Development of END-of-Life ScorING-System to identify critically ill patients after initial critical care who are highly likely to die: a pilot study.

Background. High quality palliative care should be provided for the dying patients in the intensive care unit (ICU). The aim of this pilot study is to develop a scoring system, the "END-of-Life ScorING-System" (ENDING-S), that may help to identify ICU patients at very high risk of dying after initial response to the intensive treatments and which could be used to facilitate palliative care. Methods. The characteristics of longer-term ICU patients (>4 days) who are at very high risk of dying were identified through an analysis of the literature and developed in a retrospective cohort of patients. ENDING-S Score was developed through a multivariate analysis. Model accuracy was tested through ROC and Hosmer-Lemeshow analysis for model discrimination and calibration respectively. Cross validation was used to provide internal model validation. Results. Potential predictors of death were identified and applied to 80 ICU patients. Significant variables in the multivariate analysis were the ratio of the ICU days in which the patient needs mechanical ventilation or vasoactive drugs divided by the total ICU days, the total ICU length of stay, and current sepsis. Analysis of accuracy showed a ROC-AUC equals to 0.98 (95% CI, 0.97 to 1), and agreement between the predicted probability and the observed frequency of death in the ICU was observed (P>0.05 at Hosmer-Lemeshow test). The internal validation confirms these results. Conclusion. In these preliminary results, ENDING-s shows acceptable calibration and discrimination properties. ENDING-S may raise awareness among ICU physicians about the importance of integrating palliative care into ICU daily practice.