Task force 5. Stratification of patients into high, medium and low risk subgroups for purposes of risk factor management

JACC Vol. 27, No. 5 April 1996:964-1047

CALIFF ET AL. TASK FORCE 5

1007

Task Force 5. Stratification of Patients Into High, Medium and Low Risk Subgroups for Purposes of Risk Factor Management R O B E R T M. C A L I F F , MD, FACC, CRAm, P A U L W. A R M S T R O N G , MD, FACC, J O S E P H R. C A R V E R , MD, FACC, R A L P H B. D ' A G O S T I N O , Sin, PHD, W I L L I A M E. STRAUSS, MD, F A C C

General Perspectives on Risk and Effect on Intervention The continuing increase in the prevalence of coronary artery disease coupled with increasing evidence for the clinical and economic benefits of prevention provides the clinician with an opportunity to improve patient outcomes. There are over 7 million Americans with diagnosed coronary artery disease (1); the challenge is to develop the tools and information to focus our efforts effectively. Definitive evidence exists that meaningful changes in mortality and morbidity may be achieved through prevention strategies in patients with established cardiovascular disease (2,3), and a strong rationale exists in high risk patients without documented vascular disease. Decisions made during every patient-physician contact form a small but incrementally important portion of risk intervention during the life of a person who has or is at risk for the development of coronary artery disease. Thus, a focused approach to preventing coronary events using risk stratification begins with the initial patient encounter and is continually refined as additional information is acquired and as the disease process progresses. For the same reasons that intensive intervention may be most cost-effective in patients with established cardiovascular disease (4), the concept that identification of risk level can provide a guide to the intensity of preventive measures is attractive. As the number of cardiac risk factors increases, the rise in associated risk may be multiplicative. In this task force we review the available information about level of risk as a function of various types of risk factors in the sequence in which they become available to the clinician (identification of individual risk factors is reviewed in previous task forces). An attempt is then made to provide a model for assigning patients to levels of risk in a manner that could be helpful in the development of prevention strategies. Fundamentals of risk prediction. Risk prediction has been described as an effort to predict the future from knowledge of the past (5). This activity has always been regarded as a fundamental component of the role of health care providers, but the methods of assessing risk have often been arbitrary and informal. Clinicians have tended to use heuristics, or "rules of thumb," to identify patients at high risk or low risk of poor

outcomes, to advise the patient and to develop treatment strategies. Unfortunately, this approach suffers from a lack of consensus about the definition of high and low risk, an inability to determine the accuracy of individualized predictions of risk in practice and the difficulty of synthesizing multiple risk characteristics into a specific estimate of risk for the individual. This last problem is particularly troublesome considering that multiple risk factors are known, their assessment can be expensive, and they overlap substantially. In contrast to this heuristic approach, selected research groups, exemplified by the Framingham Study, have systematically and scientifically estimated the risks of cardiovascular morbid events and mortality with mathematical functions (6,7). These functions have been shown to have validity and applicability to populations beyond their original cohorts (8-10). While substantial progress has been made in the science of risk prediction (11-14), the available methods are continuing to evolve (12). What are we trying to predict? The perspective of this section will be that the goal of clinical risk prediction is to provide the clinician and the patient with a logical estimate that an important deleterious clinical event will occur. The most important outcome is death. However, an understanding of the future risk of nonfatal clinical outcomes, particularly myocardial infarction, stroke, symptomatic heart failure, hospitalization for unstable angina as well as changes in functional capacity and overall quality of life, must be recognized as a critical element of risk evaluation. These events inevitably are intermingled with the need for and the use of revascularization procedures (surgical and percutaneous), which carry an early increase but a possible later decrease in the risk of later coronary artery disease events. Finally, the economic consequences of coronary artery disease are of increasing concern to patients and health care providers. Unfortunately, little information is available about predictors of any of these outcomes other than death. Why should we predict? Differential risk stratification within the traditional risk factors influences three partners of medical delivery: physicians, patients and insurers. Physicians. Appropriate physician care of patients with coronary artery disease depends on the ability to make indi-

1008

CALIFF ET AL. TASK FORCE 5

vidual decisions based on accurate risk discrimination; this is fundamental to ensure quality care. In addition, as the number of clinical pathways proliferates, this information will be crucial for implementation and variance analysis. Patients. Patients are becoming more informed consumers of health care. Having the patient as an active participant in informed clinical decision making can improve outcomes, utilization and quality of life (15); for patients with coronary artery disease, this process is grounded in accurate risk assessment at every point in the decision tree. Patient involvement in care, based on accurate risk assessment and the knowledge of risk reduction probabilities, can also be a powerful motivation for drug, diet and exercise compliance and other life-style changes. Ultimately, these data affect family and economic planning for the future, which cannot be approached without knowledge based on the accurate prediction of risk and its negative consequences. Insurers. Future benefit coverage decisions about new technologies, drugs and interventions will depend on the ability to assess risk and apply new actuarial models for care delivery. Cost-effective interventions that reduce risk will be supported; activities that do not affect outcome or quality of life will not be supported. Multifactorial nature of risk. Risk factors are additive and (in some cases) interactive. Predicting the risk of a clinical event in a patient with known disease requires an understanding of how biologic and clinical factors interact to provide an overall risk. Mathematical modeling is necessary to provide a quantitative assessment of the magnitude of the interactive effect. Discussion concerning risk stratification of patients with coronary artery disease has traditionally focused on a variety of noninvasive and invasive diagnostic strategies. It should be appreciated that appropriate risk stratification should not be delayed until after the results of specialized noninvasive or invasive tests are known. Rather, test results should be considered for their incremental value beyond the provider's pretest clinical impression of risk. This impression is first formed with the patient's history, physical examination and electrocardiogram (ECG). After all of this information is quantified, the value of additional testing in assessing risk can be determined. Quality of studies. The ideal information to develop statistical models to assess risk in patients with known coronary artery disease would include a complete assessment of the population: demographics, medical history, physical examination, coronary anatomy, left ventricular function, laboratory measurements and provocative or functional testing. Obviously, no single study contains this ideal information set; accordingly, this effort will attempt to synthesize information from a variety of sources to provide an overview of risk. The prognoses of patients who have or who are at risk of developing coronary artery disease have been determined from multiple sources, including population-based studies, multicenter trials, multicenter and single-center registries and individual investigators or groups (5). The Framingham Study provides a good example of population-based incidence and prevalence data; however, it does not include detailed information about invasive or noninvasive testing. Single-center

JACC Vol. 27, No. 5 April 1996:964-1047

registries are frequently more complete than multiple-site registries, although multiple-site registries are more generalizable. One of the main criticisms of the multicenter clinical trial is that only a portion of patients, for a variety of reasons, enter the trial. Registry data, such as those from the Coronary Artery Surgery Study (CASS) (16) and the Duke Databank for Cardiovascular Disease (17,18), provide a structured way of examining prognostic data but do not limit patient enrollment. The practical assessment of risk will focus on statistical models developed in the Framingham Study and the Duke Databank, since these models have been validated in independent populations. Other excellent models yielding comparable results are also available.

