Extracorporeal cardiopulmonary resuscitation (ECPR) is increasingly used for refractory cardiac arrest across the world, but its cost-effectiveness remains uncertain. We conducted this systematic review to assess the available date on the cost-effectiveness of ECPR compared with the standard care in patients with cardiac arrest.

We searched MEDLINE ALL (Ovid), EMBASE (Ovid), the Tufts Cost-Effectiveness Analysis (CEA) registry, and the Web of Science databases from database inception to 3^rd February 2025 for all economic evaluations that reported the cost-effectiveness of ECPR for refractory cardiac arrest in adults. We undertook backward and forward citation tracking to identify any additional relevant articles. We used the Evers checklist, Philips checklist, and the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist to assess the quality of the included study and the completeness of reporting. All reported costs were translated into 2025 US dollars (US$) to aid comparability.

We included 15 studies. Of these, 12 (80%) reported ahigh probabilityof ECPR being cost-effective, 1 (7%) reported alow probability, 1 (7%) suggested ECPR waspotentially cost-effective, and 1 (7%) concludedno probabilityof cost-effectiveness. Notably, the latter study specifically evaluated ECPR in patients with acute aortic dissection. The adjusted incremental cost-effectiveness ratios (ICERs) ranged from $2,564 to $276,307. When using these ICERs, ECPR remained cost-effective in 10 studies using the US willingness to pay threshold ($150,000), and in 7 studies using alternative thresholds ($40,000).

ECPR may be cost-effective in some settings, but there is marked variability in the reported cost-effectiveness of ECPR across current studies. This reflects differences in methodology, survival assumptions, and willingness to pay thresholds. Economic evaluation for appropriate implementation of ECPR across various healthcare settings remains heterogenous, highlighting the imminent need for standardised reporting in this domain.