OUP user menu

Influence of left bundle branch block on long-term mortality in a population with heart failure

Fariborz Tabrizi, Anders Englund, Mårten Rosenqvist, Lars Wallentin, Ulf Stenestrand
DOI: http://dx.doi.org/10.1093/eurheartj/ehm262 2449-2455 First published online: 1 August 2007

Abstract

Background The purpose of this study was to assess the independent contribution of left bundle branch block (LBBB) on long-term mortality in a large cohort with symptomatic heart failure (HF) requiring hospitalization.

Methods and Results We studied a prospective cohort of 21 685 cases of symptomatic HF requiring hospitalization in the Register of Information and Knowledge about Swedish Heart Intensive care Admissions in 1995–2003. Long-term mortality was evaluated by Logistic regression analysis, adjusted for multiple covariates that could influence long-term prognosis. LBBB was present in 20% (4395 of 21 685) of HF admissions. Patients with LBBB had a higher prevalence of cardiac comorbid conditions than patients with no LBBB. 1-, 5-, and 10-year mortality was 31.5 vs. 28.4%, 69.3 vs. 61.3%, and 90.1 vs. 84.7% for HF patients with and without respectively LBBB. When adjusting for comorbidity, LBBB was associated with increased 5-year mortality (OR, 1.21; 95% CI, 1.10–1.35; P < 0.001). When left ventricular ejection fraction was included in the analysis LBBB had no longer any independent influence on 5-mortality (OR, 0.99; 95% CI, 0.62–1.56; P = 0.953).

Conclusion LBBB occurs in 1/5 in HF patients requiring hospitalization and is associated with a very high mortality. However, the high long-term mortality appears to be caused by cardiac comorbidities and myocardial dysfunction rather than the LBBB per se.

  • Heart failure
  • Bundle branch block
  • Prognosis

Introduction

Left bundle branch block (LBBB) occurs in up to 30% of patients with heart failure (HF).1,2 The detrimental effect of LBBB on left ventricular systolic and diastolic function caused by regional delays of electrical activity leads to dyssynchronous left ventricular function and has been established in HF patients.3,4 In a series of trials, cardiac resynchronization therapy (CRT) improved symptoms, exercise capacity, the quality of life, and ventricular function in this population.58 The CARE-HF study of highly symptomatic HF patients with intraventricular conduction delay and ejection fraction (EF) ≤35% reported a significant reduction of all-cause mortality by CRT compared with conventional HF treatment.9 Despite the deleterious effect of LBBB on left ventricular function and salutary effect of resynchronization, studies of the significance of LBBB on mortality in patients with HF have given conflicting results. While some studies have reported LBBB as an independent predictor of mortality in HF patients,1,2,10 others have suggested that associated structural heart disease is the main cause of the higher mortality rate in this population.11,12

The purpose of the present prospective study was to assess the independent contribution of LBBB compared to no LBBB on long-term mortality in a large population with symptomatic HF requiring hospitalization.

Methods

The Register of Information and Knowledge about Swedish Heart Intensive care Admissions (RIKS-HIA) registers all patients admitted to the coronary care unit of all participating hospitals. Information, including data on 100 variables, is reported on case record forms and has been described elsewhere.13 Briefly, the register includes information on age, sex, smoking status, previous MI, atrial fibrillation, diabetes mellitus, hypertension, hyperlipidemia, previous angina pectoris, previous coronary revascularization, previous medications (‘previous’ means events occurring or medication started before the current admission), symptoms, ECG at entry, biochemical markers, echocardiography, pharmacological treatment, revascularization procedures, major complications, and outcomes during the hospital stay; and medications at discharge. (The full protocol is available at http://www.riks-hia.se). To ensure the validity of the information entered into the database a single specially trained monitor visited participating hospitals and compared information in the patients’ records, including ECG, with the information entered into the RIKS-HIA database in 30–40 randomly chosen patients for each hospital. Consequently, the proportion of the cases reviewed by the monitor was about 7% of the whole study population.

