From: firstname.lastname@example.org [mailto:email@example.com]
Sent: 08 July 2002 15:02
To: Richard Eastell
Subject: Re: HIP data analysis
I think we should look carefully at the pros and cons of Dr. Blumshon
conducting the analysis you refer to. One the plus side it does add
an extra layer of external "credibility". With this however, Industry
looses the opportunity to demonstrate its ability to be a true partner
in scientific endeavours. Beyond this, the practical issues to training
up a new statistician and the corresponding delay in "time to result" may
make the option difficult.
Finally, transferring databases which the Compnay has invested hundreds of
millions of dollars to obtain is not something to be taken lightly. That's
not to say it can't be done, but the reasons must be sufficiently important
to justify it.
Might there be some alternative approaches to the potential ciritcism? For
example, Ian this morning spoke of wanting to do a formal analysis plan
prior to any data in hand adn having this agreed by yourself and others as
needed. In hte manuscript reporting the reuslts it would be made clear a
formal analysis plan was developed and agreed by all investigators and the
statistician conducted the analyses according to plan.
I'd be happy to discuss this further if you wish. I'm in the Staines office
today and Tuesday morning, busy all day Wednesday, but then in the Geneva
office on Thursday from about mid-morning. IF there is a time you can talk,
let me know and I'l arrange the proper phone numbers be set up so we can talk.