Angela Fowler-Brown, M.D.a; Michael Pignone, M.D., M.P.H.a; Mark Pletcher, M.D., M.P.H.b; Jeffrey A. Tice, M.D.c; Sonya F. Sutton, B.S.P.H.d; and Kathleen N. Lohr, Ph.D.d
The authors of this article are responsible for its contents, including any clinical or treatment recommendations. No statement in this article should be construed as an official position from the Agency for Healthcare Research and Quality, or the U.S. Department of Health and Human Services.
Address correspondence to: Michael Pignone, M.D., M.P.H.; 5039 Old Clinic Building; UNC Hospitals; Chapel Hill, NC 27599-7110; E-mail: pignone@med.unc.edu
Select for copyright and reprint information.
The summaries of the evidence briefly present evidence of effectiveness for preventive health services used in primary care clinical settings, including screening tests, counseling, and chemoprevention. They summarize the more detailed Systematic Evidence Reviews, which are used by the U.S. Preventive Services Task Force (USPSTF) to make recommendations.
Abstract
Introduction
Methods
Results
Discussion
Acknowledgments
References
Notes
Background: Coronary heart disease is the leading cause of morbidity and mortality in the United States. Exercise tolerance testing has been proposed as a means of better identifying asymptomatic patients at high risk for coronary heart disease events.
Purpose: To review the evidence on the use of exercise tolerance testing to screen adults with no history of cardiovascular disease for coronary heart disease.
Data Sources: The MEDLINE® database from 1966 through February 2003, hand-searching of bibliographies, and expert input.
Study Selection: Eligible studies evaluated the benefits or harms of exercise tolerance testing when added to traditional risk assessment for adults with no known history of cardiovascular events.
Data Extraction: One reviewer extracted information from eligible articles into evidence tables, and another reviewer checked the tables. Disagreements were resolved by consensus.
Data Synthesis: No study has directly examined the effect of screening asymptomatic patients with exercise tolerance testing on coronary heart disease outcomes or risk-reducing behaviors or therapies. Multiple cohort studies demonstrate that screening exercise tolerance testing identifies a small proportion of asymptomatic persons (up to 2.7 percent of those screened) with severe coronary artery obstruction who may benefit from revascularization. Several large prospective cohort studies, conducted principally in middle-aged men, suggest that exercise tolerance testing can provide independent prognostic information about the risk for future coronary heart disease events (relative risk with abnormal exercise tolerance testing, 2.0 to 5.0).
However, when the risk for coronary heart disease events is low, most positive findings will be false and may result in unnecessary further testing or worry. The risk level at which the benefits of additional prognostic information outweigh the harms of false-positive results is unclear and requires further study.
Conclusions: Although screening exercise tolerance testing detects severe coronary artery obstruction in a small proportion of persons screened and can provide independent prognostic information about the risk for coronary heart disease events, the effect of this information on clinical management and disease outcomes in asymptomatic patients is unclear.
Coronary heart disease is the leading cause of death in the United States. Each year, more than 1 million Americans experience nonfatal or fatal myocardial infarction or sudden death from coronary heart disease. Coronary heart disease can also present as angina, but only 20 percent of acute coronary events are preceded by long-standing angina.1 An estimated 1 to 2 million middle-aged men have asymptomatic but physiologically significant coronary artery obstruction, which puts them at increased risk for coronary heart disease events.2,3 The economic burden of coronary heart disease is also substantial. The direct and indirect costs of coronary heart disease in the United States are projected to total $129.9 billion for 2003.1 The clinical and economic impact of coronary heart disease is the basis for considerable public health interest in the development of effective strategies to reduce the incidence of coronary heart disease events.
In 1996, the U.S. Preventive Services Task Force considered use of resting electrocardiography or exercise tolerance testing to detect asymptomatic coronary artery disease and prevent coronary heart disease events.4 The Task Force found insufficient evidence to recommend for or against using these tests to screen middle-aged and older men and women. They recommended against screening children, adolescents, or young adults.
