Fig. 1.
Time to treatment failure. (A) Cumulative probability of initial treatment failure. (B) Cumulative probability of second treatment failure. The Kaplan-Meier (KM) estimates were performed for patients who had received at least one randomized medication and one post-randomization efficacy parameter assessed. Thus, not all patients included in the KM analysis had data at month 0. Hazard ratios are based on Cox regression analysis. CI, confidence interval; NE, not evaluable.