AimTo evaluate whether a phased multidimensional intervention bundle would decrease the mortality of patients with extracorporeal membrane oxygenation (ECMO) and the complication incidence.Materials and methodsWe conducted a prospective observational study in comparison with a retrospective control group in six intensive care units (ICUs) in China. Patients older than 18 years supported with ECMO between March 2018 to March 2022 were included in the study. A phased intervention bundle to improve the outcome of patients with ECMO was developed and implemented. Multivariable logistic regression modeling was used to compare the mortality of patients with ECMO and the complication incidence before, during, and up to 18 months after implementation of the intervention bundle.ResultsThe cohort included 297 patients in 6 ICUs, mostly VA ECMO (68.7%) with a median (25th–75th percentile) duration in ECMO of 9.0 (4.0–15.0) days. The mean (SD) APECHII score was 24.1 (7.5). Overall, the mortality of ECMO decreased from 57.1% at baseline to 21.8% at 13–18 months after implementation of the study intervention (P < 0.001). In multivariable analysis, even after excluding the confounding factors, such as age, APECHII score, pre-ECMO lactate, and incidence of CRRT during ECMO, the intervention bundle still can decrease the mortality independently, which also remained true in the statistical analysis of V-V and V-A ECMO separately. Among all the ECMO-related complications, the incidence of bloodstream infection and bleeding decreased significantly at 13–18 months after implementation compared with the baseline. The CUSUM analysis revealed a typical learning curve with a point of inflection during the implementation of the bundle.ConclusionA phased multidimensional intervention bundle resulted in a large and sustained reduction in the mortality of ECMO that was maintained throughout the 18-month study period.Clinical trial registration[ClinicalTrials.gov], identifier [NCT05024786].