EXERCISE STRESS TESTING IN ASYMPtomatic

Transcription

1 ORIGINAL CONTRIBUTION Global Risk Scores and Exercise Testing for Predicting All-Cause Mortality in a Preventive Medicine Program Mehmet K. Aktas, MD Volkan Ozduran, MD Claire E. Pothier, MPH Richard Lang, MD, MPH Michael S. Lauer, MD EXERCISE STRESS TESTING IN ASYMPtomatic individuals has been shown to be of limited utility in detecting coronary artery disease. The American Heart Association, the American College of Cardiology, and the US Preventive Services Task Force do not recommend routine exercise stress testing in healthy asymptomatic individuals. 1,2 Nonetheless, recent guidelines suggest that coronary artery disease screening with exercise testing may be of value in individuals with at least an intermediate risk for adverse events. 3 As exercise stress testing has been found to be most useful for predicting fatal, rather than nonfatal events, 1 and as allcause mortality is arguably the end point at lowest risk for ascertainment bias, 2,4,5 interest has focused on the ability of the exercise test to predict mortality risk in asymptomatic population-based cohorts. 6-8 Population-based cohorts have also been used to derive risk stratification schemes 9,10 with varying end points including hard coronary events (defined as coronary death or nonfatal myocardial infarction) 9 and cardiovascular death. 10 What is not clear is whether existing stratification schemes, such as the Framingham Risk Score 9,11,12 and the Context The usefulness of exercise stress test results and global cardiovascular risk systems for predicting all-cause mortality in asymptomatic individuals seen in clinical settings is unclear. Objectives To determine the validity for prediction of all-cause mortality of the Framingham Risk Score and of a recently described European global scoring system Systematic Coronary Risk Evaluation (SCORE) for cardiovascular mortality among asymptomatic individuals evaluated in a clinical setting and to determine the potential prognostic value of exercise stress testing once these baseline risks are known. Design, Setting, and Participants Prospective cohort study of 3554 asymptomatic adults between the ages of 50 and 75 years who underwent exercise stress testing as part of an executive health program between October 1990 and December 2002; participants were followed up for a mean of 8 years. Main Outcome Measures Global risk based on the Framingham Risk Score and the European SCORE. Prospectively recorded exercise stress test result abnormalities included impaired physical fitness, abnormal heart rate recovery, ventricular ectopy, and ST-segment abnormalities. The primary end point was all-cause mortality. Results There were 114 deaths. The c-index, which corresponds to receiver operating characteristic curve values, and the Akaike Information Criteria found that the European SCORE was superior to the Framingham Risk Score in estimating global mortality risk. In a multivariable model, independent predictors of death were a higher SCORE (for 1% predicted increase in absolute risk, relative risk [RR], 1.07; 95% confidence interval [CI], ; P.001), impaired functional capacity (RR, 2.95; 95% CI, ; P.001), and an abnormal heart rate recovery (RR, 1.59; 95%, ; P=.03). ST-segment depression did not predict mortality. Among patients in the highest tertile from the SCORE, an abnormal exercise stress test result, defined as either impaired functional capacity or an abnormal heart rate recovery, identified a mortality risk of more than 1% per year. Conclusion Exercise stress testing when combined with the European global risk SCORE may be useful for stratifying risk in asymptomatic individuals in a comprehensive executive health screening program. JAMA. 2004;292: more recently described European Systematic Coronary Risk Evaluation (SCORE), 10 can be used to identify individuals seen in clinical practice for whom stress testing can identify increased risk for all-cause mortality. We sought to determine whether combining the results from the Framingham Risk Score or the European SCORE with exercise stress testing data could help to predict all-cause mortality in an asymptomatic clinical outpatient population. Author Affiliations: Departments of Medicine (Drs Aktas and Ozduran), General Internal Medicine (Dr Lang), and Cardiovascular Medicine (Ms Pothier and Dr Lauer), Cleveland Clinic Foundation, Cleveland, Ohio. Corresponding Author: Michael S. Lauer, MD, Desk F25, Cleveland Clinic Foundation, 9500 Euclid Ave, Cleveland, OH (Lauerm@ccf.org) JAMA, September 22/29, 2004 Vol 292, No. 12 (Reprinted) 2004 American Medical Association. All rights reserved.

