Implications of the Feldheim Eaton NSF research misconduct case

The following comment was published January 10, 2016 on Retraction Watch website. (The original can be found here.) It is important because it adds an additional example of the government research misconduct functions which we discussed in our manuscript, "The Essential Need for Research Misconduct Allegation Audits", published in December 2015.

 

Journalist Joseph Neff noted in his article published earlier this week about this research misconduct case that:

 

“The final NSF decision did not conclude that the scientists acted intentionally or recklessly, which would have resulted in a finding of research misconduct. That decision was a step back from the findings of the Office of Inspector General, the investigative arm of the NSF, which concluded in 2013 that Eaton, Feldheim and doctoral candidate Lina Gugliotti had acted recklessly and committed research misconduct.” (The News & Observer, January 8, 2016)

 

These results raise some additional questions regarding the research misconduct oversight system in general. First, why was the Office of Inspector General’s (OIG’s) finding of research misconduct softened by the NSF Director’s office?

 

The statement by the NSF Director’s Office in reversing the its OIG’s finding of research misconduct seems very vague compared to the much more detailed and specific claims of the NSF OIG, as well as what has been published in the press by reporter Joseph Neff.

a) From the NSF Director’s Office statement:

 

“After reviewing the evidence presented in the report, the OIG report, and [blacked out] rebuttal letter, we find that, while there are certain facts arguably supportive of recklessness, the record overall fails to provide the preponderance of evidence necessary for a determination that your actions associated with the research at issue were intentional, knowing or sufficiently reckless to rise to the level of research misconduct.”

 

Nevertheless, even the NSF Director’s Office concluded:

 

“While your conduct may not support a finding of research misconduct, it does violate the NSF Act of 1950 (the NSF Act), as amended.”

 

And the Director’s Office continued:

 

“In this case, NSF has identified significant findings that you and your co-authors failed to disclose including: 1) a clear and consistent description of the composition of the used in Publication #1; and 2) the fact that data was not properly [blacked out] prior to Publication # 1. These omissions were as significant to the published research record in question as the significant findings that you did prepare and submit for publication.”

 

So was this all really just a terrible one-time lapse by these scientists, or did the NSF’s OIG provide evidence to make a case that more likely than not this was something more, i.e. at a minimum behavior exhibiting a reckless disregard for the truth or falsity of what the authors were publishing and claiming in grants?

 

Indeed, further statements from the NSF Director’s Office seem to suggest a willful and continuing unwillingness of the accused scientists to respond to corrections with which the NSF Director’s Office apparently agreed:

 

“You chose not to follow the actions recommended by the investigation and failed to clarify fully the record. Taking into consideration the impact on the scientific record of ambiguous justifications and misleading terms, NSF now imposes that same requirement on you.”

 

b) From the NSF OIG’s statements:

 

1) “OIG Investigation concludes that:

• Act: the Subjects’ conclusions in a published paper linked to NSF-supported research were not supported by contemporaneous data.

• Intent: All three Subjects acted recklessly.

• Standard of Proof: A preponderance of the evidence supports the assessment that the Subjects’ acts were falsification and a significant departure from accepted practices and therefore constitute research misconduct.”

 

The NSF Director’s Office referenced the Black’s Law Dictionary for meaning of the term “reckless”. Since the legal meaning of this term is probably not well known to scientists, we repeat it here from the Ninth Edition of that dictionary:

 

Reckless behavior is “Characterized by the creation of a substantial and unjustifiable risk of harm to others and by a conscious (and sometimes deliberate) disregard for or indifference to that risk; heedless; rash. Reckless conduct is much more than mere negligence: it is a gross deviation from what a reasonable person would do.”

 

2) As part of their assessment in making a finding that the accused researchers were acting “recklessly”, the OIG claimed:

 

“Even after investigations commenced, none of the subjects reviewed laboratory records, but perpetuated inaccurate statements in subsequent publications.”

 

This would seem to imply a knowing resistance to correcting things which other investigating bodies seemingly had concluded needed to be corrected. Furthermore, as trained fraud investigators, NSF OIG apparently understood the importance of establishing a pattern of misbehavior:

 

“The Subjects falsified assertions in Publication 1. Subject 1 and Subject 2 engaged in a pattern of reckless perpetuation of those assertions in collaborative work repeated in Publication 4 and during the investigative activities at two universities. Even when the work was called into question, the Subjects did not review the supporting data to confirm or refute the allegations which would have been a reasonable course of action. For Subject 1 and Subject 2 this reflected their continued failure to review the data initially collected by Subject 3.”

