Readers of this guideline who are not regularly involved in the management of patients with unstable angina should refer to the list of acronyms and glossary of medical terminology included with this guideline and the list of essential definitions on page x. However, even experienced clinicians will enhance their understanding and use of subsequent chapters in this document by a review of terms used to describe CAD presentations that are often blurred in common clinical usage. The authors of this guideline propose the following definitions in discussing the management of patients with unstable angina.
Throughout this guideline, unstable angina is broadly defined as a clinical syndrome that falls between stable angina and acute MI in the spectrum of presentations of CAD (Braunwald, 1989). The intent of the definition is to include all patients with acute presentations of CAD with the exclusion of only those patients with reperfusion-eligible acute MI. In most, but not all, of these patients, these symptoms will be caused by significant CAD. Since an adequate operational definition of unstable angina for this guideline must be based on the information available at the time of the initial presentation and evaluation, patients with a spectrum of underlying problems may receive a clinical diagnosis of unstable angina until additional diagnostic information becomes available (e.g., cardiac enzymes, noninvasive testing, or results of cardiac catheterization). The three principal presentations of unstable angina are listed in Table 1.
By design, this operational definition includes patients who will subsequently be found to have had an acute MI, as well as patients who will subsequently be found not to have significant coronary disease. The Canadian Cardiovascular Society Classification (CCSC) system is used throughout this guideline to grade the severity of anginal pain and discomfort (see Table 2).
Table 3 summarizes characteristics of five common clinical care environments for patients with unstable angina.
The strength of evidence grade for each recommendation within this guideline is followed by a brief discussion of the underlying rationale. Individual studies received ratings based on experimental design and overall quality. Randomized controlled trials received the highest ratings, other well-designed studies received a lower score, and studies with design or methodologic deficiencies received the lowest rating. The strength of evidence for each recommendation is summarized as an A, B, or C rating which most closely characterizes the total scientific literature available to address the topic (see Table 4).
Often, the most basic patient management questions and the most well-accepted care strategies are the most difficult to test. For example, no randomized clinical trials are likely to be conducted to evaluate the importance of a medical history and physical examination in patients with unstable angina. Therefore, the strength of evidence grade does not always reflect the importance of the recommendation to patient care. The specific language used to formulate each recommendation conveys panel opinion of both the clinical importance attributed to the topic and the strength of evidence available.
Proper application of the action-oriented recommendations made in this guideline assumes understanding of the basic disease process and familiarity with the common clinical presentation of patients with unstable angina. The background information in thissection provides an overview of the principles necessary to place the recommendations in subsequent chapters in a balanced clinical context.
CAD is the most important cause of death and disability in the United States. Only about 10 percent of patients with CAD have unstable angina as their initial presentation if patients who experience an MI are retrospectively excluded. However, patients with established CAD (either chronic stable angina or prior MI) commonly cycle through unstable phases. As a clinical syndrome, unstable angina shares ill-defined borders with chronic stable angina, a presentation with lower risk, and with acute MI, a presentation with higher risk. Unstable angina occurs in a variety of clinical scenarios, including in patients without known CAD, with prior stable CAD, soon after MI, and following myocardial revascularization by CABG or PTCA. Patients presenting with unstable angina may undergo any of the diagnostic and therapeutic procedures used for other CAD patients. Therefore, recommendations for the management of patients with unstable angina of necessity address questions pertinent to patients with any mode of presentation of CAD.
Despite the fact that death rates for CAD are decreasing ( Gillum and Feinlieb, 1988; Feinlieb, 1984), hospital discharge rates for this disorder appear to have stabilized since 1979 ( Feinlieb, Havlik, Gillum et al., 1989). The number of hospitalizations for which the principal diagnosis was unstable angina (ICD-9-CM 411.1) increased from 130,000 in 1983 to 570,000 in 1991 ( Graves, 1993; National Center for Health Statistics, 1985). Nearly 60 percent of persons admitted to the hospital with unstable angina as their primary diagnosis were older than age 65, and 46 percent were women. The number of hospitalizations for unstable angina is greater for men than women in all age groups under 75 years of age. The ratio appears to reverse between the ages of 75 to 84, and more women than men are hospitalized for unstable angina over the age of 85 ( Feinlieb, Havlik, Gillum et al., 1989). This reversal in hospitalizations reflects the larger representation of women than men in advanced age populations and is not due to a change in the relative incidence of CAD diagnosis in men and women in this age group.
