The Predictive Validity of the Rotterdam Computed Tomography Score on Withdrawal of Life-Sustaining Therapies in Patients with Traumatic Brain Injury

Abstract

Introduction: The Rotterdam CT (RCT) score is a prognostic metric used to predict mortality in patients with traumatic brain injury (TBI) using injury characteristics observed on initial CT. Withdrawal of life-sustaining therapies (WLST) following trauma may be related to certain patient factors such as age and injury severity, but its relationship to RCT remains unknown. Objectives: The present study aims to determine whether the RCT score can be used as a reliable predictor of WLST in patients with TBI and to validate previous studies demonstrating that the RCT score can reliably predict mortality in patients with TBI. Methods: All patients ≥ 18 years old with TBI were identified from the MetroHealth TBI Database, a registry of patients admitted to MetroHealth Medical Center (Cleveland, OH) with TBI between 2018-2021. We collected factors related to RCT score, baseline demographic data, clinical presentation, neurosurgical intervention, and do not attempt resuscitation (DNAR) orders. The primary outcome measure was WLST, and the secondary outcome measure was mortality. Binary logistic regression models were created to determine factors independently associated with WLST and mortality. Results: A total of 1110 patients were identified. The median age was 61 years old and 69.3% (770/1110) were male. In this cohort, 11.3% (n = 125) of patients underwent WLST. A bivariate logistic regression model revealed that RCT score (continuous) (odds-ratio (OR) 1.6 [95% confidence-interval (95%-CI) 1.2-2.2]), Glasgow Coma Scale (continuous; GCS) (OR 0.84 [95%-CI 0.78-0.91]), Injury Severity Scale score (continuous; ISS) (OR 1.04 [95%-CI 1.01-1.07]), ventricular drain (OR 4.7 [95%-CI 1.1-19.6]), and DNAR (OR 47.8 [23.8-95.6]) were all independently associated with WLST. An additional bivariate logistic regression model revealed that RCT score (OR 2.08 [95%-CI 1.49-2.91]), ISS (OR 1.06 [95%-CI 1.03-1.10]), GCS (OR 0.81 [95%-CI 0.74-0.88]), undergoing a craniectomy (OR 0.28 [95%-CI 0.09-0.83]), and DNAR (OR 73.3 [95%-CI 34.8-145.7]) were all independently associated with mortality. These models demonstrated excellent accuracy and discrimination ability for both WLST and mortality as endpoints, as demonstrated by concordance statistics of 0.97. Conclusion: In adult TBI patients, the RCT score can be accurately used to independently predict both WLST and mortality with high predictive validity.