The Scientific Misconduct Process: Title and subTitle BreakA Scientist's View From the Inside

THE REVIEW PROCESS provided by the Research Integrity Adjudications Panel (RIAP) at the US Department of Health and Human Services Departmental Appeals Board (DAB) received a great deal of attention in the wake of the Imanishi-Kari decision issued in June 1996. As an active member of the research community, I have been interested in our ongoing attempts to address the problem of scientific misconduct, a problem that has demanded more attention as research has expanded and the competition for funding has affected every program. My perspective on this problem has been influenced by being the only scientist who has served on 2 RIAPs and worked with members of the DAB. The Imanishi-Kari case, in particular, demonstrated to me how important adversarial adjudication is for a fair and just outcome. Some critics attacked the RIAP process itself as unnecessary or lacking in expertise. I believe that the criticisms misperceive the essential role played by the RIAP in bringing finality to hotly contested cases through full, fair, and public hearings.

Under current regulations, nothing in the investigatory phase guarantees that an individual accused of scientific misconduct will be afforded such traditional due process procedures as the opportunity to confront and question witnesses or even to review directly all primary data and evidence (including that which may tend to exonerate the accused). The Office of Research Integrity's (ORI's) investigations may involve interviews with witnesses, but no one representing the respondent is necessarily present at any of these interviews. The process more closely resembles a grand jury investigation than a trial. Such a process may be valuable in producing leads and identifying potential evidence and may serve to reveal whether sufficient evidence exists to press charges, but it certainly does not constitute an open adjudication of such charges.

It has been claimed that the full-scale evidentiary hearing process of the RIAP unnecessarily prolongs the resolution of cases since, before the appeal stage is reached, institutional committees are required to include "appropriate expertise" in their process, and that the ORI "finds an individual guilty of scientific misconduct after either an institutional or an ORI investigation."¹ In addition, some have pointed to the draft reports issued by ORI, giving its analysis of the evidence and stating the charges, as providing a respondent (or accused scientist) with enough input since he or she may respond with comments or offer additional evidence.¹ But this is not sufficient.

I feel strongly that a full and fair hearing must be provided to any scientist who disputes either the facts that ORI has alleged or the legal implications to be drawn from those facts. Not only is this a matter of justice and of avoiding unfair harm to honest and talented scientists, it is also a requirement of regulations applicable to all government debarment proceedings.² Thus, whenever debarment is among the actions proposed by the government, as it was in the Imanishi-Kari case,³ the respondent must be afforded fundamental fairness, including an opportunity to confront and question any adverse witness and to present favorable evidence and witnesses. In the Imanishi-Kari case, we sustained an objection to ORI's effort to present transcripts of interviews (made without the presence of any representative of the adverse party) without making the witness available to be cross-examined precisely because of the RIAP's concern that due process as well as regulatory requirements demanded that the respondent be able to confront the witnesses. The reliability of a prior statement made behind closed doors by a witness not willing to appear at a public hearing is subject to question. Moreover, a decision based on statements taken at earlier stages without any opportunity for confrontation and cross-examination would likely be subject to reversal in court. Given the devastating impact that a finding of scientific misconduct may have on a scientist's career, even absent a proposal to debar, I believe that such safeguards should be available to ensure that the basis of any finding is established by reliable evidence.

It seems to me that 1 source of confusion has been that ORI has attempted to combine 2 roles that are better separated, ie, the prosecutorial and adjudicatory functions. In the cases on which I worked, ORI had devoted massive resources and spent extremely long periods of time to reach "final" conclusions in the cases. Yet, unless the respondent accepted ORI's conclusions without appeal, the end of ORI's lengthy investigation represented only the beginning of an adjudication to test independently the evidence in an adversarial forum. A prosecutor's task is not easy. Often it requires the gathering, sifting, and evaluation of complex evidence and the interviewing of witnesses. Nevertheless, the goal of the prosecutor is to prepare and present a case to a neutral and independent adjudicator, not to make a final decision in the case. The RIAP, as adjudicator in a de novo proceeding, has the benefit of hearing all the evidence and arguments from both sides. I believe that ORI's challenging role in prosecuting cases of scientific misconduct, whether arising through its own investigations or resulting from institutional inquiries, could be performed more efficiently if its focus was narrowed from generating comprehensive decisions to preparing thoroughly supported charges for presentation to the RIAP.

Since many of the high-profile cases disposed of before establishment of the RIAP lingered for many years before institutions and ORI or its predecessors, it is ironic that the RIAP has been targeted as a source of unwarranted delays by stating that its decisions are issued "months after the hearing."¹ In the Imanishi-Kari case, certainly one of most complex and controversial cases in this field, the parties' briefing was completed in April 1996 and the decision was issued in June. I am told that none of the other scientific misconduct cases that have gone to decision before the RIAP (5 of the 12 that have been appealed to it since 1993) has taken more than 45 days from record closure to decision.

