APPENDIX 122
Memorandum from the Lancet
1. (i) How do you ensure the integrity of
the peer-review process?
The Lancet has an in-house team of 12 physicians
and scientists who read around 8,000 submitted papers each year.
About 30% of papers are sent for external peer review. The remaining
70% are rejected because they fail on grounds of suitability for
our readership (international and clinical), topicality, originality,
or validity. The Lancet's editors select reviewers based on their
knowledge of the subject. Each editor has specialty area knowledge,
which he or she uses to select reviewers who are able to evaluate
critically each manuscript, or part of each manuscript. If an
editor perceives that he or she has a conflict of interest with
either the subject or authors of a paper, that article is reallocated
to another editor. Editors select at least three clinical/scientific
reviewers for each manuscript. The manuscript is then presented
to colleagues at a once-weekly manuscript meeting. All editors
collectively make the decision whether to proceed to statistical
review by statistical experts who regularly review for the journal,
to seek a revised paper from the authors, to seek further reviews,
or to reject the paper. Specialist reviewers in economics and
genetics are drawn from a separate database of reviewers, if required.
The process of review and revision is continued until a final
decision about acceptance or rejection is made. This group decision-making
process helps to eliminate the risk of individual editorial bias.
Editors submit a conflict of interest statement annually to senior
editorial staff.
Reviewers are usually contacted by e-mail and
asked if they are able to review the paper (they are given the
title and name of main author) within 1-2 weeks. Reviewers are
asked at this stage, and then again when they receive the manuscript,
whether they have a conflict of interest. If so, it is left to
individual reviewers to decline to review on the basis of a substantial
conflict, or to discuss their conflict with the editor, or simply
to declare any conflict on the reviewer's form. All reviewers
are asked to complete a form declaring that they either do or
do not have a conflict of interest, and to give details. Editors
then judge whether to exclude the review and choose another reviewer
or to use it. Every original research manuscript published in
The Lancet has at least three clinical reviews and one statistical
review.
Authors are asked to declare to us in their
submission letters that their manuscript is not under consideration
at another journal (to prevent duplicate publication). They are
asked to state what the contributions of each author are to the
manuscript (to ensure that all authors take responsibility for
some aspect of the work). Authors are asked to declare all conflicts
of interest at the end of the text of their manuscript. They are
also asked to explain the role of each source of funding for the
study (to ensure that all potential sources of bias are evident
to readers and reviewers). They are asked to prepare original
research manuscripts according to standard guidelines to ensure
that all necessary information to judge the methods used is present.
These processes help to ensure the integrity of the peer-review
process
In addition to these standard reviewing procedures,
The Lancet offers two additional services. First, fast-track of
clinically important work using the identical peer-review process
described above. By means of email, pre-contacting reviewers,
overnight editing (if necessary), accelerated production scheduling,
and early-online publication, we can take a paper from receipt
to publication in 4-5 weeks. Second, we offer a formal appeals
procedure for all rejected manuscripts. Appealed articles are
read by the original editor who dealt with the article plus myself.
If the appeal is accepted, it enters the usual process of peer
review once again, as described above.
A further important, and sometimes neglected,
aspect of peer review is the work conducted by our seven technical
editors. Once a paper has been accepted for publication, it undergoes
line-byline editing. Errors are often found and corrected at this
stage. Clarity is always improved. Ambiguities are removed. Manuscript
editing at this level of detail is a declining art in traditional
publishing. We recognise it as one of the key strengths of The
Lancet.
In summary, by involving 12 editors, at least
four specialist reviewers, and the relevant authors in decisions
to publish all original research in The Lancet, together with
the input of a team of technical editors, we minimise the likelihood
of bias, error, dishonesty, fraud, and misconduct. These processes
rely on the integrity of the individuals involved, and we rely
on trust between editors, reviewers, and authors.
(ii) Do you review your reviewers and, if
so, by what process?
Reviewers have to fill in a 3-page form for
each manuscript. Those who give poor reviews are less likely to
be approached again by the same editor. Evidence of poor reviewing
is entered into a computer database of all peer reviewers used
by the journal. At the weekly meeting where manuscripts and their
reviews are discussed, all editors can see the reviews and judge
for themselves the level of critique given. This gives editors
the opportunity to assess a reviewer's area of expertise and modify
our database accordingly.
