NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on Conflict of Interest in Medical Research, Education, and Practice; Lo B, Field MJ, editors. Conflict of Interest in Medical Research, Education, and Practice. Washington (DC): National Academies Press (US); 2009.

Cover of Conflict of Interest in Medical Research, Education, and Practice

Conflict of Interest in Medical Research, Education, and Practice.

Show details

4Conflicts of Interest in Biomedical Research

Biomedical research provides discoveries that may lead to new or better tests and treatments that improve individual and public health. Patients, patients’ families, physicians, other researchers, and policy makers need to trust that the design, conduct, and reporting of such research are unbiased and that the time and effort that they contribute to research will be used to advance science. Participants in clinical trials need to trust that they are not exposed to unnecessary risk. Conflict of interest policies should not only address concerns that financial relationships with industry may lead to bias or a loss of trust but should also consider the potential benefits of such relationships in specific situations.

Research partnerships among industry, academia, and government are essential to the discovery and development of new medications and medical devices that provide improved means for the prevention, diagnosis, and treatment of health problems. Historically, the federal government has taken the lead in supporting discoveries in basic science, whereas commercial firms have focused on the discovery of specific medicines and then their development through clinical trials to the regulatory approval of marketable products. (As discussed below, the development pathway for medical devices often differs from the pathway for pharmaceuticals.) Before 1980, the federal government held the patents resulting from publicly funded basic research, but very few patents were licensed for commercial development. In 1980, the U.S. Congress passed the Patent and Trademark Amendments of 1980 (P.L. 96-517, commonly known as the Bayh-Dole Act, after its sponsors). The law allowed institutions to patent discoveries resulting from federally funded research and to grant licenses for others to develop those discoveries. Universities may retain licensing and royalty fees, which they generally share with their scientists who developed the patented discovery. Since the law’s passage, patent licensing and other financial relationships linking medical researchers and research institutions with industry have expanded substantially (Schacht, 2008). Some scholars, however, have pointed to factors in addition to the legislation that may be associated with the historical increase in the numbers of patents, including a broadening of the criteria that allow materials to be patentable (particularly for life forms) and advances in biomedical research (see, e.g., Mowery et al. [2001, 2004] and Sampat [2006]; see also Diamond v. Chakrabarty, 447 U.S. 303 [1980]).

This chapter starts with a brief overview of some dimensions of university-industry collaborations in biomedical research and then summarizes data on the extent of the relationships between pharmaceutical, device, and biotechnology companies and academic research institutions and individual researchers. The next sections review concerns about these relationships and responses to those concerns. (Appendix E provides an additional discussion of the nature and importance of academic-industry collaboration in medical research.) Because many conflicts of interest at the institutional level emerge from research discoveries, the discussion of these conflicts and the responses to them presented in Chapter 8 is also relevant. The final section of this chapter offers recommendations.

COLLABORATION AND DISCOVERY IN BIOMEDICINE

The path from a scientific discovery to the marketing of a new drug, device, or biological product is typically long and complex and involves a diversity of expertise and resources. For example, basic researchers, often at academic medical centers and other research institutions, can identify new potential targets for therapies and new strategies for treatment, suggest additional diseases that may be able to be treated by existing and newly developed compounds, and suggest both how to target therapies to the patients who are the most likely to benefit and how to avoid particular treatments for patients at high risk for adverse events from those treatments. Scientists at the National Institutes of Health (NIH) also contribute to the discovery process, and important clinical research is undertaken at the NIH Clinical Center. In addition, basic scientists at biotechnology and pharmaceutical companies have made fundamental discoveries that have led to new therapies.

Scientists at pharmaceutical companies can help identify or develop drugs that may be active against new biological targets that have been identified by individuals who conduct basic research. These companies also have the critical ability to use good manufacturing practices to produce a candidate drug in sufficient quantities for clinical trials and then for large-scale commercial distribution, if the product is approved for marketing. Furthermore, they have experience with the Food and Drug Administration (FDA) drug approval process, which includes extensive requirements for preclinical and clinical testing and for manufacturing. Finally, pharmaceutical companies also supply or raise the capital needed to fund the lengthy process of bringing a product to market. Medical device companies and biotechnology companies play analogous roles in translating discoveries made through basic research into products or services for medical and public health practice, although the specific details differ from those involved with the drug approval process. (Appendix E provides a more detailed discussion of the discovery and development process.)

The committee heard testimony that collaboration between academic and industry researchers in the drug discovery process can be mutually beneficial (Benet, 2008; Cassell, 2008). When a new disease mechanism is discovered, academic and industry scientists can work together to identify promising therapeutic targets and treatment approaches. Furthermore, academic researchers can inform industry when they identify potential new targets for chemical intervention. Drug companies can then quickly scan their chemical libraries to search for compounds with potential biological activity and describe what problems they have encountered as they have tried to identify the specific targets of those compounds. This begins the long process of applied chemistry, which is needed to identify a candidate drug.

Many examples illustrate that academic collaboration with pharmaceutical and biotechnology companies can lead to dramatic therapeutic advances that save lives and improve the quality of life. Particularly dramatic are those related to therapies for human immunodeficiency virus (HIV) infection. Collaborations contributed to delineation of the pathophysiology of the disease and the development of successive new classes of drugs, including reverse transcriptase inhibitors, protease inhibitors, and entry inhibitors (Braunwald et al., 2001). These advances have transformed a uniformly fatal illness into a chronic disease that people are now generally able to survive for decades. A few other examples include the following:

