[Minutes, Advisory Committee to the Director GP-160 Meeting (1992: Bethesda, Maryland)]

140

1 percent reduction within strata.
2 As I said, the actual strata would be
3 divided up as 10 percent of these would be in the
4 lower 200. Sixty percent would be in the above 500.
5 And 30 percent would be in the middle.
6 It would be designed so that you would be
7 able to come to conclusions on all of this strata at
8 approximately the same time. And if you aggregated,
9 then, over all the strata, you could actually detect
10 overall a reduction in event rate of 25 percent.
11 Next slide. One thing that's important to
12 realize, it's obvious that the more patients we enter,
13 the sooner the required number of events will be
14 observed, but those patients experiencing very early
15 events may not be representative.
16 So I think it would be a mistake to be too
17 worried about getting an answer too quickly and try
18 and decide what's happening based on a very, very
19 small proportion of events.
20 I think many of us have seen in trials in
21 AIDS and in other diseases that survival curves,
22 event-free survival curves, sometimes don't even start
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, NW.
[phone] WASHINGTON, D.C. 20005 [phone]