JAMA Policy & The Brave New World
To the editor:
Fontanarosa and DeAngelis (1) claimed that our editorial about JAMA's
new rules for industry-sponsored studies contained "numerous errors and
misconceptions." In their elaboration, however, they mention only two
specific problems: 1) the title of our editorial suggested that the JAMA
policy applies only to clinical trials, but in fact the policy applies to
all studies; 2) our editorial states that under the new policy, authors
from industry must hire an academic statistician before submitting a paper
to JAMA, whereas the JAMA editors suggest that their new policy allows
submissions from industry without an academic statistician. JAMA will
simply refuse to accept these submissions until after an academic
statistical review is purchased by the authors.
Regarding their first claim, we agree that the title of our
editorial, which referred to trials, was misleading, but this mistake was
not ours. Our editorial's title was "The Brave New World of Journal
Submission." An editor of the BMJ changed the title after we reviewed
final proofs and without consulting us, thus confusing the issue and
making it seem that we misunderstood how broad JAMA's new policy is. We
reported this error to the BMJ on the morning it appeared in print, and it
appeared in the BMJ online rapid response on December 13 (2). The BMJ have
published a correction acknowledging this on 21 January (3). Fontanarosa
and DeAngelis must have overlooked the rapid response, and the BMJ editors
did not mention that the error was theirs when they reviewed the letter
from Fontanarosa and DeAngelis.
Their second claim surprises us. We would not have guessed that JAMA
is encouraging authors from industry to submit papers without academic
statistical approval. The editors will be bound to reject these
submissions under their new policy, though they may make a distinction
between those they might reconsider and those that they will not
reconsider. If a paper must have a statistical author from an academic
institution to be acceptable, we assumed that investigators who work in
the private sector would not consider submitting a paper to JAMA without
such an author. Why would JAMA want to review such submissions if they are
unacceptable a priori? Does this stance imply that the academic review is
merely cursory, and that the academic statistician who must be consulted
would not join the original investigators as an author of the paper? Or is
JAMA suggesting that investigators from industry should submit papers with
an incomplete list of authors, and then add another author during the
Fontanarosa and DeAngelis describe our criticism of their policies as
"blatant and unbalanced" and they take umbrage "that a journal like the
BMJ would criticise another journal's policy on such a pressing issue
without properly vetting the material for accuracy." We would have thought
that journal editors would welcome criticism of their work as warmly as
they encourage critical appraisal of the papers they publish, even if the
criticism appears in a forum outside of their control. Interestingly,
previous editors at JAMA chose to publish criticism of editorial policies
at Science and the New England Journal of Medicine (4). As for vetting for
accuracy, we sent JAMA a copy of our essay before it was published in the
BMJ, and they did not respond. Furthermore, we presume that the editors of
JAMA believe that all journals should have such a policy.
Finally, citing examples of industry malfeasance, of which we agree
there are many, does little to justify an inequitable policy, and one that
we believe will be difficult to implement. We are not aware of instances
where the malfeasance can be regarded as the responsibility of the
industry-funded statistician who authored the paper. The examples cited by
Fontanarosa and DeAngelis would probably not have been prevented by having
the analysis reviewed by a statistician from an academic institution. At
the least, some specific evidence on this point should be cited by JAMA to
justify their policy. As we stated earlier, the problems are much wider.
To suggest, as Fontanarosa and DeAngelis do, that we "do not fully
acknowledge the magnitude and implications of these problems" is
Statisticians who are members of major professional societies, such
as the Royal Statistical Society or American Statistical Association, are
subject to an ethical code of conduct. JAMA suggests by their policy that
such statisticians are less ethical than their academic colleagues. As
Senn noted (5), "independent" academic authors have their own biases, and
unlike authors from industry, they do not usually have to undergo the
intense scrutiny of regulatory authorities.
Kenneth J. Rothman, vice president, epidemiology research
RTI Health Solutions, RTI International, 200 Park Offices Drive,
Research Triangle Park, NC 27709, USA
Stephen Evans, professor of pharmacoepidemiology
Medical Statistics Unit, London School of Hygiene and Tropical
Medicine, London WC1 7HT
1. Fontanarosa PB, DeAngelis CD: JAMA's policy on industry sponsored
studies. BMJ 2006;332:177.
2. Rothman KJ, Evans S: Corrections and a rebuttal to Vandenbroucke.
BMJ Rapid response Dec 13, 2005.
3. Correction: Extra scrutiny for industry funded trials. BMJ
4. Rothman KJ: Conflict of interest--The new McCarthyism in science.
JAMA 1993; 269:2782-2784.
5. Senn S: Constant gardner syndrome. BMJ Rapid Response Jan. 20,
KJR is an employee of RTI Health Solutions, an independent non-profit research organisation that does work for government agencies and pharmaceutical companies. SE has no conflicts of interest to declare.
Competing interests: No competing interests