Precipitating conditions for unstable angina may be those that increase myocardial oxygen demand (e.g., physical exertion) or reduce myocardial oxygen supply (e.g., anemia, development of a platelet-rich thrombus on a fissured plaque, or spasm of an epicardial coronary artery). Unstable angina most often results from disruption of an atherosclerotic plaque and a subsequent cascade of pathologic processes that decrease coronary blood flow ( Davies and Thomas, 1984; Falk, 1989; Fuster, Badimon, Badimon et al., 1992; Sherman,Litvack, Grundfest et al., 1986). Most patients who die during unstable angina do so because of sudden death or an intervening MI. Therefore, no pathologic endpoint can be used to define unstable angina.
Unstable angina is often associated with significant angiographic progression of coronary atherosclerotic disease ( Moise, Theroux, Taeymans et al., 1984). Patients with unstable angina do have more complex lesions and more coronary thrombus on coronary arteriograms than patients with stable angina ( Ahmed, Bittl, and Braunwald, 1993). However, no coronary angiographic findings are pathognomonic of unstable angina. The disorder is often angiographically indistinguishable from non-Q-wave MI ( Ambrose, Hjemdahl–Monsen, Borrico et al., 1988; Arbustini, Grasso, Diegoli et al., 1991).
Unstable angina presents as a constellation of clinical symptoms and can be legitimately defined in many different ways. The strictness of the definition of unstable angina used and method of assigning deaths to this cause or other ischemic heart disease (IHD) diagnoses can greatly influence reported mortality rates. Moreover, published series of patients with unstable angina commonly begin with the definitive diagnosis of the condition and not at the onset of symptoms. Therefore, the mortality observed in any series of carefully defined patients with unstable angina will tend to understate the risk in comparison with the mortality rate expected for these patients at the time of initial presentation for acute chest pain. The diagnosis of unstable angina at the time of hospital admission carries a risk of death that is intermediate between the IHD diagnoses of stable angina and acute MI. This fact is well illustrated by data from the Duke Cardiovascular Databank describing the rate of cardiac death in 21,761 patients treated for CAD without interventional procedures at Duke University Medical Center (DUMC) between 1985 and 1992 (see Figure 1). The three patient groups were defined by the diagnoses of stable angina, unstable angina, or acute MI at the time of admission. All three groups of patients had the highest risk of cardiac death at the time of presentation, and the risk declined so that by 2 months, mortality rates were indistinguishable in all three populations.
Figure 2 shows the rate of death over time in the same
subgroup of 9,146 medically treated patients with an admission diagnosis of unstable angina
as in Figure 1 but on an expanded scale. Mortality data from 15
published series (Table 5)
In patients without a known history of CAD, consideration of the diagnosis of CAD demands assessment of whether the patient's presentation, with its constellation of specific symptoms and signs, is most consistent with CAD or with an alternative disease process.
The five major factors of the initial history and physical examination that relate to the likelihood of
CAD are ranked in order of importance (
Angina is characterized as a deep, poorly localized chest or arm discomfort that is reproducibly associated with physical exertion or emotional stress and relieved promptly by rest or sublingual NTG. Patients with unstable angina may have all the qualities of typical angina except that episodes are more severe and prolonged and may occur at rest with an unknown relationship to exertion or stress. Rest discomfort with all the features of angina but without an exertional component should be considered definite angina for purposes of estimating CAD likelihood. Some patients may have no chest discomfort but present solely with jaw, neck, ear, or arm discomfort. If these symptoms have a clear relationship to exertion or stress or are relieved promptly by NTG, they should be considered equivalent to angina. Occasionally such symptoms at rest may be the mode of presentation of a patient with unstable angina, but without the exertional history, it may be difficult to recognize their cardiac origin. Other difficult presentations of the patient with unstable angina include those without any chest (or equivalent) discomfort. Isolated, unexplained, new-onset or worsened exertional dyspnea is the most common such symptom; others include nausea and vomiting and diaphoresis. Assessment of angina should conclude with a summary statement of the patient's symptoms to one of the following four categories: definite angina, probable angina, probably not angina, and not angina (CASS, 1981).
