Author + information
- Received December 5, 2016
- Revision received February 2, 2017
- Accepted February 10, 2017
- Published online April 17, 2017.
- Varun Sundaram, MDa,b,
- Jayakumar Sahadevan, MDa,c,∗ (, )
- Albert L. Waldo, MD, PhD, (Hon)a,
- George J. Stukenborg, PhDd,
- Yogesh N.V. Reddy, MDe,
- Samuel J. Asirvatham, MDe,
- Judith A. Mackall, MDa,
- Anselma Intini, MDc,
- Brigid Wilson, PhDc,
- Daniel I. Simon, MDa and
- Kenneth C. Bilchick, MDd
- aHarrington Heart and Vascular Institute, University Hospitals Cleveland Medical Center, Case Western Reserve University, Cleveland, Ohio
- bRoyal Brompton and Harefield Hospitals, National Heart and Lung Institute, Imperial College, London, United Kingdom
- cDepartment of Medicine, Louis Stokes Veteran Affairs Medical Center, Cleveland, Ohio
- dDivision of Cardiovascular Medicine, University of Virginia Health System, Charlottesville, Virginia
- eDivision of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota
- ↵∗Address for correspondence:
Dr. Jayakumar Sahadevan, Department of Cardiovascular Medicine, Harrington Heart and Vascular Institute, University Hospitals Cleveland Medical Center, Louis Stokes Veteran Affairs Medical Center, 10701 East Boulevard, Cardiology Section (111W), Cleveland, Ohio 44106.
Background More than 20% of Medicare beneficiaries receiving cardiac resynchronization therapy defibrillators (CRT-D) have a very wide (≥180 ms) QRS complex duration (QRSD). Outcomes of CRT-D in these patients are not well-established because they have been underrepresented in clinical trials.
Objectives This study examined outcomes in patients with CRT-D in a very wide QRSD with left bundle branch block (LBBB) versus those without LBBB.
Methods Medicare patients from the Implantable Cardioverter Defibrillator Registry (January 1, 2005, through April 30, 2006) with a CRT-D and confirmed Class I or IIa indications for CRT-D were matched to implantable cardioverter-defibrillator (ICD) patients without CRT despite having Class I or IIa indications for CRT. Mortality and heart failure hospitalizations over 4 years with CRT-D versus standard ICDs based on QRSD and morphology were analyzed.
Results We analyzed 24,960 patients. Among those with LBBB, patients with a QRSD ≥180 ms had a greater adjusted survival benefit with CRT-D versus standard ICD (hazard ration [HR] for death: 0.65; 95% confidence interval [CI]: 0.59 to 0.72) compared with those having a QRSD 120 to 149 ms (HR: 0.85; 95% CI: 0.80 to 0.92) and 150 to 179 ms (HR: 0.87; 95% CI: 0.81 to 0.93). CRT-D versus ICD was associated with an improvement in survival in those with non-LBBB and a QRSD ≥180 ms (adjusted HR for death: 0.78; 95% CI: 0.68 to 0.91), but not in those with non-LBBB and a QRSD 150 to 179 ms (adjusted HR for death: 1.06; 95% CI: 0.95 to 1.19).
Conclusions Improvements in both survival and heart failure hospitalizations with CRT-D were greatest in patients with a QRSD ≥180 ms with or without LBBB, whereas patients with a QRSD 150 to 179 ms without LBBB had no improvement in survival with CRT-D, and those with a QRSD 150 to 179 ms and LBBB had only a modest improvement.
