Journal of the American College of Cardiology

2003 by the American College of Cardiology Foundation

Published by Elsevier Inc.

Vol. 42, No. 5, 2003

ISSN 0735-1097/03/$30.00
doi:10.1016/S0735-1097(03)00833-7

Heart Rate Recovery After Exercise Is a

Predictor of Mortality, Independent of the
Angiographic Severity of Coronary Disease
Deepak P. Vivekananthan, MD,* Eugene H. Blackstone, MD, FACC, Claire E. Pothier, MA,*
Michael S. Lauer, MD, FACC, FAHA*
Cleveland, Ohio
We sought to determine whether abnormal heart rate recovery predicts mortality independent
of the angiographic severity of coronary disease.
BACKGROUND An attenuated decrease in heart rate after exercise, or heart rate recovery (HRR), has been
shown to predict mortality. There are few data on its prognostic significance once the
angiographic severity of coronary artery disease (CAD) is ascertained.
METHODS
For six years we followed 2,935 consecutive patients who underwent symptom-limited
exercise testing for suspected CAD and then had a coronary angiogram within 90 days. The
HRR was abnormal if 12 beats/min during the first minute after exercise, except among
patients undergoing stress echocardiography, in whom the cutoff was 18 beats/min.
Angiographic CAD was considered severe if the Duke CAD Prognostic Severity Index was
42 (on a scale of 0 to 100), which corresponds to a level of CAD where revascularization
is associated with better long-term survival.
RESULTS
Severe CAD was present in 421 patients (14%), whereas abnormal HRR was noted in 838
patients (29%). There were 336 deaths (11%). Mortality was predicted by abnormal HRR
(hazard ratio [HR] 2.5, 95% confidence interval [CI] 2.0 to 3.1; p 0.0001) and by severe
CAD (HR 2.0, 95% CI 1.6 to 2.6; p 0.0001); both variables provided additive prognostic
information. After adjusting for age, gender, standard risk factors, medications, exercise
capacity, and left ventricular function, abnormal HRR remained predictive of death (adjusted
HR 1.6, 95% CI 1.2 to 2.0; p 0.0001); severe CAD was also predictive (adjusted HR 1.4,
95% CI 1.1 to 1.9; p 0.008).
CONCLUSIONS Even after taking into account the angiographic severity of CAD, left ventricular function,
and exercise capacity, HRR is independently predictive of mortality. (J Am Coll Cardiol
2003;42:831 8) 2003 by the American College of Cardiology Foundation
OBJECTIVES

Impaired heart rate recovery (HRR) after exercise predicts

mortality (13), even after accounting for ischemia (3,4),
chronotropic incompetence (13,5), and the Duke treadmill
score (5). There are few data on its prognostic significance
once the angiographic severity of coronary artery disease
See page 839
(CAD) is ascertained. In a recent study of male veterans,
HRR predicted death, independent of angiographic results,
but there was no correlation between HRR and the angiographic severity of disease (6). We examined the association
between abnormal HRR and mortality among men and
women referred for exercise stress testing and coronary
angiography. All-cause mortality was chosen as an unbiased
and objective end point (7).

From the *Departments of Cardiovascular Medicine, Cardiothoracic Surgery, and

Epidemiology and Biostatistics, Cleveland Clinic Foundation, Cleveland, Ohio.
This study was funded by Grant HL-66004, National Heart, Lung, and Blood
Institute of the National Institutes of Health, Bethesda, Maryland (Dr. Lauer,
Principal Investigator).
Manuscript received October 9, 2002; revised manuscript received January 1, 2003,
accepted January 11, 2003.

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

METHODS
Patient population. The cohort was derived from consecutive adults referred for symptom-limited treadmill testing
at the Cleveland Clinic Foundation between September
1990 and March 1998. All patients were undergoing their
first treadmill test at our institution. Patients were eligible if
they underwent coronary angiography within 90 days. Exclusion criteria included a history of heart failure, valvular
disease, pre-excitation, congenital disease, coronary interventions or surgery, pacemaker placement, use of digoxin,
atrial fibrillation, and absence of a recorded U.S. Social
Security number. The Cleveland Clinic Foundations Institutional Review Board approved the performance of research on this clinical data base.
Of the 2,935 patients who met these criteria, 1,384 (47%)
have been included in previous publications (1,4,5). However, we have not published any data on their angiographic
characteristics relation to HRR or whether HRR predicts
death in these patients once angiographic data are known.
Clinical data. A structured interview and chart review were
conducted before each treadmill test regarding symptoms,
medications, risk factors, cardiac history, and noncardiac
diagnoses (8). Resting hypertension was defined as systolic
blood pressure 140 mm Hg, diastolic blood pressure 90
mm Hg, or treatment with antihypertensive medication (9).