 

So this case is not about some one-time mistake or failure. The NSF OIG further claimed that,

 

“The University 1 IC [investigating committee] concluded that Subject 2 [Feldheim] was careless in that he failed to consult experts on [blacked out] and relevant data interpretation with regard to the assertion of [blacked out]. Our investigation showed (but University 1 [NC State] was unaware) that Subject 2 did consult with such an expert (Witness 1). Witness 1 told Subject 2 that the asserted [blacked out] was unlikely, warranting further care ...”

 

NSF OIG’s assessment was hardly waffling:

 

“Our analysis of the evidence shows that collectively the Subjects consciously did not demonstrate the care a reasonable person similarly situated would about the consequences of his actions and the potential resulting harm to the research community through Publication 1.”

 

So what precipitated the retreat of the NSF Director’s Office from the findings of its own OIG, and does this incident have any more general relevance for the handling of research misconduct investigation in the U.S.?

 

First, as indicated from the quotes presented above, it seems that NSF OIG was more specific and extensive than the Director’s Office in their claims. Had NSF’s Director countered its OIG’s claims, that might have provided an appropriate basis to retreat from OIG’s conclusions. Instead, quite oddly, the NSF Director’s statements seem to be largely in agreement with its OIG’s claims of a continuing and deliberate resistance of the accused scientists to assessing their own data and making corrections to the literature. If so, this suggests that there may have been something other than the facts of the case that made the NSF Director back off from a finding of research misconduct.

 

It is apparent that the National Science Foundation does not rely solely on the reports of its professional fraud investigators in making findings of research misconduct. In a manner typical of federal agencies, the NSF OIG’s findings are subject to “higher” level agency decision making that can “overrule” them; in this case, OIG’s findings were subject to review by the NSF Director’s Office. Indeed, as we pointed out in a recently published manuscript (available as a preprint at PeerJ: https://peerj.com/preprints/1577.pdf), biomedical research oversight by the ORI is in a similar position. ORI merely “recommends” findings to the Office of the Assistant Secretary of Health (OASH) in the Department of Health and Human Services. Former ORI director David Wright referred to OASH as an “intensely political environment”. That politicized environment led him to question whether OASH was an appropriate “home” for ORI. Therefore, while NSF OIG may have greater powers and professional anti-fraud structure than ORI (as also discussed in the PeerJ manuscript), the net result is that both ORI and NSF OIG are within agencies which are almost certainly subject to the lobbying of large institutions.

 

A second question has been repeatedly raised by various members of Congress and others with respect to the handling of research misconduct cases: Were the grant funds extended to these errant scientists recovered as a result of the investigatory process?

 

There is no statement to this effect in the NSF OIG’s report. Perhaps by changing a finding of research misconduct to not research misconduct, pressure to recover funds from the responsible institutions might have been avoided. To pursue such an effort is often highly contentious. This is but one of several considerations other than the plain facts of the case that arguably could have influenced the NSF Director’s Office to add up the preponderance of the evidence differently than its OIG.

 

Regardless, it is an important principle in anti-fraud efforts not to have weak penalties. This includes not only for the scientists, but for the institutions which are responsible for the training and oversight of their researchers. Otherwise, if cheating is low risk and high yield, fraud is promoted rather than suppressed. A similar situation exists in many fields, including athletics, for which reason current anti-doping efforts are quite strict. (The German legislature, for example, criminalized athletic doping last November.) Cheaters will tend to drive out those who do not cheat unless there are serious penalties. For example, we can ask whether high quality scientists did not get positions at the University of Colorado, Boulder which were given to those making the spectacular, but now apparently universally acknowledged as false, claims made by the subjects of this NSF report. Taxpayers as well as honest scientists lose not only financially but in other opportunity costs.

 

Therefore, while having an IG oversee U.S. biomedical research misconduct cases would very likely be superior to the less capable ORI (as discussed in the PeerJ manuscript cited above), there are weaknesses to the IG system as well, one of which is in common with ORI – being subject to the political nature of their agencies.

 

Nevertheless, IG’s have a further very important advantage: they report directly to Congress as well as their agency. If upon report from an IG, Congress does not feel that agency decisions are adequately protecting taxpayer interests, they can take various actions. The influence of some U.S. senators was well exhibited with respect to significantly increasing penalties in the recent Han Iowa State HIV antibody fraud case. In addition, IG’s are subject to performance audits, and they themselves can conduct agency audits. Either of those options, or a Congressional request for its GAO to do a performance audit of NSF or HHS/ORI, might also be revealing and influential. Therefore, it will be interesting to read any performance audit assessing NSF to see if concerns are raised by the outcome in this case.