Related Experiment Video
Updated: May 10, 2026
07:21
Systems Analysis of the Neuroinflammatory and Hemodynamic Response to Traumatic Brain Injury
Published on: May 27, 2022
3.2K
A Multivariable Model to Predict Risk of Dementia or Death in Older Veterans With Traumatic Brain Injury
Deborah E Barnes1, Yixia Li2, W John Boscardin3
1Departments of Psychiatry and Behavioral Sciences and Epidemiology & Biostatistics, University of California, San Francisco.
Neurology
|August 11, 2025
Summary
A new model predicts dementia or death risk in older veterans after traumatic brain injury (TBI). It uses electronic health records to identify high-risk individuals, aiding early intervention for TBI survivors.
Area of Science:
- Neurology
- Geriatrics
- Public Health
Background:
- Traumatic brain injury (TBI) increases mortality and dementia risk in older adults.
- No current models predict long-term outcomes after TBI in this population.
- Identifying at-risk individuals is crucial for timely intervention.
Purpose of the Study:
- Develop a prognostic model for 5-year dementia or death risk.
- Utilize electronic health record (EHR) data for prediction.
- Focus on older veterans with a history of TBI.
Main Methods:
- Retrospective cohort study of US veterans aged 55+ with TBI.
- Analysis of EHR data from 2002-2019.
- Multinomial logistic regression with LASSO for model development.
Main Results:
- Model accurately predicted dementia (C-statistic=0.756) and death (C-statistic=0.783).
- Key predictors included age, male sex, inpatient visits, and comorbidities (e.g., Parkinson's, diabetes, epilepsy).
- Risk varied significantly, with 3% dementia risk in the lowest decile vs. 43% in the highest.
Conclusions:
- A single prognostic model using EHR data can predict TBI-related dementia or death risk in older veterans.
- The model demonstrates good predictive accuracy.
- Further research is needed to integrate this model into clinical practice for improved care.