Features suggesting a diagnosis of not angina include:
Sharp, stabbing, or pleuritic qualities do not completely exclude an ischemic etiology. In the Multicenter Chest Pain Study, acute ischemia was diagnosed in 22 percent of ED patients presenting with sharp or stabbing pain and 13 percent of patients with some (but not full) pleuritic qualities to the presenting pain. Furthermore, 7 percent of patients whose pain was fully reproduced by palpation were ultimately recognized to have acute IHD ( Lee, Cook, Weisberg et al., 1985).
Evidence of a prior MI (history or pathologic Q-waves on the resting ECG) or a history of resuscitation from sudden cardiac death indicates a very high likelihood of significant CAD. For any clinical presentation, older patients have a higher CAD likelihood than younger patients, and at any age, men have a higher CAD likelihood than women. The likelihood of CAD and of more severe CAD increases with age, although women who have not undergone premature menopause generally lag 10 years behind men.
Findings on cardiac physical examination of a transient S3 or S4 mitral regurgitation murmur or precordial lift during an episode of discomfort signify a high likelihood of significant CAD. The presence of bruits or pulse deficits suggesting extracardiac vascular disease (carotid, aortic, peripheral) identifies patients with a higher likelihood of significant CAD.
Cardiovascular risk factors are modestly predictive of the likelihood of CAD in asymptomatic and nonacute symptomatic patients ( Chaitman, Bourassa, Davis et al., 1981; Pryor, Harrell, and Lee, 1983). A clinical diagnosis of diabetes mellitus is the most important risk factor, but cigarette smoking, hypercholesterolemia, and hypertension are also important predictors. A history of premature CAD (age <55) in a parent or sibling has inconsistently been identified as a major risk factor. This association may signify a genetic predisposition to CAD, or reflect the end result of shared environment and lifestyle characteristics, or both.
In the ED, risk factors have been found to be only weakly predictive of the likelihood of acute ischemia in men (diabetes and family history were strongest followed by smoking history); in women, no risk factor was a significant predictor of acute ischemia, possibly due to lower statistical power in this subset. A 50 percent or higher (nonsignificant) increase was seen with diabetes and hypertension ( Jayes, Beshansky, D'Agostino et al., 1992). Thus, in patients with suspected unstable angina, risk factors are far less important than the patient's symptoms and ECG findings, and presence or absence of risk factors should not be used to decide whether an individual patient should be admitted or treated for unstable angina.
Variant angina is an uncommon clinical syndrome of rest pain and reversible ST-segment elevation which may be difficult to diagnose on initial presentation. Approximately one-fourth of these patients in the United States have "insignificant" CAD and symptoms due to coronary vasospasm. Thus, this diagnosis may suggest specific management strategies, particularly the use of calcium channel blockers and nitrates and avoidance of beta blockers. However, three-fourths or more of these patients in the United States have a subtotal stenosis in a coronary artery, and a ruptured plaque is often demonstrable when their clinical course becomes unstable ( Mark, Califf, Morris et al., 1984; Waters, Miller, Szlachcic et al., 1983). Management of this latter group of patients is similar to that of other patients with unstable angina.
Cocaine use has recently been implicated as a cause of unstable angina. Three possible mechanisms by which cocaine induces myocardial ischemia are: (1) increased myocardial oxygen demand, 2) decreased myocardial oxygen supply secondary to vasospasm or coronary thrombosis, and (3) direct myocardial toxicity. Documented cocaine use should not be considered to rule out underlying significant CAD, since the drug may precipitate coronary vasospasm or acute MI in the patient with atherosclerotic CAD.
