daniel buenoWrong or fraudulent scientific articles are usually and exemplarily removed from the archives of the scientific journals that published them, but sometimes copies survive in university libraries or storage and consequently may continue to circulate, being cited by researchers who are unaware of the problem. The US researcher Philip Davis found traces of what he called “the secret life of retracted articles” when he analyzed what happened to 1,779 papers disqualified by the journals that had published them between 1973 and 2010.
He arrived at this set of proscribed articles when he researched the Medline database of the United States National Library of Medicine. His next step was to look for the records of these papers in other portals or web repositories. Davis managed to locate versions of 321 of these articles – one out of every five in the sample – lost in virtual libraries or in the archives of universities and departments. Nowhere was there any warning about the fact that the article had been disqualified.
In almost all the cases (95%), the version encountered was that of the scientific journal. Only 4% were versions produced by the author prior to the article being submitted for publication. The place with the largest number of retracted articles was the PubMed Central database, with 43% of the total (138 articles). Ninety-four (or 29%) were found in academic domains, such as laboratory or department websites; and only 10% were in the repository of institutions. He also found 24 articles (4%) in commercial websites. The odd thing is that the articles were being used to promote food supplements or surgical techniques. Records of these articles were found in the Mendeley academic social network and they had been shared, on average, by 3.4 users.
Davis suggested that scientific journals should disseminate warnings about the status of such articles in search and recovery services and that the databases should post retraction notices to the bibliographic references of papers. Another measure would be to inform virtual libraries and online bibliography organization tools about the punishment. Finally, scientific publications should sweep the bibliographic references of all their articles before they are published to avoid having papers that were removed being cited again, said Davis. The study was released by the Journal of the Medical Library Association, of the United States.Republish