Refining intensive care unit outcome prediction by using changing probabilities of mortality.

Estimating prognosis is potentially useful as a measure of ICU performance and as a guide for the clinical care of individual patients. In this study, mortality prediction models (MPMs) for patients in an adult general medical-surgical ICU were derived from data gathered at ICU admission and after 24 and 48 h of ICU care. A predictive model was developed which incorporated a sequence of probabilities collected over time in the ICU. The results of this study suggest that using serial observations may enhance substantially the usefulness of the MPM as a vehicle for helping families anticipate the patients' likely outcome.