Bleeding is a common and sometimes fatal complication of venovenous extracorporeal membrane oxygenation (ECMO). Whether lowering the intensity of anticoagulation during venovenous ECMO is safe or effective is unknown.

Is a large, multicenter randomized trial of low-intensity versus moderate-intensity anticoagulation during venovenous ECMO feasible?

In a multicenter, parallel-group, randomized, pilot trial conducted at 3 centers across the United States, we randomly assigned critically ill adults undergoing venovenous ECMO to low-intensity or moderate-intensity anticoagulation. Feasibility was assessed by enrollment rate and adherence to the assigned anticoagulation strategy. The primary efficacy outcome was major bleeding, and the primary safety outcome was thromboembolic events, both assessed between enrollment and 24 hours after decannulation.

All of the 26 patients enrolled received the assigned intensity of anticoagulation. A major bleeding event occurred in one of 12 patients (8.3%) in the low-intensity anticoagulation group and in 4 of 14 patients (28.6%) in the moderate-intensity anticoagulation group (absolute risk difference, -20.2 percentage points; 95% Confidence Interval, -48.6 to 8.1; p=0.33). One patient experienced a thromboembolic event (8.3%) in the low-intensity anticoagulation group compared to none in the moderate-intensity group (difference 8.3 percentage points, 95% Confidence Interval, -7.3 to 24.0; p=0.46). No patients died before discharge in the low-intensity anticoagulation group, compared to 2 patients (14.3%) in the moderate-intensity group, both of whom experienced major bleeding events. No patients died before discharge in the low-intensity anticoagulation group, compared to 2 patients (14.3%) in the moderate-intensity group, both of whom experienced major bleeding events.

Enrollment and separation between groups are feasible in a multicenter randomized trial of low-intensity versus moderate-intensity anticoagulation for critically ill adults receiving venovenous ECMO. A large, multicenter randomized trial is needed and appears feasible.