The Single-cutoff Trap

Abstract

Quantitative analysis of exercise electrocardiograms has been emphasized by many inves tigators. Specific problems have been found when a single cutoff is used to define a positive or a negative test: a single cutoff does not distinguish stress electrocardiography results that are slightly positive from those that are markedly positive. This may lead clinicians to un derweigh strong evidence for or against coronary artery disease. This study evaluated cli nicians' quantitative analysis of stress electrocardiograms. Two hundred and thirty-five physicians interpreted the results of mildly positive (1.2 mm ST-segment depression) and strongly pos itive (2.2 mm ST-segment depression) stress electrocardiograms. Their posttest probability estimates were too high for a mildly positive test (0.62 ± 0.02 versus actual of 0.38; p <0.001) and too low for a strongly positive test (0.77 ± 0.01 versus actual of 0.98; p <0.001). Physicians should understand decision aids and should use multiple rather than single cutoffs to interpret the results of stress electrocardiography.