A multifactor model for predicting mortality in critically ill patients: A multicenter prospective cohort study.

Area of Science:

• Critical Care Medicine
• Biostatistics
• Clinical Prediction Modeling

Background:

• Predicting mortality in critically ill patients is crucial for resource allocation and patient management.
• Existing models may not fully capture the complexity of critical illness outcomes.
• Routine clinical variables offer a readily accessible data source for predictive modeling.

Purpose of the Study:

• To develop and validate a multifactor model for predicting in-hospital mortality in critically ill patients.
• To identify key clinical variables associated with increased mortality risk upon intensive care unit (ICU) admission.

Main Methods:

• Logistic regression analysis was employed to develop the predictive model.
• A cohort of 500 patients from eight university hospital ICUs was utilized.
• Model performance was assessed using discrimination (AUC) and calibration (Hosmer-Lemeshow test).

Main Results:

• The final model incorporated lactate level, neutrophil-to-lymphocyte ratio, acute physiology score, Charlson comorbidity index, and surgery type.
• The model demonstrated strong discrimination with an area under the receiver operating characteristic curve of 0.84.
• Good calibration was observed, indicating reliable prediction accuracy.

Conclusions:

• The developed multifactor model provides an effective tool for predicting mortality in critically ill patients at ICU admission.
• Routine clinical variables can be leveraged to create robust predictive models for critical care settings.
• This model can aid clinicians in early risk stratification and timely intervention for high-risk patients.