To update the evidence review and recommendations on screening for asymptomatic coronary artery disease, the Task Force and the Agency for Healthcare Research and Quality (AHRQ) requested that the RTI International-University of North Carolina Evidence-based Practice Center perform an updated evidence review beginning in 2001.
The complete review considers resting electrocardiography, exercise tolerance testing, and electron-beam computed tomography for coronary calcium and is available at www.ahrq.gov/clinic/serfiles.htm.5 This article describes the findings on exercise tolerance testing only. The recommendations and rationale of the Task Force on screening for asymptomatic coronary artery disease are available at www.ahrq.gov/clinic/3rduspstf/chd/chdrs.htm.6
Clinicians can use 2 general approaches to prevention of morbidity and mortality from coronary heart disease. The first approach involves screening for and treating the traditional modifiable risk factors for coronary heart disease, such as hypertension, abnormal blood levels of lipids, diabetes, cigarette smoking, physical inactivity, and diet. Such an approach may incorporate explicit calculations of the patient's risk for coronary heart disease events by using risk prediction equations derived from the Framingham Heart Study or other cohort studies.7 The second strategy involves supplementation of screening based on traditional risk factors with additional tests to provide further information about future risk for coronary heart disease or to detect severe blockages of the coronary arteries that might warrant treatment.
Detection of increased risk for future coronary heart disease events may lead to intensified use of risk-reducing treatments. Some risk-reducing treatments are directed at traditional risk factors (for example, therapy with statins for hyperlipidemia), whereas others are not (for example, aspirin therapy). Revascularization by using coronary artery bypass graft surgery or percutaneous coronary intervention seeks to treat blockages of the coronary arteries. Whether revascularization will reduce the risk for coronary heart disease events in persons identified by screening is unknown.
Exercise tolerance testing is widely used as a diagnostic test in the initial evaluation of patients with symptoms suggestive of myocardial ischemia and in persons with previously recognized coronary heart disease. Although exercise tolerance testing has been applied and studied as a screening or prognostic test in asymptomatic persons, its utility in this group is controversial. The best measure of the value of screening exercise tolerance testing would come from studies that examined whether patients randomly assigned to undergo such tests had fewer coronary heart disease events or received more appropriate risk-reducing therapies than did patients assigned to receive treatments after standard risk factor assessment.
Such direct evidence is not available. However, indirect evidence suggests that screening exercise tolerance testing may be helpful in guiding medical management.8 In the Multiple Risk Factor Intervention Trial Research study, high-risk male participants were randomly assigned to receive a multimodal intervention to reduce cardiovascular risk or usual care. Among participants with an abnormal baseline result on exercise tolerance testing, those who received the intervention had a significantly lower rate of mortality from coronary heart disease during followup than did the group that received usual care. No effect was seen among men with a normal baseline result on exercise tolerance testing. It is not clear from the report of this post-hoc analysis whether the cardiovascular risk profiles of participants with an abnormal result on exercise tolerance testing at baseline differed significantly from those of participants with a normal result.
Because direct evidence on possible benefits of screening exercise tolerance testing is lacking, we used data from observational cohort studies to examine whether screening exercise tolerance testing could detect clinically significant asymptomatic obstructions of the coronary arteries or provide greater independent prognostic information about the risk for future coronary heart disease events than would be obtained solely by standard history, physical examination, and measurement of traditional risk factors. We also sought information about harms of screening, including the likelihood of false-positive results and the effect of labeling a person as being "at high risk."
To identify the relevant literature, we searched the MEDLINE® database from 1966 through February 2003 by using the exploded Medical Subject Headings coronary heart disease, exercise test, and mass screening and the keywords asymptomatic and screening. We limited the search to English-language articles on human subjects. To supplement our literature searches, we hand-searched the bibliographies of key articles, used other recent systematic reviews when available, and included references provided by expert reviewers that had not been identified by other mechanisms.
