Statistical analysis
Patients were divided into two groups based on the success of weaning off ECMO support (Weaned vs. Failed). Non-normally distributed continuous variables were compared using the Mann-Whitney U test, and categorical variables using the Chi-square and Fisher’s exact tests. Continuous data are presented as medians (interquartile ranges 25%-75%), and categorical data as numbers and fractions (%). Potential risk factors of the primary endpoint were identified by multivariate logistic regression analysis. One multivariate logistic regression analysis model was constructed using variables found to be associated with successful weaning by univariate logistic regression analysis, and the other model included variables affected to clinical outcomes in addition to these factors of univariate logistic regression (Table 3, figure 1). All p-values given are two-sided and p-values of < 0.05 were considered to indicate significance. SPSS version 22.0 (Korean version; IBM Corporation, USA) was used for the analysis.