Written evidence submitted by Dr Ben Goldacre (CT55)
1. I am a medical doctor, currently working as a Research Fellow in Epidemiology at London
School of Hygiene and Tropical Medicine. For the past ten years I have written about
problems in science for the Guardian, and in two books: Bad Science, and Bad Pharma. I am
also a co-founder of alltrials.net, a widely supported non-profit campaign group seeking to
improve access to clinical trial results.
3. Healthcare professionals and patients need the results of clinical trials to make informed
choices about which treatment is best. Currently, drug companies and researchers are allowed
to withhold the results of clinical trials, on treatments currently in use, from doctors and
patients if they wish to. This means that we are misled about the benefits and risks of
treatments. We can be misled into prescribing an expensive new drug, for example, when in
reality an older cheaper one is more effective. As a consequence, patients are exposed to
avoidable harm, and money is wasted unnecessarily.
4. Withheld results are a problem for both industry and academic trials. The best currently
available evidence, from the most current systematic review, estimates that only half of all
trials are published, and trials with positive results are twice as likely to be published. A
systematic review is the most robust form of evidence, since it is an unbiased overview of the
evidence. This systematic review is published by the NHS NIHR HTA programme.
6. The ongoing problem of withheld trial results has not been adequately addressed by any of
the initiatives in place today. The FDA Amendment Act 2007, for example, requires that
results for a subset of trials (one research site in the US, studying a currently licensed drug,
etc) are posted at clinicaltrials.gov within one year of completion. This legislation is widely
cited as evidence that the problem of missing trials has been fixed. However there was no
routine public audit of implementation, and when one was finally conducted, and published in
the BMJ in 2012, it found that this law has been ignored by four trials out of five.
7. Prayle AP, Hurley MN, Smyth AR. Compliance with mandatory reporting of clinical trial
results on ClinicalTrials.gov: cross sectional study. BMJ. 2012;344:d7373.
9. In any case, this and other interventions would have done little to improve medicine today,
even if they were effective. This problem cannot be addressed prospectively, by ensuring
access to trials finishing after 2008, or after 2013: around 85% of medicines prescribed in the
UK are “generic”, and came to the market a decade ago or more. It is the evidence from this
era of clinical trials that we need most – 2003, 1999, 1993 -to ensure prescribing today is safe
and effective. In almost all cases (although perhaps not for aspirin trials six decades ago) this
information still exists. Doctors and patients should be given access to it to make informed
10. “What is the role of the Health Research Authority (HRA) in relation to clinical trials
and how effective has it been to date?”
12. I am working with colleagues on low cost randomised controlled trials,seamlessly embedded
in routine clinical care, using the General Practice Research Database, and have submitted a
response with them on the separate issue of administrative barriers to doing clinical trials
more efficiently in the UK.
13. So far ethics committees have not sought to address the issue of withheld trial results. This is
problematic, as it is one of the key ethical problems in medical research. Patients participate
in trials in the belief that they are helping to improve knowledge and treatments for future
patients. Where trial results are withheld, those patients have been misled. I understand that
Janet Wisely, the new head of the HRA, is keen to engage on this issue. In my view there are
certain elements that should be laid down in the legislation for this body.
14. Research ethics committees should ensure that researchers do not have a previous track
record of leaving trial results unpublished, before granting them permission to conduct further
studies on trial participants. This can be done at almost no administrative cost, by simply
requesting a signed statement from the lead medic or primary investigator that they are not
witholding the results of any trials more than one year after completion. Similarly the HRA
should insist on a commitment to publication, then publicly monitor and audit compliance.
Again this would not require any significant administrative or investigative activity: an ethics
committee can simply make a diary note, write to the primary investigator, and ask for a link
to results publication, whether in an academic paper or on a results registry, one year after
completion of the trial.
15. “What evidence is there that pharmaceutical companies withhold clinical trial data and
what impact does this have on public health?”
16. I have addressed this in an attached memorandum, as suggested, because it is adapted from an
earlier document which I initially drafted as a briefing note for Earl Howe, and then as a
briefing note for alltrials.net.
17. How could the occurrence and results of clinical trials be made more open to scrutiny?
Who should be responsible?
18. In my view there has been a widespread systemic failure by regulators, industry, universities,
policy makers, and medical and academic professional bodies to take ownership of this
problem. As a consequence we have seen an incomplete patchwork of interventions that have
failed to address the core issue – we need doctors and patients to have access to all results of
all trials on all currently used treatments. Instead we have engaged with peripheral details.
19. “Clinical trial registries” are a clear illustration of this problem. Registers are public lists that
contain a small subset of all the trials that have been conducted on a medicine. They are
incomplete by design. The European Clinical Trials Register is a list of trials conducted
within Europe over the past few years. It is not a list of all the trials that have been conducted
on all the medicines currently available in Europe. It should be, or it should at least strive to
be. Clinicaltrials.gov, similiarly, is mostly trials conducted in the US, mostly from the past ten
years, and with compulsory registration only since 2007 (though even here compliance is
uncertain). These limitations reflect the early administrative origins of these registries. They
are not what trial registries should be, or need to be, to inform evidence based clinical
practice, and to achieve the clear goal of access to all results. 20. The European Clinical Trials Register should simply be a list of all the trials ever conducted,
on all the medicines currently prescribed in Europe. It should include results. Where and
when these trials were conducted is irrelevant. It is clear to me, from my experience of talking
to the public, journalists, doctors, policy makers and academics about these problems, that
many people believe a trials register to be just this: a complete list of all the trials that have
been conducted. This is indeed what they should be.