Two reviewers examined the abstracts of the articles identified in the initial MEDLINE® search and selected a subset for a full-text review. The same reviewers examined the full text of the selected articles to determine final eligibility. One reviewer extracted information from eligible articles into evidence tables, and another reviewer checked the tables. They resolved disagreements by consensus.
To be eligible, studies had to have been performed in participants with no history of cardiovascular disease or provide subset analysis for this group. Included studies on the detection of severe coronary artery obstruction reported the total number of persons screened to obtain the sample of persons with an abnormal result on exercise tolerance testing and the proportion of persons who were found to have coronary heart disease on angiography. The yield of exercise tolerance testing screening was determined by dividing the number of participants found to have abnormal results on angiography by the total number screened.
For the prognostic benefit of exercise tolerance testing, included studies reported the independent value of the test for predicting coronary heart disease events. We included studies that examined the prognostic benefit of exercise testing by using several variables, including ST-segment depression, functional capacity, chronotropic incompetence, heart rate recovery, and development of exercise-induced premature ventricular contractions. We also included studies that used nuclear medicine imaging to detect ischemia.
We excluded studies that did not use statistical methods to control for the effect of other risk factors (such as age or systolic blood pressure) on the estimate of the prognostic strength of a positive result of exercise tolerance testing. Table 1 shows information on excluded studies.
The studies used different means of characterizing the prognostic benefit of screening with exercise tolerance testing. Many studies reported outcomes in terms of independent relative risk associated with a positive (versus a negative) screening test. Others used diagnostic test terminology, such as "sensitivity and specificity" or "positive predictive value." In such cases, the terms are used to indicate test accuracy over the entire followup period rather than at 1 point in time.
To assess whether a relationship exists between sensitivity of exercise tolerance testing for future coronary heart disease and duration of followup, we examined the correlation between reported sensitivity and mean duration of followup by using STATA® statistical software, version 7.0 (Stata® Corp., Chicago, Illinois).
We rated the quality of the included articles according to criteria developed by the USPSTF Methods Work Group.9 Tables 3 and 4 show information only from studies judged "good." For the studies shown in Table 2, we considered several factors that affect quality, chiefly the percentage of patients with a positive exercise tolerance testing who underwent catheterization and how completely outcomes were assessed. We used the final set of eligible articles to create evidence tables and produce the larger evidence report, which also included evaluation of resting electrocardiography and electron-beam computed tomography to detect coronary calcium. The full evidence report was subjected to external peer review and revised on the basis of the comments received; we used the revised report as the basis for this article.
This evidence report was funded through a contract to the RTI-University of North Carolina Evidence-based Practice Center from AHRQ. Staff of the funding agency contributed to the study design, reviewed draft and final manuscripts, and made editing suggestions.
We identified 713 articles for review. We reviewed the abstracts and retained 55 articles that examined the diagnostic or prognostic significance of screening with exercise tolerance testing. After full article review, we kept 31 articles representing 29 studies that met the inclusion criteria.10-40 We identified another 11 articles for inclusion through review of reference lists and input of expert reviewers.8,41-50 Table 1 lists articles excluded during review of the full articles and the reason for exclusion.51-74
We found no studies that directly tested whether screening asymptomatic persons with exercise tolerance testing improves coronary heart disease and mortality. Similarly, we found no studies that examined the effect of screening with exercise tolerance testing on the subsequent use of risk-reducing interventions and behaviors. However, we identified fair- or good-quality observational cohort studies of asymptomatic adults that prospectively evaluated the value of exercise tolerance testing in detecting asymptomatic coronary artery obstruction14-18,22,23,25,27,28,30,31,38,75 and predicting future coronary heart disease events, such as angina, myocardial infarction, and sudden death.8,10-13,19-21,26,29,32-36,38-50 We also identified 3 good-quality studies that estimated the cost effectiveness of exercise tolerance testing to identify asymptomatic, severe, prevalent coronary heart disease.24,28,37
We identified 13 studies in 14 articles that examined the utility of exercise tolerance testing to detect asymptomatic coronary artery obstruction (Table 2).14,15,18,22,23,25,27,28,30,31,38,75 In these studies, the prevalence of abnormal exercise tolerance testing, usually defined as exercise-induced ST-segment depression of 1 mm or more, ranged from about 3 percent among aviators who were presumed healthy16 to 29 percent in a sample of diabetic persons in Finland.15,75 A portion of the participants with a positive exercise tolerance testing in each study (1 percent to 60 percent) proceeded to evaluation with cardiac catheterization. Screening with exercise tolerance testing yielded angiographically demonstrable coronary heart disease, usually defined as greater than 50 percent stenosis of a major coronary artery, in a minority of the screened patients.
