[Predicting the prognosis for severe brain injury patients: short-latency somatosensory evoked potential combined...

Objective: To explore the the effectiveness of using short-latency somatosensory evoked potential(SLSEP) combined with electroencephalogram(EEG) reactivity to predict the prognosis of severe brain injury(SBI) patients. Methods: Consecutive patients with SBI admitted in neurosurgery intensive care unit(NSICU) at Xiangya Hospital of Central South University from July 2018 to January 2019 were prospectively collected. SLSEP and EEG were recorded in these patients in NSICU within two weeks after injury onset. EEG reactivity(EEG-R) was tested during EEG signal stabilization. In addition, the concentrations of serum neuron-specific enolase (NSE) and S100 protein were also detected. All patients were evaluated with Glasgow Outcome Scale(GOS) during 12 months' follow-up. GOS grade 3 to 5 was defined as favorable group, and GOS grade 1 to 2 was defined as unfavorable group. The association of relevant predictors with patient's prognosis was assessed. The area under the receiver operating characteristic (ROC) curve (AUC) was used to evaluate each potential predictor. Results: Forty-three patients were included in the study, with 26 patients of favorable outcomes and 17 patients with unfavorable prognosis. Univariate analysis revealed that the Glasgow Coma Scale (GCS) score, the concentration of serum NSE, EEG-R, the amplitude of SLSEP were all associated with the prognosis after 12 months' follow-up. Moreover, the AUC for prediction of favorable prognosis by GCS, NSE, EEG-R, SLSEP was 0.661(95%CI: 0.493-0.829), 0.697(95%CI: 0.531-0.862), 0.718(95%CI: 0.557-0.879) and 0.758(95%CI: 0.609-0.907) respectively. However, there was no significant difference of age, gender, pupillary light reflex and S100 protein between the two groups. Furthermore, multiple logistic regression analysis showed that only SLSEP amplitude (OR=2.058, 95%CI: 0.867-4.888) and EEG-R(OR=3.748, 95%CI: 0.857-16.394) were independent predictors of favorable prognosis, and the prognostic model containing these two variables yielded an predictive performance with an AUC of 0.798. Conclusion: The higher amplitude of SLSEP and the existence of EEG-R are predictors of good prognosis in SBI patients, and the combined use of SLSEP and EEG-R in predicting the prognosis of SBI patients is more reliable.