Vogeli, Christine PhD; Koski, Greg PhD, MD; Campbell, Eric G. PhD
The often close relationship between academia—specifically, academic medicine—and industry has been highlighted in a number of sensational stories in the popular and academic press.1–3 Often, the focus of such articles is on individuals with close ties to industry, but conflicts of interest (COIs) also exist at the institutional level, and these may pose far greater threats to the integrity of science and medicine. Institutional conflicts can arise through close institutional ties to industry, via patent agreements or financial investments, for example, but they also can arise through the individual conflicts of officers or others who act on behalf of the institution. One such group is members of institutional review boards (IRBs) who review and approve research protocols on behalf of the institution. These persons serve the institution, yet they are often seasoned investigators who frequently have relationships with industry. Prior research has found that some IRB members in academic medical centers receive research funding from industry or financial compensation for participation in meetings and conferences; serve as paid consultants or members of speakers’ bureaus or scientific advisory boards; serve as an officer, executive, or paid employee in industry; receive royalties or equity; or serve on a board of directors. Altogether, more than one third of IRB members may have one or more such relationship with industry in any given year.4
The ramifications of IRB members’ industry relationships and COIs are not well understood, but the perception is that the risks to IRB functions are real. For example, in 1999, at least 2000 studies involving human subjects at Duke University were suspended after the federal government determined that Duke did not adequately protect patients involved in such studies.5 Among the 20 deficiencies cited by the Office for Protection from Research Risks (now the Office of Human Research Protections, or OHRP) was that the COIs of senior-level administrators at Duke had compromised the ability of the IRB to make independent and unbiased decisions regarding research protocols.
The concern over the potential impact of industry relationships on institutional and IRB decisions with regard to medical research has led to a number of discussions and recommendations from both federal and nonfederal organizations,6–8 with the most recent report issued by a joint Association of American Medical Colleges (AAMC) and Association of American Universities (AAU) advisory committee in February 2008.9 However, despite the promulgation of federal guidance by the U.S. Department of Health and Human Services,10 there are currently no regulatory requirements placed on IRBs. Current federal regulations state only that IRB members may not participate in discussions or decisions on any protocol in which they may have a conflicting interest, except to provide information.10 45 CFR 46.107e This policy provides no standards on what constitutes a meaningful COI, nor does it suggest policies and procedures to protect IRB-related activities from conflicts. The most recent recommendation by the joint AAMC–AAU advisory committee asks institutions to require the routine reporting of data on all financial interests from all IRB members and to establish clear policies for defining COIs, but it lacks a concrete recommendation about who should be responsible for these activities within the institution.9 Establishing a conflict of interest committee separate from the IRB is one possible approach suggested by the National Human Research Protections Advisory Committee in their comments to the deputy assistant secretary for health on draft interim guidance in 2001.11 This recommendation was based in part on an internal investigation by the National Institutes of Health (NIH), which found that only 25% of all IRBs routinely deal with investigators’ and IRB members’ industry relationships and financial COIs during IRB deliberations.12 There are no recent data on the management of conflicts during deliberations or on other tactics IRBs use to manage relationships and COIs before final deliberations.
This study seeks to provide information on IRB-specific policies and practices with regard to industry relationships and conflicts and to determine whether current policies and procedures are consistent with existing federal guidance. This study examines how IRBs at the most research-intensive medical institutions in the United States manage industry relationships and COIs among their members and whether IRB chairs feel that current policies are sufficient to ensure the integrity of the IRBs’ activities.
The sample for this study was developed in a two-step process. First, we identified the 100 medical schools and the 15 independent hospitals that received the most funding from the NIH in 2003. We included these independent hospitals because they are among the most research-intensive institutions in the country. Second, using a Freedom of Information Act request to the U.S. Department of Health and Human Services OHRP and Internet searches, we identified all chairs of all IRBs at those institutions. Of the original 115 institutions, 5 were excluded because we could not obtain the names of the chairs in a timely fashion, 8 because the list of chairs we obtained from OHRP was more than two years old, and 2 because the institution was ineligible for the study because they were home institutions for the investigators. A total of 316 IRB chairs representing 100 institutions were included in the final sample.
Survey instrument development and testing
A survey instrument was designed by researchers at the Massachusetts General Hospital in conjunction with researchers at the Center for Survey Research at the University of Massachusetts. Survey questions were drawn from previous surveys of academic–industry relationships4,13 or were developed specifically for this project. We convened four focus groups with IRB chairs and members from medical schools and teaching hospitals in the Boston area to identify topics and items that chairs could reliably report. In addition, the survey was pretested using cognitive interviews among chairs (three) and a preliminary survey of nine chairs to improve question wording and readability. The study and the final survey instrument were approved by the human research committee of the Partners HealthCare System, which serves as the IRB for the Massachusetts General Hospital.