832

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

Abbreviations and Acronyms

CAD coronary artery disease
CI
confidence interval
ESRD end-stage renal disease
HR
hazard ratio
HRR heart rate recovery
MET metabolic equivalent
PVD peripheral vascular disease

Diagnoses of diabetes mellitus and chronic lung disease

were determined on the basis of chart review, patient
interviews, and medication use. Hypercholesterolemia was
defined as a recent total cholesterol value 200 mg/dl or use
of a lipid-lowering drug. A history of coronary disease was
considered present when there were documented hospitalizations for myocardial infarction or unstable angina. All
clinical data, as well as directly measured height and weight,
were prospectively recorded on-line.
Exercise testing. Symptom-limited exercise testing procedures in our laboratory have been described (10 13). Each
patient underwent testing according to standard protocols.
Trained exercise physiologists and/or cardiology fellows
prospectively collected physiologic and hemodynamic data
during testing, including symptoms, heart rate, heart
rhythm, blood pressure, and estimated functional capacity in
metabolic equivalents (METs; where 1 MET 3.5 ml/kg
per min of oxygen consumption). Functional capacity was
defined as fair or poor for age and gender, using a previously
described scheme (8). Specifically, functional capacity was
considered poor among men if it was 8, 7.5, 7, 6, 5.5, 4.5,
and 3.5 METs for those age 29, 30 to 39, 40 to 49, 50 to
59, 60 to 69, 70 to 79, and 80 years, respectively. The
corresponding values among women were 7.5, 7, 6, 5, 4.5,
3.5, and 2.5 METs. Among men, functional capacity was
considered fair if it was not poor and it was 11, 10, 8.5, 8,
7, 5.5, and 4.5 METs for those age 29, 30 to 39, 40 to 49,
50 to 59, 60 to 69, 70 to 79, and 80 years, respectively.
The corresponding values among women were 10, 9, 8, 7, 6,
4.5, and 4 METs. We did not consider excellent functional
capacity as a separate variable, because in previous work, we
found that there were little differences in outcome among
patients with average, good, and excellent functional capacity for age and gender (8).
A chronotropic response during exercise was defined as
the percentage of heart rate reserve used at peak exercise
(10). A failure to use 80% of the heart rate reserve defined
chronotropic incompetence (10). ST segments were considered abnormal if there was at least 1 mm of horizontal or
down-sloping ST-segment depression 80 ms after the J
point in at least three consecutive beats in two contiguous
leads.
Heart rate recovery. Most patients (n 2,426) spent at
least 2 min in a cool-down period during treadmill testing at
a speed of 2.4 km (1.5 miles) per hour and a grade of 2.5%
after achieving peak work load. Those patients who under-

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

went stress echocardiography (n 509) did not have a

cool-down period. The value for HRR was defined as the
difference in the heart rate from peak exercise to 1 min after
peak exercise.
Abnormal HRR was defined as 12 beats/min for
standard exercise testing and 18 beats/min for patients
who underwent stress echocardiography (3). We have previously published prognostically based justifications for
these cut-off values (1,3).
Coronary angiography. We have described the methods of
the coronary angiographic procedures performed in our
institution (12,14). Patients were referred for coronary
angiography at the discretion of the treating physician. The
referring physicians were blinded to the patients HRR
values. Cardiologists who were blinded to the patients
HRR values and to the hypothesis of this study semiquantitatively analyzed each coronary angiogram. The results
were then entered into an angiographic data base.
We used the Duke CAD Prognostic Severity Index
(15,16) to grade the angiographic severity of CAD. This
system assigns a prognostic weight of 0 to 100 based on
analyses of the Duke cardiovascular data bank. For example,
a value of 0 corresponds to no significant coronary disease,
whereas a score of 100 corresponds to 95% left main
coronary stenosis. At least one 50% stenotic lesion is
required to define the existence of any CAD. Severe CAD
was prospectively defined as CAD having a prognostic
weight of 42 by the Duke CAD Prognostic Severity
Index. This value was chosen because it represents a
threshold where mechanical revascularization has been previously shown to reduce mortality rates (15).
Left ventricular systolic function was semiquantitatively
analyzed by contrast ventriculography or transthoracic echocardiography. We have prognostically validated visual estimates of left ventricular systolic function (3).
End points. The primary end point was all-cause mortality
during a median of six years of follow-up. Mortality was
assessed by using the Social Security Death Master Files
(17). In previous work, the Social Security Death Index has
been shown to have a very high sensitivity (5) and specificity
(18,19).
Statistical analyses. Comparisons between patients with
and without abnormal HRR were made by using the
Wilcoxon rank-sum test for continuous variables and the
chi-squared test for categorical variables. The association
between HRR and all-cause mortality was assessed by Cox
regression analyses (20). Stratified analyses of prespecified
subgroups were performed according to the angiographic
severity of CAD, gender, age, left ventricular systolic
function, functional capacity, exercise-induced angina, ischemic ST-segment changes, or medication usage (including
beta-blockers), with formal testing of interaction terms. The
Cox proportional hazards assumption was confirmed by
inspection of weighted Schoenfeld residuals (21).
When evaluating an association within any given data set,
the apparent strength of that association may be overesti-

