Grady TA, Chiu AC, Snader CE, Marwick TH, Thomas JD, Pashkow FJ, Lauer MS. Prognostic Significance of Exercise-Induced Left Bundle-Branch Block. JAMA. 1998;279(2):153–156. doi:10.1001/jama.279.2.153
From the Department of Cardiology, George M. and Linda H. Kaufman Center for Heart Failure, Cleveland Clinic Foundation, Cleveland, Ohio.
Clinical Cardiology section editors: Bruce Brundage, MD, University
of California, Los Angeles, School of Medicine; Margaret A. Winker, MD, Senior
Context.— Approximately 0.5% of all patients who undergo exercise testing develop
a transient left bundle-branch block (LBBB) during exercise, but its prognostic
significance is unclear.
Objective.— To determine whether exercise-induced LBBB is an independent predictor
of mortality and cardiac morbidity.
Design.— Matched control cohort study. Between September 1990 and February 10,
1994, 17277 exercise stress tests were performed on patients.
Setting.— Tertiary care, academic medical center.
Patients.— From the cohort, 70 cases of exercise-induced LBBB were identified.
The controls comprised 70 individuals without LBBB at rest or during exercise
that matched the 70 cases based on age, test date, sex, prior history of coronary
artery disease, hypertension, diabetes, smoking, and β-blocker use.
Main Outcome Measures.— All-cause mortality, percutaneous coronary intervention, open heart
surgery, nonfatal myocardial infarction, documented symptomatic or sustained
ventricular tachydysrhythmia, or implantation of a permanent pacemaker or
an implantable cardiac defibrillator.
Results.— A total of 37 events (28 events from the exercise-induced LBBB cases
and 9 from the control cohort) occurred in 25 patients (17 exercise-induced
LBBB patients and 8 control patients) during a mean follow-up period of 3.7
(0.9 years) (median, 3.8 years [range, 0.9-5.2 years]). There were 7 deaths,
of which 5 occurred among patients with exercise-induced LBBB. Four-year Kaplan-Meier
event rates were 19% among exercise-induced LBBB patients and 10% among controls
(log rank χ2, 5.2; P =.02). After
further adjusting for small differences in age, exercise-induced LBBB remained
associated with a higher risk of primary events (adjusted relative risk, 2.78;
95% confidence interval, 1.16-6.65; P =.02).
Conclusion.— Exercise-induced LBBB independently predicts a higher risk of death
and major cardiac events.
EXERCISE-INDUCED left bundle-branch block (LBBB) occurs during approximately
0.5% of exercise stress tests. Prior studies examining the prognostic significance
of exercise-induced LBBB have been limited by their small size, lack of matched
controls, and failure to adjust for potential confounders.1- 19
Therefore, we performed a matched control cohort study to determine whether
exercise-induced LBBB is an independent predictor for major cardiovascular
morbidity and mortality in the largest series of exercise-induced LBBB reported
Between September 4, 1990, and February 10, 1994, 17277 patients underwent
symptom-limited treadmill stress testing at the Cleveland Clinic Foundation
in Cleveland, Ohio; 70 cases (0.41%) of exercise-induced LBBB were identified.
Rest and exercise electrocardiograms (ECGs) were independently reviewed by
3 physicians to confirm the diagnosis of exercise-induced LBBB. The diagnosis
of complete LBBB was made from the 12-lead ECG if all the following criteria
were met: conduction originating above the atrioventricular node; a QRS duration
of 120 milliseconds or more; predominantly upright complexes with broad-slurred
R waves in leads I, V5, and V6; and a QS or RS pattern
in V1 with a normal intrinsicoid deflection of 35 milliseconds.6,7,19 Patients with a δ
wave or a short PR interval that suggests an accessory AV bypass tract were
excluded. Patients with permanent pacemakers and/or evidence of preexcitation
were also excluded. Exercise-induced LBBB was defined by LBBB that was documented
only during treadmill exercise stress testing. There could be no history of
LBBB and the LBBB abnormality had to resolve before the patient left the laboratory.
From the same database, a matched, controlled cohort of 70 patients
was selected. This cohort comprised individuals matched to cases by the predetermined
variables of sex, hypertension, diabetes, smoking, β-blocker use, and
history of coronary artery disease. For each case, a pool of potential control
patients was assembled. A unique control for each case was chosen by identifying
the patient closest in age and test date; no maximum differences in these
variables were prespecified. By definition, control patients' ECGs were without
LBBB at rest and during exercise.
The research protocol was approved by the Cleveland Clinic Foundation
Institutional Review Board.