R i s k Prediction in People Free of Coronary Artery D i s e a s e Multivariable risk assessment. The first level of risk prediction and stratification is for those free of coronary artery disease but who might be at high risk for coronary artery disease due to modifiable risk factors, such as hypertension, hyperlipidemia and smoking, or immutable factors, such as age and gender. To develop a method for predicting risk in this population, the Framingham Study generated a mathematical model (an accelerated-failure model) that can be used to predict the probability of developing a clinical manifestation of coronary artery disease based on a patient's profile of standard risk factors (19). Further, the Framingham Study, with the American Heart Association, generated from this function a simple "Coronary Heart Disease Risk Factor Prediction Chart" for producing 5- and 10-year predictions of the probability of developing clinical evidence of coronary artery disease. This chart is presented in Table 1. To use Table 1, data on eight risk factors are needed: gender, age, high density lipoprotein (HDL), total serum cholesterol, rest systolic blood pressure, cigarette smoking (yes or no), diagnosis of diabetes (yes or no) and the presence or absence of left ventricular hypertrophy obtained from the ECG (yes or no). Depending on the values and categories of these risk factors, points are assigned. The points are then totaled, and 5- and 10-year probabilities corresponding to individual risk are obtained. The 10-year probability can be compared with the gender- and age-specific average 10-year probability (risk) obtained empirically from the Framingham population. The ratio of the estimated probability to the Framingham ageand gender-specific risk can be used as an estimate of the patient's risk relative to an "average" individual of the same gender and age. These relative risk estimates have been shown to be valid when the Framingham functions are applied to different populations (9,20,21). For example, a 60-year old man with an HDL cholesterol of 35 mg/dl, total cholesterol of 240 mg/dl, systolic blood pressure of 160 mm Hg and who smokes cigarettes, is diabetic and has no left ventricular hypertrophy on the ECG would be assigned 14 points for being 60 years old and male and 4, 3, 4, 4, 3 and 0 points, respectively, for the HDL cholesterol, total choles-

JACC Vol. 27, No. 5 April 1996:964-1047

CALIFF ET AL. TASK FORCE 5

1009

Table 1. CoronaryHeart Disease Risk Factor Prediction Chart: Patients Without Known CoronaryArtery Disease 1. Find Points for Each Risk Factor: Women Age

Men Age

(yr)

Pts

(yr)

Pts

30 31 32 33 34 35 36 37 38 39 40 41 42-43 44 45-46

-12 -11 -9 -8 -6 -5 -4 -3 -2 -1 0 1 2 3 4

47-48 49-50 51-52 53-55 56-60 61-67 68 -74

5 6 7 8 9 10 11

Age (yr)

Pts

30 31 32-33 34 35-36 37-38 39 40-41 42-43 44-45 46-47 48-49 50-51 52-54 55-56

-2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12

Age (yr)

Pts

(mg/dl)

Pts

57-59 60-61 62-64 65-67 68-70 71-73 74

13 14 15 16 17 18 19

25-26 27-29 30-32 33-35 36-38 39-42 43- 46 4%50 51-55 56-60 61-66 67-73 74-80 81-87 88-96

7 6 5 4 3 2 1 0 -1 -2 -3 -4 -5 -6 -7

HDL-C

Total-C (mg/dl)

Pts

139-151 152-166 16%182 183-199 200-219 220-239 240 -262 263-288 289-315 316-330

-3 -2 -1 0 1 2 3 4 5 6

SBP (mm Hg)

Pts

Other

98-104 105-112 113-120 121-129 130-139 140-149 150-160 161-172 173-185

-2 -1 0 1 2 3 4 5 6

Cigarettes Diabetic (male) Diabetic (female)

Pts*

ECG-LVH

2. Sum Points for All Risk Factors (subtract minus points from total): -

-

+

Age

- - + HDL-C

-

- + Total-C

-

- + SBP

-

Smoker + Diabete------~ + E C G - L V H

Point total

4. Compare With Average 10-Year Risk:

3. Look U p Risk Corresponding to Point Total:

Probability

Probability

Pts

5 yr

10 yr

Pts

5 yr

10 yr

Pts

5 yr

10 yr

Pts

5 yr

10 yr

Age (yr)

Women

Men

--<-1 2 3 4 5 6 7 8 9

<1% 1% 1% 1% 1% 1% 1% 2% 2%

<2% 2% 2% 2% 3% 3% 4% 4% 4%

10 11 12 13 14 15 16 17 18

2% 3% 3% 3% 4% 5% 5% 6% 7%

6% 6% 7% 8% 9% 10% 12% 13% 14%

19 20 21 22 23 24 25 26 27

8% 8% 9% 11% 12% 13% 14% 16% 17%

16% 18% 19% 21% 23% 25% 27% 29% 31%

28 29 30 31 32

19% 20% 22% 24% 25%

33% 36% 38% 40% 42%

30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74

<1% <1% 2% 5% 8% 12% 13% 9% 12%

3% 5% 6% 10% 14% 16% 21% 30% 24%

*Zero points for each "no." ECG-LVH = electrocardiographic left ventricular hypertrophy; HDL-C = high density lipoprotein cholesterol; pts = patients; SBP = systolic blood pressure; Total-C = total cholesterol.

terol, systolic blood pressure, smoking, presence of diabetes and lack of left ventricular hypertrophy. This yields a total of 32 points, which corresponds to an estimated 10-year probability of having a clinical manifestation of coronary artery disease of 42%. The average 10-year risk for 60-year old men from the Framingham Study population was 21%. The present subject has a relative risk of 2.00 (42% - 21%) or a 100% higher risk. Given the 5- or 10-year probability estimates and the status of the individual risk factors in conjunction with existing guidelines, the physician can determine the seriousness of the patient's risk and plan intervention strategies. Further, the physician can estimate the potential effect of a successful intervention. For example, in the previous example, if the patient's HDL cholesterol could be elevated to 45 mg/dl, total

cholesterol reduced to 200 mg/dl, systolic blood pressure reduced to 120 mm Hg and the smoking stopped, then the 10-year probability would decrease to 16%, an absolute reduction of 26% (42% - 16%), or a relative reduction of 62% [100(42 - 16)/42]. This estimate of the potential success of the intervention should be viewed only as a guide. In the application of any intervention, there are factors not necessarily captured in the mathematical functions, and only a randomized trial can determine the net effect of an intervention. Risk associated with fibrinogen. Thrombogenic components contribute to the risk of atherosclerosis and to its clinical manifestations. One example from the Framingham Study and other studies is the significant role of fibrinogen as a risk factor for the development of coronary artery disease (22). Simple approximate multiplicative adjustments to the probabilities

1010

CALIFF ET AL. TASK FORCE 5

JACC Vol. 27, No. 5 April 1996:964-1047

Table 2. Independent Effect of Fibrinogen: Risk Adjustment Multiplication Factor for Probability Estimate Obtained From Table 1

Fibrinogen Level (mg/dl)

Male

Female

<235 235-335 >335

0.83 1.00 1.20

0.77 1.00 1.30

obtained from Table 1 that incorporate the independent effect of fibrinogen are presented in Table 2. To illustrate the use of Table 2, consider previous example of the 60-year old man used to illustrate the use of Table 1. He had a 10-year probability of developing a clinical manifestation of coronary artery disease of 42%. If he had a fibrinogen level of 350 mg/dl, then using the Table 2, his adjusted probability is 50.4% (42% × 1.20). Consideration of other vascular outcomes. While our objective is to focus on the risk of coronary artery disease, and not to be inclusive and cover the broader category of vascular disease, physicians and patients are concerned about the possibility of other outcomes. Risk factors for the development of coronary artery disease are also risk factors for other vascular events such as stroke. The effects of these risk factors--age, diabetes, smoking, cerebrovascular disease, systolic blood pressure, atrial fibrillation and left ventricular hypertrophy--on the development of stroke have been quantified by the Framingham Study Group (7,23). In the assessment of risk and stratification of patients, coronary artery disease risk factors may take on a slightly different role for other vascular diseases, such as the role of blood pressure in the development of stroke. Further, other variables not of immediate concern for one cardiovascular condition (such as atrial fibrillation) may take on a major role for other manifestations of cardiovascular disease.

