Native QRS duration predicts the occurrence of arrhythmic events in ICD recipients

Abstract

Aims Identification of implantable cardioverter/defibrillator (ICD) recipients at higher risk of future therapies may assist in pre-empting future shocks. Native QRS duration is an established predictor of overall mortality, but the role of this parameter as a clinical predictor of arrhythmic events warrants further investigation.

Methods and results In an analysis of a single-centre, 13-year ICD implantation experience (1990–2002), multiple clinical parameters including QRS duration were analysed using a multiple logistic regression model. Of 562 patients followed for at least 1 year, 98 (17%) did not receive ICD therapies (event-free, group A). Comparisons were made with a randomly selected sample of 123 patients who received ICD therapies (arrhythmic events, group B). There were no significant differences in age, gender, frequency of coronary artery disease, and degree of left ventricular dysfunction. However, QRS duration was a significant determinant of arrhythmic events (≥100 vs. <100 ms: adjusted OR 2.75, 95% CI 1.37–5.51; ≥120 vs. <120 ms: adjusted OR 1.77, 95% CI 0.97–3.23). QRS duration was also a predictor of overall mortality in the logistic regression models (≥100 ms: adjusted OR 3.72, 95% CI 1.17–11.9; ≥120 ms: adjusted OR 3.09, 95% CI 1.39–6.85).

Conclusion In this ICD population, consisting largely of secondary prevention ICD recipients, longer QRS duration predicted higher likelihood of arrhythmic events. Extent of QRS prolongation could guide the decision to initiate prophylactic anti-arrhythmic therapy in ICD patients.

Introduction

With a burgeoning population of implantable cardioverter/defibrillator (ICD) recipients, there is an acute need for clinical predictors that will assist in the management of implanted patients. An improved ability to predict an individual's risk of recurrent ventricular arrhythmias and future shocks could facilitate the initiation of targeted prophylactic interventions such as anti-arrhythmic drug therapy.

Intraventricular conduction delay, as measured by the native QRS duration, is a powerful predictor of overall mortality in patients with either a history of myocardial infarction or congestive heart failure.^1–5 There is additional evidence to suggest that increased-QRS duration may also be predictive of sudden death following myocardial infarction.⁶ However, potential interactions with ventricular arrhythmia recurrence remain unclear. In fact, the utility of native QRS duration for the purpose of predicting recurrent ventricular arrhythmia in ICD patients has not been fully assessed. We therefore evaluated the potential role of multiple clinical variables, including QRS duration, in predicting occurrence of arrhythmic events from our 13-year single centre experience.

Methods

We used the ICD database at Oregon Health and Science University to retrospectively analyse potential clinical predictors of survival with freedom from arrhythmic events (device therapies, both defibrillation or anti-tachycardia pacing) in patients with implantable defibrillators.⁷ Patients who received ICDs between January 1990 and October 2002 were considered. Those patients included in this case–control study were currently followed at Oregon Health and Science University or had died during follow-up at this institution. As a routine, patients received follow-up in the arrhythmia clinic at 1 week, 1 month, and every 3 months following ICD implantation. At every visit, evaluation included patient history and device interrogation to determine occurrence of device therapies. All patients with at least 1-year follow-up (Figure 1) who had an interpretable electrocardiogram (i.e. not paced) and who had not received defibrillation or anti-tachycardia pacing at any point during the follow-up period were included in group A (event-free). Group B (arrhythmic event) contained a representative sample of the remaining patients with an interpretable electrocardiogram and at least 1-year follow-up, who had received an ICD therapy (defibrillation or anti-tachycardia pacing). Group B was chosen by randomly identifying every fourth eligible patient, with proportional representation for each year of follow-up. Electrocardiograms obtained at the time of device implantation were used for this analysis.

Clinical variables with potential predictive value were compared between the groups. These included age, sex, QRS duration, primary cardiac aetiology, anti-arrhythmic drug therapy, coronary artery disease (CAD), left bundle branch block, and left ventricular (LV) dysfunction (ejection fraction <55%). QRS duration was measured from a 12-lead ECG obtained at the time of ICD implantation. The lead with the longest QRS duration was used. The Vaughan Williams classification of anti-arrhythmic agents was employed.

