NEW YORK — Fabricated data is the main reason biomedical and life science research articles are retracted, not error based on honest mistakes, a review of the research has found.
About 43 percent of the articles were rescinded because of made-up or manipulated data, 14 percent were retracted because the study appeared in more than one publication and about 10 percent because of plagiarism, according to the report published in the Proceedings of the National Academy of Sciences. The study is the most comprehensive analysis of retracted articles, said Arturo Casadevall, the study’s senior author.
Published studies that are withdrawn can have a negative effect on society and the public’s trust, Casadevall said. A 1998 study published in the Lancet medical journal linked a childhood vaccine for measles, mumps and rubella to autism and bowel disease and caused immunization rates in Britain to plummet. The journal retracted the study in 2010 after an investigation found the research to be flawed.
“We need to work very hard to identify what are the pressures that are leading to these kinds of behavior and we need to very rapidly find ways to clean up our act,” said Casadevall, chairman of the Department of Microbiology and Immunology at the Albert Einstein College of Medicine at Yeshiva University in New York. Still, “this is not something that’s endemic. This is a small aspect of the publication process.”
The number of retractions for fraud, though relatively small considering the thousands of papers published each year, has risen about 10-fold since 1975, the authors said.
The peak year for retractions thus far is 2007, with 83 papers withdrawn out of 867,700, or 0.00957 percent, according to Ferric Fang, the study’s lead author and professor of laboratory medicine and microbiology at the University of Washington School of Medicine in Seattle. That would be a 9.9-fold increase from 1976, when there were three retractions out of 309,800 papers, or 0.00097 percent.
The researchers reviewed 2,047 biomedical and life-science research articles that were retracted, with the earliest being published in 1973 and retracted in 1977. They found that only 21 percent of the retractions were because of error, while 67 percent were because of misconduct. The United States, Germany, Japan and China accounted for about 75 percent of all retractions because of fraud or suspected fraud.
Casadevall said the retractions might be occurring because of a rewards system that gives grants, jobs and prizes to scientists who publish the most in prestigious journals. Some researchers may be cutting corners to have their work appear in those publications, he said. Solutions to combat the problem include checklists, an enhanced focus on ethics and the establishment of uniform guidelines for retractions and retraction notices, the authors said.