Cardiac resynchronization therapy defibrillators (CRT-Ds) have been associated with marked reductions in mortality and heart failure (HF) hospitalizations in patients with heart failure and reduced ejection fraction (HFrEF) (1–3). However, 30% of patients meeting current CRT criteria do not achieve CRT response (4). Based on data from randomized clinical trials, patients with a pre-CRT QRS complex duration (QRSD) ≥150 ms benefit more than those with a QRSD of 120 to 149 ms (5). Furthermore, patients with left bundle branch block (LBBB) and a QRSD ≥150 ms have a stronger guideline-based indication for CRT-D (Class I) than patients with a QRSD 120 to 149 ms, whereas patients without LBBB and a QRSD ≥150 ms have a Class IIa indication for CRT-D (6). According to guideline-based recommendations (6), patients with LBBB and a very wide QRSD (VWQRSD) (≥180 ms) have a Class I indication for CRT-D and patients without LBBB who have a QRSD ≥180 ms have a Class IIa indication for CRT-D. However, these groups have not been well-studied because they have been underrepresented in CRT clinical trials (3,7).
A QRSD ≥120 ms can be due to either conduction block in the bundle branches, electrical uncoupling in the ventricular myocardium, or a combination of both. Electrical uncoupling in HF results from altered expression of connexins that assemble gap junctions (8). Gap junctions are critical in regulating conduction of electrical impulses within the myocardium. Decreased expression and/or disruption of gap junctions decrease the conduction velocity in the myocardium (9), leading to a wide QRSD on the electrocardiogram.
It has been shown from actual heart and computer simulation studies that with true LBBB, the QRSD is in the range of 140 ms (10); with true right bundle branch block (RBBB), the QRSD is even less. Using a realistic computer model of the human heart and torso, Potse et al. (11) demonstrated that by decreasing electrical coupling by 50% in the absence of LBBB, the QRSD increased from 90 to 120 ms; in the presence of LBBB with electrical uncoupling, the QRSD increased from 140 to 190 ms. It is also possible that a QRSD could be prolonged beyond 140 ms in the presence of left ventricular dilatation with LBBB. In the presence of significant electrical uncoupling, the benefits of CRT may be negated by slow and dispersed conduction during pacing (12). Based on these considerations, we hypothesized that 1 reason for nonresponsiveness to CRT may be the presence of significant electrical uncoupling; therefore, patients with a VWQRSD (≥180 ms) would have less favorable outcomes than patients with a moderately wide QRSD (150 to 179 ms), and different outcomes depending on bundle branch block morphology (BBB).
To evaluate outcomes in patients with a VWQRSD, we stratified patients by a QRSD and compared long-term outcomes among 24,960 Medicare patients identified in the Implantable Cardioverter Defibrillator Registry (ICDR) who received CRT-D with those of a propensity-matched control cohort receiving standard implantable cardioverter-defibrillators (ICDs) that had an indication for CRT-D.
Records were obtained for patients with CRT-Ds and standard ICDs from the original ICDR maintained by the Iowa Foundation for Medical Care under contract with Medicare from January 2005 and to April 2006 (13,14). During this period, all health care providers implanting devices funded by Medicare were required to enter patient information including demographics, history, clinical investigations, and device information into a patient registry. This Medicare ICDR was maintained by the Iowa Foundation for Medical Care through April 2006 and constitutes a distinct dataset.
Patients were classified on the basis of whether they had CRT-D or ICD devices implanted. Patients with CRT-D or a standard ICD who had an accepted Class I or Class IIa recommendation for CRT-D as per the American Heart Association/American College of Cardiology/Heart Rhythm Society guidelines were included in the analysis. All patients were required to have a left ventricular ejection fraction ≤0.35 and New York Heart Association functional class II to IV HF. In addition, patients with LBBB were required to have a QRSD ≥120 ms, and patients without LBBB were required to have a QRSD ≥150 ms. Patients without LBBB and a QRSD of 120 to 149 ms were not included in our analysis because they do not meet Class I or Class IIa recommendations for CRT-D placement. All patients with ICDs were required to meet the same criteria for inclusion as those with CRT-D.
Outcome data and linkage
Medicare hospital claims files, program eligibility records, and dates of death were also obtained for these patients for the period from ICD implantation through at least the end of 2009. Registry records were matched to Medicare eligibility and utilization data using patients’ social security numbers by the U.S. Centers for Medicare & Medicaid Services. In addition, Medicare claims data include International Classification of Diseases, 9th Revision, Clinical Modification diagnosis codes, which were used to determine which patients had hospitalizations for a primary diagnosis of HF.
