Date:         Wed, 11 Feb 2004 20:07:26 GMT
Reply-To:     John Iwaniszek <not@HOTMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         John Iwaniszek <not@HOTMAIL.COM>
Subject:      Re: Checking CRO's work

Jeff Morison wrote:

>   Thanks a bunch to all of you who responded to my
> earlier post.
>
> How to validate the SAS output provided to you by
> CRO?.
>
> Most of our SAS work  summary tables, data listings,
> and figures are generated by outside vendor,  they
> follow their own SOPs.  It is our job now to validate
> them, I am looking for some input from experienced
> programmers who have done such work.
>
> Thanks for any insight,
> Jeff.
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Finance: Get your refund fast by filing online.
> http://taxes.yahoo.com/filing.html
>

The best approach is parallel, quasi-independent programing back to the raw
data.  This combines the efficiency of SAS processing with the accuracy of
re-calculating the results.

I say quasi-independent, because in my experience there is always a need
for the QC programmer to communicate to the original
programmer/statistician regarding algos and exceptions.

At minimum, this effort requires a documentary coversheet to explain the
method used for QC, and instructions to a reviewer where to find the output
to map to the tables, and to collect things that crop up during review.
The coversheet, marked-up table, and qc output should be stored together
under another covesheet that lists programs, outputs, and other relevent
info.