Introduction
This chapter is focused on tools for risk assessment in patients with stable coronary heart disease. In general, patients with stable ischemic heart disease have a good prognosis. However, these data summarize the population average, and the clinician is able to significantly refine the estimate of risk for the individual using methods described in this chapter.The central goal of risk assessment is to guide therapeutic decision-making and, in some cases, additional diagnostic evaluation. These diagnostic and prognostic assessments, although overlapping, are not identical. The prognostic assessment is valuable because the risk of recurrent events is strongly linked to the potential absolute and relative benefits of specific therapeutic interventions. In patients with stable coronary heart disease, an estimate of risk is similarly pivotal in management such as in identifying candidates for coronary angiography and revascularization. In this chapter we will review individual prognostic markers that are associated with adverse outcomes in stable coronary artery disease (CAD). We will also review multivariable models that incorporate multiple markers to quantitatively estimate risk and examine current approaches to match therapies to individual risk of an adverse outcome.
Prognosis Overall and in Subgroups
The assessment of cardiovascular disease risk and the prevention of recurrent events in patients with established CAD represent an opportunity for major public health gain. Aligning diagnostic studies and therapeutic interventions with clinical risk is a cornerstone of secondary prevention. Previous epidemiologic studies have demonstrated that established CAD is a major risk factor for incident events. For example, data from the Framingham Study, obtained before the widespread use of aggressive medication and modification of risk factors, revealed an average annual mortality rate of 4% in patients with stable CAD. Current therapies and management have improved the prognosis of the disease substantially, with an annual mortality rate of 1% to 3% and a rate of major ischemic events of 1% to 2%. In contemporary clinical trials, patients with stable CAD have an annual rate of major cardiovascular events of 1.2–2.4% per annum.
However, substantial heterogeneity in overall risk exists amongst patients with stable CAD with baseline cardiovascular risk factors, functional characteristics, and coronary anatomy each playing an important role. For example, in the international Reduction for Continued Health (REACH) Registry—which included asymptomatic adults with risk factors, patients with stable atherosclerosis, and individuals with prior ischemic events—large variations in cardiovascular risk between subgroups of patients were observed. Patients with a prior history of ischemic events at baseline had the highest rate of subsequent ischemic events (18.3%); patients with stable coronary, cerebrovascular, or peripheral artery disease had a lower risk (12.2%); and patients without established atherothrombosis but with risk factors only had the lowest risk (9.1%) during 4-year follow-up.
As might be expected, conventional risk factors for the development of CAD —hypertension, diabetes, smoking, hypercholesterolemia, obesity, and family history —each retain their prognostic value in the context of established CAD. The prognosis for patients with stable CAD is also worsened in patients with reduced left ventricular ejection fraction, by the severity and intensity of angina pectoris, with the presence of dyspnea, and by the presence of three-vessel disease or left main disease. The estimation of the long-term risk of adverse outcomes is crucial to effectively apply measures of secondary prevention and prevent overtreatment of patients at low risk of an adverse outcome, or under-treatment of patients at high risk of an adverse outcome.
Individual prognostic markers that are associated with adverse outcomes in stable CAD are summarized in Box 17.1 .
High Risk (> 3% Annual Risk for Death or Myocardial Infarction)
- 1.
Severe resting left ventricular dysfunction (LVEF < 35%) not readily explained by noncoronary causes
- 2.
Resting perfusion abnormalities involving ≥ 10% of the myocardium without previous known MI
- 3.
High-risk stress findings on the ECG, including
- •
≥ 2 mm ST-segment depression at low workload or persisting into recovery
- •
Exercise-induced ST-segment elevation
- •
Exercise-induced VT/VF
- •
- 4.
Severe stress-induced LV dysfunction (peak exercise LVEF < 45% or drop in LVEF with stress ≥ 10%)
- 5.
Stress-induced perfusion abnormalities encumbering ≥ 10% myocardium or stress segmental scores indicating multiple vascular territories with abnormalities
- 6.
