Abstract
BACKGROUND: Venovenous extracorporeal membrane oxygenation (VV ECMO) is frequently associated with deep sedation and neuromuscular blockades, that may lead to diaphragm dysfunction. However, the prevalence, risk factors, and evolution of diaphragm dysfunction in patients with VV ECMO are unknown. We hypothesized that the prevalence of diaphragm dysfunction is high and that diaphragm activity influences diaphragm function changes. METHODS: Patients with acute respiratory distress syndrome (ARDS) requiring VV ECMO were included in two centers. Diaphragm function was serially assessed by measuring the tracheal pressure in response to phrenic nerve stimulation (Ptr,stim) from ECMO initiation (Day 1) until ECMO weaning. Diaphragm activity was estimated from the percentage of spontaneous breathing ventilation and by measuring the diaphragm thickening fraction (TFdi) with ultrasound. RESULTS: Sixty-three patients were included after a median of 4 days (3-6) of invasive mechanical ventilation. Diaphragm dysfunction, defined by Ptr, stim ≤ 11 cmH(2)O, was present in 39 patients (62%) on Day 1 of ECMO. Diaphragm function did not change over the study period and was not influenced by the percentage of spontaneous breathing ventilation or the TFdi during the 1 week. Among the 63 patients enrolled in the study, 24 (38%) were still alive at the end of the study period (60 days). CONCLUSIONS: Sixty-two percent of patients undergoing ECMO for ARDS related to SARS CoV-2 infection had a diaphragm dysfunction on Day 1 of ECMO initiation. Diaphragm function remains stable over time and was not associated with the percentage of time with spontaneous breathing. CLINICALTRIALS: gov Identifier NCT04613752 (date of registration February 15, 2021).