Prognostic assessment of comatose patients after cardiac arrest: construction of a simplified monitoring EEG risk score across different time windows.

The study population had a high rate of poor neurological outcomes, with 82% of comatose post-cardiac arrest patients experiencing poor outcomes.

• 251 comatose patients after cardiac arrest were included in the retrospective cohort study.
• 45 patients (18%) had good outcomes and 206 patients (82%) had poor outcomes.
• Patients were admitted to the ICU of the First Affiliated Hospital of Chongqing Medical University between January 2020 and December 2024.

All five EEG features incorporated into the SERS were significantly associated with poor neurological prognosis.

• Abnormal background amplitude, slow dominant frequency (δ/θ), discontinuous background, absent reactivity, and absent sleep waveforms were all significantly associated with poor prognosis.
• All associations reached statistical significance at P < 0.001.
• β-coefficients from univariate logistic regression for each EEG feature were used to construct the SERS.

The SERS demonstrated high prognostic accuracy across all three time windows assessed after cardiac arrest.

• AUCs of SERS were 0.912 at Day 1, 0.893 at Days 2–5, and 0.881 at >5 days after cardiac arrest.
• Performance was evaluated using receiver operating characteristic curves with assessment of AUC, accuracy, sensitivity, and specificity.
• The score maintained high predictive accuracy across early and later time windows, indicating temporal stability.

The SERS outperformed individual EEG features, commonly used EEG scores, and traditional clinical predictors in prognostic accuracy.

• Comparison was made against individual EEG features, commonly used EEG scores, and traditional clinical predictors at all three time windows.
• The SERS achieved AUCs of 0.912, 0.893, and 0.881 compared to lower values for comparator methods at Day 1, Days 2–5, and >5 days, respectively.
• Multivariate logistic regression was used to identify independent predictors of poor neurological outcome.

Risk stratification using SERS demonstrated clear separation of prognosis across low-, medium-, and high-risk groups at all time windows.

• Three risk groups (low, medium, high) were defined using the SERS.
• Clear separation of prognosis across risk groups was observed at Day 1, Days 2–5, and >5 days after cardiac arrest.
• This stratification supports the score's potential for clinical utility in guiding prognostic decisions.

The SERS was constructed from EEG features using β-coefficients derived from univariate logistic regression.

• EEG features included background amplitude, dominant frequency, continuity, reactivity, and sleep elements.
• β-coefficients from univariate logistic regression were incorporated into the simplified score.
• The study design was a retrospective cohort study.