Outcome prediction model for severe diffuse brain injuries: development and evaluation

Area of Science:

• Neurology
• Critical Care Medicine
• Medical Prognostics

Background:

• Intensive care resources for severe diffuse brain injury (SDBI) patients are limited.
• Predicting patient outcomes at admission is essential for optimal resource allocation.

Purpose of the Study:

• To develop a simple and effective model for predicting poor outcomes in SDBI patients.
• To aid in guiding initial therapeutic strategies for SDBI.

Main Methods:

• Retrospective analysis of 289 SDBI patients (GCS 3-8) including age, injury type, GCS, pupillary response, oculocephalic reflex, and CT findings.
• Stepwise logistic regression identified key predictors of poor outcome.
• A novel prediction model (NIMHANS model-NM) was developed and prospectively validated against existing models in 26 SDBI patients.

Main Results:

• Horizontal oculocephalic reflex, GCS motor score, and CT midline shift were significant predictors of poor outcome.
• The NIMHANS model (NM) demonstrated 75% sensitivity and 67% specificity for unfavorable outcomes.
• NM exhibited a high rate of false pessimistic predictions (33%).

Conclusions:

• The developed prediction model (NM) showed moderate accuracy in predicting unfavorable outcomes.
• Prediction models, including NM, are currently insufficient for guiding initial therapy in SDBI patients.