|
Rachel Graham is staff editor on Hospital Pharmacist |
Concerns about the reporting of clinical trial results in medical journals
have recently been expressed in a variety of forums.
For example, one of the topics of a speech given by Richard Smith, former
editor of the BMJ, was the ways in which the results of clinical trials
can be presented to appear favourably to companies with new drugs for
which they are seeking marketing approval.1
Dr Smith’s presentation was made as he received the 2004 award,
from HealthWatch, a registered charity that aims to support and encourage
the scientific testing of conventional, complementary and alternative
medicine and therapies. Examples highlighted by Dr Smith include:
· Comparing a new drug with a placebo or too low a dose of a competitor’s
drug
· Comparing a new drug with too high a dose of a
competitor’s drug, so that it looks as though the new drug has
fewer side effects
· Carrying out just a
small-scale trial comparing a new drug to a better, but more expensive
treatment, so that the results show that there is “no significant
difference” between the two agents, thereby making the new drug
appear to be good value for money
· Reporting just the
end-points of a clinical trial in which a new drug
performed well
· Reporting just the results of particular groups of patients, or patients
from particular centres in a
multi-centre trial, in which a new drug performed well
· Not publishing clinical trial results at all, if there is no subgroup
that does well
· Publishing good studies more than once by
including, for example,
different outcome
measurements or the results from different follow-up periods in separate
papers
· Publishing positive results in major journals and negative or neutral
results in minor journals
As an illustration of these methods being used, Dr Smith cited one particular
drug about which there were publications describing 84 trials on 11,980
patients. In fact, there were only 70 trials involving 8,645 patients,
but 17 per cent of the trials had been published more than once. This
was not identifiable from the published studies. He went on to demonstrate
how the multiple publication of trials could increase apparent effectiveness,
using an example of how 16 trials showing a number needed to treat (NNT)
of nine could be presented as 25 trials showing a NNT of five, by duplicating
the results of the most favourable trials.
In a similar vein, members of Parliament heard just before Christmas
from Richard Horton, editor of The Lancet, about ways in which pharmaceutical
companies attempt to influence the editing of clinical trials reports
they have submitted for publication. Dr Horton told the House of Commons
Select Committee inquiry, set up to investigate the influence of the
pharmaceutical industry, about how drug companies sometimes imply that
there is a link between their submission of a paper with revenue for
medical journals, from reprints.After submission, authors or sponsors
might then try to intervene to move the peer review process along in
directions that are less critical of their drug.2
Dr Horton gave as an example an unnamed company that had threatened to
withdraw a paper on a cyclo-oxygenase-2 inhibitor because it believed
that the review process was over-critical. The company stopped interfering
after the paper’s authors had been told that the paper would be
rejected unless the company backed off, he said.
There has also been criticism of the way in which clinical trials are
reported and analysed once a drug has received marketing approval.
A recent paper in The Lancet concluded that the concerns about the
cardiac safety of rofecoxib should have been known by the end of 2000,
and hence the drug withdrawn earlier than it was.3 Their study was
based on the results of 18 randomised controlled trials and 11 observational
studies incorporating 20,000 patients. It should, however, be pointed
out that the manufacturers of rofecoxib did not agree with this assertion,
stating that the authors had inappropriately combined heterogenous
data, which runs counter to the basic principles of meta-analysis.4
Among the conclusions that seem appropriate to draw from all this information
is that it is important that those carrying out meta-analyses of clinical
trial reports, or reviewing such literature with a view to prescribing
a drug or informing clinical practice, read the equivalent of the “small
print” in the original papers. Whatever the vigilance exercised
by pharmacists and others carrying out such activities, a fair reflection
of a drug’s safety and efficacy will be hard to achieve if trial
results have been published more than once (this not being evident from
the reports) or if unfavourable results have not been published at all.
Dr Smith’s suggestions of a register of all trials, so that unfavourable
trials do not “disappear”, critical review of trial protocols
by independent experts, publication of trials in online journals that
do not rely on commercial sponsorship for revenue and more public funding
for clinical trials, seem to be particularly worthwhile aims.1
References
1. Garrow J. Medical journals: an extension of the marketing arm
of drug companies? HealthWatch
Newsletter 2005;56:4
2. MPs hear how drug companies try to sway editors. The Pharmaceutical
Journal 2005:274:7
3. Jüni P, Nartey L, Reichenbach S, Sterchi R, Cieppe PA, Egger
M. Risk of cardiovascular events and rofecoxib: cumulative meta-analysis.
The Lancet 2004;364:2021–9
4. Vioxx “should have been withdrawn four years ago”. The
Pharmaceutical Journal 2004;273:704
|
Hospital Pharmacist