Careful examination of the ECG is crucial in the diagnosis of unstable angina ( Rouan, Lee, Cook et al., 1989). A recording made during an episode of the patient's presenting symptoms is particularly valuable, although an asymptomatic recording can be quite informative as well. Importantly, transient ST- or T-wave changes that develop during a symptomatic episode at rest and resolve when the patient becomes asymptomatic strongly suggest unstable angina and a very high likelihood of underlying severe CAD. Patients whose current ECG suggests acute IHD have added diagnostic accuracy if a prior ECG is available for comparison ( Lee, Cook, Weisberg et al., 1990).
ST-segment and T-wave changes are the primary elements upon which an ECG diagnosis of acute ischemia is based (see Table 6).
ST-segment elevation >e;1 mm in two or more contiguous leads strongly suggests the diagnosis of acute MI and possible candidacy for reperfusion therapy. ST-segment depression typically signifies ischemia or non-Q-wave infarction. Acute reperfusion therapy is usually not indicated for patients with this finding, except for those with acute posterior infarction manifesting the most marked ST-depressions in V1-V3. Inverted T-waves may also indicate ischemia or non-Q-wave infarction, especially with T-waves inverted >e;1 mm in leads with dominant R-waves. Marked symmetrical precordial T-wave inversion strongly suggests acute ischemia, particularly that due to a proximal LAD stenosis. Established Q-waves >e;0.04 seconds are less helpful in the diagnosis of unstable angina, although they do indicate a high likelihood of significant CAD with prior MI. Isolated Q-waves in lead III may be a normal finding.
Nonspecific ST- and T-wave changes, usually defined as ST-deviation or T-wave inversion <1 mm, are less helpful than the foregoing findings. In the Multicenter Chest Pain Study, when such findings were present, approximately one-fourth of the patients had acute IHD (predominantly unstable angina) ( Lee, Cook, Weisberg et al., 1985), but this was lower than the prevalence of such disease in the ED population overall. Thus, these nonspecific changes actually lowered the likelihood of MI and unstable angina, but not enough to reliably exclude either diagnosis. In the Multicenter Acute Ischemia Predictive Instrument Trial ( Pozen, D'Agostino, Selker et al., 1984), findings were similar. Along with elevation or depression of ST segments of 1 mm or more, even ST segment "straightening" (horizontal or downsloping ST segment with slight depression suggesting acute ischemia) was found to be significantly predictive of the presence of acute ischemia.
A completely normal ECG in the ED does not exclude the possibility of acute IHD, since 1 to 6 percent of such patients will eventually prove to have had an acute MI, and 4 percent or more will be found to have unstable angina ( McCarthy, Wong, and Selker, 1990; Rouan, Lee, Cook et al., 1989). In the Multicenter Acute Ischemia Predictive Instrument Trial ( Pozen, D'Agostino, Selker et al., 1984), 6 to 7 percent of ED patients with acute ischemia were found to be sent home. In a followup study ( McCarthy, Beshansky, D'Agostino, Selker, 1993) that included a search of the National Death Index, it was found that this 6 to 7 percent false-negative discharge rate included 2 percent of ED patients with acute MI who had been sent home. A study by the Multicenter Chest Pain Group showed a 4 percent false-negative discharge rate for acute MI ( Rouan, Lee, Cook et al., 1989).
Clinical and ECG characteristics that relate to the likelihood of significant CAD in groups of patients with symptoms suggestive of unstable angina at the time of initial presentation can be integrated into a summary statement about the likelihood of disease in an individual patient. Estimation of the likelihood of significant CAD is a multivariable problem that cannot be accurately quantitated from a simple table. Therefore, Table 70 is meant only to be illustrative of the general relationships between clinical and ECG findings and three arbitrary groupings of the likelihood of significant CAD. This table may be used to supplement the general clinical impression to categorize an individual patient as having low, intermediate, or high likelihood of CAD.
Since patients with unstable angina as a group are at increased risk of cardiac death and nonfatal MI, assessment of prognosis often sets the pace of initial evaluation and treatment of patients with suggestive symptoms. For all modes of presentation of IHD, a strong relationship exists between indicators of likelihood of CAD and prognosis. Those patients with a high likelihood of CAD are at a greater risk of an untoward cardiac event than patients with a lower likelihood of CAD. Therefore, assessment of the likelihood of CAD is the beginning point for determining prognosis in a patient presenting with symptoms suggestive of unstable angina. The two other important elements for prognostic assessment are the recent tempo of the patient's clinical course, which relates to the short-term risk of future cardiac events, principally acute MI, and the patient's likelihood of survival should an acute ischemic event occur.
