Long-term functional outcome after post-anoxic brain injury: model development and internal validation from a multicentre cohort

Background and objectives To identify biomarkers of long-term functional recovery after cardiac arrest (CA). Methods This is a prospective longitudinal cohort study based on the Veneto Registry of CA (July 2021- June 2023). We collected demographic, clinical, biochemical, and electroencephalography and short- and middle-latencysomatosensory-evoked-potentials (S-SSEP and ML-SSEP) within 72 hours. Functional scales were collected at discharge, 3, 6, and 12 months after the event. Multiple imputation was used to avoid statistical loss. Calibration was verified. A penalised logistic regression and a penalised continuation ratio were fitted to model the effect of covariates. The quality of prediction or fitting was verified computing the rank probability score for ordinal outcomes, the Brier score, and area under the ROC curve for binary outcomes. Decision curve analysis confirmed clinical utility. Results We included 160 persons (mean 62.80 ± SD 14.36 years; 123 males) out of 186. The interaction between continuous/nearly continuous EEG and N20/P25 predictor of survival (β = 0.840) and recovery (β = 1.160) of consciousness at 12 months after CA. ML-SSEP was a predictor of good outcome on all scales at 12 months after CA (0.739<β <1.708). Conclusion ML-SSEPs associated with standard neurophysiological assessments are robust predictors of long-term functional outcomes after CA. Significance ML-SSEPs associated with EEG and standard SSEP might be incorporated in routine clinical practice in ICU post-anoxic coma assessment.