Nonfatal myocardial infarction is, by itself, an inappropriate end point in clinical trials in cardiology.

In anumberofdouble-blinded, clinical trials in cardiology,1-4 theoccurrence ofanonfatal myocardial infarction hasserved asasecondary end point and,inatleast one,5 astheprimary endpoint. We believe thatthis practice maybeinappropriate. An assumption underlying theseparate analysis of nonfatal infarction isthatthere maybepathophysiological differences between fatal andnonfatal infarctions sothat theeffect oftreatment ononeofthese outcomes maybedifferent fromtheeffect oftreatmentontheother. However, thedifference between a fatal andanonfatal myocardial infarction isoften the result ofchance factors (e.g., having anambulance called immediately after theischemic attack, being brought toahospital where first-rate coronary intensive careisoffered), notofbiological ones. Underacompeting assumption that thedifference between afatal andanonfatal infarction ismainly a matterofseverity, inferences abouttheeffect of treatment onnonfatal infarctions maybeambiguous. Forexample, areduction intheincidence ofnonfatal infarction couldrepresent a harmful effect ifthe intervention hasnoeffect ontheoverall incidence of infarctions butincreases their severity sothatmore infarctions arefatal. Ontheother hand, areduction intheincidence ofnonfatal infarction could representabeneficial effect iftheintervention reduces the overall incidence ofinfarctions without necessarily affecting those that arefatal. Theseconflicting possibilities suggest that whenthere isresearch interest innonfatal infarctions, theyshould beanalyzed in tandemwithfatal infarctions andnotbythemselves. Ifagroupofpatients experience areduction inthe