Different Patterns of Bundle Branch Blocks and the Risk of Incident Heart Failure in the Women’s Health Initiative (WHI) Study

Abstract

Background

We evaluated the risk of incident heart failure (HF) associated with bundle branch blocks (BBB) in post-menopausal women.

Methods and Results

Cox’s regression was used to evaluate hazard ratios (HR) with 95% confidence intervals (CI) for HF among 65,975 participants of the Women’s Health Initiative (WHI) study during an average follow-up of 14 years. BBB observed in 1,676 women at baseline was categorized into left, right and indetermined-type BBB (LBBB, RBBB and IVCD, respectively). Compared to women with no BBB, LBBB and IVCD were strong predictors of incident HF in multivariable-adjusted risk models (HR 3.79, CI 2.95–4.87 for LBBB and HR 3.53, CI 2.14–5.81 for IVCD). RBBB was not a significant predictor of incident HF in multivariable-adjusted risk model but the combination of RBBB and left anterior fascicular bock (RBBB&LAFB) was a strong predictor (HR 2.96, CI 1.77–4.93). QRS duration was an independent predictor of incident HF only in LBBB, with more pronounced risk at QRS ≥140 ms than below 140 ms. RNDPV was an independent predictor in both RBBB and LBBB and in addition in LBBB, QRS/STT angle and ST J-point depression in aVL were independent predictors.

Conclusions

LBBB, IVCD and RBBB combined with LAFB are strong predictors of incident HF in multivariable-adjusted risk models but RBBB is not a significant predictor, QRS duration ≥140 ms may warrant consideration in LBBB as an indication for further diagnostic evaluation for possible therapeutic and preventive action.

Heart failure (HF) as a consequence of ischemic, hypertensive or valvular heart disease is known to be more common in bundle branch blocks (BBB) than in normal ventricular conduction. Left bundle branch block (LBBB) has been associated with excess risk of incident HF in some studies,^1–4 but right bundle branch block (RBBB) has not been found to predict incident HF.^5–7 Overall, available data are limited about specific BBB categories as predictors of incident HF, particularly in women. Therefore, the aim of the present study was to evaluate independent prognostic significance of BBB in predicting incident HF, and further to evaluate whether repolarization abnormalities associated with BBB contain additional prognostic information.

Methods

Study population and design

The Women’s Health Initiative (WHI) study is a multi-center investigation of risk factors for the prevention of common causes of mortality, morbidity, and impaired quality of life in United States- postmenopausal women. Detailed eligibility criteria and recruitment methods, randomization, follow-up, data and safety monitoring, and quality assurance have been published previously.^8–10 The study was approved by each study site’s institutional review board. All participants provided written informed consent. Of the 68,133 WHI clinical trial participants who were enrolled from 1993 to 1998, we excluded 2,158 participants from analyses: 960 with no electrocardiogram (ECG) data, 614 with inadequate quality ECG, 109 with an external pacemaker or Wolff-Parkinson-White pattern, and 475 with prevalent HF (self-report) at baseline. In the remaining group of 65,975 participants, 1,676 had BBB and 64,299 had no BBB (Figure).

Figure.

A Schematic of the decision tree with sequential steps in selection of various subgroups of study population. CVD denotes cardiovascular disease, IVCD= indeterminate type of intraventricular conduction defect, LAFB= left anterior fascicular block, LBBB= left bundle branch block and RBBB= right bundle branch block.

Outcome ascertainment

The outcome considered in the present investigation was incident HF that occurred from baseline through September 24, 2010. The follow-up period was up to 17 years (mean 14.2 years). Cardiovascular disease (CVD) at baseline was defined by myocardial infarction (MI) according to the Minnesota Code (MC)¹¹ or the Novacode¹² criteria, or a history or clinical diagnosis of MI, angina pectoris, coronary artery bypass surgery, coronary angioplasty, congestive heart failure, or stroke at the time women entered the WHI study.

