Authenticity of some published trials in question

More than 90% of a sample of randomised controlled trials (RCTs) published in Chinese journals between 1994 and 2005 did not adhere to recognised methodology for randomisation, according to a study published yesterday in Trials, casting doubt on the reliability of research that has the potential to influence medical decision-makers.

Wu and colleagues (Chinese Cochrane Centre at Sichuan University, China and Ottawa Hospital Research Institute) searched the China National Knowledge Infrastructure electronic database for reports published in the Chinese literature between January 1994 and June 2005, that were described by the authors as RCTs or claimed to have used random sequence generation or allocation concealment.

Telephone interviews with the first or co-authors of 2235 reports about randomisation methods and quality-control features of the trial indicated that only 6.8% of the studies be considered “authentic” RCTs. Although only 51.6% of trials supported by government or other official organizations were found to be authentic, all trials of pre-market drugs were identified as such. Wu et al. report that of the first-authors erroneously identifying their studies as RCTs, 85.6% did not fully understand the principles of randomisation, whilst 5.1% mislabelled their trials despite an understanding of the relevant methodology.


Randomized trials published in some Chinese journals: how many are randomized?
Taixiang Wu, Youping Li, Zhaoxiang Bian, Guanjian Liu, David Moher
Trials 2009, 10:46 (2 July 2009)
[Abstract] [Provisional PDF]

The misleading reporting of RCTs is likely a worldwide problem, but the investigators suggest a link between their results and the high proportion of positive trial results published in Chinese journals, noting that inadequate randomisation has been previously shown to result in more favourable estimates of treatment effects. They also highlight the potential for falsely reported RCTs to mislead healthcare providers and policy makers, and impact upon the findings of systematic reviews.

Wu et al. advocate improvements to the education of researchers in the principles of randomisation methodology and scientific reporting. In addition, they suggest that the development of peer review guidelines is needed to help identify poorly randomised studies before publication.

Victoria Thompson
Assistant Journal Development Editor – Trials

View the latest posts on the On Medicine homepage