Message-ID: <Pine.A41.4.44.0306061126110.82226-100000@homer33.u.washington.edu>
Date: 2003-06-06T18:46:57Z
From: Thomas Lumley
Subject: sas vs. r
In-Reply-To: <Sea2-DAV55leUBUhr3u000037b1@hotmail.com>
On Fri, 6 Jun 2003, wensui liu wrote:
> I like R more than SAS. My job is doing research on clinical trial. But
> I was told that FDA only accepts the result from SAS. Is that true? TOO
> BAD.
>
No, it isn't true. The FDA does not approve or certify statistical
software. Its main regulation relevant to statistical analysis is in 21
CFR 11, and is largely about audit trails. In fact the FDA is very
reluctant to issue blanket approval or disapproval for anything -- it
likes to keep its options open.
On the other hand, if a company has been successful in getting FDA
approval with analyses done one way, there is a substantial incentive not
to change. Why take the risk that it might slow down the approval of your
application?
-thomas