R i s k Prediction in Patients With Existing Coronary Artery D i s e a s e Standard risk factors. For patients with overt coronary artery disease, functions have been developed recently from Framingham data for estimating 2-year probabilities of a future coronary artery disease event. These functions are presented in Tables 3 and 4 for women and men, respectively, in a format similar to that of Table 1. While it is important to react to the presence of any risk factor, especially the modifiable ones, in a patient with existing coronary artery disease, Tables 3 and 4 should be useful for further clarification of the role and importance of the traditional risk factors in these patients. The variable "cigarette smoking" was not statistically significant in men. This apparent anomaly has appeared in other studies and probably relates to the fact that the smokers who survive the acute period of coronary artery disease are younger than surviving nonsmokers. The variables for ECG left ventricular hypertrophy and fibrinogen have not yet been

added to these Framingham functions, but both are highly significant, especially left ventricular hypertrophy. High levels of fibrinogen (above 310 mg/dl) and the presence of ECG left ventricular hypertrophy should be cause for concern. These functions do not include an assessment of symptoms or additional testing and therefore should be considered to be only approximate guides to risk prediction. As demonstrated in the following sections, knowledge of more detailed information about symptom status, coronary anatomy, left ventricular function and stress testing can provide more powerful prognostic estimates in patients with a previous event.

Clinical Scenarios a n d Time-Related R i s k For the purposes of clinical risk stratification for secondary prevention, the patient may be considered to be in one of five categories: stable coronary artery disease, unstable angina, acute myocardial infarction, post-coronary artery bypass surgery, or post-percutaneous coronary intervention. The latter four categories have in common with stable coronary artery disease the presence of the diseased arterial wall, and the practitioner must assume that all diseased coronary vessels are susceptible to plaque fissuring and subsequent thrombosis. They differ from stable coronary artery disease in that conditionspecific issues arise in secondary prevention. Figure 1 displays the time-related risk of future cardiac events as a function of time from the event or procedure (24,25). The patient seen at the time of discharge after acute myocardial infarction or unstable angina is in the midst of a particularly high risk period, during which the myocardium is prone to sudden electrical instability, and the initially disrupted atherosclerotic plaque is susceptible to rethrombosis. After bypass grafting, the hazard of graft occlusion due to mechanical or thrombotic causes is greatest within the first month to 1 year after operation, followed by a low risk period (Fig. 2) (26). After angioplasty, atherectomy, Rotablator or stent placement, a significant risk of restenosis exists in the first 6 months, which subsequently declines exponentially; the obvious impact of time as a risk factor and for risk stratification is thus emphasized. This section will focus on secondary prevention in the patient with stable coronary artery disease, with mention of the latter four situations where appropriate. These categories may be further modified depending on the symptomatic status of the patient. For example, those with acute ischemic syndromes form a higher risk group than those who are asymptomatic. Risk has been similarly categorized in patients with congestive heart failure symptoms (e.g., New York Heart Association functional class). Another high risk group is sudden-death survivors, whose short- and long-term event risk is much higher than that of other patient populations with stable coronary disease.

Medical History Assessment of patient risk in relation to clinical conditions emanating from established atherosclerosis depends on basic

JACC Vol. 27, No. 5 April 1996:964-1047

CALIFF ET AL. TASK FORCE 5

1011

Table 3. Risk of Coronary Artery Disease Event, Stroke or CerebrovascularDisease Death in Women With Existing CoronaryArtery Disease Age

Points by HDL-C (mg/dl)

Total-C

(yr)

Points

(mg/dl)

25

30

35

40

45

50

60

70

80

35 40 45 50 55 60 65 70 75

0 1 2 3 4 5 6 7 7

Other Diabetes Smoking

Pts

160 170 180 190 200 210 220 230 240 250 260 270 280 290 300

4 4 4 4 4 4 5 5 5 5 5 5 5 5 6

3 3 3 4 4 4 4 4 4 4 5 5 5 5 5

3 3 3 3 3 3 4 4 4 4 4 4 4 4 4

2 2 2 3 3 3 3 3 3 4 4 4 4 4 4

2 2 2 2 2 3 3 3 3 3 3 3 3 4 4

1 2 2 2 2 2 2 3 3 3 3 3 3 3 3

1 1 1 1 2 2 2 2 2 2 2 2 3 3 3

0 1 1 1 1 1 1 2 2 2 2 2 2 2 2

0 0 0 1 1 1 1 1 1 1 1 2 2 2 2

3 3

2-yr

Total Points 0 2 4 6 8 10 12 14 16 18 20 22 24 26

SBP (ram Hg)

Points

100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

0 0 1 1 2 2 2 3 3 3 3 4 4 4 4 4

Average 2-yr Risk in Women With CVD

Probability

Age

Probability

(%)

(yr)

(%)

0 1 1 1 2 4 6 10 15 23 35 51 68 85

35-39 40-44 45-49 50 -54 55-59 60- 64 65-69 70-74

<1 <1 <1 4 6 8 12 12

CVD = cerebrovascular disease; other abbreviations as in Table 1.

data from the history, physical examination, 12-lead ECG and appropriate laboratory tests. Myocardial infarction. The occurrence of a myocardial infarction is a sentinel event in the life of a patient with coronary artery disease. After the initial coronary occlusion, the risk of death or nonfatal complications is extremely high, and this risk diminishes nonlinearly with time (25). This time-related risk seems to be a function of the propensity for sudden ventricular arrhythmia. It is additive to the risk attributable to the amount of left ventricular dysfunction induced by the acute event. The ECG manifestations of infarction also have an effect on risk. Although patients with a non-Q wave myocardial infarction have a higher rate of spontaneous reperfusion, smaller infarctions and lower in-hospital mortality than those with a Q wave myocardial infarction, they tend to have a higher rate of late (subsequent) ischemic events (27). Treatment that reestablishes Thrombolysis in Myocardial

Infarction (TIMI) grade 3 coronary blood flow and restores patency, whether by thrombolytic agents or direct percutaneous transluminal coronary angioplasty, reduces risk substantially compared with patients in whom reperfusion has not been established (28-31). Severity of angina. Stable angina. The clinical history and physical examination remain mainstays of the initial evaluation of patients with coronary artery disease. A description of classic exertional angina provides a prognostic accuracy that is hard to exceed with more complex technologies. At the end of the initial evaluation, a decision must be made about the patient's short-term risk of adverse events, which dictates the type and intensity of additional evaluations. Characteristics from the initial examination can predict both the extent of coronary artery disease and survival (17,18). Anginal symptoms contribute significantly to the prediction of multivessel disease and 3-year survival. In fact, the prevalence of multivessel disease increases with age and the occurrence of typical

1012

CALIFFET AL. TASK FORCE 5

JACC Vol. 27, No. 5 April 1996:964-1047

Table 4. Risk of Coronary Artery Disease Event, Stroke or Cerebrovascular Disease Death in Men With Existing Coronary Artery Disease Age (yr)

Points

35 40 45 50 55 60 65 70 75

0 1 1 2 2 3 3 4 4

Other Diabetes

Pts 1

Points by HDL-C (mg/dl)

Total-C (mg/dl)

25

30

35

40

45

50

60

70

80

160 170 180 190 200 210 220 230 240 250 260 270 280 290 300

6 6 7 7 7 7 8 8 8 8 8 9 9 9 9

5 5 6 6 6 6 7 7 7 7 7 8 8 8 8

4 5 5 5 5 6 6 6 6 6 7 7 7 7 7

4 4 4 4 5 5 5 5 6 6 6 6 6 7 7

3 3 4 4 4 4 5 5 5 5 5 6 6 6 6

2 3 3 3 4 4 4 4 4 5 5 5 5 5 6

1 2 2 2 3 3 3 3 4 4 4 4 4 4 5

1 1 1 2 2 2 2 3 3 3 3 3 4 4 4

0 0 1 1 1 1 2 2 2 2 2 3 3 3 3

Total Points 0 2 4 6 8 10 12 14 16 18 20 22 24

2-yr Probability (%.) 2 2 3 5 7 10 14 20 28 37 49 63 77

SBP (ram Hg)