Multivariable logistic regression was used to estimate the probability of receiving CRT-D among all patients in the standard ICD groups based on 36 covariates. The Wald chi-square test statistic was used to measure the relative contribution of each individual covariate to overall model performance. Using custom code for SAS version 9.4 (Cary, North Carolina), a propensity score was calculated for each patient on the basis of the probability of receiving CRT-D versus a standard ICD in this multivariable logistic regression model. Every patient with CRT-D was matched with replacement to each patient in the ICD group on the basis of the closest propensity score (probability of receiving CRT-D). Some patients with a standard ICD were matched more than once to a patient with CRT-D (matching with replacement), based on standard propensity-matching methodology. After matching was complete, 2 CRT-D patients were found to have invalid follow-up data and were not included in the survival analyses that follow. An appropriate distributional balance in CRT-D and ICD groups was confirmed for all covariates included in the model for CRT-D and matched ICD groups using bivariable logistic regression based on the Wald chi-square test statistic. The C statistic was used to measure model discrimination for which a value of 0.5 indicates no discrimination and a value of 1.0 indicates perfect discrimination. Balance between CRT-D and ICD groups was also assessed using the standardized difference, which is preferred over standard p values (15,16). Relative density functions were plotted to demonstrate the distribution of propensity scores in the CRT-D and ICD groups.
Statistical analyses were performed using SAS 9.4. Differences in mean values of continuous variables between the CRT-D and ICD groups were assessed using the Kruskal-Wallis test statistic. The statistical significance of differences in the frequency of categorical variable values between groups was assessed using the chi-square test statistic. Patients with CRT-D and standard ICDs were grouped into 5 categories based on their BBB morphology and QRSD, reflecting the CRT-D indication: 1) LBBB with a QRSD 120 to 149 ms; 2) LBBB with a QRSD 150 to 179 ms; 3) LBBB with a QRSD ≥180 ms; 4) non-LBBB with a QRSD of 150 to 179 ms; and 5) non-LBBB with a QRSD ≥180 ms. Secondary analyses were also performed based on dichotomization of the non-LBBB groups into those with RBBB and intraventricular conduction delay (IVCD) morphologies. Differences in survival and survival free of HF hospitalization (HFH) were compared between the propensity score matched CRT-D and standard ICDs groups in the 5 prespecified subgroups based on BBB morphology and QRSD using Kaplan- Meier survival curves. The statistical significance of differences between groups was assessed using the log-rank test. Multivariable Cox proportional hazards regression analysis was used to estimate differences in overall survival, survival free of HFH, and HFH with death as a competing risk with adjustment for the concurrent effects of other covariates. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the outcomes of overall survival, survival free of HFH, and HFH with death as a competing risk were determined for CRT-D versus ICD therapy for each of the 5 BBB morphology/QRS groups. The interaction term for CRT-D and a variable representing whether the QRS was ≥180 ms or <180 ms was used to assess whether clinical outcomes were different in patients with a VWQRSD versus the remainder of the patients in LBBB and non-LBBB groups. The ASSESS option of Proc PHREG was used to plot the cumulative score residuals against time to verify the proportional hazards assumption. Conditional models were used to account for comparisons of matched groups. Competing risks analysis was used to adjust for differences in the occurrence of death as a competing event in the analysis of time to HFH events using a semiparametric Cox proportional hazards model (17–20).
We analyzed 24,960 patients, with 12,480 CRT-D patients matched with 12,480 ICD patients in the Medicare ICDR. The mean age of patients in the registry was 73.0 ± 10.5 years. Table 1 shows characteristics of patients who received CRT-D versus those who received ICD in a propensity-matched cohort. The standardized difference (a better index of balance than the p value for propensity-matched cohorts) was <0.10 for all covariates, which indicates well-matched propensity cohorts with negligible differences (15,16). The C statistic of 0.76 indicated effective propensity matching, and the overlapping relative density functions for the CRT-D and matched ICD groups shown in Figure 1 also indicate excellent balance between the groups.
