Several electroencephalographic (EEG) features —mainly the reactivity of background activity—have been suggested as reliable predictors of outcome for patients with post-anoxic coma (PAC). However, EEG in PAC often contains abundant EEG paroxysms (EP) that may hinder the detection of background EEG activity. We aimed to identify the features, among the different paroxysmal and non-paroxysmal EEG patterns, that may predict the outcome of patients with PAC.

We retrospectively reviewed the clinical and EEG characteristics of 67 patients with PAC and selected those with abundant EP. We classified EP according to several features and assessed their prognostic value for survival at 15 days. We calculated a global regularity score, as the sum of the value (1 if regular, 0 if not) attributed to each of 4 features of EP (duration, morphology, amplitude, and frequency).

The 35 patient-group with abundant EP showed a higher mortality than the group without abundant EP. Among 12 features of EP, four regularity features (regularity of EP duration, morphology, amplitude, and global regularity score) had a poor prognostic value. A global regularity score ≥ 3 showed a positive predictive value of 100 % for a poor outcome and a negative predictive value of 54 %, with good interrater consistency (Cohen's kappa = .63).

The presence of EP and their regularity features in PAC patients are strongly associated with poor outcome. We propose a global regularity score, easily derived from visual EEG inspection, that may be a reliable prognostic tool for these patients. Prospective and larger studies are needed to confirm these findings.