APACHE II score for critically ill patients with a solid tumor: A reclassification study

To improve the accuracy of the Acute Physiology and Chronic Health Evaluation (APACHE) II model for predicting hospital mortality in critically ill cancer patients.

This was a prospective cohort study of 522 patients admitted to ICU with a solid tumor. We developed the "APACHE II score for critically ill patients with a solid tumor" (APACHE II_CCP score), in which typical variables of critically ill cancer patients were added to general APACHE II score. Calibration and discrimination were evaluated by Hosmer-Lemeshow test (H-L) and area under receiver operating characteristic curve (AROC), respectively. The improvement in predicting hospital mortality with the new model was assessed using a reclassification analysis by integrated discrimination improvement (IDI), net reclassification improvement (NRI; cut-off point of 20% in risk of death) and quantitative NRI (qNRI).

The hospital mortality rate was 13%. Discrimination was superior for APACHE II_CCP score (AROC=0.91 [95% CI 0.87-0.94; P<.0001]) compared to general APACHE II score (AROC=0.62 [95% CI 0.54-0.70; P=.002]). Calibration was better using APACHE II_CCP score (H-L; P=.267 vs. P=.001). In reclassification analysis, an improved mortality prediction was observed with APACHE II_CCP score (IDI=0.2994 [P<.0001]; total qNRI=134.3% [95% CI 108.8-159.8%; P<.0001]; total NRI=41.5% [95% CI 23.7-59.3%; P<.0001]).

The performance of APACHE II_CCP score was superior to that observed for general APACHE II score in predicting mortality in critically ill patients with a solid tumor. Other studies validating this new predictive model are required.