Dunn P M, Levinson W
Internal Medicine Training Program, Good Samaritan Hospital, Portland.
Arch Intern Med. 1990 Sep;150(9):1917-9.
It is widely held that the interpretation of a diagnostic test can be biased by clinical information. Many clinicians are cautious about providing information that may bias test interpretation and result in the decreased utility of the test. We quantitatively assessed the effect of clinical information on electrocardiogram (ECG) interpretation of acute myocardial infarction (AMI). Three cardiologists were asked to give independent written interpretations of 52 ECGs devoid of clinical information. Three weeks later, the same set of ECGs was reinterpreted with clinical information that was either strongly or mildly suggestive of AMI. Of the 156 pooled ECGs, 22 changes (14%) in interpretation with the addition of clinical information were made. Eleven (50%) of the changes were from negative to positive, and 11 were from positive to negative. Changes in interpretation were not affected by the nature of the clinical information or by whether the ECG was typical for AMI. No significant changes in the sensitivity or specificity with the addition of clinical information were present. This study did not demonstrate the anticipated effect of clinical information on the ECG interpretation of AMI. Interpretation of other tests may also be unaffected by the provision of clinical information.
