NPC Archive Item: Publication bias identified for drugs receiving U.S licensing approval

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
On The Horizon Rapid Review NPC Logo

North American authors of an observational study have found evidence of incomplete and selective publication of clinical trials that supported successful licence applications to the FDA during the time period 1998 to 2000. In fact, over half of related studies remained unpublished five or more years after approval of a drug.

Readers can obtain detailed information on each drug approved by the European licensing agency (EMEA) from the European Public Assessment report (EPAR) that accompanies market authorisation. The EPAR contains brief critiques of the trials submitted for licensing purposes.

Details of trials that are to support an FDA application, and that may be used for other licensing agencies, can be found at

What is the background to this?
In the United States (U.S) it is the responsibility of the Food and Drug Administration (FDA) to approve new medicines, based on results from clinical investigations. The FDA relies on sponsors (e.g. manufacturers) to submit all the data, including data from failed trials. Publication in peer-reviewed journals is the main way in which trial results are communicated to clinicians and made available to the public. This is vital for informed clinical decision-making. Public attention has been drawn to the selective publication of trial results by controversies over drugs such as rofecoxib and rosiglitazone.

What does this study claim?
The authors found that 57% of trials supporting FDA-approved drugs remained unpublished five or more years after approval. However, so-called “pivotal trials”, larger studies and trials with statistically significant results were more likely to be published. For trials published within five years of FDA approval, likelihood of publication correlated with:

(a) pivotal status (odds ratio [OR] 5.31, 95% confidence interval [CI] 3.30-8.55),

(b) the involvement of more than 135 participants (OR 1.33, 95% CI 1.17-1.52) and

(c) statistically significant results (OR 3.03, 95% CI 1.78-5.17).

How does this relate to other studies?
Other researchers have demonstrated publication bias with trials for selective serotonin reuptake inhibitors (SSRIs), antidepressants and for paediatrics. This current study has demonstrated that publication bias is common across a diverse group of drug classes.

So what?
Publication bias may lead to inaccurate conclusions being drawn about new drugs. In the past, there was no requirement for trial results to be published. This study covered drugs licensed up to the year 2000 and developments have occurred in the reporting of trials since this time. In the U.S, new legislation has been introduced to improve the accuracy and completeness of such reporting.  It is mandatory to register all trials supporting FDA-approved drugs on the website and to post basic results. Those involved in making local decisions about recently licensed medicines may wish to check the EPAR for brief critiques of the related studies. Websites such as may also be useful for this purpose.

Study details

Lee K, Bacchetti P, Sim I. Publication of Clinical Trials Supporting Successful New Drug Applications: A Literature Analysis.
PLoS Med 5(9): e191. doi:10.1371/journal.pmed.0050191

Design: cohort study of 909 trials for 90 FDA-approved drugs (new drug entities approved between January 1998 and December 2000). Researchers identified clinical trials from PubMed and other databases up to 1 August 2006. They then matched these trials with the lists of trials submitted to the FDA by sponsors of the drugs concerned (abstracts were not considered as matches).

Outcomes: main outcome measures were time from FDA approval to publication of a full report and whether a report was published within 2 or 5 years after approval.

Results: 394 (43%) of the trials submitted to the FDA by sponsors were matched to publications in the medical literature. There were no supporting trials published for one of the 90 drugs, an antibiotic. The proportion of trials published per new drug ranged from 0% to 100% with an average of 55% of supporting trials published per new drug. Five trials were published twice and one was published three times. Trials were more likely to be published within 2 years post drug approval if they had statistically significant results and larger sample sizes. 32% of the 394 published trials had a publication date which was ahead of the relevant drug being approved by the FDA. 92% of the 394 trials had been published within 3 years of FDA approval.

Sponsorship: United States Presidential Early Career Award for Scientists and Engineers and National Institutes of Health/National Center for Research Resources University of

California San Francisco–Clinical and Translational Science Institute.

Please comment on this blog in the NPC discussion rooms, or using our feedback form