Points

100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

Average 2-yrRisk in Men With CVD Age (yr)

Probability (%)

35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74

<1 8 10 11 12 12 14 14

CVD = cerebrovasculardisease;other abbreviationsas in Table 1.

exertional anginal pain symptoms. Furthermore, the prognostic weight for typical angina contributes more than age, gender or E C G conduction defects (17). In patients with established angina, the characteristics of the anginal symptoms can be related to subsequent risk of an event (17). This risk has been quantified, characterizing the course as stable or progressive and measuring the frequency of the discomfort and the presence of ST segment changes on the E C G (32). An angina score has been constructed that quantifies the probability of death as a function of the characterization of the angina and the recent history of myocardial infarction. Table 5 provides the relative risk associated with each chest pain characteristic; these values (points) are used in the calculation of 1-year mortality (shown later in Table 7). Depending on the variable, this risk may be additive or multiplicative when more than one characteristic is present. Unstable angina. Patients with unstable angina constitute about 10% of all initial presentations with coronary artery disease. A recent clinical practice guideline (24) has synthe-

sized the published reports into empirical guidelines for risk stratification of patients with unstable angina. The clinical criteria for classification as well as for stratification of shortterm risk of death or nonfatal myocardial infarction in patients with unstable angina are shown in Table 6. Rest ECG. The presence of the changes from a prior myocardial infarction or a conduction disturbance on the routine 12-lead ECG, or both, provides independent prognostic information. In particular, left bundle branch block and nonspecific intraventricular conduction delay provide greater prognostic accuracy, but right bundle branch block and hemiblocks also provide independent predictive information (33). Congestive heart failure. Patients with congestive heart failure account for over 400,000 new ischemic events each year, including at least 200,000 deaths. In particular, patients with heart failure with left ventricular dysfunction (ejection fraction <0.40) have a reduced survival that has been shown in multiple large clinical trials (34-36); survival is inversely proportional to the degree of functional impairment. In the CASS experience,

J A C C Vol. 27, No. 5 April 1996:964-1047

25

C A L I F F E T AL. TASK FORCE 5

Table 5. Relative Risk Based on Anginal Characteristics

I I~

--

- Unstable

--Acute

angina

L

.....

"_ ~..-'-. ~ ~

0

~ ~

Nonanginal pain Atypical angina Stable angina Progressive Unstable

3 25 41 46 51

0.4 0.8 1.3 1.5 1.7

~ =~

I"

I

1

2 Months

-f

~ "1"-'~- ~ = ~ ' " 3

.....

4

F

I

5

6

since hospital admission

Figure1. Risk of future cardiac events as a functionof time since the event or procedure. MI = myocardialinfarction.

the 12-year survival was dramatically affected by the presence of heart failure at baseline (16). Survival was 81% for no heart failure, 71% for functional class I, 54% for class II, 25% for class III and 11% for class IV. Over 50% of patients in functional class IV die within 5 years; a large majority of these deaths are sudden (34-37). In addition to symptomatic heart failure, the presence of an $3 gallop on physical examination or cardiomegaly on the chest radiograph adds independent prognostic information. Peripheral and cerebral vascular disease. Angiographically significant and often severe coronary artery disease is prevalent among patients with aortic and peripheral atherosclerosis (38-41). Even in the presence of a given extent of coronary artery disease and left ventricular dysfunction, patients with clinical manifestations of peripheral vascular disease have a higher risk of death than patients without peripheral vascular disease (18). Of 30,411 patients with coronary disease followed at the Duke Databank, annual mortality for patients with a concomitant history of cerebrovascular disease is 5.1%, a rate some 25% higher than for those with coronary artery disease alone. Figure 2. Risk of occlusion due to thrombosis or mechanical obstruction as a function of time from coronary artery bypass surgery. Reprinted from Smith et al. (26), with permission.

01 J= - -

Total hazard

o01 -

~_

- -

Early hazard

- - -

Late hazard

o.OOl -

00001

1-yr Mortality Rate (%)*

*For a 60-yearold patient with no comorbidconditions.

5-

0

D "6

Points

MI

15.

8 10. ~ D

Characteristic

- - - Stable angina

2o •

~

1013

I

P

I

I

I

I

I

I

I

I

2

4

6

8

10

12

14

16

18

20

Time since bypass surgery (years)

Summary risk. These features of the clinical history and physical examination can be summarized as shown in Table 7. These data are analogous to those in Table 1 from Framingham, except that clinical features of patient characterization predominate. Diagnostic and Prognostic Testing At the time of risk assessment, a test to evaluate functional capacity or the response to a provocation such as exercise or pharmacologic stress may be included. Regardless of other available options, the history, physical examination and ECG remain the cornerstones of any prognostic eval" tion. In addition, the physician and patient may choose J obtain routine stress testing, provocative testing with imaging for ischemia or left ventricular function, rest imaging for left ventricular function or coronary angiography with left ventriculography. The goal of this section is not to recommend any specific strategy of testing, but to provide information about prognosis if a test is chosen or may be available from a recent previous evaluation. Left ventricular function. Rest left ventricuiar function is a powerful determinant of subsequent patient outcome. The importance of the rest ejection fraction by radioactive techniques was confirmed for chronic stable angina in the Veterans Administration Cooperative Study (42) and the CASS study (16) and for acute myocardial infarction in the Multicenter Postinfarction Research study groups (43) and others (44). These studies consistently have found a nonlinear relationship between systolic left ventricular function measured by left ventricular ejection fraction and death; the risk is particularly accelerated with each decrease in ejection fraction below 40% (Fig. 3) (45). Noninvasive testing modalitics. For many patients with coronary artery disease, stress testing with exercise or pharmacologic agents can provide a useful and powerful supplement to the initial clinical evaluation. However, limitations of the noninvasive testing literature include a paucity of prospective patient series and insufficient sample sizes. Up to one-half of all noninvasive testing reports are based on sample sizes insufficient to test their primary hypothesis. Additionally, 90% of all published reports are nonprospective series and subject to selection and reviewer biases. It is often impossible to compare results of case series because of the major differences in baseline characteristics between patients referred for one

1014

Table

CALIFF ET AL. TASK F O R C E 5

JACC Vol. 27, No. 5 April 1996:964-1047

6. Risk Classification for Patients With Unstable Angina Low Risk (<1%)

Intermediate Risk (]-4%)

High Risk (>4%)

No intermediate risk features but must have any of No high risk features but must have any of the following: the following: Prolonged (->20 min) ongoing rest angina, now Increased angina frequency, severity or duration resolved with moderate or high likelihood Angina provoked at a lower threshold of CAD New-onset angina with onset 2 wk to 2 mo Rest angina (->20 min or relieved with rest or before presentation SL NTG) Nocturnal angina Angina with dynamic T wave changes New-onset CCSC III or IV angina in past 2 wk with moderate to high likelihood of CAD Pathologic Q waves or rest ST depression <1 mm in multiple lead groups (anterior, inferior, lateral) Age >65 yr

At least one of the followingfeatures must be present: Prolonged (->20 min), ongoing rest pain Pulmonary edema, ischemic in origin Angina at rest with dynamic ST changes (1+ mm) Angina with new or worsening MR murmur Angina with $3 or new/worsening rales Angina with hypotension

CAD = coronary artery disease; CCSC = Canadian Cardiovascular Society classification;MR= mitral regurgitation; SL NTG = sublingual nitroglycerin. testing modality versus another. The most commonly cited comparative estimates, sensitivity and specificity, are often derived from small subgroups of the study populations. Details about patients who did not proceed to cardiac catheterization or those who were lost to follow-up (i.e., ascertainment bias) are frequently ill-described. In summary, based upon the results of a thorough synthesis of published noninvasive data, it is difficult to know the true diagnostic or prognostic accuracy of many noninvasive testing modalities used in current practice.

