RESPONDING TO ALLEGATIONS OF MISCONDUCT IN THE UNIVERSITY

unsuspecting institution without providing fair warning. This means that institutions now must make a thorough evaluation of charges before closing the books on a case even if the accused has left the institution, and be prepared to communicate their findings to others. Fear of legal challenge more than genuine liability has tended to inhibit open and ethical responses to these situations.

Impatience with the slow pace of change was evidenced by remarks and testimony at Congressional Hearings in April 1988. There was pointed concern about how universities respond to allegations of misconduct, stimulated by the testimony of several whistle-blowers (some of whom shared their difficult experiences with us at the first Workshop in September, 1987) and strong suspicion that the self-interest of these institutions prevented them from being able to regulate or investigate themselves. NIH has been harshly criticized for its tolerance of this situation, and its failure to devote greater resources to monitoring and investigating. It will require considerable persuasion, perhaps even selfpersuasion, to show credibly that the universities can and will wash their own dirty linen because they understand that is no longer in their interest to hide it under the pile.

C.

Assumptions on the prevalence of misconduct and its reporting

In order to plan logical guidelines for responding to allegations of research misconduct, it is important to make some assumptions about how much misconduct is going on, how much is detected, and how much is likely to be reported. Serious fraud, as defined above, surely is uncommon, but may take a while to be detected, since it is, by definition, a willful misrepresentation with intent to deceive. Less striking forms of fraud and other forms of misconduct are undoubtedly more common, but their precise prevalence is unknown, and subject to the same kind of uncertainty as their definitions. Let us assume, then, that misconduct of various kinds occurs regularly, something between speeding on the highway and cheating on one's income tax.

The amount of research misconduct reported to federal authorities is small, far fewer than one per major research institution, but heavily weighted by more serious violations. The complaints of misconduct examined internally by a large research institution, weighted by the exploitative types, may be several per year. How many of these result from real research misconduct is uncertain. Awareness or suspicion of improper activities, however, is clearly more common than its reporting. Presentations on research misconduct are typically followed by audience discussion in which unreported cases are mentioned. Other experience indicates that many preliminary complaints are dropped before being submitted formally, for a variety of reasons. Significant scientific error, of uncertain culpability, occurs frequently, and publication practices that do not meet today's highest standards are quite common and sanctioned to some extent by custom.

A useful working assumption is that there are many research practices not in the best traditions of science and ultimately not in the best interests of the research institutions; that relatively

RESPONDING TO ALLEGATIONS OF MISCONDUCT IN THE UNIVERSITY

of such seriousness or importance that they are investigated formally and reported to national agencies or revealed in the press. This is a conservative assumption in the sense that it requires no great faith in the perfectibility of man to believe it, but it is radical in that it implies that science could indeed improve its self-regulation in order to continue to justify public support and confidence.

There has been no shortage of well-considered discussions of what institutions should include in preparing guidelines for responding to allegations of misconduct. The Association of American Medical Colleges in 1982 published a "prototype of procedures for dealing with alleged research fraud" in a booklet on "The Maintenance of High Ethical Standards in the Conduct of Research." This is a good starting point for procedures to draw the distinction between initial review of charges and formal investigations. It recognizes the primary role of the faculty in evaluating the validity of research findings, and indicates a standard of review which could stand up to challenge in court. leaves to each institution the details of who should supervise the implementation of procedures, and recommends realistically the periodic revision of such procedures in the light of experience. one can criticize anything in the AAMC document, it would be its focus on "fraud" alone, and its optimistic presumption of "the overwhelming probability that fraudulent data will be detected soon after their presentation." It appears that in 1982, the "selfcorrecting nature of science" could still be asserted without fear

of contradiction.

It

In 1982, the Association of American Universities also published the report of its Committee on the Integrity of Research. The AAU suggested that the past quiet and informal way of dealing with "deviations from the norm" must yield to better methods for preventing, detecting, and dealing with "irregularities." Institutions were advised to prepare policies which stated clearly and explicitly the standards of quality expected in research, the responsibilities of supervisors, the procedures for dealing with misconduct ("deviations from intellectual honesty"), and the sanctions available.

