Abstract P41: Integrating Demographics, TCD and EEG Diagnostic Modalities Improves Delayed Cerebral Ischemia Prediction After Subarachnoid Hemorrhage

Abstract

Introduction: Delayed cerebral ischemia (DCI) is the leading complication of subarachnoid hemorrhage (SAH). Because DCI was traditionally thought to be caused by large vessel vasospasm, transcranial Doppler ultrasounds (TCDs) have been the standard of care. Continuous EEG has emerged as a promising complementary monitoring modality and predicts increased DCI risk. Clinical variables have also been used in DCI prediction. We hypothesize integrating these diagnostic modalities improves DCI prediction. Methods: We assessed 107 patients with moderate-severe SAH (2011-2015) who had both TCD and EEG monitoring during hospitalization. Clinical demographics, including Hunt-Hess and aneurysm treatment (clipping/coiling), were collected via retrospective chart review. Middle cerebral artery (MCA) peak systolic velocities (PSV) and the presence or absence of epileptiform abnormalities (EA), defined as seizures, epileptiform discharges, and rhythmic/periodic activity, were recorded daily. Logistic regressions were used to identify EEG, TCD, and clinical variables associated with DCI. Group-Based Trajectory Modeling (GBTM) was used to account for changes over time by identifying distinct group trajectories of MCA and EA associated with DCI risk. Results: Independent predictors of DCI in logistic regressions are: presence of high MCA velocity (PSV≥200cm/s) and presence of EA on or before day 3. There are 2 univariate GBTM trajectories of EA (%DCI in group 1=32.1, group 2=70.4) significantly associated with DCI, but MCA velocity trajectories are not significant. Logistic regression and GBTM models using both TCD and EEG monitoring improve upon models using either modality alone. Hunt-Hess score at admission and aneurysm treatment as covariates further improved model performance. The best models used both TCD and EEG monitoring modalities and clinical variables as predictors (logistic regression: Se=90%, Sp=70%; GBTM: Se=89%, Sp=67%). Conclusions: EEG and TCD biomarkers combined provide the best prediction of DCI, compared to either alone. Models that considered the timing of EA and high MCA velocities plus clinical variables improved model performance.