Launched in 2010 by American science journalists Adam Marcus and Ivan Oransky, the Retraction Watch website has become a valuable source of information on cases of scientific misconduct from around the world. The website focuses on the retraction of scientific articles, when papers are withdrawn after publication due to the discovery of fraud or errors. Now, the information it provides is being used as a basis for studies into scientific literature. An article published in the journal Scientometrics in January analyzed 1,623 retractions made between 2013 and 2015 based on parameters such as country of origin, field of knowledge, and reason for retraction. The research was coordinated by Sonia Vasconcelos, a researcher at the Federal University of Rio de Janeiro (UFRJ), and coauthored by her master’s student Mariana Ribeiro, who is studying Biosciences Education, Management, and Dissemination at the Leopoldo de Meis Institute of Medical Biochemistry.
The researchers found that retractions originating from 71 countries were announced on the website, but they focused their analysis on the 15 nations responsible for the most cases (85% of the total—see table), which generally correspond to the most productive countries in terms of scientific output. Brazil is not included in this list, despite placing 13th in the 2015 ranking of nations with the highest number of articles indexed in the SCImago Journal & Country Rank database. According to Sonia Vasconcelos, Brazil’s absence is probably due to the fact that the number of Brazilian articles shared on PubMed, a biomedical database, is lower than many other countries.
The study found that biomedical, medical, and clinical sciences account for more than 60% of total retractions between 2013 and 2015. The hard sciences, such as physics, accounted for 17% of retractions; multidisciplinary fields accounted for 12%; and human and social sciences, 8%. Vasconcelos points out that Retraction Watch tends to publish more news on cases in the biomedical sciences because the creators of the website are more familiar with these fields. “We cannot ignore the fact that historically, biomedical areas have been the most active in terms of scientific integrity and research ethics,” she says. There were a smaller number of retractions in the human and social sciences (118) than other fields, but 58% of them were related to cases of scientific misconduct, the highest proportion of all the disciplines examined. Vasconcelos believes this can be explained in part by the existence of more repeat offenders in the field, each with several articles retracted due to ethical issues.
The researchers assigned each paper featured on Retraction Watch to one country only, according to the nationality of the lead author. In total, the USA and China accounted for 41% of all retractions, with 376 and 283 retracted papers, respectively. The USA also had the largest number of retractions due to misconduct (225), followed by Japan (75) and India (61). A peculiar case was observed in Japan: just two authors—endocrinologist Shigeaki Kato, a former researcher at the University of Tokyo, and surgeon Yoshitaka Fujii, from the University of Toho—were responsible for 28 and 20 retractions respectively, representing 48% of all Japanese retractions in the period analyzed.
Vasconcelos notes that Retraction Watch features only a fraction of all retractions, taken from the most prestigious periodicals indexed in international databases. In an effort to extend its reach, the site last year created a comprehensive database, with greater coverage than is provided by its news archive. “This database will help broaden our understanding of corrections to scientific literature and will allow us to expand upon the article published in January,” says Vasconcelos. In August 2017, the Retraction Watch team announced that the database had already cataloged approximately 8,000 retractions. This number is expected to reach 15,000 in the coming months.
The authors of the Scientometrics study observed that 47% of retractions resulted from more serious forms of scientific misconduct, such as fraud and data manipulation, while errors for which there was no evidence of misconduct were responsible for 11% of retractions. They also identified a number of questionable practices that do not always characterize misconduct. “Some of them undermine the reliability of the article, such as disingenuous attribution of authorship and undeclared conflicts of interest,” explains Mariana Ribeiro.
The authors warn that the results may be biased. “The predominance of misconduct cases may be explained by the fact that Retraction Watch discusses more cases motivated by serious or intentional failures because they attract more attention from readers,” says Vasconcelos. Even so, she says, the study reflects the findings of previous research. Another study, published in Scientometrics in 2013, showed that the percentage of retractions involving allegations of misconduct increased from 55% in 2007 to 71% in 2010.
Although Brazil was not included in this analysis, a paper coauthored by Vasconcelos and published in Science and Engineering Ethics in 2016 showed that the number cases of scientific misconduct involving Brazilian authors had increased in recent years (see Pesquisa FAPESP, issue no. 240). The study analyzed more than 2,000 papers indexed in the SciELO online library and the Latin American Health Sciences Information database (LILACS) between 2009 and 2014, identifying an increase in the number of retractions in both: between 2004 and 2009, one or two retractions were issued per year; between 2011 and 2012, the average rose to seven. Plagiarism was the main reason for the retraction of Brazilian articles, accounting for 46% of the total.Republish