Population attributable fractions for 5 major risk factors among subjects free of clinical cardiovascular disease at baseline. SBP indicates systolic blood pressure; AAI, ankle-arm index; IC IMT, intima-to-media wall thickness of the internal carotid artery; and echo EF, echocardiographic ejection fraction.
Psaty BM, Furberg CD, Kuller LH, Bild DE, Rautaharju PM, Polak JF, Bovill E, Gottdiener JS. Traditional Risk Factors and Subclinical Disease Measures as Predictors of First Myocardial Infarction in Older AdultsThe Cardiovascular Health Study. Arch Intern Med. 1999;159(12):1339-1347. doi:10.1001/archinte.159.12.1339
Risk factors for myocardial infarction (MI) have not been well characterized in older adults, and in estimating risk, we sought to assess the individual and joint contributions made by both traditional risk factors and measures of subclinical disease.
In the Cardiovascular Health Study, we recruited 5888 adults aged 65 years and older from 4 US centers. At baseline in 1989-1990, participants underwent an extensive examination that included traditional risk factors such as blood pressure and fasting glucose level and measures of subclinical disease as assessed by electrocardiography, carotid ultrasonography, echocardiography, pulmonary function, and ankle-arm index. Participants were followed up with semiannual contacts, and all cardiovascular events were classified by the Morbidity and Mortality Committee. The main analytic technique was the Cox proportional hazards model.
At baseline, 1967 men and 2979 women had no history of an MI. After follow-up for an average of 4.8 years, there were 302 coronary events, which included 263 patients with MI and 39 with definite fatal coronary disease. The incidence was higher in men (20.7 per 1000 person-years) than women (7.9 per 1000 person-years). In all subjects, the incidence was strongly associated with age, increasing from 7.8 per 1000 person-years in subjects aged 65 to 69 years to 25.6 per 1000 person-years in subjects aged 85 years and older. Glucose level and systolic blood pressure were associated with the incidence of MI, but smoking and lipid measures were not. After adjustment for age and sex, the significant subclinical disease predictors of MI were borderline or abnormal ejection fraction by echocardiography, high levels of intimal-medial thickness of the internal carotid artery, and a low ankle-arm index. Forced vital capacity and electrocardiographic left ventricular mass did not enter the stepwise model. Excluding subjects with clinical cardiovascular diseases such as prior angina or congestive heart failure at baseline had little effect on these results. Risk factors were generally similar in men and women.
After follow-up of 4.8 years, systolic blood pressure, fasting glucose level, and selected subclinical disease measures were important predictors of the incidence of MI in older adults. Uncontrolled high blood pressure may explain about one quarter of the coronary events in this population.
IN MIDDLE-AGED adults, the major risk factors for a first myocardial infarction (MI) have been well characterized and include smoking, diabetes, lipid levels, and systolic and diastolic blood pressures.1- 6 Drug treatment of hypertension and hyperlipidemia is known to reduce the risk of first coronary events.7,8 For several major risk factors in older adults, however, both the levels of risk and the benefits of intervention are in general less clearly established. While the benefits of the treatment of high blood pressure are clear,9,10 the risks associated with elevated levels of cholesterol and the potential benefits of therapy in older adults remain controversial.11- 14
In recent years, advances in technology have also provided physicians with new diagnostic methods. Echocardiography is widely available, and early population-based studies suggested that echocardiographic left ventricular mass was an important predictor of MI.15 Ultrasound examination of the carotid arteries can be used to assess intima-to-media wall thickness (IMT) as a measure of atherosclerosis,16 and in a recent report from the Rotterdam Cohort Study, IMT of the common carotid artery was associated with increased risks of stroke and MI.17 In the assessment of the risk of MI, the individual and joint contributions of these subclinical disease measures and traditional risk factors have not been previously reported in population-based studies of older adults.