Categorical variables were compared using the Fisher's exact test. Continuous variables were expressed as mean±standard deviation and compared using the independent-sample t-test. Crude odds ratios were generated from logistic regression models using either occurrence of arrhythmia event or death as the dependant variables. Adjusted odds ratios were generated from two models with the same dependent variables and the following independent predictors: age (years), gender, follow-up duration (years), primary cardiac aetiology (ischaemic, idiopathic, or other), LV dysfunction (present or absent), indication for ICD placement (primary or secondary), and one of three QRS duration thresholds. Odds ratios from the multivariable logistic regression model using a ≥100 ms threshold were labelled ‘adjusted OR_≥100’ and those from the model using the ≥120 ms threshold were labelled ‘adjusted OR_≥120’ and for 140 ms, OR_≥140. Anti-arrhythmic agents were not included in the models, because they were primarily used to ameliorate ICD shocks (and were therefore not predictors of arrhythmia events). SPSS version 12 was used for all calculations.

Results

Group A consisted of 98 event-free patients and group B contained 131 arrhythmia event patients (Figure 1). A small proportion of patients who were permanently paced prior to ICD implantation were excluded because of paced-QRS (group A, n=9 and group B, n=8). Therefore, analysis was conducted for 89 group A and 123 group B patients. The groups were similar with respect to mean follow-up time, age, gender, CAD, and LV dysfunction (Table 1). Patients in group A were significantly more likely to have had an ICD implantation for primary prevention (group A 18%, group B 7%, P=0.03) or have hypertrophic cardiomyopathy (group A 4%, group B 0%, P=0.03). Patients in group B were significantly more likely to have longer QRS duration (≥100 ms: group A 64%, group B 81%, P=0.007; ≥120 ms: group A 37%, group B 51%, P=0.04) or receive an anti-arrhythmic agent (Class I: group A 1%, group B 9%, P=0.02; Class III: group A 15%, group B 33%, P=0.004). There were no significant differences in the use of beta-blocker or calcium-channel blocker therapy between the two groups.

Arrhythmic events and indication for ICD

The indication for ICD implantation had a significant impact on odds of event-free survival. The 187 (82%) patients with a secondary prevention indication (history of spontaneous ventricular arrhythmias or sudden cardiac death) had significantly higher odds of having an arrhythmic event (crude OR 2.78, 95% CI 1.17–6.62). This difference persisted when adjusted for other variables at both the 100 and 120 ms thresholds (adjusted OR_≥100 2.82, 95% CI 1.11–7.14; adjusted OR_≥120 2.72, 95% 1.08–6.85).

Arrhythmic events and QRS duration

Longer QRS duration predicted higher odds of having an arrhythmic event during follow-up. Crude odds ratios for likelihood of recurrent therapies were significantly higher for two of the three pre-determined QRS thresholds (≥100 ms: OR 2.32, 95% CI 1.24–4.31; ≥120 ms: OR 1.78, 95% CI 1.02–3.11; ≥140 ms: OR 1.24, 95% CI 0.65–2.38). When adjusted for other clinically important variables, the ≥100 ms threshold remained significant (adjusted OR_≥100 2.75, 95% CI 1.37–5.51; adjusted OR_≥120 1.77, 95% CI 0.97–3.23).

Mortality and QRS duration

Higher QRS durations predicted significantly greater odds of mortality at thresholds of ≥100 ms (crude OR 3.90, 95% CI 1.32–11.5; adjusted OR 3.72, 95% CI 1.17–11.9), ≥120 ms (crude OR 3.57, 95% CI 1.70–7.50; adjusted OR 3.09, 95% CI 1.39–6.85), and ≥140 ms (crude OR 3.11, 95% CI 1.50–6.48; adjusted OR 3.68, 95% CI 1.56–8.61).

Discussion

We observed an independent association between longer native QRS duration and occurrence of arrhythmic events (as defined by the occurrence of device therapies) following ICD implantation. In patients with QRS ≥100 ms, there was a 2.8-fold greater likelihood of arrhythmic events compared with shorter QRS durations. For QRS ≥120 ms, this likelihood was 1.8-fold. A primary prevention indication for the device also predicted a lower likelihood of therapies. As expected, QRS duration was also significantly associated with the overall mortality.

