RESUMO
OBJECTIVE: The criteria and process for liberation from extracorporeal membrane oxygenation in patients with severe acute respiratory distress syndrome are not standardized. The predictive accuracy of the oxygen challenge test as a diagnostic test in determining weaning and decannulation from venovenous extracorporeal membrane oxygenation was tested. DESIGN: A single-centre, retrospective, observational cohort study. SETTING: Tertiary referral severe respiratory failure centre in a university hospital in the United Kingdom. PATIENTS: 253 adults with severe acute respiratory distress syndrome requiring extracorporeal membrane oxygenation. INTERVENTION: None. MEASUREMENTS AND MAIN RESULTS: Patients had median age: 43 years (interquartile range: 32-52) years, extracorporeal membrane oxygenation days: 9 (interquartile range: 6-14) and acute physiology and chronic health evaluation II score 17.5 (interquartile range: 15-20). Oxygen challenge test value (PaO2-OCT) with best prediction was 31 kPa (232 mmHg; sensitivity 0.74; specificity 0.70; area under curve 0.77 (confidence interval: 0.73-0.81)). PaO2-OCT did not perform well as a prospective test to identify readiness to decannulation. Only 24 patients (10%) were decannulated 48 hours after their first positive oxygen challenge test (true positive) and 73.4% patients were false positives (positive oxygen challenge test but not decannulated). True positives had higher tidal volume (541 ± 218 vs 368 mL ± 210; p < 0.05) and minute ventilation (9.34 ± 5.36 vs 6.33 L/min ± 4.43; p < 0.05). Blood flow (3.17 ± 0.23 vs 3.53 L/min ± 0.56; p < 0.05), sweep gas flow (1.42 ±1.83 vs 3.74 L/min ± 2.43; p < 0.05) and extracorporeal membrane oxygenation minute volume at time of first positive oxygen challenge test was lower in true positives (1.66 ± 2.26 vs 4.82 ± 3.43 L/min). This was a strong predictor for decannulation within 48 hours (area under curve: 0.88, confidence interval: 0.88-0.89). CONCLUSIONS: In severe acute respiratory distress syndrome requiring venovenous extracorporeal membrane oxygenation, the PaO2-OCT is a poor predictor of readiness to decannulate from extracorporeal membrane oxygenation. Additional factors involved in the control of respiratory drive and carbon dioxide clearance, particularly native lung dead space and total minute ventilation, should be assessed.