21. There are many stories of how companies have refused to hand over information. These in
turn have generated discussions about what levers are available, and how we can force
companies (in particular) to hand over trial results. I am struck that we have never tried
simply asking, in a systematic fashion. The European Medicines Agency could say: “You
have a marketing authorisation to sell your medicine in Europe. We maintain a list of all trials
ever conducted on all uses of all medicines currently prescribed, so that doctors and patients
can make informed decisions. Here are the forms: please tell us about all the trials you hold,
or are aware of.”
22. However there are many other stages where influence could be brought to bear. IQWiG, the
German equivalent of NICE, has developed a reputation for demanding high standards of
evidence before approving a drug for use, and also for requiring all trial results to be shared
with them, and then making those public. It is through this mechanism that we have become
aware of major problems with currently used medicines such as reboxetine.
24. We could use this more robust approach in the UK. We could also insist that a treatment is
only available for prescription after all the trial results have been made publicly available.
25. Universities could also insist that all results are published, and that all collaborative contracts
between academics and industry include the right to publish, and the right of access to data.
26. There is also the matter of culture. I am concerned that the impact of withheld results on
patient care is currently a cultural blindspot in medicine and academia, even despite
systematic review evidence showing that half of all trials do not go on to be published. It is a
peculiar paradox that we spend so much money on each individual trial, investing huge effort
to ensure that they are free from bias, carefully appraising their strengths and weaknesses,
then allow so many of them to be simply deleted from the record. This non-publication
reintroduces all the biases we spend so much time and money avoiding back into the evidence
27. We need wider recognition that this is a serious problem, and that it gravely undermines our
attempts to practice evidence based medicine and make informed decisions. If a researcher
selectively deletes the unflattering data points from one single trial, in order to massage the
results and get the result they want, they are rightly regarded as being guilty of research
misconduct. This is a serious business, since they are misleading doctors and harming
patients. However, if a group of researchers delete whole trials from the overall research
picture, then there are no reputational or professional consequences, even though we know
that this will distort the apparent benefit of the treatment, just as surely as one researcher
fraudulently manipulating the results of a single trial.
28. This could be addressed in part through medical and academic membership bodies. They
could demonstrate leadership on this issue, state clearly that withholding the results of clinical
trials is research misconduct, and impose sanctions or even ejection where appropriate. 29. I am concerned that currently most major UK medical and academic bodies have done the
opposite. Many are currently signatories to a pair of documents produced by the “Ethical
Standards in Health and Life Sciences Group” that give false reassurance around the issue of
withheld trial results. The ESHLSG is co-chaired by the ABPI (the UK pharmaceutical
industry body) and the Royal College of Physicians. Engaging with industry on ethical
challenges is plainly a good thing. However the ESHLSG documents appear to make
misleading statements about the problem of withheld results. For example they make
extensive reassuring comments about current regulations, while failing to disclose the best
currently available evidence, from fully published academic papers, in leading peer reviewed
journals, which demonstrates that these regulations have been routinely ignored. I believe this
issue may be covered in more detail in a submission by the “Bad Guidelines” group, but I am
happy to give more details.
30. “Can lessons about transparency and disclosure of clinical data be learned from other
31. No. The European Clinical Trials Register is incomplete even by its own standards, with
many trials still withheld from the public form of the register, and the EMA has failed
outright to deliver on key objectives,such as their promise to carry results on their register by
2012. The US registry at clinicaltrials.gov is incomplete by design, as discussed above – it is
not retrospective, and does not cover all trials on a treatment - and even for its limited remit
the FDA Amendment Act 2007 has not been adequately implemented, with trial results still
routinely withheld, as shown by Prayle et al 2012 referenced above. Even if similar
legislation was perfectly implemented, and muscularly enforced, any beneficial impact of
getting trial results from now onwards would only be felt in several decades’ time.
32. We need to ensure that doctors and patients have access to all results of all trials that have
ever been conducted on all treatments currently in use, in order to make informed decisions
about which treatment is best. We need serious working groups to discuss how to achieve this
33. We need a credible public process to explore the details of the costs involved to industry and
academia of retrospective disclosure, in either summary results form or Clinical Study
Reports, where those exist. The EMA now discloses Clinical Study Reports on the small
proportion of trials that they hold, and the European Ombudsman said that the administrative
burden of doing so – and of removing some identifiable patient information where
appropriate -is not significant. GSK have also committed to releasing Clinical Study Reports,
in signing up to alltrials.net
34. I would also suggest a pilot of full disclosure, either for some commonly used drugs, or for
some commonly used classes of drugs. This would allow us to identify any costs, the changes
to the evidence base for current decisions, and therefore the public health benefits.
35. I believe Sir Iain Chalmers, co-founder of the Cochrane Collaboration, has made a
submission on how he has been told this problem is being fixed for three decades now. There
are many who use the reassuring language of “engagement” on missing trials, but act
inconsistently. We should not lose momentum on this important public health issue.
36. Declaration of interests:
37. I am currently a Research Fellow in Epidemiology at London School of Hygiene and Tropical
Medicine. I earn income as a doctor, academic, writer and broadcaster. In my work I discuss
problems in science, including publication bias, which is a major theme in my book Bad
Pharma. I am a co-founder of alltrials.net with the BMJ, Oxford University Centre for
Evidence Based Medicine, the James Lind Library and Sense About Science. Alltrials.net is a non-profit campaign group to improve access to clinical trial results with extremely broad