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

833

Table 1. Baseline Characteristics According to HRR

Characteristic

Normal HRR
(>12 beats/min)*
(n 2,097)

Abnormal HRR
(<12 beats/min)
(n 838)

p Value

Age (yrs)
Male gender
White race
Hypertension
Resting tachycardia (100 beats/min)
Current or recent smoker
Insulin use
NIDDM
Hypercholesterolemia
Previous CAD
Previous MI
ACE inhibitor use
Beta-blocker use
Aspirin use
Diuretic use
Nondihydropyridine calcium channel blocker use
Nitrates
Vasodilators
Peripheral vascular disease
COPD
Asthma
ESRD

59 10
1,594 (76%)
1,863 (89%)
926 (44%)
53 (3%)
407 (19%)
88 (4%)
183 (9%)
840 (40%)
1,000 (48%)
504 (24%)
244 (12%)
553 (26%)
832 (40%)
275 (13%)
193 (9%)
570 (27%)
838 (40%)
61 (3%)
56 (3%)
57 (3%)
14 (0.7%)

63 10
621 (74%)
732 (87%)
505 (60%)
72 (9%)
181 (22%)
72 (9%)
126 (15%)
317 (38%)
435 (52%)
246 (29%)
151 (18%)
271 (32%)
337 (40%)
172 (21%)
96 (11%)
278 (33%)
435 (57%)
35 (4%)
60 (8%)
23 (3%)
20 (2%)

0.0001
0.23
0.25
0.0001
0.0001
0.19
0.0001
0.0001
0.26
0.04
0.003
0.0001
0.001
0.79
0.0001
0.04
0.001
0.0001
0.08
0.0001
0.97
0.0001

*For stress echocardiography, 18 beats/min. For stress echocardiography, 18 beats/min. Data are presented as the mean
value SD or number (%) of subjects.
ACE angiotensin-converting enzyme; CAD coronary artery disease; COPD chronic obstructive pulmonary disease;
ESRD end-stage renal disease; HRR heart rate recovery; MI myocardial infarction; NIDDM noninsulin-dependent
diabetes mellitus.

mated because of idiosyncrasies of the data. To correct for

this over-optimism, we used a technique called bootstrapping. Here, 250 new data sets were assembled by randomly
selecting subjects from the main data set, with the possibility of selecting subjects once, never, or multiple times.
Furthermore, each of these new randomly constructed data
sets had only 80% of the number of subjects as the main data
set, again for the purpose of minimizing the effects of
idiosyncratic observations (22,23). A stepwise multivariable
Cox regression analysis was performed on each of these 250
data sets, using p 0.10 for model entry and p 0.05 for
retention. Those variables that were entered into at least
50% of models were considered for a subsequent set of 1,000
fixed resamplings, which were used for estimation of hazard
ratios (HRs) and confidence intervals (CIs). By fixed resamplings, we mean that the Cox models applied to each of
these 1,000 bootstrap-generated data sets contained the
candidate variables that were entered into 50% of the
original models and that no variable selection process was
used; all variables were forced in.
To assess the association between HRR, considered as a
continuous variable, and mortality, we used parametric
methods taking into account the possibility of changing the
absolute hazard of death over time (24). These enabled us to
plot estimated five-year mortality as a function of HRR,
exercise capacity, angiographic coronary disease severity,
and other variables. We chose to use a parametric approach
for this because the use of the Cox regression model for

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

estimation of absolute event risk has been considered by

some to be problematic (25,26).
The associations of varying angiographic severities of
CAD with abnormal HRR were assessed with the chisquared test.
The statistical software package SAS version 8.2 (SAS
Inc., Cary, North Carolina) was used to perform all of the
analyses. Bootstrapping SAS macros were written by Eugene H. Blackstone (available on request). The parametric
hazard analyses were performed using the PROC HAZRD,
PROC HAZPLOT, and PROC HAZPRED functions.

