Heart Rate Recovery and Impact of Myocardial Revascularization on Long-Term Mortality
Background— Although heart rate recovery (HRR) predicts mortality after exercise testing, its ability to identify patients likely to benefit after revascularization is unknown. We sought to determine whether HRR can identify patients likely to have improved survival after revascularization.
Methods and Results— A total of 8861 patients undergoing treadmill nuclear or echocardiographic testing were divided into early revascularization (percutaneous coronary intervention or bypass surgery within 3 months) and non–early revascularization groups. Prespecified subgroup analysis was performed based on the presence or absence of ischemia, normal or impaired functional capacity, and normal or abnormal HRR. The primary end point was all-cause mortality. Early revascularization occurred in 552 patients. We propensity-matched 508 early revascularization patients to 508 non–early revascularization patients on the basis of 48 possible confounders. This constituted the present study cohort. During 8-year follow-up, 232 patients died. Overall, revascularization was associated with a slight but not significant decrease in mortality (hazard ratio [HR] 0.80, 95% CI 0.62 to 1.03). A significant decrease in mortality after revascularization was present in patients with imaging evidence of stress-induced ischemia (HR 0.62, 95% CI 0.44 to 0.87). Ischemic patients with normal HRR had significantly lower mortality with revascularization (HR 0.55, 95% CI 0.34 to 0.90), whereas ischemic patients with abnormal HRR did not (HR 0.78, 95% CI 0.47 to 1.29); however, the test for interaction between these 2 groups was not significant (P=0.34).
Conclusions— In patients with imaging evidence of myocardial ischemia, an abnormal HRR is associated with a nonsignificant trend toward blunting the survival improvement associated with early revascularization. HRR does not appear to identify patients likely to have a survival benefit.
Received April 28, 2004; revision received August 27, 2004; accepted September 8, 2004.
Exercise tests provide prognostic and diagnostic information.1 Abnormal stress tests often result in revascularization. The identification of patients likely to benefit from revascularization could help to optimize its use.
See p 2778
Larger declines in heart rate after exercise, known as heart rate recovery (HRR), predict lower mortality.2,3 However, the predictive value of HRR for mortality in patients undergoing coronary revascularization is not as well studied. We aimed to determine whether HRR can identify patients likely to have improved survival after revascularization.
We followed up 8861 consecutive patients undergoing treadmill echocardiography (n=4093) or treadmill nuclear (n=4768) testing at our institution from 1990 to 1998. Exclusion criteria included heart failure (systolic or diastolic, compensated or uncompensated), atrial fibrillation, digoxin, valvular or congenital heart disease, preexcitation, and pacemaker use. Heart failure patients were excluded because they have a high likelihood of autonomic dysfunction4 and poor prognosis. Patients were divided into an early revascularization arm (n=552), defined as percutaneous coronary intervention or CABG within 3 months of the exercise test, and a non–early revascularization arm (n=8309). The Cleveland Clinic Institutional Review Board approved research using this database.
Treadmill nuclear and treadmill echocardiographic tests were performed according to established protocols.2,3,5,6 Patients underwent symptom-limited treadmill tests with a goal of achieving ≥85% of maximum predicted heart rate. With treadmill nuclear tests, test termination was followed by a 2-minute cooldown. There was no cooldown period for treadmill echocardiography because of the need to obtain poststress images immediately. An exercise test was classified as demonstrating ischemia if new wall-motion abnormalities were observed on stress images of exercise echocardiograms or if reversible perfusion defects were demonstrated on stress images of exercise nuclear tests. HRR was change in heart rate from peak exercise to 1 minute after exercise. Abnormal HRR was defined as ≤12 bpm for nuclear testing and ≤18 bpm for echocardiography.5 Functional capacity was abnormal when patients had poor or fair functional capacity. In women, functional capacity was abnormal if ≤10, 9, 8, 7, 6, 4.5, and 4 metabolic equivalents (METS) were achieved by those ≤29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, 70 to 79, and ≥80 years old, respectively. In men, functional capacity was abnormal if ≤11, 10, 8.5, 8, 7, 5.5, and 4.5 METS were achieved by those ≤29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, 70 to 79, and ≥80 years old, respectively.6,7
The primary end point was all-cause mortality, as determined by the Social Security death database.8 High specificity of 99.9%9 and sensitivity of 97%3 have been demonstrated with examination of the Social Security Death Index. Median follow-up for survivors was 8 years.
