Publication Bias in Antipsychotic Drug Trials
Turner EH, Knoepflmacher D, Shapley L
PLoS Med. 2012;9:e1001189
Turner and colleagues examined the US Food and Drug Administration (FDA) drug approval packages for 8 second-generation antipsychotics and found 24 FDA-registered premarketing trials. Four (17%) of the trials were unpublished; of these, 3 failed to show that the study drug had a statistical advantage over placebo and 1 showed that the study drug was statistically inferior to the active control drug. The remaining 20 studies (83%) were published in peer-reviewed journals.
Not evident from some of the published articles, but mentioned in the FDA review, was that active controls demonstrated superiority over the tested drug in some studies. However, in the end, the association between trial outcome and publication status did not reach statistical significance. The increase in the effect size point estimate as a result of publication bias was modest (8%) and not statistically significant. The investigators concluded that the magnitude of publication bias in antipsychotic trials was lower than that found previously in antidepressant trials, possibly because antipsychotics more consistently demonstrate superiority to placebo.
This study examined a narrow range of research that was completed purely for regulatory approval. Very few studies were unpublished; these may not have been submitted for publication because of the belief that negative and failed studies are of no interest to journals. The cynic would say that these studies were intentionally "buried." However, in the era of clinical trial registries, it is no longer possible for this behavior to go unnoticed.
Of value is the link provided to the drug approval packages that the study authors meticulously collected and posted in a digital repository at Oregon Health & Science University. These versions are an improvement over what you would download from the FDA directly because they have been assembled into PDFs, concatenated, and made searchable.
Abstract
Publication Bias in Antipsychotic Trials: An Analysis of Efficacy Comparing the Published Literature to the US Food and Drug Administration Database
Turner EH, Knoepflmacher D, Shapley L
PLoS Med. 2012;9:e1001189
Study Summary
Turner and colleagues examined the US Food and Drug Administration (FDA) drug approval packages for 8 second-generation antipsychotics and found 24 FDA-registered premarketing trials. Four (17%) of the trials were unpublished; of these, 3 failed to show that the study drug had a statistical advantage over placebo and 1 showed that the study drug was statistically inferior to the active control drug. The remaining 20 studies (83%) were published in peer-reviewed journals.
Not evident from some of the published articles, but mentioned in the FDA review, was that active controls demonstrated superiority over the tested drug in some studies. However, in the end, the association between trial outcome and publication status did not reach statistical significance. The increase in the effect size point estimate as a result of publication bias was modest (8%) and not statistically significant. The investigators concluded that the magnitude of publication bias in antipsychotic trials was lower than that found previously in antidepressant trials, possibly because antipsychotics more consistently demonstrate superiority to placebo.
Viewpoint
This study examined a narrow range of research that was completed purely for regulatory approval. Very few studies were unpublished; these may not have been submitted for publication because of the belief that negative and failed studies are of no interest to journals. The cynic would say that these studies were intentionally "buried." However, in the era of clinical trial registries, it is no longer possible for this behavior to go unnoticed.
Of value is the link provided to the drug approval packages that the study authors meticulously collected and posted in a digital repository at Oregon Health & Science University. These versions are an improvement over what you would download from the FDA directly because they have been assembled into PDFs, concatenated, and made searchable.
Abstract
Source...