We mailed the survey to the 316 IRB chairs in the fall of 2005. To guarantee complete anonymity, the survey did not contain any identifying markings. To anonymously track nonrespondents, we asked chairs to mail back an identifiable postcard independently from their survey. Respondents for whom we received a postcard were removed from the follow-up mailings. A total of three survey mailings with follow-up phone calls were used to encourage participation. Participants were not paid or otherwise compensated for their participation.
Variables and measures
In this survey we sought to understand IRB processes related to the identification of industry relationships and COIs among IRB members and the processes IRBs have (or do not have) to manage relationships once they are identified.
Two screening questions asked whether the chair was currently active and whether he or she served as an alternate chair. In a series of questions, we then asked IRB chairs to provide information about their personal characteristics, including academic rank, educational background, race/ethnicity, gender, and the number of years of experience they have had in an IRB, as an IRB chair, conducting basic research, and conducting clinical research. We also asked about their IRB setting—specifically, whether their IRB was situated in a medical school, university, or hospital—the number of meetings annually, and the average number of protocols reviewed at each meeting.
We next asked a series of questions about the processes each IRB had in place to manage member relationships. Specifically, we asked whether the IRB had a “defined process for members to disclose financial and other industry relationships to the IRB” and which IRB members were “required to disclose financial and other industry relationships.” We then asked about how these policies regarding IRB member relationships worked in practice, specifically among the subset of IRB chairs who reported that one or more IRB members had disclosed an industry relationship within the past year: “In practice to whom did IRB members with industry relationships disclose those relationships,” and, finally, “Where should the oversight responsibility for members relationships with industry reside?” IRB chairs were also asked how confident they were that their “IRB’s policies and procedures provided the appropriate level of disclosure for every study in which member-industry relationships existed” and where the “oversight responsibility for IRB members’ relationships with industry should reside.”
IRB policies regarding COIs were assessed with questions asking whether the IRB had a “written policy defining when a member’s financial or other industry relationships are a conflict of interest,” how “clear” the policy was, and, in the absence of a written policy, “who decides whether an IRB member’s relationship with industry is a conflict of interest.”
Once COIs were identified, we assessed how these conflicts were managed by asking whether a respondent’s “IRB has written policies for actions once a conflict of interest is identified” and where the current responsibility for “determining what to do once a conflict was identified” lies. Possible responses were “the IRB member,” “the IRB chair,” “the entire IRB,” a “group or individual within the organization but external to the IRB,” or a “written policy.” IRB chairs were also asked to describe the actions IRB members took when COIs were identified. This question was similar to a question asked of IRB members in a separate survey and was included in the survey of IRB chairs for comparison.4 Finally, chairs were asked how confident they felt in hindsight that “your IRB correctly managed conflicts of interest between IRB members and industry in every case.”
We analyzed the data using SPSS and standard descriptive statistical procedures. Because we surveyed all identified IRB chairs meeting our study criteria, there was no error associated with sampling, and thus statistical testing and response weighting were not necessary. Furthermore, because responses were anonymous (meaning that we could not link responses to individuals or their institutions), we could not weight survey responses for differential nonresponse by IRB type or institution, nor were we able to control for the clustering of IRBs within specific institutions.
Of the 316 IRB chairs in the overall sample, 9 were removed because they were included in the pretest, 9 because they were no longer at the institution, and 3 because they were deceased. Of the remaining 296 chairs, 211 responded, yielding a raw response rate of 71.7%. Of these, 36 were excluded from this analysis because they reported that they were no longer serving as a chairperson. An additional 58 alternate chairs were excluded from our analysis to avoid duplicative responses from a single IRB and to allow us to regard the responses of IRB chairs as proxies for their IRB as an entity. Therefore, the analysis was limited to the 107 active chairpersons.
Characteristics of participating chairs and IRBs
Approximately half of IRB chairs serving the most research-intensive institutions in the United States were full professors (54, or 50.5%) and physicians (62, or 59.6%) (Table 1). The vast majority were white (93, or 94.9%) and male (83, or 82.2%). On average, IRB chairs had more than 14 years of general IRB experience and more than 6 years of experience as IRB chairs. IRB chairs averaged more than 16 years of clinical research experience.
Consistent with our sampling strategy, the majority of IRB chairs indicated that their IRB served a medical school (90, or 84.1%). On average, chairs reported their IRB had 2.5 meetings per month and that their IRB reviewed 8.7 protocols per meeting, for an average of 22 protocols reviewed in meetings monthly, or approximately 264 protocols annually.
