Author + information
- Received December 29, 2005
- Revision received May 23, 2006
- Accepted May 30, 2006
- Published online October 17, 2006.
- John J. Hayes, MD, FACC⁎,2 ( )(, )
- Arjun D. Sharma, MD, FACC†,
- John C. Love, MD, FACC‡,
- John M. Herre, MD, FACC§,
- Anna O. Leonen, MS∥,
- Peter J. Kudenchuk, MD, FACC∥,
- DAVID Investigators1
Objectives The purpose of this study was to determine whether QRS duration or morphology increased the risk of adverse outcome in the DAVID (Dual Chamber and VVI Implantable Defibrillator) trial.
Background The DAVID trial found an increased risk of the combined end point of death and new or worsening congestive heart failure (CHF) in defibrillator recipients who were paced DDDR-70 versus VVI-40.
Methods We analyzed the combined end point in patients with abnormal QRS duration (AbQRS) (≥110 ms) compared with those with normal QRS duration (NQRS) (<110 ms).
Results The QRS data were available for 496 of the 506 patients enrolled in the trial, including 223 patients with NQRS (45%) and 273 patients with AbQRS (55%). In patients in whom defibrillators were programmed to pace infrequently (VVI-40), having an NQRS or AbQRS was not an indicator of increased risk of adverse outcome. However, among patients in whom defibrillators were programmed in a manner that promoted more frequent ventricular pacing (DDDR-70), there was a significant adverse interaction with AbQRS; this combination was independently associated with a higher risk for developing CHF or death (p = 0.017).
Conclusions Although patients with AbQRS tended to have other risk factors associated with poor outcome, the interaction of QRS duration with ventricular pacing (DDDR-70) independently contributed to a worse outcome and therefore, was a marker of patients in whom such treatment may be harmful. This should not imply that right ventricular pacing in NQRS patients is safe but rather that pacing in the context of an AbQRS is probably best avoided.
There is a growing body of evidence that the dyssynchrony induced by right ventricular pacing has an adverse effect on outcome in patients undergoing implantation of either a permanent pacemaker or an implantable cardioverter-defibrillator (ICD) (1,2). The DAVID (Dual Chamber and VVI Implantable Defibrillator) trial found an increased risk of the combined end point of death and hospitalization for new or worsening congestive heart failure (CHF) in ICD patients paced DDDR-70 versus VVI-40. It is not clear what factors influence this adverse effect. Prolonged QRS duration, an indicator of ventricular dyssynchrony, is a predictor of adverse outcome in patients with CHF (3). We sought to determine whether QRS duration or morphology was an indicator of increased risk of adverse outcome in the DAVID trial.
We performed a subgroup analysis on DAVID trial patients comparing outcome in those with abnormal QRS duration (AbQRS) (≥110 ms) with those with normal QRS duration (NQRS) (<110 ms). Outcome was also compared between those with right bundle branch block (RBBB), left bundle branch block (LBBB), nonspecific intraventricular conduction delay (IVCD), and normal QRS duration (NQRS).
The DAVID trial was a multicenter, randomized, single-blinded, parallel-arm study of patients with ICDs, comparing VVI and DDDR pacing modes. The methods and results have been previously reported (4). The DAVID patients had a standard indication for an ICD but no indication for antibradycardia pacing. Transvenous dual- chamber ICDs were implanted in all trial participants. After successful implantation, patients were randomly assigned to have the device programmed to pace infrequently (VVI mode with a lower rate of 40/min [VVI-40]) or in a manner that promoted ventricular pacing (DDDR mode with a lower rate of 70/min [DDDR-70]). Programming of the atrioventricular interval was left to the clinical judgment of the investigator, but was most commonly set at 180 ms. At 3 months’ follow-up, the pacing registry of implanted devices indicated that patients in the DDDR-70 group were ventricular paced 58% of the time compared with 2.4% in the VVI-40 group. Patients were followed for 9.2 ± 6 (mean ± SD) months.
