Author + information
- Received December 17, 2012
- Revision received March 4, 2013
- Accepted March 6, 2013
- Published online June 1, 2013.
- Zak Loring, BS∗,†,
- Daniel A. Caños, MPH∗,
- Kimberly Selzman, MD, MPH∗,‡,
- Naomi D. Herz, BS∗,
- Henry Silverman, BS§,
- Thomas E. MaCurdy, PhD§,
- Christopher M. Worrall, BS‖,
- Jeffrey Kelman, MD, MMSc‖,
- Mary E. Ritchey, PhD∗,
- Ileana L. Piña, MD, MPH∗,¶ and
- David G. Strauss, MD, PhD∗∗ ()
- ↵∗Reprint requests and correspondence:
Dr. David G. Strauss, FDA, Center for Devices and Radiological Health, 10903 New Hampshire Avenue, WO 62-1126, Silver Spring, Maryland 20993.
Objectives The goal of this study was to test the hypothesis that in recipients of cardiac resynchronization therapy defibrillators (CRT-D), conventional left bundle branch block (LBBB) diagnosis predicts better survival in women than in men.
Background New York Heart Association class I and II patients without LBBB do not benefit from CRT-D, and women have better survival after CRT-D than men. Separate analysis suggests that QRS duration thresholds for LBBB diagnosis differ according to sex, and conventional LBBB electrocardiographic criteria are falsely positive in men more frequently than in women.
Methods We analyzed Medicare records from 144,642 CRT-D recipients between 2002 and 2008 that were followed up for up to 90 months. Medicare billing data were used to determine age, sex, race, and comorbidities. Hazard ratios (HRs) were calculated to assess if conventional LBBB diagnosis had different prognostic significance according to sex.
Results In univariate analysis, LBBB was associated with a 31% reduction in death in women (HR: 0.69 [95% confidence interval (CI): 0.67 to 0.71]) but only a 16% reduction in death in men (HR: 0.84 [95% CI: 0.82 to 0.85]). In multivariable analyses controlling for comorbidities, LBBB was associated with a 26% reduction in death in women (HR: 0.74 [95% CI: 0.71 to 0.77]) and a 15% reduction in death in men (HR: 0.85 [95% CI: 0.83 to 0.87]). A significant interaction (p < 0.0001) between sex and LBBB was seen.
Conclusions LBBB diagnosis is associated with greater survival in women than in men receiving CRT-D, and this discrepancy is not explained by differences in measured comorbidities. Possible explanations for this difference include that LBBB may have different prognostic significance according to sex or that LBBB diagnosis is more often false-positive in men compared with women.
Cardiac resynchronization therapy defibrillators (CRT-D) reduce mortality and heart failure (HF) hospitalizations in patients with reduced left ventricular (LV) ejection fraction and prolonged QRS duration (QRSD) (1–5). However, recent analysis of clinical trials enrolling New York Heart Association class I and II patients found that clinical benefit was greatest in (and in some cases limited to) patients with left bundle branch block (LBBB) (6–8). In addition, analysis of Medicare patients receiving CRT-D in the American College of Cardiology Implantable Cardioverter-Defibrillator Registry demonstrated that LBBB was a strong predictor of both freedom from HF hospitalization and overall survival (9). These findings have prompted interest in developing a better understanding of how to both diagnose LBBB and integrate it into current indications for cardiac resynchronization therapy (CRT) (10,11).
Previous studies have demonstrated that women have better long-term survival after CRT than men (12,13). However, subgroup analysis of CRT-D trials to investigate the source of this sex disparity has been limited because women represented only 24% of enrolled patients (14). This underrepresentation may cause sex differences in the prognostic value of clinical variables to be overshadowed. A recent meta-analysis found that clinical benefit from CRT-D was limited to patients with QRSD ≥150 ms (14); however, this finding did not take into account QRS morphology, and QRSD ≥150 ms may thus be an indirect marker of LBBB, which has been shown to be a better predictor of CRT outcomes than QRSD (11,15). Recent studies have suggested that women with LBBB have shorter QRSD than men because women have smaller ventricles and shorter QRSD in the absence of LBBB (10,16). Thus, limiting CRT-D to patients with QRSD ≥150 ms may deny CRT-D to women with complete LBBB who are likely to benefit from this device.
Endocardial mapping studies have found that one-third of patients diagnosed with LBBB by using conventional electrocardiographic criteria do not have endocardial activation consistent with LBBB (17,18). Separate analysis suggests that QRSD thresholds for LBBB diagnosis differ according to sex and that conventional LBBB electrocardiographic criteria are falsely positive in men more frequently than in women because of their smaller ventricles and shorter QRSD than men (10,16). In the current study, we tested the hypothesis that in CRT-D patients, conventional diagnosis of LBBB would be associated with better long-term survival in women than in men even after accounting for baseline comorbidities.
