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.
3 Policies on Conflict of Interest: Overview and Evidence Current conflict of interest policies and practices have evolved over more than four decades of increasing relationships with industry in medical education, research, and practice. The increase has been accompanied by intensifying discussions about how the risks and the expected benefits of these relationships should be evaluated and balanced. Since 1995, the U.S. Public Health Service (PHS) has required most research grantees to establish policies and procedures to ensure that the design, conduct, or reporting of research funded by PHS grants not be âbiased by any conflicting financial interest of an Investigatorâ (42 CFR 50.601). The regulations, which are included in Appendix B, allow grantees considerable discretion in formulat- ing policies and procedures. To provide more specific and comprehensive guidance to academic institutions on conflict of interest policies, the Asso- ciation of American Medical Colleges (AAMC, 2001, 2002, 2008c), the As- sociation of American Universities (AAU, 2001), AAMC and AAU jointly (AAMC-AAU, 2008), and the Council on Government Relations (COGR, 2002) have issued several reports with recommendations. The Federation of American Societies for Experimental Biology (FASEB) created a conflict of interest tool kit that offers extensive online resources and guidance for aca- demic institutions, researchers, academic and professional societies, journal editors, and industry (FASEB, 2008). In 2008, the trade associations repre- senting major pharmaceutical and medical device companies revised their codes on company interactions with health care professionals (AdvaMed, 2008; PhRMA, 2008). In addition, a number of academic medical centers, professional societies, medical journals, and other institutions have revised their policies in recent years. Criticisms of current policies and their application come from different directions. Some object that policies requiring the disclosure of financial 62
POLICIES ON CONFLICT OF INTEREST 63 interests can be carried too far, encouraging âreaders to make ad hominem judgementsâ (Rothman, 2001, p. 1275) or shifting âattention away from the merits of the work and toward the biography of its authorâ (Jansen and Sulmasy, 2003, p. 40). Another critic describes disclosure policies as a kind of ânew scientific McCarthyismâ that assumes that researchers with industry ties are âtainted and untrustworthyâ (Whelan, 2008, p. A19). One researcher has criticized âconflict of interest vigilantesâ who âsearch for evidence that doctors have failed to disclose corporate connections in publications or in presentationsâ (Stossel, 2007, p. 59). He has also argued that continuing medical education disclosure policies mainly serve to protect bureaucrats rather than students, are based on ideology rather than evidence, and âare deeply disrespectful of physicians and researchersâ (Stossel, 2008, p. 476). (See Chapter 1 for additional criticisms.) Others, however, argue that conflict of interest policiesâwhen they existâare often weak, inconsistent, and inadequately administered and enforced. For example, the American Medical Student Association (AMSA) assessed the conflict of interest policies of medical schools and concluded that the policies of the majority of the schools that responded either lacked important elements or were unlikely to influence behavior (AMSA, 2008b). Whether or not one agrees with how AMSA rated the policies, the actual texts of the policies (available at or through the AMSA website) reveal considerable variability, which is consistent with the findings of this report. Members of the U.S. Congress have strongly criticized physicians and re- searchers who have failed to report substantial financial relationships with industry, as they were required to do, and have proposed that pharmaceuti- cal and medical device companies be required to report publicly their pay- ments to physicians (see, e.g., Grassley [2008b, 2009]). Also in response to concerns about the nature of financial ties between physicians and industry and the lack of disclosure of such ties, Massachusetts enacted legislation in 2008 that requires companies to report payments to physicians, researchers, and medical societies and further provides for a marketing code of conduct â In AMSAâs assessment, 9 medical schools received a rating of A and 19 received a rating of B for their policies; 44 schools received a rating of F (18 for the contents of the policies that they submitted, 9 for their refusal to submit policies, and 17 for their lack of a response after repeated requests). Another 46 schools had policies under revision. (The numbers of schools are based on the ratings listed as of February 13, 2009, at http://www.amsascorecard.org/.) The projectâs methodology, included the rating system, is available at http://amsascorecard. org/methodology and states that â[e]ach policy was graded by two independent assessors, blinded to the institution of origin. Any differences in scoring between the two assessors were resolved by a consensus process. The assessors received formal training in the use of the scor- ing system, independently evaluating and coming to a consensus on five training policies before beginning to evaluate the medical school policies.â
64 CONFLICT OF INTEREST that will prohibit or limit certain of these payments (Wallack, 2008; Lopes, 2009). This chapter outlines the basic elements of conflict of interest policies, reviews empirical data about the characteristics and consequences of those policies, and concludes with recommendations. Much of the research and descriptive information located by the committee examined the policies of academic institutions and medical journals; but the recommendations apply broadly to all institutions engaged in medical research, medical education, clinical care, or practice guideline development. The specific elements of the policies may vary according to the size, complexity, and other charac- teristics of different types of institutions (e.g., academic medical centers, professional societies, patient advocacy groups, and nursing homes). The focus in this chapter is on policies affecting individuals, primarily physicians and biomedical researchers (as explained in Chapter 1). Chapter 8 examines and makes recommendations about policies that govern institu- tional conflict of interest, which is defined to include the interests of senior institutional officials. OVERVIEW OF CONFLICT OF INTEREST POLICIES Most conflict of interest policies include the basic elements of the dis- closure of financial relationships, the prohibition of certain relationships, and the management of conflicts of interest that have been identified. All of these elements are sometimes described under the general rubric of man- aging conflicts of interest. Other common elements of conflict of interest policies include definitions, specification of who is subject to the policies, enforcement provisions, and identification of which officials or units within an organization are responsible for administering and monitoring conflict of interest policies and procedures. Depending on the circumstances and the type of institution, the person responsible for reviewing initial disclosures may be a department chair, the chair of a professional society committee developing practice guidelines, the editor or deputy editor of a journal, or the chair of a continuing medical education program. When an initial review identifies a possible conflict of interest, the case may be referred to a conflict of interest committee or a more senior official for further evalu- ation and response. Building on Chapter 2, Box 3-1 outlines a conceptual model of the â PHS rules refer to procedures to âidentify and manage, reduce, or eliminate conflicting interests.â Federal government policies for its employees are sometimes described in terms of the ââthree-Dâ method of conflict of interest regulation, that is: disclosure, disqualification and divestitureâ (Maskell, 2007, p. 3). Disqualification includes recusal from participation in a specific decision.
POLICIES ON CONFLICT OF INTEREST 65 BOX 3-1 Model of Steps Used to Identify and Respond to a Conflict of Interest Step 1 Obtain the disclosure of information about financial and other relation- ships that could constitute a conflict of interest. â No relationships reported: stop. Relationships disclosed: go to Step 2. Step 2 Evaluate the disclosuresâin light of the individualâs responsibilities or specific activities (e.g., research, teaching, and patient care)âto determine whether a conflict of interest exists. If necessary, collect additional information to assess the likelihood of undue influence and the seriousness of possible harms. â No conflict exists: stop. Conflict exists: go to Step 3. Step 3 Determine whether the relationship is one prohibited under institutional or other policies or whether the risks of the relationship are so serious that the individual should either eliminate it or forgo participation in the activity put at risk by the relationship. â Conflict elimination necessary: go to Step 5.â â Elimination not necessary: go to Step 4. Step 4 If management is appropriate, devise and implement a plan to manage the conflict. Go to Step 5. Step 5 Monitor conflict elimination or management plan and assess adherence. â Plan followed. Plan not followed: go to Step 6. Step 6 Determine the nature of the noncompliance and the appropriate re- sponse (e.g., education, penalty, or revision of the plan) and implement the response. steps that institutions with a comprehensive conflict of interest policy and implementation strategy might follow when determining whether an indi- vidual has a conflict of interest and, if so, how to respond. It shows the elimination of an identified conflict of interest as an early step, although the committeeâs experience suggests that the elimination of a conflicting relationship is often considered a last option. A given individual may be covered by several conflict of interest poli- cies. For example, a medical school faculty member may have to understand and follow the policies not only of the medical school but also those of
66 CONFLICT OF INTEREST several other institutions. Depending on his or her activities, these other policies might include those of a medical journal, a provider of continuing medical education, a professional society, or a federal advisory committee. If a faculty member is engaged in research to support an application for marketing approval of a medical product by the Food and Drug Admin- istration (FDA), the researcher can expect the studyâs sponsor to ask for the disclosure of his or her financial interests related to the company and the investigational product so that the sponsor can submit the required information to the FDA (FDA, 2001). (A recent report by the Office of the Inspector General [OIG] of the U.S. Department of Health and Human Services criticized the administration of these policies and indicated that they were deficient in several respects [OIG, 2009].) Private organizations that fund research, such as the Howard Hughes Medical Institute, also may have conflict of interest policies, which they may oversee directly rather than following the practice of the National Institutes of Health (NIH) of delegating most administrative responsibility to the research institution (Cech and Leonard, 2001). In addition, the faculty of public institutions will likely be covered by state conflict of interest policies. The committee found few reviews or studies documenting and com- paring the conflict of interest policies of institutions engaged in medical research, medical education, or clinical care. It found even less information about the implementation and effects of these policies. Most studies exam- ine the policies of academic institutions, medical and scientific journals, or government agencies. Journal articles or news stories sometime report on individual professional societies and patient or consumer groups. In addition, through its literature review, public meetings, and other information-collecting activities, the committee identified various examples of institutional policies. Although these examples are not necessarily repre- sentative, they helped the committee better understand the nature of policy variability and, in some cases, the rationale for policy differences. Institu- tions differ considerably in the conflict of interest policy information that they make public on their websites; and even if they are available, online information is not necessarily comprehensive, clear, or current. Since the committee began work, a number of medical schools, professional societies, and other groups have announced changes in their conflict of interest â The state of Washington recently changed its policies on the use of certain university resources for outside work for faculty and some other university employees to âencourage the ethical transfer of technology for the economic benefitâ of the state (University of Wash- ington, 2008). â During the study, the committee benefited from initiatives by AMSA and the Institute on Medicine as a Profession to make medical center policies available online. These databases have been useful, although they are not complete, and many schools have indicated that they are updating their policies.