The Cardiovascular Health Study (CHS) was designed to assess these risks. In this analysis, we sought to examine the association of incident MI with (1) traditional risk factors after adjustment for demographic factors, (2) subclinical disease measures before and after adjustment for traditional risk factors, and (3) the best overall set of predictors.
The CHS is a prospective cohort study of risk factors for coronary heart disease and stroke in men and women aged 65 years and older. In June 1990, 4 Field Centers completed recruitment of 5201 participants. In June 1993, the recruitment of an additional 687 African Americans was completed. Each community sample was obtained from random samples of the Medicare eligibility lists, and those eligible to participate included all persons who were living in the household of each individual sampled from the Health Care Financing Administration lists and who (1) were 65 years or older, (2) were noninstitutionalized, (3) expected to remain in the area for 3 years, and (4) gave informed consent and did not require a proxy respondent. Among those contacted and eligible, 57.3% were enrolled. The CHS design and recruitment experience are described in detail elsewhere.18,19
The baseline examination consisted of a home interview and a clinic examination. Participants answered standard questionnaires that assessed a variety of risk factors, including smoking, physical activity, and medical history of cardiovascular conditions.18 The self-reported medical conditions such as MI were validated.20 Medications were assessed by inventory at the home interview.21
Participants were asked to come to the clinic examination after a 12-hour overnight fast. Seated blood pressure, electrocardiography, and venipuncture were performed early in the examination as previously described.18 Duplicate measures of supine blood pressure in the right arm and both posterior tibial arteries were assessed by an 8-MHz Doppler probe attached to a stethoscope, and the ratio of systolic blood pressures was used to calculate the ankle-arm index. Anthropomorphic measures included weight and height. Electrocardiograms (ECGs) were read by the ECG Reading Center,22 and ECG left ventricular mass was estimated according to a new algorithm, which included the best predictors of echocardiographic left ventricular mass in the CHS population.23 The forced vital capacity and forced expiratory volume in 1 second were measured with a water-sealed spirometer (Collins Survey II; WE Collins, Braintree, Mass).
Blood samples from the fasting venipuncture were analyzed at the Central Blood Analysis Laboratory for glucose; fibrinogen; factor VII, standardized to the World Health Organization reference materials; and total cholesterol, high-density lipoprotein cholesterol, and triglycerides, standardized according to the Centers for Disease Control and Prevention as previously described.18,24 Low-density lipoprotein cholesterol was calculated according to the Friedewald equation.25
Carotid sonography was performed with sonographic units (Toshiba SSA-270A; Toshiba America Medical Systems, Tustin, Calif). A single longitudinal lateral view with measurements taken at the distal 10 mm of the far wall of the right and left common carotid arteries and 3 views with measurements centered on the site of maximum wall thickening of the proximal right and left internal carotid arteries were recorded and read by the Ultrasound Reading Center.26 The IMT was the average of the discrete maximum separately for both common and both internal carotid arteries. Readers also estimated the maximum degree of luminal stenosis. The echocardiographic examinations included 2-dimensional and Doppler methods performed with the Toshiba SSH-160A sonographic units. The CHS Echocardiography Reading Center27 read M-mode left ventricular wall thicknesses and dimensions. Readers also classified the ejection fraction in qualitative terms as borderline or abnormal.
The examination of the African American cohort in 1992-1993 was largely the same as the baseline examination of the main cohort in 1989-1990. Echocardiographic results were not available at baseline for the African American cohort, and based on the important predictors identified in the main cohort,28 ejection fraction was imputed.29- 31 In sensitivity analyses, echocardiographic ejection fraction was an important predictor after excluding the African Americans recruited in 1992-1993. To include this subgroup of the cohort, we used the imputed ejection fraction for the African Americans recruited in 1992-1993.