Table

7. Risk of Mortality at 1 Year: Clinical History Variables

1. Find Points for Each Risk Factor: Age (yr) Points 20 30 40 50 60 70 80 90 100

0 13 25 38 50 62 75 88 100

Angina (pain type) Nonanginalpain Atypical angina Typicalangina Stable Progressive Unstable

Points 3 25 41 46 51

Comorbid Factor

Points*

CVD PVD Diabetes Prior MI Hypertension Mitral regurgitation Mild Severe

20 23 20 17 8

-

Age

+

-

-

Figure 3. Risk of mortality at 2 years as a function of left ventricular ejection fraction. Reprinted from Morris (45), with permission.

+

Pain score

Duration of exercise (rain) [Maximal ST segment deviation (mm) during or after exercise x 5] (Treadmill angina index x 4),

19 38

2. Sum Points for All Risk Factors: -

Exercise ECG. A standard treadmill test is routinely performed in patients who are not taking digoxin or other drugs that can affect ST segment analysis, who are able to exercise and who have a normal rest ECG. Studies have shown that the response to exercise permits assessment of the severity of underlying coronary artery disease and the risk of death. Several features have been associated with adverse prognosis: duration of exercise, number of leads with and amount of ST depression, exercise-induced angina, maximal exercise heart rate, drop in systolic blood pressure and ventricular arrhythmia. A treadmill score was developed that uses three of these variables to predict 4-year survival in patients with suspected coronary artery disease (46). The treadmill score may be calculated by the following formula:

Comorbidity

Point total

60

3. Look Up Risk Corresponding to Point Total: Total Points 84 106 120 136 160 184

Probability of 1-yr Death 1% 2% 3% 5% 10% 20%

~

Exercise

- - -

Rest

50

Total Points

Probability of 1-yr Death

199 211 220 229

30% 40% 50% 60%

A =~ 4 0 .

~" &

20'

10-

0

*Zero points for each "no." CVD = cerebrovascular disease; MI = myocardial infarction; PVD = peripheral vascular disease.

10

I

I

I

I

I

I

I

20

30

40

50

60

70

80

Ejection fraction (%)

JACC Vol. 27, No. 5 April 1996:964-1047

CALIFFET AL. TASKFORCE5

Table 8. Survivalat Four Years According to Treadmill Score

Treadmillscore Survival Inpatients Outpatients

Low Risk

IntermediateRisk

High Risk

More than +5

-10 to +4

Below-10

98% 99%

92% 95%

71% 79%

results reveals that patients with high risk scans are more likely to have multivessel disease and worse survival. The accuracy of peffusion imaging may be further enhanced by the use of quantitative single-photon emission computed tomographic (SPECT) imaging or by the use of new imaging agents, such as technetium-99m sestamibi. Technetium-99m sestamibi provides enhanced resolution, with similar diagnostic and prognostic accuracy, compared with thallium-201 imaging. Wall motion and left ventricular function can also be simultaneously obtained. It may be particularly useful for female or obese patients, whose breast tissue may interfere with imaging interpretation. Although there are recognized imaging and radiotracer differences between thallium-201 and technetium-99m, predictive modeling with technetium-99m has demonstrated that the presence of an exercise sestamibi perfusion defect independently predicts 1-year death or nonfatal myocardial infarction in patients with stable coronary artery disease (50). Prediction of risk with noninvasive testing. Although considerable information has been accrued on prediction of risk based upon a patient's clinical risk profile, limited data are available to the clinician on the likelihood of disease or subsequent cardiac death as determined by stress myocardial perfusion imaging. Nomograms for the prediction of significant coronary disease (at least one artery with 70% stenosis), severe disease (three-vessel or left main involvement) and cardiac death are presented in Figures 4 to 6. The level of incremental risk of severe disease provides insight into the use of these technologies to predict mortality. Figures 4 and 5 provide predictive information derived from stress myocardial perfusion imaging. These estimates are based on a group of 1,800 patients undergoing cardiac catheterization and exercise or pharmacologic stress SPECT imaging from 1991 to 1995 at Duke University Medical Center. These nomograms allow a physician to estimate the incremental value of testing in populations that have different pretest clinical risk levels. Overall, stress myocardial perfusion imaging

where the angina index is as follows: no angina during exercise = 0; nonlimiting angina = 1; exercise-limiting angina = 2. The 4-year survival of patients in low, intermediate and high risk categories is shown in Table 8. Ambulatory ECG monitoring. Ambulatory ischemia has also been shown to be an independent, incremental marker of risk in patients with stable and unstable angina as well as in post-myocardial infarction patients (47-49), although inadequate numbers of patients have been evaluated to allow specific depiction of risk level in a figure. Stress imaging techniques. Patients with widespread ST depression (> 1 mm), left ventricular hypertrophy, intraventricular conduction defects (e.g., left bundle branch block) or pre-excitation are more often referred for imaging. The type of imaging agent may allow further definition of dyssynergy or left ventricular limitations or assessment of flow limitations in one or more lesions. Stress myocardial perfusion imaging. Transient perfusion defects that "fill in" on the delayed images are consistent with exercise-induced myocardial ischemia. A high risk scan may be defined as the presence of multiple perfusion defects, increased lung uptake or left ventricular dilation. The presence of multiple perfusion defects is more closely associated with multivessel disease. Scarred or fixed defects characteristically reflect infarcted tissue, although they may become ischemic after reinjection in some 30% of thallium-201-tested patients. Although the data have been derived from small samples and may be considered observational, a synthesis of research

Points

o

1o

,

Pretest Probability of CAD 30% Clinical History Score -;

20

,

32%

30

,

35%

Figure 4. Nomogram for predic-

40

I

m

42%

50

I

57%

m

80%

I

60 m

93%

I

95%

70 m

I

80 m

I

90 m

I

100 m

[

>99%

S c o r e 7 P o i n t s for each F i x e d D e f e c t (Up to 5 Defects) S c o r e 8 P o i n t s for E a c h R e v e r s i b l e D e f e c t (Up to 5 Defects)

tion of significant coronary artery disease (CAD), defined as at least one vessel with 70% stenosis, with stress myocardial perfusion imaging.

Total Sum of Points

Total Points

1015

.

• 20

o

Risk of Significant Disease .

0.3

014

, 40

0"S

016

.

0~7

, 60

, 80

018

1(30

019

120

140

-I+

--

160

180

1016

CALIFF ET AL. TASK FORCE 5

JACC Vol. 27, No. 5 April 1996:964-1047

o

10

Points • Pretest Probability of C A D Clinical History -~ Score

30

40

.

.

30%

50

.

.

32%

60

70

80

I

[

.

35%

38%

47%

90 I

60%

I

100 I

I

80%

95%

Score 2 Points for each Fixed Defect (Up to 5 Defects) Score 5 Points for Each Reversible Defect

+ Figure 5. N o m o g r a m for prediction of severe coronary artery disease

+

(CAD), defined as three-vessel or left main coronary artery disease, with stress myocardial perfusion imaging.