POLICIES ON CONFLICT OF INTEREST 67 policies and practices. Thus, even relatively recent overviews of conflict of interest policies may be somewhat out of date. DISCLOSURE: AN ESSENTIAL BUT INSUFFICIENT ELEMENT OF POLICY Disclosureâthat is, revealing to others information that may otherwise be private or confidentialâis a frequent response to concerns about con- flicts of interest in various sectors of society. Disclosure by physicians and researchers to their academic or other institution is essential because insti- tutional officials cannot evaluate and respond to individualsâ relationships with industry if they are not aware of them. Consistent with the conceptual framework outlined in Chapter 2, disclosures should provide sufficient information about the nature, scope, duration, and monetary value of re- lationships to allow institutions to assess the risk that secondary interests might unduly influence judgments about research, clinical care, education, or other primary interests. The committee distinguished disclosure to the physicianâs or researcherâs institution from disclosure beyond the institution, for example, to patients, research participants, or the public. One rationale for disclosureâespecially public disclosureâis the deterrence of questionable or inappropriate rela- tionships. As Supreme Court Justice Louis Brandeis (1914) famously ex- pressed it, âsunshine is said to be the best of disinfectants.â In a similar vein, the code of ethics of the American College of Physicians suggests that physicians considering the acceptance of gifts or other relationships with companies should ask themselves what their patients, the public, or their colleagues would think about the arrangement (Snyder and Leffler, 2005; see also Chapter 6). The Nature publishing group urges authors to avoid âany undeclared competing financial interests that could embarrass you were they to become publicly known after your work was publishedâ (NPG, 2008). Disclosure should have beneficial consequences if it leads physicians to avoid gifts, the use of industry-controlled presentations, and other re- lationships that create a risk of compromising their decisions and their professional independence. It could also have harmful consequences if physicians or researchers react by avoiding relationships that promote im- â Some analyses refer to the provision of information to institutional officials as âreport- ingâ and reserve the term âdisclosureâ for the revelation of information to members of the public (e.g., journal readers or patients) (see, e.g., AAMC ). In contrast, some policies refer to reporting of information to external groups. This report follows the common usage (including in federal policies and guidance) and applies the term âdisclosureâ to the provision of information to internal parties as well as to external parties.
68 CONFLICT OF INTEREST portant societal goals and that are accompanied by adequate measures to protect objective judgment. What Is Known About Disclosure Policies, Practices, and Consequences This section first reviews information about the characteristics of dis- closure policies and practices. It then turns to evidence about the effective- ness of disclosure. Presence and Scope of Disclosure Requirements Medical schoolsâ The most recent comprehensive study of medical school conflict of interest policies reports on a 2003 AAMC survey of member schools (response rate of 82 percent) that was designed to characterize their policies and assess the extent to which they were consistent with the asso- ciationâs 2001 recommendations on conflict of interest in clinical research (Ehringhaus and Korn, 2004). It found considerable variation. Almost all (95 percent) of the respondents reported that their policies covered all research involving human participants regardless of the funding source. Sixty-eight percent of the schools used the PHS threshold ($10,000) for in- dividuals to disclose certain financial interests to the institution, whereas 27 percent reported a lower threshold. For elements not required by the PHS regulations, more than 60 percent of the respondents requested disclosure to the institution of equity in nonpublicly traded companies, regardless of the percent share (61 percent) or the estimated valuation (64 percent). The majority requested the disclosure of royalty income either above a certain threshold (38 percent) or regardless of the amount (33 percent). In addition to requiring disclosure to the institution, policies may also require that financial relationships or conflicts of interest be disclosed to individuals who might be affected by the relationship. These might include research colleagues, research participants, journal readers, students, or â The committee also reviewed several earlier studies for additional context and under- standing of policy evolution (see, e.g., Cho et al. , Lo et al. , and McCrary et al. ). â In 2004, the Government Accountability Office reported that 79 percent of universities responding to their survey said that they had a single conflict of interest policy that covered all research. This is consistent with the recommendation of the AAU Task Force on Research Accountability that âall research projects at an institution, whether federally funded, funded by a non-federal entity, or funded by the institution itself, should be managed by the same conflict of interest process and treated the sameâ (AAU, 2001, p. 5). â The PHS regulations state that individuals do not need to report âsalary, royalties or other payments that when aggregated for the Investigator and the Investigatorâs spouse and depen- dent children over the next twelve months, are not expected to exceed $10,000â (NIH, 2008a, question C6, emphasis added). A similar rule applies to the disclosure of equity interests.
POLICIES ON CONFLICT OF INTEREST 69 TABLE 3-1â Percentage of Medical Schools Requiring Further Disclosures for Researchers with a Significant Financial Interest in Their Research Percentage of Further Disclosure Required Medical Schools To research participant in informed consent forms 74 To sponsors or funders of the research 65 To editors of journals to which papers or reports of research are 64 submitted In oral presentations of research results 60 In multicenter trials, to investigators, sponsors, and other 42 institutional review boards participating in the trial Other 23 SOURCE: Adapted from Ehringhaus and Korn, 2004. patients. Again, the AAMC survey showed variation in medical school policies (Table 3-1). A study by Weinfurt and colleagues (2006b) also reported on variations in disclosure policies. Forty-eight percent of medical schools had policies that mentioned the disclosure of researchersâ financial conflicts of interest to research participants. The policies varied in what information was to be disclosed. Medical and scientific journalsâ The International Committee of Medical Journal Editors (ICMJE) has proposed Uniform Requirements for Manu- scripts Submitted to Biomedical Journals that include explanations and provisions about conflicts of interest (ICMJE, 2008). The ICMJE website lists several hundred journals that follow these requirements, but the group does not verify the extent to which a journal does so. The World Associa- tion of Medical Journal Editors (WAME) has also made recommendations on conflict of interest policies (WAME, 2008). Even journals that adopt conflict of interest policies may not apply them equally to industry-funded journal supplements that present papers from a conference or collections of papers on a particular topic. These supplements are generally not peer reviewed and have been criticized for including articles of lower quality (Bero et al., 1992; Rochon et al., 1994). The National Library of Medicine will not cite and index articles from cer- tain types of sponsored supplements unless they include specific disclosures about âany financial relationship the guest editors and authors have with the sponsoring organization and any interests that organization represents, as well as with any for-profit product discussed or implied in the supple- ment and/or individual articlesâ (NLM, 2007, unpaged). Journals may also vary their policies for review articles and editorials
70 CONFLICT OF INTEREST or may not apply their policies to review articles and editorials, which argu- ably offer more room for bias than original research articles. For example, a 2004 editorial in the Journal of the American College of Cardiology stated that the editors generally decline publication of review articles dis- closing industry input out of concern for external influence and subtle bias (DeMaria, 2004). Editors of another journal initially declared that they would not accept review articles written by authors with conflicts of inter- est and then decided that it would accept such articles if the conflicts were not significant (e.g., they involved payments that were less than $10,000) (Drazen and Curfman, 2002). Two recent analyses found considerable variability in the conflict of interest policies of medical and scientific journals. Cooper and colleagues (2006b) found that 93 percent of biomedical journals reported that they had conflict of interest policies applicable to authors, 46 percent reported that they had policies applicable to reviewers, and 40 percent reported that they had policies applicable to editors. Fifty-seven percent reported that they published disclosures for all articles. Earlier studies reported that the percentage of biomedical journals with disclosure policies was lower (see, e.g., McCrary et al.  and Krimsky and Rothenberg ). Ancker and Flanagin (2007) were able to locate online conflict of interest policies for only 33 percent of 84 âhigh-impact, peer-reviewedâ journals in 12 sci- entific disciplines, but a subsequent survey found that 80 percent of the 49 responding journals reported that they had policies in place. Journals vary in whether they give specific guidance to authors regard- ing what financial relationships or conflicts of interest must be disclosed. Ancker and Flanagin (2007) found that 68 percent of journals provided examples of conflicts of interest and 46 percent defined the term. The committeeâs review of a convenience sample of journal policies revealed differences in the specificities of the policies. One journal advises simply, â[a]uthors are required to disclose any sponsorship or funding arrange- ments relating to their research and all authors should disclose any possible conflicts of interestâ (AJN, 2008). In contrast, the New England Journal of Medicine states that disclosures are to include âall of the authorsâ rela- tionships with companies that make products studied or discussed in the article, companies that make related products, and other pertinent entities with an interest in the topicâ (NEJM, 2008). Some journals ask authors about several specific types of relationships and also ask them to indicate explicitly if they have no relationships. One journalâs manuscript agreement form asks authors to certify that their manuscript has not been sponsored by a commercial entity and that if their manuscript includes no acknowl- edgments, it means that nonauthors have made no substantial contribution to it (AFMI, 2008).