Overall, there were 7,996 patients with LBBB and a QRSD 120 to 149 ms; 8,545 patients with LBBB and a QRSD 150 to 179 ms; 4,025 patients with LBBB and a QRSD ≥180 ms; 2,794 patients with non-LBBB and a QRSD 150 to 179 ms; and 1,600 patients with non-LBBB and a QRSD ≥180 ms. The distribution of the baseline characteristics across the 5 groups are shown in Table 2. Of note, there were no significant differences in the number of CRT-D and ICD patients in these subgroups (p = 0.28), indicating an appropriate balance of patients with CRT-D and those with a standard ICD in each subgroup. Significant differences in other covariates are typical with very large cohort sizes, and the standard differences shown in Table 1 between CRT-D and ICD groups were very small. Residual differences among the 5 groups in Table 2 were addressed effectively through subsequent statistical adjustment in the multivariable Cox proportional hazards models for the clinical outcomes.
Outcomes of patients with CRT-D compared with standard ICD
Stratification by BBB and QRSD
Figures 2A to 2J are Kaplan-Meier estimates of overall survival and survival free of HFH comparing patients who received CRT-D with those of a propensity-matched cohort receiving standard ICDs within each group of QRSD.
In patients with LBBB and a VWQRSD (Figures 2C and 2H), those with CRT-D had superior overall survival (p < 0.0001) and survival free of HFH (p < 0.0001) when compared with the matched ICD controls. This finding was also observed in patients with a QRSD of 120 to 149 ms (p < 0.0001) and 150 to 179 ms (p = 0.0004) with LBBB (Figures 2A, 2B, 2F, and 2G). In patients without LBBB, a completely different pattern was observed. In the group with a VWQRSD (Figures 2E and 2J), patients with CRT-D demonstrated superior overall survival (p < 0.0035) and survival free of HFH (p = 0.004) when compared with the group with ICD alone. Surprisingly, patients without LBBB and a QRSD of 150 to 179 ms had lower overall survival with CRT-D (p = 0.02) (Figure 2D) and no difference in survival free of HFH (p = 0.82) when compared with the matched ICD controls (Figure 2I).
The Central Illustration demonstrates forest plots showing adjusted HRs for death across all groups comparing patients with CRT-D and matched controls with ICD alone. In these models, adjusted HRs were determined with adjustment for the covariates used in the propensity matching. The adjusted HRs for overall survival for LBBB with QRS durations of 120 to 149 ms, 150 to 179 ms, and ≥180 ms were 0.85 (95% CI: 0.80 to 0.92), 0.87 (95% CI: 0.81 to 0.93), and 0.65 (95% CI: 0.59 to 0.72), respectively. Among patients with LBBB, the interaction term comparing the impact of CRT-D on survival in patients with QRS <180 ms versus patients with QRS ≥180 ms had an associated p value <0.0001, confirming a greater benefit with CRT in LBBB patients with a QRSD ≥180 ms versus those with QRS duration <180 ms. With respect to patients without LBBB, the HRs for survival for a QRSD of 150 to 179 ms and ≥180 ms were 1.06 (95% CI: 0.95 to 1.19) and 0.78 (95% CI: 0.68 to 0.91), respectively. In patients without LBBB, the interaction term comparing the impact of CRT on survival in patients with a QRS <180 ms versus patients with QRS ≥180 ms was associated with a p value of 0.0097, indicating a greater CRT-D benefit in patients without LBBB who had a QRSD ≥180 ms versus those with a QRSD <180 ms.