Incident HF during the follow-up was defined as hospitalization HF as diagnosed by a physician and the patient receiving medical treatment for HF, with corroboration of pertinent abnormalities in diagnostic testing by noncentralized review of medical records.⁹ Records were obtained and adjudicated every 6 months for self-reported HF hospitalizations, as well as possible HF events found during adjudication of other cardiovascular outcomes (e.g. MI). Three cardiologists reviewed all records independently using the WHI diagnostic criteria^13–15 to identify HF cases. Surveillance for HF did not include ascertainment or confirmation of non-hospitalized cases. The WHI criteria required a physician diagnosis of HF and medical treatment for HF during the index admission, with or without an imaging procedure showing impaired systolic or diastolic left ventricular function. The classification required a specific combination of major and minor criteria. Disagreements were resolved by consensus decision.

ECG methods

Identical electrocardiographs (MAC PC, Marquette Electronics Inc., Milwaukee, Wisconsin) were used at all clinic sites, and resting, 10-second standard simultaneous 12-lead ECGs were recorded in all participants using strictly standardized procedures. All ECGs were processed in a central ECG laboratory (initially at Dalhousie University, Halifax, NS, Canada and later at the EPICARE Center, Wake Forest School of Medicine, Winston-Salem, NC), where all ECGs were visually inspected for technical errors and inadequate quality using an interactive computer graphics terminal. The ECGs were first processed by the Dalhousie ECG program and were reprocessed for the present study using the 2001 version of the GE Marquette 12-SL program (GE, Milwaukee, Wisconsin). BBB were classified according to the MC criteria,¹¹ for complete LBBB (MC-7.1), complete RBBB (MC-7.2, QRS axis >−45 degree), intraventricular conduction defect (IVCD) (MC-7.4), and combination of RBBB and left anterior fascicular block (LAFB) (MC-7.8, RBBB and QRS axis between −45° and −120°).^11,16–17

Global interval measurements were obtained from the quasi-orthogonal XYZ leads computed from the standard 12-lead ECGs using the Horáĉek transform.¹⁸ Spatial QRS|T angle (Ѳ(QRS|STT)) was derived as the angle between the mean QRS and STT vectors. Nondipolar QRS voltage (RNDPV) was computed from singular-value decomposition of QRS as the RMS value of the pooled variance of components 4 to 8.^16,19 ECG left ventricular hypertrophy (ECG-LVH) was defined by Cornell voltage (RaVL+SV3).¹² Rate-adjusted QT interval (QTrr) was computed as a linear function of the RR interval with separate formulas derived for women no BBB and for women with BBB (listed in the footnote of Table 1).

Table 1.

Characteristic of Study Participants with incident Heart Failure in Women with Bundle Branch Block and without Bundle Branch Block at Baseline

Total N=65,975 Mean (SD), or n (%)	No Bundle Branch Block (QRS duration <120ms) N=64,299	Bundle Branch Block (QRS duration ≥120ms) N=1,676	P-value*
Age (years)	63 (7.0)	66 (6.9)	<0.001
Body Mass Index (kg/m2)	29 (5.9)	29 (5.8)	0.037
Systolic blood pressure (mm Hg)	128 (17.3)	134 (18.6)	<0.001
Race/ethnicity			<0.001
Non-Hispanic white	52461 (82%)	1427 (85%)
African-American	6506 (10%)	150 (9%)
Current smoker	5068 (8%)	61 (7%)	0.286
Hypertension	21456 (34%)	762 (46%)	<0.001
Diabetes	3916 (6%)	150 (9%)	<0.001
Hypercholesterolemia†	7227 (13%)	226 (15%)	0.009
Family history of CHD	10769 (19%)	310 (21%)	0.053
History of CVD	11689 (18%)	646 (39%)	<0.001
Atrial fibrillation (self-report)	2353 (4%)	123 (8%)	<0.001
ECG characteristics
Atrial fibrillation#	193 (0.3%)	11 (0.7%)	0.010
Ectopic complexes#	2068 (3%)	87 (5%)	<0.001
Left atrial enlargement#	3128 (5%)	132 (8%)	<0.001
ECG-LVH§	4747 (7%)	-	-
Heart rate (bmp)	66 (10.1)	67 (9.9)	<0.001
QRS duration (ms)	86 (8.5)	138 (11.3)	<0.001
QTrr†† (ms)	416 (16.6)	448 (19.3)	<0.001
Ѳ(QRS|STT)§§(°)	46 (25.1)	109 (36.7)	<0.001
RNDVPII (µV)	42 (16.3)	83 (30.8)	<0.001
STJ-point amplitude (µV)
Lead aVR	−17 (21.0)	0.5 (26.3)	<0.001
Lead aVL	8 (24.4)	−5 (26.0)	<0.001
Lead V1	−1 (26.3)	63 (53.8)	<0.001
Lead V6	14 (23.7)	−5 (29.7)	<0.001
T-wave amplitude (µV)
Lead aVR	−223 (90.2)	−165(149.2)	<0.001
Lead aVL	83 (89.7)	28 (181.3)	<0.001
Lead V1	−12 (116.0)	63 (332.8)	<0.001
Lead V6	219 (121.5)	131 (216.6)	<0.001
Outcomes
CHD death	1441 (2%)	115 (7%)	<0.001
All-cause mortality	6731 (10%)	286 (17%)	<0.001
Incident heart failure	1433 (2%)	152 (9%)	<0.001