POLICIES ON CONFLICT OF INTEREST 71 Professional societies and patient advocacy groupsâ The committee found no published reviews of the disclosure policies of professional societies. It examined a convenience sample of professional society documents and websites and found considerable variation in the content and accessibility of those policies. (Unlike professional society codes of ethics or codes of conduct, disclosure policies do not apply to members generally but are lim- ited to individuals holding positions of responsibility, for example, officers or members of policy-making committees.) Some societies had disclosure forms with a simple, open-ended question about relevant relationships, whereas other forms included specific categories of relationships and re- quired that respondents either report such a relationship or check a box stating that they had none. The policies of some professional societies that develop clinical practice guidelines are discussed further in Chapter 7. The committee did not attempt to conduct a systematic review of the policies of patient advocacy and disease-specific groups. It found little infor- mation on such policies in its initial search of organizational websites and other resources. To the extent that these groups engage in activities such as the development of clinical practice guidelines or the provision of accredited continuing medical education, many of the recommendations in Chapter 7, in this chapter, and elsewhere in this report will apply. Disclosure by Companies of Payments to Physicians District of Columbia, Maine, Massachusetts, Minnesota, Vermont, and West Virginia require pharmaceutical manufacturers to report their financial relationships with physicians; and a number of other states are considering such requirements (Wallack, 2008; Lopes, 2009; MedPAC, 2009). Minnesota and Massachusetts make the information public. Ver- mont requires the stateâs attorney general to make an annual public report based on the information that the pharmaceutical manufacturers have dis- closed. Two states also require the reporting of payments by pharmaceutical manufacturers to hospitals and nursing homes. One state requires medical device companies as well as pharmaceutical companies to report payments to physicians. In general, state policies are relatively new, and their imple- mentation and effectiveness have not been formally assessed. Some pharmaceutical and device companies have voluntarily acted to disclose publicly certain of their payments to physicians (see Chapter 6). The specific details of company plans vary and appear to be evolving as the discussion of public reporting of payments continues. Several com- panies have been required to make such public disclosures as a condition of settlements with the U.S. Department of Justice (Demske, 2008; see Chapter 6 for additional discussion). In 2007, bills were introduced in the U.S. Congress to establish a requirement for companies to report publicly
72 CONFLICT OF INTEREST their payments to physicians (S. 2029 and H.R. 5605, 110th Congress). As discussed in the final section of this chapter, the Medicare Payment Advisory Commission (MedPAC), which advises the Congress on a range of Medicare policy issues, has recommended a more comprehensive policy for company reporting of payments (MedPAC, 2009). Time Frame for Disclosure For employees and others who are involved with an institution for an extended period, disclosure policies generally require an initial disclosure and then periodic (e.g., yearly) disclosures as well as interim disclosures when new relationships arise or when specific events occur (e.g., the sub- mission of a new grant proposal or an application to license intellectual property). For researchers, policies may require the disclosure of financial ties before a study begins (e.g., to university administrators and institu- tional review board members), during a study, (e.g., to the research team, students, or research subjects), and after the study is completed (e.g., to journal editors when papers are submitted for publication in peer-reviewed journals). The conflict of interest policies that the committee reviewed varied considerably in the time periods for which disclosure is required. Typically, policies require the disclosure of relationships that are current or that oc- curred during the previous year. Some policies ask about relationships that are pending, in negotiation, or expected in the next 12 months. The PHS regulations for grantees do not specify a reporting period, except that in determining whether financial relationships exceed the $10,000 threshold for reporting, researchers must consider individual and family financial relationships projected for the next 12 months. Some organizations require disclosure for periods longer than the pre- vious year (e.g., the American Thoracic Society requires disclosure for the previous 3 years [ATS, 2008] and the Journal of the American Medical Association requires disclosure for the previous 5 years [Flanagin et al., 2006]). The requirements may vary by type of relationship. For example, the policy of the American Society of Clinical Oncology specifies disclosure within 2 years for certain relationships (e.g., honoraria and consulting ar- rangements) but not for others (e.g., research funding) (ASCO, 2007). Administrative Burden of Disclosure Policies Disclosure to multiple organizations with various policies can clearly be burdensome for individuals who have received multiple grants, write many papers, serve on various committees and advisory panels, and make many continuing medical education presentations. The committee found little em-
POLICIES ON CONFLICT OF INTEREST 73 pirical information on the administrative burden of disclosure or other ele- ments of conflict of interest policies for either individuals or institutions. A 2007 faculty burden survey undertaken for the Federal Demonstra- tion Partnership reported that respondents assigned conflict of interest monitoring an average burden rating of about 1.8 (with a rating of 1 be- ing no burden and 5 indicating a great deal of burden), whereas grants progress reporting received a rating of 3.4 (with a rating of 3 being some burden) (Decker et al., 2007). Some other government-led initiatives to streamline regulatory policies and practices mention conflict of interest policies and practices but generally do not identify them as a critical issue or problem.10 The committee found examples of efforts to make it easier for individu- als to comply with disclosure policies. For example, to assist their employ- ees in determining whether they have a relationship with a âsubstantially affected organizationâ (as described in NIH intramural conflict of interest policies), NIH has developed a searchable list of such organizations (http:// ethics.od.nih.gov/topics/sao/sao-list.aspx). Similarly, to help committee members identify and report pertinent relationships, some federal advisory committees and at least one professional society (the American Society of Clinical Oncology) have developed lists of for-profit companies that make products that might be affected by committee decisions on a particular is- sue (ASCO, 2008). â The response rate for the survey, which was directed to faculty at major research institu- tions, was less than 40 percent. The Federal Demonstration Partnership, which involves 10 federal agencies and approximately 100 institutional recipients of federal funds, is a coopera- tive initiative whose goal is to reduce the administrative burdens associated with research grants and contracts (http://thefdp.org/). An earlier partnership survey found that conflict of interest monitoring was cited among the tasks for which respondents received the least insti- tutional assistance (Wimsatt et al., 2005). 10â For example, the Research Business Models subcommittee, which is under the Committee on Science of the National Science and Technology Council, has, among other priorities, the development of âspecific guidance or regulations concerning institutional financial conflicts of interest, and to resolve differences in conflict of interest interpretations and terms and conditions of Federal grant awardsâ (http://rbm.nih.gov/priorities/sa3.htm). At NIH, the Clinical and Translational Science Awards (CTSA) program has established a research ethics oversight committee, which has in turn created a work group on conflict of interest policies to survey CTSA sites and gather information on policies. The NIH initiative to âreengineer the clinical research enterpriseâ does not feature conflict of interest policies as part of its assess- ment of clinical research policies (http://nihroadmap.nih.gov/clinicalresearch/overview-policy. asp). Nonetheless, the presentation to the IOM committee by NIH Director Elias Zerhouni stated that improving conflict of interest administration for grantees was important to NIH (Zerhouni, 2007).
