Multiple predictors and multiple outcomes after myocardial infarction

Evaluating a method of treatment depends on the ability to define outcome. While this is easy if the outcome measures are concrete, such as survival, recurrence of disease or return to work, it is very difficult when outcome depends on assessment of the quality of a patient's life. This is particularly relevant in cardiac rehabilitation. Regression analysis has been used to predict physical, psychological and social outcome in 72 patients one year after a myocardial infarction using information available while the patients were in hospital recovering from the acute illness. Some aspects of outcome were successfully predicted: employment prospects, continuing emotional upset and the patient's attitude to and complaints about residual problems and difficulties. The study has shown the value of using multiple regression techniques which allow all putative predictors to be considered together, thus facilitating the elimination of relatively unimportant measures.