From The Conversation blog, 27 August 2016. Words by Professor Jonathan Edwards, Professor Emeritus, University College London.
After nearly four months deliberation a tribunal has dismissed an appeal.PDF) made by Queen Mary University of London (QMUL) against a requirement to release data from the PACE trial – a trial that investigated treatments for chronic fatigue syndrome. This is an important step towards the increased transparency we need for balanced analysis of clinical trials where the methodology is questionable.
Chronic fatigue syndrome (CFS), otherwise known as myalgic encephalomyelitis, is a disabling condition affecting about one in 500 people. We know almost nothing about its causes, but in some cases it appears to follow infection or involve disturbances to the immune system. The key feature of the syndrome is an intolerance of physical exertion with a worsening of fatigue, sometimes a considerable time after exertion, often referred to as a crash.
Without evidence for a mechanism, treatment is by trial and error. The PACE trial compared cognitive behavioural therapy (CBT), graded exercise therapy (which uses a progressive increase in exercise), self-pacing and standard care (chiefly a leaflet containing advice). PACE was large and participants were randomly assigned to treatment groups but it was unblinded and so not in a true sense controlled. Patients knew which were favoured treatments (CBT and graded exercise) and which were intended to be the dummies (pacing and standard care). The outcome was then measured by questionnaire.
An unblinded trial with a subjective measure for the outcome would not normally be acceptable in medicine and clinical pharmacology, but seems to be in psychiatry. Specifically, the trial was criticised on the basis that chosen measures of treatment success were changed midway through the trial and criteria for recovery overlapped with criteria for being ill enough to enter the trial.
The findings of the PACE trial were published in The Lancet in 2011. The authors claimed a benefit for CBT and graded exercise, but not for self-pacing. After its publication, several applications were made through Freedom of Information Act (FOI) legislation to review the raw data to allow re-analysis.
Patients and carers are concerned that graded exercise may make things worse for those with the condition and dislike the message that continued illness is due to false beliefs and unhelpful behaviours (such as avoiding activity), which seem to be the target of CBT. Both patients and scientists are concerned that the trial results were overstated as showing a useful effect of CBT and graded exercise. In particular, a follow up study by the PACE group was interpreted as showing continued benefit from these treatments when in fact it shows no difference between groups.
Until recently, access to data has been declined, but a recent request under the FOI by Alem Matthees, a CFS sufferer from East Perth, Australia, was approved by the UK’s information commissioner. QMUL appealed but their appeal has now been dismissed by the information rights tribunal.
GOOD SCIENCE DOESN’T NEED PROTECTION
The detail of the tribunal report is interesting; it reveals the complexity of data protection and the arguments on both sides. QMUL claimed that releasing data would breach patient confidentiality, might jeopardise follow up studies and damage the reputation of the college. Matthees pointed out, and the commissioner agreed, that since the QMUL team had already made the data available to chosen colleagues the first argument was unconvincing. It was unclear why access by other researchers to the data would threaten follow up studies. Matthees argued that QMUL would be more likely to damage its reputation by losing patient trust.
The core of the dispute appears to be that the QMUL authors feel they are entitled to protect their study from those who might wish to discredit their interpretation. Hopefully, we are moving towards a consensus that there is no such entitlement. If scientific interpretation is poor it deserves no protection. If it is good it needs none.
The essence of the response from the information commissioner was that the QMUL authors had overstated their case in terms of likely damage from release of personal data. The authors and one witness painted a picture of irresponsible animal rights-type activists who might deliberately target trial patients as part of a political smear campaign. Essentially the authors were claiming that because those asking for data were patients or others associated with them they should not be considered responsible enough to have access to anonymised data. In the end they had to admit they had no grounds for this view.
From my perspective the importance of this decision is in emphasising that everyone has the right to access to scientific data and to express their opinion. Confidentiality of patient identity is essential but should not be used as a smokescreen. If scientists knew when they entered data into workbooks or spreadsheets that everyone would be entitled to scrutinise them the quality of data collection and interpretation might improve a great deal.