Intended for healthcare professionals


What have we learnt from Vioxx?

BMJ 2007; 334 doi: (Published 18 January 2007) Cite this as: BMJ 2007;334:120

Re: On Selective Learning

Merck’s lawyer writes that he “will not offer [his] views,” as he
offers his views in his critique of our article. Unfortunately his
letter, which he claims is a “correction” of our article, misrepresents
Merck’s conduct and the results of Merck’s drug trials.

Merck included some data from the prostacyclin study in its
regulatory submission but excluded the most important part – the non-Merck
authors’ conclusion that the “inhibition of urinary PGI-M by Vioxx implies
a major role for Cox-2 in the vascular biosynthesis of prostacyclin in

Furthermore, the primary FDA reviewer responsible for reviewing the
data concluded in 1999 that Merck’s pre-market Vioxx trials did contain
important signals of cardiovascular risk when he wrote:

1. Vioxx caused vascular-renal adverse events, i.e., hypertension-
and edema-type, serum creatinine increased, hyperkalemia, and to a lesser
degree proteinuria in a dose-dependent manner.
2. The vascular-renal safety profile of Vioxx is clearly distinguishable
from placebo, and is qualitatively similar to other NSAIDS.
3. The aforementioned adverse events occur at a higher rate with VIOXX at
50 mg, than with other NSAIDS at their recommended dosage. However,
whether that will translate into clinically significant differences in
vascular-renal morbidity cannot be determined from the current
osteoarthritis clinical database.

Another FDA review noted that “The data seem to suggest that in 6 -
week studies, thromboembolic events are more frequent in patients
receiving rofecoxib than placebo but do not show a clear dose response
relationship with rofecoxib.”

Both reviewers observed that the studies were not large enough to
find an effect since they were short term, excluded high risk patients and
failed to include tests designed to detect CV events. For example, Merck
failed to perform exit EKGs. These studies were also underpowered. Power
is a function of the number of expected events in a trial, not the total
number of patients studied. As the manufacturer, Merck had the
responsibility to follow up and warn about this early signal of harm.

We did not suggest that Merck failed to adjudicate CV events in
Vigor. Merck, however, did manipulate the adjudication process. During
VIGOR, but after Merck personnel were unblinded to overall results, Merck
deleted 21 cardiovascular diagnoses from the Standard Operating Procedure
list of events eligible for adjudication (SOP for all Vioxx trials).
Deletions included CHF, pulmonary edema and a variety of new EKG changes.
These changes were made on December 29, 1999 one week after the external
safety board told Merck to perform an analysis of CV events for this
particular trial. At that time Merck had no plans to analyze any of the
CV data from VIGOR or any other individual Cox-2 trial. Merck pre-dated
the change to October 3, 1999.

In regards to the Vigor DSMB’s chairman’s ownership of Merck stock,
the article simply pointed out the potential for a meaningful conflict of
interest and the fact that such stock ownership violates Merck’s own
guidelines, which we cited in the article (print citation #10). Those
guidelines say, “DSMB members should not have such a substantial financial
investment in Merck or in any of our competitors that their independence
would be questioned. All members should issue a periodic report describing
potential conflicts of interest. The DSMB chairman should review these
potential conflicts prior to each meeting, and forward them to the DSMB
chairman along with the status report.” The chairman also signed a
contract with Merck on March 6, 2000 (11 days prior to end of the Vigor
trial) to consult on the marketing and research and development of COX-2
agents including Vioxx:

The consultant will provide confidential advice, counsel, and
assistance in the development, planning and execution of commercial and
scientific initiatives related to Merck’s COX-2 Inhibitor program (the
“Consulting Service”). The Consulting Services may cover topics including,
but not limited to, market development, strategic product planning, and
educational activities as well as the design and the conduct of the
clinical and preclinical development program. (Ref. 12 in our article)

Readers should be able to decide whether the stock ownership and
consulting contract constitute a conflict of interest, and Merck should
allow readers that opportunity, rather than objecting to the disclosure of
this information.

Merck’s lawyer asserts that we omitted information from Dr. Curfman’s
deposition conducted after the submission of our article and that this
information supports Mayer’s assertion that Dr. Curfman, “…knew of the
post-cutoff data from its publication in August 2001, in the very JAMA
article the authors cite.” Mr. Mayer conflates access to the “full data”
with the issue of whether or not the data should have been included in the
NEJM VIGOR paper. The latter question centered on knowledge of different
cut off dates for analyzing CV and GI events and whether or not the “full
data” was available to Merck when they submitted the Vigor paper to the
NEJM. The NEJM authors reasonably believed that the late data was
unavailable when the VIGOR paper was submitted or that this data fell
outside of a scientifically justifiable “cut off.” Neither was the case.

Dr. Curfman’s testimony directly contradicts Merck’s lawyer’

Q. Doesn't that suggest to you that there was evidence that the
sponsor [Merck] had on CV issues that were not in the paper that were in
hand before the paper was published?

A. No.

Q. So, you read this sentence to conclude that all this evidence came
to Merck after the paper was published?

A. It doesn't specify when they had the data, and I don't think that
these authors would know. They [the authors of the JAMA article] are not
authors of the VIGOR article. They had no access to the primary data. They
weren't involved in the study. They're writing a commentary article here.
They are commentators, they are not authors, and this is not our Journal

Dr. Curfman repeatedly told Merck’s lawyer’s that their client had
played “hide and seek” with the data. Merck initially designated several
lines of this deposition “confidential,” claiming it contained trade
secrets, but in correspondence Merck told us that they withdrew their
confidentiality claim. We have made Dr. Curfman’s deposition available to
the public [] so readers can make
their own evaluation of Dr. Curfman’s knowledge. Merck complains that we
omitted key information that was available in this deposition but, but the
final version of our paper was submitted to BMJ, and the issue went to
press, before the deposition even occurred. If Merck’s lawyers believe we
misrepresented the facts surrounding their handling of Vioxx, they should
lift the veil of secrecy they have placed on their data and documents and
make them accessible on the web as the tobacco companies have done. The
drug is off the market; the only “trade secrets” are marketing
practices and methods of data manipulation. The medical community has
much more to learn from these documents.

Competing interests:
Dr. Egilman has served as an expert witness at the request of plaintiffs in the Vioxx litigation. Mr. Presler has served as a consultant at the request of plaintiffs in the Vioxx litigation.

Competing interests: No competing interests

09 April 2007
David S. Egilman
Amos H. Presler
Never Again Consulting, 8 North Main St., Attleboro, MA 02703