This study was approved by the U.S. Food and Drug Administration Research in Human Subjects Committee and the Centers for Medicare & Medicaid Services. It included all Medicare patients (107,475 male and 37,167 female subjects) who received CRT-D (International Classification of Diseases-Ninth Revision-Clinical Modification [ICD-9-CM] procedure code 0051) between July 1, 2002, and December 31, 2008, who were also continuously enrolled in Medicare Part A (inpatient hospital coverage) and B (outpatient medical coverage) for ≥6 months before CRT-D implantation.
We compared the prevalence and prognostic significance of demographic data and several comorbidities as documented in Medicare claims files by using ICD-9-CM codes. These variables include age, reason for entrance into Medicare, race, year of device implantation, preexisting comorbidities (previous myocardial infarction, hypertension, LBBB, right bundle branch block [RBBB], ischemic cardiomyopathy, diabetes mellitus, atrial fibrillation/flutter, previous stroke, previous HF hospitalizations, and end-stage renal disease). To account for risk associated with unmeasured comorbidities, we also included the Charlson comorbidity index or “Charlson score” in all models (19). The Charlson score is an index used to predict 10-year mortality based on whether a patient has certain health conditions; the specific conditions are listed in the Online Appendix. In the Cox proportional hazards models, race was classified as black or non-black. Year of device implantation was defined as the year the procedure code 0051 was recorded in the patient’s record. Preexisting comorbidities were assessed by determining if a beneficiary had an ICD-9-CM diagnosis code in the 6 months before CRT-D implantation; the Online Appendix lists the specific ICD-9-CM codes.
The primary outcome for this study was all-cause mortality. Mortality was determined from the Medicare Master Beneficiary Summary File from the Centers for Medicare & Medicaid Services, which documents date of death for beneficiaries assessed from the Social Security Administration. Secondary analyses were performed for the endpoint all-cause mortality or in-patient HF hospitalization as a primary diagnosis (ICD-9-CM code 428.x). Patients were censored if they did not reach the primary (or secondary) endpoint before December 31, 2009, or if they were no longer continuously enrolled in Medicare Part B.
Kaplan-Meier curves stratified for sex and LBBB were generated for total survival for up to 72 months of follow-up. The significance of demographic and comorbidity characteristics were assessed in univariate and multivariable Cox proportional hazards models for the total population as well as for men and women independently. The proportional hazards assumption was verified by using plots of the log (-log) survival cures and by using Cox test for continuous time interaction. Multivariable models included age, sex, race, year of device implantation, and all preexisting comorbidities, including the Charlson score. The interactions of sex and each of the comorbidities (including LBBB) were also evaluated to determine if the prognostic value of these comorbidities differed according to sex.
All analyses were conducted by using SAS version 9.2 (SAS Institute, Inc., Cary, North Carolina).
Of the 144,642 Medicare CRT-D patients included in this study, 107,475 (74%) were male and 37,167 (26%) were female (Table 1). Men (compared with women) were more commonly white (90% vs. 84%) and more frequently had ischemic cardiomyopathy (69% vs. 53%) and atrial fibrillation/flutter (56% vs. 48%). Women were more frequently black (13% vs. 7%) and more frequently had LBBB (53% vs. 39%). Other comorbidities (including Charlson score), age, reason for entering Medicare, region of residence, and year of CRT-D implantation were similar across sexes. Follow-up data were available for a median of 28 months (interquartile range: 15 to 46 months) with 5,852 patients remaining at 72 months and up to 90 months of follow-up for some patients.
After 72 months of follow-up, 57,043 patients (60% of the uncensored population) had died. Women had significantly lower mortality than men (54% vs. 62%) (Fig. 1A), and the separation of these survival curves continued to diverge over the length of follow-up. Patients with LBBB also had lower mortality than those without LBBB (56% vs. 63%) (Fig. 1B). When stratifying according to both sex and LBBB status (Fig. 1C), women with LBBB had the lowest mortality (49%), whereas men without LBBB had the highest mortality (64%). Non-LBBB women and LBBB men had intermediate mortalities (59% and 60%, respectively).
Univariate and multivariable models for death
Cox proportional hazards for overall mortality were determined for the total population and for men and women independently. Figure 2 contains forest plots of adjusted hazard ratios (HRs); Online Tables 1 and 2 contain univariate and multivariable HRs. In the overall population, all comorbidities (including ischemic cardiomyopathy and atrial fibrillation/flutter) were associated with higher rates of death, with the exception of LBBB, which was associated with an 18% lower rate of death (adjusted HR: 0.82 [95% confidence interval (CI): 0.81 to 0.84]). Presence of end-stage renal disease had the strongest associated mortality rate (adjusted HR: 2.55 [95% CI: 2.46 to 2.67]). Male sex was associated with higher mortality (adjusted HR: 1.19 [95% CI: 1.17 to 1.22]), as was black race (adjusted HR: 1.18 [95% CI: 1.14 to 1.21]). With the exception of hypertension, similar trends were seen in univariate and multivariable models. Hypertension was associated with a higher rate of death in univariate analysis (HR: 1.14 [95% CI: 1.12 to 1.17]) but a lower rate of death in the multivariable analysis (adjusted HR: 0.94 [95% CI: 0.91 to 0.96]).
