[R] gsDesign
Marc Schwartz
marc_schwartz at me.com
Tue Nov 15 15:19:50 CET 2011
Hi Dongli,
Sorry for the delay in following up.
You might want to read the dsDesignManual.pdf document, which is available in the 'inst/doc' folder in the package source tarball on CRAN, or in the package 'doc' directory in your R installation. Use:
system.file(package = "gsDesign")
to get the package top directory for your installation. The above file will be in the 'doc' sub-directory from there. It has more extensive worked examples than the default package manual.
Simple non-inferiority example from ?nBinomial, with 2:1 ratio:
n.Fix <- nBinomial(p1 = .677, p2 = .677, delta0 = 0.07, ratio = 2)
> n.Fix
[1] 2056.671
# Adjust that *up* to an integer multiple of 3
n.Fix <- 2058
# Change 'outtype' to 2 if you want to see per arm sample sizes
# eg:
> nBinomial(p1 = .677, p2 = .677, delta0 = 0.07, ratio = 2, outtype = 2)
$n1
[1] 685.5569
$n2
[1] 1371.114
# Simple default GS design using the fixed study design sample size from above,
# which is not yet adjusted for interim analyses
> gsDesign(n.fix = n.Fix)
Asymmetric two-sided group sequential design with
90 % power and 2.5 % Type I Error.
Upper bound spending computations assume
trial continues if lower bound is crossed.
----Lower bounds---- ----Upper bounds-----
Analysis N Z Nominal p Spend+ Z Nominal p Spend++
1 734 -0.24 0.4057 0.0148 3.01 0.0013 0.0013
2 1468 0.94 0.8267 0.0289 2.55 0.0054 0.0049
3 2202 2.00 0.9772 0.0563 2.00 0.0228 0.0188
Total 0.1000 0.0250
+ lower bound beta spending (under H1):
Hwang-Shih-DeCani spending function with gamma = -2
++ alpha spending:
Hwang-Shih-DeCani spending function with gamma = -4
Boundary crossing probabilities and expected sample size
assume any cross stops the trial
Upper boundary (power or Type I Error)
Analysis
Theta 1 2 3 Total E{N}
0.0000 0.0013 0.0049 0.0171 0.0233 1286.0
0.0715 0.1412 0.4403 0.3185 0.9000 1628.4
Lower boundary (futility or Type II Error)
Analysis
Theta 1 2 3 Total
0.0000 0.4057 0.4290 0.1420 0.9767
0.0715 0.0148 0.0289 0.0563 0.1000
So rather than needing 2058 from the fixed design, you actually need 2202 (1468 in one arm and 734 in the other).
I would urge you to read the manual I reference above and as Andy has noted in his reply, contact Keaven directly for further assistance with this package.
HTH,
Marc
On Nov 14, 2011, at 5:13 PM, Dongli Zhou wrote:
> Hi, Marc,
>
> Thank you very much for the reply. I'm using the gsDesign function to create an object of type gsDesign. But the inputs do not include the 'ratio' argument.
>
> Dongli
>
> On Nov 14, 2011, at 5:50 PM, Marc Schwartz <marc_schwartz at me.com> wrote:
>
>> On Nov 14, 2011, at 4:11 PM, Dongli Zhou wrote:
>>
>>> I'm trying to use gsDesign for a noninferiority trial with binary
>>> endpoint. Did anyone know how to specify the trial with different sample
>>> sizes for two treatment groups? Thanks in advance!
>>
>>
>> Hi,
>>
>> Presuming that you are using the nBinomial() function, see the 'ratio' argument, which defines the desired sample size ratio between the two groups.
>>
>> See ?nBinomial and the examples there, which does include one using the 'ratio' argument.
>>
>> HTH,
>>
>> Marc Schwartz
>>
More information about the R-help
mailing list