David Doggett wrote:
> Subject: Requiring preregistration of trials for licensing
> In the most recent JAMA there is
> a scathing editorial by Drummond Rennie that was prompted by an article in
> the same issue concerning the use of disguised multiple publications and
> inappropriate controls in pharmaceutical trials to possibly gain licensing
> and innappropriate commercial advantage.
As part of a team doing a systematic review of olanzapine
(another Eli Lilly drug), I have some sympathy with the
multiple publication issue. One study, (Tollefson 1997), has
so far hit 82 for the number of separate publications (which
admittedly includes conferences). A little bit of fun was
had trying to work out the number of people randomised to
the number of study reports, this is what we found:
"There is a close correlation between study size and number
of presentations. If regression analysis of the data in
Table 01 is undertaken on a log-graph the least-squares line
makes a close fit (R2=0.92). For every 10 people randomised
in an olanzapine versus typical drugs comparison there is a
report. A study of 10 could be predicted to merit a report,
of 100-10 presentations and of 1000-100 reports. Considering
the likely under-identification of presentations of the
larger studies this pattern of comprehensive dissemination
does not seem too unlikely"
> These problems were exacerbated
> because the authors of the JAMA meta-analysis were unable to adequately deal
> with the problems using the information in the pubished reports; and the
> study authors, institutions and pharmaceutical companies that controlled the
> original data were uncooperative. At best this is incompetent science on
> the part of the study authors; and at worst it may be cynical manipulation
> of data by the pharmaceutical companies. These sorts of things makes a
> mockery of the licensing process and the published scienific record.
I didn't fully understand the comments about the risperidone
meta-analysis, as a Cochrane systematic review has been done
and is published and maintained in the Cochrane Library, so
is maybe not so much about incompetent science on the part
of the study authors (shrewd marketing though). Certainly
it becomes more difficult when there are multiple citations
of one study, but it is manageable.
Also, in fairness to the drug companies, more and more they
are co-operating with groups like Cochrane, (Eli Lilly have
helped us enormously with identification of extra data and
confirmed the citations that we found as being part of which
study - admittedly they benefit if the review goes in their
favour and we have to be wary of selective 'drip feeding' of
information).
I suppose the point I'm making is there are changes
happening, where it becomes harder and harder for companies
to 'hide' trials, but the sooner they start registering them
at an independent site, it will become easier. There is a
problem though, as many trials are done in non-english
speaking places, and english speakers tend to ignore the
non-english published journals.
With every best wish,
Mark Fenton
Research Nurse
Cochrane Schizophrenia Group
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|