., paroxysms that would have occurred regardless of how the previous event was terminated) are difficult to distinguish. However, we do not believe that new AF episodes significantly influenced our data, for two reasons: 1) a clinical syndrome of paroxysmal AF was present in a minority (30%) of the patients; and 2) in the multivariable analysis of factors associated with ERAF, clinical AF syndrome was not important.
Second, we identified associations between several clinical and device variables and ERAF. Several variables were independently associated with ERAF. The significant variable that can most easily be influenced by device programming is AF duration. Earlier studies have suggested that a longer duration of AF is associated with a higher recurrence rate, linking AF susceptibility to arrhythmia-induced atrial electrophysiologic substrate remodeling (12–15). In contrast, our data demonstrate the opposite. An explanation for this finding may lie in the duration of AF before cardioversion. In previous studies, the duration of AF was not controlled but was generally long (days to months). In contrast, in the present study, AF duration was much shorter. We speculate that there might have been a period early after AF onset during which remodeling was insubstantial and thus not a factor in determining susceptibility to AF. During this period, AF duration may have influenced the activity of sites of origin of AF-triggering atrial premature depolarizations, with a longer AF duration being more suppressive (7). Whatever the mechanism, if substantiated, this finding may alter the way device-based atrial rhythm control strategies are implemented. For example, shock therapies may be delayed for hemodynamically stable, tolerable AF episodes.
The inverse correlation of a previous myocardial infarction with ERAF suggests that patients may be stratified for risk of ERAF based on clinical variables. If substantiated, such results may assist physicians in choosing an AF treatment strategy.
The correlation between ERAF and the requirement for multiple shocks to terminate AF probably reflects AF recurrence after successful shocks before device confirmation (Fig. 1). This would have been misclassified by the device as an unsuccessful shock.
We cite the following limitations: 1) the analysis was based on data acquired retrospectively, generated in the context of an observational trial uncontrolled for multiple factors; 2) because successful termination of AF by a shock can be preceded by unsuccessful pacing and shock therapies, it is possible that these therapies may have influenced ERAF and subacute recurrences; and 3) our data probably underestimate the incidence of ERAF within the first minute after shock.
The correlations we identified between ERAF and the analyzed variables require prospective validation.
Early recurrence of atrial fibrillation after ambulatory shock conversion is common. Factors significantly associated with ERAF, as identified in the present study, may, if subsequently validated, lead to refinement in patient selection and device programming. With respect to our speculation about the mechanism of increased ERAF incidence with decreased pre-shock duration, prospective confirmation of this concept might foster new AF control strategies not necessarily limited to those based on implantable devices.
☆ This study was sponsored by Medtronic, Inc., Minneapolis, Minnesota.
- atrial fibrillation
- confidence interval
- early recurrence of atrial fibrillation
- ERAF1 min
- successful shock followed by AF within 1 min
- ERAF1 h
- successful shock followed by AF within 1 h
- ERAF1 day
- successful shock followed by AF within 24 h
- generalized estimating equation
- Received October 11, 2001.
- Revision received March 27, 2002.
- Accepted April 5, 2002.
- American College of Cardiology Foundation
- Tieleman R.G.,
- Van Gelder I.C.,
- Crijns H.J.G.M.,
- et al.
- Yu W.-C.,
- Lin Y.-K.,
- Tai C.-T.,
- et al.
- Tse H.-F.,
- Lau C.-P.,
- Ayers G.M.
- Tse H.-F.,
- Lau C.-P.,
- Ayers G.M.
- Wellens H.J.J.,
- Lau C.-P.,
- Luderitz B.,
- et al.
- Daoud E.G.,
- Timmermans C.,
- Fellows C.,
- et al.
- Zeger S.L.,
- Liang K.Y.
- Levy S.,
- Ricard P.,
- Lau C.P.,
- et al.
- Daoud E.G.,
- Bogan F.,
- Goyal R.,
- et al.
- Fareh S.,
- Villemaire C.,
- Nattel S.