The QRS duration was reported as the average of at least 3 measured QRS complexes on the pre-implant 12-lead electrocardiogram. An NQRS was defined as <110 ms and AbQRS as ≥110 ms, based on generally accepted normal values for precordial QRS dimensions (5). Nonspecific IVCD was defined as a QRS duration between 110 and 119 ms, or ≥120 ms in the absence of a typical bundle branch block pattern. Left bundle branch block was defined as a QRS duration ≥120 ms with a wide notched R-wave in the left precordial leads (V4through V6) and lead I, or with a pattern of rR′ in the same leads. Right bundle branch block was defined as QRS duration ≥120 ms with an rsR′ complex in right precordial lead V1, a terminal R in aVR, and a terminal S in lead I.
The primary end point was a combined freedom from death and absence of hospitalization for heart failure. The determination of heart failure hospitalization was based on review of the hospital records by an events committee that was blinded to treatment group. Hospitalization had to satisfy both of the following criteria: 1) admission to hospital for more than 24 h with a clinical history of worsening symptoms of heart failure as evidenced by clinical criteria, including increased New York Heart Association functional class, orthopnea, paroxysmal nocturnal dyspnea, edema, dyspnea on exertion, or gastrointestinal symptoms attributable to heart failure; and 2) 1 or more intensive treatments for CHF within 24 h of admission, such as intravenous diuretics, intravenous inotropic medications, or placement on the status 1 heart transplant list. Follow-up occurred every 3 months.
Differences in continuous and dichotomous variables were analyzed with use of the Student ttest, the chi-square test, or the Mann-Whitney test. Event rates were estimated by the product-limit method (SPSS version 10; SPSS Inc., Chicago, Illinois). Time to primary end point was analyzed using the Kaplan-Meier method. Cox stepwise regression was used to analyze the interaction of QRS group and pacing mode on the primary end point, while accounting for the main effects of baseline covariates, particularly those that differed substantially between the QRS groups. Selection of the appropriate multivariate model was confirmed by the goodness-of-fit chi-square (Hosmer-Lemeshow) test. Statistical significance was indicated by p < 0.05.
Data on QRS duration and morphology were available for 496 of the 506 patients enrolled in the trial. There were 223 patients with NQRS (45%), which averaged 95 ± 10 ms, and 273 patients with AbQRS (55%), which averaged 138 ± 22 ms. Of the 273 patients with AbQRS, 52 (19%) had RBBB, 82 (30%) had LBBB, and 139 (51%) had IVCD morphology. Table 1shows the clinical and laboratory characteristics and medications of patients with normal and abnormal conduction. Patients manifesting abnormal conduction were older, had a lower ejection fraction, were more likely to have a history of heart failure, and less likely to have hyperlipidemia, a percutaneous coronary intervention, or coronary revascularization after trial entry than patients with normal conduction. At hospital discharge after receipt of the ICD, use of amiodarone tended to be higher, and use of other antiarrhythmic agents lower in patients with abnormal conduction, but these differences were not statistically significant. Notably, patients with abnormal QRS conduction also had a longer PR interval at baseline, and had a narrower difference between the programmed atrioventricular delay associated with atrial pacing and that associated with atrial sensing. At 3 months’ follow-up, the frequency of ventricular pacing was higher in patients with AbQRS than in those with NQRS (33 ± 41% vs. 25 ± 36% [mean ± SD]), respectively; p = 0.03).
Effect of ventricular pacing assignment and QRS conduction on outcome
The rate of survival free of hospitalization for new or worsening heart failure was compared between patients with normal and abnormal QRS conduction in whom defibrillators had been programmed to promote (DDDR-70) or not promote (VVI-40) ventricular pacing. There was no difference in this measure of outcome among patients with NQRS, whether they were assigned to be more frequently ventricular paced (DDDR-70) or infrequently paced (VVI-40) (Fig. 1)(p = 0.94). Nor were there significant differences in outcome among patients with NQRS or AbQRS if assigned to infrequent pacing (VVI-40) (Fig. 2)(p = 0.59). Conversely, those with an AbQRS in whom ventricular pacing was promoted (DDDR-70) had a worse outcome than patients in whom it was not (VVI-40) (Fig. 3)(p = 0.01). Figure 4depicts outcome for the 3 subgroups of patients with AbQRS—those with IVCD, RBBB, or LBBB. Differences in outcome between these 3 subgroups were qualitatively similar, although only those with LBBB achieved a statistically significant difference in the primary end point between DDDR-70 and VVI-40 groups. That is, patients with LBBB in whom ventricular pacing was promoted (DDDR-70) appeared to have a higher likelihood of death or heart failure than those in whom it was not (VVI-40) (p = 0.03), although there was a visually apparent divergence in the survival curves for all 3 AbQRS subgroups.
