Effect of computer reported clinical information on the cardiologist's behavior in the interpretation of electrocardiograms

The effect of computer-reported clinical information on the cardiologist's behavior in the interpretation of electrocardiograms (ECGs) was studied using 100 inpatient ECGs. Using an automated medical record system (HELP), the pertinent demographic and clinical information was printed on a clinical label. Two cardiologists independently read these 100 ECGs twice, once without the label and once with the label provided. A sample of twenty-five ECGs was chosen from the 100 ECGs for estimation of intraobserver variability. These twenty-five ECGs were independently read again with and without the clinical label. An appropriate time delay between the readings was allowed to insure independence of the readings. The following results were observed: (a) The myocardial infarction (MI) and chamber enlargement interpretations, the intraobserver variability of one cardiologist was reduced from 25 percent to 14.7 percent and a corresponding decrease from 22.7 percent to 6.7 percent for the other cardiologist. Statistically, the decreases in both cases were found to be significant. (b) The overall inter-observer variation showed statistically insignificant reduction (16.3 percent to 13 percent). However, for MI interpretation, the corresponding reduction from 24 percent to 13 percent might be clinically significant. (c) For MI, chamber enlargement and repolarization change interpretations; the frequency of agreement between the cardiologists was approximately 2.00 times that of the disagreement in the diagnoses. (d) For the interpretation of repolarization changes in the twenty-five ECGs, the frequency of changes from nonspecific to a specific interpretation increases from 2 percent to 36 percent after the introduction of the clinical label. Therefore, it was concluded that the availability of demographic and clinical information in a total hospital information system provides useful data to the cardiologist for the interpretation of ECGs.