Background: Assessment for readiness for decannulation from venoarterial extracorporeal membranous oxygenation (VA ECMO) is largely based on anecdotal data and provider experience with relatively high failure rates of 30-70% reported in literature. Compared to patients cannulated for cardiogenic shock, in-hospital cardiac arrest and post-cardiotomy, patients cannulated for out-of-hospital cardiac arrest (OHCA) extracorporeal cardiopulmonary resuscitation (ECPR) represent a unique population.

Aims: We sought to identify parameters impacting decannulation success among OHCA patients treated with ECMO-facilitated resuscitation.

Methods: Consecutive OHCA patients treated with previously described Minnesota Resuscitation Consortium ECPR protocol from 2015 to 2022 were eligible for inclusion. Hemodynamics, patient characteristics, mechanical circulatory support, and echocardiographic data were compared among patients with successful versus failed decannulations, defined as death due to cardiac causes or recannulation on VA ECMO within the same hospitalization.

Results: Among 390 ECPR protocol patients, 125 (32%) were deemed ready for decannulation with 88% successful (n=110) and 12% failed decannulations (n=15). Feature importance analysis was performed by leveraging multiple models, to identify strength of variables associated with successful decannulation. Multivariable logistic regression showed left ventricular ejection fraction (coefficient 0.14), pressors (coefficient -0.12), pulsatility (coefficient 0.08) and VA ECMO flow prior to decannulation played a role in success. The ROC-AUC for LVEF alone was 0.57, however, incorporation of additional variables improved the predictive power with an ROC-AUC of 0.857 (Figure 1).

Conclusion: Our findings accurately predict decannulation success and emphasize considering multiple factors. Prospective validation of a composite score combining these variables is warranted.