The yield of screening exercise tolerance testing was greater in higher-risk groups. Five studies in 6 articles evaluated diabetic persons,15,75 those with multiple risk factors,18,31 those with siblings with coronary heart disease17 and those who were prescreened by using a chest pain questionnaire.25 In these studies, the yield of screening for angiographically demonstrable coronary heart disease ranged from 1.2 percent31 to 9 percent.15,18 Most cases of coronary artery obstruction identified by screening were single-vessel disease, but up to 2.7 percent of screened participants had significant left main or three-vessel disease18 and as many as 1.7 percent proceeded to revascularization after screening.25 Eight studies screened unselected, low-risk patients.14,16,22,23,27,28,30,38 These studies demonstrated a yield of 0.06 percent to 1.6 percent for asymptomatic coronary heart disease on angiography.
Three studies attempted to estimate the cost-effectiveness of screening to identify prevalent coronary artery obstruction. Sox and colleagues24 used a decision-analysis model to estimate the clinical effectiveness and cost-effectiveness of exercise testing in asymptomatic adults. Their model was structured so that the benefit of screening was achieved through detection of patients with severe disease who would benefit from revascularization. Only direct costs were considered. Levels were based on reimbursement rates at the time of the study (late 1980s): $165 for exercise testing, $3,595 for angiography, and $31,178 for coronary artery bypass surgery. No discounting rate was given. Screening 60-year-old men had a cost per life-year saved of $24,600; for 60-year-old women, the cost was $47,606. For persons 40 years of age, the cost-effectiveness ratios were much higher: $80,349 per life-year saved for men and $216,496 per life-year saved for women.
The presence or absence of risk factors for coronary heart disease affected the cost-effectiveness ratios. The cost per life-year saved was $44,332 for 60-year-old men with no risk factors and $20,504 for those with 1 or more risk factors. The investigators concluded that routine screening was not warranted in general but that it may be beneficial for persons at increased risk for coronary heart disease (for example, older men with 1 or more risk factors). An earlier cost-effectiveness analysis of screening exercise tolerance testing had similar findings.37
Pilote and colleagues28 performed a cost analysis of data from their study of the clinical yield of screening exercise tolerance testing to detect unsuspected severe coronary artery obstruction. They sampled more than 4,000 persons referred to the Cleveland Clinic for screening exercise tolerance testing. Data on cost were obtained from 1994 Medicare reimbursement rates: $110 for exercise testing, $1,780 for angiography, and $27,270 for coronary artery bypass surgery. Screening identified 19 patients with severe coronary artery obstruction (0.44 percent of the cohort); of these, 14 had subsequent coronary artery bypass graft surgery. The investigators estimated a cost of $39,623 to identify 1 case of severe coronary artery disease by screening exercise tolerance testing. The estimated cost per year of life saved was $55,274.
On the basis of these studies, it appears that screening with exercise treadmill testing and performing bypass surgery on persons with severe obstructions is relatively cost effective compared with other, better-accepted types of preventive care, such as mammography in women 50 to 69 years of age.76
Exercise tolerance testing can be used to provide information about a person's risk for a future coronary heart disease event that may augment the predictive ability of traditional risk assessment. Better risk assessment may help clinicians and patients make better decisions about interventions for intermediate- and long-term risk reduction.