2 METHODS Patient Population The study population consisted of consecutive adults between the ages of 50 and 75 years who underwent routine symptom-limited exercise stress testing as part of an executive physical in the Section of Preventive Medicine at the Cleveland Clinic Foundation in Cleveland, Ohio, between October 1990 and December Participants were asymptomatic, self-referred, and were evaluated by an internist in an outpatient ambulatory setting. This evaluation included a thorough history, complete physical examination, and blood work including complete lipid panel and fasting blood glucose. Race was determined by self-report at the time of registration to the Cleveland Clinic. As part of the evaluation, all patients also underwent routine screening exercise stress testing regardless of baseline risk. No patients had known heart disease, peripheral vascular disease, history of stroke or symptomatic cerebral ischemia, symptoms suggestive of cardiac disease, uninterpretable ST segments due to left bundle-branch block, digoxin use, pre-excitation syndrome, left ventricular hypertrophy, or more than 1 mm of resting ST-segment depression, and no patients were undergoing concurrent imaging studies. Patients were also excluded if a valid Social Security number was not available or if they lived outside the United States. Performance of research based on our electronic stress test registry and the Social Security Death Index was approved by the institutional review board at the Cleveland Clinic Foundation. The requirement for written informed consent was waived. Of the 3554 eligible participants, 2140 (60%) were included in a previous article that focused on heart rate recovery as a predictor of mortality. 13 Clinical Data Methods of prospective data collection, data definitions, quality control, and online computer recording in our stress test laboratory have been described in detail elsewhere Prior to exercise stress testing, each patient participated in a structured complete history including review of available medical records to document symptoms, past medical history, medication use, cardiac risk factors, and prior cardiac and noncardiac diagnoses. Diabetes was defined as a documented prescription of a diabetic diet, use of insulin or other medications used to control blood glucose, or a fasting blood glucose level higher than 125 mg/dl (6.94 mmol/ L). Fasting total cholesterol, highdensity lipoprotein cholesterol, lowdensity lipoprotein cholesterol, and triglyceride levels were obtained using an enzymatic assay with an automated chemistry analyzer. Appropriate standardization of the assays was performed in the time interval of the study in compliance with qualitycontrol measures. Exercise Testing Exercise testing in our laboratory has been described in detail elsewhere Briefly, all patients underwent symptom-limited exercise treadmill testing primarily according to Bruce or modified Bruce protocols. 19 During each stage of exercise and recovery the following characteristics were recorded: symptoms, blood pressure, heart rate, cardiac rhythm, and exercise work load in metabolic equivalents (METs; 1 MET=3.5 ml/kg per minute of oxygen consumption). Patients were encouraged to continue exercising until symptoms limited further continuation even after reaching 85% of maximum predicted heart rate. An ischemic ST abnormality, which was assessed visually by 2 independent readers, was defined as a 1-mm horizontal or downsloping ST-segment depression occurring 80 milliseconds after the J point; ST-segment depression had to be noted in at least 3 consecutive beats in at least 2 contiguous leads. Heart rate recovery was defined as the difference between heart rate at peak exercise and 1 minute later, that is, 1 minute into a 2-minute upright cool down period in which patients walked 1.5 mph at a grade of 2.5%. Failure of the heart rate to decrease by more than 12 beats during the first minute following exercise constituted abnormal heart rate recovery. 13,15 The cut-off value of 12/ min was recently validated in a cohort of more than patients based on random forest methods. 20 Functional capacity was considered impaired if the exercise workload was less than the 25th percentile value for sex (men 9.5 METs; women 7.5 METs). Frequent ventricular ectopic activity during the first 5 minutes of recovery was defined as the presence of 7 or more ventricular premature beats per minute, ventricular bigeminy, ventricular trigeminy, ventricular couplets, ventricular triplets, sustained or nonsustained ventricular tachycardia, ventricular flutter, torsade de pointes, or ventricular fibrillation. 