Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
C=PrER 5 IONS ~ preparing rations, the cc~ni~ considered score Ian 60 pectorals developed ~ me six panels and plenary discussions of the Meteor 6-8 workshop, regulatory portals published by Me National Aspires of Health and the Department of Health and Amman Services (l~i) in the Sumter 19 Federal Resister (~;~, 1988a, 1988b), a draft report on scientific my onduct released by Me HHS Office of the Or General ~ late ~ (USI~HS, 1988c), and legislative proposals developed by I; of Congress (Lipton, 1988a). The - cc~mi~e also reviewed ~ background papers carnmissioned for the worl~hc'? (my, 1988; Jasanoff, 1988) and selected literature frog the project bibliography, including reports on Maturity of Search am scientific ~scor~uct published by the American Association for the ~ of Scions, the Association of American Universities, the Ass aviation of American Medical Colleges, arm Sigma Xi discussed in Copter 2. Af ~ r r~riewirx~ this ma ~ rial, the cc~mitt== formulated the following rrYx'T~enlations. They describe the actions that the committee belle vex are most appropriate to be taken at this time by government agencies, universities, professional organizations, and journals in seeking to promote Integrity and quality in health sciences rls~irch. RE~ONS FOR TO NATIONAL INgrnm~ OF HFAIQff 1. The National Institutes of Health shard establish an office to Promote responsible reteach Practices. This office should be coordinated with an expanded NIH effort to evacuee institutional investigations of misconduct in scientific reteach. The primary function of the office Should be to foster and monitor he development of high professional stanx arcs of research practice by all grantee and applicant institutions. The National Institutes of Health needs to respond vigorously to the perception that science has not maintained high standards of integrity. Although the committee believes that the primary responsibility for developing and implementing professional research standards rests with principal research investigators, medical schools, universities, and other research institutions and professional organizations, the members conclude that NIH must provide leadership and motivation to induce the research community to define professional standards for research practice, monitor levels of supervision, arm provide education in professional stanzas and the ethics of research. Taking sew; to promote the development of these standards at the local level is one of the nest important actions that NIB can take to improve the integrity of research. 23
Pony academic institutions and ~dic-~1 schools have recently developed policies art procures for report to allegations of scientific m~s~nduct. Sane institutions did so ~ response to cases of misconduct. Others are doing so now In r ~ onse to regulatory requirements. m ese policies and procedures are important, but they are insufficient to promote integrity and quality in the research environment. The committed believes that additional standards are necessary to define the means by which the institution will promote the responsible conduct of research. As with the requirements for safeguarding human and animal research subjects, the proposed office will provide institutions an opportunity to involve the ~ faculties and administrations in developing ~ licies and procedures that are consistent with traditional practices and that also meet federal standards. It is the perception of the committ^= that institutional self-regulation will be strengthened as a result of this approach. The committee recommends that the propose] office be established as an administrative unit within the Office of the Director of NIH and have oversight responsibility for all NIH grant and contract awards. The office should assist local research institutions in developing professional standards of research practice and define those areas that must be addressed by local policies and procedures, including the way the institution monitors the research environment, and how it p y s to respond to charges of misconduct In research. The areas to be addressed by local institutions should include: o policies for recording and retention of research data; o professional standards for training and supervision; O education In p~vfession~1 standards and the responsible conduct of research; policies and procedures for responding to allegations of misconduct; - designation of an institutional official to address concerns related to the conduct of research; and description of the process by which the university faculty, staff, and students--are kept informed of institutional and professional research standards and policies. o o o There are some potential disadvantages with the establishment of an office to promote responsible research practices as proposed. For example, the additional amount of paperwork required would involve time of scientists and university officials, as well as the efforts of some administrators at the NIH and comparable agencies, and for these reasons could add some additional expense to research. In addition, the mission of this office might be inappropriately broadened. The proposed NIH office should disseminate information about institutional policies and procedures governing professional standards and assist in their development; receive regular reports, possibly on a biannual basis from institutional officials charged with addressing 24
research concerns, describing developments that affect the integrity of health sciences research; convene occasional forums for these officers and the general research community to prcmcke broad awareness of important trends or new issues that influence the conduct of research; interact with other NIH officer that monitor compliance with regulations on research involving human s Ejects and animals, the use of hazardous substances, or recomb mant-DNA materials; and work closely with, but be distinct from, the NIH office charged with evaluating the adequacy of institutional misconduct investigations. The proposal to locate the office for the promotion of responsible ~. . _ . research In NIH is based on the relationship ot the sc~ent313c~lly trained administrators at NIH with the researx h community, whose activities they understand and whose expertise they often solicit in evaluating specific problems. NIH and other research funders can exercise an influential rode by organizing conferences and workshops in which their grantees and professional organizations exchange information about and experience with the development of standards to promote responsible research practices. m ese meetings could highlight useful models, identify problems and barriers to the development of effective research guidelines, and suggest collaborative solutions to problems that may appear intractable to individual institutions. 