Example 1. ECG for Chest Pain.
Consider the ambulatory patient with chest pain. (This example is from page 82-83 of Sackett DL, Haynes RB, and Tugwell P, Clinical Epidemiology: A Basic Science for Clinical Medicine (First Edition). Boston: Little, Brown, and Company, 1985. The data are from a study by Bartel HG, Behar VS, Peter RH, Orgain ES, and Kong Y. Graded exercise stress tests in angiographically documented coronary heart disease. Circulation 49: 348, 1974.)
Patients in the Bartel et al study had
various types of chest pain and were tested using both exercise
electrocardiography and coronary arteriography. Let us assume
the coronary arteriography indicates the patient's true disease
status: whether there is coronary artery stenosis of at least
70%. The study provides the number of patients who had the disease
and had a positive exercise ECG result (137 True Positives), who
had the disease but had a negative ECG (90 False Negatives), who
did not have the disease and had a positive ECG (11 False Positives),
and who did not have the disease and had a negative ECG (112 True
Negatives). What does this allow us to say about the accuracy
of the Exercise ECG?
Count of Cases
This study provides you with
the counts of cases in the cells of the table, so you can use
the 2x2probs calculator. The True Positives go in the upper left
cell of the 2 by 2 table. The False Positives are in the upper
right. The False Negatives are in the lower left, and the
True Negatives in the lower right cell. After you have hit the
"return" button, your table should look like this:
Resulting Statistics
Scroll below
and see the statistics which can be calculated from this table.
You can see that the sensitivity of Exercise ECG (the probability
the test will be positive, if the patient has the disease) is
.60 (60%), and the specifity (the probability the test will be
negative, if the patient does not have the disease) is .91. In
this study, the proportion of patients with disease was .65.
If a patient with chest
pain, like those in this study, has a positive Exercise ECG, the
probability of significant coronary artery stenosis goes from
the .65 before the test to .93 (the predictive value of a positive
test). If the Exercise ECG is negative, the probability decreases
from .65 to .55 (the predictive value of a negative test).
