Introduction
Of the more than 8 million visits to emergency departments (EDs) in the United States for chest pain or other potential ischemic symptoms, only a small minority will yield a diagnosis of an acute coronary syndrome (ACS). Less than 5% of patients so affected will prove to have ST-elevation myocardial infarction (MI), and only approximately one fourth will receive the final diagnosis of non–ST-elevation ACS (NSTE-ACS). Identifying patients who can be discharged early, with minimal additional testing, while also ensuring that patients at high risk for cardiac events are appropriately triaged to receive more advanced evaluation and management, remains one of the most common and challenging clinical scenarios that cardiologists, emergency medicine physicians, and primary care providers face on a daily basis.
Chest and abdominal pain are the two most common symptoms prompting emergency room visits in the United States (see Chapter 6 ), with a volume that has remained relatively stable over the past decade. However, practice in the ED and inpatient evaluation of these patients has evolved considerably. Compared with a decade ago, more patients hospitalized with chest pain now undergo advanced imaging studies, such as echocardiography, computed tomography, or cardiac magnetic resonance imaging, as evidenced by a shift in the frequency of such imaging from only 3.4% in 1999 to 15.9% of patients in 2008. Patients presenting with chest pain were much more likely to be hospitalized or transferred to another institution or to die than patients with other chief complaints such as abdominal pain. Despite the fact that the rate of admission, transfer, or death has declined over the past decade, from 42.5% in 1990 to 35.2% in 2008, the burden of evaluating patients with chest pain in the ED and efforts to address the typically high rate of complications continue to create tremendous demands upon the health care system.
The clinical presentation of patients with suspected MI and the key considerations in the general approach to their evaluation are discussed in Chapter 6 . The principles behind the optimal use of cardiac troponin (cTn) are addressed in Chapter 7 . Other biomarkers are discussed in Chapter 8 , and the use of imaging is described in Chapter 9 . This chapter reviews specific strategies and algorithms integrating each of the elements of clinical, laboratory, and imaging data to identify those patients for whom the probability that an ACS is the cause of their symptoms is deemed to be very low. An efficient approach to evaluating such “low-probability” patients should minimize the time until diagnosis, reduce the need for additional testing, and limit the duration of hospitalization, while avoiding erroneous discharge of the patient with an ACS. Older studies suggest that as many as 2% of patients with acute MI may be erroneously discharged from the ED. In the current era, an acceptable “miss” rate for MI generally is viewed as less than 1%.
Defining the “Low-Probability Patient”
Before defining the specific population of patients who can be safely discharged from the ED, it is worthwhile making the distinction between the labels of “low probability” and “low risk,” because these two terms often are interchanged freely in discussing chest pain and ACS. In characterizing such patients, it is preferred to specify “low probability” for the presence of ACS, rather than “low risk,” which most commonly is employed in estimating the likelihood of subsequent cardiovascular events (see also Chapter 6 ). Risk stratification is a critical step in the evaluation of patients with documented ACS and incorporates many of the same clinical characteristics used for diagnosis. However, the risk estimates and clinical implications in this patient population are much different from those in patients undergoing evaluation for suspected ischemic symptoms. For example, a patient may have a high probability for ACS and therefore merit hospitalization but be at moderate or low risk for subsequent cardiovascular events (see Figure 6-2 ). Risk stratification in patients with documented ACS is reviewed in detail in Chapter 11 .
Considerations for Defining Probability and Ruling Out ACUTE Coronary Syndrome
The evaluation of the patient with suspected ischemic symptoms integrates, at a minimum, the patient’s comorbid conditions, history and presentation, and electrocardiographic findings. Most patients, except those with the very lowest probability for ischemia, also will have at least one biomarker of necrosis (i.e., cTn) measured in their evaluation. Decisions regarding subsequent noninvasive testing in general, and which modality in particular is most appropriate, remain controversial. The most challenging aspect of identifying patients with a low probability for ACS is the absence of a single “gold standard” test for this clinical entity. Troponin assays identify myocardial injury but not the underlying cause (see Chapter 7 ). The diagnosis of MI is made on the basis of the clinical scenario, testing results, and ultimately, medical judgment.
Depending on clinical characteristics and electrocardiography findings, most patients can be classified into groups of very low, low, intermediate, or high probability (see also Chapter 6 ). This first, immediate estimation of probability is important, because the value of all subsequent testing is dependent on the pretest probability of disease. Typically, additional testing is most useful in those patients with intermediate pretest probability ( Figure 12-1 ). In patients with a high pretest probability, even a “negative” subsequent test would not be reassuring, because a higher-than-acceptable false-negative rate is likely. Conversely, in the very-low-probability patients, a positive test result is much more likely to be a false positive than to represent the identification of “true” disease. In practical terms, this circumstance might correspond to measuring cTn in a 25-year-old woman with a history of 3 days of chest discomfort relieved with an antacid. Alternatively, an initially normal cTn level would not be entirely reassuring in a 75-year-old patient with diabetes and vascular disease presenting with typical chest discomfort.
