Failure to Admit Chest Pain Patients with Acute Ischemia
abstract & commentary
Synopsis: The "relatively low rate of misdiagnoses" will be difficult to reduce further and improvement in ECG interpretation ability and recognition that accelerating angina is a reason for admission are two factors amenable to improvement.
Source: Pope HJ, et al. N Engl J Med 2000;342:1163-1170.
A large body of clinical literature has evolved regarding the optimal approach to individuals presenting with chest pain to the emergency room. Chest pain units become popular, and many hospitals have developed algorithms and guidelines for the triage and evaluation of such individuals. This study sought to analyze those factors relating to inadvertent discharge from the emergency department (ED), as well as to assess the outcomes of such individuals. This was a multicenter study called the Acute Cardiac Ischemia Time-Insensitive Predictive Instrument (ACI-TIPI), which involved 10,689 patients seen in a total of 10 institutions. The latter represented a wide spectrum of hospitals throughout the country. The study period was seven months (May-December 1993). Data were collected in the ED, during hospitalization, and at 30 days. Patients who were discharged from the ED were asked to return 24-72 hours later for repeat evaluation, including an ECG and myocardial muscle creatine kinase isoenzyme (CK-MB). Follow-up rate was 99%. Seventeen percent of the entire cohort met the criteria for acute cardiac ischemia; approximately half had acute myocardial infarction (MI) and half with unstable angina. Another 6% were diagnosed with stable angina, 21% were felt to have nonischemic cardiac problems; the majority (55%) had noncardiac problems. The ultimate proportion of subjects with unstable angina who were missed and/or were not hospitalized was 2.3% (22/956); for acute MI, it was 2.1%. There was considerable range among the 10 sites, with one institution missing as many as 4.3% unstable angina patients and one institution missing 11.1% of acute MI patients. Eighty percent of the patients who had a misdiagnosis were seen by an attending physician; half of these were felt to have noncardiac chest pain. A few had an erroneous ECG interpretation. Eighty-six percent of the patients with unstable angina were seen by an attending physician; half were felt to have stable angina. Misinterpretation of the ECG by the ED staff was felt to occur in three of 19 patients. Clinical features were analyzed that were related to failure to hospitalize. Multivariate analysis concluded that women younger than 55 years of age, nonwhite patients, shortness of breath, and a normal ECG were major confounding factors. Of the 19 patients with a misdiagnosis of acute MI, 70% were rehospitalized within 30 days, compared to 17% of those individuals initially hospitalized with this diagnosis; 50% of the unstable angina patients sent out of the ED were hospitalized within the first month, compared to 20% of those with this diagnosis who were initially admitted. The mortality rate for hospitalized and nonhospitalized patients with AMI at 30 days was approximately 10%; however, when risk adjustment was performed, the hospitalized patients had a 5.5% mortality rate vs. 9.8% in the inappropriately discharged patients. The risk ratio of mortality for failure to hospitalize with MI was 1.9, and for unstable angina it was 1.7 (30-day mortality, 5.0 vs 2.1% for those who were hospitalized). Pope and associates conclude that the overall rate of misdiagnosis of MI was acceptably low and comparable to that in prior studies.
A small number of errors was made in not detecting minor degrees of ST elevation (11%) "which represents an important and potential contribution to failure to admit these patients." Younger women were less likely to be admitted with acute ischemia. This is possibly related to sex bias and atypical presentation, such as shortness of breath, heart failure, and abdominal pain. The observation that nonwhites with unstable angina were sent home more than twice as often as whites, and missed acute MI was four times more than whites, has not been described previously. Overall, 5.8% of blacks with acute MI were not hospitalized compared to 1.2% of whites. In the unstable angina population, a number of individuals thought to have stable angina actually had new or increasing angina symptoms within three days. A normal or almost normal ECG was quite common in individuals inappropriately discharged; 53% of the MI and 52% of the unstable angina patients were in this category: three-fourths of the missed MIs were non-Q. Pope et al conclude that the "relatively low rate of misdiagnoses" will be difficult to reduce further and suggest that improvement in ECG interpretation ability, and recognition that accelerating angina is a reason for admission, are two factors amenable to improvement. Nonwhites were 4-5 times as likely not to be hospitalized with an acute MI, particularly if the initial ECG was normal or nondiagnostic. Those institutions with chest pain units did not have a lower rate of misdiagnosis. Pope et al suggest that the small number of hospitals is a study limitation, particularly in that there were no rural hospitals and no hospitals without ED physicians. They suggest that improved training in ECG interpretation and recognition of those factors that may obscure the diagnosis, particularly in women and nonwhites, is needed. They believe that diagnostic techniques and evaluation strategies should continue to be assessed, as well as efforts to better understand those.