CHD=coronary heart disease; CVD=cardiovascular disease;

^*P-value for Student's t-test for difference between mean values or for z-test for proportions in BBB and No-BBB groups;

^†Hypercholesterolemia identified by self-reported use of cholesterol-lowering drugs.

^#Atrial fibrillation, ectopic complexes, and Left atrial enlargement by MC classification.

^§ECG-LVH= Left ventricular hypertrophy by Cornell Voltage (RaVL + SV3) ≥ 2200 µV for no BBB group.

^††QT_rr= QT adjusted for rate as a linear function of RR interval with the formula as QT_rr=QT−157*(60/heart rate −1) − 0.86*(QRS duration −138)for BBB group and QT_rr=QT −185*(60/heart rate −1) for no BBB group.

^§§Ѳ(QRS|STT)= Spatial angle between the mean QRS and ST-T vectors.

^IIRNDVP= QRS nondipolar voltage from singular value decomposition.

Because our previous data revealed that some additional repolarization abnormalities were associated with mortality risk among women with BBB,²⁰ we also examined their possible added risk for HF.

Statistical methods

Frequency distributions of ECG variables were first inspected to rule out anomalies and outliers possibly due to measurement artifacts. Descriptive statistics were used to determine mean values, standard deviations, and percentile distributions for continuous variables, and frequencies and percentages for categorical variables. Cox's proportional hazards regression was used to assess the associations of ECG variables with the risk of HF for unadjusted single-ECG variable models and for multivariable-adjusted ECG variable models with an adjustment for demographic and clinical factors (age, ethnicity, body mass index, smoking status, hypertension, diabetes mellitus, history of CVD, hypercholesterolemia, family history of coronary heart disease (CHD), systolic blood pressure, heart rate, and study component/arm in hormone therapy /dietary modification /calcium and vitamin D trial participants). In addition to single ECG variable models, independent ECG predictors for incident HF were identified after considering collinearity between ECG variables. In the group with no BBB, several variables were highly correlated (r >0.5). It was observed that QRS/T angle was functionally a dominant determinant for T amplitudes and it was highly correlated with T wave amplitudes in leads of aVR, V1 and V6. These variables were removed and the remaining ECG variables were entered simultaneously into multiple ECG variable models first unadjusted for non-ECG variables and then multivariable-adjusted for demographic and clinical factors. In BBB groups, few variables were significant predictors in the unadjusted risk models and their correlations were low.

Predictor ECG parameters were first evaluated as continuous variables and then dichotomized for HF risk evaluation, with the upper quintile as the test group and the remaining four quintiles as the reference group. This was done in order to have reasonable sized (20%) test groups. Hazard ratios (HR) were evaluated for increased values of ECG predictor variables (quintile 5) with quintiles 1–4 as the reference group (HR =1). However, quintile 1 (decreased or more negative values) was used as the test group for T wave amplitude in aVL and ST J-point depression in aVL, with quintiles 2–5 as the reference group.

