Current issues of ACP Journal Club are published in Annals of Internal Medicine


4 characteristics helped differentiate non-Q-wave MI from unstable angina

ACP J Club. 1995 Nov-Dec;123:73. doi:10.7326/ACPJC-1995-123-3-073

Source Citation

Cannon CP, Thompson B, McCabe CH, et al, for the TIMI III Investigators. Predictors of non-Q-wave acute myocardial infarction in patients with acute ischemic syndromes: an analysis from the Thrombolysis in Myocardial Ischemia (TIMI) III Trials. Am J Cardiol. 1995 May 15;75:977-81.



To develop and validate a set of criteria from baseline characteristics that could differentiate between non-Q-wave acute myocardial infarction (MI) and unstable angina in patients who present with ischemic chest pain.


Evaluation of baseline characteristics to identify predictors of MI. These predictors (risk factors) were compared with creatinine kinase (CK) and CK-MB levels at baseline and at 4 and 12 hours. Predictors were identified with data from half the sample (n = 750) and were validated with data from the other half (n = 750; Thrombolysis in Myocardial Ischemia III [TIMI III] Trial).


25 clinical centers in North America.


1470 adults presenting with ischemic cardiac pain at rest that lasted ≥ 5 minutes to 6 hours and who had evidence of ischemic heart disease. Exclusion criteria were age ≥ 76 years, MI within the previous 21 days, angioplasty within the past 6 months, previous bypass surgery, pulmonary edema, cardiogenic shock, contraindications to thrombolytics, or left bundle-branch block.

Description of Test and Diagnostic Standard

50 clinical and electrocardiographic (ECG) variables (patient characteristics, previous cardiac history and medications, features of chest pain, and changes in ECG) were analyzed (step-wise logistic regression) to identify independent predictors of MI. MI was confirmed by CK and CK-MB levels.

Main Outcome Measures

Sensitivity and specificity.

Main Results

4 risk factors were found (absence of previous coronary angioplasty, duration of pain ≥ 60 min, ST-segment deviation on initial ECG [P for each factor < 0.001], and recent-onset angina [P = 0.002]). 7% of patients with no risk factors and 71% of patients with 4 risk factors had MI. The sensitivity and specificity for 1 risk factor were 99% and 6%; for 2 risk factors, 83% and 37%; for 3 risk factors, 55% and 78%; and for 4 risk factors, 17% and 97%, respectively.


In patients presenting with ischemic chest pain, 4 baseline characteristics (no previous cardiac angioplasty, duration of pain ≥ 60 min, ST-segment deviation, and recent-onset angina) were somewhat useful for distinguishing between non-Q-wave myocardial infarction and unstable angina.

Source of funding: National Heart, Lung, and Blood Institute.

For article reprint: Dr. E. Braunwald, Department of Medicine, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115, USA. FAX 617-732-6439.


The results of this study by Cannon and colleagues will probably not alter the management of patients presenting with ischemic chest pain because of the poor predictive ability of the decision rules and the uncertain clinical importance of making the diagnosis at admission.

Notwithstanding the fact that the 4 baseline characteristics are the best "predictors" for the development of a non-Q-wave MI, the ability to predict an MI using them is poor whether they are considered alone or in combination. For example, in the presence of 1 and 2 characteristics, the positive likelihood ratios (the odds of a patient with MI, as compared with unstable angina, having these characteristics) are 1.05 and 1.32, respectively, and are therefore not helpful. When either 3 or 4 characteristics are present, the resulting likelihood ratio is modestly improved to 2.5, but only 55% of patients with MI are in this category. The best predictability is obtained when all 4 characteristics are present. The resulting high specificity (97%) should help establish an MI; however, relying on this prediction rule would identify only 15% of all patients with MI.

Currently, because the early management of patients with unstable angina and non-Q-wave MI does not differ, it is not apparent that classification of patients within the first 24 hours is important. Further, although the prognosis of patients with unstable angina and non-Q-wave MI differs at 6 and 12 weeks, it is not clear whether making this distinction at hospital admission will influence future outcomes. Finally, although the authors suggest that some efficiencies in patient triage might be realized using these prediction rules, the low overall 42-day mortality rate shown in TIMI IIIB (2.4%) suggests that most patients with unstable angina or non-Q-wave MI could be treated in a step-down unit.

David Massel, MD
Victoria Hospital London, Ontario, Canada

Author's Response

The commentator should recall that clinical diagnosis is not perfect: Goldman and colleagues (1) found that the positive predictive value for predicting MI from ECG information was 29%. The positive predictive value of our rule ranges from 33% to 71% for 1 to 4 risk factors. We offer it to clinicians as a no-cost tool to help identify patients at higher risk.