Background: Extracorporeal membrane oxygenation (ECMO) for acute respiratory failure is still a matter of debate.
Methods: We performed a structured search on Pubmed, EMBASE, Lilacs, and the Cochrane Library for randomized controlled trials and observational case-control studies with severity-paired patients, evaluating the use of ECMO on severe acute respiratory failure in adult patients. A random-effect model using DerSimonian and Laird method for variance estimator was performed to evaluate the effect of ECMO use on hospital mortality. Heterogeneity between studies was assessed with Cochran's Q statistic and Higgin's I(2).
Results: Three studies were included on the metanalysis, comprising 353 patients in the main analysis, in which 179 patients were ECMO supported. One study was a randomized controlled trial and two were observational studies with a propensity score matching. The most common reason for acute respiratory failure was influenza H1N1 pneumonia (45%) and pneumonia (33%). ECMO was not associated with a reduction in hospital mortality (OR = 0.71; CI 95% = 0.34 - 1.47; P = 0.358). If alternative severity-pairing method presented by the two observational studies was included, a total of 478 cases were included, in which 228 received ECMO support. In the former analysis, ECMO had a benefit on hospital mortality (OR = 0.52; CI 95% = 0.35 - 0.76; P < 0.001).
Conclusion: Extracorporeal membrane oxygenation benefit on hospital mortality is unclear. Results were sensitive to statistical analysis, and no definitive conclusion can be drawn from the available data. More studies are needed before the widespread use of ECMO can be recommended.
Keywords: Acute respiratory failure; Extracorporeal membrane oxygenation; Meta-analysis.