P39-T Predicting outcome in patients with moderate to severe traumatic brain injury using electroencephalography. Issue 7 (July 2019)

• Record Type: Journal Article
• Authors: Haveman, Marjolein E.; van Putten, Michel J.A.M.; Beishuizen, Albertus; Hom, Harold W.; Eertman-Meyer, Carin J.; Tjepkema-Cloostermans, Marleen C.
• Is Part Of: Clinical neurophysiology. Volume 130:Issue 7(2019:Jul.)
• Page Start: e50
• Publication Date: 2019-07
• Subjects: Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2019.04.402
• Languages: English
• ISSNs: 1388-2457

Abstract

Abstract : Background: We developed a multifactorial model combining quantitative electroencephalography (qEEG) measurements and clinically relevant parameters for outcome prediction of patients with moderate to severe TBI. Material and methods: Fifty-seven patients with moderate to severe TBI were included: a training ( n = 38) and validation set ( n = 19). Continuous EEG was registrated during the first 7 days of ICU admission. Patient outcome at 12 months was dichotomized as poor (Extended Glasgow Outcome Score, GOSE 1–2) or good (GOSE 3–8). Prediction models were created using a Random Forest classifier based on 23 qEEG features, age and mean arterial blood pressure (MAP) at 24, 48, 72 and 96 h after TBI and combinations of two time intervals. After optimization, we added parameters from the International Mission for Prognosis And Clinical Trial Design (IMPACT) predictor, comprising core, CT and laboratory parameters at admission. Predictive ability for the prediction of these models was assessed in both training (using leave-one-out) and validation set. Results: Our best model used 8 qEEG parameters at 72 and 96 h after TBI, MAP, age and 9 other IMPACT parameters. This model had high predictive ability on both training set (AUC = 0.94, specificity for poor outcome = 100%, sensitivity = 75%) and validation set (AUC = 0.81, specificity = 75%, sensitivity = 100%). The IMPACT predictor independently predicted both groups with an AUC of 0.74 (sensitivity = 65%, specificity = 81%) and 0.84 (sensitivity = 88%, specificity = 73%) respectively. Conclusions: Multifactorial Random Forest models using qEEG parameters in combination with clinical admission, laboratory and CT parameters have the potential to reliable predict outcome in patients with moderate to severe TBI patients.