Similar forest plots are shown in Figure 3 for the combined outcome of death or HFH. Results for the analysis of HFH adjusted for death as a competing risk and adjusted for the covariates used in the propensity matching are shown in Figure 4. The greatest risk reduction for HFH was observed in patients with LBBB and a QRSD ≥180 ms; patients with LBBB and a QRSD of 150 to 179 ms also had a risk reduction. A statistically significant difference was confirmed for the interaction term for CRT-D and a QRSD ≥180 ms versus a QRSD <180 ms (p < 0.0001). There was no reduction in HFH in patients with LBBB and a QRSD 120 to 149 ms or in any of the groups without LBBB. Even with this neutral effect of CRT-D for HFH in patients without LBBB, there was still a net reduction in the combined endpoint of death or HFH in patients without LBBB and a QRSD ≥180 ms, as shown in Figures 2 and 4, which was completely driven by the risk reduction in death during follow-up in this group.
Subgroup analysis of patients without LBBB
The analysis of the subgroups of patients without LBBB on the basis of whether an IVCD or RBBB was present demonstrated consistent effects in the subgroups. In patients with a QRSD of 150 to 179 ms, overall survival with CRT-D versus the matched ICD controls was similar for both patients with an IVCD (HR: 1.10; 95% CI: 0.94 to 1.28) and those with RBBB (HR: 0.95; 95% CI: 0.79 to 1.13), as shown in the forest plot in the Central Illustration, panel B. In patients with a VWQRSD, there was a consistent overall survival benefit with CRT-D versus standard ICDs for both patients with an IVCD (HR: 0.82; 95% CI: 0.69 to 0.98) and those with RBBB (HR 0.57, 95% CI 0.44 to 0.75) (Central Illustration, panel B). The effect of CRT-D on the combined endpoint of death or first HFH in patients with non-LBBB and a QRSD of 150 to 179 ms was similar for those with IVCD (HR: 0.91; 95% CI: 0.80 to 1.03) and RBBB (HR: 0.92; 95% CI: 0.79 to 1.07), as shown in Figure 3B. In patients with a VWQRSD, CRT-D was favored over the ICD for the combined endpoint of death or first HFH in both patients with an IVCD (HR: 0.85; 95% CI: 0.73 to 0.99) and those with RBBB (HR: 0.66; 95% CI: 0.52 to 0.84) (Figure 3B). A similar analysis for the time to first HFH accounting for the competing risk of death shows similar HRs for HFH with QRS ≥180 ms in subgroups without LBBB (Figure 4B).
Contrary to our hypothesis, this large Medicare ICDR-based study demonstrated that a VWQRSD was associated with superior clinical outcomes after CRT-D compared with outcomes for a propensity-matched cohort receiving standard ICD, regardless of whether LBBB was present or absent.
Among patients with LBBB and a QRSD ≥180 ms, patients who received CRT-D had a much greater adjusted survival compared with matched ICD patients when compared with those having LBBB and a QRSD <180 ms. In patients without LBBB and a QRSD ≥180 ms, there was an improvement in both overall survival and survival free of HFH with CRT-D versus matched standard ICD patients, but there was no improvement in these outcomes in patients without LBBB and a QRSD of 150 to 179 ms. Improvements in both overall survival and survival free of HFH were strongest in patients with QRSD ≥180 ms in the presence or absence of LBBB, whereas currently indicated patients with a QRSD of 150 to 179 ms either had no improvement in outcomes with CRT-D (in the case of non-LBBB) or only a modest improvement (in the case of LBBB).
Electrical uncoupling in HF and its relevance in CRT: Relationship to prior studies
To our knowledge, outcomes with CRT-D in patients with a VWQRSD (≥180 ms) have not been systematically evaluated in a large cohort of patients. HF results in altered expression of connexins, consequently leading to dysregulation of gap junctions. Gap junctions play a vital role in conduction of electrical impulses in the myocardium, dysregulation of which causes electrical uncoupling (9). The presence of a VWQRSD reflects not only the conduction disease of the bundles, but also the electrical uncoupling from diseased myocardium. There is a potential for the effects of CRT to be negated by slow and dispersed conduction because of the contribution of electrical uncoupling at a VWQRSD (12).
