Abstract
Following recent revelations about drug companies withholding clinical trial evidence from drug regulators which provide licences for the sale of drugs and, despite there being heavy regulations in this area, the BMJ has launched a consistent and sustained campaign for more open data with a focus on the following drugs: Tamiflu and Relenza, both to treat influenza, Incretin mimetics and Rosiglitazone (Avandia), both to treat type 2 diabetes, and Rofecoxib (Vioxx) for osteoarthritis.
The case for regulators having access to unbiased evidence seems self-evident. Without it, they cannot possibly discharge their substantive function to protect public health. With some success, a team of scientists, which includes some of the original investigators of Tamiflu sponsored by Roche, will undertake a meta-analysis of both published and unpublished data. In addition, the Cochrane Neuraminidase Review Group have gained access to unpublished trials to update their own analysis from 2009 (Doshi, 2009; Jefferson et al., 2009). However, it remains the case that the majority of Roche’s 74 trials remain unpublished over a decade after completion (Cohen, 2009; Smith, 2009), WHO recommends Tamiflu, but has not vetted the Tamiflu data, EMA approved Tamiflu, but did not review the full Tamiflu dataset, and the CDC and ECDC encourage the use and stockpiling of Tamiflu, but did not vet the Tamiflu data (BMJ, 2013).
Peter Gøtzsche, Director of the Nordic Cochrane Centre, however, wants more open data and, in 2010, he sought a ruling from the European Ombudsman on the right of the public to scrutinize those data (BMJ, 2013). The devil seems now definitely in the detail. AbbVie and InterMune have recently obtained an interim ruling from the General Court of the European Union to prevent the European Medicines Agency from data sharing on the grounds that the patient level data requested ‘did not meaningfully contribute to the scientific review or evaluation of our products‘ (BMJ, 2013).
Where does this leave us? We could ask whether it is factually true that providing patient level data is unnecessary for the scientific review of a new product. A full scientific or peer review may require at least the ability to repeat the analysis undertaken, oftentimes by an in-house statistician, to corroborate the aggregate result. Indeed, the BMJ will not now accept papers for publication without a commitment to make patient level data available for independent analysis. However, the legal case rests on public access of patient level data and thus includes providing data for lay consumption. A review of the risks and expected benefits of a new drug may not require potentially overwhelming sets of numbers without thorough analysis and interpretation. These functions, however, are not mutually exclusive.
We must then ask why the companies seem so reluctant to make patient level data more open. There is a cynical and a not-so-cynical response. The cynical response is that commercial interests bias the analysis in ways the companies do not want revealed in this direct way (although any bias may still be revealed in other ways such the tendency for companies to publish positive results). The more generous response is to suggest that the issue of patient confidentiality cannot successfully be resolved through anonymization. We know that there have been cases where individuals have been identified despite patient-level data seemingly being anonymized (Gymrek et al., 2013). However, although this is indeed an important and substantive issue, it is one the companies have strangely neglected and the European Court not raised in its ruling. It would, though, be curious for companies not to pursue a more robust line in research integrity and leave the responsibility for regulating the ethical concerns over confidentiality and public rights to access data to the European Medicines Agency. As the ban on the public sharing of patient level data is only an interim ruling, the European Court still has the chance to grasp the nettle and provide governance in its etymological Greek sense of charting a course.
Editor’s choice
In this issue, Ann Cook timely picks up the issue of privacy in the context of clinical trials in ‘The truth about the truth: What matters when privacy and anonymity can no longer be promised to those who participate in clinical trial research?’, while Mollie Gerver argues that blanket anonymity is not in any case always appropriate in her article, ‘Exceptions to blanket anonymity for the publication of interviews with refugees: African refugees in Israel as a case study’. What seems critical in both articles is the basis on which consent to participate was granted. If privacy is indeed merely a question of individual control over personal data (and thus of consent), then this conclusion seems correct. However, there are other important concerns which place limits on how well consent can protect privacy interests and provide a good case for additional regulatory protection (Edwards, 2012).
Our review article in this issue is on the topic of non-inferiority trials and offers a fairly long and sometimes technical read. Marco Annoni’s article, ‘The ethics of inferiority trials: A consequentialist analysis’, provides an excellent background for those interested in placebo controlled trials and consequentialist thinking in applied ethics more generally.
To round off, there is a new case study, ‘Vaccine for the Norovirus’, and a commentary on the last study, ‘The missing kitten’, both written by Georgia Testa.