2. What are your views on the impact that
a "pay-to-publish" business model would have on your
peer-review process?
If one assumes that all editorial decisions
are hermetically sealed from the publishing environment in which
those decisions are made, one could safely consider that a "pay-to
publish" business model would have no impact on our peer-review
process. This assumption is at best naive, at worst flawed.
The reality is that editors face—and sometimes
try to resist—many indirect pressures on their decision
making. Take, for example, the desire to win higher subscription
and web-usage numbers. For a medical journal such as The Lancet,
our readership is mostly clinical—that is, it consists of
individuals qualified to practice medicine. That audience drives
us to select material for publication that is likely to influence
clinical thinking, clinical research, or clinical practice. This
seems to us a valuable pressure—it provides an incentive
for us to produce a journal designed to best suit the needs of
our readers. There is also pressure to produce a journal that
meets the needs of authors. In part, this requirement means an
efficient and effective peer-review process. It also means ensuring
that the journal's Impact Factor is competitive.
There are less benign pressures that increasingly
press themselves on editors. It is important to appreciate these
adverse pressures if one is to understand the potential impact
of a "pay-to publish" business model on editorial decision
making. One of the most difficult areas for The Lancet is research
sponsored by the pharmaceutical industry. In a traditional publishing
environment, especially one that is more commercially oriented,
the pressure to publish research that will earn revenues from,
for example, the sale of reprints is intense. The pharmaceutical
industry is seen as an important partner for publishers—in
revenue-generating research papers, supplements etc. Any criticism
of industry by editors is also a matter of particular sensitivity.
Although these pressures are not direct on The Lancet's editorial
team, we are increasingly made aware of their existence an environment
that can, at times, be uncomfortable and could be, if it were
not resisted, damaging to the journal's independence.
In exactly the same way, a "pay-to-publish"
business model would inevitably create an environment in which
editors perceived a pressure, especially where profit margins
were tight, to accept more papers (and so generate more revenue).
The publication of more papers, in our daily experience, would
be bound to lower the overall quality of The Lancet a severely
damaging result. If editors tried to resist this pressure, as
I am sure we would, our quality would be preserved at the expense
of our financial viability—an extremely unwelcome balance
of forces. Furthermore, in a "pay-to-publish" model,
there would likely be a perverse shift in the type of submission
sent to us. We would receive fewer papers from authors who might
find it hard to pay a submission or publication fee. We would,
instead, receive papers from those who could pay. I would be concerned
that research sponsored by industry would have no access barrier
to publication, whereas those authors not backed by a wealthy
sponsor could face serious difficulties. Such "pay-to-publish"
journals may therefore receive and publish more industry-driven
research, thus compromising their independence and neutrality.
One of the vital safeguards of scientific integrity would have
been erased.
3. (i) Does The Lancet take responsibility
for the papers that it publishes after publication?
Yes, 100%, and forever. That responsibility
transfers, across time, from one group of editors to another.
If an error is discovered, and if it is serious enough to need
correcting, then a correction will be printed. The Lancet's website
and MEDLINE link that correction to the original published article.
(ii) For example, in deciding to publish a
paper, what weight is attached to the possibility that certain
campaigners are likely to attach an exaggerated significance to
certain papers which support their view?
Editors always consider the way in which articles
may be used or interpreted once published, not only by "campaigners"
but also by doctors, scientists, the media, and members of the
public. The possibility of a paper being misused would never stop
the journal working on an article; it becomes just one of many
factors taken into consideration before a final decision is made.
If a controversial paper is accepted, we weigh very carefully
indeed the claims made in the article (ensuring that they are
properly qualified where necessary), we frame the work in a responsible
clinical or public health context by commissioning a linked commentary
to be published in the same issue, and we consider if and how
to present the article to the media in an accompanying press release.
In more recent years, we have seen a trend for authors and their
institutions/sponsors to press release and publicise their own
work, separately from the journal. While understandable, this
process carries its own risks since different messages can be
transmitted, confusing media and public alike. We try to work
with all parties to ensure that the same and appropriate scientific
or clinical message is published.
(iii) What steps do you take to respond to
the distortion or misrepresentation of papers published in your
journal in cases of great public interest?