74 CONFLICT OF INTEREST Accuracy and Completeness of Disclosures Although most organizations are reluctant to publicize violations of their policies, instances of incomplete and inaccurate disclosure periodically make news. For example, in 2008, investigations by U.S. Senate Finance Committee staff led to a front-page story in the New York Times on the failure of three Harvard faculty members to disclose in fullâeven after they were asked to file amended disclosure formsâthe substantial payments that they had received from pharmaceutical companies over the period from 2000 to 2007 (Harris and Carey, 2008; see also Grassley, 2008b). (The Senate committee staff obtained the data through separate inquiries to companies and medical schools and then compared the responses.) In some cases, it appeared that the disclosures that had been omitted involved companies whose products the researchers were investigating. The present Institute of Medicine (IOM) committee understands that one of the ques- tions about these cases is whether the institutionâs disclosure policy actually requested all the information specified in the PHS regulations, but further details of investigations into the matter had not been released as this report was being completed. Although the IOM committee did not examine the issue, it notes that journalists often fail to report the sources of funding for research that they publicize (see, e.g., Cook et al. ). In addition, journalists themselves may report stories involving pharmaceutical, medical device, or biotech- nology companies with which they have conflicts of interest, for example, the acceptance of company-sponsored travel or prizes for reporting or the reliance on opinion leaders suggested and paid by companies (see, e.g., Schwartz et al. ). Newspapers have also publicized examples of failures by NIH in- tramural scientists to disclose relationships with industry as required by agency rules and examples of scientists who have maintained relationships that would likely not have been approved under the rules. For example, journalists reported the apparent failure of dozens of NIH scientists to disclose relationships with industry, although only 20 or so actual cases were confirmed in a subsequent investigation performed by NIH (see, e.g., Weiss ). Another story reported on a researcher who was found by an internal investigation to have âactivelyâ chosen in âat least 38 separate instances . . . not to adhere to policies because it was inconvenient or time- consuming; he knew it was likely his participation [with the pharmaceutical companies] would have been disapprovedâ (Willman, 2006). A report from the OIG of the U.S. Department of Health and Human Services criticized the agency for not obtaining adequate documentation for the outside finan- cial relationships that it explicitly approved (OIG, 2005). Although cases of nondisclosure may receive considerable publicity,
POLICIES ON CONFLICT OF INTEREST 75 the frequency and extent of deliberate or unintentional underreporting is unknown, and alternative methods for improving the accuracy of disclosure have not been tested.11 Weinfurt and colleagues (2008c) reported incom- plete and inconsistent disclosure in articles on coronary stents published in 2006. They found that 75 authors disclosed at least one relationship with a pharmaceutical company or other organization, but for only 2 of those authors was that relationship disclosed in all of the authorsâ articles. Weinfurt and colleagues did not, however, take into account whether some journals either did not require certain relationships to be disclosed or chose not to publish the disclosures with an article. If a national system of public disclosure of payments by pharmaceutical, medical device, and biotechnol- ogy companies is enacted, institutions could verify the disclosures that they receive. Monitoring and Enforcement The committee found no peer-reviewed studies on the monitoring or enforcement of disclosure requirements specifically or conflict of interest policies generally. One study of journal policies reported that of the 28 jour- nals that had a disclosure policy for authors, 13 had policies that were silent on procedures for responding to an authorâs failure to make a disclosure (Ancker and Flanagin, 2007). As a means of informing readers and also of promoting adherence to their policies, journals from time to time report on cases in which authors did not disclose pertinent relationships with industry (see, e.g., Petersen , Armstrong , Chabner , DeAngelis and Fontanarosa , and Ross et al. ). They sometimes require these authors to write a letter to the editor acknowledging the error (see, e.g., Kurth et al. , Matteson and Bongartz , and Henschke and Yankelevitz ). A few journals have more stringent penalties. For example, after prob- lems with authorsâ failures to disclose, the journal Environmental Health Perspectives adopted a policy that (1) imposes a 3-year ban on the publica- tion of articles by authors who have âwillfully failed to disclose a compet- ing financial interestâ and (2) provides for the publication of a retraction if the editors conclude that they would have rejected the article initially had they known of the undisclosed relationships (EHP, 2009). In general, how- ever, journals decline âto become the COI [conflict of interest] investigative 11â Although the committee did not locate assessments of different disclosure forms, two studies have assessed procedures for obtaining information about the contributions of the listed authors to a submitted manuscript (e.g., analysis of data and drafting of the manuscript) (MarusiÄ et al., 2006; Ivanis et al., 2008). One found that open-ended forms yielded signifi- cantly less information than forms with explicit response categories (MarusiÄ et al., 2006). Those studies were also replicated using different disclosure formats.
76 CONFLICT OF INTEREST squadâ and âcount on . . . authors to be forthright with usâ (Goldsmith, 2006, p. 2148). In addition to exposing offenders to negative publicity, journal reports about failures to make the necessary disclosures may have other consequences for authors. In one case, the Mayo Clinic required investigators found to have made incomplete disclosures to a journal to undergo an internal investigation and to participate in remedial activities (Matteson and Bongartz, 2006). AAMC has recommended that academic medical centers specify the possible sanctions for noncompliance with policies governing conflicts of interest in research involving human subjects and then regularly assess com- pliance (e.g., through internal audit mechanisms and other self-evaluation strategies) (AAMC, 2001). A 2003 AAMC survey, which did not review actual policies but which relied on responses to survey questions, found that 80 percent of respondents reported that their policies had sanctions for violations (Ehringhaus and Korn, 2004). The AMSA assessment cited earlier suggests variability in the oversight and enforcement of conflict of interest policies. On the basis of a review of medical school policies, the report categorized institutions as either having or not having provisions for oversight and enforcement (AMSA, 2008b).12 Of the 58 schools that initially responded to the survey and supplied writ- ten policies for review, 55 percent were characterized as having oversight policies, 45 percent were characterized as having enforcement policies, and 34 percent were characterized as having both.13 A report by the Council on Government Relations (COGR; an as- sociation of research universities) also suggested inadequacies in the pro- cedures used to promote compliance with conflict of interest policies. It concluded: While virtually all research universities and organizations have written pol- icies governing individual financial conflicts of interest in research-related areas, most institutions are still developing formal and informal education programs to assure that the policies are well understood and that compli- ance by affected faculty and researchers is fully in place. (COGR, 2002, unpaged) 12â As described in footnote 1, two independent, trained reviewers read the policies that the medical schools submitted (without identifying information) and then rated them according to specified criteria. For the administration and oversight categories, the reviewers gave yes or no answers to these two questions: Is it clear that there is a party responsible for general oversight to ensure compliance? Is it clear that there are sanctions for noncompliance? 13â Some schools that at first failed to provide relevant policies have since supplied or indicated that they will supply additional information (personal communication, Gabriel Silverman, AMSA Scorecard Director, AMSA, June 6, 2008).
POLICIES ON CONFLICT OF INTEREST 77 Effectiveness of Disclosure Policies A physicianâs or researcherâs disclosure of financial relationships, either to the institution or to a broad audience, is a necessary step for identifying and avoiding or managing conflicts of interest, but it also has important limitations. First, disclosure alone does not resolve conflicts of interests or prevent the harms that may result from a conflict. Second, some evidence suggests that the disclosure of a conflict of interest may have little effect or may even be counterproductive in some circumstances. Experimental studies in psychologyâ Several experimental studies in psy- chology raise general questions about the effectiveness of disclosure of a conflict of interest and even suggest the potential for unintended adverse consequences. For example, in two sets of experimental studies of disclo- sure by individuals in an advice-giving role, researchers concluded that the disclosure of conflicts of interest significantly benefited the advice givers but hurt the interests of those to whom the disclosure was made (Cain et al., 2005). Although the authors of those studies noted that the findings should be treated as no more than evidence that disclosure can potentially have unintended consequences, they caution that most of the mechanisms that produce the effects found are likely to exist except when the recipients of the advice are savvy and experienced. The disclosure of financial relationships can also be ineffective for reasons unrelated to those discussed in the studies just cited. For example, when a large amount of information is disclosed (e.g., on prescription inserts or in certain informed-consent forms), critical points can get lost among less important details. That is, the disclosure of more information may, in some situations, be counterproductive. (Appendix D provides an additional review of the relevant psychological research.) Journal readersâ Two randomized studies suggested that the disclosure of an authorâs financial interests can reduce journal readersâ perceptions of the believability and importance of research reports. One study found that journal readers found an article to be less âinteresting, important, relevant, valid, and believableâ when the authors were disclosed to be employees of a (fictitious) pharmaceutical company instead of employees of an ambulatory care center (Chaudhry et al., 2002, p. 1392). The other study found that readers rated âimportance, relevance, validity, and be- lievabilityâ lower if it was disclosed that the authors had stock holdings rather than nothing to disclose and if it was disclosed that the authors had received a research grant from a company rather than nothing to disclose (Schroter et al., 2004).