2. By 1992, NIH should require all grantee and applicant institutions to provide assurances that they have adopted policies and pro secures to encourage responsible research practices. Research applicants should affirm their familiarity with these policies and prooe~ures and should also propose how they plan to store research data in the course of their study. The proposed NIH office should require assurances that applicant institutions have adopted guidelines a ~ oversight mechanisms for the responsible conduct of research analogous to those required for research involving human subjects or animals. To enable the proposed office to carry out its functions, NIH should require applicant institutions to forward a copy of their research standards when developed. NDH should plan ~nnnunce ~ specific date after which grant applications that do not include the required assurances will not qualify for funding. The areas to be addressed by local institutions are listed under Reccmmeniation 1. The subjects of the propose] standards for the responsible conduct - of research involve difficult and controversial matters. Development of policies and procedures for satisfying these requirements will take time, because they require debate and education within the local institutions. At the same time, it is important that these institutions initiate a process to develop formal guidelines in onier to comply with 25
funding requirements. Thus, the committee suggests a three-y~r time period, at the most, for the implementation of the institutional assurances requirement. When the assurances are received, the NIH office for the promotion of responsible research should monitor the adequacy of local institutional guidelines and statements of compliance by investigators with these standards. The office should notify applicants and grantees from institutions that have not. developed appropriate guidelines that they will no longer qualify for NIH-funded research support. The office should provide educational support and guidance for institutions that have difficulties in developing appropriate standards. The committee also encourages NIH to require their research grantee= art applicants to describe within each prodded study the m ~ anises for data retention and consider the issue of data are<=== in the course of their research. While institutions are being charged to develop overall policies to dead with these issues, the individual grant will serve as the best location for detailed specifications of the data arctic an] storage procedures to be followed. m e appropriateness of these prooe~ures can then be evaluated as part of the study section review of the research proposal. m is approach provides for the increasingly common and difficult circumstances of multi-institutional or other collaborative studies, in which concerns about intellectual ownership may not become serious problems if they are resolved before the research is performed. In such circumstances, no single institution can provide oversight based on its . . con policies. 3. NIH should not implement random data audits as a mechanism for _ ensuring the responsible conduct of investigator-initiated research. Data audits are occasionally used by government research agencies when there is reason to question an investigator's reported research results. These "for-cause" audits are often useful in identifying incidents of fabrication, falsification, or serious distortion of research data. For-cause audits may result In punitive actions, such as critical prosecution, disqualification of an investigator or, in the cane of the Food and Drug Administration (FDA), restriction of his or her access to new drugs (Shapiro and Charrow, 1985~. In recent regulatory proposals the HHS Inspector General and the Department of Health and Human Services have suggested various means to enhance the detection of research fraud, such as the use of random or systematic data audits by government investigatory bodies. These random audits differ from the for-cause audits because they are conducted without reason to suspect or question the reported research results. It has been argued that data audits are useful and accepted by investigators conducting studies subject to FDA oversight and by 26
investigators conducting mwlticentPr collaborative studies funded by the National Cancer Institute (NCI) (Shapiro and Charrow, 1985; 1isook, 1986). In the case of FEA-regulated studies, systematic review of data is performed in the final stage of new drug applications to ensure compliance with formal, muhIa1ly agr~P~-upon' predefined protocols and to determine that all data have been correctly collected and recorded. This form of data aunt can be carried out by individuals with minimal ~ ~ e _ ~ ~ ~ SCl~tl: -1C tralRllilg. Similarly, contract work and multi-institutional studies for NCI or NIH require precise adherence to protocols accepted by all investigators at the initiation of the study. Spook agents of the Eta report in these studies allow for a technical review similar to that used in the FDA agents. In these studies, the investigators know about and agree to the use of data aunts as a form of quality control in the development of the study contract and the research protocol. Uniform practices and prearranged agreements characterize these stories. Thus, this research activity lends itself to a~;tina and is enhancer TV the systematic audit requirement. , i_ Some investigator-initiated, grant-supporbed research involves therapeutic trials, other interventions, or measurements on human subjects following clearly specified protocols that have been filed with Institutional Review Cards. Data audits of these studies cod be corx~uc~ by auditors without extensive scientific expertise. On the other harm, Tech of Be investigator-initiat~ research that NIH supports is basic or discovery research. The nature of the research arm its susceptibility to scrutiny by nonscientists are dramatically different fern clinical trials. Protocols sunlit In the research application are often changed ~ preliminary results reveal unexpected findings. The research is more exploratory and less repetitive. me technical difficulty of comparing raw data with reported results is far greater, because the data are not the end result of the investigation. The task of auditing and interpreting data in basic research would require scientists with considerable expertise in the subject matter to interpret most laboratory records. There are other potential problems and limitations to the extension of random or systematic data audits to investigator-' nitiated research. It is possible that the introduction of auditors into the scientific laboratory would diminish spontaneity and creativity In research. The anticipation of visits from auditors charged with the task of - identifying inconsistencies between recorded data and reported results might preoccupy researchers and administrators with defensive recor~keeping, perhaps at the expense of the substance of the research. 27
