Neuroworsening from a normal Glasgow Coma Scale Motor Score in the emergency department is an early predictor of neurosurgical intervention, hospital outcomes, and longitudinal disability in traumatic brain injury: A TRACK-TBI Study.

Abstract

Neuroworsening portends poor outcomes after traumatic brain injury (TBI) and is protocolized in intensive care unit (ICU) settings. The utility of neuroworsening assessments in non-ICU settings for intervention and prognostication requires further understanding. This study assessed relationships among neuroworsening in the emergency department (ED), clinicoradiological injury, blood-based biomarkers, neurosurgical interventions, and outcomes in TBI patients without Glasgow Coma Scale-Motor Score (GCS-M) impairment at ED arrival. Adult subjects from the 18-center Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI; ClinicalTrials.gov #NCT02119182) Study with ED arrival GCS-M = 6 and ED disposition GCS-M were analyzed. Neuroworsening was defined as ED disposition GCS-M < 6. Subjects received clinically-indicated head computed tomography (CT) scan within 24-hours (h) post-TBI. Clinical characteristics, acute plasma TBI biomarker levels (glial fibrillary acidic protein (GFAP), ubiquitin c-terminal hydrolase-L1 (UCH-L1); pg/ml), neurosurgical procedural interventions, hospital outcomes, and 3- and 6-month outcomes (Glasgow Outcome Scale-Extended (GOSE)) were compared. Multivariable logistic regressions examined predictors of neurosurgical interventions and unfavorable outcomes (GOSE ≤ 4) using adjusted odds ratios (AOR [95 % confidence intervals (CI)]). Cox proportional hazards model examined hospital discharge rate over time using adjusted hazard ratios (AHR). In 1210 subjects, 36 (3.0 %) had ED neuroworsening. Neuroworsening was associated with features of more severe injuries, including ICU admission (91.7 % vs. 30.3 %, p < 0.0001), post-traumatic amnesia duration (>24 h: 26.7 % vs. 4.2 %, p < 0.0001), and traumatic intracranial injuries on CT (72.2 % vs. 39.7 %, p = 0.00020). Neuroworsening subjects had higher GFAP (median = 1400 [Q1-Q3:864-3663] vs. 306 [82-839], p < 0.0001) and UCH-L1 (median = 459 [287-1036] vs. 170 [94-322], p < 0.0001), neurosurgical procedural interventions (38.9 % vs. 2.1 %, p < 0.0001), in-hospital mortality (8.6 % vs. 1.0 %, p = 0.018), hospital length of stay (6.9 days [Q1-Q3:4.8-16.8] vs. 2.2 days [1.3-4.0], p < 0.0001), and 3- and 6-month unfavorable outcomes (26.1 % vs. 3.5 %, p = 0.00040; 26.1 % vs. 3.7 %, p = 0.00050). Neuroworsening independently predicted neurosurgical interventions (AOR = 18.7 [95 % CI: 7.9-44.1], p < 0.0001), lower discharge rate [AHR = 0.35 [0.24-0.50], p < 0.0001), 3-month unfavorable outcome (AOR = 9.8 [3.0-31.9], p = 0.00010), and 6-month unfavorable outcome (AOR = 11.0 [3.1-38.7], p = 0.00020). ED neuroworsening is an early indicator of clinicoradiological TBI severity, and predicted neurosurgical procedural interventions, longer hospitalizations, and 3- and 6-month unfavorable outcomes. Higher blood-based TBI biomarker levels were associated with ED neuroworsening, suggesting their potential role to aid in the assessment of TBI patients at high risk of neurological deterioration.