Policies for disclosure of industry relationships within IRBs
Nearly three fourths (79, or 74.5%) of IRB chairs reported that their IRB had a defined process for members to disclose industry relationships to the IRB (Table 2). Overall, two thirds (70, or 66.0%) of IRB chairs reported that voting members were required to disclose relationships with industry, and 23.6% (25) reported that that nonvoting members were required to do so. However, among the 25.5% (27) of IRB chairs who responded that their IRB did not have defined processes for the disclosure of industry relationships, only 1 (3.7%) responded that voting members were required to disclose relationships.
Among the 54.7% (58) of chairs who responded that one or more IRB members had disclosed an industry relationship in the past year, most responded that, in practice, the member disclosed the relationship to the entire IRB (44, or 75.9%), and half (31, or 52.5%) responded that the IRB member disclosed the relationship to a group or organization separate from the IRB but within the institution it serves. Asked where the oversight responsibility for member industry relationships should reside, 25.2% (27) responded within the IRB, and 12.1% (13) responded that the IRB member should be responsible for monitoring their own relationships. IRB chairs most often responded (56, or 52.3%) that an individual or group external to the IRB but within the organization it serves should have oversight responsibility for member–industry relationships. Even among the subset of chairs (44) who reported that IRB members disclosed their relationships to the entire IRB, the most common response (18, or 40.9%) was that oversight responsibility should be held by an individual or group separate from the IRB but within the institution it serves.
Overall, 19.8% (21) of IRB chairs were not very or not at all confident that their current IRB processes and procedures provided the appropriate level of disclosure for every study in which an IRB member had a relationship with industry. Lack of confidence in disclosure processes or procedures was much more pronounced among chairs without a defined process for members to disclose financial and other industry relationships to the IRB: 61.5% (16 of 26) of chairs without written policies were not very or not at all confident in comparison with 6.3% (5 of 79) of chairs with policies.
COI policies in IRBs
The majority of IRB chairs (73, or 68.2%) reported that their IRB had a written policy or policies that define when an IRB member’s industry relationship should be considered a COI (Table 3). The remaining chairs reported that their IRB either did not have a policy defining COI (24, or 22.4%) or that they were not aware of such a policy (10, or 9.3%). Among chairs with a formal written policy of what constitutes a COI, half (35, or 48.6%) felt that their policy defining COIs was “very clear,” and 41.7% (30) felt it was “clear enough.”
Among IRBs without a formal written definition of a COI, the responsibility for determining whether an industry relationship should be considered a COI within the context of the IRB was held by a broad array of persons or groups. One in five (7, or 20.6%) chairs reported that within their IRB it was up to the individual IRB member to self-identify a COI. The same numbers responded that the responsibility lay with the IRB chair (7, or 20.6%), the entire IRB (7, or 20.6%), or a group or individual within the organization but separate from the IRB (7, or 20.6%).
Actions IRBs took when COIs were identified
Nearly three fourths (79, or 73.8%) of chairs serve IRBs that have written policies that define the appropriate actions to take when a COI is identified (Table 4). Among chairs who identified one or more COIs in the past year, 31.1% (14) reported that decisions about how to best manage the conflict were made by the entire IRB, 26.7% (12) by the IRB chair, 15.6% (7) by a group or organization separate from the IRB but within the organization it serves, and 13.3% (6) by the IRB member.
Among chairs with a member COI in the past year, all (47, or 100%) reported that within their IRB, members never voted on protocols on which they had a COI (Table 5). Fewer chairs (29, or 63.0%) responded that IRB members with COIs never participate in the general discussion regarding the protocol with which they had a COI, or that IRB members with COIs always left the room when the protocol in question was under consideration (30, or 63.8%). Overall, 34.0% (16) of IRB chairs who experienced a member COI within the last year were very confident, and an additional 61.7% (29) were moderately confident that COIs were correctly managed in every case within their IRB.
This national survey of IRB chairs serving the most research-intensive medical institutions in the United States provides the most current assessment of policies and practices of IRBs related to member relationships with industry and COIs within IRBs. The results display a diversity of approaches to disclosure and management of IRB member–industry relationships that is not surprising given the lack of clear federal policies and procedures. We found that voting members were not always required to disclose relationships and that IRBs have diverse practices for managing industry relationships. IRB chairs were not confident that current procedures for the identification of relationships and management of conflicts were effective, but there was no clear consensus on where oversight responsibility for member industry relationships should lie.
Although federal regulations do not specifically require that IRBs develop policies and procedures, they are clearly and strongly recommended in guidance from both the federal government and professional associations and are key elements of current accreditation processes for human research protection programs.7,9,10 The most recent guidance from the AAMC–AAU advisory committee recommends that institutions require the reporting of all financial interests by IRB members on their appointment and at least annually thereafter.9 However, remarkably, one third of IRBs in the nation’s medical schools and major academic medical centers report that they do not have such a requirement in place, and almost all of these lack a defined disclosure process.
