Long-term neurological outcomes after extracorporeal cardiopulmonary resuscitation for refractory cardiac arrest: a 14-year single-centre cohort study

Background: Extracorporeal Cardiopulmonary Resuscitation (ECPR) via Veno-Arterial Extracorporeal Membrane Oxygenation (ECMO) is a viable treatment for refractory cardiac arrest (r-CA). Data on long-term outcomes and predictors of favourable neurological prognosis remain limited, and definitive patient-selection recommendations are lacking. Methods: We conducted a retrospective observational study of adult patients treated with ECPR for in-hospital (IHCA) and out-of-hospital (OHCA) r-CA at an Italian ECMO centre between 2011 and 2024. The primary outcome was long-term neurological performance, measured by the Cerebral Performance Category (CPC) scale six months after hospital discharge. Multivariable and latent class analyses assessed independent predictors and explored distinct pre-ECPR phenotypes. Results: Among 295 consecutive patients (117 IHCA; 178 OHCA), 17.3% achieved CPC 1-2 at six months (28.2% IHCA vs 10.1% OHCA; p < 0.0001), and 4.4% survived with severe long-term neurological sequelae (CPC 3-4). Independent predictors of CPC 1-2 were younger age (OR 0.95 per year, 95% CI 0.92-0.98), an initial shockable rhythm (aOR 2.7; 95% CI 1.11-7.04), and shorter low-flow duration (OR 0.95 per minute increase, 95% CI 0.93-0.97). Stepwise selection based on these criteria progressively increased the proportion of favourable survivors but excluded a small proportion who might have recovered. Conclusions: These results emphasise the importance of establishing pre-treatment selection criteria to optimise ECPR use and enhance long-term neurological outcomes. Age, initial rhythm, and low-flow time are key determinants, and exploratory phenotype-based analyses suggest multidimensional patient characterisation may complement traditional selection criteria.