Furthermore, the introduction of data allots will not identify all dishonest researchers. As the program became known, unscrupulous investigators in some fields might become adept at fabricating their raw data as well as their reported results. Thus, there might still be fraud, but it would be occurring earlier In the process of reporting. Finally, the costs of an aunt program for basic and disco very research are unknown and might be prohibitive if the program is other than very limited in scope, considering the extensive data that would have to be reviewed by skilled scientists. Government policies currently encourage regulatory approaches that provide the least net cost to society (U.S. Office of the President, 1981). To date, the number of cases of scientific fraud that have keen publicly reported is small relative to the amount of research that is conducted. Given that an audit program could have deleterious effects on the research environment, would not detect all cases of fraud, and would be associated with unknown and possibly large costs, the committee does not favor its current intrcOuction into basic or discovery Ask. 4. NIH should adopt professional stand ares to ensure responsible research practices by its intramural scientists. NIH is in a key position to develop and implement responsible research practices in its intramural research program. NIH should adopt professional standards that clarify the basic guidelines to be observed by intramural investigators in the conduct of research. m e intramural research standards should address, but not be limited to, the recording and retention of research data, the training and supervision of young scientists, and authorship and other publication practices. NIH should take the lead in studying the issue of rights and responsibilities of all relevant parties to research data within the intramural program and prepare model guidelines for data sharing and access. NIH should promote the quality and Integrity of research conducted within the intramural program by adopting incentives and guidelines that reduce the pressure to publish. NIH should designate administrative officers or scientific staff members to promote responsible research practices within the intramural program. These individr~As should be available to provide mediation and counseling services for staff who wish to express concerns about questionable training and research practices. They should also work with the office for the promotion of responsible ~· . · · . research co Disseminate gulaellnes developed By the Intramural research program to other public and private research institutions. 5. NIH should adopt policies to limit the newer of publications that can be considered as part of any grant application, in order to emphasize quality over quantity. 28
NIT is in a key position to demonstrate Cat ~ali~ ral:her Can q~anti~ of publications is the proper measure of scientific adhiev~nt in the Oral romaji pr~rmn. qhr~ Be Division of Research Grants, shad sections. and councils. for example. NIH , . , dhadc3 add pot icies ~t 1 emit the namer of pubs ications considers }fly initial review g~6 as part of a grant application. This action, along with the deve~c~nent of a ~eJ sylvan for promotion of responsible relearn Induct, -will demonstrate that NIH is firmly China efforts to promote research q~i~r in Be hearth Icier;. Among ff,e factors that predisEx~e to sloppy or dishonest sole we is the emphasis on quantity of publications that pervades the research community. Although it is clearly not the only such factor in the research environment, it is one that is exacerbated by the competition for research support. NIH, which has been in a pivotal position in the creation of the competitive research environment, can alleviate this problem by developing procedures ~ at reduce the incentive for large bibliographies. FOLIOS FUR UNIVERSE AND aTHER RESE~I CENTERS 6. Universities bedim] schools. and other research organizations should adopt guidelines to clarify the expectations of each institution about the professional StaDaard5 to be ok served by Investigators In the conduct of research. Recognizing that many medico schools and universities have begun to develop policies and procedures to handle allegations of scientific m~soonduct, the committee recommends that all academic institutions, median schools, and research institutes develop guidelines for the responsible conduct of research. These should be developed in consultation with the faculty, research staff, and students and should reflect the standards of accepted practice of the research community. m e guidelines should specify desirable behavior and also provide a basis for identifying unacceptable research practices. At a minimum, these guidelines should be in accord with the requirements discussed In Recondensation 1. In addition to these requirements, the committee encourages institutions to address= other areas in developing their research conduct guidelines. m e committee recognizes the difficulties in reaching a consensus or standard that can be applied by NIH within the next few years to such matters as the appropriate conditions for sharing research data, authorship and publication criteria, regulations governing financial conflict of interest concerns, and efforts to reduce excessive publication pressures, particularly in appointment and promotion decisions. However, Harvard Medical School has recently adopted a set of research guidelines that address many of these issues. A faculty committee at the University of Michigan has also reach ended 29