Comment by Jonathan Abrams, MD
In the same issue of the New England Journal of Medicine, there is an editorial by Mehta and Eagle dealing with this article, as well as a review article by Lee and Goldman, entitled "Primary Care: Evaluation of the Patient with Acute Chest Pain." These three articles represent a valuable primer for all individuals who have an interest in the problem of chest pain/myocardial ischemia decisionmaking in the ED. When one examines the national experience, it is common that only a small minority of individuals who come to the ED with chest pain actually have a valid diagnosis of acute MI or unstable angina. The 17% rate in this study is higher than in many, but not all, reports. Results from the ACI-TIPI trial are reassuring, and suggest that there is a relatively low level of misdiagnosis that may be difficult to reduce significantly. Pope et al do not report the algorithm used at the 10 study sites to exclude myocardial ischemia, and therefore it is difficult to precisely compare this trial to other reports in the literature. The recommended approach is to obtain biochemical markers of ischemia (i.e., CK, CK-MB, and troponin) at presentation to the ED, as well as 6-8 hours later. It is well known that these markers may be normal on admission, and that it may take many hours for them to become abnormal. It is probably more important to obtain a second set of enzymes and troponin than a second ECG. After myocardial necrosis is excluded, many institutions proceed with some type of a stress test before the patient is released from the ED. How often this occurred in the present study is not reported. Pope et al state that "serial measurements of creatine kinase MB were carried out," but further information is not provided. No details regarding stress testing are available.
The fact that nonwhites and women (particularly younger than the age of 55) frequently have atypical chest pain presentation is of great importance to ED personnel and other physicians who evaluate these patients on a daily basis. A heightened awareness for possible ischemia is critical. Optimal ECG reading skills are important. It is unrealistic to expect that ED physicians will be as competent as a board-certified cardiologist in recognizing subtle or minimal ECG changes that suggest myocardial ischemia or acute injury. The issue regarding the value of chest pain units is important and unresolved. Pope et al raise the question as to whether these are truly effective and efficient, whereas Mehta and Eagle suggest that a chest pain center may decrease the frequency of a missed diagnosis. It is likely we will not have an answer to this question for some time. The national data thus far do not show a major change in the misdiagnosis rate in established chest pain units.
The other side of the coin is also critical; far more patients than need be are admitted to the hospital and ultimately receive a diagnosis of noncardiac chest discomfort. Thus, the cost of care and diagnostic testing of individuals presenting with chest discomfort to the ED is enormous. Therefore, the focus of our approach should be to try not to miss the occasional patient who does not present with typical features or a diagnostic ECG, but also not to overadmit or overtest patients who have features of noncardiac disease. Mehta and Eagle emphasize the use of careful history and physical examination and assessment of traditional risk factors as well as enhanced ECG skills. They also call for routine exercise testing before discharge to help identify those patients at risk for an acute coronary syndrome. They estimate that hospitalizations for noncoronary pain result in health care costs of more than $5 billion annually; prior literature suggests that 2-8% of individuals with acute MI are actually released from the ED. Furthermore, the fear of malpractice suits drives excessive admissions and overtesting. As with so much in medicine, clinical judgment and skills are mandatory. Employment of a strategy using the history, the ECG, and serial biochemical markers remains the best approach for the new millennium.
Experience with chest pain triage suggests that:
a. missed acute coronary syndrome patients are less than 5%.
b. atypical presentations are a clinical challenge.
c. inaccurate ECG reading remains an issue.
d. All of the above