RESULTS
Baseline, exercise, and angiographic characteristics. Of
2,935 eligible patients, 838 (29%) had abnormal HRR.
Baseline characteristics according to HRR are summarized
in Table 1. Patients with abnormal HRR were older, more
likely to have hypertension or diabetes, and more likely to
have a history of myocardial infarction.
Exercise and angiographic characteristics are presented in
Tables 2 and 3. Patients with abnormal HRR were more
likely to have chronotropic incompetence and less likely to
have abnormal ST-segment changes. Severe CAD was
present in 421 patients (14%). Patients with abnormal HRR
were more likely to have severe CAD (p 0.02). However,
abnormal HRR had a sensitivity of only 31% and a
specificity of 76% for the detection of any CAD.

834

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

Table 2. Exercise Characteristics According to HRR

Characteristic
Peak METs
Men
Women
Poor functional capacity
Fair functional capacity
Peak heart rate (beats/min)
Chronotropic incompetence
(no beta-blocker)
HRR (beats/min)
Angina, nontest-limiting
Angina, test-limiting
Abnormal ST-segment changes

Normal HRR
(n 2,097)

Abnormal HRR
(n 838)

p Value

8.5 2.3
6.5 1.7
247 (12%)
537 (26%)
146 22
489 (23%)

6.9 2.2
5.3 1.6
203 (24%)
289 (34%)
134 22
298 (36%)

0.0001
0.0001
0.0001
0.0001
0.0001
0.0001

23 8
476 (23%)
49 (2%)
578 (28%)

85
175 (21%)
18 (2%)
168 (20%)

0.0001
0.28
0.75
0.0001

Data are presented as the mean value SD or number (%) of subjects.

HRR heart rate recovery; METs metabolic equivalents.

Mortality and HRR. During the median follow-up period

of six years, there were 336 deaths (11%). Abnormal HRR
predicted death (19% vs. 8%; unadjusted HR 2.5, 95% CI
2.0 to 3.1; p 0.00001). Of the 336 patients who died, 162
(48%) had an abnormally low HRR. Other univariate
predictors of mortality included older age, severe CAD, low
ejection fraction, fair or poor functional capacity, a low
chronotropic response index, and male gender (Table 4).
Abnormal HRR provided additive prognostic information to the angiographic severity of CAD (Fig. 1). There
was no interaction between abnormal HRR and the angiographic severity of CAD, gender, age, left ventricular
systolic function, functional capacity, exercise-induced angina, or medication usage (including beta-blockers).
Multivariable Cox regression analyses. Results of bootstrap resamplings and multivariable Cox regression analyses
are shown in Table 5. Variables that entered the final model
included age, use of aspirin, abnormal HRR, poor and fair
functional capacity, non-test-terminating angina, current or
recent smoking, end-stage renal disease (ESRD), low ejection fraction, male gender, resting tachycardia, peripheral
vascular disease (PVD), severe CAD, chronotropic incompetence in the absence of a beta-blocker, insulin use, and
nifedipine use. After adjustments were made for age, gender, left ventricular function, resting heart rate, chronotropic
response index, exercise capacity, exercise-induced angina,

ischemic ST-segment depression, the presence or absence of

hypertension, diabetes, PVD, hyperlipidemia, tobacco
abuse, chronic lung disease, ESRD, the use or nonuse of
beta-blockers, nondihydropyridine calcium channel blockers, lipid-lowering therapy, vasodilator medications, and
angiographic severity of CAD, a low value for HRR
emerged as a strong predictor of death (adjusted HR 1.6,
95% CI 1.2 to 2.0; p 0.0001).
Even after excluding patients with exercise-induced angina and ischemic ST-segment depression, abnormal HRR
was still predictive of mortality (adjusted HR 2.1, 95% CI
1.2 to 3.5; p 0.009). We also performed a stratified
analysis according to functional capacity. There were 204
deaths in the 1,276 patients with poor or fair functional
capacity. In this group, abnormal HRR strongly predicted
the risk of death (HR 2.6, 95% 2.0 to 3.4; p 0.0001). In
patients with normal functional capacity, abnormal HRR
still predicted death, although the association was not as
strong (HR 1.7, 95% CI 1.2 to 2.4; p 0.0067). There was
no interaction between functional capacity and the risk of
death associated with abnormal HRR (p 0.07). It is
important to note, though, that decreased functional capacity was itself a powerful independent predictor of death
(Tables 4 and 5).
We also constructed a Cox model in which the CAD
severity index was considered as a continuous variable. The

