The ventilator is generally set to moderate levels of mechanical support (e.g. tidal volume 6 mL/kg, plateau pressure <30 cmH₂O, PEEP 5–12 cmH₂O, FiO₂<0.6) [7]. Similarly, in the CESAR trial, a peak pressure <30 cmH₂O and FiO₂ <60% were required [2]. In the EOLIA (ECMO to rescue lung injury in severe ARDS) trial, patients were ventilated in BIPAP/APRV mode under ECMO with a PEEP >12 cmH₂O and upper pressure limited to 23–24 cmH₂O [8]. Weaning trial was indicated when the tidal volume generated by the driving pressure was >200 mL. The device may be withdrawn if (1) PaO₂ was >60 mmHg with SaO₂ >90% with FiO₂ on the ventilator <60% and inspiratory plateau pressure < 30 cmH₂O and (2) echocardiography reveals no signs of acute cor pulmonale for at least 1–12 h.

During the trial, ECMO flows of <2 L/min should be avoided to reduce the risk of thrombus forming in the circuit. However, maintaining normal circuit flow with the sweep gas turned off prevents thrombus forming in the circuit but allows the patient to be tested off extracorporeal support. In most instances, after stopping systemic anticoagulation for 1 h, the ECMO cannulas can be removed without surgical repair of the vessel, but simply by a topical pressure pulling for 30 min. As the cannula is removed, the patient should be on Trendelenburg position and have received a short-term pharmacological paralysis or perform a Valsava maneuver to reduce the risk of air embolism. Routine venous Doppler ultrasound following decannulation is warranted to detect deep vein thrombosis in the cannulated vessel. Its prevalence following ECMO was estimated to 8.1/1000 cannula days [9].