Spike In Retractions Highlights Problems In Research

Oct 17, 2017

Across academia, faculty and graduate students perform all kinds of research. Their findings lead to new medicines, better understanding of space and time, or how the brain ticks. Universities brag about the economic impact their research has. Every year, for example, North Carolina schools bring in more than $3 billion of research funding.

When scientists make discoveries, they publish their work in academic journals. These include prestigious publications like Nature or the New England Journal of Medicine. Other scientists build on these journal publications to make even more discoveries that advance human knowledge.

For the most part, these scientists act ethically. They present the data as they find it. They offer warnings about what the discoveries don't show, or limitations of the findings.

In some cases, however, researchers commit fraud. They edit their findings, or fabricate data to make their research look better.

Duke University, the prestigious institution in Durham, has recently dealt with two major fraud cases. Duke faces a whistleblower lawsuit that it used fraudulent research to receive $200 million of research funding. The lawsuit stems from 15 papers retractions tied to research by Erin Potts Kant, a former Duke scientist. In 2011, fraudulent research from Duke scientist Anil Potti led to a 60 Minutes expose.

These are extreme examples. But they are not the only examples. Across academia, researchers are seeing more of their work retracted. In some cases, journals have retracted entire issues. In the two decades from 1982 through 2002, there were only five retraction notices combined from Duke, NC State, UNC and Wake Forest. In the 15 years since, there have been a combined 93 retractions from those universities.

The increase in retractions has more to do with better watchdogs than with an increase in fraud. Now that almost all papers are posted online, there are more eyes on every journal article. Also, software to detect if images have been manipulated has given more sophisticated tools to editors to catch misconduct.

In a statement, Duke Health responded that it has dealt with faculty. Of the 58 retractions in the database since 1998, "41 are related to four situations that involved research misconduct, which Duke has investigated and closely cooperated with other investigating agencies," according to the statement. "The remaining cases appear to be the result of a variety of causes that aren’t necessarily related to research misconduct or the actions of Duke faculty or staff.”

The data come from the Retraction Watch database. Retraction Watch is a blog launched in 2010 by Ivan Oransky and Adam Marcus. They noticed there were hundreds of scientific journals publishing some 2 million academic papers every year. In some cases these papers were retracted. But the retraction notices were often buried or hidden out of sight. This bothered Oransky and Marcus, so they started tracking retractions.

"Some retractions are actually hidden completely out of sight," said Oransky. "Even if the retraction notice ran somewhere in the journal, if it's not connected to the paper that you're looking at, it's completely useless. It might as well be a tree falling in the forest. Nobody's hearing it. Nobody knows that it's been retracted."

It's important to note that retractions are the last resort. Journals also publish corrections, which in academia are called errata or corrigenda. They publish editor's notes or warnings, too.

Ivan Oransky
Credit Retraction Watch

"Retractions are the nuclear option of correction in the scientific literature," Said Oransky. "It means that you can't rely on the paper that you are looking at anymore."

Universities take scientific research very seriously. "N.C. State's research expenditures are in the $500 million range. That's a considerable amount that we are very proud of, but it also carries a lot of responsibilities," said Mladen Vouk, N.C. State associate vice chancellor of research development. "I think what we have had in the past was people who just do honest research, and they don't have any bad intentions. Mistakes are possible, yes. That'll happen."

If there is any intentional misconduct, however, Vouk says the university would come down on the faculty. "You can have just random errors. That's one extreme where people just make mistakes," said Vouk. "And then there are cases where people do – presumably deliberately – copy something that they shouldn't. They don't attribute results to someone that they should. They actually falsify the data, potentially. Those are bad cases. Those are the ones we have zero tolerance on. If it comes to our attention, we really do not tolerate that."

State of Things: Frank Stasio talks with Jason deBruyn about this story

Vouk said the university has taken steps to improve its quality control processes. That includes labs in which students share equipment with professionals. "And that actually helps in situations like this because sometimes mistakes happen simply because it's a graduate student that's learning how to use equipment," Vouk said. "It's not that we shouldn't teach graduate students how to use equipment – we should. But when you have professional results coming out – doing it in a shared facility which is professionally maintained and run – helps reduce these kinds of errors, for instance. So that's one of the big efforts."

Duke University archway
Credit Julie Schoonmaker / Duke University

Although there has been a dramatic spike in retraction notices, academia agrees there is actually less scientific misconduct happening. Policing of research fraud has gotten a lot better. And the peer review process has tightened.

"In many ways it's good news that there are more retractions. I know that sounds a little bit strange, but in many ways it's good because it means people are actually paying attention," said Oransky. "I don't really trust a newspaper that has never run a correction notice. And I probably don't trust a newspaper that runs a correction notice on every story either. But somewhere in the middle there is sort of, if you will, the right number of corrections, retractions. Whether it's a newspaper, whether it's a scientific journal. And we're sort of getting there with some journals that are really cleaning house. But it's very much a Herculean task of trying to clean out the stables. And I think the stables still need quite a fair amount of cleaning out."