(Up to 5 Defects)

Total S u m of Points Total P o i n t s

"o

l'o

do

Risk of Severe Disease

i0 " 4~

5'o

011

0.05

go

012

40

do

013

014

may contribute as much as 40% of the predictive information for significant coronary artery disease (Fig. 4) and 20% of the predictive information for severe disease (Fig. 5). Since testing is not expected to be of value in patients who have low pretest risk, pretest probability begins only in patients with a moderate to high likelihood of coronary artery disease. However, perfusion imaging likewise may not be helpful for patients who have a very high pretest clinical risk profile. As shown in Figure 4, if the pretest risk estimate exceeds 90%, the risk of disease increases from 70% to 90%. This incremental information might not be sufficient to alter patient management and thus would not be clinically meaningful. It should be noted that these estimates are based on catheterized patients; they may overestimate risk as compared with patients evaluated in a nuclear laboratory. Pharmacologic stress perfusion imaging. In patients unable to perform maximal stress tests, a number of pharmacologic stressors may be used, including dipyridamole, adenosine and dobutamine. No evidence exists that one is superior, although dipyridamole has been used for more than 10 years and has a

0

Points

10

20

30

40

50

60

. . . . . . . . . . . . . .

Pretest Probability of S u r v i v a l C l i n i c a l I n d e x -i.5

Ejection Fraction.

90

99%

98%

-'~

.~;.5

6

8'0

94%

0;s

7"0

85%

i

t~5

6"0

4o

16o

do

015

o.6

2~5

70

5'0

4'0

~

(kpm)

1100

10])0

91)0

800

700

660

5(30

360

2'0

1'o

6

200

~6o

6 m i

Total

Total Score .0

. 20 .

40

.

.

60

R i s k of One Year Mortality

.80 . 100 . 0[~5

.

120

oll

.

140

.160 . 012

180

017

+

375

3'0

400

1io

1oo

8O

÷ W o r k Load

15o

more substantial body of literature supporting its effectiveness as a diagnostic and prognostic tool. Risk stratification with dipyridamolc-thallium-201 perfusion imaging (51) indicates that the presence of a perfusion defect is able to provide independent prognostic information. Dobutamine stress echocardiography has become an increasing popular method for risk stratification (52,53). Dobutamine-induced ventricular wall motion abnormalities have been significantly associated with myocardial viability and restenosis in several observational patient series (53). In patients undergoing vascular surgery, the presence of new or worsening dyssynergy was associated with a 5- to 14-fold increase in risk of subsequent death or reinfarction (52). Exercise radionuclide angiography. Regardless of the radionuclide technique (first-pass or gated), a review of the published reports on exercise radionuclide angiography reveals that exercise ejection fraction is the best univariate and multivariable predictor of a cardiac event (54-60). In one large series, the continuous relationship between the risk of death and exercise ejection fraction had the same shape as the

64%

~

-

.

200

220

240

0'.3 0~4 015 0~6 0~7

260

Figure 6. N o m o g r a m for prediction of cardiac death at 1 year with exercise radionuclide angiography.

JACC Vol. 27, No. 5 April 1996:964-1047

CALIFF ET AL. TASK FORCE 5

Table 9. Coronary Artery Disease Prognostic Index

Extent of CAD

PrognosticWeight (0-100)

5-yr Mortality Rate (%)*

1-vessel disease, 75% >l-vessel disease, 50-74% 1-vessel disease, ->95% 2-vessel disease 2-vessel disease, both ->95% 1-vessel disease, ->95% proximal LAD 2-vessel disease, ->95% LAD 2-vessel disease, ->95% proximal LAD 3-vessel disease 3-vessel disease, ->95% in at least 1 3-vessel disease, 75% proximal LAD 3-vessel disease, ->95% proximal LAD

23 23 32 37 42 48 48 56 56 63 67 74

7 7 9 12 14 17 17 21 21 27 33 41

Exercise echocardiography. Stress echocardiography has been reported to have improved diagnostic accuracy over the exercise ECG alone. Stress echocardiography is not affected by breast attenuation (as is perfusion imaging), and it allows the evaluation of wall motion in multiple planes. Extent of coronary artery disease. One of the most important prognostic factors in patients with coronary artery disease is the extent and severity of underlying coronary artery disease (62). A number of prognostic indexes have been developed that relate the severity and extent of coronary artery disease to the subsequent risk of adverse outcomes. The simplest depiction of the extent of disease into one-vessel, two-vessel, three-vessel or left main disease has provided a useful clinical shorthand; recent long-term follow-up from the CASS study indicates the continued utility of this simple classification. Further evaluations have led to the clinically obvious conclusion that proximal coronary stenoses are more important for prognosis than distal coronary lesions. This concept is the basis of the "jeopardy score" (63), in which the prognostic significance of lesions is weighted as a function of location within the vessel. More recently, Mark et al. (64) proposed a new prognostic coronary artery disease index (Table 9). This hierarchical index takes into account information about the lesion severity and location with prognostic weights ranging from 0 (no coronary artery disease) to 100 (95% left main stenosis). The index was developed by specifically analyzing the relationship between the location of the lesions and the risk of cardiac death in medically treated patients. This index is able to stratify patients who appear to have a substantial survival benefit from revascularization strategies (64-66). Figure 7 depicts a nomogram that integrates information from the history, physical examination, and cardiac catheterization findings for prediction of the risk of death.

*Assuming medical treatment only. CAD = coronary artery disease; LAD = left anterior descending coronary artery.

mortality-rest ejection fraction relationship, but the exercise ejection fraction was a more powerful predictor (56). Multiple studies have shown that patients with exercise ejection fractions <40% have very low rates of survival (56,59). The summary odds of a cardiac event are fourfold higher for patients with an abnormal peak radionuclide angiogram versus those with a normal examination (61). In patients with a normal ejection fraction, the presence of ventricular wall motion or functional abnormalities during exercise has been associated with a worsening 4-year survival (61). Figure 6 details the results of 780 patients undergoing gated equilibrium radionuclide angiography at Duke University. In this model, 1-year prognosis is best predicted by a combination of peak bicycle work load, exercise ejection fraction and the patient's pretest clinical risk. In this case, the combination of ejection fraction and functional capacity data provides substantial improvement over clinical history estimates.

0

1017

10

20

30

40

50

60

70

80

90

100

Points Age (in years) Ejection Fraction

20 7*5

25 7~0

3(I

35

6'5

40 6'0

45 5'5

CAD Index

Figure 7. Nomogram for prediction of

5-year survival from clinical, physical examination and cardiac catherization findings. Asymp = asymptomatic; CAD = coronary artery disease; MI = myocardial infarction; Symp = symptomatic.

50

55

6~)

5'0

4'5

40

3;

5t6

65 6~

23237 42 48

65

70

75 30 "]4

gO 25

85

90

2'0

1'5

93

-1-

lJO

;

82

"~

New MI/Unstable Angina Score

"J"

0--Asymptomatic 26--6-24hrsPost MI Sex: Score 3 Points for Females 4--Stable Angina 31--Under 6hrs Post M[ 9--Progressive angina Mitral regurgitation: Score 4 Points per Level 13--Unstable Angina i7--Asymp Post MI 22--Symp Post M1 Vascular Disease Index: Score 10 Points per Level

"l"

Congestive Heart Failure: Score 18 Points for Any

"l"

Comorbid Conditions: Score 16 Points for Any

"l"

New M1/Unstable Score:

"1" "1"