  • an anticoagulant (abciximab), which is a monoclonal antibody against the platelet glycoprotein IIb/IIIa, that has been shown to prevent thrombotic complications of coronary angioplasty (EPIC Investigators, 1994; Tcheng et al., 2003);
  • pulmonary surfactant, which improves survival in neonates with respiratory distress syndrome and which was developed by a number of academic researchers at different universities working in close collaboration with several pharmaceutical companies (personal communication, Jeffrey A. Whitsett, Chief, Section of Neonatology, Perinatal and Pulmonary Biology, Cincinnati Children’s Hospital Medical Center, December 9, 2008);
  • rituximab, a monoclonal antibody against the CD20 marker on B cells, which is effective in patients with certain types of lymphoma and leukemia, rheumatoid arthritis, and multiple sclerosis and in preventing the rejection of transplanted organs (Maloney et al., 1997; Edwards et al., 2004; Hauser et al., 2008);
  • bortezomib, a proteasome inhibitor, which improves survival in patients with multiple myeloma (San Miguel et al., 2008); and
  • imatinib, a tyrosine kinase inhibitor, which has greatly prolonged the survival of patients with chronic myelogenous leukemia (Druker et al., 2006).

Compared with the drug development process, the development of complex medical devices tends to be a more continuous process of innovation and refinement that involves frequent alterations in device design, materials, manufacturing processes, or other characteristics. Examples of medical devices that have been developed as a result of close academic industry collaborations include implanted defibrillators (Jeffrey, 2001), prosthetic heart values (Gott et al., 2003), and mechanical ventilators (Keszler and Durand, 2001). Advances in many technologies, such as pulse oximetry for the monitoring of anesthesia and phototherapy for the treatment of disease, highlight the results that may accrue from a combination of research collaboration and communication with senior clinicians about their experiences (Mike et al., 1996; Dicken et al., 2000; McDonagh, 2001; Severinghaus, 2007; Vreman et al., 2008).

Nevertheless, advances in medical devices may result in conflicts of interest. For example, the process of device refinement (particularly when the refinements are minor or are not associated with well-designed clinical studies) is at the center of controversies over whether some consulting arrangements between orthopedic surgeons and the manufacturers of orthopedic devices represent fair payments for technical services or are inducements for the surgeons to use the device.

To promote further progress in moving discoveries from basic science into successful products, NIH has developed major initiatives to strengthen early translational research, which focuses on transforming specific discoveries into clinically useful products or services (see, e.g., NIH [2008d] and CTSA [2009]). At academic centers, this research may involve populations of individuals with rare diseases or biological agents that do not have obvious commercial potential. Such research may, nonetheless, lay the foundation for companies to develop successful products or at least for company licensing of compounds or agents for which university research has provided proof-of-concept data but for which companies must take the next steps.

INDUSTRY FUNDING AND RELATIONSHIPS IN BIOMEDICAL RESEARCH

Growth and Magnitude of Industry Funding

Industry funding for biomedical research has been growing in recent decades and is now the largest source of funding for such research in the United States. Between 1977 and 1989, the proportion of the total funding for clinical and nonclinical research supplied by industry grew from 29 to 45 percent (Read and Campbell, 1988; Read and Lee, 1994). Between 1995 and 2003, the yearly figures (which are based on sources of information somewhat different from those for 1977 to 1985) ranged from 57 to 61 percent (Moses et al., 2005; see also Hampson et al. [2008]). This funding supports work in the laboratories of pharmaceutical, device, and biotechnology companies; contracts for research conducted by universities and other nonprofit research institutions; and contracts with commercial contract research organizations that carry out clinical trials in academic and private practice settings.

Extent of Academic-Industry Relationships

Industry relationships with academic biomedical researchers are extensive. A 2006 national survey of department chairs in medical schools and large independent teaching hospitals found that 67 percent of academic departments (as administrative units) had relationships with industry (Campbell et al., 2007b). In addition, 27 percent of nonclinical departments and 16 percent of clinical departments received income from intellectual property licensing. Among the department chairs, 60 percent had relationships with industry, including serving as a consultant (27 percent), a member of a scientific advisory board (27 percent), a paid speaker (14 percent), an officer (7 percent), a founder (9 percent), or a board member (11 percent) for a company. In some universities, companies fund individual departments, multidisciplinary research centers, or campuswide research programs (Bero, 2008).

For individual academic researchers, studies from the 1990s show that they have widespread relationships with industry. In a 1996 survey, 28 percent of life sciences faculty who conducted research received support from industry sources (Blumenthal et al., 1996a,b). The prevalence of support was greater for researchers in clinical departments (36 percent) than for those in nonclinical departments (21 percent). In a 1998 study, 43 percent of academic scientists in the 50 most research intensive universities reported receiving research-related gifts (independent of a research grant or contract) during the preceding 3 years (Campbell et al., 1998). The most widely reported gifts received from industry were biomaterials used in research (24 percent),1 discretionary funds (15 percent), research equipment (11 percent), and trips to professional meetings (11 percent). Among those receiving gifts, 66 percent viewed them as important to their research.

A study of disclosures at the University of California at San Francisco found that by 1999, approximately 8 percent of principal investigators at the institution reported personal financial ties to the sponsor of a particular research project (Boyd and Bero, 2000). Thirty-four percent of these involved temporary speaking engagements, 33 percent involved consulting relationships, 32 percent involved paid positions on a scientific advisory board or board of directors, and 14 percent involved equity in a firm (more than one type of involvement for a single research project was possible).

Although evidence is limited and not recent, some research suggests that faculty members who have research relationships with industry are more productive in certain respects than faculty who do not have such relationships. One study found that researchers in the former group are significantly more likely than researchers in the latter group to report that they are involved with a start-up company (14 versus 6 percent) or that they have applied for a patent (42 versus 24 percent), have had a patent granted (25 versus 13 percent), have a patent licensed (18 versus 9 percent), have a product under review (27 versus 5 percent), or have a product on the market (26 versus 11 percent) (Blumenthal et al., 1996a). That study also reported that these faculty reported that they had published significantly more articles in peer-reviewed journals in the previous 3 years than faculty without industry funding (15 versus 10 articles) (Blumenthal et al., 1996a). In general, a greater number of biomedical patents should benefit society, since patents are usually a key step in the development of new therapies or diagnostic tests. Likewise, greater publication productivity should, in general, advance scientific knowledge.