development of a code for ethical conduct in scholarship. Other institutions, such as the University of Texas at Housbon, have adopted policies on authorship and plagiarism that address selected aspects of these issues. The requirement that all research institutions develop professional research stand ares as a rendition of NIH funding will stimulate much debate and reflection among scientists. The committee believes this discussion itself will improve the conditions under which science is conducted and the environment that shapes the training of young scientists. 7. Universities should provide formal instruction in good research practices. Thls instruction should not be limited to formal courses but should be incorporated into various places in the undergraduate and graduate curricula for all science students. The lack of formal discussion about responsible research practice and the ethics of research is a serious flaw in the professional training of young scientists and clinicians. Many medical schools and universities-= have traditionally avoided this topic and have relied instead upon the faculty to communicate the standards and traditions of research practice through personal example and Restoring. Others have suggested that this information can best be communicated through guest presentations or occasional lectures in traditional science courses. Although these approaches are often useful, they are no longer adequate because of the size and complexity of the modern research environment. The committee believes that instruction ~ the standards and ethics of research is essential to the proper education of scientists. Some organizations, including Sigma Xi and the American Association for the Advancement of Science, have published materials appropriate for this instruction (Sigma Xi, 1986; ChaLk, 1988~. The committee urges~professional organizations to expand their training activities to help and encourage faculty to offer seminars or courses on the standards and ethics of research. Private and public research funding agencies are encouraged to support projects that will create and disseminate model curricula and supporting materials related to the responsible conduct of research. The committee emphasizes that the value of mentoring should not be overlooked in institutional efforts to communicate responsible research practices. The challenge to the universities is to identify ways to support and reward professional mentoring, to ensure that investigators communicate responsible research standards in their interaction with trainees and students, and to supplement this informal communication with instruction designed to expand the student awareness of the ethical and professional dimensions of research work Rut the training experience. 30
8. Universities should designate one or more administrative officers or faculty members to promote responsible research practicer within the institution. The institution should also provide mediation and counseling servi ~ for faculty, staff' and students who wish to express concerns about professionally questionable training or research practices. Universities should not rely upon formal complaints of scientific misconduct as the sole means of monitoring the integrity and causality of the research conducted under their auspices. They need continuing mechanisms to review and evaluate ~ e research and training environment of their institution. They also need personnel who think critically about the Integrity ad quality of the research environment and ways ~ which it calls be iTr~roved. Ibese or other sell irxlividuals shard be available to discuss, on a confidential basis,, incidents that raise questions of j~c or professional behavior or the a~prwriate stains that shard guide ~ conduct. Because even serious ~scorx3uct mav be recognized oTlly by E~r~ mini esridenoe or suspicion, it is important to be Men ~ questions of all levels of aunt seriousness. he ir~ih~ions may wish to designate such an indivi~1 for each collie or major research unit. _ .. . . · . · . · · · . . ~ · . The current use of ombu~spersans by some universities and medics schools deserves consideration as one means of providing mediation and counseling services to those who raise concerns about questionable practices. The organizational factors that pro mote or inhibit the effectiveness of onbudspersons '-- ~ -~---~----~---~---~ deserve further analysis. In cna~lr~ suostar~ru practices Some universities may wish to form faculty committees to monitor the conduct of research within their institution. These committees Thy develop other T~anisms both to identify and correct subs~car~rd practices within the university art to promote stanzas of excellence the performers of research. 9. Universities art offer research institutions should at heathen the integrity and quality of research by modifying incentives and academic guidelines in order to reduce the pressure for excessive publication. Publication is the end result of nearly all scientific research, and researchers are too often evaluated for academic appointments, promotion, and funding primarily by their number of publications. A prolific researcher may have several hundred publications, and evaluating committees cannot possibly read them all. Although some members of evaluating committees may read some of the papers and know the reputations of the researcher and of the journals in which the papers are published, the increasing tendency over the past 30 years or more is to allow the number of publications to serve as an index of quality. 31
Quantity of publications as an into of quakier may be r~sonabJe, insofar as the number of pet ications suggests Dirty and pr~ctivity, arid may be a valid indicator of fire r~a~ - perfo~e. However, as large bibliographies became the norm, it has became necessary to undertake smaller and -her projects to attain the highest possible rate of publication. R#s archers bend to divide a large or complicated study into many parts so that each can be published separately, resulting in multiple, overlapping, and trivial papers. This is also a feature of collaborative studies, which are increasingly common in today's science. The committee believes that the pressure to attain a long bibliography leads to poorer rather than higher quality work. It also leads to the widespread practice of including as authors those who did not contribute substantially to the work. This is sometimes done out of generosity towards a younger colleague, sometimes at the insistence of the laboratory chief, and sometimes because it is thought that including the name of a well-known senior researcher will enhance the chances of publication in a particular journal. Not only does the pressure to publish