Statistical analysis

The SAS 9.3 statistical package was used to perform all analyses in this study. We examined differences in continuous variables between the two cohorts by using a Student’s t-test and we examined differences in dichotomous variables of the potential confounders between the two cohorts by using a Pearson χ² test.

The incidence rate is expressed per 100 000 person-years. The cumulative incidence of acute pancreatitis was assessed using the Kaplan-Meier estimator, with significance based on the log-rank test. The Cox proportional hazard regression model was used to analyse the risk of acute pancreatitis. Age, sex and baseline comorbidities were adjusted in multivariate analysis. Crude and adjusted HRs are presented along with 95% CIs. Each type of drug was separately analysed as a time-dependent effect in the Cox proportional hazard regression model. The HRs of each type of drug could be explained as follows: in any given month, if a patient used the given type of drug, the risk of acute pancreatitis would averagely increase (HR>1)/decrease (HR<1) compared with a patient who did not use the given type of drug. The results of all statistical tests were considered significant if the two-sided p value was ≤0.05.