Closing the barn door: Coping with findings of research misconduct by trainees in the biomedical sciences

Abstract

The proportion of research misconduct cases among trainees in the biomedical sciences has risen, raising the question of why, and what are the responsibilities of research administrators and the research community to address this problem. Although there is no definitive research about causes, for trainees the relationship with a research mentor should play a major role in preventing actions that constitute research misconduct (fabrication, falsification and plagiarism). Examination of the limited literature and of the number of cases closed by the US Office of Research Integrity (ORI) between 2009 and 2013 raises questions about the mentor-student relationship and what it should be accomplishing. But many gaps in policy and its implementation inhibit this role. There is no acknowledgement of mentorship in federal regulations and research on how to teach research integrity is woefully underdeveloped, especially for international trainees. And some institutional research integrity officers may have had little preparation for the role.

Introduction

Cases of research misconduct (RM) continue unabated around the world. In the US, ORI reports new allegations rising from 86 in 1993 to 154 in 2012 (Office of Research Integrity [ORI], 2012). Others recently reported include lack of acknowledgement in the peer reviewed literature of findings by the FDA of fabricated and/or falsified data (Seife, 2015). The ongoing parade of misconduct raises questions about why researchers, students and staff falsify, fabricate and plagiarize and how these actions might be prevented. This paper suggests that re-examination of information about graduate students committing research misconduct is necessary. It also suggests actions that can be taken to ameliorate the situation.

In the years 2009 to 2013, 26 of the 56 cases of research misconduct reported in the US Office of Research Integrity (ORI) newsletter involved graduate trainees including postdocs and medical residents. This figure is up from 30% for cases between 1990 and 2004 (Wright et al., 2008).

For having intentionally fabricated and/or falsified data or plagiarized in Public Health Service (PHS)-supported research, these individuals entered into agreements to be excluded from serving in any advisory capacity to PHS and frequently to be excluded from contracting or subcontracting with an agency of the US government for a period of two to five years. Sixteen of the 26 were also required to have an ORI-approved plan for supervision of any subsequent PHS research prior to any future application submission. For 10 of the 16, their employing institutions had to certify to ORI that data provided by these individuals were legitimate.

These statistics raise several questions. Why are trainees such a large proportion of cases? Are they reported more frequently than are scientists in the professorial ranks? Is their misconduct related to immaturity in learning scientific norms or to pressures to succeed or both? Are their mentors lax in reviewing source data and setting standards? Wright et al. (2008) found these mentoring problems in three-fourths and two-thirds respectively of trainee cases closed by ORI prior to 2005.

Although there is no definitive research about the problem, research misconduct may occur because of: 1) sociopathology, 2) increasing pressure on researchers, 3) ignorance of research standards and ethical norms or some combination of these causes (Wright et al., 2008). Some argue misconduct is behavior confined to a few bad apples. An implicit assumption in the bad apple explanation is that sociopathology, or just not being able to “cut it”, are the sources of the problem.

The research mentor-trainee relationship should moderate pressure and assure that research standards and ethical norms are taught and learned. But in the US federal research misconduct regulations there is no definition of the responsibilities of the research mentor and no requirement regarding mentoring (Wright et al., 2008), leaving little direction for what actions and outcomes are expected of mentors. (Mentors are faculty supervisors employed by the university that will award the research degree and are assigned to guide graduate students through the research process. Other terms for this role may be used in other countries.) In addition, scholarship regarding researcher moral development or the efficacy of research integrity teaching is woefully underdeveloped (Evans, 2011).

Since 1989, NIH has required trainee instruction in responsible conduct of research (RCR). RCR training is required only of those supported on NIH training mechanisms. Even among top-funded US institutions, only half require RCR training of all students (Resnik and Dinse, 2012).

Despite this investment, little is known about what works (Mazmanian et al., 2014; Kalichman, 2013). A summary of the few available studies suggests that current RCR instruction is largely ineffective and in some cases may be harmful, deriving from student avoidance of ethical problems or overconfidence in their ability to handle them (Antes et al., 2009).

As a strategy for economic competitiveness, many countries are heavily investing in PhD education in science and technology (Mulvany, 2013). Among the large number of international trainees in the US, many are first introduced to research practice in their home countries and are bewildered by US expectations for RCR. These authors have also found international trainees to be less likely to accept US norms than are their US-trained counterparts. The normalcy of plagiarism in many international environments, vague policies about what constitutes plagiarism, and difficulty writing in English (Heitman and Litewka, 2011) may put them and their mentors at risk of misconduct.

In addition to the usual sanctions in a finding of research misconduct, ORI’s requirements for supervision and in some instances certification of trainees’ subsequent work are likely meant to provide them with guidance and oversight to develop the research skills and integrity they need. Current regulations target the actions of FFP after they occur, without regard for the developmental situation of trainees or the institutional environment in which they are operating. But this is “shutting the barn door after the horse is gone.” A whistleblower, sometimes the mentor, has already reported evidence of fabrication and/or falsification or plagiarism. Federal misconduct regulations or institutions themselves ought to require some prospective monitoring of or outreach to trainees.

