Date:         Tue, 19 Aug 2008 10:48:01 -0700
Reply-To:     vontressms@CS.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         vontressms@CS.COM
Organization: http://groups.google.com
Subject:      Re: Validating SAS for FDA?
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

On Aug 19, 9:41 am, Lex Jansen <lexjan...@gmail.com> wrote:
> You might want to look at this draft guidance as well:http://www.fda.gov/cdrh/osb/guidance/1601.html
>
> Lex Jansen
>
> On Aug 19, 9:46 am, John Uebersax <jsueber...@gmail.com> wrote:
>
>
>
> > Mark,
>
> > > FDA does accept R though and they are beginning to accept WinBugs.
>
> > I just caught this part.  Actually, that's what my post was leading up
> > to.  The real issue I'm faced with is convincing project managers to
> > allow WinBUGS.
>
> > An alternative has been proposed -- to use a SAS macro for MCMC
> > estimation.  This makes no sense to me, since it would seem at least
> > as difficult to validate this SAS macro as to validate a WinBUGS
> > program.
>
> > I don't really understand how the FDA could not accept WinBUGS
> > results,  I could understand them rejecting a Bayesian analysis, but
> > not the WinBUGS program itself.
>
> > For what it's worth, my provisional plan is to demonstrate that
> > WinBUGS accurately replicates textbook examples, to document this, and
> > to place the document on file.
>
> > John Uebersax- Hide quoted text -
>
> - Show quoted text -

This guidance is the one that opens the door to the use of Winbugs at
CDRH. I don't know about CDER, but I suspect they are accepting it for
Bayesian Adaptive Designs for proof of concept studies. I don't know
about confirmatory studies. I doubt CDER permits Bayesian analysis for
confirmatory studies.

The device guidance still needs a lot of work if you intend to apply
it. There is a lot of up front agreement with FDA that you need to get
before you start a study using it. You also have to submit your
scripts.

Mark