The tempo of the patient's disease is judged from the cardiac history and ECG and from an examination of the patient during a symptomatic episode. The key elements of the history are: the current frequency of episodes, the change in frequency over the last 2 months (and particularly the last week), any increase in severity or duration of symptoms and in particular occurrence of episodes lasting >20 minutes, progression from effort or stress-related symptoms to symptoms occurring at rest, new onset of nocturnal symptoms, or a significant decrease in the amount of stress or effort necessary to provoke symptoms ( Califf, Mark, Harrell et al., 1988; De Servi, Ghios, Ragni et al., 1985). New onset angina is an adverse prognostic event ( Roberts, Califf, Harrell et al., 1983), but its risk is defined by other variables from the history, such as the tempo, frequency, and severity of symptoms ( Califf, Mark, Harrell et al., 1988; White, Lee, Cook et al., 1990).
Whenever possible, the patient should be examined and have an ECG recorded during a symptomatic episode. The key prognostic elements from the physical examination include any evidence of acute CHF (i.e., new or worsening rales, an S3), a new or worsening MR murmur, and systemic hypotension. The major prognostic elements from the ECG are dynamic shifts in the ST-segment (>e;1 mm ST-depression or elevation) or T-wave inversions that resolve, at least partially, when symptoms are relieved ( Bosch, Theroux, Pelletier et al., 1991; Karlson, Herlitz, Pettersson et al., 1993). With these characteristics it is possible to separate patients into three risk groups as defined in Table 8.
The four most important factors related to the likelihood of survival should an acute ischemic event occur in all groups of patients with CAD, including those with unstable angina, are LV function, extent of obstructive coronary artery disease, age, and comorbid conditions. Assessment of LV function is the single strongest predictor of subsequent cardiac death in patients with CAD, and this relates to the lowered reserve of cardiac function in these patients which makes them less tolerant of further ischemia or infarction. The extent of coronary disease defines both the likelihood of an acute coronary event and the likely availability of collateral supply should such an event occur. Thus, coronary events are both more frequent and more likely to be fatal in patients with significant CAD of all three coronary arteries than in patients with significant one-vessel disease. Advanced age is an independent marker of risk that may relate to the lower reserve of cardiac function during stress in the elderly as well as to diminished function of other important organ systems. Important comorbid conditions that greatly influence survival in patients with unstable angina include renal failure, chronic obstructive pulmonary disease (COPD), cerebrovascular disease, and malignancy or other chronic systemic disease.
Based on the disappointing results of using customarily available clinical data, investigators have proceeded to evaluate mathematically based decision aids and newer cardiac imaging techniques to optimize ED triage of such patients.
The goal of mathematically based diagnostic aids for acute cardiac ischemia and acute MI is to improve physicians' use of clinical information by quantifying risk in the face of uncertainty ( McCarthy, Wong, Selker, 1990; McNutt and Selker, 1988; Wasson, Sox, Neff, Goldman, 1985). The first such diagnostic aid was Sawe's (1972) "clinical diagnostic index," which predicted acute MI based on nine clinical variables derived by discriminant analysis. Tested prospectively it was 100 percent sensitive for acute MI, but its very poor specificity (16%) limited its applicability to actual practice. Based on 655 ED patients with chest pain, Tierney and colleagues created a multivariable model predicting acute MI based on the clinical presentation and ECG that was more specific (86% vs. 78%), but less sensitive (81% vs. 87%) than physicians. Hypothetical integration with physicians' triage decisions did not significantly improve accuracy, and its prospective trial has not been reported ( Tierney, Roth, Psaty et al., 1985).
