To assess the prognosis of cancer patients in an intensive care unit (ICU), to compare the capabilities of severity scoring systems to predict hospital death, and to improve prediction by adding new variables.
PATIENTS AND METHODS
Cohort study in a medical-surgical ICU of a university hospital. Demographic and oncologic characteristics were collected along with death records for all nonsurgical cancer patients admitted between January 1995 and June 2000. Severity scores and risk of death were calculated.
In the cohort of 250 patients studied, the hospital mortality rate was 58% and the ICU mortality rate was 38.8%. The best predictions were made with the third version of the Acute Physiology and Chronic Health Evaluation (APACHE III), the total maximum Sequential Organ Failure Assessment (SOFA) score, and the total maximum Multiple Organ Dysfunction Score (MODS). The APACHE II and the Simplified Acute Physiology Score (SAPS), version II, were good predictors, whereas the systems of the International Council on Mining and Metals overestimated hospital mortality and the Modality Prediction Model at 0 and 24 hours (MPM0 and MPM24) and the Logistic Organ Dysfunction System underestimated it. The total maximum SOFA and MODS scores had the greatest discriminating capability and the SOFA0, the MODS0, MPM0, and MPM24 had the poorest. All assessment systems except the APACHE III improved when we added new mortality-associated variables: prior functional status, diabetes, radiographic lung infiltrates, mechanical ventilation, and vasoactive support.
Medical oncology patients should not all be denied intensive care. None of the systems assessed offer clinically relevant advantages for predicting hospital mortality in nonsurgical oncology patients in the ICU, although we recommend the SAPS II because it includes oncologic variables, is easy to score, and has good prognostic capability.
CITATION Rev Esp Anestesiol Reanim. 2007 Aug-Sep;54(7):405-13