Objective: To assess the intrarater and interrater reliability of electrocardiogram (ECG) interpretation in critically ill patients and to assess the effect of knowledge of cardiac troponin values on these reliability estimates.
Design: Prospective cohort study.
Setting: Fifteen-bed medical-surgical intensive care unit.
Patients: Consecutive adults admitted over a 2-month period.
Measurements and results: All consecutive 12-lead ECGs were interpreted independently by two raters for the presence of myocardial ischemia or infarction and secondarily for specific ischemic ECG abnormalities. The ECGs were first interpreted blinded to the patient's troponin levels and reinterpreted on two separate occasions, blinded and unblinded to the troponin values. Results are reported using chance-independent agreement (phi) with associated 95% confidence intervals. For the presence of ischemia or infarction, the intrarater reliability ranged from fair to moderate (phi = 0.35 [95% confidence interval = 0.16, 0.52] and 0.59 [0.33, 0.77] for the two raters, respectively); interrater reliability was slight when blinded to troponin levels (phi = 0.18 [0.03, 0.32]) and increased to moderate when the raters were unblinded to troponin values (phi = 0.52 [0.33, 0.66], p value for the difference = .004). For specific ECG changes, the intrarater and interrater reliability were low for T-wave flattening, whereas detection of a left bundle branch block showed high reliability.
Conclusions: ECG interpretation in critically ill patients for the presence of myocardial ischemia or infarction showed moderate reliability at best; however, there was high reliability for specific ECG changes. Knowledge of the patient's troponin values increased the reliability for all studied ECG changes and resulted in a statistically significant increase in the interrater reliability for diagnosing myocardial ischemia or infarction. Additional studies assessing the appropriate methods of diagnosing myocardial ischemia and infarction and assessing the reliability of these diagnostic tests in critically ill patients are required.