Mortality data were obtained by merging the RIKS-HIA database with the National Death Register, which includes the vital status of all Swedish citizens in 1995 through 2005. The causes of death were so far only available through 2003. Previous history of stroke, renal failure, chronic pulmonary disease, dementia, cancer, history of HF, and peripheral vascular disease were obtained by merging with the National Patient Register, which includes diagnosis on all patients hospitalized in Sweden from 1987 and forward. All patients for whom data were entered into RIKS-HIA were informed of their participation in the registry (patients could request to be excluded) and the long-term follow-up. The registry and the merging with registries were approved by the National Board of Health and Welfare and the Swedish Data Inspection Board.

The Ethics Committee of the Uppsala University Hospital has approved the study.

Between 1995 and 2003, all patients with symptomatic left ventricular failure requiring hospitalization in the CCU and with HF as final diagnosis were included. Patients with pacemaker QRS complexes or typical symptoms of acute ischaemic cardiac event with electrocardiographic or biochemical criteria of acute coronary syndrome were excluded. In patients with rehospitalization due to HF during follow-up period, the first admission was considered as the time for start of follow-up. LBBB was diagnosed according to standard definitions:14 QRS-duration greater than or equal to 0.12 s, predominantly upright QRS complexes with slurred R waves in leads I, V5, and V6, and QS or rS pattern in V1. ECG was classified according to two variables: rhythm (sinus, atrial fibrillation or flutter, pacemaker rhythm, other rhythm); and QRS complex (normal, LBBB/pacemaker complex, new Q wave, old or uncertain date of Q wave, other pathological QRS complex). Starting in a hierarchical order from alternative one in each of the two variables, the first applicable alternative that was correct for that part of ECG was the coding for it. Left ventricular ejection fraction (LVEF) was determined by echocardiography before discharge and coded into RIKS-HIA in 6 groups: not performed, performed (no information about LVEF), normal (LVEF>50%), mild dysfunction (LVEF 40–50%), moderate dysfunction (LVEF 30–39%), severe dysfunction (LVEF < 30%). Before 2001, LVEF was not a part of the mandatory protocol, and echocardiography was registered only as performed or not performed. Hence, analyses including LVEF were available in only a minority of patients and survival analyses including LVEF could only be traced for 5 years.

Statistical analysis

Different patient strata were compared by χ2 tests for categorical variables and the t-test for continuous variables. All reported P-values are two-sided and a P-value < 0.01 was considered significant. These comparisons were however of little interest since mass significance was reached in most variables due to the large number of patients in the main study population. Logistic regression analyses for 1-, 2-, 3-, 5-, and 10-year survival were performed including 24 covariates: age, sex, Killip class, previous MI, history of HF, history of atrial fibrillation, peripheral vascular disease, diabetes mellitus, history of stroke, renal failure, chronic pulmonary disease, dementia, cancer within 3 years, history of hypertension, history of coronary artery revascularization, and medication before study entry (including ACE-inhibitors or angiotensin II receptor blockers, antiplatelet therapy, anticoagulants, beta-blockers, calcium channel blockers, digitalis, diuretics, lipid-lowering drugs, and long-acting nitrates) to identify whether LBBB by itself significantly influenced long-term mortality. In a subgroup where LVEF was available, from 2001 and forward, this was added to the 24 covariates. Several interactions between variables including age, gender, LBBB, and LVEF were tested but these had no significant influence on the results. Only patients who would have been possible to follow for the full duration of each predefined time period (1, 2, 3, 5, and 10 years) were allowed into each of the separate logistic analyses for the risk of death during that period of years. Adjusted survival curves were constructed by Cox regression analysis including the same variables but stratifying by the absence or presence of LBBB on ECG. Statistical analyses were performed with the statistical program SPSS version 14.0 software (SPSS Inc.).