Our analysis of patients in the Medicare ICDR demonstrates that patients with HFrEF and a VWQRSD had better clinical outcomes with CRT-D when compared with propensity-matched ICD controls regardless of the BBB morphology. Our study suggests that, in patients with HF, although a VWQRSD is possibly a marker of advanced electrical and myocardial remodeling, the pathophysiologic progression of disease can still be partially reversed with CRT-D irrespective of BBB morphology. The benefits observed in patients with a VWQRSD could plausibly be due to a greater degree of mechanical dyssynchrony, because studies using cardiovascular magnetic resonance strain imaging have shown that patients with the greatest mechanical dyssynchrony stand to benefit the most from CRT (21–23). To our knowledge, these results have not been reported in previous studies, and these data from the ICDR facilitated evaluation of this important physiologic hypothesis because of the larger sample size of subgroups of interest.
Relevance of data from large registries
Although randomized clinical trials form the basis of guideline recommendations, they have their own limitations (1,2,4,5). The Medicare ICDR included data from an older population with more comorbidities than patients included in clinical trials. In addition, the follow-up period of patients in this registry was significantly longer, with a median follow-up of 48 months, in contrast to 26 months in MADIT CRT (Multicenter Automatic Defibrillator Implantation With Cardiac Resynchronization Therapy) (24) and 16 months in the COMPANION (Comparison of Medical, Pacing, and Defibrillation Therapies in Heart Failure) trial (3).
In patients with HFrEF, the current guidelines have recommended implantation of CRT-D devices in patients without LBBB and a QRSD >150 ms (Class IIa), which would also include patients with a VWQRSD ≥180 ms (1,6). There have been no randomized controlled trials specifically addressing CRT-D outcomes in patients without LBBB, and the current guidelines are based on subgroup analyses and a meta-analysis of randomized trials in which most patients demonstrated LBBB at baseline. The present analysis of the Medicare ICDR demonstrates that outcomes in patients without LBBB morphology and a QRSD of 150 to 179 ms were similar to those of matched ICD controls. This suggests that the benefit of resynchronization pacing in patients falling into the current Class IIa guideline-based indication for CRT-D, without LBBB and a QRSD of ≥150 ms, may only be realized in patients with an even wider QRSD. The analysis also suggests that the greatest benefit for CRT-D in LBBB may only be realized in patients with a QRSD approaching or exceeding 180 ms. Reconciliation of these results with results from prior clinical trials may be achieved based on the following: 1) CRT benefit in patients with a VWQRSD of at least 180 ms may drive the benefit in the group of patients with a wide QRSD; 2) real-world outcomes in patients implanted apart from clinical trials may differ from those in the highly selected patients enrolled in clinical trials; and 3) follow-up times in the present analysis were longer compared with some clinical trials.
In January 2005, the U.S. Centers for Medicare & Medicaid Services established the original Medicare ICDR only as an initial data collection mechanism, and hospitals had to transition to the American College of Cardiology-National Cardiovascular Data Registry by April 2006. Our access to these data was limited to patients from the original Medicare ICDR, which enrolled patients between January 2005 and April 2006. Although guideline recommendations for CRT-D implantation have changed in recent years, we analyzed only patients with a current CRT indication based on the most recent guidelines, and the vast majority of patients were on guideline-directed medical therapy for HF including angiotensin-converting enzyme inhibitors and beta-blockers. Although device technology has also evolved during this time, the fundamental CRT procedure remains essentially the same. The Medicare ICDR we used had a wide range of patient information, but certain important patient characteristics, such as biomarkers and right ventricular function, were not available in this database. Furthermore, data on whether the left ventricular pacing site were optimal relative to electrical and mechanical activation and also in relation to scar were not available in this database. We were also not able to obtain information on follow-up electrocardiograms and echocardiograms, which reflect electrical and mechanical changes with CRT.