Table 3. Angiographic and Ventriculographic Characteristics According to HRR

Characteristic

Normal HRR
(n 2,097)

Abnormal HRR
(n 838)

p Value

Duke CAD index

Any CAD
Severe CAD (prognostic score 42)
LMCA disease
Proximal LAD disease
LCx disease
RCA/PDA disease
Low ejection fraction (40%)

19 (032)
1,279 (61%)
280 (13%)
79 (4%)
476 (23%)
593 (29%)
757 (36%)
537 (25%)

23 (037)
576 (69%)
141 (17%)
41 (5%)
230 (27%)
282 (34%)
368 (43%)
269 (32%)

0.0001
0.0001
0.02
0.16
0.007
0.004
0.0001
0.0004

Data are presented as the median value (interquartile range) or number (%) of subjects.
CAD coronary artery disease; HRR heart rate recovery; LAD left anterior descending; LCx left circumflex
coronary artery; LMCA left main coronary artery; PDA posterior descending artery; RCA right coronary artery.

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

835

Table 4. Risk of Death in Prespecified Subgroups on Univariate Analysis

Variable

HR
(95% CI)

Chi-Squared
Statistic

p Value

Old age (65 yrs)

Low CRI (no beta-blocker)
Low EF
Severe CAD
Any CAD
Poor functional capacity
Low HRR
Nondiagnostic ST-segment changes
Resting tachycardia (100 beats/min)
Fair functional capacity
Male gender
Abnormal ST-segment changes

2.1 (1.72.6)
1.9 (1.52.4)
1.9 (1.52.3)
2.0 (1.62.6)
1.9 (1.52.4)
2.8 (2.13.7)
3.2 (2.05.2)
1.6 (1.22.1)
2.0 (1.43.1)
2.0 (1.62.6)
1.5 (1.02.3)
1.2 (0.91.5)

43
34
32
30
24
23
22
13
12
10
4
1

0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0004
0.0005
0.0013
0.04
0.32

CI confidence interval; CRI chronotropic response index; EF ejection fraction; HR hazard ratio; other abbreviations
as in Table 1.

association between HRR and death was unchanged. The

CAD severity index was also predictive of the risk of death
(adjusted HR 1.1 for 20-point increase, 95% CI 1.0 to 1.3;
p 0.03).
Angiographic severity of CAD, HRR, and mortality in
women. There were 720 women (25%), among whom 217
(30%) had abnormal HRR and 62 (9%) had severe CAD.
During follow-up, 65 died (9%). Abnormal HRR was
independently predictive of death, even after accounting for
age, coronary disease severity, functional capacity, ejection
fraction, smoking, use of aspirin, and a history of ESRD
(adjusted HR 1.5, 95% CI 1.2 to 1.9; p 0.002). Severe
angiographic CAD was also an independent predictor of
death in women (adjusted HR 1.4, 95% CI 1.1 to 1.9; p

0.01), and its strength of association was similar to that of

men.
Impact of revascularization. During the first three months
after exercise testing, 435 patients (15%) underwent coronary bypass grafting and 368 (13%) underwent percutaneous
revascularization. There was no association between having
an abnormal HRR and undergoing subsequent bypass
grafting (16% vs. 14%, p 0.21) or percutaneous revascularization (13% vs. 12%, p 0.33). However, severe CAD
was associated with subsequent coronary artery bypass
grafting (45% vs. 10%, p 0.0001) but negatively associated with a subsequent percutaneous coronary intervention
(10% vs. 13%, p 0.04). After taking into account
subsequent revascularization by adding a revascularization

Figure 1. Kaplan-Meier plot relating heart rate (HR) recovery and angiographic severity of coronary artery disease (CAD) to risk of death.

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

836

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

Table 5. Results of Cox Multivariate Proportional Model*

Variable

Percentage of Bootstrap
Models Entered

Adjusted HR
(95% CI)

Age (10-yr increase)

Aspirin use
Abnormal HRR
Poor functional capacity
Fair functional capacity
Non-terminating angina
Current or recent smoking
ESRD
Low ejection fraction
Male gender
Resting tachycardia
Peripheral vascular disease
Severe CAD
Chronotropic incompetence (no beta-blocker)
Insulin use
Nifedipine use

100
99
97
93
89
84
83
83
82
79
79
72
68
65
61
51

1.9 (1.72.2)
0.6 (0.50.7)
1.6 (1.22.0)
2.1 (1.52.9)
1.7 (1.32.2)
0.6 (0.40.8)
1.5 (1.12.0)
3.1 (1.56.2)
1.5 (1.21.9)
1.5 (1.12.0)
1.9 (1.12.8)
1.8 (1.12.6)
1.4 (1.11.9)
1.3 (1.01.7)
1.5 (1.02.2)
1.4 (1.01.9)