Propensity matching matches patients with controls when observational data are being analyzed.10 Of 552 patients undergoing early revascularization, we were able, using nonparsimonious logistic regression, to propensity-match (based on 48 variables; Table 1) 508 patients to 508 patients not undergoing early revascularization. These 1016 patients composed the present study cohort. This cohort was subdivided into prespecified subgroups based on (1) presence/absence of ischemia on stress testing, (2) normal/abnormal functional capacity, and (3) normal/abnormal HRR. All subgroup analyses performed in the propensity-matched cohort were also performed in the entire patient sample.
Statistical analysis was performed with SAS software, version 8.2 (SAS Inc). A probability value of <0.05 was considered statistically significant.
Comparisons between patients undergoing early revascularization and those not undergoing early revascularization were performed with the χ2 test for categorical variables and the Wilcoxon rank-sum test for continuous variables. Continuous variables are listed as mean±SD. The effect of revascularization on mortality was assessed with Cox proportional hazards, with adjustments made for the propensity score.12 Covariates used to calculate propensity scores for the proportional hazards models included all variables listed in Table 1. Survival was estimated by the Kaplan-Meier method.13 Formal interaction testing assessed whether any interaction between revascularization and HRR or between revascularization and impaired fitness existed. There was no interaction of revascularization with impaired fitness nor was there interaction of revascularization with abnormal HRR in any of the subgroups. The Schoenfeld residual test was performed for all models in Tables 2 and 3⇓ to assess whether the Cox proportional hazards assumption was met.14
In supplementary analysis, HRR was analyzed as a continuous variable among all patients undergoing nuclear imaging who had imaging evidence of ischemia. Mortality estimates were adjusted for the variables listed in Table 1.
Supplemental analysis was also performed to assess early revascularization as a time-dependent variable. Subjects were coded 0 until time of revascularization, 1 thereafter, and 0 for those who did not have revascularization within 3 months of treadmill test. We controlled for confounders by collapsing possible confounding variables (Table 1) into 1 propensity score.
Among all 8861 patients, those undergoing early revascularization were older, had more prior coronary artery disease, and had more stress test abnormalities than those not undergoing revascularization (Table 1). In the 1016-patient propensity-matched cohort, of the 508 patients not undergoing early revascularization, 70% underwent nuclear stress testing, and 30% underwent echocardiographic stress testing. In the 508 subjects undergoing revascularization, 65% underwent nuclear stress testing, and 35% underwent echocardiographic stress testing (P=0.11 for comparison for test modality between revascularization and nonrevascularization groups). After propensity matching, no significant difference in baseline characteristics or exercise test results existed between the revascularization and nonrevascularization arms (Table 1).
Among all propensity-matched patients, revascularization was associated with a nonsignificant trend toward lower mortality (hazard ratio [HR] 0.80, 95% CI 0.62 to 1.03, P=0.09; Figure 1). In prespecified subgroup analysis, patients with ischemia during stress testing had lower mortality after revascularization than ischemic patients who did not undergo revascularization (Table 2). In contrast, patients without ischemia had similar mortality rates regardless of revascularization. When this subgroup analysis was applied to all 8861 subjects, a similar pattern emerged (Table 2).
Mortality in Patients Demonstrating Ischemia on Stress Testing
Ischemic patients from the propensity-matched cohort were then classified by functional capacity and HRR. Patients with ischemia and impaired functional capacity had lower mortality with revascularization than those with no revascularization (Table 3). Subjects with ischemia and normal HRR also had lower mortality with revascularization. Patients with ischemia and abnormal HRR were at high risk of dying, but revascularization did not improve survival. When the same analysis was applied to the entire patient sample who had ischemia during stress testing, similar trends emerged (Table 3).