All analyses were performed with SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).

Results

Study group characteristics

Detail of demographic, clinical, and ECG characteristics of the study population are summarized in Table 1. The mean age for the participants was 63 years (SD 7.0), 82% were non-Hispanic white, 10% African American, 34% had hypertension, 6.2% diabetes, and 19% had a history of CVD or ECG evidence of MI. BBB was present in 2.5% (1,676/65,975) including 740 with RBBB, and 139 with RBBB combined with LAFB, 680 with LBBB, 117 with IVCD, and total 936 with the combination LBBB/RBBB&LAFB/IVCD. With the exception of current smoking and family history of CHD, all other differences between women with/without BBB in demographic, clinical and ECG variables in Table 1 were statistically significant (P<0.05 or less). During an average 14 years of follow-up, 9.1% of participants with BBB had incident HF and 2.2% in the no-BBB group.

Bundle branch blocks and incident heart failure

The group with no BBB was used as the reference group in evaluating the risk of incident HF by BBB category (Table 2). In unadjusted risk model-1, RBBB was a significant predictor of incident HF (hazard ratios (HR) 1.78 with 95% confidence intervals (CI) 1.25–2.53). Risk levels were substantially higher for other BBB categories: with over 5-fold increased risk for LBBB and for RBBB combined with LAFB, and approaching a 6-fold increased risk for IVCD. In fully adjusted multivariable risk model-2, RBBB was no longer a significant predictor of incident HF but the other BBB categories remained strong predictors with a 3-fold or higher increased risk. The risk of incident HF for the pooled LBBB group was comparable to LBBB alone. Evaluating interaction between QRS duration as a continuous variable and BBB categories produced P=0.047 for unadjusted risk model and P=0.037 for multivariable adjusted model, suggesting that QRS duration may in part account for the differences in incident HF risk between BBB categories.

Table 2.

Hazard Ratios with 95% Confidence Intervals for Incident Heart Failure by Bundle Branch Block Category during average 14 years Follow-up

	Event Rate n/N	Events/1000 person years (95% CI)	Model-1 Unadjusted HR (95% CI)	Model-2 Multivariable* HR (95% CI)
(A) All bundle branch blocks (1,585/65,975)†
No BBB	1433/64299	1.63 (1.49,1.78)	1.00 (reference)	1.00 (reference)
RBBB Isolated	32/740	3.33 (2.25,4.36)	1.78 (1.25–2.53)	1.03 (0.68–1.58)
RBBB&LAFB	18/139	10.9 (6.11,15.7)	5.39 (3.38–8.58)	2.96 (1.77–4.93)
IVCD	18/117	12.8 (7.44,18.2)	6.80 (4.27–10.8)	3.53 (2.14–5.81)
LBBB	84/680	10.3 (8.19,12.4)	5.41 (4.34–6.74)	3.79 (2.95–4.87)
(B) Pooled LBBB/RBBB&LAFB/IVCD§ (1,553/65,235)
No BBB	1433/64299	1.63 (1.49,1.78)	1.00 (reference)	1.00 (reference)
Pooled-LBBB	120/936	10.7 (8.87,12.5)	5.58 (4.62–6.73)	3.59 (2.91–4.43)

^*Model-2 adjusted for age, ethnicity, body mass index, smoking status, hypertension, diabetes mellitus, CVD status at baseline, hypercholesterolemia, family history of CHD, systolic blood pressure, heart rate, and study component/arm groups (hormone therapy /dietary modification /calcium and vitamin D).

^†P value for unadjusted risk model-1 for interaction between QRS duration as a continuous variable and BBB categories (BBB*QRSd) P=0.047 and for multivariable adjusted model-2 P=0.037.

^§Pooled LBBB group combining LBBB, RBBB & LAFB and IVCD.