Subjects were excluded from the analysis if (1) they reported an MI prior to entry into CHS; (2) they had on their baseline ECG evidence of a previous MI defined as the presence of major Q waves or the combination of minor Q waves and ST-T wave changes32; or (3) during follow-up, they were found to have had an MI that predated their entry into CHS.33
Participants were contacted every 6 months, and the contacts alternated between a telephone interview and a clinic examination, which included an ECG. At each contact, participants were asked about cardiovascular events and all hospitalizations. Discharge summaries and diagnoses were obtained for all hospitalizations. For all potential incident cardiovascular events, additional information, including cardiac enzyme determinations and serial ECGs, was collected.
The algorithm for classifying MI, which includes elements of chest pain, cardiac enzyme levels, and serial ECG changes, has been published.33 For participants whose fatal event did not meet the criteria for a definite fatal MI, deaths were classified as definite fatal coronary heart disease if the participants had chest pain within 72 hours of death or had a history of ischemic heart disease. Final classification of all cardiovascular events was determined by the consensus of the members of the Morbidity and Mortality Committee.33 To identify clinically unrecognized events, annual CHS clinic ECGs were read serially by the ECG Reading Center, and the development of new Q waves (major evolution of Q waves or moderate evolution of Q waves with major ST-T wave evolution; NOVACODES C1 and C222) in a patient without an intervening clinically recognized coronary event was counted as a new silent MI.
Events that were judged by the Morbidity and Mortality Committee to be the consequence of a procedure such as surgery or angioplasty were eliminated from this analysis (n=5 for fatal events and n=45 for nonfatal events), and follow-up for these subjects was censored at the time of their procedure-related event. For subjects with an MI during follow-up, event times were computed as the time to the first definite event. For subjects with a silent MI during follow-up, the event times were set to the midpoint between the serial annual ECGs that identified the new Q waves. For subjects without events, censoring times were calculated as according to the last date of follow-up or the date of death.
Although participants with a prebaseline MI were excluded, some had a history of coronary heart disease, which was defined as a history at baseline of angina, coronary angioplasty, coronary artery bypass surgery, or use of nitroglycerin at baseline. Clinical cardiovascular disease was defined as a history at baseline of coronary disease, congestive heart failure, stroke, or carotid endarterectomy. Several continuous variables were dichotomized: (1) systolic blood pressure less than 140 mm Hg vs 140 mm Hg or greater; (2) fasting glucose level less than 7 mmol/L vs 7 mmol/L or greater (<126 vs ≥126 mg/dL), which is the new recommended definition for diabetes34; (3) ankle-arm index less than 0.9 vs 0.9 or greater; and (4) internal carotid IMT less than 1.79 vs 1.79 mm or greater (80th percentile for subjects without clinical cardiovascular disease at baseline).
We used SPSS-PC software for data analysis.35 Techniques included t tests for continuous variables, χ2 tests for categorical variables, and Cox proportional hazards models for multivariate analysis.36 Population attributable fraction was calculated according to the formula given by Rothman37 and uses information about the prevalence of a risk factor and its relative risk to estimate that risk factor's contribution to disease incidence in the population as a whole. All P values represent 2-sided tests.
The CHS has a large number of potential predictor variables, and in this analysis, we screened a representative sample of risk factors. The variables age, sex, race, and indicator variables for site were included in all models. Based on the initial bivariate screening, we selected a limited number of variables for multivariate analysis and submitted them to stepwise analysis in 2 separate groups. The first group included traditional major risk factors, which were systolic blood pressure, current smoking, fibrinogen level, fasting glucose level, and high-density lipoprotein cholesterol level. The second group included measures of subclinical disease, and for each examination component we selected a single predictor variable. The subclinical disease measures were ankle-arm index, maximum internal carotid IMT, ECG left ventricular mass, echocardiographic ejection fraction, and forced vital capacity. Data on left ventricular mass as assessed by echocardiography were missing for about one third of subjects, so we chose ECG left ventricular mass instead.23 Criteria for entry and exit in the stepwise models were P<.05 and P>.10, respectively.
