Confidence interval of risk difference by different statistical methods and its impact on the study conclusion in antibiotic non-inferiority trials