pen to paper, and judge whether the quantity of
material supports a whole paper, a brief report or
even more than one paper.
Authorship on papers is a matter of substan-
tial debate. Under some circumstances, literally
dozens of coauthors will clamor to be listed, and
this phenomenon is not restricted to the publica-
tion of huge multicenter clinical trials. Clinical
trials are a specific case of this general, peren-
nial problem, to which Rafal (1991) has pro-
vided a somewhat humorous guide. There are
two solutions.
The first solution is the prospective promulga-
tion of a set of criteria that every author must meet.
Many journals publish their own specific guide-
lines or criteria, and these do not differ greatly in
qualitative terms. In the practicality of publishing
clinical trials, the following would be typical:
(a) The principal investigator(s) is/are authors
unless so numerous as to require a team desig-
nation (see below).
(b) The statistician(s) who personally accept(s)
responsibility for the statistical analysis in the
corresponding document(s) that is/are sub-
mitted to regulatory authorities should sign
off on the paper and be named as author(s).
(c) Key members of the clinical team within the
pharmaceutical company may (but not neces-
sarily need to) be authors.
(d) All named authors should be able to personally
defend the paper after publication, and be
familiar with (but not necessarily have person-
ally performed) all the methods employed in
the clinical trial.
(e) There shouldbe nocircumstances where ‘guest
authorship’ or ‘gratitude authorship’ is awa-
rded; all authors’ participation must have been
fundamental to the conduct and success of the
clinical trial.
(f) All authors should be prepared to disclose all
conflicts of interest and the sources of financial
support for the clinical trial.
The second solution is to publish the paper under
the name of the team that conducted the trial, rather
than the personal names of the participants. The
acknowledgments can then list all those who took
part (e.g. The Subcutaneous Sumatriptan Interna-
tional Study Group, 1991). A hybrid variant is also
sometimes used, where a one (or a few) lead
author(s) is named and stated to represent the rest
of the team (e.g. Cadyet al., 1991)^1.
The advantages of this tactic are that there is at
least one person who accepts responsibility for
defense of the paper after publication. A further
advantage is that this can be used to motivate
investigators in multisite studies: the protocol can
state that the investigator who recruits the most
completed patients, without violations, will be
named the first author in any publication.Isolated abstracts and posters
An argument can be made that the isolated abstract
format is not a good vehicle for the publication of
clinical trials. Indeed, the inclusion and exclusion
criteria in most clinical protocols alone exceed the
word limit of most journal article abstracts. Too
often, the publication of an abstract or poster is a
criterion used by companies to justify the time and
expense of sending staff to a conference: authors
then generate and submit unimportant abstracts,
principally for use as tickets to venues that attract
them for ulterior reasons.
There are a fewexceptionsto this generalization,
however. Legitimate retrospective analysis of the
database of a clinical trial that has been previously
published in full sometimes can make an isolated
abstract, provided the full reference is provided,
and an educated audience at, say, an academic
conference, will be aware of the potential biases
of this technique. Similarly, the open-label toler-
ability extension to a previously published con-
trolled trial might be usefully published as a
poster. But these are minor exceptions to the gen-
eral principle that in order to assess the validity of a
clinical trials report, far more detail is needed than(^1) See especially the footnote to the first column on page 2831
and the acknowledgments.
568 CH42 PUBLISHING CLINICAL STUDIES