78 CONFLICT OF INTEREST Research participantsâ Several studies have suggested that disclosures to prospective research participants of investigatorsâ financial relationships have little impact on decisions to participate in research (see, e.g., Kim et al. ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ , Hampson et al. , Weinfurt et al. [2006a, 2008a,b]), and Gray et al. . ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ In a survey of participants in clinical trials for the treatment of cancer, more than 70 percent of the respondents would still have enrolled in the clinical trial even if the researcher had financial ties to the pharmaceutical company sponsoring the trial or had received royalty payments (Hampson et al., 2006). Only 31 percent wanted the researcherâs financial interests to be disclosed. Other studies have described hypothetical clinical trials to individu- als with chronic diseases and varied the kind of information presented about the researchersâ financial relationships with the sponsors of the trial (Weinfurt et al., 2008a,b). The respondentsâ willingness to participate in a hypothetical clinical trial varied substantially, depending on the type of financial relationships. The respondents were more concerned when the researchers held equity in the sponsoring company than when the research- ers received a payment to cover the cost for each participant in the study. Trust in the researchers decreased somewhat after the disclosure of equity interests. Other factors, such as the benefits and the risks of the clinical trial, had more of an impact on the respondentsâ decision to participate in the trial. These studies of research participation can be criticized on methodolog- ical grounds for not explaining the risks of conflicts of interest (e.g., bias in the conduct of research and the failure to publish negative findings) or not linking the responses to actual decisions about research participation. It is not known whether the respondents might have been more concerned about researchersâ financial relationships with sponsors if they had been given background information about the risks. Patientsâ Several surveys in the 1990s suggested that many patients were not aware of industry gifts to physicians but were relatively tolerant of most gifts. One study suggested that, overall, patients were considerably more likely than physicians to believe that gifts from pharmaceutical companies influenced physician practice, but only 54 percent of patients were aware of such gifts (Gibbons et al., 1998; see also Blake and Early  and Mainous et al. ). On a different but related issue, one study of the disclosure of information about physician payment mechanisms in man- aged care plans found that disclosure did not reduce patientsâ trust in their physicians and might even have âa mild positive impactâ on trust (Hall et al., 2002, p. 197; see also Pearson et al. ). Other studies have sug- gested that patients are interested in information about how their physi- cians were paid or, more generally, what financial incentives the patientsâ
POLICIES ON CONFLICT OF INTEREST 79 health plan imposes on participating physicians (Kao et al., 2001; Levinson et al., 2005). (Chapter 6 briefly discusses conflicts of interests created by physician payment methods.) PROHIBITING OR ELIMINATING CONFLICTS OF INTEREST Prohibition as a Preventive Strategy Some institutions have conflict of interest policies that prohibit certain financial relationships outright because their risks are considered to greatly outweigh any potential benefits. As described further in Chapter 5, a 2008 report by AAMC recommended that academic medical centers prohibit a wide range of financial relationships with industry. Several medical schools (e.g., the University of Pittsburgh, the University of Texas Medical Branch at Galveston, and the University of California system) have policies prohib- iting gifts, and some prohibit participation in company speakers bureaus (e.g., the University of Massachusetts, the Mayo Clinic, and the University of Louisville).14 Also in 2008, the Pharmaceutical Research and Manufacturers of America revised its Code on Interactions with Healthcare Professionals to state that companies should not offer pens, notepads, and other non- educational items to health care professionals. In Massachusetts, recent legislation gives these guidelines legal force by requiring the public health department to establish âregulations for a marketing code of conduct . . . that shall be no less restrictive than the most recent versionâ of the codes on interactions with health care providers of the Pharmaceutical Research and Manufacturers of America and the Advanced Medical Technology As- sociation (see Chapter 111N, section 2, Massachusetts Senate No. 2863). Thus, policies may forbid both the giving and the receiving of certain gifts. (Implementing regulations were published in March 2009 [see Lopes, 2009; see also Chapter 6].) Some conflict of interest policies prohibit certain relationships but allow exceptions. For example, federal policies covering NIH and other employees of the U.S. Department of Health and Human Services state that its employees may not have an âemployment relationshipâ with drug, medical device, or biotechnology companies; grantees; health care provid- ers; or health insurers. They also may not be paid to teach, speak, write, or edit for such organizations. The policies allow for prior approval of certain 14âExcept for the information for the University of California system (University of Cali- fornia, 2008), this information comes from policies summarized by AMSA (2008b) and then checked by reference to documents on the AMSA website or through links to those documents.
80 CONFLICT OF INTEREST exceptions if prohibition of a relationship is not ânecessary to ensure public confidence in the impartiality or objectivity with which HHS programs are administeredâ (HHS, 2005, p. 51572). To cite another example, AAMC recommends that medical schools set a ârebuttable presumption that an individual who holds a significant financial interest in research involving human subjects may not conduct such research . . . [except when] the circumstances are compellingâ (AAMC, 2001, p. 7). (The ârebuttable presumptionâ concept is taken from the law and refers to assumptions that are taken to be true unless they are explicitly and successfully challenged in a particular case.) A compelling circum- stance would exist, for example, if a researcher with a conflict of interest has unique expertise or skill with implanting and adjusting a complex new medical device and this expertise is needed to carry out an early-stage clini- cal trial safely and competently. Generally, some kind of management plan would then be devised. This approach is discussed further in Chapter 4. Prohibition or Elimination as a Management Strategy The options for managing conflicts of interest discussed in the next section all permit the continuation of a relationship in some situations in which a conflict exists. In certain cases, however, continuation of the relationship is not acceptable because of the severity of the threat that it poses to the primary interest. In that case, an individual with a conflict of interest may agree to end the relationship that creates the conflict, for example, by selling company stock, resigning from a company governing or advisory board, or ceasing to consult for a company. Alternatively, an individual with a conflict of interest may decide to forgo participation in such an activity rather than eliminate the financial relationship in question. Some relationships with conflicts of interest may be difficult to eliminate, for example, the relationship with a spouse because of a conflict of interest involving the spouseâs employment. The committee found no systematic assessment of the adoption, imple- mentation, or effectiveness of policies prohibiting certain financial rela- tionships with industry. Somewhat more information is available on the management of conflicts of interest. EVALUATING AND MANAGING CONFLICTS OF INTEREST The management of a conflict of interest is necessary when an assess- ment of an individualâs financial relationships identifies a conflict of interest and when disclosure alone is inadequate but elimination of the conflict is a requirement that is too severe. AAMC has recommended that medical schools create conflict of interest committees to make these assessments and
POLICIES ON CONFLICT OF INTEREST 81 propose management strategies, when appropriate. Professional societies may rely on senior staff or members (e.g., chairs of guideline development panels) for assessments of relationships and responses. The management options will vary depending on the nature of the conflict and the activity under consideration. Examples of management options follow: â¢ asking an individual with a conflict of interest to reduce the value of a financial relationship so that it falls below a threshold amount; â¢ requiring that an individual forgo participation in committee votes, deliberations, or decisions about a topic related to that individualâs conflict of interest; â¢ modifying the design of a research project or having a researcher with no conflict of interest serve as the principal investigator; or â¢ providing an observer to monitor and evaluate the content of a continuing education course conducted by an individual with a conflict of interest for bias. What Is Known About Management Policies, Practices, and Consequences The available data suggest that institutions vary considerably in how they oversee and manage conflicts of interest. Ehringhaus and Korn (2004) reported that 76 percent of medical schools responding to the 2003 AAMC survey had established, as recommended by AAMC, a standing committee to evaluate conflict of interest disclosures, and 21 percent included at least one committee member from outside the institution, also as recommended by AAMC. Eighty-one percent of the medical schools responding to the AAMC survey allowed investigators with a significant financial interest to conduct research involving human participants when compelling circum- stances exist. Only 61 percent of the respondents indicated that they had adopted the rebuttable presumption or a similar strategy, and only 26 per- cent indicated that they had a definition of the compelling circumstances or similar conditions that would allow rebuttal of the presumption. Even within a single university system, conflict of interest practices may vary (see, e.g., several studies of the University of California system reported by Boyd et al. , Lipton et al. , and Boyd and Bero ). For example, within the University of California system, some campuses have standing committees that meet at least monthly, whereas others convene committees on an ad hoc basis (Boyd et al., 2004). Some but not all campuses include committee members from outside the campus community. Some committees are structurally linked through centralized computer systems to other oversight bodies, such as the campus institu-
82 CONFLICT OF INTEREST tional review board, whereas others do not share financial information within the university unless they are asked to do so. Assessing Risks of Disclosed Relationships If an organizationâs policy requires more than just disclosure, the next step is a review to assess whether a disclosed relationship constitutes a conflict of interest and what risks or potential benefits the relationship presents. As described earlier, a department chair or similar individual may review disclosures and identify conflicts of interest or may refer potential conflicts of interest for further review by a conflict of interest committee or other group or official. The IOM committee found little systematic investigation of the insti- tutional practices and or criteria used to assess financial relationships and conflicts of interest. One small qualitative study of a university system found that individual conflict of interest committees made decisions on a case-by-case basis, taking into account multiple considerations (e.g., the extent and the nature of the financial relationship and the type of research and research design) and following no rigid formula (Boyd and Bero, 2007). The committees rarely made a direct assessment of the likelihood that an investigator would act improperly. Some specific advice on assessing the severity of conflicts of interest is available. The AAMC-AAU report on conflict of interest in research involv- ing human subjects describes several considerations that should be taken into account when the risks and possible benefits of allowing an investiga- tor with a conflict of interest to participate in such research are assessed (AAMC-AAU, 2008) (Box 3-2).15 It also discussed the application of these questions to 10 illustrative cases.16 The FDA has developed guidance on whether an individual with a con- flict of interest should be allowed to serve on one of its advisory committees (FDA, 2008b). Some of the questions roughly correspond to the consid- erations identified in Chapter 2. For example, one question is whether a âparticular matterâ under consideration by a committee will have a direct and predictable effect on the financial interests of any organization? . . . A âpredictableâ effect . . . is a real, as opposed to a 15â Ingeneral, this report follows the practice of recent IOM reports in referring to research participants rather than research subjects (see, e.g., IOM [2001, 2003, 2004]; see also NBAC ). When quoting and sometimes when referring to AAMC and other reports that employ the latter usage, the report follows their practice. 16â The 2002 report by COGR also included an analysis of cases, and some university educational materials likewise feature analyses of case studies as a means of providing an understanding of the risks presented by financial relationships (see, e.g., Columbia University [undated]).