21 European SCORE and Framingham Risk Score The European SCORE was developed by the European Society of Cardiology in 2003 as an estimation of 10- year risk of fatal cardiovascular disease in European nations to be used in primary prevention. The SCORE was designed to estimate total cardiovascular risk rather than risk of coronary heart disease alone by totaling calculated coronary and noncoronary components. 10 Citing difficulties in attaining Framingham study end points, the authors used only fatal atherosclerotic cardiovascular events in the total risk estimation for the SCORE. In calculating the SCORE for a 10-year risk estimate, the variables used included age, sex, total cholesterol, systolic blood pressure, and smoking status. The original formula allows for the use of coefficients representing high- and lowrisk cohorts as defined by geographic location. In the current study, we used high-risk coefficients because the lowrisk coefficients were based on Southern European cohorts known to have substantially lower rates of coronary artery disease events than those seen in North America. 22,23 The Framingham Risk Score 9,11 was derived from the Framingham Heart 2004 American Medical Association. All rights reserved. (Reprinted) JAMA, September 22/29, 2004 Vol 292, No

3 Study Cohort and was designed to predict 10-year risk of hard coronary events, including mortality due to coronary heart disease and nonfatal myocardial infarction. The Framingham Risk Score is calculated by taking into account age, sex, smoking status, total cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, and diabetes. 9,11,12 For this analysis, we used the sexspecific Framingham equations of D Agostino et al 9 because they include diabetes. End Points The primary end point of this study was all-cause mortality during a mean of 8 years of follow-up. The censoring date was February 28, Death of a patient was determined using the Social Security Death Master Files. 24 All-cause mortality was chosen as an objective, easily ascertainable, and unbiased end point. 2,4,5 Previous work has shown the high specificity of the Social Security Death Index. 25 Furthermore, this index has been shown to be 97% sensitive when applied to patients in the Cleveland Clinic Stress Laboratory. 13 Statistical Analyses Survival analyses were performed by construction of Kaplan-Meier plots, calculation of log-rank statistics, and by Cox proportional hazards regression. The proportional hazards assumption was confirmed by inspection of curves relating the logarithm of cumulative hazard to the logarithm of follow-up time. To compare the ability of the Framingham Risk Score and the European SCORE to predict death, Akaike Information Criteria and weights were calculated for separate nonnested models. 26,27 We also calculated the c- index, corresponding to the receiver operating characteristic curve area, for censored data using a method described by Harrell et al. 28,29 For models that included global risk and exercise variables, we identified the best models (those that considered candidate variables as categorical, continuous, or both) by use of Akaike Information Criteria and weights. 26,27 Differences between baseline and exercise characteristics according to the tertiles of the better global risk stratification scheme were compared using the Kruskal-Wallis or 2 tests as appropriate. A 2-sided P value of less than.05 was considered statistically significant. All analyses were performed using SAS statistical software (version 8.2, SAS Institute Inc, Cary, NC), except for the calculation of the c-index, which was performed using S-PLUS software (version 6.2, Mathsoft Inc, Seattle, Wash). 30 Predicted mortality risk based on Cox models was calculated using the baseline option of the PROC PHREG of the SAS statistical package. RESULTS Global Risk Scores and Prediction of Death The study cohort consisted of 3554 patients comprised of 2871 (81%) men and 683 (19%) women. There were 114 deaths; only 12 occurred among women. The Framingham Risk Score was not predictive of death (for 1% predicted increase in absolute risk of hard coronary events, relative risk [RR] of death, 1.07; 95% confidence interval [CI], ; P=.36), whereas European SCORE was strongly predictive (RR of death, 1.09; 95% CI, ; P.001 for each 1% increase in predicted absolute risk of cardiovascular death). By Akaike weights, there was a higher than 99% probability that the