# Predicting chronic subdural hematoma risk in elderly patients with mild traumatic brain injury

## TL;DR

This study identifies risk factors for chronic subdural hematoma in elderly patients with mild brain injuries and creates a prediction model to help doctors assess risk.

## Contribution

A new risk prediction model for chronic subdural hematoma after mild TBI in elderly patients is developed and validated.

## Key findings

- Pathological CT findings at presentation were the strongest predictor of chronic subdural hematoma.
- A risk model achieved an area under the curve of 0.76 for predicting CSDH.
- Male sex, older age, and renal failure were strong predictors of CSDH.

## Abstract

Mild traumatic brain injury (TBI) is a frequent presentation in the emergency department. Chronic subdural hematoma (CSDH), most often caused by TBI, is increasing in incidence due to population aging and widespread anticoagulant use. Identifying mild TBI patients at high-risk of CSDH remains challenging. This study aimed to identify risk factors and to create a simple prediction model for CSDH following mild head injury.

We conducted a historical cohort study including patients aged 65 and older who presented with mild TBI to the emergency department of a large tertiary district medical center between 2000 and 2021. The primary outcome was the development of CSDH within 2–12 weeks after the head injury. A prediction model including demographics, comorbidities, and chronic-medication-use was developed.

Of 7,246 eligible patients, 92 developed CSDH. Age (OR = 1.03, 95% CI: 1.00–1.06, per one-year increase), male sex (OR = 2.31, 1.48–3.59), renal failure (OR = 2.16, 1.27–3.67), chronic use of anticoagulant medications (OR = 1.64, 0.77–3.49), and pathological computed tomography (CT) at presentation (OR = 12.73, 8.19–19.79), were most strongly associated with development of CSDH. Among patients who developed CSDH, 58% had pathological CT findings at presentation. A simple score-based risk model had an area under the receiver operating characteristic curve of 0.76 (0.67–0.85).

Pathological CT findings following mild TBI were the strongest predictor for CSDH. Male sex, older age, and renal failure were also strong predictors. The prediction model allows efficient bedside risk stratification and opens the possibility for targeted surveillance protocols in high-risk populations.

The online version contains supplementary material available at 10.1007/s00701-026-06788-5.