Statement of Ltc. Robert R. Redfield

Drs. Vahey and McCarthy but the Acting Department Chief in my
absence was not asked to participate which meant that no RV21
Principal Investigator was present to discuss the RV21 data. I
have agreed with the statistical analysis performed by Dr. McCarthy
and have used this data in all subsequent presentations. Also
during this week, Dr. Birx was requested by Dr. Vahey to provide
the 15 patients presented at Amsterdam. Dr. Birz was unable to
locate the information and suggested they await my return to which
Dr. Vahey agreed.
I returned unexpectedly on 27 August to find that a meeting
was scheduled the' following day to go over the data presented in
Amsterdam. On 28 August, I reviewed in detail my analysis of the 15
patients pre ented at Amsterdam with COL Burke after which he
*.rfVie Irat he had a meeting that afternoon with Drs. Vahey,
Brundage and McCarthy. The purpose of the meeting was to assess
the disparity between analyses presented by asme and others. Again,
no Principal Investigator on RV21 was present at this meeting. Dr.
McCarthy presented a report (dated 28 August and addressed to Dr.
Vahey and not to- my earlier request) of his analysis of
15 patients but they were not the same 15 I presented at Amsterdam.
He concluded that while there was no statistical evidence of
significant effect of intervention, the findings did not exclude
the possibility that there were significant effects of therapy
(overall or in subgroups of treated patients) that were obscured in
his analysis by such factors as small sample size, variability of
RNA/DNA measurements, effects of uncontrolled confounding factors,
e.g., age, CD4+). This report was leaked to the press.
Two days before (29 August) the Chantilly meeting began I
received a MFR from COL Burke of his meeting on 28 August. It
stated four factors attributing to the disparity between the
analysis of 26 patients and what I reported in Amsterdam: a.) the
use of viral load divided by CD4 cells, since the natural history
group was less than perfectly matched and had more patients with
greater CD4 cell count decreases which artificially raised the
reported differences over time, was suboptimal and led to
misinterpretation; b.) for unexplained reasons, patients enrolled
and vaccinated earlier tended to have falling burdens, while those
vaccinated later tended to have increasing burdens; the first 7
vaccinees looked best, the first 15 (including the 7) looked next
best whereas data from the 26 showed no difference from the natural
history cohort over time; c.) the use of a half-log rise or fall
was arbitrary and post hoc grouping led to marginal but unreal
significance; and d.) the second table was organized so that
patients who showed no change in viral burden being grouped with
those whose viral burden decreased gave the impression to casual
observers that a large number had falling viral burdens.
I responded to this KYR verbally at several a tings in the
month of September and in writing 16 October wit a handwritten
apology for misunderstanding that a written response was requir L
In this response I stated: a.) the decision to compare preliminary
RV21 data to a natural history cohort was discussed at several
Department meetings, and the decision, while liable to questioning