Fig. 2. Sensitivity analysis.

Last observation carried forward (LOCF) method was used to impute missing values if at least one post-baseline value is available. Dropout patients with only the baseline value are considered as ‘failures’. Worst case: dropout patients independently of the time of dropout are considered as ‘failures’. Stratified: within each country, centres with <10 patients randomized were pooled for the stratified analysis. n = number of patients without signs of anterior chamber inflammation; m = total number of patients valid in the analysis set. Difference between treatment groups is calculated as (levofloxacin + dexamethasone) vs. (tobramycin + dexamethasone). FAS full analysis set, PP per protocol.