Results

Patient material and baseline characteristics

Nineteen hospitals participated in the registry in 1995, which gradually increased to 72 of all 77 Swedish hospitals in 2003. During this period, there were 395 435 admissions to the coronary care units of which 27 235 admissions had HF as the final diagnosis. Of these, 1088 (4%) had pacemaker rhythm with pacemaker QRS complexes, and 4462 admissions were patients that were already included in the study at a previous date during the 9-year period of the present RIKS-HIA study. These two subgroups of admissions were excluded from the study (Figure 1). A total number of 21 685 patients were included in the study of whom 4395 (20%) had LBBB. In the baseline characteristics, it could be noted that LBBB patients were slightly older and had a significantly higher prevalence of concomitant cardiac disease in terms of previous MI and history of HF (Table 1). Consequently, the LBBB population had a significantly higher proportion of pharmacological therapy for these conditions on admission compared with no LBBB patients, and Killip classes were worse for the LBBB group (Table 1).

Figure 1

Schematic representation of selection procedure of patients admitted for heart failure from The Register of Information and Knowledge about Swedish Heart Intensive care Admission (RIKS-HIA). ACS, acute coronary syndrome; HF, heart failure; LBBB, left bundle branch block.

View this table:
Table 1

Characteristics of 21 685 heart failure patients on admission to intensive heart care units in Sweden

ECG on admissionNo. LBBB (n = 17 290)LBBB (n = 4395)P-value
Age (SD)74.7 (10.8)76.1 (9.7)<0.001
Male (%)54.558.1<0.001
Concomitant risk factors/history of disease (%)
 Current smoker13.410.9<0.001
 Stroke16.515.10.023
 Renal failure3.62.90.021
 Chronic pulmonary disease12.111.40.201
 Dementia1.00.70.075
 Cancer within 3 years4.54.60.795
 Previous MI45.549.5<0.001
 History of heart failure48.358.0<0.001
 Atrial fibrillation35.234.80.778
 Peripheral vascular disease8.58.70.716
 Diabetes mellitus27.527.60.843
 Hypertension37.336.70.414
 Previous PCI7.06.00.037
 Previous CABG10.112.7<0.001
Treatment on admission (%)
 ACE-inhibitor / AII-blocker39.248.6<0.001
 Antiplatelet therapy48.849.00.849
 Warfarin15.217.40.018
 Beta-blocker45.243.90.114
 Calcium channel inhibitor16.215.00.056
 Digitalis22.127.3<0.001
 Diuretics oral63.470.2<0.001
 Lipid lowering15.416.20.175
 Nitroglycerin29.934.2<0.001
Index events (%)
 Circulatory arrest at entry0.81.20.044
 Killip class 143.938.5<0.001
 Killip class 235.836.5
 Killip class 39.811.8
 Killip class 410.413.2
  • AII indicates angiotensin II receptor; ACE-inhibitor, angiotensin converting enzyme inhibitor; CABG, coronary artery bypass graft surgery; PCI, percutaneous coronary intervention.

In-hospital intervention and long-term medical treatment

The median time of the in-hospital stay was not different between the LBBB and no LBBB group, 4 (IQR 2–8) days and 4 (IQR 2–7) days, respectively. Patients with LBBB were more likely to receive intravenous nitroglycerin, and intravenous diuretics compared with the no LBBB group (Table 2). The percentage of patients undergoing PCI or CABG during the in-hospital period was not different between the groups. At discharge, a higher proportion of LBBB patients were on ACE-inhibitors or angiotensin II receptor blocker, warfarin, digitalis, diuretics, and nitroglycerin compared with no LBBB patients (Table 2).

View this table:
Table 2

Proportion of in-hospital complications, interventions, and long-term medical treatment in 21 685 heart failure patients

ECG on admissionNo LBBB (n = 17 290)LBBB (n = 4395)P-value
IV beta-blocker13.312.50.184
IV nitroglycerin24.928.5<0.001
IV inotropic agents3.64.70.001
IV/SC anticoagulant18.217.30.187
IV diuretics67.872.4<0.001
PCI/CABG during in-hospital stay1.61.20.054
Medical treatment at discharge (%)
 ACE-inhibitor / A II-blockers60.667.8<0.001
 Antiplatelet therapy57.056.90.914
 Warfarin20.922.40.040
 Beta-blocker57.254.50.003
 Calcium channel blocker14.612.0<0.001
 Digitalis30.735.4<0.001
 Diuretics oral85.089.0<0.001
 Lipid lowering20.320.30.939
 Nitroglycerin35.438.00.002
In-hospital complications (%)
 New atrial fibrillation14.113.10.778
 In-hospital mortality7.47.60.449
  • AII indicates angiotensin II receptor; ACE-inhibitor, angiotensin converting enzyme inhibitor; CABG, coronary artery bypass graft surgery; IV, Intravenous; PCI, percutaneous coronary intervention.

