Comparing R&D and Commercial: Are All Data Created Equal?
Twenty-fifth Annual Midwest Biopharmaceutical Statistics Workshop
21 May 2002
Rafe M. J. Donahue, Ph.D.
Principal Consultant Data Mining
Research Triangle Park, North Carolina
In the early summer of 2001, I made a move from the safe and friendly confines of Medical Affairs
Statistics at GlaxoSmithKline (GSK) to the unknown world of Data Mining Technologies, a part of our
Commercial IT department. Since arriving at GSK from Hoechst Marion Roussel (HMR) and it predecessor,
Marion Merrell Dow (MMD), in 1996, I had been involved in the various statistical components of Phase IV
clinical trials. While this certainly was a change from the Phase I, II, III world I had experienced at HMR
and MMD since 1992, it wasn’t vitally different. All the classic statistical elements of clinical trials were
there: protocols, analysis plans, sample size calculations, primary analysis reports, secondary and exploratory
analyses, patient listings and summaries, project development plans, and the like.
There were, however, certain things about Phase IV that were different. The general feel was slightly
dissimilar since the focus of the development plan was rarely a pair of adequate and well-controlled trials each
with p-value less than 0.05. Thus, there was more latitude in study design. Principles of sound statistical
and scientific conduct were still essential but the foci of the development plans were varied. Prior to a
regulatory submission in Phases I, II, and III, all activity converges toward a single point: the submission.
After approval in Phase IV, though, the program is more free to expand to other areas and is no longer
focused solely on “the submission.”
But the move from pre-submission work to post-submission work was nothing compared to the change
from clinical trials to the world of Commercial.
In the year that I have spent exploring this brave new world of comm