Total Points = Total Points 5 Year Survival

60

80 ' 100 ' 120 ' 1~10 ' 160 ' 180 ' 200 ' 220 ' 240 ' 260 .99

.98

.95

.9

.8

.7 .6 .5 .4 .3

.i .(J5 .01

1018

CALIFF ET AL. TASK FORCE 5

Summary and Future Work As detailed in this task force report, considerable observational and clinical trial data have been accrued to identify risk markers from the initial examination and from a variety of noninvasive test results. Substantial information is available regarding the risk of future coronary artery disease events in populations without clinically evident coronary artery disease, based on risk factors readily available from simple evaluations (age, blood pressure) and laboratory measurements (lipids). Less is known about the complex interplay of symptoms, risk factors, coronary anatomy and results of additional testing in patients who have established coronary artery disease. A case in point is the incremental impact of noninvasive test results beyond the information obtained from a patient's clinical history. A negative test may provide a great deal of reassurance to the patient as well as information regarding enhanced survival. Further, patients with a high risk stress test (e.g., multiple perfusion defects, ejection fraction <0.35) have worse survival. However, noninvasive testing is of value only if it adds substantially to the clinical history and thus would improve clinical decision making. Although some information is available to document the independent prognostic value of noninvasive testing beyond the initial clinical examination, integration of this information in real time to determine when further testing may be useful or when it may be unhelpful in further risk stratification remains a challenge (56,61,67). The incremental cost of additional prognostic information is another issue that merits investigation. In many circumstances, a less expensive evaluation may provide similar information. The value of a simple clinical prediction rule should not be underestimated. Clinically based models have been equally able to predict outcome or the results from noninvasive testing, or both, thus obviating the need for further risk stratification from subsequent procedures (18,62,68). There are many unresolved issues regarding not only the independence of risk measures but also their clinical relevance in populations where indecision is greatest. These issues, which affect daily clinical decision making, must be addressed in the future with more rigorous methodologies.

References 1. American Heart Association. Heart and Stroke Facts. Dallas: American Heart Association, 1994. 2. Siegel D, Grady D, Browner WS, Hulley SB. Risk factor modification after myocardial infarction. Ann Intern Med 1988;109:213-8. 3. Haskell WL, Alderman EL, Fair JM, et al. Effects of intensive multiple risk factor reduction on coronary atherosclerosis and clinical cardiac events in men and women with coronary artery disease. The Stanford Coronary Risk Intervention Project (SCRIP). Circulation 1994;89:975-90. 4. Goldman L, Gordon D J, Rifkind BM, et al. Cost and health implications of cholesterol lowering. Circulation 1992;85:1960-8. 5. Mark DB. Assessment of prognosis in patients with coronary artery disease. In: Roubin GS, Califf RM, O'Neill WW, Phillips HR III, Stack RS, editors. Interventional Cardiovascular Medicine. Principles and Practice. New York: Churchill Livingstone; 1994:165-85. 6. Gordon T, Kannel WB. Multiple risk functions for predicting coronary heart disease: the concept, accuracy, and application. Am Heart J 1982;103: 1031-9.

JACC Vol. 27, No. 5 April 1996:964-1047 7. Wolf PA, D'Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham study. Stroke 1991;22:312-8. 8. Smith KW, McKinlay SM, McKinlay JB. The validity of health risk appraisal for coronary heart disease: the results from a randomized field trial. Am J Public Health 1991;81:466-70. 9. Laurier D, Chau NP, Cazelles B, Segond P, PCV-METRA Group. Estimation of CHD risk in a French working population using a modified Framingham model. J Clin Epidemiol 1994;47:1353-64. 10. Gray D, Hampton JR, Shaw LK, Bernstein SJ, Pryor DB. Successful application of a predictive model of coronary diseases [abstract]. Circulation 1992;86 Suppl I:I-4. 11. Truett J, Cornfield J, Kannel W. A multivariate analysis of the risk of coronary heart disease in Framingham. J Chronic Dis 1967;20:511-24. 12. Daley J, Shwartz M. Developing risk-adjustment methods. In: lezzone LI, editor. Risk Adjustment for Measuring Health Care Outcomes. Ann Arbor (MI): Health Administration Press, 1994:199-238. 13. Hlatky MA, Califf RM, Harrell FE Jr, et al. Clinical judgment and therapeutic decision making. J Am Coil Cardiol 1990;15:1-14. 14. Selker HP, Griffith JL, D'Agostino RB. A time-insensitive predictive instrument for acute myocardial infarction mortality: a multicenter study. Med Care 1991;29:1196-211. 15. Kasper JF, Mulley AG, Wennberg JE. Developing shared decision-making programs to improve quality of health care. Qual Rev Bull 1992;18:183-90. 16. Emond M, Mock MB, Davis KB, et al. Long-term survival of medically treated patients in the coronary artery surgery study (CASS) registry. Circulation 1994;90:2645-57. 17. Pryor DB, Shaw L, Harrell FE Jr, et al. Estimating the likelihood of severe coronary artery disease. Am J Med 1991;90:553-62. 18. Pryor DB, Shaw LK, McCants CB, et al. Value of the history and physical in identifying patients at increased risk for coronary artery disease. Ann Intern Med 1993;118:81-90. 19. Anderson KM, Wilson PWF, Odell PM, Kannel WB. An updated coronary risk profile: a statement for health professionals. Circulation 1991;83:35662. 20. Gordon T, Garcia-Palmieri M, Kagn A, Kannel WB, Schiffman J. Differences in coronary heart disease in Framingham, Honolulu, and Puerto Rico. J Chronic Dis 1974;27:329-44. 21. Leaverton PE, Sorlie PD, Kleinman JC, et al. Representativeness of the Framingham risk model for coronary heart disease mortality: a comparison with a national cohort study. J Chronic Dis 1987;40:775-84. 22. Kannel WB, D'Agostino RB, Belanger AJ. Update on fibrinogen as a cardiovascular risk factor. Ann Epidemiol 1992;2:457-66. 23. D'Agostino RB, Wolf PA, Bclanger AJ, Kannel WB. Stroke risk profile: adjustment for antihypertensive medication. Stroke 1994;25:40-3. 24, Braunwald E, Mark DB, Jones RH, et al. Unstable Angina: Diagnosis and Management. Rockville (MD): Agency for Health Care Policy and Research, 1994. 25, Gilpin EA, Koziol JA, Madsen EB, Henning H, Ross J Jr. Periods of differing mortality distribution during the first year after acute myocardial infarction. Am J Cardiol 1983;52:240-4. 26, Smith LR, Harrell FE, Jr., Rankin JS, et al, Determinants of early versus late cardiac death in patients undergoing coronary artery bypass graft surgery. Circulation 1991;84: Suppl III:II1245-53. 27. Gibson RS. Non-Q-wave myocardial infarction: diagnosis, prognosis and management. Curr Probl Cardiol 1988;8:1-72. 28. Schroder R, Neuhaus KL, Linderer T, Bruggemann T, Tebbe V, Wegscheider K. Impact of late coronary artery reperfusion on left ventricular function one month after acute myocardial infarction: results of the ISAM study. Am J Cardiol 1989;64:878-84. 29. The GUSTO Angiographic Investigators. The effects of tissue plasminogen activator, streptokinase, or both on coronary-artery patency, ventricular function, and survival after acute myocardial infarction. N Engl J Med 1993;329:i615-22. 30~ Fibrinolytic Therapy Trialists (FIT) Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative overview of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Lancet 1994;343:311-22. 31. Grines CL, Browne KF, Marco J, et al. A comparison of immediate angioplasty with thrombolytic therapy for acute myocardial infarction. The Primary Angioplasty in Myocardial Infarction Study Group. N Engl J Med 1993;328:673-9.

