The Pre-ECPR score for predicting favorable neurological outcomes after extracorporeal cardiopulmonary resuscitation: protocol for external validation in the Prague OHCA trial cohort

TL;DR

This paper describes a protocol to validate a new score for predicting good neurological outcomes in patients undergoing ECPR.

Contribution

The study introduces a pre-specified validation framework for the Pre-ECPR score in an independent OHCA cohort.

Findings

The Pre-ECPR score showed better discrimination (AUC 0.79) than ELSO criteria (AUC 0.63) during development.

The validation uses 92 ECPR-treated patients from the Prague OHCA trial with systematically collected data.

The study will evaluate calibration, discrimination, and risk stratification of the Pre-ECPR score.

Abstract

•Protocol for external validation of Pre-ECPR score for patient selection in ECPR.•In development, the score achieved AUC 0.79 vs 0.63 for ELSO selection criteria.•Prague OHCA-trial provides an ideal external validation cohort: 92 ECPR patients.•Prospective data with all physiological predictors systematically collected.•We present a pre-specified validation framework with blinded outcome assessment.

Protocol for external validation of Pre-ECPR score for patient selection in ECPR.

In development, the score achieved AUC 0.79 vs 0.63 for ELSO selection criteria.

Prague OHCA-trial provides an ideal external validation cohort: 92 ECPR patients.

Prospective data with all physiological predictors systematically collected.

We present a pre-specified validation framework with blinded outcome assessment.

No clear consensus exists for extracorporeal cardiopulmonary resuscitation (ECPR) patient selection in refractory cardiac arrest. Evidence suggests that current binary criteria yield either low favorable outcome rates, when being liberal, or exclude potential survivors, when being restrictive. The Pre-ECPR score, which uses logistic regression to give a single predicted probability of favorable outcome (modified Rankin Scale 0–3), demonstrated significantly better discriminatory ability than the Extracorporeal Life Support Organization “Example of selection criteria for ECPR”. This protocol outlines the methodology for external validation of the Pre-ECPR score using the ECPR-treated patients from the randomized Prague Out-of-Hospital Cardiac Arrest (OHCA) trial.

The validation cohort comprises the 92 ECPR-treated patients of the Prague OHCA trial. The Pre-ECPR score incorporates eight predictors: age, no-flow time/initial rhythm combination, total cardiac arrest duration, signs of life, pupil dilation, end-tidal CO2, regional cerebral oxygen saturation, and arterial pH, all measured during resuscitation. Model performance will be evaluated through calibration (calibration-in-the-large, calibration slope, and calibration plots); discrimination (area under the receiver operating characteristics curve and Tjur’s discrimination index); and risk stratification analysis at probability thresholds. If needed, model recalibration or refitting will be performed, with internal validation based on k-fold cross-validation. Outcome probabilities will be calculated blinded to study outcomes.

This external validation will indicate whether the Pre-ECPR score retains its predictive performance in an independent OHCA population. The results will provide essential evidence for its potential to improve ECPR patient selection beyond conventional binary selection criteria.

Figures