European SCORE is a better predictor of death. Similarly, the c-index corresponding to the receiver operating characteristic curve area was higher for the assessment using the European SCORE (0.73) compared with the Framingham Risk Score (0.57). Given the superiority of the European SCORE for predicting mortality, all further analyses were based on it. Baseline Characteristics Baseline characteristics of the study population by tertile from the SCORE are shown in TABLE 1. Patients in the higher tertiles were more likely to be older men with more adverse risk profiles. These patients had higher systolic and diastolic blood pressures, were more likely to have higher total cholesterol and low-density lipoprotein cholesterol levels; lower high-density lipoprotein cholesterol levels; to be smokers; and to have non insulin-dependent diabetes mellitus. Patients in the higher tertiles were also more likely to be taking aspirin, -blockers, nondihydropyridine calcium channel blockers, angiotensinconverting enzyme inhibitors, and/or thiazide diuretics. Exercise Characteristics Exercise characteristics by tertile from the SCORE are summarized in TABLE 2. Individuals in the higher tertiles from the SCORE were more likely to have impaired fitness for age and sex and also more likely to have an abnormal heart rate recovery or frequent ventricular ectopy during recovery. However, there was no association between the SCORE tertile and the occurrence of ischemic ST-segment changes, except when we specifically focused on marked STsegment depression of 2 mm or higher. Only 1 patient, who was in the third tertile from the SCORE, had exerciseinduced angina. Predictors of Death During a mean of 8 years of follow-up, 114 (3%) patients died. Seventy-six (6%) deaths occurred among patients in the third tertile from the SCORE compared with a total of 38 (2%) deaths among patients in the first and second tertiles combined together. Patients in the third tertile from the SCORE had a RR of death nearly 4 times that of patients in the first tertile (TABLE 3). In unadjusted proportional hazards models, the exercise stress test variables that predicted death included abnormal heart rate recovery, impaired fitness for age and sex, and frequent ventricular ectopic activity in recovery (Table 3). Ischemic ST-segment changes were not predictive of mortality, which was true even when we focused on marked ( 2 mm) ST-segment depression. In a subset analyses of patients in the highest SCORE tertile (patients with the highest global risk for death), ische JAMA, September 22/29, 2004 Vol 292, No. 12 (Reprinted) 2004 American Medical Association. All rights reserved.

4 Table 1. Baseline Characteristics According to Tertiles of European SCORE* Characteristic First Tertile (n = 1184) Second Tertile (n = 1185) Third Tertile (n = 1185) P Value Predicted risk of cardiovascular death 0.14 ( ) 3.0 ( ) 6.6 ( ).001 in 10 y (median, 25th-75th percentiles), % Framingham Risk Score (median, 25th-75th percentiles) 3.9 ( ) 3.9 ( ) 4.6 ( ).001 Age, mean (SD), y 53 (3) 56 (4) 62 (6).001 Men, No. (%) 672 (57) 1084 (92) 1115 (94).001 Blacks, No. (%) 23 (2) 25 (2) 15 (1).26 Body mass index, mean (SD) 27 (5) 28 (4) 28 (4).001 Cholesterol, mean (SD), mg/dl Total 206 (33) 216 (33) 223 (36).001 High-density lipoprotein 52 (16) 47 (13) 47 (13).001 Low-density lipoprotein 130 (31) 143 (31) 148 (32).001 Triglycerides, mean (SD), mg/dl 122 (80) 138 (87) 145 (90).001 Blood pressure, mean (SD), mm Hg Systolic 118 (13) 127 (14) 139 (17).001 Diastolic 81 (9) 86 (10) 89 (10).001 Current smoker, No. (%) 40 (3) 100 (8) 242 (20).001 Diabetes, No. (%) 21 (2) 25 (2) 43 (4).009 Glucose level, mean (SD), mg/dl 94 (16) 99 (20) 101 (23).001 Medication use, No. (%) Aspirin 109 (9) 131 (11) 167 (14).001 -Blockers 37 (3) 47 (4) 73 (6).001 Nondihydropyridine calcium channel blocker 17 (1) 21 (2) 48 (4).001 ACE inhibitors 34 (3) 73 (6) 89 (8).001 Statin 59 (5) 58 (5) 60 (5).98 Thiazide 29 (3) 41 (4) 70 (6).001 Abbreviations: ACE, angiotensin-converting enzyme; SCORE, Systematic Coronary Risk Evaluation. SI conversion factors: To convert low-density lipoprotein, high-density lipoprotein, and total cholesterol to mmol/l, multiply by ; glucose to mmol/l, multiply by ; triglycerides to mmol/l, multiply by *The ranges for the SCORE in each of the tertiles were less than 2.2, 2.2 to 4.2, and greater than 4.2. Calculated as weight in kilograms divided by the square of height in meters. Defined as fasting glucose level of 125 mg/dl (6.94 mmol/l) or higher or use of diabetic