The effects of computer-assisted electrocardiographic interpretation on physicians' diagnostic decisions

To evaluate the effect of computer-assisted interpretation of electrocardiograms (ECGs) on diagnostic decision making by primary care physicians.

Randomized controlled trial.

Primary care physicians' outpatient clinics in or near the Minneapolis/St. Paul metropolitan area.

Forty family physicians and general internists who were members of either of two large consortia of clinics.

Subjects evaluated ten clinical vignettes accompanied by ECGs and reported their diagnostic impressions. The vignettes were based on actual patient visits. Half of the subjects received ECGs with computer-generated reports, the other half received the same ECGs without reports.

ECG reading time; agreement with the clinical diagnosis; agreement with the computer report; diagnostic confidence.

The subjects receiving the reports were more likely to agree with the clinical diagnoses of the original cases, particularly for two vignettes in which the diagnoses were uncommon and were mentioned in the reports. The subjects receiving the reports were also more likely to make diagnoses that were consistent with the reports, even when the reports were erroneous. Those receiving the reports spent, on average, 15 seconds less time looking at each ECG, a 25% decrease.

In simulated cases, primary care physicians appear to use computer interpretations of ECGs when available, as shown by enhanced diagnostic accuracy and modestly reduced time spent reading the tracings. However, erroneous reports may mislead physicians. Since the effects of computerized ECG interpretation do not appear to have been uniformly favorable in this mock-clinical setting, it will be important to evaluate the effects of this technology in actual practice.