In-hospital complications, in-hospital, and long-term mortality

The incidence of new atrial fibrillation or death during the in-hospital stay did not differ between LBBB and no LBBB groups (Table 2). In-hospital mortality in the study population was 7.4% (1609 of 21 685 patients) which was not statistically different between LBBB and no LBBB patients, 7.6% (335 of 4395) and 7.4% (1274 of 17 290) respectively, P = 0.45. Unadjusted and adjusted long-term survival is shown in Table 3, and Figure 2A and B. The unadjusted risk of death was higher for LBBB patients at all studied time periods (Table 3); 1-, 5-, and 10-year mortality was 31.5 vs. 28.4%, 69.3 vs. 61.3%, and 90.1 vs. 84.7% for HF patients with respectively without LBBB. The adjusted mortality was slightly increased at 3 and 5 years but not at earlier or later observations (Table 3). Information about echocardiographic results including LVEF was available for analysis in 3327 of 21 685 (15.3%) patients of whom 679 (20.4%) with LBBB. As mentioned in the methods section, the survival analyses in the subgroup of study population with LVEF available could be traced for 5 years. Patients who had echocardiography performed had, irrespectively of LBBB or not, significantly better survival than the total cohort even after adjusting for admission year (OR between 0.72 and 0.77 when analysed for 1-, 2-, 3-, or 5-year mortality, P < 0.001 in all analyses). The proportion of LBBB patients with LVEF < 40% was significantly higher than the no LBBB group, 70 and 51% respectively, P < 0.001. In this subgroup, with moderate to severe left ventricular dysfunction, 1-year mortality was 23% in LBBB and not significantly different from that in no LBBB patients, 20%, P = 0.31 in a univariate analysis. The corresponding figures for subgroup of patients with LVEF ≥ 40% were 20 and 17% respectively, P = 0.48. One-year mortality in patients with LVEF <40% did not significantly differ compared with those with LVEF ≥ 40% in the LBBB group, P = 0.40. Unadjusted cumulative survival for patients where LVEF was available is shown in Figure 3A and adjusted survival in Figure 3B. After adjusting for age, comorbidity, prior medication, and LVEF, there remained no significant difference in 1-, 2-, 3-, or 5-year survival in HF patients with or without LBBB on ECG (Table 4).

Figure 2

Estimated survival after admission for heart failure patients with or without left bundle branch block. (A) Unadjusted data. (B) Adjusted for differences in clinical characteristics and concomitant diseases by Cox multivariable hazards regression. Solid line, no left bundle branch block; broken line, left bundle branch block. Numbers at risk indicated below the diagram.

Figure 3

Estimated survival after admission for heart failure patients with or without left bundle branch block where left ventricular ejection fraction was available. (A) Unadjusted data. (B) Adjusted for differences in clinical characteristics and concomitant diseases by Cox multivariable hazards regression. Solid line = no left bundle branch block; broken line, left bundle branch block. Numbers at risk indicated below the diagram.

