Concordance between EMA Good Clinical Practice inspections and medical literature concerning drugs that have not received marketing authorization in the European Union: a meta-research survey

Abstract Objectives To describe the differences between European Medicine Agency (EMA) Good Clinical Practice (GCP) inspection reports and medical literature on drugs with withdrawn or refused applications. Design A retrospective study comparing studies included in European Public Assessment Reports (EPARs) and the corresponding articles published in the medical literature. Data sources We screened all EPARs released by the EMA from inception to April 2024 for drugs that were refused or had a withdrawn application. In those EPARs, we looked for mentions of GCP inspections and details about them, then we searched for related publications on the inspected studies on bibliographic databases. Eligibility criteria All EPARs mentioning good clinical practice inspection were included in this survey. Data extracted Two reviewers independently gathered information on the GCP inspections and their findings, including the EMA’s opinion on their impact on the study data. The reviewers checked related publications for mentions of the inspection and any subsequent correction, retraction, or expressions of concern related to its findings. Main outcome measures The main outcome was the mention of the GCP inspection findings in the publication of the inspected studies. We also assessed whether there was any mention of these findings in a correction, retraction, or expression of concern. Results Out of 285 EPARs screened, 57 (20%) mentioned a GCP inspection. 58 distinct studies with inspections had 61 publications. For 17 publications the inspection occurred after the publication, for 20 the inspection happened before the publication, and for 24 the date of the inspection was unknown. Only 1 publication (2%) addressed the inspection findings. Moreover, there were no corrections, retractions, or expressions of concern related to inspection findings. Among the 61 publications, 26 (43%) were related with 24 distinct studies that had an inspection that casted doubts on data reliability, but none mentioned the inspections at or after the time of publication. Conclusions This meta-research survey indicates that health authorities’ GCP inspections are not reflected in the published literature, even when the inspections have put the data reliability in doubt. Journals should clearly specify which aspects of those studies are trustworthy and which ones are not. Trial registration osf.io/pa9fq/