X hits on this document

495 views

0 shares

0 downloads

0 comments

153 / 170

9

Vesnarinone, Losartan, Amlodipine, and a lot from my area in heart failure; that initial

10

trials went one direction with small events. Then when the real trial was done with larger

11

events, it went dramatically in the other direction. So I have this just sense of unease

12

anytime you deal with just such small numbers, granted, you can do the same, you can do

13

multiple statistical tests on the same numbers and that helps, but it doesn’t get rid of the

14

fact they are just really small numbers. So, one of the things that I propose, is later on, is

15

to do kind of a simple sensitivity analysis saying, so how many more patients would have

16

had to shown up in the Saxagliptin group to move the point estimate or the confidence

17

interval into the 1.3, 1.8 test.

1

series, actually the DCCT trial was an example that was almost stopped because of

2

harm, because of low event rates, but then because they didn’t believe it even though it

3

was a statistically significant test with confidence intervals that were positive, were

4

actually worse. So the point estimate was on the wrong side, the confidence intervals at

5

95%, confidence intervals were on the wrong side. They decided to keep going, because

6

I said, you know, we don’t really have enough events to evaluate what’s going on here,

7

and they decided to continue the trial and low and behold they found a big benefit.

19

moved to have changed, and my guess is that number would be not big, I don’t know

20

how big, but probably not big, and that may give a better sense of - that’s a true

21

sensitivity analysis. Saying okay, if patients kind of moved one direction or the other,

22

how confident are we and how sensitive are these numbers to the fact that we only have

23

40 events. I am not a statistician. I don’t know the ways to do that, but that would have

24

been helpful to me.

152

8

We have a whole - after seventeen years. We have lots of examples from

18

How many patients out of that 4000 plus trials would have had to have

Scribes, LLC Toll Free 1-800-675-8846 www.scribesllc.com

Document info
Document views495
Page views495
Page last viewedThu Dec 08 20:12:30 UTC 2016
Pages170
Paragraphs7975
Words54271

Comments