View this table:
Table 3

Unadjusted and by logistic regression adjusted long-term mortality in heart failure patients with or without left bundle branch block on ECG admitted to intensive heart care units in Sweden

Length of follow-upN, dead/n, total number at risk (%)χ2, P-valueAdjusted OR (95% CI)Logistic regression, P-value
LBBBNo LBBB
1 year1385/4390 (31.5)4911/17280 (28.4)<0.0011.01 (0.93–1.09)0.882
2 years1918/4390 (43.7)6717/17280 (38.9)<0.0011.07 (0.99–1.15)0.094
3 years2093/3869 (54.1)7208/15126 (47.7)<0.0011.12 (1.03–1.22)0.008
5 years1988/2867 (69.3)6806/11102 (61.3)<0.0011.21 (1.10–1.35)<0.001
10 years254/282 (90.1)676/798 (84.7)0.0250.94 (0.62–1.42)0.759
  • Adjusted odds ratio (OR) calculated by logistic regression analysis including 24 variables as described in the methods section in the calculations above.

View this table:
Table 4

Unadjusted long-term mortality in the subgroup of heart failure patients with left ventricular ejection fraction available stratified by the presence of left bundle branch block on ECG and adjusted odds ratio calculated with same 24 variables plus left ventricular ejection fraction

Length of follow-upN, dead/n, total number at risk (%)χ2, P-valueAdjusted OR (95% CI)Logistic regression, P-value
LBBBNo LBBB
1 year141/679 (20.8)469/2648 (17.7)0.0670.88 (0.69–1.11)0.285
2 years154/494 (31.2)485/1818 (26.7)0.0480.93 (0.73–1.19)0.572
3 years158/369 (42.8)409/1225 (33.4)0.0011.12 (0.85–1.47)0.416
5 years67/127 (52.8)205/463 (44.3)0.0890.99 (0.62–1.56)0.953

Cause of death

There was no major difference in the cause of the death between patients with or without LBBB; ischaemic heart disease 47.6 vs. 43.3%, arrhythmia 2.3 vs. 2.6%, HF 6.9 vs. 6.8%, other cardiac diseases 10.3 vs. 7.6%, stroke 4.1 vs. 5.0%, cancer 5.4 vs. 8.4%, and other causes of death 22.3 vs. 28.0%.

Data validity

When 1972 randomly chosen computer forms from 38 hospitals containing 161 280 variables points were reviewed by an external monitor, there was 97% agreement between the registered information and the source data in the patient records among the 24 covariates included in the logistic and Cox regression analyses. With regard to the ECG variable, in 0.6% of the 1972 cases, LBBB was miscoded as another type of QRS complex, and in 2.2%, a QRS complex that really was not LBBB had been incorrectly coded as LBBB.

Discussion

This study is the first to include a large enough population to allow an extensive multivariable statistical analysis to evaluate contribution of LBBB per se on long-term mortality in a HF population. In the present study, the prevalence of LBBB in the HF population was 20%, consistent with previous studies1,2 indicating that it is representative for a general HF population. The study also provided a unique long-term perspective, with almost complete follow-up until death in all patients, showing the very high 8–10% yearly mortality in symptomatic HF patients with as well as without LBBB. The most important new finding in the present study was that after adjustment for differences in clinical characteristics, comorbidities and myocardial function, LBBB per se was not associated with higher mortality in patients with symptomatic HF requiring hospitalization. The finding that the extent of comorbidity and degree of left ventricular dysfunction explain the prognostic impact of LBBB in highly symptomatic HF population contrasts to previous findings. However, previous reports, which however, included a limited number of patients,1,10 only investigated patients with dilated cardiomyopathy10 or did not use the same extensive adjustment for covariates for mortality.10 In a previous study of 5517 unselected outpatients with HF LBBB was found to be associated with an increase of 1 year mortality by 36% after adjustment for different clinical variables.2 In that study, all-cause mortality was increased by 93% from higher score of NYHA classification (III–IV), and by 75% from previous hospitalization due to HF. In contrast, all our patients had HF symptoms that required hospitalization, which resulted in a substantially higher 1 year mortality (29%) compared to the Baldasseroni's report (12%). These factors may explain the discrepancy between these two studies and also suggest that the more severe symptoms in the hospitalized HF population may override the contribution of LBBB on long-term prognosis. The lack of significant independent risk contribution of LBBB to mortality in patients with underlying cardiac disease is also supported by a study consisting of a large cohort with acute myocardial infarction reported by Stenestrand et al.15 Even in this study, the higher 1 year mortality in LBBB population could mainly be explained by the significantly higher prevalence of comorbid conditions and left ventricular dysfunction rather than LBBB per se after adjustment for different risk factors.