If

In

The AAU report indicated that "members of the academic Community have a responsibility to report what they believe to be lack of integrity in research." This recognizes the collective conscience which should supersede the collegiality of the individual laboratory. The committee argued that "research methods and results should always be open to inspection, evaluation, and criticism. this spirit, all involved should be encouraged to accept an investigation of alleged misconduct as part of the process of the search for truth." This is in harmony with the concept of encouraging more open criticism of research practices with less onus attached to having to acknowledge error.

The University of Michigan created a Joint Task Force on Integrity of Scholarship, which reported in 1984 on the full range of issues that concern us today, from the conditions that can

RESPONDING TO ALLEGATIONS OF MISCONDUCT IN THE UNIVERSITY

enhance or discourage integrity, to the development of procedures related to misconduct. The principal recommendations included definitions of the kind of offenses covered, the prompt notification of deans when major misconduct is charged, and two distinct stages of investigation. The report stressed the importance of establishing the credibility of the review procedures and suggested that confidentiality might be less important than credibility, especially at the stage of formal review. Major recommendations included development of a code of ethical conduct in scholarship, based on the comprehensive analysis of the research process that formed the first part of the report, and incorporation of ethical considerations at all levels of administration and teaching.

The

Despite this leadership, academic inertia prevailed widely until the real impetus for action came with the decision of the major granting agencies, NIH and NSF, to publish policies and procedures to guide both their own staffs and the awardee institutions in managing fraud or allegations of misconduct. NIH Guide of July 18, 1986 contained interim policies, which have been recently advanced toward regulation by a Notice of Proposed Rulemaking published in the Federal Register on September 19, 1988 (42 CFR Part 50; FR 53: 36347), concerning responsibilities of awardee institutions. The NSF final regulations were published in the Federal Register of July 1, 1987 (45 CFR Part 689).

Both agencies assigned to the awardee institution the primary responsibility for preventing, detecting, and handling misconduct in the research programs that they support. The NIH specifically endorsed the AAMC and AAU reports cited above, and required institutions to adopt policies and procedures to deal with allegations of research misconduct. The required features included providing for prompt inquiry, protecting both those who bring "good faith" accusations and the accused, notifying the granting agency if an inquiry results in a formal investigation, and documenting findings even if no investigation is pursued. Procedures should ensure fairness and minimize conflict of interest; formal investigations should be prompt but thorough, with notification of the funding agency of the progress and outcome of the process.

A.

IV. POLICIES AND THEIR IMPLEMENTATION

Policy development at the University of California, San Diego

In response to all these suggestions, and to the wellpublicized deficiencies in Harvard's initial review of the accusations against Darsee, as Associate Dean for Academic Affairs I worked with a faculty-selected committee to develop procedures for our School of Medicine; the existing Academic Senate Code of Faculty Conduct of the general campus was rarely invoked, required involvement of the Chancellor to institute any investigations, and did not spell out the way charges should be initially handled. We started with the concern that reviews be moved out of the involved departments as soon as it was determined that charges were not groundless or (as in the wording of some institutions) frivolous. In the working version of the policy we based the necessity to proceed on the level of confidence as to the innocence of the accused on the part of the individual supervising the confidential

· RESPONDING TO ALLEGATIONS OF MISCONDUCT IN THE UNIVERSITY

preliminary inquiry. Any "reasonable doubt" necessitated a further investigation to be carried out as expeditiously as possible by the associate dean with a small ad hoc committee chosen for its expertise.

This "further investigation" step, equivalent to the "inquiry" stage of the PHS regulations, specified participation of faculty from outside the department, and provided the dean with a consensus on whether a formal investigation would be necessary. In our procedures this ad hoc committee had the discretion to interview anyone concerned, and it was at this stage that the accused had to receive formal notice of the charge that had been developed, and advised of the status of the investigation. Our procedures do not address the question of keeping the identity of the accuser confidential; we believed that it was unlikely to remain a secret in most laboratories, and although we do not provide for direct confrontation of the accuser by the accused in the formal investigation, we do not believe that the accuser should remain anonymous at this stage except under unusual circumstances. The finding by this group that there was merit to the charges would be followed by a formal investigation, with due process as will be discussed.

