June 19, 2008
Study Casts Doubt On U.S. Research Integrity
A new study has found that misconduct at research institutions within the United States may be more common than previously believed.
A survey conducted in 2006 of 2,212 mainly biomedical scientists at 605 universities and other research institutions showed that 9 percent reported they had personally witnessed fabrication, plagiarism or falsification. Many were hesitant to report the misconduct, the survey found.Thirty-seven percent of alleged misconduct cases were never reported to the institution involved for review, perhaps out of fear of reprisals for reporting a colleague or a desire to protect the flow of research funding.
"There's more misconduct, or potential for misconduct, out there than probably anyone has appreciated before. And a good part of that goes unreported," James Wells, who helped conduct the survey, said in a Reuters telephone interview. Wells serves as director of the Office of Research Policy at the University of Wisconsin.
"Usually what happens is that somebody very close to the research has to observe this going on. And they have to step forward and report it to their institution in order for something to happen. And they can very often be jeopardizing themselves," said Wells, who conducted the survey with two experts from the U.S. Health and Human Services Department's Office of Research Integrity.
The study's findings highlight concern among members of Congress and others about the integrity of research in the United States and abroad. There is skepticism about potential financial conflicts of interest by scientists who get paid by drug companies, and of study results potentially warped by the influence of pharmaceutical industry research funding.
U.S. Sen. Charles Grassley, for example, accused well-known Harvard University psychiatrist Dr. Joseph Biederman and others of failing to fully disclose payments received from pharmaceutical companies.
Wells said his survey did not look specifically examine such financial conflicts of interest. Instead, it queried scientists about whether they had witnessed acts that would meet the government's definition of research misconduct, which includes plagiarism, fabrication or falsification in conducting research, reporting findings or seeking grants.
In total, 192 scientists, or 8.7 percent, said they observed or had direct evidence of researchers in their own departments committing such misconduct over the previous three academic years, reporting a total of 265 incidents of misconduct. But Wells and his team then evaluated the allegations and determined that some did not meet the federal misconduct definition. Allowing for these, that left 201 incidents of misconduct reported by 164 scientists, or 7.4 percent of the respondents. The findings suggest that more than 2,300 annual cases of misconduct may be occurring at U.S. research institutions.
The survey respondents cited examples of misconduct that included altering data to "improve" findings, submitting false data to obtain a grant and misrepresenting findings.
In a report about the study, Wells and his colleagues wrote that the HHS research integrity office receives only about two dozen reports each year of research misconduct, a mere "tip of the iceberg".
"It's really the universities' responsibility to police this. And as we've seen in the (financial) conflict-of-interest field, they do a very poor job," Merrill Goozner, who leads the Integrity in Science Project at the Center for Science in the Public Interest, told Reuters.
Wells said it was hard to determine whether the problem was getting worse, or whether it is any worse in the U.S. than elsewhere, since there have not been any previous comparable national studies regarding U.S. research integrity.
One of the most well known examples of research fraud in recent years involved South Korean researcher Hwang Woo-suk, who in 2006 admitted to fabricating stem cell data.
The study was published Wednesday in the journal Nature. The full report can be viewed at http://www.nature.com/nature/journal/v453/n7198/full/453980a.html.
On the Net: