Results of pivotal trials and statistically significant
results from larger trials more likely to be published.
The results of more than half of all supporting trials for FDA-approved
drugs remain unpublished five or more years after drug approval, according to
“Regardless of the cause, publication bias harms the public good by
impairing the ability of clinicians and patients to make informed clinical
decisions and the ability of scientists to design safer and more efficient
trials based on past findings,” the researchers wrote. The results were
published in PLoS Medicine.
Researchers from the University of California San Francisco conducted a
cohort study of trials supporting new drugs approved between 1998 and 2000.
They set out to determine the proportion of trials whose results were submitted
to the FDA in support of newly approved drugs that are published in biomedical
journals a clinician, consumer or policymaker might search.
“Our study shows that results from more than half of these trials
are not available in the medical literature, and results that show the drug
working are more likely to be published than results that show the drug not
working; this pattern of publishing makes it look like these drugs are better
than the company’s own studies would show in total,” Ida Sim, MD,
PhD, associate professor in the division of general internal medicine,
UCSF, told Endocrine Today.
There were 909 trials supporting 90 approved drugs in the FDA reviews;
43% of the studies had published results. Seventy-six percent of the subset of
trials described in the FDA-approved drug label and classified as pivotal
trials had published results, according to the study.
“Because medical literature (ie, journal publications) is the most
influential way to communicate trial results to the scientific community and to
the public, we need to be vigilant that the literature contains a fair
presentation of the total evidence for and against expensive new drugs,”
In multivariate logistic regression, publication correlated with
statically significant results (OR=2.96; 95% CI, 1.24-7.06) and larger sample
size (OR=1.47 per twofold increase in sample size; 95% CI, 1.15-1.88) for
pivotal trials five years post-approval, according to the findings. These were
predictive of publication at two years post-approval as well.
“Our data provide a baseline for evaluating publication bias as the
new FDA Amendments Act comes into force, mandating basic results reporting of
clinical trials,” the researchers wrote. – by Christen Haigh
PLoS Med. 2008;doi:10.1371/journal.pmed.0050191.
The authors do not quantify the tendency for editors to reject papers of insignificant difference and select papers with significant outcomes noted. Thus, some of the bias may have no nefarious intent on the part of the trial sponsor, but may merely reflect the realities of the journal publication process.
– Alan J. Garber, MD, PhD
Endocrine Today Chief Medical Editor
Likelihood of Publication in
Multivariable Logistic Regression for All Trials Five Years Post-approval
|Statistically significant results
||OR=3.03; 95% CI, 1.78-5.17
|Larger sample size
||OR=1.33 per twofold increase in sample size; 95% CI,
||OR=5.31; 95% CI, 3.30-8.55