Criticism of the US federal agency charged with investigating scientific fraud is growing as allegations of wrongdoing have reached an all-time high. However, fewer than one in five cases has led to a formal investigation and only a handful of cases has led to censure, writes Jon Marcus.
In 2004, the most recent year for which figures are available, the Office of Research Integrity heard 267 allegations of research misconduct. But the ORI concluded investigations in 23 cases.
Research misconduct was found in eight cases, and the scientists involved were barred from receiving federal research money for various lengths of time, usually three years.
Dissatisfaction with the system, which critics call imperfect, has increased in the wake of several high-profile cases, including one connected to discredited Korean researcher Woo Suk Hwang. Dr Hwang faked virtually all the data in two journal articles purporting to show that it was possible to produce stem cells from custom-cloned human embryos.
The articles' co-author Gerald Schatten, of the University of Pittsburgh, was found by his university to have committed "research misbehaviour" as opposed to "misconduct". The university said that this case was not being investigated by the ORI because no government funding had been used in the fraudulent research.
Observers also noted that it was rarely in a university's interest to report misconduct by researchers.
Chris Pascal, ORI director, said that much like congressional committees "an institution can have a natural preference for not finding research misconduct. It can lead to embarrassment. It may lose funds."
Offences that were disclosed, Mr Pascal said, were usually reported by "somebody who is in the laboratory or the department and is familiar with the research being done so they have enough knowledge to know that something's wrong".
It was just such a situation that led to the ORI's investigation of Xiaowu Li, a postdoctoral fellow at the University of California, San Francisco.
The inquiry concluded that photographs that he claimed showed human pancreatic cancer cells were actually of mouse melanoma cells.
Jason Lilly, a postdoctoral fellow at Cornell University, was found to have made copies of a single image of a genetic assay, altered them and claimed they were all different assays.
Ali Sultan, an assistant professor at the Harvard School of Public Health, was discovered to have plagiarised the work of another researcher.
Charles Rudick, a graduate student at Northwestern University, was found to have altered photos of recorded nerve signals.
ORI investigations typically take six months. These follow reviews by the relevant university, which take an average of ten months.
"If the ORI reviews the final report of the institution and it does not agree with it, only then will it conduct some kind of an investigation," said Adil Shamoo, University of Maryland professor, editor of the journal Accountability in Research and a longtime critic of the ORI process.
"The established community does not want the ORI to exist," Professor Shamoo said. "They accept it because they don't have a choice."
He added: "To me, the best deterrent is education and training, and it doesn't exist. People are not born knowing what the regulations are and what ethical norms are."