Missing homeopathy trial data ‘lets misinformation thrive’

Almost 40 per cent of registered homeopathy trials are going unpublished while most published, experiments are unregistered with authorities, says a new study which warns that the practices help to “substantially overestimate” the effectiveness of alternative medicine.

Based on a review of clinical trial registries run by the US, the European Union and the World Health Organisation, researchers in Austria and the US found the results of 38 per cent of homeopathy trials registered since 2002 – 90 in total – were never published, according to a new paper published in the journal BMJ Evidence-Based Medicine. Of these published experiments, 25 per cent altered or changed the primary outcomes in publication, it adds.

The paper also found some 53 per cent of published homeopathy trials were unregistered, despite the 2005 policy of the International Committee of Medical Journal Editors that the journals it oversees should only publish the results of clinical trials that have been prospectively recorded in a public registry.

This kind of selective publication, combined with the release of potentially less rigorous pre-registered trial data, had wrongly allowed meta-analyses to indicate evidence for homeopathic treatments was more robust than it really was, according to the study’s lead author, Gerald Gartlehner, from the department for evidence-based medicine and evaluation at Danube University Krems, in Austria.

“Homeopathy is a huge market – billions of euros are spent on it each year – and operates completely outside regulatory processes, yet, at the same time, there are many trials, some of which claim lots of things about its effectiveness,” he told Times Higher Education.

“There are three meta-analyses which are most often cited – since most of the time, negative results aren’t published, you are likely to be missing a large body of evidence showing homeopathy doesn’t work,” said Professor Gartlehner, who called it a “type of cherry-picking”.

The scale of the problem is likely to be even worse than the paper, titled “Assessing the magnitude of reporting bias in trials of homeopathy: a cross-sectional study and meta-analysis”, indicates, given the studies that the analysis was unable to capture, said Professor Gartlehner. “There is probably a large, dark area of unpublished and unregistered studies that we will never know about,” he added.

Non-reporting of clinical trial findings in general remained a problem despite tougher rules from the EU and the US, where fines of up to $10,000 (£7,615) can be imposed for non-publication, said Professor Gartlehner.

To combat the issue, research funders should take a much tougher line on those who withhold results, recommended Professor Gartlehner. “If you are a public funder of clinical research and you have an author scientist that takes money to conduct trials and does not publish the results, you should put them on a blacklist and never fund them again,” he said.

jack.grove@timeshighereducation.com