*See Methods section of text for details regarding bootstrap resampling methods and modeling techniques. Models in which
the variable was retained at p 0.05.
CI confidence interval; HR hazard ratio; other abbreviations as in Table 1.

term into a supplementary Cox model, the association

between HRR and mortality was unaffected.
Type of recovery protocol. There were 509 patients (17%)
who underwent stress echocardiography, which mandated
assuming a left lateral decubitus position immediately after
exercise, as opposed to the more standard upright cooldown. During follow-up, 41 (8.1%) of these patients died.
Even after adjusting for age, gender, functional capacity,
and angiographic severity of coronary disease, HRR was
predictive of mortality among patients undergoing stress
echocardiography (adjusted HR 1.9, 95% CI 1.0 to 3.5; p
0.06), similar to patients undergoing other types of testing
(adjusted HR 1.9, 95% CI 1.5 to 2.4; p 0.0001; p 0.98
for interaction).
Parametric analyses. Parametric analyses confirmed an
independent association of HRR with mortality, even after
accounting for age, gender, left ventricular function, angiographic severity of coronary disease, functional capacity, and
other possible confounders. No significant interactions were
noted. Figure 2 shows how HRR was associated with
predicted five-year mortality as a function of functional
capacity.
Previously published data. Among the 1,551 patients on
whom no HRR data have been published, 450 (29%) had
abnormal HRR and 144 died during follow-up. Abnormal
HRR was associated with death in unadjusted analyses (16%
vs. 7%; unadjusted HR 2.5, 95% CI 1.8 to 3.4; p 0.0001)
and after adjusting for the severity of coronary disease, along
with multiple other confounders (adjusted HR 1.8, 95% CI
1.3 to 2.5; p 0.0009). There was no interaction between
having been included in a previous publication and the
association of HRR with death (p 0.89 for interaction
term).
Impact of beta-blockers and heart rate-lowering calcium
blockers. There were 1,614 patients (55%) who were taking neither beta-blockers nor heart ratelowering calcium

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

blockers. Of these, 405 (25%) had abnormal HRR; during

follow-up, 170 died (11%). Abnormal HRR predicted death
by itself (19% vs. 8%; unadjusted HR 2.6, 95% CI 1.9 to 3.5;
p 0.0001) and after adjustment for the severity of
coronary disease and other confounders (adjusted HR 1.4,
95% CI 1.0 to 1.9; p 0.04).
Of note, among all 2,935 patients, there was no interaction between beta-blocker use and the association of HRR
with death (p 0.34 for interaction). Similarly, there was
no interaction with calcium channel blocker use (p 0.40
for interaction). Of the 824 patients taking beta-blockers,
231 (28%) had a heart rate 60 beats/min at baseline.
Among these patients, the risk of death associated with
abnormal HRR was significant (15% vs. 8%; HR 2.1, 95%
CI 1.3 to 3.5; p 0.003). The risk of death associated with
abnormal HRR did not differ among patients who had a

Figure 2. Association of heart rate recovery, considered as a continuous

variable, with five-year predicted death rate as a function of exercise
capacity. Results of parametric analyses. BPM beats/min; METs
metabolic equivalents.

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318

baseline pulse 60 beats/min (19% vs. 9%; HR 2.2, 95% CI

1.0 to 4.6; p 0.04).

