Severity-of-illness scores for neutropenic cancer patients in an intensive care unit: Which is the best predictor? Do multiple assessment times improve the predictive value?

Abstract

To use three severity of illness scores to estimate the probability of hospital mortality among patients with cancer and neutropenia; to compare the performance of these scores, calculated at admission to an intensive care unit (ICU); and to test the improvement in estimation obtained by taking into account the first 72-hr period. Collection of data for every neutropenic patient hospitalized in the ICU during a 4-yr period. A comprehensive cancer center. Ninety-four patients were neutropenic at ICU admission. Their vital status was measured at hospital discharge. The new Simplified Acute Physiology Score (SAPS) II improved the estimation of hospital mortality compared with the original SAPS score. Using a simple score based on the number of acute organ system failures (OSFs) to classify the patients, good discrimination between survivors and nonsurvivors was observed (area under the receiver operating characteristic curves, 79 +/- 5 [SD] %). The relationship between successive scores and outcome was explored using recursive partitioning. Patients were first classified according to their OSF value on the first day of hospitalization in the ICU with a cutoff of two organ failures, and classification was then improved by taking into account the OSF score on the third day. For cancer patients hospitalized in an ICU for a neutropenic episode, the severity of illness and the risk of death can be accurately assessed by the SAPS II score and the number of acute organ failures at admission. The OSF values on the first and third days of hospitalization both provided information, allowing the classification of patients into groups with different probabilities of hospital mortality.