medications (including insulin). mic ST-segment changes again failed to predict death (HR, 1.18; 95% CI, ; P=.63). There were no interactions between abnormal heart rate recovery and impaired physical fitness for prediction of death. In a multivariable model, independent predictors of death were a higher SCORE (for 1% predicted increase in absolute risk, RR, 1.07; 95% CI, ; P.001), impaired functional capacity (RR, 2.95; 95% CI, ; P.001), and an abnormal heart rate recovery (RR, 1.59; 95% CI, ; P=.03). ST-segment changes did not predict risk of death, even when focusing on marked ( 2 mm) STsegment depression. Addition of exercise variables to the SCORE increased the c-index from 0.73 to Variables that were not significant predictors of death in this model included diabetes, body mass index, levels of Table 2. Exercise Stress Test Characteristics According to Tertiles of European SCORE* Characteristic First Tertile (n = 1184) Second Tertile (n = 1185) Third Tertile (n = 1185) Mean (SD) No. (%) Mean (SD) No. (%) Mean (SD) No. (%) P Value Metabolic equivalents Men 11 (2) 672 (57) 11 (2) 1084 (91) 10 (2) 1115 (94).001 Women 9 (2) 512 (43) 8 (2) 101 (9) 7 (2) 70 (6).001 Peak heart rate 166 (12) 164 (13) 155 (15).001 Impaired functional 162 (14) 243 (21) 499 (42).001 capacity for age/sex Ischemic ST-segment changes 1mmand 2mm 81 (7) 56 (5) 78 (7).06 2 mm 39 (3) 59 (5) 58 (5).08 Recovery Abnormal heart rate 150 (13) 155 (13) 244 (21).001 Frequent ventricular ectopy Abbreviation: SCORE, Systematic Coronary Risk Evaluation. *One patient had angina, which was not test terminating. 19 (2) 38 (3) 50 (4).001 triglycerides and high-density lipoprotein cholesterol, black race, and use of medications including -blockers, aspirin, angiotensin-converting enzyme inhibitors, nondihydropyridine calcium blockers, statins, and thiazide di American Medical Association. All rights reserved. (Reprinted) JAMA, September 22/29, 2004 Vol 292, No

5 Table 3. Univariable Analyses of Mortality Figure 1. Combined 9-Year All-Cause Mortality Rate According to SCORE and Exercise Test Results 8-Year Death Rate, % Exercise Test Normal Abnormal 10 P = P<.001 P = No Tertile of Predicted Risk by European SCORE An exercise test was considered abnormal if functional capacity was impaired (lowest sex-specific quartile) or if heart rate recovery was abnormal ( 12 beats). SCORE indicates Systematic Coronary Risk Evaluation. uretics. This model, which considered SCORE as a continuous variable and functional capacity and heart rate recovery as dichotomous variables was superior by Akaike weights to a purely continuous model. In the purely continous model, death was predicted by both results from the SCORE (for 1% predicted increase in absolute risk, RR, 1.07; 95% CI, ; P.001) and No. of Participants No. (%) of Deaths RR (95% CI) P Value Tertile of SCORE First (1) 1.00 Second (2) 1.47 ( ).25 Third (6) 4.39 ( ).001 Impaired functional capacity for age/sex Absent (2) 1.00 Present (7) 4.15 ( ).001 Ischemic ST-segment change None (3) mm (4) 1.16 ( ).68 2 mm (3) 0.87 ( ).78 Recovery Abnormal heart rate Absent (3) 1.00 Present (7) 2.66 ( ).001 Frequent ventricular ectopy Absent (3) 1.00 Present (7) 2.52 ( ).02 Abbreviations: CI, confidence interval; RR, relative risk, SCORE, Systematic Coronary Risk Evaluation. exercise capacity in METs (RR, 1.29; 95% CI, ; P.001 for each 1 MET decrease), but not by heart rate recovery. SCORE, Exercise Stress Testing, and Risk of Death FIGURE 1 shows the 8-year mortality rate according to tertiles from the SCORE when combined with a normal or abnormal exercise stress test result. An abnormal stress test result was defined as impaired functional capacity for age and sex or an abnormal heart rate recovery. In the highest tertile from the SCORE, an abnormal exercise stress test result identified a mortality risk of more than 1% per year. In the 2 lowest tertiles from the SCORE, the absolute mortality risk was low irrespective of exercise stress test results. FIGURE 2 shows the predicted 8-year mortality rate from the SCORE, exercise capacity, and heart rate recovery, which were considered dichotomous variables corresponding to the multivariable model with the best Akaike Information Criteria value. A mortality rate exceeding 1% per year was noted when the results from the European SCORE were relatively high and there was both impaired functional capacity and an abnormal heart rate recovery. Follow-up Procedures A total of 190 patients (5%) underwent either nuclear or echocardiographic stress imaging testing