The associations reported above do not prove causality. Industry may fund scientists who are more productive or whose research has more commercial potential. Alternatively, industry may provide funding that allows scientists to be more successful commercially and academically, or such support may encourage funded scientists to be more active commercially.

CONCERNS ABOUT RELATIONSHIPS WITH INDUSTRY

Despite their benefits, relationships with industry create conflicts of interest that can undermine the primary goals of medical research. Where there are conflicts, legitimate and serious concerns can be raised about the openness of research and potential bias in the design, conduct, and reporting of research (see, e.g., Gross [2007]). Whether or not the conflicts actually lead to unwarranted secrecy or biased results in particular cases, they have the potential to threaten the reputation of the research enterprise if they are not avoided or identified and managed responsibly.

The review below does not cover marketing activities disguised as research, in particular, so-called seeding trials that companies design to change the prescribing habits of participating physicians rather than to gather scientifically valid information. These studies, which potentially expose study participants to risk without investigating scientifically significant questions, are discussed in Chapter 6.

Industry Funding of Research and Reduced Openness in Science

A fundamental tenet of academic science is that information, data, and materials should be shared. Such sharing could be at risk in academic-industry collaborations. A 2003 National Research Council report identified “the commercial and other interests of authors in their research data and materials” as major obstacles to information sharing (NRC, 2003, p. 1).

A 1995 survey of life sciences faculty in the 50 most research intensive institutions found that 14 percent of those with funding from industry reported that trade secrets had resulted from their research, whereas 5 percent of those without funding from industry did so (Blumenthal et al., 1996a). Trade secrets were defined as information that is kept secret to protect its commercial value. In some cases, this finding may represent the normal and necessary protection of key information prior to the filing of a patent on intellectual property, with the resulting enhanced opportunity for successful commercialization. (Unlike trade secrets, patents require the disclosure of information but protect property interests in a discovery for a defined period.) A 1993 study of academic genetics research found that faculty with research funding from industry were significantly more likely to delay publication of their research results by more than 6 months to allow the commercialization of their research (Blumenthal et al., 1997).

The situation may have changed since the 1993 study cited above because some basic science journals have adopted more stringent policies on data sharing and withholding (see, e.g., NRC [2003], NPG [2007], and Piwowar and Chapman [2008]). In any case, not only journals but also the research institutions themselves could better maintain the integrity of research to the extent that they adopt more stringent policies on data sharing.

A related concern involves access to data. In some industry-supported research, the investigator lacks full access to the study data and depends almost entirely on company statisticians for analysis (Bombardier et al., 2000; Silverstein et al., 2000; Curfman et al., 2005). The conflict in such situations raises reasonable concerns about the integrity of the data. To address this problem, some journals have recently decided not to publish the results of studies funded by industry unless there is full access to the data and independent repetition of the data analyses by academicians or government employees not affiliated with the sponsor (DeAngelis et al., 2001). In addition, many universities have recently added a requirement for access to study data to the terms of their research contracts with industry.

Research Funding from Industry and Pro-Industry Findings in Published Research

Several systematic reviews and other studies provide substantial evidence that clinical trials with industry ties are more likely to have results that favor industry. One meta-analysis found that clinical trials in which a drug manufacturer sponsors clinical trials or the investigators have financial relationships with manufacturers are 3.6 times more likely to find that the drug tested was effective compared to studies without such ties (Bekelman et al., 2003).2 Another meta-analysis that included non-English-language studies found that studies that favored a drug were four times more likely to be funded by the maker of the drug than any other sponsor (Lexchin et al., 2003). A more recent literature review found that 17 of 19 studies published since the preceding two meta-analyses reported “an association, typically a strong one, between industry support and published pro-industry results” (Sismondo, 2008, p. 112). Similarly, another review found that industry-funded studies were more likely than other studies to conclude that a drug was safe, even for studies that found a statistically significant increase in adverse events for the experimental drug (Golder and Loke, 2008).

In addition, a study of materials submitted to the FDA in support of successful new drug applications found that clinical trials with statistically favorable results were almost twice as likely to be published as industry-funded studies that did not have favorable results (Lee et al., 2008). Overall, the results of more than half of clinical trials submitted to the FDA in support of a new drug application remained unpublished more than 5 years after approval of the drug. Furthermore, comparisons of information submitted to regulatory agencies with information on the same trials published in the medical literature have found changes in the ways that the results of the trials were reported so that the published results appeared to be more favorable than the results reviewed by regulatory agencies. Such selective reporting of trial results includes additions of favorable outcomes, deletions of unfavorable outcomes, and changes in the statistical significance of the outcomes reported (Hemminki, 1980; Melander et al., 2003; Chan et al., 2004a; Rising et al., 2008; Turner et al., 2008). Recent requirements for web-based reporting of clinical trial results are described below.

Other studies have found that research funded by industry was more likely to report conclusions that favored the sponsor’s drug, even if the results did not in fact support such conclusions. For example, studies that have examined clinical trials involving specific clinical specialties or particular clinical problems have found an association between industry sponsorship and results that favor industry. Examples include clinical trials of statins for the treatment of elevated cholesterol levels (Bero et al., 2007), breast cancer studies (Peppercorn et al., 2007), clinical trials of new antipsychotic drugs (Heres et al., 2006), and various nutrition-related studies (Lesser et al., 2007; see also Perlis et al. [2005]).

