Clinical Trial of an Algorithm for Outcome Prediction in Acute Circulatory Failure

The authors evaluated prospectively an index for outcome prediction previously developed retrospectively from the cardiorespiratory data of a series of 113 critically ill postoperative general surgical patients. A predictive score generated by nonparametric multivariate analysis of the observed value for each cardiorespiratory variable and the frequency distributions of survivors' and nonsurvivors' values of that variable at each stage of postoperative shock. An overall global predictive index was then generated from the sum of the weighted predictive scores in each variable. This predictive index was tested prospectively in a new series of 156 operations and was found to be 94% accurate for the values of the last available data set, suggesting that the method satisfactorily predicts outcome. This index may be used as an objective measure of the severity of illness; i.e., it may be used to track the clinical course of postoperative general surgical patients during periods of critical illness. It was concluded that the predictive index aids in evaluation of monitored cardiorespiratory variables, improves interpretation of physiologic alternation, and facilitates clinical decision-making of critically ill patients at the bedside.