The Predictive Value Of Aeeg In Patients With Disturbance Of Consciousness In Nicu

Objectives： Using Amplitude-integrated electroencephalography（aEEG）in the patients with disturbance of consciousness in NICU toinvestigate the predictive value of aEEG，and compare the performancesof aEEG with the Acute Physiology and Chronic Health Evaluation(APACHE) II and the Glasgow Coma Scale （GCS）。Patients andMethods：This study was undertaken over a2-year period (from December2011to February2013)in First affiliated hospital of chongqing medicaluniversity NICU, and included78consecutive patients(45men,33women)who were hospitalized for disturbance of consciousness. Their agevaried from18to83years, with mean±standard deviation of60.77±18.64years.The duration of each aEEG monitoring was not less than30min。TheAPACHE II score and GSC was recorded on the day of admission to thehospital。 Performances of these scoring systems were presented ascalibration and discrimination, which were measured by theHosmer–Lemeshow goodness-of-fit test and the area under the receiveroperating characteristic (ROC) curve。Results：twenty-six patient（s33.3%）died in14-days。The three modles all had a good calibrationin predicting the number of early deaths for the whole group(aEEG16deaths [95％CI(9，26)]，APACHE II31deaths[95％CI(26，35)]，GCS24deaths[95％CI(16，32)]。CI31．07to56．67)．For the divided groups，aEEG and APACHE IIhad a good calibration of prediction(aEEG χ2=0，P=1,d.f.=1；APACHE,χ2=8.041， P=0.429,d.f.=8）, while GCS showed poorcalibration（χ2=18.687，P=0.005，d.f.=6）。A11the three models showedacceptable discrimination with area under the ROC curve higher than0.7（aEEG0.781，APACHE II0.757，GCS0.768）． No significances werefound when comparing the area under the ROC curve of the three models。Conclusions： aEEG has a good prediction value in patients withdisturbance of consciousness for poor outcome and has a good calibrationof prediction Compare with the two conventional scale modles，while thediscrimination has no significances。...