raf - IMHO, I think it is not that it is speeding up, but that events are imminent in the population for this study... so the longer the study goes on, the closer to an event occurring within each patient gets shorter. Yes, not all patients started the study at the same time. But the way I interpret that you were looking at the data implies a new start date each time. For example, if you break it down into 4 quarters, would the events occur 20% in Q1, 20% in Q2, 20% in Q3, and 20% in Q4? I seriously doubt it. I think it would be more like 10% in Q1, 20% in Q2, 30% in Q3 and 40% in Q4. Consider this to be Q5 of the study now... so a greater number of events are occurring the further you move into the study. Hence, faster and faster timing to reach X number of events than at the beginning of the study.