In the present study, there was no difference in 1-year mortality between LBBB patients with LVEF ≥ 40% and those with LVEF < 40%. Dyssynchronous septal motion of the left ventricle in the presence of LBBB with subsequent difficulties in appropriate assessment of LVEF by conventional echocardiography has been reported previously and might lead to underestimation of LVEF in LBBB patients.1618 Wide QRS, especially in the presence of LBBB, has been correlated to LV dyssynchrony.3,4 Therefore, this has been proposed as a key prerequisite for selection of patient for CRT.9,19,20 However, the correlation between the QRS width and regional electromechanical LV dyssynchrony has not been completely clarified21 and for a given QRS width there is a considerable scatter in response to CRT, responsive patients with narrow complexes and less responsive ones with wide complexes exist.22 Furthermore, a reduction in QRS interval does not necessarily correlate with an improvement of left ventricular haemodynamic variables.23 Recently, echocardiographic techniques such as tissue doppler imaging have been suggested to offer superior sensitivity and specificity than ECG in selecting HF patients with dyssynchrony.24 In a tissue doppler echocardiographic study of patients with HF and narrow QRS (≤120 ms), dyssynchronous left ventricular contraction was demonstrated in 51% of patients.25 Furthermore, in recently published studies, including patients with HF and echocardiographic evidence of LV dyssynchrony, the clinical and functional improvement of CRT was similar in patients with narrow (≤120 ms) and wide (>120 ms) QRS-complexes.26,27 The CARE-HF study of patients with NYHA Class III or IV HF, wide QRS-complexes (≥120 ms), and EF ≤ 35% showed a significant reduction of all-cause mortality by CRT compared with conventional HF treatment.9 The risk reduction of death by CRT in HF patients with dyssynchronous left ventricular contraction indicates that regional LV dyssynchrony may have an important detrimental contribution on long-term prognosis in HF population. The present report highlights that many other HF patients have as severe outcome as those with wide QRS-complexes. Therefore, it needs to be investigated whether the survival benefit of CRT can be obtained in the HF patients without prolonged QRS interval who have LV dyssynchrony. Accordingly, a better definition of LV dyssynchrony criteria, other than a simple QRS interval, is warranted in order to select HF population for CRT.18

Reliability of data

The current RIKS-HIA was first used in 1991 and has become a reliable source of information on consecutive patients admitted to participating units. ECG had a high validity with 97% correct coding in a large sample from the study cohort. With the exception of patients with pacemaker QRS complexes, there were no exclusions in the HF population resulting from the presence or absence of specific risk factors, comorbidities, anticipated adverse effects, participations in clinical trials, or contraindications to certain medications. The representativeness of the cohort was also strengthened by the inclusion of all consecutive patients with HF from general population at the centres with different levels of care from 94% of the hospitals within an entire country.

Study limitations

The low proportion of the study population with available LVEF and the lack of confirmation of echocardiography findings by a core laboratory are limitations in this study. As the population with available LVEF had better long-term survival the results from this subgroup might not be relevant for the whole HF population. Still the unadjusted difference in survival between patients with LBBB compared to those without LBBB were very similar to the whole HF cohort. It is also important to mention that ECGs were not reviewed by a core laboratory but in spite of this limitation, the validity of the ECG report was high.

Conclusion

LBBB occurs in 1/5 in HF patients requiring hospitalization and is associated with a very high mortality. However, the high long-term mortality in this population appears to be caused by cardiac comorbidities and myocardial dysfunction rather than the LBBB per se. These findings could have important implications on how to select patients that will benefit from CRT. Future studies are needed where LVEF is properly evaluated to investigate further the influence of LBBB on HF and the selection of patients for CRT in order to improve these patients very poor long-term survival.

Conflict of interest: none declared.

Footnotes

  • These authors have contributed equally to the manuscript.

References

View Abstract