within 3 months of their stress test; only 16 (8%) of these had a finding of either ischemia or scar. Coronary angiography was performed in 21 patients (0.6%) within 3 months; 3 underwent coronary artery bypass graft surgery and 1 underwent percutaneous revascularization within 1 month of angiography. Two of these patients had greater than 70% stenosis of the left main arterial trunk. COMMENT In a large cohort of asymptomatic patients undergoing a comprehensive executive health evaluation, we found that European SCORE predicted absolute risk of all-cause mortality better than the Framingham Risk Score. Exercise stress test results also predicted mortality risk, particularly when attention was focused on exercise capacity and heart rate recovery. While previous studies of similar cohorts have shown that exercise testing can predict mortality, 31 by taking into account a quantitative assessment of baseline risk, we show that exercise testing only predicted a clinically meaningful increase in mortality risk when applied to those individuals who had a higher baseline risk as assessed by the SCORE. The use of composite risk scores for prediction of cardiovascular risk has gained popularity in the medical literature, but their practical use has been called into question. 32 In the United States, the Framingham Risk Score has been incorporated into guidelines for cholesterol management, 11 aspirin use, 12 and consideration of screening studies. 3 The Framingham Risk Score 33 has been validated in some populations 9 but not in others. 34 Criticisms of the Framingham Risk Score include use of hard to define nonfatal events and the relatively small and homogeneous sample from which it was derived. 10 Because our analysis focused on mortality only, this may explain why the Framingham Risk Score did not perform well in our population JAMA, September 22/29, 2004 Vol 292, No. 12 (Reprinted) 2004 American Medical Association. All rights reserved.

6 The European SCORE, which we used in the current study, was derived from a large sample, focused on fatal cardiovascular events, and took into account regional differences in risk. To the best of our knowledge, it has not been previously applied to a North American cohort. Although the SCORE end point was cardiovascular mortality (an end point that is subject to ascertainment bias 2,4 ) and was based on European country-specific calculations, we found a remarkably strong association between an increasing SCORE and an increased risk of all-cause mortality. This may be because the vast majority of the deaths in our cohort were of a cardiovascular nature, although assessment of cause of death is inherently problematic. 35 The value of exercise stress testing for cardiovascular screening has been questioned recent recommendations argue neither for nor against this practice. 2 One major problem with using exercise stress testing for screening is the poor accuracy of ST-segment changes for diagnosing coronary artery disease. 36 More recently, we and others have focused on other measures that have far greater prognostic value, particularly exercise capacity, 16 heart rate recovery, 13,15 and ventricular ectopy. 21 In the current study, there was no association between ST-segment changes and mortality, whereas exercise capacity and heart rate recovery were predictive, as we have previously shown. Why ST-segment depression was not predictive of mortality is unclear, although this finding is consistent with those reported by others. 37 A major new finding in the current study is that the European SCORE can be used to identify individuals for whom an abnormal exercise stress test foretells a clinically meaningful increase in mortality risk. An abnormal exercise stress test result among patients in the highest tertile of the SCORE predicted a mortality rate of more than 1% per year, whereas the absolute risk of death was quite low in the 2 lower tertiles, irrespective of the exercise stress test result. Thus, if a clinician contemplates Figure 2. Eight-Year Predicted All-Cause Mortality Rates 8-Year Death Rate, % referring a patient for a screening exercise stress test, it might be reasonable to limit such referrals to those with a higher risk as determined by the tertile from the SCORE. There are a number of important limitations that must be considered. First, there are no randomized trial data showing that any type of cardiovascular screening test reduces mortality. It is not clear that knowing about a patient s increased risk, for example by combining SCORE and exercise data, will lead to any kind of improved outcome. Nonetheless, our study may have practical utility in that physicians can confidently use the SCORE to identify patients for whom exercise stress testing will have minimal if any clinical value. Furthermore, current guidelines suggest that asymptomatic individuals who are identified as being at potentially high risk should be considered candidates for particularly aggressive medical management. 