Objective: 

Analysis of the prevalence and risk factors for weaning failure from venovenous extracorporeal membrane oxygenation (VV-ECMO) in patients with severe acute respiratory insufficiency.

Design: 

Single-center retrospective observational study.

Setting: 

Sixteen beds medical ICU at the University Hospital Regensburg.

Patients: 

Two hundred twenty-seven patients with severe acute respiratory insufficiency requiring VV-ECMO support between October 2011 and December 2017.

Interventions: 

None.

Measurements and Main Results: 

Patients meeting our ECMO weaning criteria (Spo₂ ≥ 90% with Fio₂ ≤ 0.4 or Pao₂/Fio₂ > 150 mm Hg, pH = 7.35–7.45, positive end-expiratory pressure ≤ 10 cm H₂O, driving pressure < 15 cm H₂O, respiratory rate < 30/min, tidal volume > 5 mL/kg, ECMO bloodflow ≈ 1. 5 L/min, sweep gas flow ≈ 1 L/min, heart rate < 120/min, systolic blood pressure 90–160 mm Hg, norepinephrine < 0.2 µg/[kg*min]) underwent an ECMO weaning trial (EWT) with pausing sweep gas flow. Arterial blood gas analysis, respiratory and ventilator parameters were recorded prior, during, and after EWTs. Baseline data, including demographics, vitals, respiratory, ventilator, and laboratory parameters were recorded at the time of cannulation. One hundred seventy-nine of 227 (79%) patients were successfully decannulated. Ten patients (4%) underwent prolonged weaning of at least three failed EWTs before successful decannulation. The respiratory rate (19/min vs 16/min, p = 0.002) and Paco₂ (44 mm Hg vs 40 mm Hg, p = 0.003) were higher before failed than successful EWTs. Both parameters were risk factors for ECMO weaning failure (Paco₂: odds ratio [OR] 1.05; 95% CI, 1.001–1.10; p = 0.045; respiratory rate: OR 1.10; 95% CI, 1.04–1.15; p < 0.001) in multivariable analysis. The rapid shallow breathing index [42 (1/L*min), vs 35 (1/L*min), p = 0.052) was higher before failed than successful EWTs. The decline of Sao₂ and Pao₂/Fio₂ during EWTs was higher in failed than successful trials.

Conclusions: 

Seventy-nine percent of patients were successfully decannulated with only 4% needing prolonged ECMO weaning. Before EWT only parameters of impaired ventilation (insufficient decarboxylation, higher respiratory rate) but not of oxygenation were predictive for weaning failure, whereas during EWT-impaired oxygenation was associated with weaning failure.