Myocardial infarction. How representative are autopsied subjects with this clinical entity?

We documented the differences in clinical features between 86 autopsied and 54 nonautopsied subjects who died of myocardial infarction to ascertain any bias that might be present in use of postmortem data. More than 200 historical, clinical, noninvasive, and hemodynamic aspects were compared. Of noninvasive aspects examined in all subjects, only 13 had significant differences (by chi 2 or unpaired t tests) between autopsied and nonautopsied subjects. However, there had been greater impairment of vital signs and hemodynamic aspects during early hospitalization in the autopsied vs the nonautopsied subjects. Further, the one-month survival rate was lower in autopsied subjects (31% vs 72%; P less than .01). We conclude that patients in severe congestive heart failure or shock and, presumably, with relatively large or complicated myocardial infarcts are more likely to die early and be autopsied. Those with better cardiac function live longer, and often die after having been released from the initial hospitalization; these subjects, presumably with smaller and uncomplicated infarcts, do not undergo autopsy. Before correlating pathological and clinical data from subjects with acute myocardial infarction, it is important to carefully analyze bias inherent in the selection of subjects to undergo autopsy.