It is true that the Imanishi-Kari case required 19 months from the filing of the appeal, through the 6 weeks of hearings, oral arguments, and the briefing process until final issuance of the decision (interrupted by 2 government shutdowns). However, it must be remembered that this appeals process was the first opportunity that Dr Imanishi-Kari had to receive definitive rulings on legal issues, to test all the evidence amassed against her, to cross-examine adverse witnesses, and to present documents and witnesses supportive of her position. In a de novo proceeding such as this before the RIAP, the panel had to reach a resolution based on evidence presented to it, rather than on a review of the record on which the challenged ORI decision was based. Therefore, ORI had to be prepared to present evidence adequate to support its proposed findings and recommended actions, rather than simply to explain why it reached its decision. Whereas some feel that requiring ORI to call witnesses whom it had already repeatedly interviewed was redundant,¹ it seems to me that the indispensable step was the presentation of the witnesses in an open hearing. I suggest that the best way to avoid reversals after years of investigation and expenditure of substantial resources by the government may be to clarify ORI's focus on determining which cases have merit, developing its cases adequately, and then promptly presenting for adjudication those that are disputed by the respondent. There is little value in changing who hears the appeals (which, as discussed below, is unlikely to change the outcome) or tinkering with the appeals process.

In fairness, it should also be pointed out that highly politicized and public cases (such as those involving Drs Sharma,⁴ Imanishi-Kari, and Gallo⁵ ) are not the norm and may not be a good basis on which to draw valid generalizations. More "typical" cases may involve less complex scientific and legal disputes, more straightforward issues of disputed fact, a record less compromised by long delays that can affect both witnesses' memories and the availability and integrity of primary data, and generally a less heated and contentious history to be sorted through. Such cases may more fairly represent ORI's ability to handle matters before it. Nevertheless, the RIAP has demonstrated repeatedly its ability to provide credible and timely resolutions even of the most controversial cases.

Debra Parrish, who previously served with the ORI and was one of the attorneys for ORI in at least one of those appealed cases in which ORI failed to prove misconduct (the Sharma case), describes the guidelines governing the RIAP's procedures as "ad hoc,"¹ as if the panel's process were decided arbitrarily. The RIAP guidelines, which we followed, were developed with ORI's involvement and were published in the Federal Register.⁶ The guidelines, like local rules of practice before a court or administrative body, are intended merely as a procedural guide. They provide latitude to the RIAP to modify procedures where necessary to achieve the goal of fairly and promptly developing a complete record to support a sound decision. The general powers of the DAB in hearing cases under its regulations (incorporated by reference in the guidelines) allow authority to adapt its process to meet the needs of a particular case. An example of the use of this sort of flexibility was the decision to prevent delay by bifurcating the hearing in the Imanishi-Kari case and going forward with the evidentiary presentation on most of the scientific and factual issues while reserving consideration of the forensic issues until they were ready for presentation.

The presence of a scientist on the RIAP in no way obviates the need to create a full record for decision. This record must include any scientific information on which the charges depend, in a form that is amenable to court review, if necessary. The preexisting knowledge or assumptions of a scientist member cannot substitute for expert witness testimony, stipulations, documentation, or other forms of scientific evidence on the record as a basis for a decision that can be upheld in the event of a court appeal.

Parrish's view of the use of a glossary in a scientific misconduct case is misconceived.¹ The submission of glossaries by the parties is not evidence of the inability of RIAP members to understand simple scientific terms, such as protein or amino acid. In fact, in the Imanishi-Kari case and, I presume, in others, the glossary submissions served 2 functions. First, those terms about whose meaning parties agreed could be treated as proven for purposes of the case, saving hearing time. But second and more important, major differences in the parties' interpretation and use of critical scientific terms, such as hybridoma, clone, and double producer, were first revealed and explored through their submitted glossaries. We found that ORI's case relied, for example, on a definition of double producers that was not accepted by any scientific witness, even ORI's own expert.

The impression that RIAP attorneys have difficulty with basic-level science conflicts with my experience in working with the panel members. When I joined the RIAP, I found that the 2 other panel members who would serve with me, as well as the 2 staff attorneys assigned to the panel, had already immersed themselves in the fundamentals of immunology and genetics, and I was impressed with the extent of their comprehension. To be sure, there were instances in which some technical point had to be discussed and explained, but the necessary clarifications were made to colleagues who had the sophistication and intelligence to make this task easy. Looking back at the hearings, it is clear to me that the attorneys were better prepared to deal with the science than I was to deal with administrative law. My colleagues were patient and efficient in teaching me what I had to know about the legal process.

There is no reason to disparage the expertise and status of the regular members of the RIAP by referring to them as "staff lawyers."¹ In fact, as I understand it, the 5 members of the DAB are appointed by the secretary of the US Department of Health and Human Services. Each one has many years of experience in conducting adversarial hearings and in reviewing the decisions of administrative law judges. (In addition, experienced administrative law judges from the Civil Remedies Division of the DAB are available to serve as panel members.) The 2 other members of the panel who served with me in the Imanishi-Kari case had between them more than 30 years of experience in adjudication. Board members have been hearing debarment cases involving scientific misconduct for more than a decade.