In the Multicenter Chest Pain Study of 12,140 patients, a model was developed using data from the history, physical examination, and ECG to predict the probability of acute MI and improve triage to the ICU ( Goldman, Cook, Brand et al., 1988; Goldman, Weinberg, Weisberg et al., 1982). Features associated with a higher probability of acute MI included: duration of symptoms <48 hours prior to ED evaluation, history of angina or prior MI, pain duration >e;1 hour, pain worse than prior angina or equivalent to prior MI, age >e;40, ST- or T-wave changes of ischemia or strain not known to be old, and radiation of pain to the neck, left shoulder, or left arm. Features lowering the probability of acute MI include: radiation of pain to the back, abdomen, or legs; "stabbing" quality of pain; and reproduction of pain by palpation. Combinations of these characteristics yielded 14 subgroups with a probability of acute MI ranging from 1 to 77 percent.
The Multicenter Chest Pain Study model has been shown to stratify 36 percent of patients presenting with chest pain to the ED into a low-risk subgroup ( Lee, Juarez, Cook et al., 1991). After a 12-hour observation period, 81 percent of these low-risk subjects remained free of recurrent pain and had at least one normal and no abnormal cardiac enzyme determinations. These uncomplicated low-risk patients were judged to be suitable, after the 12-hour observation period, for further evaluation and therapy in an unmonitored hospital setting. Evidence of acute MI was subsequently obtained in 0.5 percent of this cohort, and 0.6 percent died of cardiac causes during the hospitalization, all after day 3. Further results from this research project showed that initially uncomplicated "rule-out MI" patients can probably be cared for safely in an intermediate-care unit, thus reserving ICU admission for patients with definite or high probability for acute MI and patients who have developed early complications ( Fiebach, Cook, Lee et al., 1990).
Pozen, Selker, and colleagues developed and validated a quantitative predictive instrument to improve the diagnosis of acute cardiac ischemia (unstable angina or acute MI) and subsequent triage decisions in the ED ( Pozen, D'Agostino, Selker et al., 1984). They identified seven major predictive factors: (1) age; (2) sex; (3) the presence or absence of chest pain or pressure, or left arm pain; (4) whether or not chest pain or pressure was the patient's most important presenting symptom; (5) the presence or absence of ECG Q-waves; (6) the presence and degree of ECG ST-segment elevation and depression; and (7) the presence and degree of ECG T-wave peaking or inversion ( Selker, Griffith, and D'Agostino, 1991). This model was shown to have a sensitivity of 95 percent and a specificity of 78 percent for diagnosis of acute cardiac ischemia.
In addition to the use of acute cardiac ischemia as the clinical endpoint instead of acute MI alone, this work is also different from prior work in that instead of only including patients with chest pain, it included all ED patients presenting with symptoms suggestive of acute cardiac ischemia including chest pain or left arm pain, abdominal pain or nausea, shortness of breath, and dizziness or lightheadedness. (These inclusion criteria were based in the Imminent MI Rotterdam [IMIR] Study criteria, which have been shown to capture more than 90 percent of all patients in a community with acute cardiac ischemia) ( Van der Does, Lubsen, Pool et al., 1976). In controlled prospective trials of the instrument's use, first at Boston City Hospital and then in the Multicenter Predictive Instrument Trial ( Pozen D'Agostino, Selker et al., 1984), it reduced false-positive CCU admissions by 30 percent without an increase in false-negative discharges to home.
The Pozen/Selker model is designed to predict acute cardiac ischemia (acute MI plus unstable angina), while the Goldman model predicts the probability of acute MI. Both were originally developed as tools to improve the cost-effective use of cardiac ICU admissions from the ED. Differences in the models can be attributed to the use of different endpoints, different statistical methodologies, and variability in different clinical data sets. Both models emphasize the importance of prolonged or severe chest pain (or equivalent symptoms), evidence of prior MI, and ST- and T-wave changes on the ECG. Although these predictive statistical models have been prospectively tested in diverse practice settings and hospitals with encouraging results, the panel feels that their routine use in clinical medicine is still some years off. However, practitioners and hospitals should be encouraged to use such an approach as a supplement to the traditional less structured clinical evaluation, if they wish to do so.
Return to AHCPR Guideline List
Return to HSTAT Home Page