At the NASW annual conference this past weekend, I happened to sit next to a scientist who studies ethical violations in scientific papers. Over lunch, Harold (Skip) Garner, a professor of biochemistry and internal medicine at the Unviersity of Texas Southwestern Medical Center, told me and the other folks at our table that research shows an increase in the incidence of data manipulation in published papers, ranging from manipulating results on submitted manuscripts and images to plagiarism, duplication and fabrication.
I've blogged about this problem before here, and it is, of course, not news to the many medical journals who have had to invest in sophisticated software to unearth such shenanigans. (It was with software like this that the editors of the New England Journal of Medicine discovered that the researchers who submitted a seminal trial of Vioxx to them in 2000 had erased data showing that the Cox-2 painkiller increased the risk of heart problems among patients. (Unfortunately, the journal only discovered that manipulation years later).
Even so, I was curious as to what Garner thought might be beyond this growth industry in cheating. Garner offered a two-fold reason. First, the opportunity is there: ever more sophisticated technologies allow scientists to cut and paste and slice and alter data submitted electronically. Second, as funding for NIH grants grows scarcer, there is greater competition and more pressure on scientists to publish and win those grants. Garner himself admits to such pressures. "We're always worried about getting the next grant," he said.
A third reason Garner didn't mention is the increasing commercialization of science and medicine, which has spawned an army of researchers on the take from health product companies. And a fourth reason is an appalling lack of enforcement by federal and state agencies that are supposed to monitoring scientific misconduct. According to The New York Times, a new Congressional report shows that the FDA takes years to investigate researchers accused of scientific fraud, which means that many of these scientists remain eligible to conduct research even after they have been convicted of fraud by local law enforcement.
Similarly, the Office of Research Integrity at the NIH only investigates one in 100 cases of suspected data manipulation and it rarely resolves the cases it does investigate. The commitment to policing scientific fraud should come from the top, but I haven't heard anything about this problem from Francis Collins, the new director of the NIH.
The NIH can certainly do far more to investigate allegations of fraud and restrict grants to labs and universities where there is evidence of irregularities. But it can't clean house by itself. Universities and academic institutional review boards (IRBs) have got to get a lot tougher in policing ongoing research studies and requiring the disclosure of financial conflicts of interest that tempt researchers into manipulating data in the first place.
Btw, I'm glad to see that Ed Silverman has Pharmalot up and running again, if only for news nuggets like this. You go Congresswoman McCollum!