Performance of prognostic models in critically ill cancer patients – a review

INTRODUCTION Prognostic models, such as the Acute Physiology and Chronic Health Evaluation (APACHE) II or III, the Simplified Acute Physiology Score (SAPS) II, and the Mortality Probability Models (MPM) II were developed to quantify the severity of illness and the likelihood of hospital survival for a general intensive care unit (ICU) population. Little is known about the performance of these models in specific populations, such as patients with cancer. Recently, specific prognostic models have been developed to predict mortality for cancer patients who are admitted to the ICU. The present analysis reviews the performance of general prognostic models and specific models for cancer patients to predict in-hospital mortality after ICU admission. METHODS Studies were identified by searching the Medline databases from 1994 to 2004. We included studies evaluating the performance of mortality prediction models in critically ill cancer patients. RESULTS Ten studies were identified that evaluated prognostic models in cancer patients. Discrimination between survivors and non-survivors was fair to good, but calibration was insufficient in most studies. General prognostic models uniformly underestimate the likelihood of hospital mortality in oncological patients. Two versions of a specific oncological scoring systems (Intensive Care Mortality Model (ICMM)) were evaluated in five studies and showed better discrimination and calibration than the general prognostic models. CONCLUSION General prognostic models generally underestimate the risk of mortality in critically ill cancer patients. Both general prognostic models and specific oncology models may reliably identify subgroups of patients with a very high risk of mortality.