Perhaps the most important lesson of the Imanishi-Kari case is the rare good news in this field—something works! More specifically, despite much ink spilled over the essential incompatibility of legal and scientific culture, lawyers and scientists can work effectively together and successfully resolve even the most complex cases—both in terms of scientific issues and legal complications.

Certainly, legal processes and scientific methods are quite different approaches to problem solving and truth seeking, but their distinct strengths can support rather than undercut each other in dealing with allegations of scientific misconduct. It is the common business of law to weigh behavior against preestablished standards using well-developed procedures to preserve fairness, gather evidence, and test assertions. In the past, ad hoc approaches to resolving allegations of misconduct in science without legal involvement have been criticized for failing to give notice of what conduct might be sanctioned and denying due process to those accused. Nevertheless, when the conduct at issue is the doing of science, the law cannot operate in a vacuum oblivious to the real-world scientific context; the way in which science is normally done is an essential element of weighing what was done in the particular case. Further, justice is best done when done openly, and the involvement of a peer scientist in the process may help ensure the credibility of the ultimate decision for other scientists.

However, the attitude that science can police itself and that any lapse will inevitably come to light because false results will not be reproducible fails to address the following complications: (1) false results may well be discovered but dishonest researchers may not; (2) no profession that depends on public trust and public funding is immune from accountability; and (3) when a charge of misconduct can destroy careers, collegiality cannot substitute for due process.

It is seldom acknowledged that, in almost every situation where allegations arise about research, 2 very different problems are raised. First, the accuracy, reliability, and validity of the research data are crucial to those who need to reproduce or apply the questioned results. In such a context, it makes little difference if the data are unreliable because of poor interpretation, erroneous recording, miscommunication among the researchers, or experimental artifacts. Regardless of the cause, the consequence is that science cannot build on unreliable data.

However, the integrity of an individual researcher is the central legal and scientific issue when determining future eligibility for government funding. It matters little if the research turns out ultimately to be correct, if the researcher showed himself or herself willing to lie or cheat to arrive there. (It is perfectly possible to lie about one's work and, being an educated guesser, to turn out to be accurate without being honest.) The government as funder must focus on acts that impugn the integrity of the researcher. The scientific community is properly concerned with the reliability of both the data and the researcher. But the concept of scientific misconduct, with its concomitant implications for funding eligibility and institutional liability, should be focused on the integrity of the researcher.

At times, this distinction has been difficult to understand for scientists participating as witnesses, whistleblowers, respondents, or even observers. Sometimes there is a feeling that the right questions are not being asked or answered. However, the bottom line is that decisions that reflect on an accused scientist's integrity and affect his or her future career result from the development of a full record before a panel in open hearings with both the accused and the accuser having every opportunity to develop and test all the relevant evidence. Scientists have a valid and important role in this process, but scientific misconduct hearings are not a substitute for and do not serve the same purposes as peer review mechanisms.

It has been suggested that including 1 or more scientists in every panel would result in ORI success in proving misconduct more often. This does not reflect the facts or even the position of ORI. In the Imanishi-Kari case, several prior bodies composed entirely of scientists had decided that scientific misconduct had not occurred. The ORI rejected those conclusions in proferring its charges. The RIAP, which unanimously concluded that ORI failed to support its charges with adequate evidence, included a scientist (me). There is no reason to believe that inclusion of 1 or more scientists in the appellate panels in other high-profile cases in which ORI did not prevail would have changed the outcome. Further, ORI has not sought to include a scientist in all panels even though the RIAP's guidelines have always permitted it. And ORI has prevailed in cases in which no scientist was included in the panel.⁷

Despite the fact that participation of a scientist is not necessary or requested in every case, it is troubling to read the assertion that "many scientists refuse to be on the list of scientists" willing to serve as panel members, allegedly because they fear the "herculean" task of explaining science to the other panel members and the risk of being outvoted.¹ There is no instance I am aware of where a scientist panel member has been outvoted. Certainly, the magnitude of the cases in which I was involved was daunting and required significant commitment of time and energy. However, the major difficulty was in analyzing the massive record, not in educating my colleagues. I do understand that, before my appointment, an effort in the Imanishi-Kari case by the National Academy of Sciences to identify potential panel members under a procedure developed with ORI did not yield prompt or satisfactory results. However, it is not clear that this experience supports a generalization about the willingness of scientists to participate in the process, given the unique circumstances involved in that particular case, which may have discouraged or excluded involvement by many scientists with relevant expertise. In any case, I certainly recommend participation in a panel, which, though demanding, is a valuable and enlightening experience. Furthermore, I call on my peers to view such service as a professional obligation. It is hardly reasonable for scientists to call for an increased involvement in the resolution of misconduct charges and then refuse to serve when called on to do so.

While the entire area of scientific misconduct remains controversial, and the loss by ORI of several widely publicized cases has resulted in increased scrutiny of the process, it would, in my view, be a mistake to respond by attacking the appeals process that has brought openness and closure to these long-running cases. Scientists can make valuable contributions as participants in panels reviewing misconduct findings and should make themselves available to do so. But they should do so confident that they are participating in a process that works.