A Nomogram for Predicting 28-Day Mortality in Critically Ill Patients With Prostate Cancer: A Retrospective Cohort...

Area of Science:

• Oncology
• Critical Care Medicine
• Biostatistics

Background:

• Critically ill patients with prostate cancer (PCa) represent a complex and understudied group.
• There is a lack of precise prognostic tools for this specific demographic.
• This study aimed to develop and validate a nomogram for predicting 28-day mortality in PCa patients requiring intensive care.

Purpose of the Study:

• To establish and validate a robust nomogram for predicting 28-day mortality risk in critically ill prostate cancer patients.
• To compare the nomogram's performance against traditional prognostic metrics.

Main Methods:

• Retrospective analysis of data from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database.
• Patients were divided into training (70%) and validation (30%) sets.
• Least Absolute Shrinkage and Selection Operator (LASSO) regression with cross-validation was used for predictor selection and model construction.
• Model performance was assessed using ROC curves, calibration plots (Hosmer-Lemeshow tests), and decision curve analysis (DCA).

Main Results:

• The final model identified six key predictors of 28-day mortality.
• The nomogram demonstrated superior discriminative accuracy with an AUROC of 0.814 (training) and 0.809 (validation), outperforming SAPS II, SOFA, and CCI.
• Strong calibration was indicated by high p-values (>0.05) in Hosmer-Lemeshow tests.
• DCA confirmed the nomogram's superior net clinical benefit compared to conventional scoring systems.

Conclusions:

• The developed 28-day mortality prediction model for critically ill PCa patients enables effective early risk stratification.
• This tool has the potential to guide timely interventions and improve survival rates in this high-risk population.