Prior studies have shown higher rates of all-cause, cardiac and arrhythmic mortality in patients with prolonged QRS duration and ischaemic heart disease.^1,3,5,8 The increased mortality among these patients could be due to arrhythmias, pump failure, or both. A study of 777 patients referred for electrophysiological testing showed an independent, three-fold association between QRS duration and sustained monomorphic ventricular tachycardia (VT) inducibility, supporting a relationship between QRS duration and ventricular tachyarrhythmias.⁹ Our observations are consistent with these findings. However, a recent retrospective analysis from a randomized controlled trial of ICD therapy modes has presented contradictory findings.¹⁰ Among 431 patients who experienced a total of 95 events after 1-year follow-up, there was no statistically significant difference in event rates based on the QRS duration. Compared with the present study, this previous work from Buxton et al.¹⁰ was an analysis of fewer events (95 vs. 123), shorter follow-up (12 months vs. 48 months), and a greater proportion of primary prevention ICDs (48 vs. 12%). These important differences, in particular the secondary prevention numbers and duration of follow-up, are potential explanations for the discordance in findings between the two studies.

In a registry of 165 ICD patients with functional Class III heart failure and secondary indications for implantation, QRS duration ≥150 ms correlated with increased overall as well as cardiac mortality.² This study also reported that episodes of recurrence of fast VT were higher with longer QRS duration, but the overall freedom from recurrent events was not significant and the analysis was limited to a cut off of QRS duration at 150 ms. In the present study, longer QRS duration was also associated with higher mortality. However, longer QRS duration was also predictive of arrhythmic event occurrence, independent of other factors.

These findings have potential implications for the management of ICD patients. While ICD therapies can be effective, these can also cause significant discomfort and it is the common goal of the physician and the patient that these should be prevented. In a patient with QRS duration over 120 ms, it may be useful to have a lower threshold for initiating a therapeutic agent for the prevention of ventricular arrhythmia. Furthermore, although indications for ICD implantation are expanding, there is significant need for the development of clinical tools to identify the patient at highest risk who will derive maximal benefit from the ICD. As a potential predictor of high ICD usage among recipients, prolonged native QRS duration may assist in future optimization of patient selection for ICDs. Clearly, optimization of selection criteria for ICD recipients will require a multi-pronged approach^11–16 and any single parameter is likely to be insufficient.

Limitations

This was a retrospective, single-centre study. It may have been prone to some selection bias due to follow-up requirements for inclusion. There may be potential for referral bias, although this was likely minimized because a large volume of the institution's ICD implantations were performed under contract with a large local health maintenance organization. Group B patients were more likely to have been treated with anti-arrhythmic agents. As this is the group that received therapies, initiation of such drugs for the prevention of recurrent arrhythmias is an expected finding. However, as most of the patients were on Type III anti-arrhythmic agents (amiodarone and sotalol that do not affect myocardial conduction velocity), an influence on QRS duration is unlikely.

Conclusion

In this ICD population consisting largely of secondary prevention patients, longer native QRS duration was independently associated with higher likelihood of arrhythmic events as defined by the occurrence of ICD therapies. These data have implications for optimizing selection of ICD patients that are more likely to benefit from initiation of prophylactic anti-arrhythmic drug therapy.

Acknowledgements

We would like to thank Ron Oliver, Karen Griffith and Karen Paladino for their invaluable and dedicated effort in maintaining the implantable defibrillator database at our institution, which made this analysis possible. S.S.C. is supported by United States Centers for Disease Control and Prevention/ATPM TS-0660, the Donald W. Reynolds Clinical Cardiovascular Research Center Grant to Johns Hopkins University, NHLBI HL-04-001 and PHS Grant 5 M01 RR000334. E.C.S. and B.T.J. are recipients of post-doctoral fellowship awards from the American Heart Association. M.Z. is the recipient of an Advanced Clinical Training Fellowship grant from the Heart Rhythm Society.

References