Cox regression analysis was performed, adjusting the primary outcome for differences in baseline variables, including age, ejection fraction, revascularization, history of heart failure, history of diabetes, serum creatinine, PR interval, and use of nitrates, followed, in turn, by QRS duration, promoted ventricular pacing (DDDR-70), and the interaction between QRS duration and ventricular pacing (DDDR-70), as depicted in Table 2.The regression analysis identified a significant effect upon outcome stemming from the combination of an AbQRS, ventricular pacing (DDDR-70), and the interaction between AbQRS and pacing corresponding to that seen in Figures 3 and 4(p = 0.017). That is, the adverse effects on the primary outcome when ventricular pacing was promoted (DDDR-70) were much more pronounced in the AbQRS than in the NQRS group. When not adjusted for baseline variables, there was only a trend toward an interaction between QRS duration and treatment upon the primary outcome (p = 0.13), probably owing to imbalances in other predictive characteristics between groups. There was an insufficient number of cases to distinguish whether this adverse effect was independently predicted by the specific conduction abnormality (LBBB, RBBB, or nonspecific IVCD), as was suggested by the Kaplan-Meier survival analysis. For example, despite the apparent relationship between LBBB and treatment on outcome observed in the univariate analysis, the interaction between the two, either unadjusted or adjusted for baseline variables, was not statistically significant (p = 0.45 and 0.42, respectively). This may have been because no such interaction truly existed or because the analysis was underpowered. Finally, in support of our prospective selection of a conventional cut point of ≥110 ms to define abnormal QRS duration, when the regression analysis was restricted to patients who already manifested such abnormal conduction, further increments in QRS duration were not in themselves found to be a significant predictor of the primary end point (p = 0.89).
Across the spectrum from narrow to markedly prolonged, QRS duration has been shown to be an independent marker of mortality in patients with and without congestive heart failure (1,3,6). Recent studies have also observed that, when combined with severe left ventricular dysfunction, right ventricular pacing worsens clinical outcome (4,7), particularly when it occurs frequently (8). In addition, multiple studies have shown the benefits of biventricular pacing in patients with prolonged QRS duration and CHF suggesting that a widened QRS is not just a marker of disease severity, but a possible indicator of a treatable functional impairment (9,10). Taken together, these observations support the existence of an adverse and potentially modifiable interaction between heart failure, QRS duration, and pacing that was addressed by the present study.
When ventricular pacing was promoted (DDDR-70) among patients with impaired heart function, we found that those with a prolonged QRS duration (≥110 ms) had a higher risk of death or new or worsening congestive heart failure compared with patients with a baseline narrow QRS (<110 ms). When ventricular pacing was not promoted (VVI-40), having a normal or an abnormal QRS duration was not an indicator of increased risk of adverse outcome. Nor, apart from pacing considerations, was increasing QRS duration found to independently predict an adverse outcome among patients in whom QRS was already prolonged at baseline (≥110 ms). However, in combination with assignment to more frequent ventricular pacing (DDDR-70), a prolonged QRS was independently associated with a higher risk of an adverse outcome. These results suggest that patients with some degree of baseline dyssynchrony may be more susceptible to the adverse effects of further dyssynchrony induced by right ventricular pacing. Although patients with a wide QRS tended to have other risk factors associated with poor outcome (such as increasing age, worse ejection fraction, and higher likelihood of prior congestive heart failure), the adverse interaction of QRS duration with assignment to more frequent ventricular pacing (DDDR-70) contributed to a worse outcome independent of these other risk factors and, as such, was a marker of patients in whom such treatment should be avoided.
