Venovenous Extracorporeal Membrane Oxygenation Cannulation by Intensivists and Surgeons: A Single-Center Retrospective Noninferiority Analysis of Complications and Outcomes From 2018 to 2023.

Objectives: To evaluate the safety and efficacy of venovenous extracorporeal membrane oxygenation (ECMO) cannulation by nonsurgeon intensivists (critical care medicine intensivists [CCM]) compared with cardiothoracic surgeons (CTS).

Design: Retrospective, single-center observational study using a noninferiority framework to evaluate outcomes across cannulating physician training backgrounds. The primary outcome was the rate of cannulation-related complications. Secondary outcomes included in-hospital mortality and resource utilization. Noninferiority was assessed using a predefined margin corresponding to an odds ratio of 1.55 (15% higher complication rate for CCM vs. CTS).

Setting: Quaternary academic medical center with a multidisciplinary ECMO program serving the southeastern United States, including in-center and remote cannulations.

Patients: Adults with refractory respiratory failure who underwent venovenous ECMO cannulation and initiation.

Interventions: Cannulation and initiation of venovenous ECMO within an established program.

Measurements and main results: A total of 533 cannulation events in 231 patients from January 1, 2018, to December 31, 2023, were analyzed. Patient characteristics, pre-cannulation factors, predictive scores, hospital courses, cannulation-related complications, and in-hospital mortality were compared between CCM and CTS groups. At the time of cannulation, CCM-initiated cases had lower rates of vasoactive medication use and mechanical circulatory support and were more often performed in remote settings, reflecting differences in practice patterns. In a generalized linear mixed model adjusting for cannulation site, body mass index, and respiratory failure etiology, CCM was noninferior to CTS for cannulation-related complications, with an adjusted odds ratio of 0.84 (95% CI, 0.47-1.50); the upper confidence limit remained below the predefined noninferiority threshold of 1.55. Complication rates were 12% for CCM and 15% for CTS. In-hospital mortality was 29%, with no significant difference or evidence of noninferiority between groups.

Conclusions: Venovenous ECMO cannulation by nonsurgeon intensivists was noninferior to that by CTS with respect to complication rates. These findings support the safety of intensivist cannulation in multidisciplinary ECMO programs and highlight the feasibility of flexible models for ECMO delivery.