Trainees surely should expect to receive guidance and mentorship in practicing research ethics. If they are found to have “intentionally, knowingly or recklessly” (42 CFR, Part 93.104) committed research misconduct, it is reasonable to ask what went wrong in their development. Research administrators follow institutional policy for handling allegations and investigations. But policy as 42CFR, Part 90, section 93.300 does not address responsibilities that educational institutions assume for the development of the ethics of students.

Doctoral students

While PhD education should be the primary locus of socialization into research ethics, there is almost no literature about how successfully that is accomplished. A survey from Norway of PhD students in all of that country’s medical faculties found that “10% did not find it inappropriate to report experimental data without having conducted the experiment; 38% did not find it inappropriate to try a variety of different methods of analysis to find a statistically significant result; 13% agreed that it is acceptable to selectively omit contradictory results to expedite publication; and 10% found it acceptable to falsify or fabricate data to expedite publication”, although no participants reported they had fabricated, falsified or plagiarized data or publications (Hofmann et al., 2013: 1).

A smaller Swedish study found that students reported that they had experienced exploitation by being asked by their mentors to do work not related to their doctoral studies, abuse through public humiliation, and had their ideas and/or data misappropriated (Lofstrom and Pyhalto, 2014).

But perhaps the ultimate occurred with a group of doctoral students at the University of Wisconsin who reported their advisor for research misconduct, which she was found to have committed. One of those students describes the losses as including a mentor without a lab, many research projects never published, and students being advised to find a new laboratory home and start over on a new project. Those students now worry whether their roles as whistleblowers will affect their hireability (Allen and Dowell, 2013). Whistleblower protection policies assume threat to employment and do not address the educational impact on students when they act as whistleblowers. This must be explicitly addressed at institutions which host students. Also to be addressed is the university’s responsibility in helping students in such a situation to finish a degree.

With one exception the PhD students in the current study, among those found to have committed research misconduct between 2009 and 2013, ceased publishing (as documented in Pub Med), although those found in later years of this period may yet regain the ability to publish. It’s as if they simply vanished.

Postdocs

About half of biomedical postdoctoral fellows in the US are temporary residents, many of whom received PhD training outside the US (Ghaffarzadegan et al., 2014). Many are supported by research assistantships from NIH funds, which provide the institution with 50% indirects but on which they may not receive the documented formal and informal mentorship that is required of NIH-sponsored training grants, which yield 5% indirects. Such training grants require mentor plans and are judged by the quality of the mentor’s training record and prior trainee outcomes (Rockey, 2014).

This constitutes a perverse incentive in US policy, largely benefiting senior researchers who depend on large numbers of graduate students and postdocs, well beyond the number who will find jobs (Stephan, 2012; Rockey, 2014). Indeed, a study of mentors affiliated with the Clinical and Translational Science Award (CTSA) program found 39% with four or more CTSA mentees, with most having additional non-CTSA mentees (Miyaoka et al., 2011), raising a question of mentor capacity.

Out of the 26 student cases settled by ORI between 2009 and 2013, 14 already held PhDs, and the misconduct was identified during the postdoctoral training (54% of student cases). One would expect such individuals to have been socialized to RCR during their PhD programs. Through publication records, four were verified as coming from a foreign country for a US postdoc and returning to their country of origin after the finding of misconduct (China, India, Japan), continuing to publish. Two others had been publishing from their country of origin prior to the US postdoc (China, Korea); recent publications show they are now working in US institutions. Japan also has suffered from multiple egregious cases of research misconduct, indicating a lack of research oversight and cultural reluctance to act on suspicions of peers (Agency for change, 2014; Normile, 2014). Many others could not be tracked because their publication records stopped or they could not be identified.

It is important to note that the major country supplying the US student or workforce in biology and other fields (17%) is China; the second is India (Franzon et al., 2012). Both countries have serious local problems with research misconduct. Conservative estimates indicate that a third of Chinese researchers have engaged in practices that include data fabrication and plagiarism (Cao, 2014). China has been very successful at recruiting overseas Chinese-born scientists back to their country of origin (Xie et al., 2014), raising the question of whether findings of RM in a US postdoc have any impact on a subsequent scientific career back in the homeland. India has no specific laws pertaining to scientific fraud. In a small study of medical colleges and hospitals, 56% of questionnaire respondents reported knowledge of alteration or fabrication of research data (Dhingra and Mishra, 2014). The rate reported largely in the US and UK summarized in a systematic review was 14% (Fanelli, 2009).