DISCUSSION
Among patients who had exercise stress testing and who
had coronary angiography within 90 days, an attenuated
HRR after exercise predicted mortality. The mortality risk
of abnormal HRR was comparable to having angiographically severe CAD (15,16). In fact, abnormal HRR provided
additive prognostic information to the angiographic severity
of CAD.
Previously, our group found that abnormal HRR predicted mortality in multiple patient groups (15). Only one
previous study (6) has evaluated the prognostic significance
of abnormal HRR once the angiographic severity of CAD is
ascertained. In a male veteran population, Shetler et al. (6)
found abnormal HRR to be predictive of increased mortality, independent of the angiographic severity of CAD.
Attenuated HRR was not helpful in predicting the presence
of significant angiographic coronary disease.
Our study confirms and extends these findings in several
important respects. First, 25% of our patients were women,
among whom HRR predicted death, independent of coronary disease severity, to the same extent as in men. Second,
our study included patients who underwent either standard
exercise stress testing with a 2-min upright cool-down
period or stress echocardiography, which, per protocol, did
not have an upright cool-down period. Irrespective of the
testing protocol used, we found abnormal HRR to be
independently predictive of death, even after accounting for
the angiographic severity of coronary disease. Third, we also
confirmed that HRR is a poor diagnostic test for angiographic CAD, with a sensitivity of only 31% and a specificity of 76%. Finally, our use of bootstrap resamplings and
parametric analyses enabled us to deal with excessive overoptimism inherent when analyzing only one data set and to
explore the importance of HRR when considered as a
continuous variable.
Our study adds to growing evidence that abnormal HRR
adversely affects mortality, independent of ischemic burden
(1,5). The drop in heart rate after exercise is thought to
represent withdrawal of the sympathetic nervous system
and, more importantly, reactivation of the parasympathetic
nervous system (27,28). Reduced vagal activity has been
shown to adversely impact mortality in other patient populations (29,30).
At this time, additional study is needed to delineate
therapeutic options for patients with abnormal HRR and to
determine whether HRR is a modifiable risk factor. To
date, no therapy has been systematically investigated for its
ability to attenuate the risk associated with abnormal HRR.
Preliminary evidence does suggest that cardiac rehabilitation
improves HRR (31).
Our study was limited because it involved a single
tertiary-care referral center and thus was open to biases of

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

837

patient selection and referral patterns for coronary angiography. Moreover, the use of coronary angiography alone as
a risk factor for future cardiac events has been called into
question (32). Due to inherent problems of vessel overlap,
vessel tortuosity, and vessel resolution, stenoses can be overand/or underestimated. In addition, the degree of luminal
stenoses may not correlate with the propensity for plaque
rupture. Finally, our study did not identify specific causes of
death, but used all-cause mortality as the end point.
We found that HRR provides additive prognostic information to the angiographic severity of coronary disease. Our
findings provide further evidence supporting the routine
incorporation of HRR into exercise testing interpretation.
Reprint requests and correspondence: Dr. Michael S. Lauer,
Desk F25, Department of Cardiovascular Medicine, Cleveland
Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195.
E-mail: Lauerm@ccf.org.

REFERENCES
1. Cole CR, Blackstone EH, Pashkow FJ, Snader CE, Lauer MS.
Heart-rate recovery immediately after exercise as a predictor of
mortality. N Engl J Med 1999;341:13517.
2. Cole CR, Foody JM, Blackstone EH, Lauer MS. Heart rate recovery
after submaximal exercise testing as a predictor of mortality in a
cardiovascularly healthy cohort. Ann Intern Med 2000;132:5525.
3. Watanabe J, Thamilarasan M, Blackstone EH, Thomas JD, Lauer
MS. Heart rate recovery immediately after treadmill exercise and left
ventricular systolic dysfunction as predictors of mortality: the case of
stress echocardiography. Circulation 2001;104:19116.
4. Diaz LA, Brunken RC, Blackstone EH, Snader CE, Lauer MS.
Independent contribution of myocardial perfusion defects to exercise
capacity and heart rate recovery for prediction of all-cause mortality in
patients with known or suspected coronary heart disease. J Am Coll
Cardiol 2001;37:1558 64.
5. Nishime EO, Cole CR, Blackstone EH, Pashkow FJ, Lauer MS.
Heart rate recovery and treadmill exercise score as predictors of
mortality in patients referred for exercise ECG. JAMA 2000;284:
13928.
6. Shetler K, Marcus R, Froelicher VF, et al. Heart rate recovery:
validation and methodologic issues. J Am Coll Cardiol 2001;38:
1980 7.
7. Lauer MS, Blackstone EH, Young JB, Topol EJ. Cause of death in
clinical research: time for a reassessment? J Am Coll Cardiol 1999;34:
618 20.
8. Snader CE, Marwick TH, Pashkow FJ, Harvey SA, Thomas JD,
Lauer MS. Importance of estimated functional capacity as a predictor
of all-cause mortality among patients referred for exercise thallium
single-photon emission computed tomography: report of 3,400 patients from a single center. J Am Coll Cardiol 1997;30:6418.
9. The Fifth Report of the Joint National Committee on Detection,
Evaluation, and Treatment of High Blood Pressure (JNC-V). Arch
Intern Med 1993;153:154 83.
10. Lauer MS, Francis GS, Okin PM, Pashkow FJ, Snader CE, Marwick
TH. Impaired chronotropic response to exercise stress testing as a
predictor of mortality. JAMA 1999;281:524 9.
11. Lauer MS, Pashkow FJ, Snader CE, Harvey SA, Thomas JD,
Marwick TH. Gender and referral for coronary angiography after
treadmill thallium testing. Am J Cardiol 1996;78:278 83.
12. Lauer MS, Pashkow FJ, Harvey SA, Marwick TH, Thomas JD.
Angiographic and prognostic implications of an exaggerated exercise
systolic blood pressure response and rest systolic blood pressure in
adults undergoing evaluation for suspected coronary artery disease.
J Am Coll Cardiol 1995;26:1630 6.
13. Marwick TH, Mehta R, Arheart K, Lauer MS. Use of exercise
echocardiography for prognostic evaluation of patients with known or
suspected coronary artery disease. J Am Coll Cardiol 1997;30:8390.

