because the s/c trial is 2:1, there are 2x as many
Post# of 154852
(Total Views: 863)
Posted On: 07/22/2020 11:45:26 PM
Re: ClosetInvestor #44648
because the s/c trial is 2:1, there are 2x as many leronlimab patients as placebo patients, so an equal number of deaths in each arm means that leronlimab has reduced mortality by 50% compared to the placebo, and an equal number of deaths is a good sign that leronlimab is helping significantly.
They will test whether the log(odds) of dying are less with leronlimab than they are with placebo, or whether log(die/live for placebo) - log(die/live for leronlimab) is significantly greater than zero. I whipped up a little table of how many deaths could be observed in the leronlimab arm and still have p-value less than 0.05 for a each given number of deaths in the placebo arm (assuming a total trial size of 150).
P L
3 0
4 1
5 3
6 4
7 5
8 7
9 8
10 10
11 11
12 13
13 14
14 16
15 18
16 19
17 21
18 23
19 24
20 26
21 28
22 30
23 32
24 34
25 35
26 37
27 39
28 41
29 43
30 45
This is the Mantel-Hanszel test , which is very common in epidemiology and clinical trials (e.g., [url=DOI: 10.1056/NEJMoa2015301]Goldman et al. 2020[/url].
They will test whether the log(odds) of dying are less with leronlimab than they are with placebo, or whether log(die/live for placebo) - log(die/live for leronlimab) is significantly greater than zero. I whipped up a little table of how many deaths could be observed in the leronlimab arm and still have p-value less than 0.05 for a each given number of deaths in the placebo arm (assuming a total trial size of 150).
P L
3 0
4 1
5 3
6 4
7 5
8 7
9 8
10 10
11 11
12 13
13 14
14 16
15 18
16 19
17 21
18 23
19 24
20 26
21 28
22 30
23 32
24 34
25 35
26 37
27 39
28 41
29 43
30 45
This is the Mantel-Hanszel test , which is very common in epidemiology and clinical trials (e.g., [url=DOI: 10.1056/NEJMoa2015301]Goldman et al. 2020[/url].
(3)
(0)