All patients underwent symptom-limited treadmill stress testing usually
according to the Bruce or modified Bruce protocol. During each stage of exercise
and recovery, data on symptoms, rhythm, heart rate, blood pressure, workload
in metabolic equivalents (METs), and ST segment changes were prospectively
collected and recorded online until recovery was complete. Participants were
encouraged to achieve at least 85% of their maximum age-predicted exercise
heart rate (calculated as 220−age of the participant). Participants
were not allowed to lean on handrails during exercise testing. Among controls,
an ischemic ST-segment response was considered present if there was 1 mm of
horizontal or down-sloping ST-segment depression 80 milliseconds after the
By both chart review and structured interview, the following characteristics
were identified: age, sex, a history of diabetes, a history of hypertension
(defined as resting systolic blood pressure of >140 mm Hg, and/or a resting
diastolic blood pressure of >90 mm Hg, and/or taking antihypertensive medication),
use of a β-blocker, a history of coronary artery disease, and current
or past smoking.21
Cholesterol values predating the exercise test or occurring within 90
days following the stress test were identified for 63 (90%) of the case patients
and 59 (84%) of the control patients. Hypercholesterolemia was defined as
having a total cholesterol value of 6.21 mmol/L (240 mg/dL) and/or taking
The presence of a Q or QS wave was defined as having an amplitude of
0.25 mm or more and a duration exceeding 20 milliseconds in the majority of
beats in any 1 lead except aVR.6,7
Left ventricular function and ejection fraction were determined by echocardiography,
ventriculography, and/or nuclear multigated acquisition scintigraphy. Left
ventricular function was quantified in 59 patients (84%) with exercise-induced
LBBB and 32 controls (46%). Information regarding degree of coronary artery
disease was obtained by reviewing coronary angiograms. Narrowing of 50% or
more was considered significant (Table 2).
The prospectively defined primary end point in this study was a composite
of all-cause mortality, percutaneous and/or surgical revascularization, nonfatal
myocardial infarction, and need for a permanent pacemaker and/or an implantable
cardiac defibrillator with documented symptomatic or sustained ventricular
tachycardia or ventricular fibrillation by either telemetry and/or Holter
monitor. Follow-up was obtained by a structured chart review and telephone
interview. If a patient had died, the next of kin was interviewed and the
death certificate was reviewed. Follow-up was performed by 2 physicians who
were blinded to results of the stress test. Follow-up was obtained for 100%
of patients. The duration of follow-up was a mean (SD) of 3.7 (0.9) years
with a median of 3.8 years (range, 0.9-5.2 years).
Continuous variables are described as mean (SD). Differences in nonmatched
baseline characteristics were compared between the 2 groups using the Student t, Wilcoxon rank sum, χ2, and Fisher exact
tests as appropriate.
Cumulative incidence rates of primary events according to presence or
absence of exercise-induced LBBB were calculated using the Kaplan-Meier product
limit method. Differences between event-free survival curves were compared
using the log-rank test. The Cox proportional hazards model was used to further
adjust for age and extent of coronary artery disease at the time of exercise
testing.22 The validity of the proportional
hazards assumption was confirmed by examination of weighted residuals.23 Analyses were not specifically performed for matched
pairs. All statistical analyses were performed using SAS 6.12 statistical
package (SAS Inc, Cary, NC).
Before performing the analyses, a power calculation was performed that
assumed 36% as being the lowest rate of developing a cardiovascular event
for cases; this event rate was obtained from the literature reporting outcomes
for similar populations.15 It was assumed that
cardiovascular events among all patients undergoing exercise stress testing
occurs at a 3% rate over a similar period based on the results of the follow-up
period in this study, which found a 1% rate per year or 3% to 4% over 3.7
years. A matched control cohort study with a study population of 70 subjects
and 70 controls was confirmed to have more than 90% statistical power to detect
The cases and controls had similar baseline characteristics except for
age (Table 1). For both the case
and control groups, for those who had coronary angiography prior to exercise
testing, the mean time between coronary angiography and exercise stress testing
was 3.5 (3.5) years with a median of 2.3 years (range, 0.01-13.1 years). There
were no marked differences in the distribution of severity of coronary disease
between the 2 groups (Table 2).
Indications for stress testing among patients with exercise-induced
LBBB were follow-up for known coronary disease in 43 (61%), evaluation for
possible coronary disease in 23 (33%), and arrhythmia evaluation in 4 (6%).
The corresponding values for indications for stress testing among controls
were 43 (61%), 22 (31%), and 3 (4%), along with 2 (4%) who were referred for
other reasons (P=.98 for differences in indications
for stress testing between the 2 groups). No marked difference in exercise
characteristics were noted (Table 3).
The mean length of follow-up for all patients was 3.7 (0.9) years with
a median of 3.8 years (range, 0.9-5.2 years). Primary end points were more
common among exercise-induced LBBB cases than controls (Table 4). Figure 1 illustrates
the event-free survival of the 2 groups over the period of follow-up. Four-year
cumulative event rates were 10% (8 patients) in the control cohort and 19%
(17 patients) in the case cohort (log-rank χ2, 5.2; P=.02). After adjusting for the small differences in age in the 2 groups,
the relative risk was 2.78 (95% confidence interval [CI], 1.16-6.65; P =.02).
In additional analyses, we considered the impact of a history of coronary
artery disease on outcome, a variable for which controls had been matched
to cases. Of the 86 patients with known coronary artery disease, 19 (22%)
went on to develop a prespecified end point while 6 (13%) without a known
history of coronary artery disease went on to develop a prespecified end point.