However, we believe that these limitations are outweighed by the many advantages of the study design, including the population-based perspective provided by analysis of registry data for CRT-D and standard ICD patients with analysis of real-world outcomes with long-term follow-up in a large number of patients, multivariable adjustment for more than 35 covariates, use of a well-matched control group, and isolation of the resynchronization pacing component of CRT-D on clinical outcomes.
Our analysis of the large Medicare ICDR suggests that in patients with HFrEF and a VWQRSD, CRT-D may be associated with superior clinical outcomes with either LBBB or without LBBB when compared with a propensity-matched cohort receiving standard ICDs. Specifically, improvements in survival with CRT-D were greatest in patients with a QRSD ≥180 ms both with and without LBBB, whereas currently indicated patients with a QRSD 150 to 179 ms either had no improvement in outcomes with CRT-D (in the case of patients without LBBB) or a modest improvement (in patients with LBBB). These results promise to be very helpful for clinical decision-making with respect to CRT-D implantation versus standard ICD implantation because patients without LBBB appear to require a greater QRSD than previously realized to experience substantial benefit from resynchronization pacing, whereas the benefit of CRT-D in LBBB appears to be substantially improved with a QRSD ≥180 ms.
COMPETENCY IN MEDICAL KNOWLEDGE: In Medicare recipients with a QRSD ≥180 ms, use of CRT-D devices is associated with lower rates of HF hospitalization and mortality, irrespective of QRS morphology. Patients with QRS durations 150 to 179 ms and LBBB have only modest improvement in survival with CRT-D therapy; patients without LBBB had no improvement.
TRANSLATIONAL OUTLOOK: Randomized trials are needed to assess the benefit of CRT therapy in patients with QRS durations shorter than 180 ms.
Dr. Sahadevan is supported by Ohio Third Frontier and the Clinical and Translational Science Collaborative grants. Dr. Waldo has received a research contract from and served as a consultant to Gilead Sciences; consulted on a clinical trial steering committee for Atricure; consulted for Milestone Pharmaceuticals and Biosense-Webster; served on a scientific advisory board for Daiichi-Sankyo; served on a trial steering committee for St. Jude Medical; and served as a speaker for Bristol-Myers Squibb and Pfizer. Dr. Reddy is supported by T32 HL007111 from the National Institutes of Health. Dr. Mackall has served as a consultant for and undertaken clinical research for St. Jude Medical. Dr. Bilchick is supported by grant R03 HL135463 from the National Institutes of Health. Drs. Sundaram and Sahadevan contributed equally to this work.
- Abbreviations and Acronyms
- bundle branch block
- confidence interval
- cardiac resynchronization therapy defibrillator
- heart failure
- heart failure hospitalization
- heart failure with reduced ejection fraction
- hazard ratio
- Implantable Cardioverter Defibrillator Registry
- intraventricular conduction delay
- left bundle branch block
- QRS complex duration
- right bundle branch block
- very wide QRS complex duration
- Received December 5, 2016.
- Revision received February 2, 2017.
- Accepted February 10, 2017.
- Daubert J.C.,
- Saxon L.,
- Adamson P.B.,
- et al.
- Fornwalt B.K.,
- Sprague W.W.,
- BeDell P.,
- et al.
- Tracy C.M.,
- Epstein A.E.,
- Darbar D.,
- et al.
- Potse M.,
- Krause D.,
- Bacharova L.,
- Krause R.,
- Prinzen F.W.,
- Auricchio A.
- ↵ICD Registry. Available at: https://www.cms.gov/Medicare/Medicare-General-Information/MedicareApprovedFacilitie/ICDregistry.html. Accessed February 27, 2017.
- Bilchick K.,
- Stukenborg G.J.
- Lau B.,
- Cole S.R.,
- Gange S.J.
- Ramachandran R.,
- Chen X.,
- Kramer C.M.,
- Epstein F.H.,
- Bilchick K.C.
- Bilchick K.C.,
- Kuruvilla S.,
- Hamirani Y.S.,
- et al.
- Bilchick K.C.