• T Berghmans, M Paesmans, and J P Sculier.
• Critical Care Department and Thoracic Oncology, Institut Jules Bordet, Centre des Tumeurs de l'Université Libre de Bruxelles, Rue Héger-Bordet, 1, 1000 Brussels, Belgium. thierry.berghmans@bordet.be
• Support Care Cancer. 2004 Apr 1;12(4):234-9.

ObjectiveTo evaluate the effectiveness of a specific oncologic scoring system-the ICU Cancer Mortality model (ICM)-in predicting hospital mortality in comparison to two general severity scores-the Acute Physiology and Chronic Health Evaluation (APACHE II) and the Simplified Acute Physiology Score (SAPS II).Patients And MethodsAll 247 patients admitted for a medical acute complication over an 18-month period in an oncological medical intensive care unit were prospectively registered. Their data, including type of complication, vital status at discharge and cancer characteristics as well as other variables necessary to calculate the three scoring systems were retrospectively assessed.ResultsObserved in-hospital mortality was 34%. The predicted in-hospital mortality rate for APACHE II was 32%; SAPS II, 24%; and ICM, 28%. The goodness of fit was inadequate except for the ICM score. Comparison of the area under the ROC curves revealed a better fit for ICM (area 0.79). The maximum correct classification rate was 72% for APACHE II, 74% for SAPS II and 77% for ICM. APACHE II and SAPS II were better at predicting outcome for survivors to hospital discharge, although ICM was better for non-survivors. Two variables were independently predicting the risk of death during hospitalisation: ICM (OR=2.31) and SAPS II (OR=1.05).ConclusionsGravity scores were the single independent predictors for hospital mortality, and ICM was equivalent to APACHE II and SAPS II.

27 keywords...

hide…