Excluded from this analysis were 403 men and 297 women who reported at baseline a previous history of MI; 97 men and 104 women with evidence of an MI on their baseline ECG; and 28 men and 13 women who, during follow-up, were found to have an MI that predated their entry into the CHS. Of 5888 participants, the 1967 men and 2979 women who were at risk of a first MI were included in this analysis and followed up for an average of 4.8 years.
Among all events (Table 1), 29 (9.6%) of 302 were definite fatal MIs, and 39 (12.9%) represented definite fatal coronary heart disease. Another 40 (13.2%) MIs were detected only by the presence of new Q waves on annual in-clinic ECGs. The event rate in men was significantly higher than the rate in women (Table 2), and the incidence was strongly associated with age in both men and women (P<.01).
Table 3 summarizes the baseline risk factors in men and women with and without clinical cardiovascular disease at baseline. In men and women, risk factors such as blood pressure and glucose level were generally higher in subjects with clinical cardiovascular disease. Lipid levels were similar in the 2 groups. The prevalence of smoking was actually lower in those with clinical cardiovascular disease at baseline. The subclinical disease measures, including ejection fraction, carotid IMT, and ankle-arm index, differed between those with and without clinical cardiovascular disease at baseline.
Table 4 summarizes the bivariate associations between the incidence of MI and the risk factors that were considered for multivariate analysis. The unadjusted associations for many of the variables were strong and highly significant. With the exception of smoking, glucose level, and forced vital capacity, adjustment for age, sex, and clinical heart disease tended to reduce the strength of the associations.
Table 5 summarizes the findings for traditional risk factors. After adjustment for age, sex, race, site, and clinical cardiovascular disease, the stepwise model included systolic blood pressure and fasting glucose level. The risk for current smoking was marginal (risk ratio, 1.41; 95% confidence interval, 0.99-2.00; P=.06). In analyses stratified on sex, the point estimates for the risk factors were similar in men and women. Table 6 summarizes the findings for subclinical disease measures after adjustment for demographic factors and traditional risk factors. Three variables—echocardiographic ejection fraction, ankle-arm index, and internal carotid IMT—entered the stepwise model. While internal carotid IMT was a significant predictor in both men and women, echocardiographic ejection fraction was a significant predictor only in women, and ankle-arm index only in men.
Since the forced entry of all the traditional risk factors into the model in Table 6 may represent overadjustment, we repeated the analysis of subclinical disease measures after adjustment only for age, sex, race, site, and clinical cardiovascular disease (Table 7). Forced vital capacity and ECG left ventricular mass still did not enter the stepwise model. Ankle-arm index was primarily a predictor in participants who did not have clinical cardiovascular disease at baseline.
In Table 8, we restricted the analysis to subjects without clinical cardiovascular disease at baseline. In this analysis, we adjusted for age, sex, race, and site, and permitted the traditional and subclinical disease measures to compete for entry into the stepwise model. Among all subjects, the significant predictors were the same—systolic blood pressure, fasting glucose level, ankle-arm index, echocardiographic ejection fraction, and internal carotid IMT. With the exception of ejection fraction, the point estimates for these risk factors were similar in men and women. In sex-specific stepwise models, systolic blood pressure, glucose level, ankle-arm index, and internal carotid IMT entered for men; and for women, systolic blood pressure, echocardiographic ejection fraction, and ankle-arm index.
Figure 1 provides estimates of the population attribution fraction for these 5 predictors of MI among subjects free of clinical cardiovascular disease at baseline. Continuous variables were dichotomized. The range of prevalences was wide, from 5.2% for abnormal ejection fraction up to 39.5% for elevated systolic blood pressure. The population attributable fractions were highest for the traditional risk factors of elevated systolic blood pressure (24.0%) and fasting glucose level (12.8%). For each of the 3 subclinical disease measures, the population attributable fractions were less than 10%.
