Statement of John F. Brundage, M.D.

" rs,,

slides very rapidly including the tables and p values. I had no
objection to the slides. As a matter of fact, I felt that he had
understood from my tutoring several days previously.
A meeting on 31 July 92 brought forth the issue of different
analytic results on the data presented at Amsterdam. COL Burke
called after the meeting and requested that I analyze the data from
the 26 patients in the gpl60 Phase I trial. Dr. McCarthy analyzed
the same data which resulted in his memos of 20 and 28 August 92.
At the time we did our analysis I had not seen the analyses of Dr.
McCarthy because we felt the analyses should be independent.
Both Dr. McCarthy's and my analyses were reviewed at a meeting
28 August 92 in COL Burke's office to determine what and how data
should be presented at the Chantilly meeting. There was not
disagreement regarding results or conclusions of analyses among COL
Burke, LTC Redfield, Drs. McCarthy and Vahey and myse
While there was consensus regarding our data analyses
My interactions with LTC Redfield go back to the article in
the New England Journal of Medicine (NEJM). Tom Wierzba of the
HMJF was providing data support for the study at that time. We
were concerned that differences of CD4+ decline for responders
versus non-responders were simply manifestations of better inherent
prognosis (i.e., that responders are inherently better off than
non-responders). If so, the course of the two subgroups combined
would be expected to be no better (or possibly worse) than expected
based on our natural history experience. To examine this
possibility, we utilized medical surveillance data that describes
patients evaluated at Army Medical Centers. I worked with a
programmer/analyst/statistician to select 10 patients who were
matched by age, ethnic group and baseline CD4+ count for each
vaccine treatment patient. LTC Redfield had no hands-on role in
the methods or conduct of this analysis. The results of the
analysis were discussed in the NEJM paper. I wrote the portion of
the Methods section relevant to this analysis and it accurately
describes the analytical method that was planned and conducted
without knowing the results.
In March 92, LTC Redfield asked me if the 28 patients were
doing better or worse based on comparison with the natural history
group. A comparison to our natural history experience showed that,
as a group, the patients in the trial were in the 20-30th
percentile range of the distribution expected for groups of this

2