Charlson M E, MacKenzie C R, Ales K L, Gold J P, Fairclough G F, Shires G T
Department of Medicine, Cornell University Medical College, New York, NY 10021.
J Clin Epidemiol. 1989;42(1):25-34. doi: 10.1016/0895-4356(89)90022-x.
The objective of this study was to evaluate different approaches to the diagnosis of post-operative myocardial infarction. A total of 232 patients, mostly hypertensive and/or diabetic patients, who were undergoing elective non-cardiac surgery were evaluated pre-operatively. They were followed serially from the day of operation to discharge or the sixth post-operative day with daily clinical evaluations, electrocardiograms, creatine kinase and creatine kinase isoenzymes. In total 22% (51/232) of the patients had post-operative ECG changes in two or more leads. Only 1% developed new Q waves; most of the changes involved changes in the T or ST segments. Seventy percent of patients who had changes in their electrocardiogram were completely asymptomatic. The highest risk of ECG changes or symptoms occurred on the day of operation and the first post-operative day; evidence of post-operative infarction was infrequent after the second post-operative day. Creatine kinase levels rose an average of 250-300 IU on the first and second post-operative day (also the peak time for post-operative ECG changes), reducing its utility as an adjunct to the diagnosis of post-operative infarctions. Importantly, 52% (12/23) of the patients who had greater than or equal to 5% MB isoenzyme had neither ECG changes nor symptoms; the diagnosis of a myocardial infarction should not be made in these patients. In summary, most patients who experience ischemia or infarction post-operatively are asymptomatic. Symptoms should not be required for the diagnosis of post-operative infarction. Seemingly minor differences in criteria can produce major discrepancies in post-operative myocardial infarction rates (from 1 to 9%). The development of a final set of criteria will require further study but the diagnosis of post-operative infarction should probably be based on ECG changes, their duration and consistency, and the association of a positive MB fraction.
