Some university researchers still believe that if their clinical trial publishes its outcomes in a peer-reviewed journal, they do not also have to upload its summary results onto trial registries.
That is wrong. Here are the facts:
Best practices set out by the World Health Organization (WHO) require the results of all clinical trials to be uploaded onto a trial registry within that timeframe.
Posting results onto registries accelerates medical progress because the 12-month timeline permits far more rapid results sharing than the slow academic publication process allows.
Posting results onto registries minimises the risk of a trial never reporting its results and becoming research waste, which can happen when a principal investigator dies or leaves their post during the prolonged process of submitting an academic paper to a succession of medical journals.
Results posted on registries are easier to locate and are open access.
Research shows that trial results posted on registries typically give a more comprehensive and accurate picture of patient-relevant trial outcomes than corresponding journal articles do.
Registry reporting facilitates comparison of trial outcomes with a trial’s originally stated aims, and thus discourages harmful research malpractices such as the ‘silent’ suppression, addition, or switching of selected outcomes, HARKing, and p-hacking.
Results on trial registries enable the more rapid and reliable identification of potential safety risks posed by medicines already on the market.
There is plenty of research showing that publications in peer-reviewed journals often provide a misleading picture of trial results.
For example, ‘outcome switching’ is a form of evidence distortion that involves moving the goal posts of a trial after it has been completed. To use an analogy, researchers shoot their arrows first and then draw bull’s eyes around wherever the arrows have landed, making their archery look far more impressive than it really is. In essence, reporting successes in treating patients based on trial data using switched outcomes makes drugs and devices appear far more effective than they really are.
A team of researchers at the University of Oxford reviewed papers published in the world’s top five medical journals to find out how widespread outcome switching is. Many doctors rely on these journals for guidance on how to treat their patients. The Oxford team found that 58 out of 67 trial publications contained altered outcomes. In total, 354 pre-specified outcomes had not been reported, while 357 new outcomes had been silently added.
Only nine trials out of the 67 had been accurately reported in journals.
Meanwhile, a survey of 390 consulting biostatisticians in the United States recently found that statisticians are frequently asked by researchers to massage their data. The most frequently received requests for inappropriate analysis and reporting involved:
Stress only the significant findings, but underreport non-significant ones (55% of respondents had received such requests)
Report results before data have been cleaned and validated (44%)
Do not show plot because it did not show as strong an effect as you had hoped (42%)
Conduct too many post hoc tests, but purposefully do not adjust alpha levels to make results look more impressive than they really are (40%)
Results posted on clinical trial registries provide a far more reliable picture than journal articles do.
To a large extent, this is because trial registries force researchers to report results against the targets they set before the trial began, and also because registry managers perform quality control on the results submitted.
Please, universities, no more excuses. Post all your clinical trial results onto registries, without exceptions and without delay, as demanded by the World Health Organization.
Remember that uploading results onto a registry poses no barrier to subsequent publication in an academic journal.
The International Committee of Medical Journal Editors has explicitly stated that the posting of summary results onto trial registries is not considered prior publication by academic journals. Thus, academic journals will accept articles reporting a trial’s outcomes even if that trial’s outcomes have already been made public in a trial registry. Because results reporting on registries is typically faster than academic publication, making trial results public on registries before they are published in an academic journal is now the norm in best practice scientific communications. More details here.
Universities that want to improve their clinical trial transparency can use TranspariMED's collection of transparency tools to strengthen their policies, processes and performance.
To see what clinical trial results posted directly onto registries actually look like, check out the results for this breast cancer trial on the American registry Clinicaltrials.gov, and this kidney cancer trial on the European registry EUCTR.
The author would like to thank Rita Kessler of Revue Prescrire for drawing his attention to the survey of US biostatisticians cited above.