Pacing in bundle branch block
Although it could not be established as an independent predictor, the adverse outcome in patients with increased QRS duration in whom ventricular pacing was promoted (DDDR-70) appeared to be more strongly associated with LBBB than with RBBB or nonspecific IVCD. This observation is consistent with the known mechanical and hemodynamic consequences of LBBB on left ventricular function, which is simulated, and arguably might even be exaggerated, by right ventricular pacing (11). However, nonsignificant trends toward worse outcome in patients with RBBB and IVCD when paced also were observed which were qualitatively similar to those with LBBB. It is possible that ventricular pacing is detrimental in the presence of an underlying conduction disorder in any of its forms, and this trial was simply underpowered to establish these differences. It is also possible that the subgroup of patients with IVCD, which included those with more minor degrees of QRS prolongation, might be expected to behave more like patients with NQRS and have less detrimental effects from ventricular pacing. Similarly patients with RBBB (in whom the benefit from resynchronization therapy is less certain (12) may, by the same token, sustain less harm from the change in ventricular activation sequence caused by right ventricular pacing.
Pacing in patients with a normal QRS
In our study we found no difference in the primary end point among patients with a normal QRS in whom ventricular pacing was (DDDR-70) or was not (VVI-40) promoted. This should not necessarily be taken to indicate there are no adverse effects from right ventricular pacing in patients with a narrow baseline QRS. A subgroup analysis of the MOST (Mode Selection Trial) pacing trial showed an increased risk of development of CHF and atrial fibrillation in patients with a “normal” QRS who received ventricular pacing (13). One explanation for this apparent discrepancy between MOST and DAVID lies in their selection of a wider QRS (<120 ms) to define normal than the more conventional definition (<110 ms) used in our trial. In the context of a wider than normal QRS, both trials could be said to have observed ventricular pacing to be detrimental, whereas this may or may not have been the case when the QRS was truly narrow. Another explanation for this discrepancy is the difference in the frequency of ventricular pacing between the 2 studies. The DDDR group in MOST was 90% ventricular paced compared with 58% paced in DAVID. The VVI group was 58% paced in MOST compared with 2.4% in DAVID. These differences stem from the different programming parameters as well as the fact that MOST patients had a primary indication for antibradycardia pacing. The MOST substudy observed an increasing risk of adverse outcomes associated with the frequency of ventricular pacing. It is therefore possible that patients in DAVID with a narrow QRS complex would have had a higher event rate had their frequency of ventricular pacing been greater or their duration of follow-up been longer.
This study is limited by the post-hoc nature of the analysis. The DAVID trial was not primarily designed to study the effects of QRS duration on outcome or in relation to ventricular pacing. The results do suggest, however, that patients with baseline electrocardiographic evidence of dyssynchrony may be particularly prone to the adverse effects of right ventricular pacing. Increased attention to programming and novel pacing modes are needed to avoid right ventricular stimulation where possible. Prospective studies are required to define better alternatives, such as biventricular pacing, for broader groups of patients who require ventricular pacing.
Supported by a grant from St. Jude Medical, Sylmar, California. A complete list of the DAVID Investigators and Coordinators is published in reference 4(JAMA 2002;288:3115–23).
- Abbreviations and Acronyms
- abnormal QRS
- congestive heart failure
- Dual Chamber and VVI Implantable Defibrillator
- implantable cardioverter-defibrillator
- intraventricular conduction delay
- left bundle branch block
- Mode Selection Trial
- normal QRS
- right bundle branch block
- Received December 29, 2005.
- Revision received May 23, 2006.
- Accepted May 30, 2006.
- American College of Cardiology Foundation
- Sweeney M.O.,
- Hellkamp A.S.,
- Lee K.L.,
- Lamas G.A.,
- MOST Investigators
- Rowlands D.J.
- Sweeney M.O.,
- Hllkamp A.S.,
- Ellenbogen K.A.,
- et al.,
- MOST Investigators