To determine whether combining electroencephalography (EEG) with transcranial doppler (TCD) thresholds of vasospasm improves prediction of delayed cerebral ischemia (DCI) after subarachnoid hemorrhages (SAH).

DCI is a devastating complication of SAH. TCD remains the standard to monitor vasospasm, but is limited. EEG also predicts increased DCI risk (Vespa 1997; Kim 2017) and may surpass TCD (Rosenthal 2018). We hypothesize that EEG and TCD combined improves DCI prediction.

We retrospectively assessed 108 patients with moderate to severe SAH (2011-2015) with TCD and EEG. Middle cerebral artery (MCA) peak systolic velocity (PSV) and the presence of abnormal EEG patterns, (ictal-interictal continuum abnormalities; IICAs) were recorded daily. TCD velocities and their rates of increase between DCI and controls were compared (Wilcoxon Rank-Sum and slopes of linear mixed-effects models). The time relationship between TCD, IICAs, and DCI was visualized using cumulative probability curves. Multivariate logistic regression was used to identify significant TCD and IICA predictors of DCI.

MCA TCD median PSV were significantly higher in DCI vs controls on day of DCI (Wilcoxon Rank-Sum, p=0.03). Linear mixed effects modeling of TCD slopes from -4 to 0 days before DCI are significantly different (p<0.01). We found that IICAs, especially epileptiform discharges (ED) often precede high MCA velocities (PSV>200cm/sec) and DCI based on cumulative probability curves. High MCA PSV (p<0.01) and some IICA subtypes (+ED, +GPD, -GRDA) increase odds of DCI based on logistic regression (p<0.01). This regression model predicts DCI with an accuracy of 70% with an AUC of 0.738 (95% CI: 0.703–0.774) and improves upon accuracy of MCA or IICA presence alone (47% and 56%).

Our data suggest that combining both TCD values and IICA presence improves DCI prediction.