By putting in one more preliminary step than in the usual recommendations, we allowed laboratory supervisors or department chairs to receive and evaluate complaints first, though an individual could go to the associate dean directly, or subsequently if not satisfied with the initial reception. Those who have responsibility for responding to charges of misconduct should act in awareness of the uncertainties in evaluating such charges and the mixed motives of the accusers. There is great latitude for informal confidential inquiry in this phase. It is not reasonable to assume that all complaints warrant regular inquiry and investigation.

If it is accepted that it is important to maintain a high ethical standard as part of excellence in research, then there must be provision for receiving and evaluating a variety of complaints and charges. Most procedures have dealt with charges of fraud or plagiarism, but not with the spectrum of other offenses against good science and ethics that are more common. The ethical atmosphere may depend on the way an institution reacts to the variety of initial charges or "expressed concerns" that may be received, and to decide which warrant the full investigative treatment. This can be facilitated by developing guidelines listing proper and improper research practices, a major project in itself, but an important part of due process (Mishkin). This will have been discussed in some detail at the recent Institute of Medicine Workshop on the Responsible Conduct of Research.

Administrative procedures for dealing with fiscal

irregularities or violations of animal or human subject rules are often quite separate, and are less likely to contain the safeguards and due process of general misconduct procedures, but accordingly are less likely to include serious penalties in their scope of disciplinary actions. We have included in the scope of our

RESPONDING TO ALLEGATIONS OF MISCONDUCT IN THE UNIVERSITY

procedures complaints of violation of the regulations governing animal and human research, but in practice these rules are

transgressed frequently, and supervision and correction provided by the dedicated committees and their staffs, so it is unusual for a problem to reach the dean's office.

C. Formal investigations

If a formal investigation is required, a larger committee is formed by the dean, and the accused is guaranteed an opportunity to appear and present evidence. The accused is advised to retain counsel, and the committee consults its own counsel. At this time, notification of granting agencies or other concerned authorities takes place. When appropriate, journals may be requested to stop processing manuscripts from the accused, with a promise that a full evaluation will be later reported. The committee may be charged with other matters by the dean, on an ad hoc basis, such as a request to recommend how to avoid future misconduct, or to review a larger portion of the work of the accused than that figuring in the present complaint. The findings and recommendations will be submitted in writing to the dean. We did not specify the time frame for this process, but are affected by NIH regulations which require substantial progress or completion of the investigation in 120 days. Flexibility is important, however, since it may take several weeks to assemble a select academic group for a committee meeting.

In our original working procedures, if an ad hoc committee found there was substance to the charges, a formal investigation was required. That worked appropriately when serious fraud was suspected, as in the Slutsky case (Engler RL et al, New Engl J Med 317:1383, 1987; Friedman PJ, Amer J Roentgenol 150:27, 1988). Later, when a less serious violation of research regulations was charged and the accused was initially found by the ad hoc committee to have been disingenuous about the matter, both our counsel and the faculty assembled to pursue the review considered that far too much had been made of it, and that we had followed our procedures too literally. After a modest reprimand to the accused faculty member, we revised the policy to allow an ad hoc committee to recommend a sanction or settlement by the dean when charges or findings did not warrant the major step of conducting a formal investigation.

D. The personal hazard of bringing charges of misconduct

Critics of the present self-regulatory mechanisms of science find that "whistle-blowers" have been treated with skepticism and rejection by their institutions. This should be no surprise. The scientific ideal of independence has led to tolerance of much deviant behavior, including a certain amount of research misconduct, but formally bringing charges of misconduct is a threatening departure from the customs and the mutual tolerance of scientists. While skepticism and challenge are accepted ways of reacting to the results and theories of others, it is seriously different to claim that someone has cheated or lied, not just made a mistake. It is obviously difficult to challenge a more senior person, and other feelings inhibit challenging a young colleague publicly. The advice given to and acted on by trainees and junior investigators has been to distance themselves from improper research activities, rather