838

Vivekananthan et al.
Heart Rate Recovery and Coronary Disease

14. Brener SJ, Pashkow FJ, Harvey SA, Marwick TH, Thomas JD, Lauer
MS. Chronotropic response to exercise predicts angiographic severity
in patients with suspected or stable coronary artery disease. Am J
Cardiol 1995;76:1228 32.
15. Jones RH, Kesler K, Phillips HR 3rd, et al. Long-term survival
benefits of coronary artery bypass grafting and percutaneous transluminal angioplasty in patients with coronary artery disease. J Thorac
Cardiovasc Surg 1996;111:101325.
16. Mark DB, Nelson CL, Califf RM, et al. Continuing evolution of
therapy for coronary artery disease: initial results from the era of
coronary angioplasty. Circulation 1994;89:201525.
17. Curb JD, Ford CE, Pressel S, Palmer M, Babcock C, Hawkins CM.
Ascertainment of vital status through the National Death Index and
the Social Security Administration. Am J Epidemiol 1985;121:754
66.
18. Boyle CA, Decoufle P. National sources of vital status information:
extent of coverage and possible selectivity in reporting. Am J Epidemiol 1990;131:160 8.
19. Newman TB, Brown AN. Use of commercial record linkage software
and vital statistics to identify patient deaths. J Am Med Inform Assoc
1997;4:2337.
20. Cox S. Regression models and life tables. J Am Stat Soc 1972;34:187
220.
21. Cain KC, Lange NT. Approximate case influence for the proportional
hazards regression model with censored data. Biometrics 1984;40:
4939.
22. Chen CH, George SL. The bootstrap and identification of prognostic
factors via Coxs proportional hazards regression model. Stat Med
1985;4:39 46.
23. Collett D, Stepniewska K. Some practical issues in binary data
analysis. Stat Med 1999;18:2209 21.

Downloaded From: http://content.onlinejacc.org/ on 04/29/2015

JACC Vol. 42, No. 5, 2003

September 3, 2003:8318
24. Blackstone EH, Naftel DC, Turner MEJ. The decomposition of
time-varying hazard into phases, each incorporating a separate
stream of concomitant information. J Am Stat Assoc 1986;81:61524.
25. Nieto FJ, Coresh J. Adjusting survival curves for confounders: a review
and a new method. Am J Epidemiol 1996;143:1059 68.
26. Lee J, Yoshizawa C, Wilkens L, Lee HP. Covariance adjustment of
survival curves based on Coxs proportional hazards regression model.
Comput Appl Biosci 1992;8:237.
27. Imai K, Sato H, Hori M, et al. Vagally mediated heart rate recovery
after exercise is accelerated in athletes but blunted in patients with
chronic heart failure. J Am Coll Cardiol 1994;24:1529 35.
28. Pierpont GL, Stolpman DR, Gornick CC. Heart rate recovery
post-exercise as an index of parasympathetic activity. J Auton Nerv
Syst 2000;80:169 74.
29. La Rovere MT, Bigger JT Jr., Marcus FI, Mortara A, Schwartz PJ,
and the ATRAMI (Autonomic Tone and Reflexes After Myocardial
Infarction) Investigators. Baroreflex sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction.
Lancet 1998;351:478 84.
30. La Rovere MT, Pinna GD, Hohnloser SH, et al. Baroreflex sensitivity
and heart rate variability in the identification of patients at risk for
life-threatening arrhythmias: implications for clinical trials. Circulation 2001;103:20727.
31. Hao SC, Chai A, Kligfield P. Heart rate recovery response to
symptom-limited treadmill exercise after cardiac rehabilitation in
patients with coronary artery disease with and without recent events.
Am J Cardiol 2002;90:7635.
32. Topol EJ, Nissen SE. Our preoccupation with coronary luminology:
the dissociation between clinical and angiographic findings in ischemic
heart disease. Circulation 1995;92:233342.