In a Cox model that included exercise-induced LBBB, age, and documented coronary
disease, the association of exercise-induced LBBB with the outcome measures
was unchanged with an adjusted relative risk of 2.73 (95% CI, 1.14-6.56; P =.02). Thus, the association of exercise-induced LBBB
with predefined end points was independent of documented coronary disease.
Among patients with a history of coronary artery disease, of the 7 patients
who went on to coronary artery bypass surgery, 5 were in the exercise-induced
LBBB group and 2 were in the control group; there were no deaths among these
patients. Of the 15 patients who went on to percutaneous intervention, 10
patients were in the exercise-induced LBBB group and 5 were in the control
group; there was 1 death in each group.
When left ventricular function was analyzed, a low ejection fraction,
defined as less than 40%, did not influence the association of exercise-induced
LBBB with predefined end points. In a Cox model that included exercise-induced
LBBB, age, and known low ejection fraction, the association of exercise-induced
LBBB with adverse events was unchanged with an adjusted relative risk of 2.77
(95% CI, 1.15-6.67; P =.02).
The median heart rate when exercise-induced LBBB developed among the
cases was 121 beats per minute (25th to 75th percentiles; range, 105-140 beats
per minute). At the time of exercise-induced LBBB, the median blood pressure
was 168/88 mm Hg (25th to 75th percentiles; range, 164-182/84-98 mm Hg). Blood
pressure and heart rate did not fluctuate at the time when exercise-induced
LBBB developed. The mean metabolic equivalents achieved at the onset of exercise-induced
LBBB was 7.2 (2.5) with a median of 7.1 (range, 3.2-12.8) while the mean rate
pressure product was 26508 (4915) with a median of 26434 (range, 18732-29975).
There was no association between the heart rate at which LBBB occurred and
the risk of developing adverse events.
The prognostic significance of exercise-induced LBBB has been variably
reported in the literature.5- 19
Interpretation of prior studies is limited by lack of control groups, the
wide spectrum of patients referred for exercise testing, crossover to coronary
revascularization, and small sample sizes. Our investigation represents the
largest published series to date and is the first to our knowledge to utilize
a matched control cohort method. Despite the achievement of a comparable work
capacity, heart rate, and rate-pressure product, our study population demonstrated
a significantly lower event-free survival compared with the matched control
Wayne et al14 observed that exercise-induced
LBBB occurs most commonly in the presence of underlying heart disease, particularly
coronary artery disease. However, a smaller series reported that this phenomenon
was associated with normal coronary arteries.13
Two larger series by Bounhoure et al8 and Williams
et al15 reported a high prevalence of coronary
artery disease (64% to 75%) but a variable incidence of cardiac events (36%
to 50%). However, without considering a control group, the independent prognostic
impact of exercise-induced LBBB is difficult to gauge.
Our series complements the findings of Schneider et al4
from the Framingham Heart Study regarding the significance of newly acquired
LBBB. During an 18-year period of observation, of 5209 subjects followed,
56 patients (1.1%) developed LBBB. Fifty percent of these patients died within
10 years after the development of the abnormality and only 11% remained free
of clinically apparent cardiovascular events. Because the left bundle-branch
of the cardiac conduction system has a dual blood supply involving both the
left anterior descending coronary artery and posterior descending coronary
artery (a branch of a dominant right coronary artery and/or a dominant left
circumflex coronary artery), exercise-induced LBBB may be a clinical marker
for a greater degree of underlying coronary artery disease and/or ischemic
Some important limitations in our study need to be noted. Outcomes measured
in this study may reflect some degree of bias since patients with an exercise
stress test abnormality, such as exercise-induced LBBB, may have had closer
follow-up. Hence, earlier detection and treatment of these patients may have
had some role in their higher frequency of primary events. Markedly more deaths
occurred among the case patients, though, arguing against clinician bias being
the major reason for the difference in event rates. Another limitation is
that angiography was not performed in all patients and often preceded the
exercise test by more than 2 years, precluding complete adjustment of the
analysis for baseline coronary artery disease. However, the data available
indicate a similar degree of severity between the groups.
Because exercise-induced LBBB is a comparatively infrequent finding
during exercise testing, our case and control populations are small. Thus,
any conclusion about lack of difference regarding certain parameters, for
example, ejection fraction, may be due to insufficient sample size. Nonetheless,
to our knowledge, this analysis is the largest study to date examining the
prognostic significance of exercise-induced LBBB. We did not systematically
record the length of time the LBBB persisted after exercise and what impact
this might have on prognosis. As our hospital is a major tertiary referral
center, our patient population may not reflect the population at large who
are undergoing exercise stress testing, but the frequency of exercise-induced
LBBB in our population was 0.41%, a value which closely parallels that reported
in the literature.8- 10,12- 15,19
Though exercise-induced LBBB is comparatively infrequent, approximately
10000 new cases will be observed annually in the United States, where more
than 2 million exercise stress tests are currently performed each year.24 Thus, the independent contribution of exercise-induced
LBBB to prognosis warrants attention among clinicians whose patients undergo