In additional analysis, results were similar when we excluded the African American cohort recruited in 1992-1993. Systolic blood pressure was much more strongly associated with the incidence of MI than diastolic blood pressure, and after adjustment for systolic blood pressure, there was no association with diastolic blood pressure. None of the lipid measures (total, high-density lipoprotein, and low-density lipoprotein cholesterol or triglycerides) was associated with the risk of MI in this population. In the CHS, the maximum IMT of the internal but not the common carotid artery was associated with the incidence of MI. Analysis using quintiles for systolic blood pressure, ankle-arm index, and carotid IMT suggested generally linear trends in risk for these continuous variables while the elevated risk associated with glucose level was largely confined to those in the highest quintile (glucose level >6.3 mmol/L [>114 mg/dL]). There was no significant interaction between glucose level and drug treatment for diabetes (P=.79) or between systolic blood pressure and drug treatment for hypertension (P=.31). Excluding definite fatal coronary heart disease and excluding silent MI by annual ECG had little effect on the predictors or their estimated risk ratios.
During an average of 4.8 years of follow-up, we identified a first coronary event in 302 (6.1%) of 4946 older adults. The incidence was strongly associated with age, sex, and the presence of other clinical cardiovascular disease. Traditional risk factors, including systolic blood pressure and fasting glucose level, were strongly associated with the incidence of MI in this analysis. The results for smoking were marginal, and there was no association with lipid levels. Several measures of subclinical disease were also important predictors. Internal carotid IMT, ankle-arm index, and echocardiographic ejection fraction were associated with the incidence of MI. In this analysis, forced vital capacity and ECG left ventricular mass did not enter the stepwise models although they were associated with MI in bivariate analyses.
In the CHS, the age-sex specific incidences of MI were 16.3 and 5.8 per 1000 person-years in men and women, respectively, aged 65 to 74 years at baseline and 28.7 and 12.9 in men and women aged 75 to 84 years at baseline. These rates are similar to those reported from 30 years of follow-up (through 1978) in the Framingham Heart Study.6 For the end point of coronary heart disease without angina,6 the respective age-sex specific incidences from Framingham were 18 and 8 per 1000 person-years in men and women aged 65 to 74 years and 27 and 14 in men and women aged 75 to 84 years. While mortality from coronary heart disease has declined precipitously in the past 30 years, the incidence in older adults appears to have changed little. The CHS findings of a relatively high incidence of coronary disease in older adults is consistent with the findings from other studies.38,39 In the Minnesota Heart Survey,39 mortality from coronary disease declined much more dramatically than the occurrence of hospitalizations. Both the improvements in medical care and the detection of less severe events in recent years may be responsible for these trends. Nonetheless, 13.2% of coronary events in the CHS remained clinically silent, detected only as Q wave changes on annual in-clinic ECGs.
In general, the findings from the CHS for traditional risk factors are consistent with those of many previous studies. Although smoking was only marginally significant after adjustment for age, sex, race, clinical cardiovascular disease, systolic blood pressure, and fasting glucose level, the point estimate of a 41% increase in risk is consistent with estimates of other studies that have examined the effect of smoking in older adults.40- 42 In the CHS, both systolic blood pressure and glucose level were strongly associated with the incidence of MI in older adults. Diabetes is a well-known risk factor in middle-aged adults,43,44 and several recent studies have given us a new appreciation for the importance of diabetes as a risk factor for ischemic heart disease in older adults.45,46 In older adults, systolic blood pressure is a major risk factor for coronary disease,9,47,48 and the data from the CHS suggest that even mild elevations above 140 mm Hg—an elevation present in about 40% of the population—may be important.
There was little association between the incidence of coronary disease and plasma lipid levels in the CHS. Follow-up was relatively short in the CHS, and there is some evidence that the strength of the association between cholesterol level and risk increases with longer follow-up.49 While the level of risk appears to be smaller in the elderly than in the middle-aged, the general consensus seems to be that lipid levels are a risk factor for coronary disease in older adults.4,11- 14,46
In this analysis, a number of measures of subclinical disease were also associated with the incidence of coronary disease in older adults. Internal carotid IMT, borderline or abnormal echocardiographic ejection fraction, and low ankle-arm index were all independently associated with coronary disease in the CHS cohort even after adjustment for demographic factors, systolic blood pressure, and fasting glucose level. It is interesting to speculate that since ejection fraction is depressed prior to infarction in some subjects, ischemic hibernation may antedate myocardial necrosis.