Stress-induced LV dilation
- 7.
Inducible wall motion abnormality (involving more than two segments or two coronary beds)
- 8.
Wall motion abnormality developing at low dose of dobutamine (≤ 10 mg/kg per min) or at a low heart rate (< 120 beats/min)
- 9.
Multivessel obstructive CAD (≥ 70% stenosis) or left main stenosis (≥ 50% stenosis) on CCTA
Intermediate Risk (1–3% Annual Risk for Death or Myocardial Infarction)
- 1.
Mild to moderate resting LV dysfunction (LVEF of 35–49%) not readily explained by noncoronary causes
- 2.
Resting perfusion abnormalities involving 5–9.9% of the myocardium in patients without a history or previous evidence of MI
- 3.
ST-segment depression of ≥ 1 mm occurring with exertional symptoms
- 4.
Stress-induced perfusion abnormalities encumbering 5–9.9% of the myocardium or stress segmental scores (in multiple segments) indicating one vascular territory with abnormalities but without LV dilation
- 5.
Small wall motion abnormality involving one to two segments and only one coronary bed
- 6.
One-vessel CAD with ≥ 70% stenosis or moderate CAD stenosis (50–69% stenosis) in two or more arteries on CCTA
Low Risk (< 1% Annual Risk for Death or Myocardial Infarction)
- 1.
Low-risk treadmill score (score ≥ 5) or no new ST-segment changes or exercise-induced chest pain symptoms when achieving maximal levels of exercise
- 2.
Normal or small myocardial perfusion defect at rest or with stress encumbering < 5% of the myocardium ∗
∗ Although the published data are limited, patients with these findings will probably not be at low risk in the presence of either a high-risk treadmill score or severe resting LV dysfunction (LVEF < 35%).
- 3.
Normal stress or no change in limited resting wall motion abnormalities during stress
- 4.
No coronary stenosis > 50% on CCTA
CAC , Coronary artery calcium; CAD , coronary artery disease; CCTA , coronary computed tomography angiography; ECG , electrocardiogram; LV , left ventricular; LVEF , left ventricular ejection fraction; MI , myocardial infarction; VF , ventricular fibrillation; VT , ventricular tachycardia.
Prognosis in Subgroups
Coronary Artery Spasm
Although the pathophysiology is incompletely understood, known triggers for coronary vasospasm include smoking, electrolyte disturbances (potassium, magnesium), cocaine use, cold stimulation, autoimmune diseases, hyperventilation, or insulin resistance. The symptoms vary from silent myocardial ischemia to angina and even myocardial infarction. Long-term survival is usually good as long as patients are on calcium antagonists and avoid smoking. The incidence of cardiac death among patients with coronary artery spasm is up to 10% during 3 years of follow-up. The prognosis of vasospastic angina depends on the extent of underlying CAD and on disease activity (frequency and duration of spastic episodes), the amount of myocardium at risk, and the presence of severe ventricular tachyarrhythmias or advanced atrioventricular block during ischemia. The prognosis of vasospasm may be better in Japanese patients than patients of European ancestry, potentially due to differences in baseline characteristics, ascertainment of individuals of less severe disease, and fewer patients of Japanese ancestry having multivessel coronary spasm and/or reduced left ventricular function.
Women
Cardiovascular disease remains the leading cause of death in women and is responsible for 42% of premature deaths in women under the age of 75 years. Although coronary heart disease develops 5–10 years later in women than in men and women have historically been at lower risk for CAD, more recent data indicate that the prevalence of cardiac events in men is decreasing, whereas women are experiencing an increase in cardiac events, including myocardial infarction. Women have been underrepresented in cardiovascular clinical trials to date, representing 30% of participants in trials conducted since 2006, thus diminishing the quality of the evidence base available to guide therapy.The increasing recognition of heart disease in women is likely to stimulate key additional research in coming years. The considerable decline in mortality from CAD is mainly caused by population-level improvements in risk factors and by improvements in primary and secondary prevention.