Mortality in Patients With Ischemia During Stress Testing and Impaired Functional Capacity
Patients with ischemia during stress testing and impaired functional capacity had lower mortality after revascularization than those who were not revascularized. This group was subdivided by HRR to determine whether HRR could identify which patients were likely to demonstrate improved outcomes with revascularization. In this subset, those with abnormal HRR had high event rates but did not have improved survival with revascularization, whereas those with normal HRR had lower mortality after revascularization (Figure 2A). In the entire patient sample, similar to the propensity-matched cohort, ischemic patients with impaired functional capacity and abnormal HRR did not have lower mortality after revascularization, whereas those with normal HRR did (Figure 2B).
Subjects Unable to Be Propensity Matched
We were unable to propensity match 44 of the 552 patients who were revascularized. Seventy percent of the nonmatched subjects had multivessel ischemia compared with 26% of the propensity-matched cohort. Multivessel scar was present in 7% of the nonmatched subjects compared with 20% of the propensity-matched cohort. Thus, the nonmatched patients were ideal revascularization candidates and were unable to be matched because few subjects had extensive ischemia and little scar without undergoing revascularization. There were 7 deaths (16%) in nonmatched subjects and 21 deaths (20%) in propensity-matched subjects.
HRR as a Continuous Variable
In the subset with nuclear evidence of ischemia (all having a cooldown period), there was a lower mortality rate with revascularization at high values of HRR (Figure 3). As HRR decreased, mortality rates increased, and any difference between the 2 groups disappeared.
Revascularization as a Time-Dependent Variable
Because revascularization is a stochastic phenomenon that occurs after stress testing, analysis of revascularization as a time-dependent variable was performed. In all subjects, early revascularization was not associated with mortality benefit (adjusted HR 0.95, 95% CI 0.74 to 1.22, P=0.68). In subjects with ischemia by imaging stress test, revascularization was associated with a nonsignificant trend toward lower mortality (adjusted HR 0.77, 95% CI 0.55 to 1.07, P=0.11). In patients with ischemia and normal HRR, revascularization was associated with lower mortality (adjusted HR 0.57, 95% CI 0.35 to 0.93, P=0.02), whereas in patients with ischemia and abnormal HRR, it was not (adjusted HR 1.07, 95% CI 0.69 to 1.68, P=0.75).
We estimated the required sample size, assuming power of 0.90 and α=0.05, for a prospective trial assessing whether HRR can predict mortality benefit from revascularization in patients with imaging evidence of ischemia. In ischemic patients with normal HRR, the use of our estimated 8-year mortality rate of 19% in the nonrevascularized group and 11% in the revascularized group (Table 3) results in a sample size of 882 total subjects. In ischemic patients with abnormal HRR, application of our 8-year mortality estimates of 35% in the nonrevascularized arm and 33% in the revascularization arm (Table 3) results in 23 774 subjects. If instead we assume a mortality rate of 30% in the revascularization group, then 3762 patients are needed.
We aimed to determine whether HRR can identify patients likely to reap a survival benefit from coronary revascularization. We studied patients undergoing exercise stress testing and stratified them by ischemia, functional capacity, and HRR. We found that ischemia, as has been reported,15 strongly predicted improved survival associated with early revascularization. Second, although ischemic patients with normal HRR had improved survival after revascularization, abnormal HRR in ischemic subjects was associated with a nonsignificant trend toward blunted survival benefit after revascularization.
In prespecified subgroup analyses, ischemic patients with abnormal functional capacity but normal HRR had improved survival after revascularization, whereas ischemic patients with abnormal functional capacity and abnormal HRR did not. The former may represent an intermediate level of risk in which abnormal functional capacity represents deconditioning, but autonomic function is not substantially impaired. In contrast, those ischemic patients with abnormal functional capacity and abnormal HRR may represent patients in such poor physiological shape, with advanced deconditioning and autonomic perturbations, that revascularization alone is not enough to affect positively their outcomes.