In the initial management of patients with suspected ischemia, it often is more important to rule out unstable ischemic syndromes than to definitively “rule in” the diagnosis of ACS. Assessing pretest probability of disease is as important in excluding such disease as in its ultimate diagnosis. In deciding whether a test is appropriate to rule out a disease, one must consider the underlying prevalence of the disease under consideration and the specificity of the test to exclude disease. For example, as discussed further on, a computed tomography (CT) angiogram would be useless to exclude ACS in a patient with documented coronary disease. Similarly, measuring high-sensitivity cTn in a patient with end-stage renal disease and severe left ventricular hypertrophy will yield results that can be challenging to interpret.
Stable versus Unstable Coronary Artery Disease
Another important distinction relevant to the evaluation of patients with suspected ACS is the difference between stable coronary artery disease and an unstable coronary lesion precipitating ACS. Although the presence of known coronary artery disease increases the probability of developing ACS, patients with stable coronary artery disease commonly present with nonischemic or even noncardiac causes of chest symptoms. Patients can have underlying asymptomatic coronary artery disease and present with symptomatic gastrointestinal reflux disease (GERD) symptoms. Thus, diagnosing the presence of coronary artery disease does not equate to a diagnosis of ACS—a mistake that is commonly made in both clinical practice and clinical investigation. The clinical scenario of presentation is essential to making the diagnosis of ACS.
Identifying the Low-Probability Patient
Although many different algorithms are available for defining a low probability for the presence of ACS, all classification systems incorporate elements of history, physical examination, electrocardiogram, and cardiac biomarkers (see also Chapter 6 ). Defining low probability is, in this sense, excluding those clinical features that identify a high-probability patient. Specifically, patients are considered to have a low probability for ACS if they do not have typical chest pain symptoms, including angina similar to previous angina, or chest and left arm pain or discomfort (see Figure 6-4 ). Low-probability patients tend to be younger and have fewer identifiable cardiovascular disease risk factors. Under this paradigm, it is difficult for example to categorize a patient older than 70 years of age with a history of diabetes as inherently having a low probability for ACS, regardless of the clinical scenario. On physical examination, the low-probability patient should be free of any evidence of volume overload or extracardiac vascular disease. Patients are considered to be in the low-probability group if their ECG is normal in terms of ischemic changes or demonstrates, at most, T-wave flattening or inversions of less than 0.1 mV. Perhaps most important, initial levels of biomarkers in a low-probability patient should be normal.
Epidemiology of Low-Probability Patients
A majority of patients who present with chest pain have either a low- or an intermediate-probability for presence of unstable coronary artery disease. It is difficult to accurately estimate the exact proportion of patients classified as low probability, because of variable definitions of low probability across clinical studies, which also span heterogeneous cohorts. With those caveats, between 25% and 40% of all patients presenting with chest pain appear to have a low probability for ACS.
Accelerated Diagnostic Protocols
Most strategies to identify and appropriately triage patients with a low probability for ACS are based on some type of accelerated diagnostic protocol. These protocols may be implemented in the ED, in dedicated chest pain units, and on inpatient wards. Current reimbursement in United States rewards shorter hospital stays, with the goal of more patients discharged directly from the ED and limiting in-hospital stays to less than 24 hours. Thus, the metrics of a successful algorithm, in addition to ensuring a very low rate of missed ACS cases, include time until discharge and need for subsequent testing.
Risk Scores
A variety of clinical risk scores have been proposed for the evaluation of patients with suspected ischemic symptoms. Some are specifically derived from broad populations of patients with chest pain, whereas others either implement or modify existing clinical risk scores that were originally derived in patients with established ACS. A key principle of such scores is that the chance of a final diagnosis of an MI is extremely small in a low-probability population. Comparison of the performance of one score versus another is challenging because of differences in the inclusion criteria between studies and which biomarker of necrosis was measured. For example, the same risk score tool will perform substantially differently if cTn is used instead of creatine kinase–myocardial biomarker (CK-MB) (see Chapter 1 and Chapter 7 ).