Although the risk factors for CAD in women and men are similar, their distribution differs over time and between regions. Smoking seems to be associated with a higher relative risk of CAD in women than men, and the prevalence of hypertension increases more with age in women than men, resulting in higher rates of stroke, hypertrophy of the left ventricle, and diastolic heart failure. Diabetes is associated with a higher risk of CAD in women than in men. Previously, circulating estrogens were believed to have a beneficial effect on the risk of CAD, but exogenous hormone administration has not led to a similar effect.
Women and men of every age presenting with stable angina have increased coronary mortality relative to the general population, and several studies have indicated gender-related bias in care of both acute and chronic CAD. However, in a large international prospective population (CLARIFY) of outpatients with stable CAD, the rates for cardiovascular clinical outcomes were similar between men and women at 1-year follow-up.
Diabetes Mellitus (See Chapter 24 )
Diabetes mellitus doubles the risk of major cardiovascular complications in patients with and in patients without established cardiovascular disease, such that the majority of patients with diabetes die of cardiovascular diseases. Patients with angina and concomitant type 2 diabetes mellitus often have more diffuse and extensive CAD compared with those without type 2 diabetes mellitus. Furthermore, patients with CAD and type 2 diabetes may also have a greater burden of angina leading to a worsening prognosis. If diabetes mellitus is accompanied by other coronary risk factors or target organ damage, the patient is considered to be at very high risk and maximal preventive efforts are warranted. The control of risk factors appears to be efficacious in preventing future major adverse cardiovascular events in patients with stable CAD and diabetes mellitus. The clinical manifestations of cardiovascular disease in diabetic patients are similar to those in nondiabetic patients. In particular, angina, myocardial infarction, and heart failure are the most prominent clinical manifestations in patients with diabetes and tend to occur at an earlier age. The cardiac assessment of symptomatic ischemia in diabetic patients should follow the same indications as for patients without diabetes. The Bypass Angioplasty Revascularization Investigation 2 Diabetes trial showed that patients with CAD and diabetes had the same risk of cardiovascular events and mortality regardless of whether or not they had angina symptoms. Therefore, the management of CAD in these patients should not be predicated solely on the presence or absence of angina symptoms. However, routine screening for cardiovascular disease in asymptomatic patients is not currently recommended. The greater degree of plaque burden together with comorbidity (e.g., renal failure) and smaller distal vessels in patients with diabetes influence the prognosis and may guide the choice of coronary revascularization strategy.
Chronic Renal Failure
Chronic kidney disease is a risk factor for CAD and has a major impact on outcomes and therapeutic decisions within stable CAD. There are several risk factors in patients with chronic kidney disease that interact with the medical and diagnostic management of stable CAD and accelerate the development of CAD. Cardiovascular disease mortality is increased in patients with end-stage renal disease, therefore these patients should be monitored for symptoms suggestive of CAD. In CAD patients, the risk of sudden cardiac death is increased by 11% for every 10 mL/min decline in glomerular filtration rate. Myocardial perfusion imaging carries prognostic value in end-stage renal disease patients who are asymptomatic for CAD, although the screening for asymptomatic patients is not currently in routine clinical use. The work-up of suspected CAD in symptomatic patients with renal disease follows the same patterns as in patients with normal renal function. However, the presence of impaired renal function increases the pretest probability of CAD in patients who report chest pain, and noninvasive test results need to be interpreted accordingly. In addition, the use of iodinated contrast agent should be minimized in patients with preterminal renal failure and in dialysis patients with preserved urine production, in order to prevent contrast-induced nephropathy. Similarly, special attention should be paid to the drugs that are renally cleared and may need dose down-adjustment or substitution.
The same treatment options should be initiated in patients with CAD with or without renal insufficiency.Thus, treatment for risk modification should be initiated. However, mortality rates and the risk of complications are high in this type of patient compared to those without impaired renal function. In general, coronary bypass surgery is associated with higher procedural mortality and a greater likelihood of hemodialysis in nonhemodialysis-dependent patients after revascularization, while available studies suggest a trend toward better long-term survival, as compared with percutaneous coronary intervention (PCI).