In view of the high risk for RBBB-LAFB combination, we also considered RBBB combined with left posterior fascicular block (LPFB). However, there were only 22 women with this combination, and although none of these 22 developed a new-onset HF during the follow-up, the sample size is too low to make any conclusions.

To further evaluate potential influence of QRS duration differences between BBB categories, QRS duration was entered as a categorical variable dichotomized at 140 ms and risk levels for RBBB and LBBB were compared (Table 3). Focusing on the multivariable adjusted risk model, risk levels for RBBB were not significantly higher than for No-BBB group. Risk levels for LBBB were higher at stratification level 120 to 139 ms (HR 2.99, CI 1.93–4.62) and particularly at level ≥140 ms than for No-BBB group (HR 4.32, CI 3.21–5.83). P values for the interaction term between QRS duration group and the two BBB categories were P=0.005 for the unadjusted model and P=0.038 for the fully adjusted multivariable model, indicating again that QRS duration accounts in part for the differences in risk levels between RBBB and LBBB.

Table 3.

Hazard Ratios with 95% Confidence Intervals for Incident Heart Failure for RBBB and LBBB with QRS Duration Categorized at 140ms Threshold using Women with No BBB as the Reference Group

BBB Category*	Event Rate (n/N)	Events/1000 person years (95% CI)	Model-1 Unadjusted HR (95% CI)	Model-2 Multivariable† HR (95% CI)
QRS duration<120ms	1433/64299	1.63 (1.49,1.78)	1.00 (reference)	1.00 (reference)
QRS duration 120–139ms	47/800	4.62 (3.54,5.69)	2.41 (1.80–3.23)	1.57 (1.11-2.21)
RBBB	19/489	3.00 (2.19,3.81)	1.58 (1.00–2.48)	0.89 (0.51–1.53)
LBBB	28/311	7.27 (4.18,10.4)	3.77 (2.59–5.49)	2.99 (1.93–4.62)
QRS duration ≥140ms	69/620	9.06 (6.36,11.8)	4.89 (3.84–6.23)	3.18 (2.42–4.18)
RBBB	13/251	3.95 (2.52,5.39)	2.18 (1.26–3.76)	1.36 (0.70–2.62)
LBBB	56/369	12.9 (7.44,18.5)	6.87 (5.26–8.98)	4.32 (3.21–5.83)

^*For interaction between categorized QRS duration and BBB group overall P=0.005 for unadjusted model and P=0.038 for multivariable adjusted model.

^†Adjusted for demographic and clinical factors listed in the footnote of Table 2.

Repolarization abnormalities as independent predictors of the risk of incident heart failure

Hazard ratios for incident HF for repolarization-related ECG abnormalities are listed in Table 4 for LBBB, RBBB, and No-BBB groups. Independent predictors of incident HF are identified by Model-2 and the fully adjusted Model-3. Focusing on the fully adjusted risk Model-3, QRS duration was an independent predictor of incident HR only for LBBB. Apart from QRS duration, additional independent predictors in LBBB were heart rate, Ѳ(QRS|STT), RNDPV and ST J-point depression in aVL, with increased risk approximately 2-fold for heart rate, Ѳ(QRS|STT) and RNDPV, and 2.5 for ST J-point depression exceeding 25 µV in aVL. RNDPV was a strong independent predictor of incident HF in RBBB category, with an over 3-fold increased risk. Apart from RNDPV, neither QRS duration nor any other repolarization was significant independent predictors of incident HF in women with isolated RBBB.

Table 4.

Hazard Ratios with 95% Confidence Intervals for Significant Independent ECG Predictors of Incident Heart Failure in Women with LBBB, RBBB and No BBB

