A Call to Action from AllTrials

- 1 Comment

“Industry and academics must share all trial results: without this information, doctors are misled, and patients are harmed.” That’s Ben Goldacre summing up the motivation behind the campaign AllTrials, and yes, it’s just that simple.

Launched in January 2013, AllTrials is an initiative on behalf of Sense about Science, Bad Science, the BMJ, and others calling for practical steps from regulators, industry, funders, and publishers and learned societies in order to ensure all trials are registered and reported. By highlighting the value of negative results, the AllTrials initiative is part of a growing effort to eliminate publication bias toward positive results and complete the scientific record. AllTrials is not concerned with patient data, regulation of which differs by country and the discussion of which can become a Pandora’s Box no one wants to open, but with the registering and reporting of clinical trials. It wants out in the open three things: 1) Knowledge that a trial has been conducted, 2) A summary of the trial’s results, 3) Full details about the trial’s methods and results.

To start with, you might check out the interview Ben did with our magazine Biome on why we need the AllTrials campaign. I particularly enjoyed Ben going head-to-head with John Castellani of PhRMA, the pharmaceutical industry representative body in the US, in the BMJ. The idea is an old one and is included in the Declaration of Helsinki, the World Medical Association’s ethics statement on medical research involving people. That was adopted in 1964. We’ve been doing clinical trials since at least 1774. We haven’t moved quickly, one might say.

Last week, however, AllTrials released a detailed plan on how all clinical trials can be registered and reported—pushing these efforts forward with useful, practical directions. The guidelines released specify actions on behalf of regulators, researchers, funders, and publishers and societies. They’re brief and worth a read.

Although this plan points to critical improvements in registries and in regulation, what interests me most is the funder–publisher side of things. We all agree clinical trials need to be reported; now what should we be doing to ensure that happens?

When it comes to study and protocol registration, funders and publishers need to focus on enforcement and monitoring. We can shout all we want about encouraging authors to register their trials that we’re either funding or publishing, but without any real enforcement, we’ll at best be missing out on a significant portion of clinical trials. If we look at OA compliance rates for Wellcome Trust authors (only 55% last year), we can see that effective compliance only comes with enforcement.

In that same vein, AllTrials states: “It should be impossible to obtain funding for a trial, including funding from Government, or to sell a product, or to obtain permission to do a clinical trial, without proving registration.” Agreed. Now how to enforce this?

For Funders—in addition to funding applications explicitly stating that the trials will be registered and results reported, trial registration IDs should be requested and monitored, similar to the monitoring of OA publications through FundRef. If this does not happen, funds should be withheld. 

For Publishers—we should publish reports only from registered trials and request the trial number at submission. Journals should also “commit to making it clear on a trial report if previously undisclosed trials come to light after publication.” We already do this here at BioMed Central for our journal Trials. The trial registration number is included in the article and links out to the trial record in the given registry, and we allow authors to publish information and data that come to light after original research publication (for example, further analysis of results that brings to light new findings or new data elicited from existing information).

We do this through our Threaded Publications initiative, the aim of which is to clearly connect all published content relating to a trial, including the trial ID, in one secure place. Details of this initiative are more fully explained in our latest blog about Threaded Publications. To briefly summarise, it works through the CrossRef DOI (used across publishers) and the CrossMark tool (a tool for conveying additional information or metadata about an article, such as an updated version or corrigendum; see figure). For clinical trial publications, this allows a truly coherent record of the trial—connecting its registry ID with all publications on the trial.

 

Part of our job as publishers is to maintain this coherency in the research we publish. Digital publications demand a rethinking of how we do this; a simple reference list no longer suffices. We invite the clinical research community to help us do this by contributing to our document on the metadata included in the CrossMark tool.

More can be done, though. Yesterday, the BMJ announced that in order to better ensure authors were fully reporting their clinical trials, they would require authors to sign a publication transparency declaration upon submission. The declaration would confirm that no important aspects of the study were omitted and would hopefully encourage authors to think more carefully about their adherence to the relevant reporting guidelines. If authors were later found to have misrepresented their study by excluding information on critical aspects of the study or the like, this would be evidence of scientific misconduct, for which the author would have to account. The BMJ invites other publishers to support this initiative by implementing the same declaration and points publishers to the EQUATOR initiative.  

The AllTrials plans also demand that a summary of results of the clinical trial be publicly available where the trial was registered no longer than a year after the trial’s completion. Such results should include information on all outcomes measured and statistical analyses. This is a giant step from registration  and a declaration at submission. Registering a trial is fairly painless, but there is large attrition when it comes to fulfilling the reporting of results. Indeed, in 2012 an audit was conducted on the clinicaltrials.gov registry; only a fifth of those trials registered had reported results within a year of the trial completion. What can we do as funders and publishers to help combat this attrition?

For Funders—again, funds should be withheld until results are shared. For funding applications Funders should also require explicit statements that results will be made available. Perhaps as standard funding applications should also demand a reporting of past trial sharing. Shouldn’t researchers who have a history of sharing their results be rewarded?

For Publishers—journals should make clear that they will not prohibit authors from publishing trial reports when summary results are posted in a registry.

Finally, when a full report is produced for marketing authorisation or the like, this should be made publicly available. The detailed information such reports contain on methods, analysis, results, and conclusions of clinical trials offers valuable insight for doctors and other researchers. It is crucial that doctors have such reports on medicines when prescribing them in order to make a fully informed decision. Indeed, as Dr Richard Lehman laments, “It is a scandal that doctors like myself often prescribe treatments without knowing their true benefits and harms, because research evidence from human trials has been withheld. That means that over my 35 years as a GP, I have unintentionally spent large sums of NHS money on treatments that did not work, and some patients have suffered avoidable harm.”

As taxpayers and patients, any reader will find Dr Lehman’s statement a powerful motivator. Enforcement for the latter, however, must come from regulators and the Pharma Industry itself. Its realisation will take seeing this information as a matter of medical evidence rather than commercial confidentiality. ­­­­

As funders and publishers, we must do our part in taking responsibility for the sharing of trial data. AllTrials has clearly laid out what needs to happen. It’s time to get to work.

 

  • Christopher Southan

    This is a good move. However, as we outlined in http://www.ncbi.nlm.nih.gov/pubmed/23159359 an essential componant of clinical data transparency is that the molecular structure of the trial compounds should be available in public databases. This should be made conditional for publishing results. The benifits for open data mining are obvious.