Tools for Risk Assessment
Medical History (See Chapter 7 )
The approach to the patient with stable CAD starts with the medical history, in which several variables can provide important prognostic information and serve as an effective gatekeeper. Additional findings of heart failure or atherosclerosis in noncoronary vascular beds are associated with a poorer prognosis. Traditional models have estimated the likelihood of obstructive CAD rather than the risk of clinical events. The pattern and duration of chest pain, and the frequency of chest pain, in addition to traditional risk factors for atherosclerosis, confer prognostic information. In a study published in 2015, a model based on medical history–taking alone was able to identify a majority of patients with low risk (1%) of future clinical events (myocardial infarction and death) during 3 years of follow-up.
Resting Electrocardiogram (See Chapter 10 )
A normal resting electrocardiogram (ECG) is common in patients with stable CAD and may assist the clinician in differential diagnosis and defining the mechanisms of chest pain. Stable angina pectoris patients with an abnormal ECG are at greater risk of adverse outcomes than those with a normal ECG. A normal resting ECG suggests underlying normal left ventricular function whereas presence of a left bundle branch block on an ECG is associated with multivessel disease, impaired left ventricular function, and a poorer prognosis. ECG evidence of left ventricular dysfunction (left bundle branch block, nonspecific intraventricular conduction delays) is also a well-characterized indicator of adverse prognosis and increases the likelihood of future cardiac events almost twofold to fourfold.
Exercise Testing (Treadmill Test) (See Chapter 10 )
Exercise ECG is an important tool for risk stratification in patients with stable CAD. The exercise capacity is measured by maximum exercise duration, workload, and metabolic equivalent level. Maximum exercise capacity is one of the strongest prognostic markers and there are no major differences between the specific variables used to measure exercise capacity. The prognostic information is incorporated in the Duke treadmill score, which is well validated, and patients with a normal treadmill test have an excellent prognosis. The Duke treadmill score classifies patients into three risk groups: low, moderate, and high. Mean annual mortality is 0.25% in the low-risk group and 5% in the high-risk group. Cycle ergometry is an alternative to treadmill testing and is widely used in Europe. The work intensity can be adjusted by variations in resistance and cycling rate and is typically calculated in watts.
Echocardiography (See Chapter 11 )
Echocardiography has a range of uses in ischemic heart disease including diagnosis, risk stratification, and clinical decision-making. Quantitative indices of global and regional systolic function are also valuable in describing left ventricular function, determining prognosis, and evaluating treatment outcome. Measurement of left ventricular ejection fraction is useful for risk stratification and is a strong predictor of adverse outcomes. Reduced left ventricular ejection fraction is associated with a high risk of cardiovascular death. Echocardiography is also important for excluding other conditions such as significant valvular heart disease, pulmonary hypertension, or hypertrophic cardiomyopathy.The introduction of global longitudinal strain measurement may complement the traditional measurement of ejection fraction in the future, as the prognostic value of global longitudinal strain appears to be superior to that of ejection fraction for predicting major adverse cardiac events.
The sensitivity of stress echocardiography averages approximately 88% (range, 76–94%), and its specificity is 83% for detecting myocardial ischemia in patients with stable CAD and carries prognostic information. However, the diagnostic performance is dependent on the operator skills to obtain good image quality (adequate images can normally be obtained in more than 85% of patients, and the test is highly reproducible). The accuracy of stress echocardiography is in line with stress myocardial radionuclide perfusion imaging. A normal result portends a good prognosis, whereas an abnormal result indicates an increased risk of cardiac events.
