A prognostic model for long-term outcome in moderate-severe traumatic brain injury using multimodal neuromonitoring: a retrospective cohort analysis

Abstract

Background Moderate-to-severe traumatic brain injury (msTBI) remains a leading cause of long-term disability and mortality. While multimodal neuromonitoring (MMM) provides real-time insights into cerebral pathophysiology, its integration into prognostic models for long-term outcomes requires further validation. Objective To develop and evaluate a prognostic model for 6-month neurological outcomes in msTBI patients using acute-phase multimodal monitoring parameters, clinical variables, and comorbidities. Methods We conducted a retrospective cohort study of 143 adult patients with msTBI [Glasgow Coma Scale (GCS) 4–12] who underwent multimodal intracranial monitoring (ICP, CPP, PbtO₂, and cerebral microdialysis) for ≥48 h at Hospital between June 2022 and May 2025. Demographic, clinical, and monitoring data from the first 5 days were collected. The primary outcome was neurological function at 6 months, assessed by the Glasgow Outcome Scale-Extended (GOSE), dichotomized as favorable (GOSE 5–8) or unfavorable (GOSE 1–4). Multivariable logistic regression was used to identify independent predictors, and a prognostic model was constructed and internally validated using ROC analysis and the Hosmer-Lemeshow test. Results Of 143 patients, 86 (60.1%) had favorable outcomes and 57 (39.9%) had unfavorable outcomes (including 29 deaths). Independent predictors of unfavorable outcome were: older age (adjusted OR 1.055 per year, 95% CI: 1.015–1.098), lower admission GCS (OR 0.713, 95% CI: 0.537–0.947), elevated lactate/pyruvate ratio (LPR) (OR 1.129 per unit, 95% CI: 1.044–1.220), reduced brain tissue oxygen tension (PbtO₂) (OR 0.842 per mmHg, 95% CI: 0.730–0.972), and pre-existing coronary heart disease (OR 3.866, 95% CI: 0.998–14.976). The final model demonstrated excellent discriminative performance with an AUC of 0.882 (95% CI: 0.820–0.944) and good calibration (Hosmer-Lemeshow p = 0.523). A dose–response relationship was observed between the number of abnormal monitoring parameters and risk of poor outcome. Conclusion A prognostic model incorporating multimodal neuromonitoring parameters (LPR and PbtO₂), age, admission GCS, and coronary heart disease accurately predicts 6-month outcomes in msTBI patients. These findings support the clinical utility of integrated neuromonitoring for early risk stratification and personalized neurocritical care.