When male and female patients were analyzed separately (Figs. 2B and 2C, Online Table 2), all comorbidities except LBBB (and hypertension in multivariable analyses) maintained their association with higher rates of mortality. In the univariate analysis, LBBB was associated with a 31% lower mortality rate in women (HR: 0.69 [95% CI: 0.67 to 0.71]) but only a 16% lower mortality rate in men (HR: 0.84 [95% CI: 0.82 to 0.85]). Controlling for comorbidities, LBBB was still associated with a 26% lower mortality rate in women (adjusted HR: 0.74 [95% CI: 0.71 to 0.77]) compared with a 15% lower mortality rate in men (adjusted HR: 0.85 [95% CI: 0.83 to 0.87]) (Fig. 2D). Evaluating the interaction of sex and LBBB demonstrated that after accounting for other comorbidities, the presence of LBBB was associated with a lower mortality in women than in men (p < 0.0001). (All interaction p values are listed in Online Table 3). In contrast, ischemic cardiomyopathy and atrial fibrillation/flutter were associated with lower rates of death in men compared with women (adjusted HRs: 1.02 vs. 1.10 and 1.25 vs. 1.34, respectively).
Univariate and multivariable models for HF hospitalization or death
Similar patterns were seen in all models for the combined outcome of HF hospitalization or death (Figs. 3 and 4, Online Tables 4 and 5). All comorbidities were associated with increased rates of HF hospitalization or death with the exception of LBBB, which was associated with an 18% lower rate of event (adjusted HR: 0.82 [95% CI: 0.81 to 0.83]) (Fig. 4A). As with mortality alone, the reduced rate of HF hospitalization or death associated with LBBB was more substantial in women (31% lower event rate) compared with men (16% lower event rate) (Online Table 5). This difference in the prognostic significance of conventional LBBB diagnosis according to sex remained significant after controlling for baseline characteristics (26% lower event rate in women compared with 15% in men) (Fig. 4, Online Table 5). Gender interaction p values for HF hospitalization or death are listed in Online Table 6.
The diagnosis of LBBB is associated with significantly greater survival in women than in men receiving CRT-D, and this discrepancy is not explained by differences in comorbidities. Men more frequently had ischemic cardiomyopathy and atrial fibrillation/flutter, which have been associated with worse CRT-D outcomes (20,21). Although differences in baseline risk profiles may contribute to the differences in CRT-D outcomes according to sex, multivariable analysis demonstrated that after accounting for baseline comorbidities, a large difference in the association of LBBB with death remained (15% lower mortality in males vs. 26% in females). Furthermore, a significant interaction between sex and LBBB confirmed that the LBBB diagnosis in and of itself carries a different prognosis for female and male CRT-D recipients. These findings suggest that the sex difference in LBBB mortality rates is independent of differences in baseline risk profiles.
LBBB and CRT
When the left bundle branch is blocked, the LV lateral wall is activated significantly later than the septum, which creates dyssynchronous contraction; dyssynchrony can be minimized by CRT to improve cardiac output. Other conditions that prolong QRSD (e.g., RBBB, LV hypertrophy, intramural conduction delay) maintain coordinated LV activation by the rapidly conducting LV Purkinje system. Previous work has suggested that the reduction in mortality and HF hospitalizations associated with CRT may be limited to patients with LBBB (6,7,22). Retrospective analysis of the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial–Cardiac Resynchronization Therapy) study found that LBBB patients receiving CRT-D had a 53% reduction in death or HF events, whereas patients with RBBB or nonspecific LV conduction delay receiving CRT-D had a 24% increase in HF events or death (although these results were not statistically significant) (6). Furthermore, the outcome was even worse in patients with nonspecific LV conduction delay than RBBB. Newly developed, stricter diagnostic criteria for LBBB suggest that many patients who receive a diagnosis for LBBB according to conventional electrocardiographic criteria (defined in this study by using the ICD-9-CM code) and do not meet the strict LBBB criteria may belong to the nonspecific LV conduction delay group (10). One study comparing CRT outcomes between patients who met strict LBBB criteria versus those who only met conventional LBBB criteria found that those meeting the strict criteria had better echocardiographic response and higher event-free survival, which was independent of QRSD (23).