Stress Perfusion Scintigraphy and Cardiac Magnetic Resonance Imaging (See Chapter 12 , Chapter 13 )
Myocardial perfusion imaging using single photon emission computed tomography (SPECT) is a useful tool in risk stratification of patients with stable CAD. On the one hand, among individuals with stable CAD and a normal stress imaging result, the annual cardiac mortality and myocardial infarction rate is similar to the general population. On the other hand, stress-induced reversible perfusion defects of greater than 10% of the total myocardium are associated with a poor prognosis. However, myocardial perfusion imaging has limited sensitivity for the detection of high-risk CAD, but a normal global coronary flow reserve (CFR) seems to be helpful in excluding the presence of high-risk CAD on angiography. A growing body of evidence supports the prognostic ability of absolute flow when quantified by cardiac positron emission tomography (PET) showing that intact CFR is associated with a favorable prognosis during follow-up of up to 5 years. Mechanistically, a reduced CFR leads to a worse prognosis either through a severe, focal defect and its future risk of plaque rupture with an acute coronary syndrome, or through a global flow reduction that serves as a marker for diffuse disease and overall CAD burden.
Evidence of the prognostic value of stress cardiac magnetic resonance and outcome is more limited but, in general, the same principles as for SPECT are shared. Thus, stress-induced reversible wall abnormalities of greater than 10% of the left ventricle are associated with a high-risk situation ( Fig. 17.1 ).
Coronary Computed Tomography Angiography (See Chapter 13 )
Noninvasive coronary computed tomography angiography (CTA) is very sensitive in detecting obstructive CAD, but is limited in its positive predictive value. Therefore, at present, the strength of CTA is its ability to exclude significant CAD with a high negative predictive value. In addition, CTA does not assess the functional significance of visualized lesions and often leads to further evaluation with either stress testing or invasive angiography, or both. However, new technologies with CTA are now available to estimate the functional significance of individual coronary lesion flow (fractional flow reserve).
Risk stratification using CTA is well established and large prospective international multicenter studies have demonstrated that the extent and severity of CAD is associated with all-cause mortality and demonstrated the independent prognostic value of both obstructive as well as nonobstructive CAD by CTA. However, the clinical event rate is very low in the absence of any coronary plaque or with plaque, but without stenosis ( Fig. 17.2 ).
Coronary Angiography (See Chapter 14 )
Coronary angiography provides important information both in the diagnosis of CAD and in assessing the risk of cardiovascular events. In the stable angina setting, coronary angiography provides information on the number of vessels involved. The severity is assessed by the overall number of lesions, lesion location, severity, and the extent of involvement of branch vessels.
The classification of disease into single-, double-, or triple-vessel or left main CAD is the most widely used and can be translated into prognostic information ( Fig. 17.3 ). The SYNTAX (Synergy Between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery) score extends this simple classification, provides a detailed risk assessment of the severity of epicardial CAD, and has been validated.
The angiogram is not always sufficient to characterize the coronary atheroma, and therefore advanced invasive imaging techniques, such as intravascular ultrasound and optical coherence tomography, are evolving as additional tools, but so far have not translated into prognostic information in patients with stable CAD.
The functional significance of a stenosis can be measured by fractional flow reserve (FFR). FFR is calculated as the ratio of distal coronary pressure to aortic pressure measured during maximal hyperemia. A normal value for FFR (> 0.80) indicates that a stenosis is not flow limiting and the prognosis is therefore excellent (< 1% risk of cardiovascular event).
Genetic Testing (See Chapter 3 )
Long known to be a heritable condition, genetic analyses have validated more than 50 loci across the genome that are independently related to risk of coronary disease. A genetic risk score based on a combination of 27 such variants was recently shown to predict risk of recurrent coronary events in participants from the CARE (Cholesterol and Recurrent Events) and PROVE-IT TIMI 22 (Pravastatin or Atorvastatin Evaluation and Infection Therapy–Thrombolysis in Myocardial Infarction 22) clinical trials. For example, those in the top quintile of the genetic risk score had a hazard ratio of 1.81 (95% confidence interval [CI] 1.22–2.67) for incident events. Furthermore, enhanced relative and absolute risk reduction was noted with statin/higher-intensity statin therapy within this subgroup. This finding has led to the hypothesis that identification of patients at increased genetic risk may allow for tailored therapy.