Traditionally, studies of the predictive value of exercise tolerance testing on future coronary heart disease have examined ST-segment response to exercise as the risk predictor. Most of these studies reported the total number of coronary heart disease events (fatal and nonfatal myocardial infarction, new-onset stable or unstable angina, and coronary death) as their main outcome. Others reported death from coronary heart disease or from all causes as the main outcome or as secondary outcomes. The mortality rate from coronary heart disease, and particularly the total mortality rate, may be less subject to ascertainment bias than is the total number of coronary heart disease events and, hence, may be more valid measures. However, whether from coronary heart disease or other causes, death is uncommon in the generally healthy, asymptomatic patients enrolled in these studies, making it difficult to estimate the ability of exercise tolerance testing to predict such events.
We identified 15 studies in 18 articles that examined the relationship between ST-segment response to exercise and risk for future coronary heart disease events (Table 3).8,11-13,19-21,26,29,32,33,36,39-42,45,50 Thirteen of these studies (in 16 articles) found that ST-segment response during exercise predicted future coronary heart disease events.8,11-13,19-21,26,29,33,36,39-41,45,50 In 1 of these studies, only coronary heart disease events occurring during exercise was considered as the outcome;12 we therefore excluded it from analysis of the predictive utility for coronary heart disease events. Two studies found that ST-segment response to exercise alone did not predict future coronary heart disease events.32,42
Of the studies that found ST-segment response to be predictive of future coronary heart disease events, 6 (published in 8 articles) selected persons for participation on the basis of the presence of 1 or more risk factors: diabetes,13 multiple risk factors,8,33,39,50 hyperlipidemia,26,41 and sedentary lifestyle and obesity.29 The prevalence of an abnormal exercise tolerance testing, usually defined as ST-segment depression of 1 mm or more, ranged from 12 percent to 52 percent. After adjustment for other risk factors, the independent relative risk for coronary heart disease events associated with an abnormal ST-segment response to exercise in these higher-risk groups ranged from 3.58,50 to 21.0.13 Sensitivity for occurrence of coronary heart disease events over the duration of the studies (3 to 8 years) ranged from 30 percent to 100 percent. The positive predictive value of an abnormal exercise tolerance testing ranged from 7.1 percent26,41 to 46 percent.29
Seven studies (published in 8 articles) found ST-segment response to exercise to be predictive of future coronary heart disease events in an unselected, low-risk sample.11,19-21,33,36,40,45 The prevalence of an abnormal test tended to be lower than that in the higher-risk sample, ranging from 3 percent33 to 20 percent.11,21 The independent relative risk for coronary heart disease events associated with an abnormal exercise tolerance testing ranged from 1.640 to 21,33 with the majority of the values between 2.0 and 5.0. Gibbons and colleagues33 reported a higher relative risk in low-risk persons (21.0) than did the other investigators; however, the absolute event rate was low (0.08 to 2.8 events/1000 person-years) and the confidence interval was wide (6.9 to 63.3). The sensitivity of exercise tolerance testing for coronary heart disease events was 10 percent45 to 70 percent.11,21 The positive predictive values ranged from 2.2 percent33 to 24 percent.19
Two of the studies added nuclear perfusion imaging to exercise electrocardiography.19,32 These studies reported positive predictive values of about 50 percent. However, imaging is likely to increase screening program costs.19,32 As might be expected, the sensitivity of an abnormal exercise tolerance testing decreased as the duration of followup increased (r = -0.56). Data from these cohort studies suggest that the majority of asymptomatic persons with an abnormal exercise tolerance testing do not go on to have coronary heart disease events, at least within the time frame of followup. Persons who do have events often develop angina rather than experience myocardial infarction or sudden death. The prevalence of an abnormal result on exercise tolerance testing and its predictive value among asymptomatic persons is greater in those at higher risk. These data are consistent with those of other investigators and policymakers who have suggested that the value of exercise tolerance testing is greater when it is applied to patients with 1 or more risk factors for coronary heart disease because selection of a higher-risk cohort for screening increases the prevalence of disease and positive predictive value.10 Bruce and associates10 reported that in the Seattle Heart Watch Study of 4158 asymptomatic men and women, a positive result on exercise tolerance testing in the absence of risk factors provided little predictive value. However, among patients with 1 or more other risk factors for coronary heart disease, the occurrence of 2 different types of abnormal response to exercise tolerance testing (exercise risk predictors) was associated with a 15-fold increase in risk compared with patients who had a normal result.
