Statistical analysis

Categorical variables were described as number (percentage); while continuous variables were reported as mean [standard deviation (SD)]. Baseline characteristics of the early and late ERCP groups were compared using the chi-square test for categorical variables and the student t test for continuous variables. Logistic regression was used to compare the odds of the three outcomes (in-hospital mortality, 30-d mortality, and 30-d readmissions) between the two groups. Covariates that were considered for inclusion into the model were age, sex, mortality index, and presence of severe cholangitis. The readmission index was used for 30-d readmission outcome. Covariates with P < 0.2 in univariate analysis were introduced into the model and retained if the P value was < 0.05. To assess the effect of severe cholangitis on the association of ERCP timing and outcomes, we performed an additional analysis in which we added an interaction term (timing*severity) to stratify this association based on cholangitis severity. Results of logistic regression were expressed using adjusted odds ratio (aOR) and 95% confidence interval (CI). We used a multivariable generalized linear regression model to compare the length of stay and total hospitalization costs between patients in the early and late ERCP groups. A 2-tailed P of 0.05 was used as the threshold for statistical significance.