There are two parts to our procedures: one before
and one after publication. Before publication, editors ensure
that the interpretation sections of the discussion and summary
do not go farther than the results allow. Statistical reviewers
can be helpful here. Sometimes the authors will be asked to alter
their interpretation, especially to tone it down. Editors may
ask authors to list any limitations of their study. The Lancet
has a comprehensive set of guidelines for authors on conflict
of interest, as already discussed. The role of the funding source
is also made clear. For clinical trials, the CONSORT checklist
is used by editors to ensure that all the CONSORT points have
been described in the paper (eg, an effect size as well as relative
risks). CONSORT now insists that authors list major side-effects
in the paper in absolute and percentage terms, and The Lancet
ensures that the main or any unexpected side-effects are also
reported in the summary. As already mentioned, we invariably commission
an expert to write a signed commentary on controversial papers,
to put the study and its findings, together with any general implications,
into context. A study may be included in a press release, written
by a dedicated press officer and checked by an editor with a clinical
qualification, and it will include the main results and any adverse
events in simpler language than the journal might use. It would
highlight, where appropriate, any limitations of the study. A
commentary linked with the paper would also be mentioned in the
press release, with a quote from the author of the commentary.
The press release includes contact details for the main author
of the paper and of any commentary. If appropriate, one of the
senior editors will be available for interview by a journalist.
After publication, the journal has a correspondence
column that acts as a forum for critical debate about the work
we publish. We will also, in cases of substantial clinical or
public health interest, consider papers that report contrary findings—that
is, we position ourselves as a place where the self-correcting
nature of science can be played out transparently and accountably.
In occasional instances, we might refer to distortions and misrepresentations
in editorials, which are written by the editors, and signal major
errors or new material facts relating to work we have published
by notifying the press, for example, by means of a news release.
4. How does The Lancet implement the COPE
guidelines on good publishing practice?
As one of the founding members of COPE in 1997,
I, together with other Lancet editors, have been actively involved
in devising the COPE guidelines, which were first published in
1999. COPE started out as a sounding board for editors. The committee
meets bi-monthly and gives advice to editors about how to handle
suspected cases of publication and research misconduct. In 2001,
COPE matured into a more formal body with a COPE Council committee,
an elected chair and vice-chair, and subcommittees for research
into misconduct and for education about good research practice.
Two Lancet editors are members of the COPE Council and are joint
chairs of the education subcommittee. The Lancet fully endorses
the COPE Guidelines on Good Publication Practice (it took part
in their drafting) and we have been actively involved in devising
a basic Code of Conduct for Editors, which will become binding
for all COPE members shortly. In addition, one Lancet editor will
be working on a gold-standard Code of Conduct for Editors. The
Guidelines on Good Publication Practice are divided into 11 major
points:
(a) Study design and ethical approval
The Lancet requires statements by authors on
ethics approval where appropriate. However, editors and peer reviewers
will, in addition, judge the ethical conduct of a study independent
of such a statement. For all randomised controlled trials, we
require the submission of the protocol of the study to scrutinise
the planned design and to compare this with the submitted paper.
We expect all other studies to state a rationale for the research
undertaken. All case reports, clinical pictures, or other easily
identifiable material will only be published afier The Lancet
has seen an appropriate signed informed consent form.
The Lancet sends every paper with data analyses
(and the protocol in the case of clinical trials) to statistical
advisors who have specific guidelines to assess methods, analysis,
reporting and interpretation of data. In addition, a CONSORT table
and figure is normally required for all randomised controlled
trials.
All authors are required to sign an authorship
and contributor statement explaining their role in the planning,
design, analysis, and reporting of a study. A summary of the contribution
of each author is published at the end of the paper. The corresponding
author signs a statement taking responsibility that all authors
have seen and agree with a final version of the paper before acceptance.
Should authorship disputes arise, a paper will not be published
before these are resolved satisfactorily.
(d) Conflicts of interest
All authors have to sign a conflict of interest
statement explaining whether they have any financial and personal
relationships with other persons or organisations that could inappropriately
influence their work. Examples of financial conflicts include
-employment, consultancies, stock ownership, honoraria, paid expert
testimony, patent applications, and travel grants, all within
three years of beginning the work submitted. If there are no conflicts
of interest, authors should state that there are none. These conflict
of interest statements are published under a separate heading
at the end of the paper. All reviewers have to state their conflict
of interest on the reviewer form and are advised to speak to the
editor handling a paper if they have a conflict of interest before
proceeding with the review. All editors declare any conflict of
interest to the Editor and Deputy Editor and will take themselves
out of the peer-review process if a conflict would bias their
decision-making in relation to a particular paper.