Several possible explanations can be offered for the association between industry support and results that are favorable to the sponsor. First, pharmaceutical and biotechnology companies seek to invest in products that will be shown to be effective and safe; hence, compounds that enter clinical trials have been selected as being likely to succeed. (That is, for-profit companies may be more risk adverse than nonprofit sponsors and fund mostly studies that seem likely to produce favorable results.) Second, investigators might have become persuaded by their own research that a drug is efficacious and, as a result, develop financial relationships with trial sponsors to help promote the future clinical development or use of the drug. Third, industry studies might be less rigorously designed or designed in a way that will bias the findings in favor of a drug, leading to false-positive conclusions that an intervention is effective, or they might be well designed but not actually conducted according to the protocol (Bero and Rennie, 1996; Steinman et al., 2006). Fourth, sponsors may be more likely to fully publish the results of studies with favorable findings (Rising et al., 2008).

The findings of three systematic reviews do not support the suggestion that industry-sponsored trials are poorly designed. They concluded that the quality of industry-sponsored trials is comparable to that of studies funded by other sources (Bekelman et al., 2003; Lexchin et al., 2003; Hampson et al., 2008). The methodologies used in those assessments of the quality of trials did not, however, take into account such issues as the appropriateness of the control intervention, the clinical relevance of the research question, and whether the findings of the studies were fully published (Lexchin et al., 2003; Hampson et al., 2008).

In addition, it is sometimes suggested that journals prefer to publish articles that report positive findings rather than equivocal or nonexistent relationships. Several studies, based on self-reports from the authors of unpublished studies, suggest instead that authors’ decisions to not submit manuscripts with the findings of their studies account for the majority of unpublished studies (Dickersin et al., 1987, 1992; Dickersin, 1990; Easterbrook et al., 1991). Similarly, a more recent study—based on inquiries to investigators about trial results that were not published—suggested that “studies were not published because they were not submitted” (Rising et al., 2008, p. 1568).

Box 4-1 summarizes several incidents that have added to concerns about bias in the reporting of industry-funded studies. Most involve an alleged failure to publish negative findings from industry-sponsored clinical trials or long delays in publication. These incidents involved a number of pharmaceutical companies and different types of drugs. Sometimes the information became known only after legal proceedings led to the disclosure of confidential internal industry documents.

Box Icon

BOX 4-1

Examples of Biased Reporting in Clinical Research. In a pivotal trial of celecoxib for treatment of arthritis, only data on outcomes at 6 months were presented, even though the original protocol called for the trial to be of a longer duration and the (more...)

In addition, systematic reviews that look at meta-analyses rather than individual clinical trials as the unit of analysis also find an association between industry funding and conclusions that favor the sponsor’s product. One study found that industry-supported reviews had more favorable conclusions, noted fewer reservations about the methodological limitations of the trials included, and were less transparent than reviews conducted by the Cochrane Collaboration.3 All seven industry-sponsored reviews recommended the experimental drug without reservation, whereas none of the Cochrane Collaboration reviews did (Jorgensen et al., 2006). Another study, a review of meta-analyses of clinical trials of treatments for hypertension, found that meta-analyses conducted by individuals with financial ties to a single drug company were not more likely than meta-analyses conducted by individuals who received funding from other sources to have results that favored the sponsor’s drug. Financial ties to a single company were, however, associated with favorable conclusions by the authors of the meta-analyses. Among meta-analyses conducted by individuals with financial ties to one drug company, 27 of 49 (55 percent) reported favorable results from the meta-analysis, but 45 of 49 (92 percent) reported favorable conclusions. The authors of the review suggested that there was a “discordance between the data that underlie the results and the interpretation of these data in the conclusions” (Yank et al., 2007, p. 1204).

Thus, although there is little direct evidence that industry sponsorship has led to deliberate skewing of the results or reporting, there are multiple cases in which industry sponsors have withheld important study results and in which the conclusions presented in the reports appear to overstate the study findings. The risk of undue influence in research exists. The risk is particularly relevant in clinical trials, when the prospect of direct harm to patients (as well as research participants) is a more immediate concern than is the case for most nonclinical research. In this case, conflict of interest policies may help prevent an erosion in public confidence beyond that which may result from research that documents bias or the withholding of data.

Ghostwritten research articles also raise concerns about bias as well as the ethics of author attribution. A conflict of interest is inherent in this practice when the industry sponsor has more control over the article than the nominal authors. Chapter 5 discusses ghostwriting and also participation on speakers bureaus and recommends that academic medical centers forbid faculty from accepting the authorship of ghostwritten articles and participation in speakers bureaus.

Terms of Research Contracts

Some academic health centers allow provisions in research contracts that give industry sponsors important control over the reporting of research findings. In a 2004 survey involving academic medical centers, 7 percent of respondents reported that their institution would allow industry sponsors to revise manuscripts or decide whether results should be published, and more than 5 percent reported that they were unsure about the answers to both questions (Mello et al., 2005a). Half allowed the sponsor to draft the manuscript, whereas only 40 percent prohibited that practice. Seventeen percent of the responding institutions reported disputes over control or access to the data from research. Such disputes also figured in some of the incidents cited in Box 4-1 (see, e.g., Rennie [1997] and Kahn et al. [2000]).

Funding arrangements with contract research organizations have also raised concerns about inappropriate control by industry sponsors (Bodenheimer, 2000; Mirowski and Van Horn, 2005; Shuchman, 2007; Lenzer, 2008). For example, the International Committee of Medical Journal Editors (ICMJE) has expressed concern about the role of contract research organizations that conduct the majority of industry-funded trials, often without the protections that many university research contracts require, including rights of access to the source data and rights to publica tion (Davidoff et al., 2001). Although the committee found no systematic assessments or comparisons of bias in research conducted by these organizations, any lack of such controls over unilateral industry influence raises concerns.

