T68. The kind of EEG reactivity assessed by quantitative analyses may predict good and poor outcome in comatose neurosurgical ICU patients

Abstract

EEG reactivity is defined as a change in frequency or amplitude upon stimulation. However, knowledge about the kind of EEG changes to different external stimulations and their relation to prognosis in comatose patients is sparse. This is despite the fact that EEG reactivity often is regarded as an important factor in outcome prediction and may influence decisions on when to stop life-sustaining therapy. In this study we aimed to quantitatively describe EEG changes in comatose neurosurgical patients due to standardised stimulations and relate the changes to outcome. A standardised stimulation protocol comprising four different pain stimuli (squeezing the nail-side with a pen in both upper limbs and one lower limb and sternal massage), one auditory stimulation (clapping) and one visual stimulation (eye opening) was used. Stimuli were applied for 30 s preceded by a resting period of at least 2 min. Patients were 39 comatose (Glasgow coma scale ⩽8) neurosurgical intensive care unit patients monitored with EEG because they did not wake up as expected. EEG power in the delta, theta, alpha, and beta band was calculated in the first 2, 5, 10, 15, 20, and 30 s of the stimulation epochs and in similar time intervals in the preceding resting epochs. EEG reactivity was quantified as the power measured in z-scores relative to average power of six resting epochs. Outcome was measured after 3 months on the Glasgow Outcome Scale. Significant reactivity was present in 848 out of 5348 analysed epochs (15.9%). Reactivity as an increase in EEG activity was more common (13.4%) than reactivity as a decrease in EEG activity (2.5%). Pain, without any difference between the four stimulation methods, was the most provocative stimulation type (20.4%) followed by sound (8.7%) and eye-opening (6.7%). Increase in EEG activity was related to poor outcome (specificity 100% (CI: 52–100%); sensitivity of 39% (CI: 23–58%)), was more common (13.4% of tests), and grew continuously during the 30-s stimulation epoch. Decrease in EEG activity was related to good outcome (specificity 100% (CI: 87–100%); sensitivity 33% (CI: 6–76%)), was rarer (2.5%), and peaked after 10–15 s of stimulation. Absence of reactivity did not predict poor outcome very well (specificity of 33% (CI: 6–76%); sensitivity 61% (CI: 42–77%)). Grand average analyses revealed that discrimination between good (n = 6) and poor (n = 33) outcome was best for pain stimulation in the theta and alpha bands 15 s after stimulation and for sound stimulation in the theta and alpha bands about 5–10 s after start of stimulation, while eye-opening did not discriminate. Knowledge about different kinds of EEG reactivity and their relation to outcome comprise an important step towards the development of a more objective, reliable and standardised procedure for evaluation of EEG reactivity.