lead to Me practices of repetitive Publication. trivial work. arm loose authorship. but it mav _ , ~ _ _ , _ _ _ _ , ~ , , _ , ~ . ~ also Apt researchers to engage on more serlcus m~scorK1uct to achieve a publishable result. His miscor~uct may range freon sloppiness, to trimnir~ arm selecting data for more compelling results, to major fraud. How far a researcher is willing to I ~ e along this s ~ tr ~ of misconduct is, of course, a function of personal character as sell as of the external pressures. Cbvicusly, many are willing to engage in forms of repetitive publication although it is likely that only a few are willing to commit fraud. ~· ~· · · ~ ~ ~ ~ _" _ It is of interests however, that several recent highly publicized instances of fraud in biomedical research occurred in very productive laboratories where the head of the laboratory had a long bibliography and there was great emphasis on frequent publication (Woolf, 1986~. One way of dealing with the deleterious effects of excessive publication pressure is to allow only a limited number of publications to be considered for academic appointment, promotion, or funding. Harvard Medical School, which at one time required a researcher to have a minimum number of publications to be considered for appointment as assistant professor on the basic science faculty, now has guidelines suggesting maximum numbers of publications to be considered for promotion or appointment to each faculty level: 5 for assistant professor, 7 for associate professor, and ten for full professor (Tbsteson, 1988~. For such a scheme to have the desired effect of reversing the trend toward greater numbers of publications, it will be essential that the candidate submit a list of only the maximum number of publications allowed (presumably those considered the best) without mentioning others. Only in this way can the emphasis on numbers be changed. 32
It is sometimes objects that this will deprive those evaluating a rewarder freon learnir~ of the indivi - ~1 's worst work. But pi Me T=St important factor thatch be Me best the r~er can c30 and not the worst. the hope is that each stay acid beagle ately more s ~ cantial and that evaluators w ~ d find it rot only possible, but also near to read the limited number of publications. It is difficult to say what the effect of such a system would be on the incidence of misconduct in scientific research. Almost certainly, the p~-actioes of repetitive publication and honorary authorship would be sharply curtailed. as would stockiness and more benign forms of cutting corners. neC~Y;c~rily be much affected, it is likely that in a less cluttered system, misconduct would be easier to detect. , Although major fraud might not 10. Academic departments and research units should monitor the supervisory and training practices of their faculty and research staff to ensure that adequate oversight is provided for young scientists. Several sociological analyses of seleabed professions, such as medicine are] law, have concluded ~t the nest significant determinant of cc~mpliar~ce with professional norm is the social settle of professional practice (Bayles, l9Bl; Marlin, 1966; Freidson, 1970~. In keeping with this firmly, there is a rum new for scientific institutions to across the social ~vin~mner~t of their faculty, staff, and sets are to identify organizational e1 ~ Its, incentives, and barriers that shape their understanding of, and adherence to, responsible research standards. As a first step in implementing this approach, the committee recommends that universities and research centers develop new policies to guide the training and supervisory practices of their research . _ _ ~ ~ ~ _ _ a Ermine. ~ these pollcles sncula require that a primary supervisor He designated for each trainee and that the supervisor provide adequate review of the trainee's research performance. me committee considered, but did not endorse, a proposal that institutions limit the number of trainees assigned to a senior investigator. Although this approach may be desirable, there is insufficient information at this time to suggest that numbers ad one significantly affect the quality of research supervision. m ere is a critical need for more analysis of the organizational and social components of research settings that influence the quality of research performance. There are many different steps that should be considered by department chairs and laboratory chiefs in strengthening the research environment. The most intense surveillance of research practice should occur on an ongoing basis within the confines of the research unit. Research personnel should be explicitly instructed in the 33
prier Cans of designing experiments arm volleying experimental Am. Oratory chiefs chard insist that raw data be preserved within Books and over repositories ~ a way that is readily interpretable by qualified Others beyond those directly responsible for correcting the I. In addition, there should be frequent interaction between the principal investigator of the research unit and the individuals conducting the research. Effective interactions require frequent review and interpretation of raw data. There should also be regular meetings in which the individuals within a research unit are able to scrutinize and critique each other's work. Conversely, laboratory or clinical research practices that encourage cxxlxur~nentalization, secrecy, or isolation within the research unit should be viewed as incompatible with the conduct of good research. Departmental officers and laboratory chiefs should discourage practices in which experimentalists routinely describe their work only to their supervisor and not to their peers. The committee also recommends that universities, especially the medical schools, expand the ~ use of interdepartmental reviews and visiting committees as mechanisms to promote the integrity and quality of research training prc grams. These evaluation procedures are traditionally used in gL-dbuate research prcqrams throughout the universities, but they have not been commonly applied to other research settings, such as medical schools and hospitals that conduct clinical research. 11. Academic departments and research units should adopt authorship policies to improve the publication practices of their faculty, staff, and students. Authorship of a scientific report is a responsibility as well as a privilege. It implies that a person has contributed essentially and substantially to the study and is able and willing to defend the work publicly. This does not mean that such author participated in all parts of the study, but it does mean that all authors have familiarized themselve_ with the general nrin~;nl-= of ~11 antic of the study. ~ _ ~= ~,= ~,=~ Authors who have not substantially contributed to the study are often added to papers, a situation that reflects the importance to each researcher of lengthening a personal bibliography (Hush, 1988~. Sometimes the head of the laboratory, who may have obtained the funding for the study, requires recognition as an author. Sometimes authorship is added willingly, because it is thought that the name of a well-kncwn, senior researcher as author will increase the report's chances of being published. In general there is a tendency to add names as a form of collegiality. 34
