MIRO study: Do the results of a randomized controlled trial apply in a real population?

Abstract

BackgroundThe aim of our study was to evaluate the external validity of the MIRO randomized controlled trial findings in a similar nationwide setting “real life” population, especially the benefit of a hybrid approach in esophageal resection for pulmonary complication. The external validity of randomized controlled trial findings to the general population with the same condition remains problematic because of the inherent selection bias and rigid inclusion criteria. MethodsThis study was a cohort study from a National Health Database (Programme de Medicalisation des Systemes d’Informations) between 2010 and 2022. All adult patients operated on using Ivor Lewis resection for esophageal cancer were included. We first validated the detection algorithm of postoperative complications in the health database. Then, we assessed the primary outcome, which was the comparison of postoperative severe pulmonary complications, leak rate, and 30-day mortality between the 2 surgical approaches (hybrid versus open) over a decade. ResultsBetween 2010 and 2012, 162 of 205 patients in the MIRO trial were anonymously identified in the health care database. No difference between randomized controlled trials and healthcare database measurements was found within severe respiratory complications (24% vs 22%, respectively) nor within leak rate (10% vs 9%, respectively). After application of selection criteria according to the MIRO trial, 3,852 patients were included between 2013 and 2022. The hybrid approach was a protective factor against respiratory complications after adjustment for confounding variables (odds ratio = 0.83; 95% confidence interval = 0.71–0.98, P = .025). No significant difference in the 30-day mortality rate or 30-day leakage rate between the types of approach was reported. ConclusionThis national cohort study demonstrates the external validity of the MIRO randomized controlled trial findings in a real-life population within France.