Issues Involving Research Participants or Students

As Chapter 3 discussed, academic medical centers vary in their policies on disclosure to research participants of investigator’s conflicts of interest. It also noted that several surveys suggest that participants in clinical trials currently are not highly concerned about investigators’ financial conflicts of interest. Most respondents report that their decision to enroll in a clinical trial would not be greatly affected by learning that the researcher had a financial relationship with the sponsor. Some respondents even believed that “a greater financial interest would make the investigator do a better job” (Weinfurt et al., 2006a, p. 903).

It is not clear, however, whether participants in clinical trials understand how conflicts of interest could potentially compromise study designs and the protection of research subjects or how they could contribute to bias in the reporting of the results—with the possible consequence being harm to future patients. Furthermore, it is not clear that it is reasonable to expect the average participant to understand these issues. In any case, even if research subjects are not worried about conflicts of interest, other important members of the public may be concerned. As noted in earlier chapters, the political and economic support of the research enterprise depends critically on the confidence of the opinion leaders in government, the media, and academia. When they have doubts about the integrity of the enterprise, that essential support may begin to erode.

Concerns have also been raised about how researcher conflicts of interest might affect their advice about or supervision of the research of medical students, residents, fellows, and junior faculty (AAMC, 2008b; AAMC-AAU, 2008). For example, in their recent report on conflict of interest policies in human subjects research, the American Association of Medical Colleges (AAMC) and the Association of American Universities (AAU) noted the potential for the exploitation of these individuals by conflicted senior investigators or advisers. Such exploitation is unethical and also has the potential to bias the design, conduct, and findings of research. Areas that may raise problems with undue influence include decisions about an individual’s inclusion or exclusion from a research project; the focus, design, and conduct of a study; the publication of research findings (including the suppression of publication); and the treatment of intellectual property interests.

RESPONSES TO CONCERNS ABOUT CONFLICTS OF INTEREST IN RESEARCH

Limits on Conduct of Research by Investigators with Conflicts of Interest

As discussed in Chapter 1, much of the impetus for conflict of interest policies in universities stems from concerns about industry-funded biomedical research and investigators who have financial stakes in the outcomes of their research. In 1995, the U.S. Public Health Services (PHS) issued regulations that require institutions receiving PHS research funding to develop conflict of interest policies that require the disclosure and management of certain financial relationships between researchers and industry (see Appendix B). Chapter 3 noted reviews by NIH and others that questioned the adequacy of policy adoption and the implementation of the PHS regulations by research institutions, which in turn, raised additional concerns about the adequacy of government oversight of institutional compliance.

Because the PHS regulations were not specific on many issues and because some studies indicated shortfalls in their implementation, AAMC issued a report in 2001 with recommendations to help academic medical centers develop sound conflict of interest policies for research involving human subjects (AAMC, 2001).4 A key policy recommendation called for institutions to establish a “rebuttable presumption” that researchers may not conduct research involving human participants when they have a financial stake in its outcome. This presumption can be rebutted when compelling circumstances justify the researchers’ involvement.5

A 2003 AAMC survey indicated that only 61 percent of medical schools had adopted the rebuttable presumption in their policies (Ehringhaus and Korn, 2004). In addition, only a minority of the medical schools with such a policy had defined the compelling circumstances that would support an exception.

To further promote the adoption of conflict of interest policies governing research involving human participants, AAMC joined with AAU to issue a second report that offered additional guidance and support for policy development and implementation (AAMC-AAU, 2008). The report reemphasized the importance of the rebuttable presumption. It also presented informative case studies and a template for analyzing these cases to illustrate how different situations can be evaluated for the existence of a conflict of interest, the risks presented by the conflict, the options for eliminating or managing a conflict, and the compelling circumstances that might justify the participation of an investigator with a conflict of interest in research with human participants. Among the examples of risks cited in the template is the extent to which the reputation of the researcher with a conflict of interest or his or her institution could be damaged, even if a plan for managing the conflict is created and implemented.

Unlike the PHS regulations that cover both clinical and nonclinical research, the 2008 AAMC-AAU recommendations focused on clinical research. One recommendation did, however, call for medical center conflict of interest committees to review investigator conflicts of interests in certain nonclinical studies. Examples include those that can be “reasonably anticipated … to progress to research involving human subjects within the coming 12 months” (p. 9).

The committee found much less information and analysis about conflict of interest policies affecting nonclinical biomedical research than about policies affecting clinical research. Universities and medical schools may have different policies for different kinds of research or may apply different criteria to evaluate conflicts of interest in research that does not involve humans (as reported in Chapter 3). One university, however, recently adopted a conflict of interest policy that explicitly states that “[t]o protect against the risks that may accompany relationships with Interested Businesses, it is not ordinarily allowable for an Individual who has a Significant Financial Interest in an Interested Business to Conduct Research involving that Interested Business” (Columbia University, 2009).

Although an immediate risk to research participants does not exist in basic research, the potential for bias in basic research does exist. The result could be the initiation of clinical trials based on flawed basic science. In general, a weighing of risks against expected benefits should allow conflict of interest committees to apply policies while taking into account differences in clinical and nonclinical research, including differences in what constitutes a reasonable justification for researchers to be involved in research in which they have a financial stake.

Terms for Research Contracts

AAMC has not proposed comprehensive formal recommendations on the terms of research contracts with industry, but it has issued two reports with suggestions and recommendations that respond to concerns about the integrity of clinical trials (Ehringhaus and Korn, 2004, 2006). The first report provides a checklist of topics, including publication rights and intellectual property, to be covered in research contracts. Among other elements, one or both reports call for contracts to explicitly grant researchers free access to study data, to include no restrictions on publication (except for a slight delay for sponsor review and possible filing of a patent application), and to require a good faith and timely effort to publish the results of research in a peer-reviewed journal.