This finding is surprising and concerning. IRBs without a defined process for disclosure or requirements for disclosure will likely not only have incomplete disclosure but also will lack the ability to appropriately manage COI within the IRB. Within our study, all IRB chairs responded that no members with COIs had voted on an IRB protocol in the past year. However, a companion survey of IRB members conducted at the same hospitals and medical schools found that only 64.5% of members with a COI never voted on the affected protocols, and 19.4% responded that they always voted on protocols when they personally had a COI.4 Chairs’ lack of awareness of IRB member actions likely stems from deficits in reporting requirements and points to a significant need for a strengthening of IRB policies and practices with regard to the reporting of industry relationships.
Our finding that in two thirds (66.0%) of IRBs, voting members were required to disclose industry relationships, is similar to but not entirely consistent with findings from Ehringhaus and colleagues14 that 89% of U.S. medical schools require IRB members to report significant financial interests to the institution. This inconsistency may illustrate the distinction between policy and practice and the difficulty of applying and making known institutional policies throughout all levels of these large and complex institutions.
In practice, when industry relationships were disclosed, it was most often to IRB chairs; however, chairs almost unanimously agreed that they did not want to hold the oversight responsibility for IRB members’ relationships with industry. The most common response among chairs of IRBs with and without defined disclosure processes was that oversight responsibility for relationships should lie with a group or individual separate from the IRB but within the organization it serves. Examples of such a body would be a COI committee or a senior member of the medical school or hospital administration with responsibility for the IRB under the terms of his or her federalwide assurances.
The impact of various policies currently in place on full disclosure and management of industry relationships needs to be explored before defining best practices. However, the lack of confidence expressed by knowledgeable chairpersons in the current disclosure process, especially among chairs of IRBs without defined processes for members to disclose relationships to their IRB, suggests that institutions currently lack the information to appropriately manage industry relationships and identify COIs within IRBs. Disclosure of industry relationships should be the norm for individuals involved in clinical research, including people serving on the IRB.
Our survey also found gaps in the identification of COIs in that a significant minority of medical school IRBs (31.7%) do not have or are not aware of policies defining when relationships are considered COIs. Among these, one in five leaves the identification of conflicts up to the individual member with the relationship, reducing the objectivity of these decisions. A recent review of IRB-specific policies at NIH-funded medical schools nationwide found that 46% lacked policies that defined COI, but the review did not include institutional policies that may apply to IRB faculty members.15 However, both our work and prior work suggest that a significant proportion of IRBs are not adequately protecting IRB-related activities from COIs in accordance with OHRP guidelines.
This study has several limitations that must be considered. First, our survey was designed to collect information on IRB policies and procedures with regard to relationships and conflicts. We did not collect information on the genesis of or the extent to which these IRB-specific policies and practices were integrated into a broader institutional approach toward relationships and conflicts. Second, we surveyed IRB chairs to assess the state of IRB policies and procedures regarding IRB members’ relationships with industry and the COIs those relationships can create. Although chairs’ assessments may be subjective and may inaccurately estimate true behaviors, especially behaviors that are viewed as inappropriate, there are currently no other sources for this type of data. Chairs’ responses do illustrate their understanding of the situations and their recollection of policies and procedures within their IRBs. Broader institutional policies and conflicts at the institutional level were beyond the scope of this study. Third, the chairs who responded to our survey may not be representative of all IRB chairs in all medical schools. To ensure anonymity to our survey respondents, we did not track the institutions of specific responders. However, we did exclude alternate chairs from our final sample so that each chair’s response represented one IRB. Our anonymous survey process also prevented us from comparing the characteristics of responders and nonresponders. Finally, our survey provided a snapshot of IRB policies and practices in 2005. Although policies and practices may have evolved, in the absence of external requirements such as federal guidance, the likelihood of significant change is small.
Overall, our study shows significant variation among academic medicine IRBs in all aspects of the disclosure and management of IRB members’ relationships with industry and a concerning level of unawareness of industry relationships among IRB members. Failure to address the issues raised could call into question the ability of medical school IRBs to appropriately manage their members’ industry relationships and appropriately identify COIs. Importantly, existing guidance from respected, authoritative sources, including professional organizations (e.g., the AAMC) and federal funding and oversight agencies (e.g., OHRP, NIH, National Science Foundation) should be reemphasized and used to develop consistent policies and procedures to be adopted by these institutions and, thereby, avoid confusing inconsistencies and their potentially damaging consequences.