Goldman and colleagues proposed one of the earliest comprehensive algorithms three decades ago. Many of the subsequently developed algorithms have utilized risk scores such as Thrombolysis in Myocardial Infarction (TIMI) Risk Score, the Global Registry of Acute Coronary Events (GRACE) Risk Score, and the PURSUIT Risk Score, which originally were derived and validated in clinical trials or registries of patients with confirmed ACS (see Chapter 11 ). By design, these risk scores predict major cardiovascular outcomes in patients with ACS and are neither sensitive nor specific enough alone to use for the diagnosis of ACS. Even modifying the TIMI risk score by giving more weight to abnormal cTn levels or ECG changes may still not be sensitive enough at scores of 0 to allow early discharge.
Other studies derived new scores with the specific goal of identifying the population of low-probability patients who could be discharged safely without subsequent testing; such scores have seen limitations to wide implementation, however, because the false-negative or “miss” rate has not consistently been demonstrated to be below the commonly accepted threshold of 1%. One of the most extensively validated risk scores for chest pain is the HEART Score, which evaluates patients on five domains—history, ECG, age, risk factors, and troponin— to generate a score from 0 to 10 (see Figure 6-6 ). In validation studies, the risk of a subsequent cardiac event was less than 1% in the roughly one third of patients with a low HEART Score (0 to 3).
All of the diagnostic risk scores use a biomarker of cardiac necrosis. The inclusion of cTn improves the diagnostic performance of all risk scores, but as discussed further on in this chapter, a single measurement of cTn is not always sufficiently sensitive to exclude a diagnosis of ACS.
A dynamic risk score is one that will incorporate additional data collected after the initial presentation. Most algorithms for identifying low- or intermediate-probability patients need to be dynamic, because no single test or algorithm is sufficiently sensitive or specific when based on the initial clinical and biochemical assessment alone.
“Rule Out Myocardial Infarction”
Historically, patients presenting with suspected ischemic syndromes were given the classic diagnosis of “rule out MI” (“R/O MI”), which was actually a diagnostic strategy rather than a diagnosis. The strategy included serial ECGs and biomarker measurements 8 hours apart (usually a total of three), followed by an exercise stress test. With the transition from CK-MB to cTn, many institutions now omit the third biomarker assessment and shorten the interval between measurements to 3 to 6 hours (see Chapter 7 ). Even with a reduced number of biomarker measurements and shorter time intervals, the evaluation may require 18 to 36 hours, depending on the availability of a stress testing protocol. Accordingly, efforts have now focused on identifying better algorithms that will facilitate the early triage and discharge of patients with a low probability for ACS.
Shortening the interval between repeat biomarker assessments is one of the easiest ways to accelerate a dynamic protocol, but the question of how short the interval can be remains controversial (see also Chapter 7 ). With earlier generations of biochemical assays, poor analytic performance and relatively high diagnostic cutpoints limited the detection of an early rise in biomarkers or low levels of cardiac injury. This shortcoming was partially overcome in the original R/O MI protocols by requiring 6 to 8 hours between CK-MB assays. Even early cTn assays, although superior to CK-MB assays, lacked the sensitivity to detect necrosis early after event onset, and use of serial measurements at similar intervals was required to maintain adequate sensitivity. With the introduction of newer-generation assays for cTn (not including the high-sensitivity assays discussed further on), troponin testing can be done at reduced intervals of 0 and 3 hours, with some algorithms using 2-hour intervals.
Integrated Algorithms
Most contemporary chest pain algorithms incorporate an assessment of probability based on clinical presentation combined with two sequential cTn measurements. One relatively simple algorithm, the North American Chest Pain Rule (see Figure 6-6 ), accurately identified a low-probability population by combining just one troponin assay with three clinical features in patients 40 years of age or younger, with use of two sequential troponin measurements for patients 40 to 50 years of age. Unfortunately, this algorithm did not perform well in patients older than 60 years, highlighting the problems of excluding ACS in patients with higher pretest probability of disease.
The HEART Score, which used data from clinical presentation, was updated into the HEART Pathway by integrating serial troponin measurements at 0 and 3 hours ( Figure 12-2 ). Patients deemed to have a low probability for ACS based on the Heart Pathway (HEART Score 0 to 3 and negative serial cTn assays), who accounted for roughly 50% of the population studied, were discharged home from the ED without further testing. In a randomized trial compared with standard care, the HEART Pathway decreased the frequency of cardiac testing, lowered length of hospital stay by 12 hours, and increased the proportion of patients discharged early. No patients within the low-risk group experienced a subsequent cardiac event.
At our institution, we have integrated the HEART Score into a dynamic algorithm for the diagnosis of suspected ACS. In our algorithm, patients with a low HEART Score at admission can be discharged if symptom onset was more than 6 hours before presentation. Because we do not utilize a high-sensitive troponin assay, if a patient presents within 6 hours of symptom onset, we order a second cTn at 3 hours. If that value is also below the 99th percentile, then the patient can be discharged home without further testing ( Figure 12-3 ).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