The findings for subclinical disease are similar to the results of other studies. Low ankle-arm index is a strong predictor of mortality in older adults.50 In the Kuopio Ischemic Heart Disease Risk Factor Study,51 intimal-medial thickening of the common carotid artery was associated with a 2-fold increase in the risk of coronary heart disease. In the Atherosclerosis Risk in Communities Study, Chambless and colleagues52 also reported a strong graded association between carotid IMT and the incidence of coronary heart disease. Although echocardiographic left ventricular mass was also a risk factor in the Framingham Heart Study,15 ECG left ventricular mass was not an independent risk factor in this analysis from CHS. Forced vital capacity was an important predictor in Framingham,53 but did not enter the stepwise models in this analysis.
At the outset, we identified a limited number of candidate variables, and by doing so, we did not examine the association between the incidence of MI and a large number of other variables that may be important predictors. Moreover, in stepwise analysis, the entry of additional variables and the estimates of their relative risks depend importantly on the variables that are already in the model.
The CHS is a population-based study, and 57.3% of eligible subjects enrolled in the CHS. Population attributable fractions depend on estimates of both prevalence and risk level. The prevalences of hypertension and diabetes were actually lower in those who were eligible and enrolled than in those who were eligible but did not enroll.19 If, as seems likely, the associations of risk factors and disease incidence are similar between the enrolled and the unenrolled, the population attributable fractions for these risk factors in the CHS may be slightly underestimated.
Compared with the measurement of blood pressure or fasting glucose, the subclinical disease measures represent more costly high-technology methods of assessing risk. Some of them, including echocardiographic ejection fraction and internal carotid maximum IMT, were important independent risk factors for coronary disease in this study, and clearly their use significantly improved the ability to predict coronary events in older adults. For the clinician seeing patients who have multiple risk factors, including those risk factors defined by subclinical disease measures, it is important to emphasize that these patients have high rates of coronary events.54,55 From the point of screening populations, however, these high-risk patients are uncommon. Among subjects without clinical cardiovascular disease at baseline, the prevalences of low ankle-arm index and borderline or abnormal ejection fraction were relatively low, found in only 9.5% and 5.2.% of the population at baseline, respectively (Figure 1).
In this population, elevated levels of blood pressure and glucose clearly pose the greatest hazard to the health of the public ( Figure 1). Systolic blood pressure elevated above 140 mm Hg may explain one quarter of the coronary events in this study. For stroke in older adults, elevated systolic blood pressure may explain about one third of cerebrovascular events.31 While the health effects of the aggressive treatment of elevated glucose levels in patients with type 2 diabetes are currently under study,56 the effectiveness of antihypertensive therapy, especially low-dose diuretic therapy, is well established in older adults.7 The absolute risk reductions associated with the low-dose diuretic therapy to treat elevated systolic pressure are twice as high in diabetic as in nondiabetic individuals.57 Both in terms of the levels of risk and the prevalence of the conditions, the treatment of even modest levels of systolic hypertension and the prevention of glucose intolerance with diet and physical activity are likely to have the largest effects in preventing the incidence of coronary disease in older adults.
Accepted for publication September 29, 1998.
This research was supported by contracts N01-HC-85079, N01-HC-85080, N01-HC-85081, N01-HC-85082, N01-HC-85083, N01-HC-85084, N01-HC-85085, and N01-HC-85086 from the National Heart, Lung, and Blood Institute, and from the NWO (Nederlandse Organisatie voor Wetenschappelijk Onderzoek). Dr Psaty is a Merck/SER Clinical Epidemiology Fellow (sponsored by the Merck Co Foundation, Rahway, NJ, and the Society for Epidemiologic Research, Baltimore, Md).