38 Second, we used all-cause mortality as our end point and did not attempt to ascertain a cause of death. The problems with using cause-related end points as opposed to the hard, objective, and unbiased end point of all-cause mortality have been discussed elsewhere. 4,5 Third, the study cohort came from an executive health clinic in a major academic medical center. Our patients likely had ready access 4 Abnormal Heart Rate Recovery Normal Heart Rate Recovery Abnormal Heart Rate Recovery Normal Heart Rate Recovery European SCORE 10-Year Risk for Fatal Cardiovascular Disease, % Impaired Functional Capacity Preserved Functional Capacity According to European SCORE, functional capacity, the presence or absence of an abnormal heart rate recovery, and mortality rates. SCORE indicates Systematic Coronary Risk Evaluation. to health care, a higher socioeconomic status, and a high level of interest in their long-term health. Whether our results may be generalized to other populations will require further study. Fourth, we did include patients with diabetes, but this was not part of the SCORE. In a multivariable model that included SCORE, exercise variables, and diabetes, only SCORE and exercise variables were predictive of increased risk. Fifth, we did not have detailed data on medication use or procedures after exercise stress testing. Finally, most of the patients in our study were men, meaning that our results may not necessarily apply to women. Stratified analyses by sex could not be done because there were only 12 deaths among women. Of note, recent work from a populationbased cohort found that exercise capacity and heart rate recovery predicted allcause and cardiovascular mortality in women even after accounting for standard risk factors. 6,7 There continues to be insufficient evidence to support exercise stress testing as a screening tool in asymptomatic healthy individuals. Our results should not be interpreted to advocate the use of exercise stress testing to this end. However, the use of a global risk stratification scheme such as the European SCORE may be helpful in referral to and interpretation of exercise stress testing American Medical Association. All rights reserved. (Reprinted) JAMA, September 22/29, 2004 Vol 292, No

7 Author Contributions: Dr Lauer had full access to all of the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Aktas, Lang, Lauer. Acquisition of data: Aktas, Ozduran, Pothier, Lang. Analysis and interpretation of data: Aktas, Pothier, Lauer. Drafting of the manuscript: Aktas, Lauer. Critical revision of the manuscript for important intellectual content: Aktas, Ozduran, Pothier, Lang, Lauer. Statistical analysis: Pothier, Lauer. Obtained funding: Lauer. Administrative, technical, or material support: Aktas, Ozduran, Lang, Lauer. Study supervision: Lauer. Funding/Support: Dr Lauer and Ms Pothier were supported by research grant HL from the National Heart, Lung, and Blood Institute. Role of the Sponsor: The National Heart, Lung, and Blood Institute played no direct role in the study design, conduct, data management, data analysis, manuscript preparation, review, or authorization for submission. Disclaimer: Dr Lauer was not involved in the editorial evaluation or decision making regarding this article. REFERENCES 1. Gibbons RJ, Balady GJ, Bricker JT, et al. ACC/ AHA 2002 guideline update for exercise testing: summary article: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2002;106: Fowler-Brown A, Pignone M, Pletcher M, Tice JA, Sutton SF, Lohr KN. Exercise tolerance testing to screen for coronary heart disease: a systematic review for the technical support for the US Preventive Services Task Force. Ann Intern Med. 2004;140:W9-W Greenland P, Smith SC Jr, Grundy SM. Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests. Circulation. 2001;104: Lauer MS, Blackstone EH, Young JB, Topol EJ. Cause of death in clinical research: time for a reassessment? J Am Coll Cardiol. 1999;34: Gottlieb SS. Dead is dead artificial definitions are no substitute. Lancet. 