Analysis of retracted articles on medicines administered to humans

Abstract

Retraction of an article is the public recognition that something went wrong.1 Retractions can be due to honest errors by authors or journals. However, most of them translate a misbehaviour by the authors in the conduct or report of the study. In 2047 retracted biomedical and life sciences papers indexed by PubMed as of May 2012, research misconduct (ie, fabrication, falsification, and plagiarism)2—or research fraud3—was the reason behind 53% of all retractions, whereas duplicate publication and errors were responsible for 14% and 21%, respectively.4 Since October 2018, the information of the Retraction Watch database (RWdb) is publicly available.5 This is the largest and most comprehensive database worldwide on retracted articles in all disciplines.6 In the RWdb, there are six authors (J. Boldt, Y. Fujii, J. Iwamoto, S.S. Reuben, Y. Saitoh, and Y. Sato) with multiple retracted papers—all for research misconduct and many other misbehaviours—that extensively worked on the research of medicines in humans. The aim of this analysis was to describe the main reasons for retraction of articles describing the administration of medicines to humans. The search was done on 17-20 May 2019 aiming to capture all retracted articles that were originally published between 1 January 1970 and 31 December 2018. The search used the following descriptors of the RWdb: Subject: “Medicine-pharmacology” and Nature of notice: “retraction.” Consecutive searches were conducted with three different Article types: “Clinical study,” “Research article,” and “Meta-analysis.” Entries excluded were the following: those authored by the six researchers mentioned above, to prevent distortion of the whole analysis; in vitro, animal studies and all others that were not clinical trials, observational research, PK studies, and meta-analysis. The searches yielded 655 retracted articles; 485 were excluded: 148 authored by one or more of the six fraudsters mentioned above, 194 were in vitro studies, 69 animal studies, seven duplications, and 67 misclassified retracted articles (eg, educational, microbiology, and review article). The information retrieved from the 170 retracted articles was that included in Table 1. Retracted papers were published between May 1975 and November 2018. With regard to the medical specialty involved, and to ensure a fine tune classification, the specialty that seemed more relevant to the topic addressed was chosen from the two or more that were included for each article on the RWdb. Most articles were retracted for more than one reason. However, falsification/fabrication and plagiarism were usually mutually exclusive, except for two articles wrote by authors from China. Hence, research misconduct was involved in 22 (13%) retracted articles. On the other hand, retracted articles that were due to honest errors (“errors without other issues”) accounted for 11% of all. Journal errors—ie, publication of duplicate articles—were very rare (2%). So, taken into consideration those retracted articles with limited or no information (10%), authors were found to be consciously and/or recklessly responsible for their misbehaviour in 131 (77%) retracted papers. Twenty-nine per cent of retracted articles were subject to investigation by the journal, institution/company, US Office for Research Integrity, or a third party. Oncology and cardiovascular accounted for 37% of all retraced papers of 19 different medical specialties and dentistry. Authors worked in 43 different countries of the five continents; those from USA and China were involved in 45% of all retracted papers. Only three retracted articles were replaced: one cohort study published in JAMA Oncology, one placebo-controlled randomized trial published in Journal of Endourology, and, finally, a meta-analysis published in Lancet; the three were published in 2017-2018. The main limitations of this study were the limited number of retracted papers included and that all the information was captured from a database that has its own peculiarities. So, for instance, the classification of the “types of article” could be improved. Thus, there were nine in vitro studies classified as “clinical study,” and 67 misclassified retracted articles included as “research article.” This, however, is a common issue in science.7 Finally, some retracted articles have been missed since, most likely, not all retractions were hosted on the RWdb; however, this does not invalidate the study findings. Further studies are warranted to have a clearer and more precise picture of retracted articles on the use of medicines in humans. Yet this study helps to know which reasons for retraction were common—once retractions by the six frequent fraudsters have been excluded. In any case, honest errors were more frequent in biomedical research retracted articles (21%)4 that found in this study (13%), which is of concern. The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request) and declare no support from any organization for the submitted work and no other relationships or activities that could appear to have influenced the submitted work.

Ähnliche Arbeiten