Role of the research administrator

Institutions receiving federal research funds must manage a number of regulatory requirements including protection of human and animal subjects, financial conflict of interest, biological risks. Among these regulations, in place since 1989, are those requiring assurance that policies and procedures are in place conforming to 42 CFR 93 to investigate allegations of research misconduct, defined as fabrication and/or falsification and/or plagiarism. The institutional official, the Research Integrity Officer (RIO), administers these policies and procedures.

Although every institution with an assurance of compliance must name a RIO, the role is not delineated in the federal regulations. Two studies have found that in the aggregate, preparation for the role is rare, most have handled few cases, and many RIOs report inconsistent legal and logistical support (Wright and Schneider, 2010). When compared with responses of expert RIOs to usual research misconduct situations (sequestering evidence, threat of retaliation, and coordination of responsibilities with the institutional review board), many RIOs did not demonstrate a high level of skill (Bonito et al., 2012).

What are our responsibilities as a research community?

It is clear that system-wide reforms are necessary. Leshner (2012) notes there is far too much variability in training for the 50,000 postdocs in the US. Calls a decade ago to establish standards, norms and expectations for mentors, mentees and their institutions have still not been addressed (Leshner, 2012). Alberts, Kirschner, Tilghman and Varmus (2014) note that the severe imbalance between limited funds available for research and the still-growing numbers of researchers and personnel in the scientific community in the US have created a hypercompetitive atmosphere, which can heavily influence the integrity of graduate students and postdocs. Devereaux (2014) describes the growing gap between scientific ideals and the institutional reward system (e.g. science asks for collaboration and openness but rewards competition and “getting there first”). Yet, despite a more perilous environment, next to nothing is known about how to educate to reduce misconduct.

Even in the absence of reforms in the broader governance of science, institutions can adopt policies aimed at preventing research misconduct among graduate students and can evaluate them to see if they are effective. First, accountable mentoring must be assured, undergirded by mandatory and effective RCR education tailored specifically to graduate students. Particular attention must be paid to at-risk non-English speaking groups and those from countries with research ethics standards discrepant from those expected in the host country.

Mentors should receive training for their roles and be expected to review source data and set standards (Wright et al., 2008), and these expectations should be built into the institution’s code of conduct for faculty employment. Investigations of misconduct allegations against a student should require an accounting of the mentor’s activities in carrying out these duties. And NIH policy should require all students supported by NIH grants (whether research or training grants) to be assigned an accountable mentor.

Students in training programs have described mentor pressure to behave unethically. A survey of the MD Anderson Cancer Center found a third of student respondents feeling pressure to prove the mentor’s hypothesis, even if the trainee’s data did not support it. Twenty percent of students reporting in this survey said they had been pressured to publish findings about which they had doubts (Mobley et al., 2013). And an Australian survey found student criticism that university academic integrity policy was not enforced (Mahmud and Bretag, 2014). These kinds of experiences/perceptions can undo any RCR program.

RCR education should incorporate these and other concerns of doctoral and postdoctoral students. Their roles as beginners make them vulnerable to uncertainties about correct courses of action in situations they will encounter and dubious about the consequences of acting on their judgments, especially against authority figures. And as a condition of accepting graduate students, including postdocs from non-English-speaking countries, institutions must seek to improve their writing of English manuscripts, to a documented standard of acceptability, and provide help in teaching how to express their ideas without being tempted to, or even out of respect, plagiarize. In general, guidance on effective research ethics for international trainees and US trainees in international research settings has been vague and irregular (Heitman and Litewka, 2011).

As typically conducted, much RCR is not especially effective. It is important for institutions to adopt models shown to be more effective to obtain the best outcomes for the investment. The most successful programs are case-based and interactive, requiring practice in identifying ethical issues and strategies for working through problems (Antes et al., 2010). Nedeker (2014) recommends actively engaging students with role-play, debate and use of authentic examples and formative evaluation during instruction. Mumford, Steele and Watts (2015) suggest evaluation of RCR programs in four areas: behavior, cognition, reaction and institutional outcomes, and review measures available to do so.

The scientific community and individual institutions are not doing what is needed to discourage misconduct among young investigators. Education and mentoring clearly have important roles to play but how and with what accountability remain unclear. What is clear is that focusing on punishment is not the best route to discourage unethical conduct.

Conclusion

Ongoing issues with research misconduct among trainees in the biomedical sciences require action by research training institutions and by the scientific community. The absence of federal regulations addressing responsible mentoring and lack of a research base for education in the responsible conduct of research constitute gaps in our knowledge. But development of best practices that are evidence-based can and should be undertaken by individual institutions and/or by consortia of research institutions. Accountable mentoring and effective RCR are basic. Graduate students operating in a hypercompetitive, hierarchical environment with a widening gap between scientific ideals and everyday practice are vulnerable. This is especially the case if they are not fluent in English and have been educated in a culture whose standards are not congruent with those in the host country.

The ultimate test of more rigorous standards will be a decrease in the percentage of research misconduct cases in students and their satisfaction that they have been supported in attaining scientific integrity.