1997;349: Mora S, Redberg RF, Cui Y, et al. Ability of exercise testing to predict cardiovascular and all-cause death in asymptomatic women: a 20-year follow-up of the lipid research clinics prevalence study. JAMA. 2003; 290: Gulati M, Pandey DK, Arnsdorf MF, et al. Exercise capacity and the risk of death in women: the St James Women Take Heart Project. Circulation. 2003;108: Lauer MS, Okin PM, Larson MG, Evans JC, Levy D. Impaired heart rate response to graded exercise: prognostic implications of chronotropic incompetence in the Framingham Heart Study. Circulation. 1996;93: D Agostino RB Sr, Grundy S, Sullivan LM, Wilson P. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001;286: Conroy RM, Pyorala K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003;24: Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002;106: Lauer MS. Clinical practice: aspirin for primary prevention of coronary events. N Engl J Med. 2002;346: 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: 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: 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: 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 singlephoton emission computed tomography: report of 3,400 patients from a single center. J Am Coll Cardiol. 1997;30: 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: 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: Fletcher GF, Balady G, Froelicher VF, Hartley LH, Haskell WL, Pollock ML. Exercise standards: a statement for healthcare professionals from the American Heart Association. Circulation. 1995;91: Ishwaran H, Blackstone EH, Pothier CE, Lauer MS. Relative risk forests for exercise heart rate recovery as a predictor of mortality. J Am Stat Assoc. In press. 21. Frolkis JP, Pothier CE, Blackstone EH, Lauer MS. Frequent ventricular ectopy after exercise as a predictor of death. N Engl J Med. 2003;348: Haq IU, Ramsay LE, Yeo WW, Jackson PR, Wallis EJ. Is the Framingham risk function valid for northern European populations? a comparison of methods for estimating absolute coronary risk in high risk men. Heart. 1999;81: Menotti A, Lanti M, Puddu PE, Kromhout D. Coronary heart disease incidence in northern and southern European populations: a reanalysis of the Seven Countries Study for a European coronary risk chart. Heart. 2000;84: Boyle CA, Decoufle P. National sources of vital status information: extent of coverage and possible selectivity in reporting. Am J Epidemiol. 1990;131: Newman TB, Brown AN. Use of commercial record linkage software and vital statistics to identify patient deaths. J Am Med Inform Assoc. 1997;4: Akaike H. A Bayesian analysis of the minimum AIC procedure. Ann Inst Stat Math. 1978;30: Burnham KP, Anderson DR. Model Selection and Inference: A Practical Information-Theoretic Approach. New York, NY: Springer; Harrell FE Jr, Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med. 1984;3: Harrell FE Jr, Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests. JAMA. 1982; 247: Harrell FE Jr. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York, NY: Springer- Verlag Inc; 2001;493: Wei M, Kampert JB, Barlow CE, et al. Relationship between low cardiorespiratory fitness and mortality in normal-weight, overweight, and obese men. JAMA. 1999;282: Topol EJ, Lauer MS. The rudimentary phase of personalised medicine: coronary risk scores. Lancet. 2003; 362: Wilson PW, D Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation. 1998; 97: Prediction of mortality from coronary heart disease among diverse populations: is there a common predictive function? Heart. 2002;88: Lloyd-Jones DM, Martin DO, Larson MG, Levy D. Accuracy of death certificates for coding coronary heart disease as the cause of death. Ann Intern Med. 1998;129: Froelicher VF, Lehmann KG, Thomas R, et al, for the Veterans Affairs Cooperative Study in Health Services #016 (QUEXTA) Study Group. The electrocardiographic exercise test in a population with reduced workup bias: diagnostic performance, computerized interpretation, and multivariable prediction. Ann Intern Med. 1998;128(12 pt 1): Goraya TY, Jacobsen SJ, Pellikka PA, et al. Prognostic value of treadmill exercise testing in elderly persons. Ann Intern Med. 2000;132: Grundy SM, Pasternak R, Greenland P, Smith S Jr, Fuster V. Assessment of cardiovascular risk by use of multiple-risk-factor assessment equations: a statement for healthcare professionals from the American Heart Association and the American College of Cardiology. Circulation. 1999;100: JAMA, September 22/29, 2004 Vol 292, No. 12 (Reprinted) 2004 American Medical Association. All rights reserved.