Requirements to Register and Report on Clinical Trials

Congressional, journal editor, and other requirements for the registration of clinical trials are, in part, a response to concerns about conflict of interest in industry-sponsored research and research reporting. The registration of clinical trials and the provision of key details about the trial protocol and the data analysis plan ensure that basic methods for the conduct and analysis of the findings of a study as well as the primary clinical end points to be assessed and reported are specified before the trial begins and before data are analyzed. The substitution of ad hoc or secondary end points for primary end points and other important departures from the protocol can thus be detected in reports of the findings of a trial. Clinical trials registries also allow others to determine whether the results from a trial have not been presented or reported at all. Researchers carrying out critical literature reviews can then contact the investigators to try to obtain unpublished results. After ICMJE stated that clinical trial registration should be considered a prerequisite for the publication of research articles, the numbers of trials registered increased substantially (Zarin et al., 2005).

In 2007, the U.S. Congress expanded the types of clinical trials of drugs, biologics, and devices—and the kinds of information about these trials—that must be registered (P.L. 110-85). To further address the problem of withholding negative findings, it also required the creation of a link from the ClinicalTrials.gov registry to a database of reports of basic results for applicable trials.6 The reported results are to include basic demographic and baseline information, findings for primary and secondary outcomes, and a point of contact.

Study Methodology, Data Analysis, and Research Reporting

To the extent that the design of clinical trials is standardized and publicized, the implementation of conflict of interest policies is also assisted. Abuses and patterns of abuses can be more readily detected, which may make more evident the need for changes or reforms in the policies. Efforts to improve the design of clinical trials and other types of research stretch back decades and include a range of techniques, including the random assignment of subjects to intervention and control groups and the blinding of investigators and participants to treatment assignment. In addition, NIH has supported programs to train physician investigators to conduct rigorous clinical research. Experts in research methodology, statistics, and evidence-based medicine have developed techniques to limit bias in research and have codified standards and checklists for reporting research findings. These standards and checklists cover various types of studies, including clinical trials (see, e.g., Moher et al. [2001]), evaluations of clinical tests (Bossuyt et al., 2003), epidemiological studies (see, e.g., von Elm et al. [2007], but see also the comments of the editors of Epidemiology [Editors, 2007]), and meta-analyses (see, e.g., Moher et al. [1999] and Stroup et al. [2000]).

ICMJE now specifies a format for the reporting of results and refers authors to the CONSORT checklist for the reporting of the findings of randomized clinical trials (see, e.g., Moher et al. [2001], CONSORT Group [2007], and von Elm et al. [2007]) (Table 4-1). Standards for the reporting of methods and results help editors, reviewers, and readers assess the validity of a research paper. Studies suggest that these standards also improve the design and conduct of the research itself (see, e.g., Plint et al. [2005]).

TABLE 4-1. Checklist for Reporting Clinical Trials from CONSORT 2001 Statement.

TABLE 4-1

Checklist for Reporting Clinical Trials from CONSORT 2001 Statement.

In addition to these standards for the conduct and reporting of the results of clinical trials, the FDA has suggested that it is desirable for the data-monitoring committees for clinical trials to have statistical reports prepared by statisticians who are independent of the trial sponsors and clinical investigators (FDA, 2001). For industry-funded clinical trials “in which the data analysis is conducted only by statisticians employed by a company sponsoring the research,” the Journal of the American Medical Association requires that a statistical analysis also be conducted by an independent statistician at an academic institution, such as a medical school, academic medical center, or government research institute, that has oversight over the person conducting the analysis and that is independent of the commercial sponsor (Fontanarosa and DeAngelis, 2008, p. 95; see also a review of opinions about this requirement in Rockhold and Snapinn [2007]).

Peer Review and Journal Policies on Disclosure

Peer review is a key step used to detect and reduce bias in publications and improve the quality of research reporting. Effective review depends on independent reviewers who are not biased by their own financial relationships with industry. As described in Chapter 3, journals vary in the extent to which they apply conflict of interest policies to reviewers. Meaningful peer review is also assisted by the previously described standards for the reporting of methods and data in manuscripts.

In response to concerns about the reporting of research results described earlier in this chapter, medical journals have moved toward increasingly specific requirements for disclosure of authors’ financial interests (see ICMJE [2008] and WAME [2008] for the statements of two associations of medical journal editors). Still, as described in Chapter 3, journal policies remain variable. The completeness and accuracy of disclosures are continuing issues for medical journals as well as for academic medical centers and other institutions. These concerns have led to action in some states and recommendations for the federal government to establish a policy that requires companies to report payments to physicians, researchers, and institutions, as outlined in the preceding chapter. Chapter 3 includes a committee recommendation supporting such a program.

Issues Involving Research Participants and Students

As described in Chapter 3, AAMC recommended in 2001 and again in 2008 policies that require some form of disclosure of investigator conflicts of interest to research subjects, and many medical schools have adopted those policies. Chapter 3 also reviewed some of the findings from a set of coordinated research projects and activities to investigate the views of research participants and ways of informing them. This research is itself a major response to concerns about practical and ethical issues in managing conflicts of interest in research, for example, balancing the disclosure of information with the design of an informed consent form and process that does not overwhelm research participants.

AAMC has also recommended the disclosure of investigator conflicts of interests to other members of the research team. It also advised that schools prohibit “agreements with sponsors or financially interested companies that place restrictions on the activities of students or trainees or that bind students or trainees to non-disclosure provisions” (AAMC, 2001, p. 20). In a later statement about the responsibilities of biomedical graduate students and their advisers, AAMC states that advisers should “recognize the possibility of conflicts between the interests of externally funded research programs and those of the graduate student” and should commit that those conflicts will not be allowed to interfere with the student’s thesis or dissertation research (AAMC, 2008b. 6). The research adviser also agrees to discuss authorship policies and intellectual property policies related to disclosure, patent rights, and publication. In addition, in a series of questions that should be asked when assessing the risks of allowing an investigator with a conflict of interest to conduct research with human participants and the possibility that a conflict can be appropriately managed, the AAMC-AAU report includes questions about whether the “the roles of students, trainees, and junior faculty and staff [are] appropriate and free from exploitation” and whether special protections are needed for “vulnerable members” of the research team (AAMC-AAU, 2008, pp. 25 and 28, respectively). One protection might be to provide such individuals with access to independent senior faculty members for independent review and guidance when questions and concerns arise.

