Prêmio CBMM
Imprimir Republish

Good Practices

Insufficient care taken with review articles

009_BoasPraticas_247._altaVeridiana ScarpelliReview articles, those scientific works that organize data from existing literature in a given area of knowledge, are not always written as carefully they should be to avoid citing papers that contain inaccurate data or about which there is a suspicion of misconduct. This is the conclusion of a study by researchers from Switzerland, France and Germany, published in the journal BMJ Open. They examined 118 review articles published in 2013 in four medical journals: Annals of Internal Medicine, The British Medical Journal, The Journal of the American Medical Association and The Lancet.

Researchers investigated whether the authors had followed six procedures that are important for preventing inclusion of erroneous or fraudulent data in review articles. They concluded that in only half the works did the authors employ three or more of those procedures that can help detect problems.

These procedures are the following: checking results of the cited papers against gross data from clinical tests; contacting the people responsible for the articles so as to obtain access to unpublished results; evaluating whether similar papers have been published about the same finding; verifying whether sponsors interfered with the articles; analyzing possible conflicts of interest; and checking whether the research projects described in the papers had been approved by ethics committees. In order to determine whether these recommendations were adopted, researchers analyzed the review articles with an eye towards the methods used and contacted the authors to ask additional questions.

The study observed that 11 of the 118 review articles had not taken any of the procedures into consideration. In 79 (66%) of them, their authors had searched through raw data; in 73 (62%) they had contacted the authors of the original papers; in 81 (69%) they looked for duplicate articles, but in only 5 (4%) had they examined conflicts of interest. It was also found that in only three review articles (2.5%) had the reviewed studies been approved by ethics committees.

The researchers reported that few authors of review articles had raised questions about signs of possible misconduct that they had noticed during their reviews of literature. Seven authors admitted that they had included in their review some papers on which suspicion had been raised, such as signs of plagiarism or manipulation of images, but only two had included a warning in their review articles. “When review authors suspect misconduct in some article, it seems as though they don’t know what to do with that information,” says Nadia Elia, a researcher at the University of Geneva, in Switzerland and principal author of the study, in speaking to the Retraction Watch site.