leers are different Opinions as to the mind level of participation By to qualify for authorship. For example, ore primal has been suggests by the Intonational Cc~nitt~ of Medics Journal Factors (ICMlE, 1988~: "Authorship Bit ~ d be bass only on substantial contributions to (a) car ~ ption and design, or analysis and interpretation of data; (b) drafting the article or revising it critically for important intellectual content; an] (c) final approval of the version published. Conditions (a), (b), and (c) must all be met." Although these criteria have been accepted by many medics journal cantors, they have not been accepted by the research community as a hole and have been qylestioned by those who think they are too restrictive. A~rdi~ ~ he ISLE criteria, the contribution rest be substantial, hit it need not be to all parts of the study. Incus, sc ~ >ne who conceives of arx] designs a term hy sexy wok d be included as an author even if the indiviarmn did not participate in the analysis and Interpretation. However, merely generating the data, no matter how laborious a task, without involvement in design or analysis does not qualify a person for authorship. Nor does simply acquiring the funding or supervising the laboratory where the work is done, if the investigator is not involved in the substance of the research. the committee did not endorse these particular criteria, but believes that they represent a useful starting point for the development of institutional guidelines. It is important to limit authorship to those who mate substantial contributions. First, it is confusing and deceptive to include supernumerary authors. More important, gift co-authorship diminish== the concept of responsibility and makes it easier for the writer to publish fraudulent data. When criteria for authorship are loose, co-authors may evade all questions about the integrity of the work. If stringent criteria for authorship were widely accepted, this is less likely to happen. Senior researchers can provide important encouragement to junior researchers by designating them as co-authors. However, unjustifiable authorship should not be used as encouragement, no matter how generous the impulse, because it erodes the concept of responsibility. Preferably, a junior researcher should be involved in a study early enough and broadly enough to satisfy the criteria for authorship. The committee considered, but opposes, a proposal that universities review research manuscripts prior to their submission to journals or professional meetings. This review is unnecessary and intrudes upon traditions of academic freedom by suggesting that some form of administrative or departments clearance is required prior to the communication of research findings. The committee believes that scientific journals and research institutions can address the problem of supernumerary authors by 35
establishing criteria for authorship and requiring each author of a report to specify a contribution that satisfies the criteria. In addition, the committee recommends that each academic department and research center maintain an up-to-date record of the publications of its faculty, staff, and students. The deparLmenta~ record will help facilitate colloquial review of published research and will also highlight Native publication p~-ac~ti~. If questions are raised ascot inappropriate assignment of authorship credit, the de~rtrarnt art its institution Uphold take appropriate steps to investigate the allegation. RE~ONS FOR ~E=SIC~L AND SCIENTIFIC O~NIZ=rCNS AND Jay 12. Professional and scientific organizations representing the research community should develop educational and training activities and materials to improve the integrity of research. These organizations should assist universities in identifying substandard research and training practices that compromise the integrity or quality of research. Professional organizations. including the various discinlinarv _ - - - , ~ a e ~ ~ ~ societies, play an important role in developing consensus about the goals and values that should shape research practice. Scientific meetings and journals often provide young scientists with a first glimpse of the issues of importance to the research ccmmNnity as a whole and allow greater insight into the social context of individual or institutional research projects. The committee hats taken note (Chapter 2) of several current projects conducted by scientific and educational organizations that address issues of research misconduct. These include a series of workshops conducted by the National Conference of lawyers and & dentists and the "framework" project currently sponsored by the Association of American Universities and the Association of American Medical Colleges. The committee believes that much more can be done by these and other organizations to promote the responsible conduct of research. Some topics that are suitable for development include the issue of rights and responsibilities with respect to sharing research data and unique biological materials; the professional duties and expectations that shape the career development of young scientists; and the appropriate means of acknowledging scientific achievement and contributions in collaborative efforts. There are many practices that most researchers would recognize as distasteful although they may not be viewed as serious scientific 36