We appreciate the comments, criticisms, and suggestions that Teri Manolio, MD, MHS, provided on earlier drafts of the manuscript.
Reprints: Cardiovascular Health Study, Coordinating Center, Century Square, Suite 2105, 1501 Fourth Ave, Seattle, WA 98101.
Participating Institutions and Principal Staff
Field Center in Forsyth County, North Carolina: Bowman Gray School of Medicine of Wake Forest University, Winston-Salem, NC: Gregory L. Burke, Alan Elster, Walter H. Ettinger, Curt D. Furberg, Edward Haponik, Gerardo Heiss, Dalane Kitzman, H. Sidney Klopfenstein, Margie Lamb, David S. Lefkowitz, Mary F. Lyles, Cathy Nunn, Ward Riley, Maurice Mittelmark, Grethe S. Tell, James F. Toole, Beverly Tucker; Bowman Gray School of Medicine–ECG Reading Center: Farida Rautaharju, Pentti Rautaharju. Field Center in Sacramento County, California: University of California, Davis: William Bommer, Charles Bernick, Andrew Duxbury, Mary Haan, Calvin Hirsch, Paul Kellerman, Lawrence Laslett, Marshall Lee, Virginia Poirier, John Robbins, Marc Schenker, Nemat Borhani. Field Center in Washington County, Maryland: The Johns Hopkins University, Baltimore, Md: M. Jan Busby-Whitehead, Joyce Chabot, George W. Comstock, Linda P. Fried, Joel G. Hill, Steven J. Kittner, Shiriki Kumanyika, David Levine, Joao A. Lima, Neil R. Powe, Thomas R. Price, Jeff Williamson, Moyses Szklo, Melvyn Tockman; MRI Reading Center–The Johns Hopkins University: R. Nick Bryan, Carolyn C. Meltzer, Douglas Fellows, Melanie Hawkins, Patrice Holtz, Michael Kraut, Grace Lee, Larry Schertz, Earl P. Steinberg, Scott Wells, Linda Wilkins, Nancy C. Yue. Field Center in Allegheny County, Pennsylvania:University of Pittsburgh, Pittsburgh, Penn: Diane G. Ives, Charles A. Jungreis, Laurie Knepper, Lewis H. Kuller, Elaine Meilahn, Peg Meyer, Roberta Moyer, Anne Newman, Richard Schulz, Vivienne E. Smith, Sidney K. Wolfson.
Echocardiography Reading Center (Baseline)–University of California, Irvine: Hoda Anton-Culver, Julius M. Gardin, Margaret Knoll, Tom Kurosaki, Nathan Wong; Echocardiography Reading Center (Follow-up)–Georgetown Medical Center, Washington, DC: John Gottdiener, Eva Hausner, Stephen Kraus, Judy Gay, Sue Livengood, Mary Ann Yohe, Retha Webb; Ultrasound Reading Center–Tufts, New England Medical Center, Boston, Mass: Daniel H. O'Leary, Joseph F. Polak, Laurie Funk; Central Blood Analysis Laboratory–University of Vermont, Colchester: Edwin Bovill, Elaine Cornell, Mary Cushman, Russell P. Tracy; Respiratory Sciences–University of Arizona-Tucson: Paul Enright; Coordinating Center–University of Washington, Seattle: Alice Arnold, Annette L. Fitzpatrick, Bonnie K. Lind, Richard A. Kronmal, Bruce M. Psaty, David S. Siscovick, Lynn Shemanski, Lloyd Fisher, Will Longstreth, Patricia W. Wahl, David Yanez, Paula Diehr, Maryann McBurnie; National Heart, Lung, and Blood Institute Project Office: Diane E. Bild, Robin Boineau, Teri A. Manolio, Peter J. Savage, Patricia Smith.