Although it is possible that LBBB truly has different prognostic significance according to sex, recent studies have suggested that current diagnostic criteria for LBBB misdiagnose up to one-third of patients and that this misclassification occurs more frequently in men than in women (10,17,18,24). Women have shorter normal QRSD than men (87.1 ± 8.7 ms vs. 92.7 ± 9.3 ms, respectively) (16) in normal conduction; thus, recommendations for sex-specific LBBB criteria use a QRSD threshold of 130 ms for women and 140 ms for men (10,24). These thresholds suggest that many patients diagnosed with LBBB by using conventional criteria may have false-positive findings and that the potential misclassified population is larger in men than in women (all male patients with QRSD 120 to 140 ms vs. 120 to 130 ms for women). Therefore, it is possible that LBBB predicts better outcomes in women because among those diagnosed with LBBB, women more frequently exhibit truly dyssynchronous activation of the left ventricle and are thus more likely to benefit from resynchronization therapy.
Sex differences in HF and CRT
In both LBBB and non-LBBB patients, women in the current study had a lower rate of mortality than men throughout follow-up. This finding is consistent with previous studies in both the United States and Europe, which demonstrated that women have better long-term survival after CRT than men (25–27). Although the source of this disparity cannot be determined from our study, it is possible that sex differences in true LBBB prevalence is a major factor driving the outcome disparity in men and women with LBBB. For non-LBBB patients, other studies have demonstrated that women also more frequently have nonischemic cardiomyopathies and smaller myocardial scar sizes than men, both of which are associated with better outcomes after CRT (12,13,25–28).
The current study relied on assessing baseline characteristics and outcomes as documented by Medicare billing data and ICD-9-CM codes. It is possible that coding or entry errors may have resulted in misdiagnosis of baseline characteristics, including LBBB, previous HF hospitalization, and CRT-D implantation. In addition, some diagnostic codes are broad (e.g., 427.3x refers to atrial fibrillation or atrial flutter) and may result in overestimation or underestimation of some comorbidities. However, this possibility is equally likely in men and women, and the large sample size makes these results more robust and less prone to errors due to miscoding. In addition, previous work has demonstrated that adjudicating Medicare claims resulted in changes in <3% of claims, suggesting that the data entry errors are minimal (29). The sex differences demonstrated in this study may be the result of indirect effects from unmeasured variables. All models were designed to be robust and include all available comorbidities likely related to the outcomes. In addition, the study would be strengthened by including analysis with QRSD, New York Heart Association class, and LV ejection fraction; however, these variables are not available in Medicare billing records.
The results of this study demonstrated that among Medicare beneficiaries undergoing implantation with CRT-D, LBBB predicts better outcomes among women than men. One possible explanation for this sex disparity is that men may have more false-positive LBBB diagnoses than women. To our knowledge, this is the first time that LBBB has been shown to portend significantly better long-term survival benefit in women than men receiving CRT-D. Developing more patient-specific selection criteria for CRT-D may reduce the risks and costs associated with inappropriate therapy. Future studies should investigate appropriate QRSD thresholds in men and women that best identify CRT-D candidates.
This project was supported by the Centers for Medicare & Medicaid Services/U.S. Food and Drug Administration (FDA) SafeRx Project and the FDA Office of Women's Health. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- cardiac resynchronization therapy
- cardiac resynchronization therapy defibrillators
- heart failure
- hazard ratio
- International Classification of Diseases-Ninth Revision-Clinical Modification
- left bundle branch block
- left ventricular
- right bundle branch block
- QRS duration
- Received December 17, 2012.
- Revision received March 4, 2013.
- Accepted March 6, 2013.
- American College of Cardiology Foundation
- Zareba W.,
- Klein H.,
- Cygankiewicz I.,
- et al.
- Gold M.R.,
- Thebault C.,
- Linde C.,
- et al.
- Bilchick K.C.,
- Kamath S.,
- Dimarco J.P.,
- Stukenborg G.J.
- Dupont M.,
- Rickard J.,
- Baranowski B.,
- et al.
- Macfarlane P.,
- Oosterom A.V.,
- Pahlm O.,
- Kligfield P.,
- Janse M.,
- Camm J.
- Vassallo J.A.,
- Cassidy D.M.,
- Marchlinski F.E.,
- et al.
- Auricchio A.,
- Fantoni C.,
- Regoli F.,
- et al.
- Goldenberg I.,
- Moss A.J.,
- Hall W.J.,
- et al.
- Boston Scientific Corporation. Multicenter Automatic Defibrillator Implantation Trial With Cardiac Resynchronization Therapy (MADIT-CRT): sponsor's executive summary, P010012/S230. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/MedicalDevices/MedicalDevicesAdvisoryCommittee/CirculatorySystemDevicesPanel/UCM204607.pdf. Accessed May 30, 2012.
- Bleeker G.B.,
- Kaandorp T.A.,
- Lamb H.J.,
- et al.
- White J.A.,
- Yee R.,
- Yuan X.,
- et al.