More recent studies of the value of exercise testing in asymptomatic persons have examined the utility of other exercise-associated risk markers, including functional capacity, chronotropic incompetence, heart rate recovery, and development of exercise-induced premature ventricular contractions, for predicting patients' risk for coronary heart disease events or death (Table 4).21,34,35,42-49 In contrast to ST-segment response, these exercise indicators may not directly detect ischemic myocardium, but they probably indicate other cardiovascular derangements, such as abnormal autonomic regulation, that predict coronary heart disease events. In general, these findings are associated with moderate increases in risk for coronary heart disease after adjustment for other risk factors for coronary heart disease (relative risk, 1.7 to 3.5). Some factors are common: For example, failure to achieve target heart rate was noted in 21 percent of patients in the Framingham Offspring Study.44
Two recent studies contribute important information on the predictive value of exercise tolerance testing in asymptomatic women.42,43 The majority of other studies that we identified did not include women or did not provide subgroup analysis of the predictive value of screening exercise tolerance testing for women. Mora and colleagues42 analyzed data from the female participants in the Lipid Research Clinics Prevalence Study, many of whom had hyperlipidemia. They found that unlike in studies whose samples comprised predominantly men, ST-segment response did not predict future risk for coronary heart disease events (relative risk, 0.88 [95 percent CI, 0.48 to 1.61]) in women.42 Low exercise capacity, along with low heart-rate recovery after exercise, was an independent predictor of death from coronary heart disease (relative risk, 3.52 [95% CI, 1.57 to 7.86) and of all-cause death (relative risk, 2.11 [95% CI, 1.47 to 3.04]) in women.
Gulati and coworkers43 sampled asymptomatic female volunteers living in the Chicago area. They found that exercise capacity predicts risk for all-cause death in women. For every increase in exercise capacity of 1 metabolic equivalent, the relative risk for death was 0.83 (95% CI, 0.78 to 0.89). The predictive utility of exercise markers other than ST-segment response in these 2 studies of women is consistent with the results of similar studies in which most participants were men.
Exercise tolerance testing is frequently used as part of an evaluation of middle-aged persons before they begin an exercise program. Few data are available to determine the effectiveness of this approach in reducing the risk for activity-related coronary heart disease events. Siscovick and colleagues12 analyzed the effectiveness of exercise tolerance testing to predict activity-related coronary heart disease events in the Lipid Research Clinics cohort of asymptomatic hypercholesterolemic men. After an initial exercise tolerance test, the cohort was followed for an average of 7.4 years; during that time, the investigators used retrospective record review to identify coronary heart disease events that were associated with moderate or intense activity. The cumulative incidence of activity-related coronary heart disease events during followup was 2 percent. An abnormal ST-segment response to exercise at the time of entry into the study was associated with a relative risk of 2.6 (95% CI, 1.3 to 5.2) for activity-related coronary heart disease events. The sensitivity of exercise testing for predicting the events was 18 percent, and the predictive value of a positive test for coronary heart disease events during exercise was 4 percent. Of the persons who had an activity-associated coronary heart disease event, 80 percent had an initially normal ST-segment response to exercise; 94 percent of persons with abnormal ST-segment response to exercise did not have an activity-associated event during followup. Thus, exercise testing appears to have limited ability to detect persons who will have exercise-related coronary heart disease events.
Other than information on the frequency of false-positive results, we found no studies that examined the potential harms of screening. No study reported rates of complications from angiography of asymptomatic persons, measures of anxiety from knowledge of an abnormal test result, or adverse events from medical therapy initiated because of an abnormal test result.