	Reference Quintiles	Test Quintile Threshold	Model-1*	Model-2*	Model-3*
LBBB (N=680)
Heart rate	Q1–4	Q5=(>75)	1.94 (1.22–3.09)	1.84 (1.13–2.98)	2.01 (1.09–3.70)
QRS dur. (ms)	120–139	≥140	1.77 (1.12–2.79)	1.59 (1.00–2.58)	1.87 (1.01–3.46)
RNDVP(µV)	Q1–4	Q5=(≥107)	1.83 (1.15–2.93)	1.64 (1.00–2.70)	1.89 (1.05–3.41)
STJ aVL (µV)	Q2–5	Q1=(≤−25)	2.60 (1.67–4.06)	1.85 (1.10–3.12)	2.50 (1.35–4.61)
Ѳ(QRS|STT)† (°)	Q1–4	Q5=(≥153)	2.01 (1.26–3.19)	1.41 (0.80–2.47)	1.97 (1.05–3.70)
RBBB (N=740)
RNDVP(µV)	Q1–4	Q5=(≥112)	2.53 (1.24–5.19)	2.81 (1.16–5.34)	3.31 (1.09–10.0)
QRS dur. (ms)	120–139	≥140	1.34 (0.66–2.72)	1.14 (0.55–2.37)	1.00 (0.37–2.74)
No-Bundle Branch Block at Baseline (N=64,299)
Heart rate	Q1–4	Q5=(≥75)	1.78 (1.58–1.99)	1.52 (1.35–1.70)	1.26 (1.10–1.44)
QRS dur. (ms)	Q1–4	Q5=(≥95)	1.64 (1.46–1.84)	1.29 (1.14–1.46)	1.12 (0.97–1.28)
QTrr# (ms)	Q1–4	Q5=(>427)	1.80 (1.61–2.01)	1.36 (1.21–1.53)	1.29 (1.14–1.47)
Ѳ(QRS|STT)† (°)	Q1–4	Q5=(≥63)	2.79 (2.51–3.10)	1.52 (1.33–1.73)	1.36 (1.18–1.58)
RNDVP(µV)	Q1–4	Q5=(≥53)	1.46 (1.29–1.64)	1.28 (1.13–1.44)	1.27 (1.11–1.46)
STJ aVR (µV)	Q1–4	Q5=(>0)	2.02 (1.80–2.26)	1.35 (1.17–1.57)	1.40 (1.18–1.65)
TaVL (µV)	Q2–5	Q1=(≤0)	2.17 (1.95–2.42)	1.22 (1.08–1.38)	1.23 (1.06–1.41)

^*Model-1: Unadjusted single ECG variable risk model (included heart rate, QRS duration, QTrr, Ѳ(QRS|STT), RNDVP, ST amplitude at J-point in leads of aVR, aVL, V1 and V6, and T wave amplitude in the leads of aVR, aVL, V1 and V6, separately); Model-2:Significant predictors in Model-1 entered simultaneously and each adjusted for the other ECG covariates; Model-3:ECG variables in Model-2 entered simultaneously and adjusted for demographic and clinical factors listed in the footnote of Table 2.

^†Ѳ(QRS|STT)= spatial angle between the mean QRS and ST-T vectors.

^#QT_rr = rate adjusted QT using formulas listed in the footnote of Table 1.

Several repolarization abnormalities were independent predictors of incident HF in the group of women with no BBB. These included heart rate, rate-adjusted QT, Ѳ(QRS|STT) and RNDPV, each with an approximately 1.3-fold increased risk of incident HF.

Other ECG and non-ECG factors as predictors of incident heart failure

Several non-ECG factors were associated independently with the risk of incident HF (Table 5). As expected, CVD at baseline and diabetes were strong predictors for incident HF in women with and without BBB, and also hypertension was a significant predictor. The variables with independent risk for incident HF in addition to BBB included atrial fibrillation, with a more than 3-fold increased risk of HF in both BBB and no-BBB groups. Ischemic ST-T and any major ECG findings according to MC classification were associated with substantially increased risk of incident HF in women with no BBB. In addition, the presence of ectopic atrial and ventricular complexes at baseline increased the risk of HF by 69% in women with no BBB and an increase of 167% in women with BBB group.

Table 5.

