The primary efficacy analysis assessed the noninferiority of the mortality rate of intravenous colistin plus levofloxacin compared to intravenous meropenem plus levofloxacin in the MITTs population.
Noninferiority of colistin is concluded if the lower limit of the two-sided 95% confidence interval for the difference in mortality rates (colistin minus meropenem) is ≤10%.
In order to achieve a power of 80% to reject the null hypothesis with a significance level of 5%, assuming as stated in previous studies that the survival rate is 80%with standard treatment [11], and the same with experimental treatment, with a non-inferiority limit of ≤10%, 198 patients are needed in each group. Taking into account that microbiological documentation of VAP is obtained in approximately 80% of the episodes, and therefore about 20% of randomized patients will not have a microbiological diagnosis and will not be useful for evaluation of effectiveness, this requires a 25% increase in sample size. This increases the sample size of intention-to-treat patients included to a total number of 496.