JACC Vol. 27, No. 5 April 1996:964-1047 32. Califf RM, Mark DB, Harrell FE Jr, et al. Importance of clinical measures of ischemia in the prognosis of patients with documented coronary artery disease. J Am Coil Cardiol 1988;11:20-6. 33. Bounous EP Jr, Califf RM, Harrell FE Jr, et al. Prognostic value of the simplified Selvester QRS score in patients with coronary artery disease. J Am Coil Cardiol 1988;11:35-41. 34. The SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N Engl J Med 1991;325:293-302. 35. The CONSENSUS Trial Study Group. Effects of enalapril on mortality in severe congestive heart failure. N Engl J Med 1987;316:1429-35. 36. Cohn JN, Johnson G, Ziesche S, et al. A comparison of enalapril with hydralazine-isosorbide dinitrate in the treatment of chronic congestive heart failure. N Engl J Med 1994;325:303-10. 37. Califf RM, Bounous P, Harrell FE, et al. The prognosis in the presence of coronary artery disease. In: Braunwald E, editor. Congestive Heart Failure: Current Research and Clinical Application. New York: Grnne and Stratton, 1982:31-40. 38. Reigel MM, Hollier I.H, Kazmier FJ, et al. Late survival in abdominal aortic aneurysm patients: the role of selective myocardial revascularization on the basis of clinical symptoms. J Vasc Surg 1987;5:222-7. 39. Crawford ES, Bomberger RA, Glaeser DH, Saleh SA, Russell WL. Aortoiliac occlusive disease: factors influencing survival and function following reconstructive operation over a twenty-five year period. Surgery 1981;90: 1055-67. 40. Gersh BJ, Rihal CS, Rooke TW, Ballard DJ. Evaluation and management of patients with both peripheral vascular and coronary artery disease. J Am Coil Cardiol 1991;18:203-14. 41. Ashton CM, Petersen NJ, Wray NP, et al. The incidence of perioperative myocardial infarction in men undergoing noncardiac surgery. Ann Intern Med 1993;118:204-10. 42. Detre KM, Peduzzi P, Hammermeister KE, Murphy ML, Hultgren HN, Takaro T. Five-year effect of medical and surgical therapy on resting left ventricular function in stable angina: Veterans Administration Cooperative Study. Am J Cardiol 1984;53:444-50. 43. Multicenter Post-Infarction Research Group. Risk stratification and survival after myocardial infarction. N Engl J Med 1983;309:321-36. 44. Sanz G, Castaner A, Betriu A, et al. Determinants of prognosis in survivors of myocardial infarction: a prospective clinical angiographic study. N Engl J Med 1982;306:1065-70. 45. Morris KG. Use of radionuclide angiography following acute myocardial infarction. In: Califf RM, Mark DB, Wagner GS, editors. Acute Coronary Care. 2nd ed. St. Louis: Mosby-Year Book, 1995:797-813. 46. Mark DB, Shaw L, Harrell FE Jr, et al. Prognostic value of a treadmill exercise score in outpatients with suspected coronary artery disease. N Engl J Med 1991;325:849-53. 47. Marmur JD, Freeman MR, Langer A, Armstrong PW. Prognosis in medically stabilized unstable angina: early Holter ST-segment monitoring compared with predischarge exercise thallium tomography. Ann Intern Med 1990;113:575-9. 48. Goodman SG, Freeman MR, Armstrong PW, Langer A. Does ambulatory monitoring contribute to exercise testing and myocardial perfusion scintigraphy in the prediction of the extent of coronary artery disease in stable angina? Am J Cardiol 1994;73:747-52. 49. Langer A, Minkowitz J, Dorian P, et al. Pathophysiology and prognostic significance of Holter-detected ST segment depression after myocardial infarction. The Tissue Plasminogen Activator: Toronto (TPAT) Study Group. J Am Coll Cardiol 1992;20:1313-7. 50. Stratmann HG, Tamesis BR, Younis LT, Wittry MD, Miller DD. Prognostic value of dipyridamole technetium-99m sestamibi myocardial tomography in

CALIFF ET AL. TASK FORCE 5

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63. 64.

65.

66.

67.

68.

1019

patients with stable chest pain who are unable to exercise. Am J Cardiol 1994;73:647-52. Younis LT, Byers S, Shaw L, Barth G, Goodgold H, Chaitman BR. Prognostic importance of silent myocardial ischemia detected by intravenous dipyridamole thallium myocardial imaging in asymptomatic patients with coronary artery disease. J Am Coil Cardiol 1989;14:1635-41. Shaw LJ, Eagle KA, Gersh BJ, Miller DD. Meta-analysis of intravenous dipyridamole thallium-201 imaging (1985-1994) and dobutamine echocardiography (1991-1994) for risk stratification prior to vascular surgery. J Am Coll Cardiol 1996;4:787-98. Madu EC, Ahmar W, Arthur J, Fraker TD Jr. Clinical utility of digital dobutamine stress echocardiography in the noninvasive evaluation of coronary artery disease. Arch Intern Med 1994;154:1065-72. Jones RH, McEwan P, Newman GE, et al. Accuracy of diagnosis of coronary artery disease by radionuclide measurement of left ventricular function during rest and exercise. Circulation 1981;64:586-601. Johnson SH, Bigelow C, Lee KL, Pryor DB, Jones RH. Prediction of death and myocardial infarction by radionuclide angiocardiography in patients with suspected coronary artery disease. Am J Cardiol 1991;67:919-26. Lee KL, Pryor DB, Pieper KS, et al. Prognostic value of radionuclide angiography in medically treated patients with coronary artery disease: a comparison with clinical and catheterization variables. Circulation 1990;82: 1705-17. Upton MT, Palmeri ST, Jones RH. Assessment of left ventricular function by resting and exercise radionuclide angiocardiography following acute myocardial infarction. Am Heart J 1992;104:1232-43. Pryor DB, Harrell FE, Jr., Lee KL, et al. Prognostic indicators from radionuclide angiography in medically treated patients with coronary artery disease. Am J Cardiol 1984;53:18-22. Jones RH, Johnson SH, Bigelow C, et al. Exercise radionuclide angiography predicts cardiac death in patients with coronary artery disease. Circulation 1991;84 Suppl I:I-52-8. Morris KG, Palmeri ST, Califf RM, et al. Value of radionuclide angiography for predicting specific cardiac events after acute myocardial infarction. Am J Cardiol 1985;55:318-24. Kesler K, Shaw L, Heinle FK, Borges-Neto F, Jones R. Prognosis in patients with coronary artery disease by equilibrium radionuclide measurements of left ventricular function [abstract]. J Am Coil Cardiol 1996;27:286A. Pryor DB, Bruce RA, Chaitman BR, et al. Task Force I: determination of prognosis in patients with ischemic heart disease. J Am Coil Cardiol 1989;14:1016-25. Califf RM, Phillips HR, Hindman MC, et al. Prognostic value of a coronary artery jeopardy score. J Am Coll Cardiol 1985;5:1055-63. Mark DB, Nelson CL, Califf RM, et al. Continuing evolution of therapy for coronary artery disease: initial results from the era of coronary angioplasty. Circulation 1994;89:2015-25. Yusuf S, Zucker D, Peduzzi P, et al. Effect of coronary artery bypass graft surgery on survival: overview of 10-year results from randomised trials by the Coronary Artery Bypass Graft Surgery Trialists Collaboration. Lancet 1994;344:563-70. Califf RM, Harrell FE Jr, Lee KL, et al. The evolution of medical and surgical therapy for coronary artery disease. A 15-year perspective. JAMA 1989;261:2077-86. Mark DB, Hlatky MA, Harrell FE Jr, Lee KL, Califf RM, Pryor DB. Exercise treadmill score for predicting prognosis in coronary artery disease. Ann Intern Med 1987;106:793-800. Silver MT, Rose GA, Paul SD, O'DonneU CJ, O'Gara PT, Eagle KA. A clinical rule to predict preserved left ventricular ejection fraction in patients after myocardial infarction. Ann Intern Med 1994;121:750-6.