WTF?… »
« v…
the devil’s still in the details…

Posted on Friday 27 September 2013

Following up on the last post, this is what the AllTrials campaign proposes [the document contains an Appendix with an outline of the specifics]:
September 2013

The AllTrials campaign calls for all past and present clinical trials to be registered and their results reported. Clinical trials are investigations designed to assess the effects – wanted and unwanted – of healthcare interventions in people. The Declaration of Helsinki, which is the World Medical Association’s statement of principles for medical research involving people, states that every investigator running a clinical trial should register it and report its results. Millions of volunteers have participated in clinical trials to help find out more about the effects of treatments on disease, yet that important ethical principle about reporting has been widely ignored. Information on what was done and what was found in these trials could be lost forever to doctors and researchers, leading to bad treatment decisions, missed opportunities for good medicine, and trials being repeated. This is what led to the AllTrials campaign in January 2013, a campaign which is now supported by thousands of individual patients, clinicians and researchers across the world, and by hundreds of organisations representing millions of people.

This document sets out more information about achieving a situation globally where all trials are registered and results reported. It is an achievement that will involve regulators and registries, clinical trial funders, universities and institutes, professional and learned societies and medical journals, patients and researchers.

This document is part of a continuing discussion which many different organisations are working on elaborating further over coming weeks and months.
Please email views and contributions to:

What trial information needs to be registered and reported?
There are four levels of information in clinical trial reporting:
  1. knowledge that a trial has been conducted, from a clinical trials register;
  2. a brief summary of the trial’s results;
  3. full details about the trial’s methods and results;
  4. individual patient data from the trial.
The AllTrials campaign is concerned with the first three. There are now initiatives in many countries to work out how individual patient data can be shared with other researchers.
  1. Registration:
    Planned clinical trials should be registered, with a summary of the trial protocol, before the first participant is recruited. Past trials that were not registered should now be registered retrospectively. This is essential if the trial was on medicines or interventions that we currently use [this includes some trials conducted before registries were established].
  2. Summary results reporting:
    A summary of results should be publicly available where the trial was registered , within o ne year of completion of the trial. Summary results from all past trials of medicines currently in use should be made publicly available on a register now. Summary results include information on the primary and any secondary outcomes measured and statistical analysis. This is part of the structured information that global registries should support.

  3. A full report:
    Trial sponsors or others who produce a full report for marketing authorisation or any other purpose should make this publicly available. The narrative reports of adverse events and individual patient data in a full report can be redacted and available on request to researchers, in the same way that reports of adverse incidents currently are, with a commitment that no reasonable request will be refused.
  4. Individual patient data:
    The AllTrials campaign is not calling for individual patient data to be made publicly available.
      There are currently initiatives in many countries looking at how to improve sharing of this level of information for the benefit of future research. This offers significant opportunities, such as: improving the accuracy of estimates of benefits from a treatment, through individual patient data meta-analyses; and identifying subgroups of patients who respond better, or worse, to a specific treatment. Patient groups, medical research funders and trialists have raised concerns about the inability to reuse past research. They are keen to develop consent protocols that will optimise the ability to reuse findings, and want legislators to look at whether new data protection regulations impose unnecessary burdens and restrictions on reuse of past research.
I’m not the one to parse things in small print. Registering and reporting AllTrials is great. Will there be enough in what is described here to thoroughly vet questionable studies? I don’t mind them cutting deals to get the job done, but if another Paxil Study 329 gets published, I want someone to be able to look at the data promptly. To be honest, I don’t understand the various initials well enough to tell, but I liked the wording from the committee in the last post [v…] better than number 4. above:
We are not in favour of placing anonymised individual patient-level data [IPD] in the public domain in an unrestricted manner, as the risk to patient confidentiality is too great. Instead, specific individuals should be provided with controlled access to IPD through carefully managed and secure “safe havens”. Access should be facilitated by an independent “gatekeeper”, responsible for ensuring that data is handled responsibly and in a way that makes a useful contribution to scientific knowledge.
This is no time to lose in the small print…
    September 27, 2013 | 6:37 PM

    I think caution is really justified here! These transparency proposals will rise or fall on the question of “patient-level data.” If all you have is a summary, then you never know that the 10 patients who “dropped out” or were “lost to follow-up” happened to be the ones getting worse … or that the 10 patients with “emotional lability” expressed their “lability” not by yelling at the nurses, but by trying to commit suicide.

    A story recorded on David Healy’s blog continues to haunt me — from a woman with Crohn’s disease who was in the initial trials of AbbVie’s drug Humira. When her colonoscopy indicated she was actually getting worse on Humira she was removed from the study and her doctor was informed that she had dropped out “voluntarily.”

    Healy has indicated that he fears both AllTrials and the MPs as well are being taken for a ride by EFPIA and GSK … and this is good evidence that he’s not just being persnickety. After all, who is this “independent” gatekeeper and how will they decide which inquiries are “responsible” and likely to make a “real contribution to science”? From what I understand, GSK have turned down requests from leading members of the Cochrane Collaboration, and if those guys are not “responsible” enough for them then they must be confusing “responsible” with “lapdog.” Sheesh.

    September 27, 2013 | 10:34 PM

    I don’t know how well alltrials is really prepared to address data manipulation. An assumption is made that unpublished replications show the deficits, and that those replications have been hidden on purpose for that reason.

    Without access to the full data it is hard to actually vet anything. Data manipulation is hard to address without full access to the FDAs full data. The below publication goes into some detail in the freely available full paper.

    “Publication Bias in Antipsychotic Trials: An Analysis of Efficacy Comparing the Published Literature to the US Food and Drug Administration Database”

    “Throughout the FDA reviews and journal articles, the primary analyses involved modified intention-to-treat methods [17] for handling dropouts: mixed-effects model repeated measures [18] was used for some of the iloperidone trials (in both FDA reviews and journal articles); last observation carried forward (LOCF) [19] was used for all other trials.”

    In the case you mentioned Johanna, that would be “last observation carried forward”. If the person can’t stay on the drug, it’s not effective. Deleting the dropout datapoints or pretending a benefit was found where there really *wasn’t* any benefit is certianly manipulating data.

    In any case, alltrials should make things better even if they can’t do everything at once.

    In the end it all comes down to who has the keys. The goverment is driven by ass-covering, and companies are driven by minimizing losses.

Sorry, the comment form is closed at this time.