Multivariable Risk Prediction Models in Chronic Coronary Artery Disease
Informal methods of risk prediction, such as identification of clinical signs, symptoms, or biomarkers associated with adverse outcomes, have long been applied in the treatment of chronic CAD to identify individuals who may benefit from more intensive therapy. However, informal methods of risk stratification or use of single markers to predict risk have significant disadvantages. First, observed rates of adverse outcomes may vary substantially among individuals who have a prognostic marker associated with adverse outcomes due to the presence of other observed and unobserved risk factors. For example, both age and the presence of comorbidities such as diabetes and heart failure influence the prognosis of patients with chronic CAD. Second, the ability of clinicians to estimate the likelihood of patients’ outcomes using clinical signs and symptoms may be poorly related to patients’ observed risk of adverse outcomes.
Global risk scores that quantitatively estimate patients’ absolute risk of outcomes using multiple variables can avoid these disadvantages. Quantitative risk scores can allow physicians to predict patients’ absolute risk of adverse outcomes using greater information and with greater consistency than a single sign or symptom. Indeed, use of multivariable risk prediction models to guide therapy for the primary prevention of cardiovascular disease has become common and has been the focus of intensive research over the past two decades.
The Framingham Coronary Heart Disease risk model has been widely applied to predict risk of coronary heart disease among individuals without cardiovascular disease and has been recently extended to prediction of cardiovascular events including stroke. For the 2013 American Heart Association/American College of Cardiology (AHA/ACC) guidelines on the initiation of statin therapy, the Pooled Cohort equations were developed to predict cardiovascular risk and identify individuals at greater than 7.5% risk of a hard atherosclerotic cardiovascular event (cardiovascular death, myocardial infarction, or stroke) over 10 years for potential initiation of statin therapy. In the United Kingdom, the QRISK2 score has been developed to predict risk of a cardiovascular event and has also been incorporated into National Institute for Health and Care Excellence guidelines for the initiation of statin therapy. A 2011 systematic review on cardiovascular risk prediction identified more than one hundred different risk models aiming to predict incident cardiovascular disease.
In contrast to the intensive research into risk models for primary prevention of cardiovascular disease, research into risk prediction of cardiovascular disease among individuals with chronic CAD has been limited. Although the 2014 AHA/ACC and 2013 European Society of Cardiology guidelines on treatment of stable CAD recommend risk assessment of patients, neither recommend a multivariable risk assessment model, as the 2013 AHA/ACC guidelines for initiation of statin therapy do with the Pooled Cohort equations, likely due to the lack of an established global risk score. This section will review the comparatively limited research on multivariable risk prediction models in chronic CAD, including current multivariable risk models that use traditional risk factors, research into novel markers to improve risk prediction, and limitations of current models for risk prediction in chronic CAD.
Risk Prediction Using Multivariable Models
Although numerous statistical methods have been developed to characterize risk prediction models, they can be broadly characterized by measures of calibration and discrimination. Calibration refers to the ability of models to accurately predict the risk of an event observed over a period of follow-up. For example, if a model estimates the risk of an event in a group of participants to be 7% over a given period of follow-up, while only 3.5% of participants are observed to actually have an event, the model would be considered poorly calibrated. Calibration can be assessed by dividing participants into subgroups (often tenths of participants) and comparing the predicted risk in each subgroup to the observed risk. Calibration is often quantified using the Hosmer-Lemeshow chi-square statistic, which, if significant ( p < 0.05), indicates a lack of calibration. Discrimination, in contrast to calibration, refers to the ability of models to discriminate future cases from noncases. It is often quantified using the C-statistic, which refers to the probability that a randomly selected case has a higher predicted risk than a noncase. A risk model can be discriminatory, but not well calibrated, as Fig. 17.4 illustrates.