POLICIES ON CONFLICT OF INTEREST 83 BOX 3-2 Risks and Potential Benefits to Consider in Assessing the Severity of a Researcherâs Conflict of Interest â¢ Risks to human subjects: to what extent could the conflict of interest in- crease the risk (considering the role specified for the researcher with the conflict of interest in recruiting or treating research participants)? â¢ Risks of bias in data collection, analysis, and reporting: to what extent could the researcher with the conflict of interest compromise the integrity of the data? â¢ Risks to reputation: to what extent could the reputation of the researcher with the conflict of interest or the researcherâs institution be damaged, even if the institution establishes a plan to manage the conflict? â¢ Expected benefits to medicine, science, and public health: how do the expected benefits of allowing the research to proceed compare with the risks? SOURCE: Adapted from AAMC-AAU, 2008. speculative, possibility that the matter will affect the financial interest. It is not necessary, however, that the magnitude of the gain or loss be known, and the dollar amount of the gain or loss is immaterial. . . . [M]ost poten- tial advisory committee recommendations pertaining to marketing status, labeling, post-marketing requirements, and device classification or reclas- sification would ordinarily have a âdirect and predictable effectâ on finan- cial interests. . . . Financial interests that ordinarily will not be affected in a direct and predictable manner include a grant or contract between an organization and the employeeâs university to conduct research on a product that is not the subject of the particular matter before the advisory committee or a competitor product. (FDA, 2008b, pp. 10â14) FDA rules involving clinical investigators also take into consideration aspects of the study designâfor example, the use of objective end points, blinding, or the participation of multiple investigatorsâthat might re- duce the potential of an investigatorâs interests to bias the study results (21 CFR 54.5). (The rules cover financial disclosures and the management of the relationships of clinical investigators in studies that companies plan to use to support FDA approval of the marketing of a medical product.) Management Strategies Survey data indicate that medical schools employ various strategies to manage conflicts of interest in research (Table 3-2). Disclosure to some outside party seems to be a common and preferred response to an identi-
84 CONFLICT OF INTEREST TABLE 3-2â Percentage of Medical Schools Citing Different Management Policy Options When Researchers Have a Significant Financial Interest in Their Research Policies Suggested or Required by Organizations Permitting Percentage of Participation After a Conflict of Interest Is Identified Medical Schools Monitoring the research project 87 Eliminating the investigatorâs significant financial interest 83 Disclosing significant financial interests to human subjects on the 86 consent form Using either internal or external data safety monitoring boards 54 Regularly auditing the informed consent and research subject 51 enrollment process Involving a patient representative during the consent and enrollment 26 process Involving a patient representative during recruitment of research 22 subjects SOURCE: Adapted from Ehringhaus and Korn, 2004. fied conflict of interest (see, e.g., Boyd and Bero  and Ehringhaus and Korn ). One analysis of cases in which researchers disclosed their financial re- lationships found that university conflict of interest committees determined that 26 percent of the cases reviewed involved conflicts of interest that needed management (Boyd et al., 2004).17 The three most commonly ap- plied management strategies were requiring disclosure in publications and presentations (40 percent of the managed cases), appointing an oversight committee to protect the interests of students involved in the project (21 percent of the managed cases), and eliminating the relationship during the period of the project (22 percent of managed cases). The least common management approach was eliminating the conflict of interest or prohibit- ing the research. The IOM committee is not familiar with any evaluations of the imple- mentation or the consequences of different management strategies. This is a significant deficit. At one of the committeeâs public meetings, an experi- 17 Financial ties were most often with pharmaceutical companies or biotechnology com- panies. Across the seven campuses involved in the analysis, payment for consulting activities accounted for 54 percent of the financial disclosures, equity holdings accounted for 38 percent of the disclosures, payment for talks accounted for 14 percent, scientific advisory board mem- bership accounted for 13 percent, membership on a companyâs board of directors accounted for 12 percent, and being a company founder accounted for 7 percent. Over this period, in- vestigators became more likely to have multiple financial ties with a single company, such as financial ties through the receipt of consulting income, honoraria, and stock.
POLICIES ON CONFLICT OF INTEREST 85 enced clinical researcher questioned the strategy of appointing an oversight committee to monitor research involving an investigator with a conflict of interest (Benet, 2008). In that scientistâs view, so many decisions need to be made in the course of a research project that it is not realistic to expect a faculty member to want to or have time to participate in the close and effective monitoring of another faculty memberâs research. In addition, monitoring imposes costs that might be judged in some cases to exceed the potential benefits. In Chapter 9, the committee recommends the development and funding of a program of research on conflict of interest. The outcomes of conflict of interest policies, both positive and negative, would be a key issue for consideration in such a program of research. Knowledge and Attitudes Regarding Conflict of Interest Policies A few studies suggest that many investigators do not understand their institutionâs conflict of interest policies and may be skeptical about them. In one in-depth qualitative study of clinical investigators, less than half of the respondents could accurately describe their institutionâs policies (Boyd et al., 2003). In addition, many respondents believed that the indi- vidual investigator, the professional society, and the public at largeânot the universityâwere the appropriate monitors of conflicts of interest. Although many respondents recognized the general risks associated with conflicts of interest, they believed that they were personally not at risk for bias result- ing from financial relationships, a common finding in the research reviewed for this report. In another, web-based survey of researchers at a single medical center (response rate of less than 40 percent), 17 percent of the respondents were not aware of the institutionâs conflict of interest policies and 60 percent could correctly describe at least one (but not all) of the policies (Lipton et al., 2004). With respect to consequences, 43 percent of the respondents believed that the policies discouraged a faculty memberâs ability to start new companies, 31 percent believed that the policies discouraged consultation with companies, and 21 percent believed that the policies discouraged spon- sored research but another 21 percent thought that they encouraged such research. Although 14 percent believed that the schoolâs policies hindered their own research agenda, 82 percent believed that it had no effect. Among the respondents who actually had a financial relationship that was subject to committee review, 91 percent said that they were satisfied with how the review was handled, but some of the remaining 9 percent who were not satisfied had very negative attitudes toward the process.
86 CONFLICT OF INTEREST Policy Dissemination and Education Strategies AAMC has advised academic medical centers to provide education and training about their conflict of interest policies to all faculty, staff, students, and trainees (AAMC, 2001). In 2002, NIH reported that the policies of some research institutions were difficult to locate and were sometimes interspersed in various other institutional policies on issues such as ethics, purchasing, and consulting. It recommended that institutions present their conflict of interest policy âas a complete, self-contained document with citations and web links to supporting policies, procedures, and Federal and state regulations, as appropriateâ (NIH, 2002, unpaged). The Office of Extramural Research at NIH recently created an online tutorial on conflict of interest and other materials intended to help investigators understand and comply with NIH policies (the tutorial is available at http://grants.nih. gov/grants/policy/coi/). The IOM committeeâs review of the policies and other information on conflict of interest from academic medical centers and universities showed that they vary considerably in the informational resources that they make available to their faculty and staff. Some schools provide online resources that are intended to help people easily find relevant institutional policies and resources (including individuals who can answer questions about the policies). Examples include the University of Minnesota, which has a web- based training module on conflict of interest (University of Minnesota, 2008), and Stanford University, which has frequently asked question units on conflict of interest and related university policies, as well as a quiz and other resources (Stanford University, undated). A professional society may publicize its policies by publishing them in the societyâs journal(s). It may also make the policies accessible to the public on its website. Compliance and Enforcement The earlier discussion of compliance with and the enforcement of disclosure policies reviewed information about compliance with and the enforcement of policies as they apply to individuals. The discussion in this section focuses on the extent to which research institutions follow appli- cable PHS rules. A 2002 review of a sample of grantee policies undertaken by the NIH Office of Extramural Research found that institutional policies often did not reflect the requirements of the PHS regulations (NIH, 2002). In 2007, NIH reported on 18 targeted site reviews regarding grantee compliance with PHS conflict of interest policies. It found no instances of intentional noncompliance and concluded that the institutions that it visited generally
POLICIES ON CONFLICT OF INTEREST 87 had âimplemented the Federal regulation thoughtfully and with diligenceâ (NIH, 2007). It did, however, report some problems with timely and con- sistent reporting and suggested the need for improvements in several areas, including educational and enforcement procedures, the clarity of the forms used to report conflicts of interest, and definitions. A 2008 report by the OIG of the U.S. Department of Health and Human Services criticized NIHâs oversight of grantee institutions (OIG, 2008). Although NIH accepted some of the reportâs suggestions, it rejected taking a more active oversight role, particularly requiring and reviewing detailed conflict of interest reports from institutions. Doing so would âef- fectively, if not legally, transfer the locus of responsibility for managing [financial conflicts of interest] from the grantee institution to the Federal Governmentâ (Zerhouni  in OIG [2008, pp. 20â21]). The OIG disagreed that collection of the information would usurp grantee responsi- bilities, and it argued that without some details about the nature (and not just the existence) of the conflicts that were identified, NIH lacks important information that it needs to oversee and enforce PHS regulations. Also in 2008, NIH announced the development of and began testing an electronic reporting and tracking tool that that would allow grantee institu- tions to prepare and submit required conflict of interest reports and search past reports. Consistent with one of the OIG reportâs recommendations, the tool would also provide a central web-based location for grantee conflict of interest reports received across NIH (Bravo, 2008; see also NIH [2008b]). In addition, NIH has initiated procedures and training to ensure proper NIH staff oversight of conflict of interest issues involving grantees. The IOM committee identified some publicly reported instances of NIH enforcement of PHS policies. For example, in October 2008, after congressional inquiries and reports of apparent major inaccuracies in a researcherâs financial disclosures to Emory University, NIH suspended a $9 million grant for a study led by the researcher and instituted special condi- tions for the institutionâs other studies conducted with the support of NIH grants (Harris, 2008; Kaiser, 2008). Subsequently, the university removed the individual from his post as department chair and significantly restricted his outside activities (Shelton, 2008). RECOMMENDATIONS Empirical data on conflict of interest policies are limited, have method- ological shortcomings, and tend to focus on academic institutions. Some in- stitutions do not make their policies easily accessible. Institutions also revise their policies, which limits the usefulness of older studies. Nonetheless, the available evidence points to substantial variations in institutional require- ments for the disclosure of financial relationships or conflicts of interest.