Hazard Ratios with 95% Confidence intervals for Incident Heart Failure for Demographic and Clinical Risk Factors in Women with/without BBB at Baseline

	No-BBB Group (N=64,299)		BBB Group* (N=936)
	Unadjusted	Multivariable Adjusted†	Unadjusted	Multivariable Adjusted†
Age, by 5 years	1.21 (1.19–1.23)	1.19 (1.17–1.21)	1.21 (1.13–1.28)	1.20 (1.12–1.29)
Body Mass Index, by 10 kg/m2	1.95 (1.82–2.09)	1.61 (1.48–1.76)	1.26 (0.93–1.70)	1.27 (0.87–1.84)
Hypertension (yes vs. no)	2.72 (2.44–3.02)	1.37 (1.20–1.57)	2.03 (1.39–2.95)	1.67 (1.06–2.63)
Diabetes (yes vs. no)	5.04 (4.46–5.69)	2.82 (2.44–3.26)	2.32 (1.26–4.28)	2.38 (1.15–4.93)
Self-report AF§(yes vs. no)	2.92 (2.46–3.46)	1.37 (1.11–1.70)	2.59 (1.63–4.12)	1.98 (1.11–3.53)
CVD at baseline (yes vs.no)	3.29 (2.96–3.65)	2.29 (2.00–2.61)	1.97 (1.37–2.83)	1.94 (1.25–3.00)
ECG AF at baseline (yes vs. no)	8.66 (6.24–12.0)	3.70 (2.50–5.48)	3.15 (0.77–12.8)	6.88 (1.44–32.8)
Ectopic complexes (yes vs. no)	2.34 (1.92–2.85)	1.69 (1.34–2.13)	2.50 (1.45–4.30)	2.67(1.45–4.92)
ECG-LVH & major STT (yes vs.no)	7.09 (5.65–8.89)	2.55 (1.95–3.34)
Major ST-T abnormality(yes vs.no)	3.39 (2.95–3.89)	1.82 (1.55–2.14)
Major ECG abnormality (yes vs. no)	5.33 (4.62–6.16)	2.34 (1.96–2.79)

^*BBB Group contains combined BBB categories excluding RBBB.

^†Unadjusted single variable risk model; Multivariable adjusted for demographic and clinical factors listed in the footnote of Table 2.

^§Other the symbols for abbreviations as in the footnote of Table 1.

Discussion

The key findings of our study were the following: 1) LBBB, IVCD and RBBB combined with LAFB were all strong predictors of incident HF in unadjusted risk model and remained strong predictors in fully adjusted multivariable risk model; 2) RBBB was a significant predictor of incident HF only in unadjusted risk model; 3) In women with RBBB, RNDPV was a strong independent predictor of incident HF. Apart from RNDPV, neither QRS duration nor any repolarization-related ECG variable was a significant independent predictor; 4) In women with LBBB, QRS duration was an independent predictor of incident HF. Apart from QRS duration, additional independent predictors in LBBB were heart rate, Ѳ(QRS|STT), RNDPV and ST J-point depression in aVL.

LBBB, QRS prolongation and increased risk of heart failure

The fact that QRS duration was a significant independent predictor of incident HF in women with LBBB may have potential clinical implications. We had data on time to HF for QRS duration dichotomized at 150ms indicating that in women with LBBB and QRS duration ≥150ms the average time to incident HF was 3.6 years, and in women with RBBB and QRS duration ≥150ms was 4.7 years.

Depolarization sequence at the time of QRS peak is normally directed to inferior-left anterior. In LBBB depolarization proceeds from right-anterior to left-posterolateral direction. Repolarization direction is also altered from the normal spatial orientation from superior-right posterior to inferior-left-anterior.²¹ In LBBB repolarization direction changes from the predominantly reverse normal repolarization sequence and becomes more concordant with depolarization sequence. Delayed activation of particularly of LV posterior and lateral wall leads into dyssynchrony of ventricular contraction. Delayed left ventricular excitation in LBBB leads also into delayed left ventricular repolarization and relaxation. Dyssynchrony of ventricular contraction and relaxation has become an important consideration in resynchronizing therapy for HF patients.^22–24 chronic dyssynchrony of ventricular excitation may be the main mechanism in electrophysiological remodeling in LBBB.

