# Risk Assessment of Venous Thromboembolism in Neurocritical Patients: Construction and Validation of a Clinical Prediction Model

## TL;DR

This study creates a prediction model to assess the risk of blood clots in critically ill neurology patients, helping doctors prevent such events more effectively.

## Contribution

The study introduces a validated clinical prediction model specifically tailored for venous thromboembolism risk in neurocritical patients.

## Key findings

- The model achieved strong discrimination with an AUC of 0.763 in training and 0.809 in testing.
- Key predictors included age, NICU length of stay, APTT, D-dimer, and clinical factors like tracheotomy and antibiotic use.
- The model showed good calibration and consistent clinical net benefit across risk thresholds.

## Abstract

Venous thromboembolism (VTE) remains a significant challenge in neurocritical care, with limited tailored risk assessment tools available for clinical practice. This study aimed to develop and validate a practical prediction model to support VTE prevention strategies in neurocritical patients.

A total of 605 neurocritical patients were retrospectively enrolled in the neurologic intensive care unit (NICU) from May 2022 to April 2024. The eligible patients were randomly divided into a training dataset and a testing dataset in a ratio of 7:3. Variables with significant univariate effects in the training dataset were selected for multivariable stepwise regression analysis. The model fitting goodness was tested using the Hosmer–Lemeshow test, the area under the receiver operating characteristic (ROC) curve (AUC) was used to evaluate the model discrimination, and decision curve analysis (DCA) was used to test the clinical value of the model.

The final model identified age, length of stay in the NICU, activated partial thromboplastin time (APTT), D‐dimer, tracheotomy duration, pulmonary infection, antibiotic use, and dehydrating agents as key predictors. The nomogram demonstrated excellent discrimination (AUC 0.763, 95% CI: 0.714–0.811 in the training dataset; AUC 0.809, 95% CI: 0.732–0.885 in the testing cohort), strong calibration (Hosmer–Lemeshow test: p = 0.126 [training]; p = 0.823 [testing]) and consistent clinical net benefit across prevention thresholds (10%–40% risk probability).

The risk prediction model developed in this study can effectively predict VTE occurrence in neurocritical patients with good discrimination and clinical utility, providing a valuable tool for identifying high‐risk individuals and performing early prevention and treatment measures.