88 CONFLICT OF INTEREST Variations exist in who is required to report on a conflict of interest, when reporting is required, and what relationships and what details about these relationships are to be reported (e.g., the exact amounts of payments rather than payments above a threshold or within dollar categories). Variations also exist in what relationships are prohibited, what criteria are considered in evaluating financial relationships, what strategies are employed when a conflict of interest is identified, and what is done to monitor and promote adherence to policies. These extensive variations raise concerns that some institutions may not have sufficient data to make determinations about the extent and the nature of an individualâs financial relationships or to judge the severity of a conflict of interest. Some institutions may also lack adequate procedures for evaluating and eliminating or managing identified conflicts. The committee expects that there are many explanations for the varia- tions in policies, including the press of other issues demanding attention, a reluctance to propose changes that may spark controversy and dissension, and cultural traditions that vary in how restrictions on the pursuit of per- sonal gain are viewed. Absent outside pressures and oversight, variation in conflict of interest policies may encourage an unhealthy competition among institutions to adopt weak policies and shirk enforcement. It may also aid investigators who want to avoid restrictions on their pursuit of secondary financial interests. The recommendations presented in this chapter and in this report are intended to discourage such undesirable institutional and individual be- havior but not to damage beneficial collaborations. If institutions do not act voluntarily to strengthen their conflict of interest policies, such inaction may prompt government regulation. (The recommendations below focus on individual conflicts of interest. Chapter 8 presents recommendations on conflicts of interest at the institutional level.) Adopting Conflict of Interest Policies The committeeâs first recommendation deals with institutional basics: the adoption of a policy and the creation of a conflict of interest committee. The details of the policies may vary, depending on an institutionâs mission and other characteristics, but certain features are fundamental to credible and meaningful conflict of interest policies. RECOMMENDATION 3.1â Institutions that carry out medical re- search, medical education, clinical care, or practice guideline develop- ment should adopt, implement, and make public conflict of interest policies for individuals that are consistent with the other recommenda- tions in this report. To manage identified conflicts of interest and to
POLICIES ON CONFLICT OF INTEREST 89 monitor the implementation of management recommendations, insti- tutions should create a conflict of interest committee. That committee should use a full range of management tools, as appropriate, including elimination of the conflicting financial interest, prohibition or restric- tion of involvement of the individual with a conflict of interest in the activity related to the conflict, and providing additional disclosures of the conflict of interest. Recommendation 3.1 calls on all institutions that conduct medical research, offer medical education, provide clinical care, or develop practice guidelines to adopt comprehensive conflict of interest policies for their employees. These policies should cover all those whose decisions and judg- ments affect their institutionâs missions and primary interests. Consistent with the committeeâs charge, the recommendation refers only to relation- ships with pharmaceutical, medical device, and biotechnology companies. In practice, individual institutions will design their policies to cover other relevant relationships. These might include consulting or speaking ar- rangements with health insurance companies, leadership positions with professional organizations, teaching at other institutions, and service on government advisory committees. (As described in Chapter 2, some of these relationships may present conflicts of commitment.) So that those who rely on academic medical centers, medical journals, professional societies, patient advocacy groups, and other institutions may assess an institutionâs conflict of interest policies, the policies should be publicly available, for example, on the institutionâs website. Although the details will vary, it is also important for institutions to disseminate and explain their policies to those who are subject to them. Strategies might include the provision of an education module and the inclusion of a set of frequently asked questions. Recommendation 3.1 also calls on academic medical centers and other institutions to create conflict of interest committees to manage conflicts of interest involving individuals. This reiterates a recommendation of AAMC, which found in its 2003 survey that not all medical schools reported that they had such committees. Professional societies and other institutions would also benefit from conflict of interest committees that would imple- ment their policies. For example, a conflict of interest committee for a professional society would review conflicts that arise in different aspects of the societyâs work, including the development of clinical practice guidelines and the conduct of society meetings and educational programs. (For some very small institutions, the formation of a formal committee may not be necessary if the relevant responsibilities are clearly defined and assigned to appropriate staff or, possibly, volunteers.) A conflict of interest committee should bring experience and consistency to evaluations of financial relation-
90 CONFLICT OF INTEREST ships with industry and decisions about those relationships, although the specific details (e.g., how risks and potential benefits are assessed and what management options are considered) may vary, depending on the activity in question. The recommendation mentions monitoring as an activity of the conflict of interest committee, but in practice, the details of monitoring may best be handled by an administrative unit, with the conflict of interest committee providing more general oversight. Improving Information for Identifying and Evaluating Conflicts of Interest Disclosure as an Element of Policy The disclosure of financial relationships with industry is only one part of a comprehensive conflict of interest policy, but it is nonetheless an es- sential step. Unless institutions know about these relationships, they cannot assess them and determine whether additional stepsâsuch as the elimina- tion or management of a relationshipâare necessary. Recommendation 3.2 identifies key features of policies on disclosure. Recommendations in Chapters 4, 5, 6, and 7 provide guidance about the elimination or manage- ment of conflicts of interest in the contexts of medical research and educa- tion, patient care, and practice guideline development, respectively. RECOMMENDATION 3.2 As part of their conflict of interest poli- cies, institutions should require individuals covered by their policies, including senior institutional officials, to disclose financial relationships with pharmaceutical, medical device, and biotechnology companies to the institution on an annual basis and when an individualâs situation changes significantly. The policies should â¢ request disclosures that are sufficiently specific and comprehen- sive (with no minimum dollar threshold) to allow others to assess the severity of the conflicts; â¢ avoid unnecessary administrative burdens on individuals making disclosures; and â¢ require further disclosure, as appropriate, for example, to the conflict of interest committee, the institutional review board, and the contracts and grants office. Conflict of interest policies should cover individuals who have discre- tion in the conduct of research and educational activities, the provision of clinical care, and the development of clinical practice guidelines. (Senior officials are also covered by Recommendation 8.1 in Chapter 8, which examines institutional conflicts of interest.) Disclosures should be made
POLICIES ON CONFLICT OF INTEREST 91 at least annually and more often if an individualâs situation changes. They should also be updated during the year if an individualâs situation changes significantly, for example, because an existing relationship expands (e.g., a faculty member who is a company consultant is also appointed to the companyâs governing board) or because a new relationship (e.g., a new consulting arrangement) is created that is relevant to a specific activity (e.g., participation in a panel developing a clinical practice guideline). In addition to requiring disclosure of conflicts of interest to the institutional review board and the other entities listed in the recommendation, policies may also cover additional disclosures, for example, to entities responsible for continuing medical education program oversight. Elements of a disclosure policy may vary depending on the institution, but the disclosures should be sufficiently specific to support the identifica- tion of conflicts of interest and an evaluation of their severity. For example, if information on the dollar value of relationships is reported in categories rather than specific amounts, the highest categories should reach into the hundreds of thousands of dollars. The committee recommends the elimina- tion of minimum thresholds for individual reporting of financial relation- ships. As discussed earlier in this chapter, the 1995 PHS regulations specify a $10,000 threshold, which applies to the individual and his or her spouse and dependent children. Most PHS grantees have adopted this threshold, al- though approximately one-quarter require reporting regardless of the dollar value of the relationship. The committee recognizes that elimination of the minimum threshold would add to the burden both for those reporting and for those reviewing relationships but believes that it is important to increase the accuracy of reporting and provide institutions with a more complete picture of an individualâs financial relationships across different reporting categories (e.g., consulting, advisory committee service, and speaking). The committee also notes research that suggests that even small payments may put an individual at risk of unconscious bias. In their joint report on conflict of interest in human subjects research, AAMC and AAU also recommended removing minimum (de minimis) thresholds (AAMC-AAU, 2008). NIH should seek revisions in the PHS regulations to eliminate the threshold, but NIH grantees should act without waiting for such revisions. Greater Consistency in Disclosure Policies The committee recognizes that the objective of achieving sufficient specificity in disclosures may sometimes be in tension with the objective of minimizing the administrative burdens of disclosure. To the extent that the consensus process proposed in Recommendation 3.3 is successful, it may help resolve these tensions by promoting greater consistency across institutions. Greater consistency should simplify the demands on those who