m~scorx3uct. Several parts convent at the workshop suggest emit of Seth practice, inch: misuse of statistics, selective ~n~:~tation of r~h fistic, unjustified seer are r~1ization of research projects, incomplete an of m~ntriblltions f m n colleagues or train s, and inappropriate publication practices, such as divided publication, repetitive publication, or incomplete or inaccurate publication. Most of thee= practices' although deplorable, do not fit the prc posed federal definitions of scientific misconduct and they are not suitable for investigation and punitive action by research institutions. In the long run, however, substandard practices may damage or seriously compromise the Integrity of scientific research a ~ the quality of the research environment. Furthermore, these practices, if left uncorrected, may evolve into research fraud or ocher serious deviations four professional sban~ards (Bosk, 1981). The committee recommends that prof-~-~cional organizations develop forums and publications to identify substandard practices that compromise the integrity or accuracy of scientific research. Once consensus about these practices has been achieved, they should be addressed by NIH and the university research guidelines a ~ oversight mechanisms. 13. Scientific journals should develop policies to promote responsible authorship practices including procures for responding to allegations or indications of misconduct in published research or reports submitted for publication. The committee commissioned one background paper on issues of scientific m~soonduct in connection with authors hip and publication practices (Hush, 1988). mese issues were also addressed by a workshop panel whose members included both research scientists and editors of various research journals. There are a host of topics related to The responsible conduct of research that need to be addressed by scientific journal editors and publishers. Several organizations, including the Council on Biology Factors, the International Committee of M~dim~1 Journal Editors, and the Amari ~ Medical Association, are developing programs and policies that address many of these concerns (Culliton, 1988b; AMA, 1988~. m e topics that require immediate attention by scientific journals include repetitive publication, supernumerary authorship, institutional responsibilities for discI06ure and notification of research misconduct in publication, the use and misuse of peer-review, and the appropriate response to suspicions or confirmations of misconduct in published work or work submitted for publication. The committee endorses proposals made at the workshop that encourage journals to require all authors submitting a manuscript to 37
the journal to sign a statement that defines their individual contributions to He manuscript. Doors of I reports might be as} - explicitly to asinine r~nsibility for the ~ntegrit~r of He data. Alff,a~h the Soiree believes it is praise to require insti ~ tional review and approval of manuscript submissions as a means of assuring integrity in published work, the members conclude that journals can do much more to ensure that authorship criteria are taken seriously. In developing their policies, journals need to work closely with research institutions to encourage good publication practices that will protect the variation that is important for discipline-specific journals. The committ== also believes that journals should disclose to appropriate persons at the research institutions substantial allegations or indications of research misconduct that are detected in the course of peer review. Many journals are presently in the position in which research institutions found themselves before federal regulations required the development of misconduct policies and procedures. The editors are often ill-prepared to deal with cases that are brought to their attention, and they are reluctant to c~11 attention to complaints of unprofessional behavior that have not been formally substantiated. Journals have an obligation to publish retractions of published reports that have been found erroneous by the original authors or that have been declared frand~nent by appropriate authorities at the research institutions. The committee recommends that science journal editors develop a uniform system for reporting serious violations of professional standards to research institutions so that institutional officers can be informed in a timely manner of the nature of these complaints. The committee does not encourage editors of journals to conduct random data audits to ensure the responsible conduct of research (see Recommendation 3~. Editors seldom have the resources or the expertise to carry out such audits. 14. The National Academy of Sciences should pursue the issues and findings developed by the Institute of Medicine In this report and examine their relevance an___eplication to other fields of scientific research. The committee= has examined the issue of the responsible conduct of research with particular emphasis on the health sciences, as requested by the National Institutes of Health. Thrcoghout this study, however, questions have been raise] about the significance and applicability of this issue to other fields of scientific inquiry. 38
Scientific ~sco~ct is not limit ~ bit arid behavioral ~ Cohn, 1986~. Attach serious cast of scientific fray have altars ~st camn~nly in heath sciences r~, there are also examples of Cuba falsification, fabrication, and plagiarism in chemistry, geology, physics, and other fields of scientific and scholarly inquiry (John, 1986; Broad and Wade, 1982~. The committee urges the Academy to develop additional activities to pursue the issues and findings discussed in this report. Topics that seem to warrant immediate attention by the Academy include the following: o Con~rr~s about scientific misconduct ark responsible research practices are currently very visible In the health sciences. Alff~a~h he committee believes that the issues address in this study are relevant to over fields of ~r~, it has rot attempted to judge the value of its ~ ations for all of science. The Academy could play an important role by ebz~u~nung the need for policies and procedure to assure responsible research practices throughout the scientific community and by highlighting factors that differentiate the quality ass~n~n~e programs of different disciplinary fields. O The traditional procedures used by various research institutes and fiel ~ of study to detect and correct errors an] to encourage quality In research and scientific training are not well defined or well understood. IN particular, training evaluation systems (such as 'visiting committees'') deserve further characterization and analysis to determine their strengths and limitations. There is a need to highlight effective methods of professionalization that should be more broadly disseminated t ~ out the research community. 