Anticoagulant Therapy and Myocardial Infarction-Reply

In Reply.—  We agree with Dr Selzer in wanting a study in which "anticoagulant therapy or its lack is the only variable." This is, in fact, the reason that we specifically proposed that a well-designed clinical trial be conducted.We cannot, however, agree with his contention that unless an experimental trial is conducted, it is worthless to analyze observational data. He raises the possibility that the lower case-fatality rate in our anticoagulated group could have been from a spuriously high rate in the nonanti-coagulated group. However, since our sample included nearly all hospitalized myocardial infarction patients in Baltimore, the possibility of selection biases that might have artifactually affected mortality is unlikely. That our figures do not match what Dr Selzer believes are "the best mortality figures" may simply reflect the fact that studies done in different populations rarely, if ever, yield identical results or that the "best available" figures may