92 CONFLICT OF INTEREST must understand and comply with the disclosure requirements of multiple institutions. RECOMMENDATION 3.3 National organizations that represent aca- demic medical centers, other health care providers, and physicians and researchers should convene a broad-based consensus development pro- cess to establish a standard content, a standard format, and standard procedures for the disclosure of financial relationships with industry. To achieve greater consistency in institutional disclosure requirements, Recommendation 3.1 calls for a broad-based national consensus develop- ment process. This undertaking would be convened by national organiza- tions representing academic medical centers, other health care providers, physicians, and researchers and would also include representatives of pro- fessional societies; consumer and patient advocacy groups; accreditation, certification, and licensing agencies; medical journals and organizations of medical journal editors; health plans and insurers; government agencies, including NIH and the FDA; and organizations with expertise in database development and management. The process used by AAMC to develop its recent recommendations on relationships with industry in medical educa- tion offers one model for the process, although the task would be narrower and more detailed in its focus on definitions of the financial relationships to be disclosed, reporting formats, and similar matters. The committee appreciates that different disclosures may be required for different purposes. For example, the information that a medical journal needs from the authors of a manuscript differs from the information that a government agency may require for members of an advisory panel. For similar institutions (e.g., for medical journals as a category and for similar government advisory panels as a category), the objective would be to de- velop a consensus on a common format. A major task for the consensus development process would be to agree on the categories of relationships that need to be disclosed and the type of information about each relationship that is needed to evaluate it. Consulting is an example of a category that needs further specification. That term can cover relationships that range from the provision of promotional or mar- keting support to a company to the offering of objective technical advice on scientific advances, products in development, or research study design. The institution of standard categories, definitions, and similar agreements should reduce confusion, misunderstandings, and misinterpretations. In technical terms, the task for the consensus group would be to specify the elements for a relational database, including the definitions and attri- butes of these elements. Once the elements are specified, the expectation is that software developers would create programs that physicians and
POLICIES ON CONFLICT OF INTEREST 93 TABLE 3-3 Candidate List of Categories of Financial Relationships with Industry to Be Disclosed Research grants and contracts Consulting agreements Participation in speakers bureaus Honoraria Intellectual property, including patents, royalties, licensing fees Stock, options, warrants, and other ownership (excepting general mutual funds) Position with a company Company governing boards Technical advisory committees, scientific advisory boards, and marketing panels Company employee or officer, full or part time Authorship of publications prepared by others Expert witness for a plaintiff or a defendant Other payments or financial relationships researchers could use on their computers to enter, store, and update infor- mation on their financial relationships. The software would then format the information as needed for disclosures for various purposes (e.g., submission to an academic medical center or a medical journal). It would be similar to reference software that allows authors to format references to meet the specifications of different journals. As a starting point, Table 3-3 presents a candidate list of basic catego- ries of the relationships to be disclosed. Each requires further definitions, and some might require subcategories. The committee did not propose a specific format for the provision of information about these relationships. It is important, however, that any format promote completeness and speci- ficity, for example, by requiring individuals to check one box if they have a particular relationship, to check another box to declare explicitly that they do not have the relationship, and to provide certain details about an indicated relationship (e.g., its value, the company involved, and the nature of the work). In addition to the categories of relationships to be disclosed, the consen- sus process needs to address several other key questions. For example, what details of relationships need to be reported (e.g., the amount of income and the name of the company)? How should amounts be reported? Would it be preferable to have individuals making disclosures check a box indicating the range of income from a relationship or should they provide specific dollar amounts? Will a single time frame (e.g., the relationships in existence dur- ing the previous 12 months) be adequate for all purposes? How should the financial relationships of close family members (e.g., spouses or domestic partners, dependent children, and parents) be considered?
94 CONFLICT OF INTEREST Company Reporting of Payments to Individuals and Institutions Recommendations 3.2 and 3.3 involve disclosures by individuals to organizations. The next recommendation proposes requirements for com- panies. Several state laws and proposals for additional state or federal rules reflect concerns about inaccurate and incomplete disclosures. As discussed earlier, these laws and proposals vary, for example, in the types of compa- nies and payments or relationships that they cover and in provisions for public reporting. In response to proposals for additional state and national legislation, several industry groups and individual companies have sup- ported some form of company disclosure while seeking to minimize the administrative burdens of such reporting and to protect information that might reveal business strategies to competitors (Finance Committee, U.S. Senate, 2008). Recommendation 3.4 calls for a broad national reporting program. RECOMMENDATION 3.4 The U.S. Congress should create a national program that requires pharmaceutical, medical device, and biotechnol- ogy companies and their foundations to publicly report payments to physicians and other prescribers, biomedical researchers, health care institutions, professional societies, patient advocacy and disease-specific groups, providers of continuing medical education, and foundations created by any of these entities. Until the Congress acts, companies should voluntarily adopt such reporting. A national law covering company payments to physicians, research- ers, and medical institutions would be a useful supplement to policies that require individual physicians, researchers, and others to disclose financial relationships to institutions. It should provide that company-reported pay- ments be readily available on a searchable public website that allows the aggregation of all payments made to an individual or organization, al- though some personal identifying information might be restricted to protect individuals against, for example, identity theft. Such a database could help institutions and potentially others to monitor adherence to institutional dis- closure policies. It would not substitute for institutional conflict of interest policies. It also would not eliminate conflicts of interest. One objective of drafting and implementing legislation and explaining it to the public and those affected would be to discourage the inference that all reported rela- tionships are bad and to avoid harm to constructive collaborations. The committee did not investigate program options and administration in detail, but it generally supports the approach to company reporting dis- cussed by MedPAC during several public sessions in 2008 and presented in MedPACâs March 2009 report (MedPAC, 2008a,b,c,d, 2009). Consistent
POLICIES ON CONFLICT OF INTEREST 95 with the committeeâs recommendation but in contrast to state policies and some other proposals for federal policy, MedPACâs proposal covers not only payments to physicians but also payments to a range of organizations, including medical schools, professional societies, and providers of continu- ing medical education. The committeeâs proposal would add payments to biomedical researchers. The MedPAC recommendation would add pay- ments to pharmacies and pharmacists, health plans, and pharmacy benefit managers as well as payments by medical supply companies. Companies could include clarifying details about the context of a payment (e.g., speci- fying whether the payment is a research grant that covers all project costs and not just the investigatorâs salary). The committee considers these to be reasonable provisions for a company reporting program. Implementing regulations would need to specify clear definitions and exact categories for the reporting of payments. The consensus-building activity proposed in Recommendation 3.3 could contribute to this specifica- tion and promote consistency with institutional disclosure policies. As proposed by MedPAC, the database of company-reported informa- tion would be public, but the physicianâs National Provider Identifier (NPI) would not be given.18 The entire database would be available to researchers who enter into confidentiality and data use agreements with the secretary of the U.S. Department of Health and Human Services. The database would be searchable by manufacturer; recipient name, location, and specialty (if applicable); type of payment; name of related product (if applicable); and year. The MedPAC report did not include an estimate of the costs to the government of creating and maintaining the systems but notes that the costs would be higher than those of state systems, only one of which makes the data public, but not in a searchable database. In MedPACâs proposal, company reporting would be required annu- ally, but reporting for a clinical trial could be delayed until the trial was publicly registered or until FDA approval related to the development of a new product was granted (but not later than 2 years after the payments were made). The national policy would preempt state policies, to the extent that they cover the same categories of payments and recipients, and would provide for civil penalties for noncompliance. Legislation introduced in the U.S. Congress in 2009 includes similar provisions on these points (Grassley, 2009). In addition to the proposal on company reporting, MedPAC has also proposed that the Congress require hospitals and other providers to report (and the government to post on a public website) on physiciansâ direct 18â The NPIï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ is a unique number mandated by the U.S. government for most U.S. physicians that is available in a publicly accessible database that links it to the physicianâs name, practice location, business office location, license numbers, and other identifiers.
96 CONFLICT OF INTEREST or indirect ownership shares in the facility (MedPAC, 2009; see further discussion in Chapter 6). The provision of recommendations on conflicts of interest arising from physician ownership of facilities was outside this committeeâs charge. The reporting program proposed by MedPAC would make considerable additional information available to researchers, patients, and others. A discussion of the pros and cons of establishing a broader system of disclosure is presented in Appendix F. Recommendations in the Following Chapters The recommendations in this chapter call for institutions to adopt con- flict of interest policies consistent with the recommendations in this report and for individual and cooperative institutional efforts and legislative ac- tions to strengthen policies on the disclosure of individual and institutional financial relationships with industry. The next four chapters of this report offer additional recommendations related to policies and practices in the specific areas of medical research, medical education, patient care, and the development of clinical practice guidelines. Chapter 4 calls for institutions to generally bar researchers with a conflict of interest from conducting re- search with human participants except when the investigatorâs expertise is essential to the safe and rigorous conduct of the research. Chapters 5 and 6, among other recommendations, call for physicians and researchers to forgo and institutions to prohibit or end certain relationships with industry that present unacceptable risks of undue influence over professional decision making or a loss of public trust. Chapter 7 includes recommendations for reducing industry influence in the development of clinical practice guidelines and increasing the levels of disclosure of organizational and individual financial relationships. Chap- ter 8 recommends that institutions establish policies at the board level to identify, limit, and manage institution-level conflicts of interest. The final chapter calls for a range of organizations to develop incentives to promote the institutional adoption and implementation of the policies recommended here. It also calls for the development of a research agenda to evaluate and guide improvements in conflict of interest policies and procedures.