Recurrent myocardial infarction. 2. Possibilities of prediction

1306 men less than 68 years of age who survived a first myocardial infarction (MI) during 1968-1977 were followed up between 2 and 12 years. The mean follow-up time was 6.5 years. The patients were unselected and paid regular visits to a Post-MI Clinic where treatment was standardized. The autopsy rate was high and the follow-up of endpoints was complete. The diagnosis of a non-fatal reinfarction was based on conventional clinical criteria, and the diagnosis of a fatal reinfarction on autopsy findings of a recent myocardial injury and/or a fresh coronary thrombus. The patients were randomly assigned to two halves. One was used only for derivation of the predictive models, and the other only for validation. Common clinical variables judged to be prognostically important were selected. Among variables available at discharge from hospital a history of hypertension, angina pectoris or diabetes before the MI and the maximal serum ASAT during the MI were independently related to reinfarctions during the follow-up. A predictive index was formed and validated. The rate of reinfarction among risk quartiles in the validation sample increased from 24 to 38% (P = 0.003). The aetiologic fraction (the percent of reinfarctions predicted by the index) was 24%. Among variables from the follow-up only cessation of smoking after the MI had independent predictive power. A new predictive index including this variable was formed and validated. The reinfarction rate in the risk quartiles increased from 15 to 39% (P less than 0.001). The aetiologic fraction was 44%. When only reinfarctions occurring before the median follow-up time of 21 months were considered, the aetiologic fraction was 62%.