LBBB radiofrequency ablation in dogs has been reported by Spragg et al. to result in significantly reduced conduction velocity and APD in the late-activated, lateral wall of dyssynchronous hearts compared to the anterior wall.²⁵ These authors concluded that already in the absence of LV dysfunction, dyssynchrony induces regionally specific changes in conduction and repolarization.

In an earlier WHI study with 9 years of follow-up, RNDPV in women without BBB was found to be an independent predictor of HF.⁸ Similarly, in the ARIC study RNDPV was a significant independent predictor of incident HF in women without BBB but not in men.²⁶ Our study results confirm RNDPV as a significant independent predictor of incident HF in women without BBB (Table 4). In our study, RNDPV was also a significant predictor for incident HF in women with BBB, including RBBB. One can speculate about possible mechanisms for the observed excess risk for incident HF for RNDPV when added as a covariate with BBB and repolarization-related variables. Nondipolar components from singular-value decomposition of QRS (the RMS value of the pooled variance of components 4 to 8 in the present study) may reflect fractionated QRS and sudden shifts in the orientation of depolarization sequence. It is not clear why RNDPV was an independent predictor of incident HF risk in women with normal ventricular conduction but not in men in the previous study cited.²⁶

Our results in relation to work done by other investigators

LBBB has been associated with excess risk of incident HF in some studies.^2–4 RBBB has not been found to predict incident HF.^5–7 In a report from the Heart Outcomes Prevention Evaluation (HOPE) trial, baseline LBBB was an independent predictor of HF, sudden death, CVD death, and all cause death, but baseline RBBB was not associated with increased CVD risk.^3,7

Dhingra et al in Framingham Heart Study with population of 1,759 participants and 12.7 year follow-up¹ showed that longer QRS duration (≥100ms) was associated with increased HF risk. Also, Aro et al in their study with population of 2,049 men found QRS duration threshold of 110ms as an optimal cut point to define a prolonged QRS duration as a risk factor.²⁷ Our study also showed that women with QRS duration 110–119ms had a near 2-fold risk of future HF using women with QRS duration <110ms for the reference group (not shown in the table).

Potential clinical implications

This large population study in women with specific categories of BBB demonstrates that more pronounced QRS prolongation in LBBB is associated with increasing risk of incident HF and a shorter time to the onset of HF. Thus, QRS duration in patients with LBBB may warrant consideration for additional diagnostic evaluation and possible preventive action.

Study limitations

Our study involved only women. Previous studies were conducted in males only or did not stratify study population by gender. The question of gender differences in the risk associated with BBB remains open. We used only hospitalizations to identify incident HF events. Although this is a limitation, consideration of incident HF hospitalizations, relative to incident outpatient HF, is important because of associated disease severity and cost. WHI population includes healthy, post-menopausal women, and our findings cannot be generalized to men or to younger age groups of women.

Conclusions

LBBB, IVCD and RBBB combined with LAFB are strong predictors of incident HF but RBBB was not a significant predictor in multivariable risk models adjusted for demographic and clinical factors. In women with RBBB, RNDPV is a strong independent predictor of incident HF. In women with LBBB, QRS duration, heart rate, Ѳ(QRS|STT), RNDPV and ST J-point depression in aVL were independent predictors of incident HF. QRS duration exceeding 140 ms in women with LBBB may warrant consideration as an indication for further diagnostic evaluation for possible therapeutic and preventive action.

Short commentary for CIRCHF/2013/000217-R2.

While most studies have found left bundle branch block (LBBB) to be associated with excess risk of incident heart failure (HF), the data are inconsistent about RBBB and indetermined type of ventricular conduction defect (IVCD). We examined the association between different BBB categories with the risk of HF in 65,975 women from the Women's Health Initiative. After adjustment for cardiovascular risk factors and potential confounders, RBBB was not associated with increased risk of incident HF but LBBB, IVCD and RBBB combined with LAFB were strong predictors. This study is the first large population study in women investigating how specific categories of BBB influence incident HF risk. A better understanding of the association the relationship between the type of ventricular conduction defect and incident HF may help to design strategies for improved risk stratification which could ultimately improve outcomes.