Stephen J.LurieMD, PhD, Senior EditorIndividualAuthor
Copyright 2002 American Medical Association. All Rights Reserved.
Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To the Editor: Dr Welch and colleagues1 reported that patients with acute myocardial infarction
(AMI) with initially normal or nonspecific initial electrocardiogram (ECG)
results had lower mortality rates than those who presented with diagnostic
ECG results. These conclusions are based on odds ratios derived from models
comparing mortality in patients with normal or nonspecific ECG results and
in patients with ECG signs specific to AMI. Although odds ratios are important
measures of association, they do not describe the ability of information—here
normal, nonspecific, or specific ECG results—to help the clinician predict
the death or survival of a given patient.2
To provide this information, the authors should have focused on the positive
and negative predictive values.3 The relevant
positive predictive value is the proportion of patients dying in the diagnostic
ECG group (11.5%); the relevant negative predictive values would be the proportions
of patients surviving in the normal (94.3%) and the nonspecific (91.3%) groups.
These predictive values are not impressive.
Salmi LR. Predictive Value of the Electrocardiogram in Acute Myocardial Infarction. JAMA. 2002;287(6):711-712. doi:10.1001/jama.287.6.711