0 The rights and responsibilities swrll~ulling access to research data, methods, and materials, deserve analysis to assist in the development of policies and procedures that affirm the traditional openness of the scientific p-m-cess while Plan nmt£~;na ;mnort~nt intellectual property interests. _ _ ,= ~.= ~ Although the issues of data access and data sharing have been discussed in many settings, there does not yet appear to be real acnsensus over the basic standard Fat dhaald govern individual practice (a Ration 15). Sound policies require thoughtful analysis by exams in Irony disciplines. This area provides an opporiuni~r for the Academy to build upon its earlier work in analyzing data sharing practices In the social sciences and to rY~xn'~end practices that are consistent with the norms of good science (Fienberg et al., 1985~. o With the requirement that there be exposure to professional stdrlilnis and ethical issues in the training of scientists will come the practical need to develop ways to achieve this instruction. The committee seeks to maintain creativity and flexibility In 39
- institutional approaches fulfillir~ ~ his requ~x~. There is a need for an = ~ of existing instructional materials. There is also a new for national ford; to cti~te effective educational materials, incl~irx~ talcs, case st~ies, Video tapes, or booklets:, since few institutions can cla ~ to have the practical expertise that will be require! for sound beaching materials. o Although it is clear that poorly supervised scientists are more likely to develop unacceptable research practices, it is not known with any precision what constitutes adequate supervision. m e perception needs to be a~]rPc~PO that there are large laboratories with absentee chiefs, who are training a correspondingly large number of young scientists and failing to indoctrinate them adequately into the professional as well as the technic stanzas of science. 15. Art interdisciplinary cc~nnitt~ should] be cor~vened to shady the issue of rights ark responsibilities of all relevant parties to rearm data and to prepare model guidelines for data sharing and data access. The ~ntegri~r of science is directly influenced by the verifiability arx] repr ~ ucibili~r of ~ h data. The c ~ nitty= believes that the issues surrounding access to and sharing of research data are especially complex, and the members wish to highlight this topic as one that clearly deserves further consideration. This topic is appropriate for individual or collaborative studies by organizations such as the National Academy of Sciences, the American Association for the AcvanKement of Science, and the Association of American Universities. As mentioned previously, the National Research Council has conducted-a study on data-sharing practices in the social sciences. Sharing Research Data provides a useful reference and good beginning for a broader examination by professional and scientific organizations of the principles that should guide research practice In all fields of science. BAAS has also published an important report on this topic (Science as Intellectual Property) and has hosted several symposia to identify and evaluate current data sharing practices (Nelkin, l983~. ALAS nas also Published an important resort on this tonic The committee believes a sbudy-of the rights and responsibilities of all parties to research data is especially important, given the perceptions expressed by many works hip participants that commercial and entrepreneurial interests as well as competitive academic pressures are affecting traditional data sharing practices. The committee suggests that this study not be limited to the sharing of research data, but include related issues such as the appropriate conditions for sharing research methods and materials. 40
16-. There are many issues that deserve further analysis to enhance institutional and policy efforts to discourage scientific misconduct and to improve the integrity and quality of research. The committee reccmmen~s that professional and scientific organizations initiate studies to understand and encourage responsible research practices. IN its review of the topic of fraud and quality assurance ~ hearth sciences research, the committee identified a number of areas in which information is insufficient to warrant specific policy recommendations. The committee urges professional and scientific organizations, as well as individruH investigators, to undertake studies in these areas to racist in the development of informed policies and procedures to promote the responsible conduct of research. The following topic_ seem most appropriate for further analysis at this time: o The effectiveness of mechanisms currently used to monitor quality ~ Xanadu ate research programs. The proposed study should explore (1) how mechanisms such as visiting committees and research r ~ systems actually work, (2) the experience of selected institutions In using these systems to determine research quality, and (3) the means by which they cculd be introduced into other research settings. o The existence of ethical concerns among research scientists and trainees and The ways in which these concerns are ad~r==sed. A few surveys have been oon~uc~P~ to determine professional awareness of incidents of m~soonduct (Sigma Xi, 1988; Tangney, 1987, 1988~. Much greater understanding is needed of the types and prevalence of ethical concerns including, ~ t not Ignited to, concerns about m~soonduct- among scientists and train ~ . It would be useful, for example, to study the opinions of individual scientists about the manner ~ which institutions develop and monitor risible ~ practices. O the roles ark responsibilities: of r~a~ staff In the laboratory or cl~ni~1 Leant center. He rapid growth ark ~lexit~r of heath scier~es rearm In recent decades has spat t~rxious diversification ark variation within the Rae ~ at. There is same confusion over the role~--ar~ responsibilities~of junior and senior personnel. Studies designed to analyze the expectations of the performance of different personnel in the laboratory and clinical research centers will identify other factors that affect the integrity of research. O The detection and correction of error in science and scholarship. Various disciplines rely on systematic mechanisms, such as the replication studies required by some chemistry journals, to 41
identify research errors. The quality of research would be improved by characterization and analysis of the procedures COPE by research institutions, journals, professional groups, and individual investigators in various disciplines to detect and correct error. ., 42