RECOMMENDATIONS

Relationships between industry and research institutions and researchers are common and are often mutually beneficial. They also serve society by generating valuable preventive, diagnostic, and therapeutic products. At the same time, these individual and institutional relationships have risks that could jeopardize the integrity of scientific research and conflict with the ethical conditions for the conduct of research with humans. Analyses indicate that they are associated with decreased openness in sharing data and findings, and cases in which negative findings are not published in a timely fashion or at all raise concerns. Some studies also suggest that meta-analyses sponsored by a single company tend to present conclusions favorable to industry sponsors even when the actual findings of the analyses are not favorable. Moreover, when investigators themselves have a financial stake in the outcomes of their research, it creates conflicts of interest, which may lead to bias and the erosion of confidence in the research enterprise.

Chapter 2 discussed why conflicts of interest matter even if they do not actually lead to undue influence or bias in a particular case. Correlations or associations in studies such as those reported here are enough to support concerns over potential conflicts of interest. The purpose of conflict of interest policies is preventive: the policies are intended to remove or reduce relationships that create a risk of undue influence or erosion of confidence in the research enterprise.

As described in this chapter and in Chapter 3, research institutions vary in their conflict of interest policies, including the extent to which they have adopted and implemented PHS conflict of interest regulations and policies recommended by AAMC and AAU. Government and press investigations and payment data reported by companies have revealed failures of individual researchers to fully and accurately disclose their financial relationships with industry, as required by institutional or government policies.

The preceding section of this chapter provided an overview of recommendations for action that should be taken by research institutions, research sponsors, investigators, and medical journals to protect the integrity of biomedical research, safeguard research participants, and preserve public trust. The recommendation below focuses on one specific concern: the conduct of research with human participants by investigators with a financial interest in the outcome of that research. The discussion of the recommendation is followed by a review of standards for nonclinical research and a suggestion that NIH take a lead role in further examination of the involvement of conflicted investigators in this kind of research.

Clinical Research

It is critical that the public trust that research institutions are protecting the integrity of the medical research on which clinical practice and education depend. Such protection is especially important in clinical research because bias in the design, conduct, or reporting of the findings of such research may expose human participants to risks without the prospect of gaining valid, generalizable knowledge and may ultimately expose much larger numbers of patients to ineffective or unsafe clinical care.

Recommendation 4.1 calls for research institutions to allow researchers with a conflict of interest to conduct research involving human participants only when a researcher’s participation is truly essential and is also managed to limit risk. This recommendation is similar to the AAMC “rebuttable presumption” described earlier in this chapter.

RECOMMENDATION 4.1 Academic medical centers and other research institutions should establish a policy that individuals generally may not conduct research with human participants if they have a significant financial interest in an existing or potential product or a company that could be affected by the outcome of the research. Exceptions to the policy should be made public and should be permitted only if the conflict of interest committee (a) determines that an individual’s participation is essential for the conduct of the research and (b) establishes an effective mechanism for managing the conflict and protecting the integrity of the research.

This recommendation covers principal investigators and others who share substantial responsibility for the design, conduct, or reporting of the findings of clinical studies. Relevant financial interests often involve stock or other ownership in a company making a product that could be affected by the results of a study, including not only a product under study but also a product that is an alternative to the intervention under study. (Although AAMC recommended no minimum threshold for the initial disclosure of financial interests, it suggested that “significant interest” should generally be defined as a financial interest of $10,000 or more.)

In exceptional cases, a clinical investigator may be judged to be essential if his or her participation is determined—after careful assessment—to be necessary for the safety, reliability, or validity of the research, circumstances that AAMC described as compelling. Often cited as examples are situations in which inventors of a medical device or investigators responsible for certain kinds of breakthrough scientific discoveries are crucial to research, especially early-phase studies, because of their “insights, knowledge, perseverance, laboratory resources” or access to “special patient populations” (AAMC-AAU, 2008, p. 6; see also Witkin [1997] and Citron [2008]).

A specific example of a compelling situation might involve the participation in a pilot study of the inventor of an implanted medical device that requires a complex, new surgical procedure that has not been mastered by others. The reasons for allowing a researcher with a conflict of interest to participate in a pilot or early-phase study or other investigation in a particular situation should be persuasive to others who are presented with the facts of the case. In most cases of a conflict of interest, no compelling argument that the investigator’s participation is essential can be made. Even if the investigator’s participation is essential, the elimination of the conflict of interest (e.g., through the sale of stock) is the preferred step. If an exception is granted, it should be made public.

If an exception is made for an investigator with a conflict of interest, the next step is for the conflict of interest committee to establish a strategy for managing the conflict and a plan for monitoring the strategy’s implementation during the course of the research. For instance, the plan might specify that the researcher with the conflict of interest not serve as the principal investigator. It might also restrict the researcher recruiting subjects; obtaining informed consent; assessing the clinical end points; analyzing data; or writing the results, conclusions, and abstracts for publications reporting the findings of the study. The plan might, however, allow the researcher to participate in aspects of study design, fund raising, and manuscript review.

