Prognostic scale for death in critically ill cancer patients due to neurological complications

Abstract

Introduction: The scales used to predict death in cancer patients are scarce and have not been adequately validated in populations with neurological complications. Objective: To design a prognostic scale for death in critically ill cancer patients due to neurological complications. Method: An observational, prospective cohort study was conducted from January 2017 to 2024 in the Oncology Intensive Care Unit of the Institute of Oncology and Radiobiology in Havana, Cuba. The study population consisted of 265 patients. Clinical, oncological, laboratory, and life support variables were studied. Binary logistic regression analysis was used to determine the variables with the best prognostic power for the model. The final variable was mortality in the Intensive Care Unit. Results: The final model consisted of the following variables: ECOG III (OR: 3.62; 95% CI: 1.05 - 12.46; p= 0.041), absence of pupillary reflex (OR: 9.35; 95% CI: 2.24 - 38.99; p= 0.002), intracranial hypertension (OR: 17.91; 95% CI: 3.02 - 105.91; p= 0.001), hyperglycemia (OR: 5.85; 95% CI: 2.07 - 16.49; p= 0.001), hypoxemia (OR: 19.09; 95% CI: 2.91 - 61.67; p= 0.000) and artificial mechanical ventilation (OR: 9.59; 95% CI: 2.59 – 35.51; p = 0.001).Conclusions: The prognostic score classified mortality into 3 risk groups: low (0 to 2 points), moderate (3 to 6 points), and high (7 to 13 points). Mortality was higher in patients with higher scores. Internal validation showed excellent calibration and discrimination.