In early attempts to develop multivariable risk prediction models in CAD, quantitative measures of discrimination and calibration were inconsistently reported ( Table 17.1 ). For example, one of the first multivariable models to be developed for patients with CAD, published in 1988, was a risk score developed from a database of all patients undergoing cardiac catheterization at Duke. Stepwise selection was used to identify significant predictors of risk of death or myocardial infarction over a median 22 months of follow-up, with ejection fraction, number of diseased vessels, left main stenosis, angina score, age, and sex included in the final model. The dataset was divided into a training set, used to develop the model, and validation set, which was used to roughly examine the calibration of the final model. Kaplan-Meier curves for the model in the training set and the validation set overlapped, suggesting that the model was reasonably calibrated. However, quantitative measures of discrimination and calibration, such as the C-statistic or predicted versus observed event rate, were not reported.
| Name of Model (Author) | Year of Publication | Endpoint(S) | Population | Predictor Variables | Discrimination | Calibration | validated in an External Cohort? | Validated by External Researchers? | Other Limitations |
|---|---|---|---|---|---|---|---|---|---|
| Duke (Califf et al.) | 1988 | Death and nonfatal MI | 5886 participants with CAD | Demographic and clinical characteristics | Not assessed | Visually assessed through KM curves | No | No | Single center, baseline 1971 |
| LIPID (Marschner et al.) | 2001 | CHD death and nonfatal MI | 8557 participants with stable CAD and a history of myocardial infarction | Demographic and clinical characteristics | Visually assessed | Visually assessed | No | Yes, C-statistic = 0.61 | Baseline 1990 |
| TIBET (Daly et al.) | 2003 | Cardiac death, nonfatal MI, unstable angina | 682 participants with stable angina | Demographic, clinical, and noninvasive test variables | Not assessed | Not assessed | No | No | Baseline before 1995 |
| ACTION (Clayton et al.) | 2005 | Death, MI or stroke | 1063 participants with stable angina | Demographic, clinical, and noninvasive test variables | Visually assessed | Visually assessed | No | No | Baseline 1996 |
| Olmsted County (Miller et al.) | 2005 | 1.Death 2.Cardiac death 3.Cardiac death or nonfatal MI | 3546 participants undergoing stress testing for CAD | Demographic and clinical variables | Not assessed | Not assessed | No | No | Baseline 1987 |
| Euro Heart Angina (Daly et al.) | 2006 | 1. Death and nonfatal MI 2. Cardiovascular event | 3031 participants with stable angina | Clinical and test variables | C-statistic = 0.74 | Not assessed | No | No | |
| PEACE (Hsia et al.) | 2008 | Sudden cardiac death | 8290 participants with stable CAD | Demographic, clinical, and test variables | C-statistic = 0.71 | Not assessed | No | No | |
| Duke SCD (Atwater et al.) | 2009 | Sudden cardiac death | 37,258 participants with angiographic CAD | Demographic, clinical, and test variables | C-statistic = 0.75 | Visually assessed | Yes, C-statistic = 0.64 | No | Baseline 1985 |
| VILCAD (Goliasch et al.) | 2012 | Death | 547 participants with stable CAD | Demographic, clinical, and test variables | C-statistic = 0.77 | Not assessed | Yes, C-statistic = 0.73 | No | |
| EUROASPIRE (De Bacquer et al.) | 2013 | Cardiovascular death | 5216 participants with CAD | Demographic, clinical, and test variables | No | No | No | No | Baseline 1995 |
| EUROPA (Battes et al.) | 2013 | 1. Cardiovascular death 2. Cardiovascular death, nonfatal MI, and cardiac arrest Other endpoints | 12,218 participants with stable CAD | Demographic and clinical variables | C-statistic = 0.70 | Visually assessed | No | No | |
| CALIBER (Rapsomaniki et al.) | 2014 | 1. Death 2. CAD death or nonfatal MI | 102,023 participants with CAD | Demographic and clinical variables | C-statistic = 0.81 | Visually assessed | Yes, C-statistic = 0.74 | No |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