Nonclinical Research

Most of the discussion of conflicts of interest in research has focused on clinical research. This emphasis reflects concerns that research participants might be harmed or that bias might contribute to the making of incorrect decisions about approving new drugs and devices or changing clinical practice. Because conflicts of interest in various kinds of nonclinical research have been little investigated, the committee found it difficult to evaluate arguments about the extent and the consequences (or the lack of consequences) of investigator and institutional conflicts of interest in this sphere of research. It thus did not make a formal recommendation about conflicts of interest in nonclinical research. The committee did, however, hear testimony that new models of academia-industry collaboration are needed to promote basic scientific discoveries and the development of new therapies while also addressing concerns about conflicts of interest (Moses, 2008; see also Moses and Martin [2001]).

No matter the type or stage of research, certain fundamentals still apply. All researchers should be subject to an institution’s disclosure policies, as described in Chapter 3, and the institution’s conflict of interest committee or its equivalent should be notified when investigators have financial stakes in the outcomes of their research. Similarly, following the conceptual framework presented in Chapter 2, once a financial relationship or interest has been disclosed, it should be evaluated for determination of the likelihood that it will have an undue influence that will lead to bias or a loss of trust. If a risk is judged to exist, a conflict of interest committee might conclude that the implementation of safeguards is necessary. Such safeguards could consist of a management plan that includes the involvement of a researcher without a conflict of interest in certain aspects of the research and disclosure of the conflict to coinvestigators and in presentations and publications.

Additional studies on the extent of financial relationships in nonclinical research and their consequences, as well as the consequences of conflict of interest policies, are needed to establish a sounder base of evidence for future policies. Given its extensive and direct relationships with basic scientists, NIH could play a central role in gathering such evidence. As discussed in Chapter 9, NIH could fund research on conflicts of interest in nonclinical scientific research. Furthermore, NIH could convene working groups and public meetings to promote a fuller understanding—empirical, conceptual, and practical—of conflicts of interest in nonclinical research and propose responses. Such meetings might identify good practices in developing academia-industry relationships in nonclinical research and suggest how such relationships might be developed in ways that promote constructive collaboration while appropriately addressing concerns about conflicts of interest. The development of illustrative case studies might help institutions better understand and manage conflicts of interest in nonclinical research.

Other Relevant Recommendations in This Report

The adoption of the recommendations made elsewhere in this report would also affect researchers, research institutions, and companies. These recommendations call for standardization of the procedures used to disclose conflicts of interest to harmonize the requirements of different institutions and reduce the disclosure burdens on individuals (Recommendation 3.3), implementation of methods for the easier verification of certain financial disclosures (Recommendation 3.4), limitations on certain relationships with industry (e.g., acceptance of gifts and participation in promotional activities) for academic medical center personnel (Recommendation 5.1), and promotion of reforms in industry policies on consulting and research grants (Recommendation 6.2).

Chapter 8 includes a recommendation that responsibility for the oversight of institutional conflicts of interest be lodged in the governing boards of institutions (Recommendation 8.1). Many conflicts of interest at the institutional level involve research or proposed research in which a university or medical school has a financial stake related to its interests in patents or start-up companies.

In addition, the committee recommends that other public and private organizations create incentives to support the adoption of the recommendations made in this report (Recommendation 9.1). As one example, NIH could expand its recent efforts to provide more guidance and oversight to grantee institutions covered by the PHS regulations, issue regulations directing grantees to adopt institutional conflict of interest policies (Recommendation 8.2), and take a lead role in the development of a research agenda on conflict of interest (Recommendation 9.2). NIH could also consider requiring investigators funded by NIH awards to be trained on conflict of interest principles and policies. (NIH has a new training module on conflict of interest that could be tailored for investigators.) Other public agencies that support academic biomedical research, for example, the U.S. Department of Defense, could also provide guidance compatible with that presented in this report.

Taken together, the changes recommended here should not burden socially valuable collaborations between academic researchers and industry. Rather, they should help justify and maintain public trust in their integrity.

Footnotes

1

One reviewer of the report observed that companies view the provision of these proprietary materials as a service to the academic community and that they may, in any case, not have a mechanism for charging for them.

2

“A study was included if it met the following criteria: (1) its stated primary or secondary purpose was to assess the extent, impact, or management of financial relationships among industry, investigators, or academic institutions; (2) it contained a section describing study methods; (3) it was written in English; and (4) it was published following the passage of the Bayh-Dole Act of 1989” (Bekelman et al., 2003, p. 455). “The main outcomes were the prevalence of specific types of industry relationships, the relation between industry sponsorship and study outcome or investigator behavior, and the process for disclosure, review, and management of financial conflicts of interest” (p. 454).

3

The Cochrane Collaboration describes itself as “an independent, nonprofit, international organization that develops and disseminates systematic reviews of health care interventions and promotes the creation and use of evidence to guide clinical and policy decisions” (see http://www​.cochrane.org/docs/descrip.htm). It relies primarily on volunteers who conduct reviews according to specific standards. It has policies intended to limit bias and restrict financial conflicts of interest in its activities.

4

As noted in Chapter 1, this report generally follows the practice of recent Institute of Medicine reports in referring to research participants rather than to research subjects.

5

In the words of the AAMC report, the rebuttable presumption means that an “institution will presume, in order to assure that all potentially problematic circumstances are reviewed, that a financially interested individual may not conduct the human subjects research in question” (AAMC, 2001, p. 12). The report goes on to say that the “rule is not intended to be absolute: a financially interested individual may rebut the presumption by demonstrating facts that, in the opinion of the COI [conflict of interest] committee, constitute compelling circumstances … [and] would then be allowed to conduct the research under conditions specified by the COI committee and approved by the responsible IRB [institutional review board]” (p. 12).

6

In addition, the Pharmaceutical Research and Manufacturers of America has coordinated the creation of a voluntary online resource to provide information to physicians about the results of clinical trials (see http://www​.clinicalstudyresults.org/).

Copyright © 2009, National Academy of Sciences.
Bookshelf ID: NBK22940

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.2M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...