O-42 Dynamic functional connectivity of the EEG in postanoxic coma

Abstract

Listen

Translate article

Background Predicting neurological recovery of patients with postanoxic coma remains a challenge. We investigate predictive properties of dynamic interactions between neuronal assemblies. Methods We performed a retrospective study of 709 comatose patients after cardiac arrest. We extracted 5 min EEG epochs every 2 h after resuscitation, up to 72 h, and applied automated artefact rejection. Per frequency band (Theta (4–8 Hz), Alpha (8–13 Hz)), instantaneous frequencies were obtained using the ridge algorithm. A link between two EEG locations was assumed if they shared the same instantaneous frequency. Link Rate (LR) was the mean number of links per second, Link Duration (LD) the mean duration per interaction. Differences between patients with good (CPC 1–2) and poor (CPC 3–5) outcome were compared. Results 47 patients were lost to follow-up. Patients with good outcome (n = 300) showed a higher LR 10–40 h after cardiac arrest in the theta band (good: 1.94 ± 0.21 Hz, poor: 1.86 ± 0.40 Hz (median ± IQR) at 24 h) and between 12–24 h in the alpha band (good: 1.72 ± 0.15 Hz, poor: 1.66 ± 0.33 Hz at 24 h). LR stabilized after 44 h in the theta band and decreased in the alpha band. LD with a good outcome was always lower than with a poor outcome in the theta band (good: 85 ± 6 ms, poor: 88 ± 2 ms at 24 h), and for the first 42 h in the alpha band (good: 91 ± 5 ms, poor: 96 ± 2.5 ms at 24 h). Conclusions During the first 40 h after cardiac arrest, lower LR and higher LD are associated with poor outcome. Analysis of predictive values for individual patients are ongoing and will be presented.