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


Emergency physicians often misread electrocardiograms for patients with cardiac ischemia

ACP J Club. 1993 Jan-Feb;118:16. doi:10.7326/ACPJC-1993-118-1-016

Source Citation

Jayes RL Jr, Larsen GC, Beshansky JR, D'Agostino RB, Selker HP. Physician electrocardiogram reading in the emergency departmentaccuracy and effect on triage decisions: findings from a multicenter study. J Gen Intern Med. 1992 Jul-Aug;7:387-92.



To determine the accuracy of physician electrocardiogram (ECG) reading in the emergency department when diagnosing patients with suspected acute ischemic heart disease.


Comparison of emergency physicians' readings of ECGs with expert ECG reader codings.


6 New England hospitals.


1912 patients (age ≥ 30 y) who presented at any of the 6 emergency departments with chief complaints of chest, jaw, or left arm pain; shortness of breath, abdominal pain, or nausea; and fainting or lightheadedness. Patients with right or left bundle-branch block, left ventricular hypertrophy, or paced rhythms were excluded. 439 emergency department physicians participated.

Description of test and diagnostic standard

The emergency physicians coded their readings of the ST segments and T waves based on the Minnesota code. The ECGs were later coded by expert physician ECG readers who were blinded to the emergency department physicians' readings. Expert physicians' interrater agreement was > 90%.

Main outcome measures

ECG codings were compared for recognition of ST-segment depression or elevation of ≥ 1 mm, ST-segment straightening or depression of < 1 mm, T-wave depression of ≥ 1 mm, T-wave flattening, and T-wave elevation.

Main results

Compared with expert ECG readers, emergency physicians misinterpreted 20% of the ST segments and 16% of the T waves as normal (false-negative readings). Emergency physicians also reported abnormalities in 7% of ST segments and 9% of T waves when these changes were absent (false-positive readings). The emergency department physicians had sensitivities of 59% and 64% for ST-segment and T-wave changes, respectively. Specificities were 86% and 83% for ST-segment and T-wave changes, respectively. Suboptimal triage occurred in 22% of patients when ST segments were misread as normal compared with 11% when ST segments were read correctly (P < 0.001).


Physicians in the emergency department often misread the ST segments and T waves of patients with possible acute cardiac ischemia. Misreading ST segments resulted in a higher likelihood of inappropriately triaging these patients.

Sources of funding: Agency for Health Care Policy and Research and Technology Assessment and the National Library of Medicine Medical Information Program.

For article reprint: Dr. H.P. Selker, Division of Clinical Care Research, New England Medical Center, Box 1031, 750 Washington Street, Boston, MA 02111, USA. FAX 617-636-8023.


The interpretation of the first ECG in the emergency room determines to a large part whether a patient is considered a candidate for thrombolytic therapy and whether a patient is admitted to a critical care unit, a step-down monitored bed, an unmonitored bed, or sent home. This article indicates that errors in reading ECGs are more likely to occur by individuals with less formal training in ECG interpretation and that errors in interpretation lead to mistakes in management.

When faced with a patient with chest pain in the emergency department, clinical decision making involves assessment of the ECG, of the symptoms, and of the overall risk for coronary artery disease. Two articles (1, 2) have tried to distill this decision-making process into useful algorithms, both of which attribute key importance to interpretation of the ECG.

Candidates for thrombolytic therapy should be rapidly and unerringly identified; unfortunately, in this study 16% of ECGs with ST elevation were incorrectly identified as normal. To help solve this problem, rapid access to interpretations by expert ECG readers would ideally be available around the clock to the emergency department physician. This should not prove overly difficult in the era of facsimile machines and couriers. Computer ECG interpretations continue to improve (3), which might also provide some backup. Emergency physicians' ECG readings should routinely be over-read by experts to provide feedback and ongoing instruction. Finally, formal ECG training should be emphasized for physicians who might find themselves in a position of making important decisions based on their own interpretations.

Ellis W. Lader, MD
Benedictine HospitalKingston, New York, USA


1. Goldman L, Cook EF, Brand DA, et al. A computer protocol to predict myocardial infarction in emergency department patients with chest pain. N Engl J Med. 1988;318:797-803.

2. Pozen MW, D'Agostino RB, Selker HP, et al. A predictive instrument to improve coronary-care-unit admission practices in acute ischemic heart disease. A prospective multicenter clinical trial. N Engl J Med. 1984;310:1273-8.

3. Elko PP, Weaver WD, Kudenchuk P, Rowlandson I. The dilemma of sensitivity versus specificity in computer-interpreted acute myocardial infarction. J Electrocardiol . 1992; 24(Suppl):2-7.