Alternatively, although chronic imbalances of the autonomic nervous system that favor sympathetic stimulation or blunt vagal activation adversely affect prognosis,16 and even though abnormal HRR reflects autonomic nervous system abnormalities, revascularization may simply not improve these autonomic abnormalities. The predictive value of HRR for mortality is independent of coronary artery disease severity,7 which suggests that coronary artery disease and abnormalities in HRR reflect different pathological processes.
Our end point was all-cause mortality. However, revascularization is also performed for angina relief, an end point that the present study was not designed to analyze.
Revascularization comprised both CABG and percutaneous coronary intervention. Survival benefit after revascularization has been demonstrated with CABG over percutaneous coronary intervention in certain instances.17 Stratification of the present analysis by mode of revascularization would have been preferable; however, this would have limited the number of events in each subgroup and precluded robust analysis.
Three months after stress testing was prespecified as the time point to separate the early revascularization and non–early revascularization groups. Owing to the lag between stress test and revascularization dates, patients could die while awaiting revascularization. Because we could not determine intention to treat immediately after the stress test but instead classified patients on the basis of treatment received, any deaths that occurred after the stress test but before planned revascularization would be classified as a death in the nonrevascularization arm, which potentially could make the nonrevascularization results appear worse. However, within 3 months of stress testing, only 2 deaths occurred in the revascularization group and 2 in the nonrevascularization group. Our results were similar when revascularization was analyzed as a time-dependent variable.
We could not compare ejection fraction between the revascularization and nonrevascularization groups because ejection fraction was not available among patients undergoing nuclear imaging. Among all patients, we were unable to assess the effect of HRR on revascularization as a continuous variable because we used both echocardiographic and nuclear stress tests. Echocardiographic and nuclear stress tests have different normal ranges for HRR,5 thus rendering problematic the direct comparison of absolute HRR. However, we did assess HRR as a continuous variable in the nuclear treadmill group (Figure 3) and found that, consistent with the categorical findings, as HRR decreased, so did the survival improvement associated with early revascularization.
The difficulty of interpreting marginally significant probability values in the context of multiple testing should be noted. Although our subgroup analysis was prespecified, significant probability values could be spurious in light of multiple testing and marginally significant probability values.
Finally, this was a nonrandomized observational study in which propensity matching was performed. Owing to the lack of randomization, physician and patient choice plays a role in treatment decisions. As such, our analysis is subject to the limitations and biases of a nonrandomized trial.
Conclusions and Implications
We confirm that ischemia on imaging testing appears to be a powerful predictor of mortality benefit after revascularization. In patients with imaging evidence of myocardial ischemia, an abnormal HRR is associated with a nonsignificant trend toward blunting the survival improvement associated with early revascularization. HRR does not appear to identify patients likely to have a survival benefit. Further studies are needed to elucidate whether ischemic patients with abnormal HRR derive other benefits from revascularization, perhaps focusing on nonmortality end points such as myocardial infarction, repeat revascularization, or symptom relief.
Drs Lauer and Blackstone and Ms Pothier receive support from the National Heart, Lung, and Blood Institute (grant HL-66004).
Nolan J, Batin PD, Andrews R, Lindsay SJ, Brooksby P, Mullen M, Baig W, Flapan AD, Cowley A, Prescott RJ, Neilson JM, Fox KA. Prospective study of heart rate variability and mortality in chronic heart failure: results of the United Kingdom Heart Failure Evaluation and Assessment of Risk Trial (UK-HEART). Circulation. 1998; 98: 1510–1516.
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: 1911–1916.
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: 641–648.
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: 754–766.
Newman TB, Brown AN. Use of commercial record linkage software and vital statistics to identify patient deaths. J Am Med Inform Assoc. 1997; 4: 233–237.
Cox D. Regression models and life tables. J R Stat Soc B. 1972; 34: 187–220.
Hachamovitch R, Hayes SW, Friedman JD, Cohen I, Berman DS. Comparison of the short-term survival benefit associated with revascularization compared with medical therapy in patients with no prior coronary artery disease undergoing stress myocardial perfusion single photon emission computed tomography. Circulation. 2003; 107: 2900–2907.