All original research and all review material
is peer reviewed. Original research papers are seen by three clinical
and one statistical reviewer before publication. The Lancet peer
review is largely confidential but if the reviewer wishes to disclose
his or her name to the authors we will do so. Reviewers are pre-contacted
and reminded if necessary to ensure a speedy review process. Descriptions
of our peer review, selection, and appeals processes are published
in our information-for-author section once a month in print and
continuously on our website. We regularly monitor our acceptance
rates and publication times and strive to keep delays to a minimum
without compromising the quality of peer review.
(f) Redundant publication
We do allow re-publication of a paper in a foreign
language with disclosure of the original English-language publication.
Any other form of redundant publication is unacceptable and a
note of duplicate publication would be published if this has indeed
been confirmed.
Permissions for any previously published figures,
illustrations, or tables are required. Any undisclosed plagiarism
would be pursued by the editors.
All action points under this subheading are
observed by Lancet editors. In particular, we base our decision
to accept or reject a paper on its importance, originality, and
validity without stifling innovation or preliminary observations,
provided their significance is not overstated. We do consider
studies that challenge previous work published in the journal
carefully. We do not exclude studies reporting negative results
if they are clinically important. All published original studies
are peer reviewed. We treat all submitted papers as confidential.
We do correct the record if a published paper is subsequently
found to contain major flaws. We follow due process when misconduct
is suspected and contact the authors first before taking the matter
further if needed. Our instructions to authors make it clear that
authors must take informed consent from patients included in their
research.
We state our policy on presentation of results
in meetings in our instructions for authors. Presentation of data
at a scientific meeting has no conflict with submission to The
Lancet. We have a dedicated press officer, who chooses 3-5 papers
from each issue, writes summaries of their findings, which are
then checked by the authors and an editor, and he sends those
out under embargo 36 hours before publication.
Editorial decision-making is not influenced
by reprint value. Information on upcoming publications is confidential
and known by non-editorial departments of The Lancet only up to
eight days before publication (when journalists will be given
this information also). We have a stringent advertising policy
that does not allow unsubstantiated claims to be made or juxtaposition
next to related research material. All advertisements are checked
and approved by an editor. Reprints are published as they appear
in the journal unless a correction has had to be added.
(k) Dealing with misconduct
When misconduct is suspected, we follow due
process as recommended by COPE, in that we seek clarification
or explanation from the authors first. If that is either not satisfactory
or not forthcoming, we take the matter further and write to the
employer, the institution, or any other relevant bodies (such
as licensing authorities) to request that they investigate the
claim. If serious misconduct is proven, we would take appropriate
steps, such as retraction of a paper or notification of duplicate
publication. If there are serious concerns that cannot be proven,
we would consider publishing an expression of concern. We aim
to present all such cases at COPE for the record and for further
advice.
5. Can you outline your policy on retraction?
What is a partial retraction, and how is this covered by the COPE
guidelines?
The Lancet's policy on retraction follows the
guidance provided by COPE in its Guidelines on Good Publication
Practice, Sections 3 (Serious Misconduct) and 5 (Sanctions).
A partial retraction is simply the retraction
of a very specific part of a previously published paper. Neither
COPE nor the International Committee of Medical Journal Editors
(nor, for that matter, the World Association of Medical Editors)
has a separate policy on partial retraction. Rather like the difference
between statute law and case law, partial retraction is, nevertheless,
a well-described procedure in the scientific literature. For example,
— Retraction of part of a dataset
See Herman TS. Int J Radiat OncolBiol Phys 1995;
32: 897-98.
See Thompson JA et al. Proc Natl Acad Sci USA
1992; 89: 7849.
— Retraction of an interpretation
Zhang L et al. Science 2004; 303: 467.
In the case of the 1998 paper by Dr Wakefield
and colleagues, we noted that the press coverage of the partial
retraction distinguished clearly between the differing elements
of the work.
March 2004
|
