Institutional financial conflicts of interest policies at Canadian academic health science centres: a national survey.

Financial conflicts of interest (fCOI) are of particular concern in the conduct of human subject research, whether they occur at the level of individual investigators or at the level of the institution.1 Institutional fCOI can occur when an institution that hosts the research, or a senior institutional official acting on behalf of the institution, has a financial interest in the study outcome.2,3 Such conflicts can be detrimental to research subjects,4 lead to an inappropriate degree of control over what should be an independent research agenda,5 and have serious implications, even in the absence of research misconducts.6,7 Moreover, undisclosed fCOI undermine the public’s confidence in science. For example, senior National Institutes of Health (NIH) officials were allowed to receive income as consultants to drug companies.6 The concern that this income might inappropriately influence their work resulted in the formation of a national Blue Ribbon committee to examine NIH conflict-of-interest policies.7 The committee’s 2004 report recommended that senior management should not consult with companies whose interests could influence the outcomes of their research6 ultimately led to a moratorium on industry-paid consultancies held by any NIH employee. Despite increased scrutiny of this issue on the part of government,6 the media,8 and the public,9 academic institutions have been slow to develop policies related to institutional fCOI.1 We conducted a content analysis of the institutional fCOI policies in use at Canadian academic health science centres (AHSCs) to identify gaps in policy coverage and to guide policy improvement. Our research methods are more fully described in Box 1. In brief, we collected institution-level fCOI policies from all 16 AHSCs (16 medical schools and 47 teaching hospitals as well as their 16 partner universities) from August 2005 to February 2006. These centres are the major sites of academic research involving humans in Canada. We contacted the vice president (VP) of research (or equivalent) at each site and asked him or her to identify 3 key institutional fCOI policies at their institution. To evaluate policy comprehensiveness, we compared each unique policy to our “standard” of 16 core items relevant to institutional fCOI derived from the key COI documents.2,10,11 Box 1 Description of methods All 16 universities (100%), their 16 medical schools (100%), and 42 (89%) of the teaching hospitals responded to our request for policies. Nine (56%) universities, 9 (56%) medical schools, and 15 (36%) teaching hospitals responded that they had no policies on institutional fCOI. Of the 72 policies identified, 34 were shared within AHSCs. Approximately a quarter (26%) of universities, medical schools, and teaching hospitals identified more than one institutional fCOI policy. Only 6 (16%) included “institutional conflicts of interest” in the policy title. Five (13%) policies were cross-referenced to other relevant policies from their institution. Of the 2 teaching hospitals that used internal and partner-university policies, neither cross-referenced the other’s policies. The ability of some leaders within an AHSC to identify a fCOI policy, while others based in the same AHSC could not, suggested that communication within AHSCs about fCOI policies was limited. For example, one teaching hospital said that its submitted fCOI policy was also used by its parent university; however, the parent university reported having no relevant policy. Table 1 describes a content analysis of the 38 unique policies. Definitions that informed our content analysis are listed in Box 2. Of the policies analyzed, 2 contained no items of relevance to institutional fCOI. On average, individual policies contained 20% of the 16 core “standard” items: no individual policy contained more than 65% of the core fCOI items. Even when the content of up to 3 policies per site was combined, less than half of the core items were addressed. Less than a quarter of policies addressed royalties, equity interest, or ownership interests. Box 2 Glossary of terms Table 1 Comprehensiveness of institutional fCOI policies at Canadian academic health sciences centres (universities, medical schools and teaching hospitals) Our results demonstrate that more than half of Canadian universities, half of medical schools, and more than a third of teaching hospitals had no institutional fCOI policy at the time of our survey. This is consistent with a 2006 survey of 86 deans of US medical schools (response rate 86/125), which found that fewer than half (38%) reported adopting an institutional fCOI policy.18 Further, policies were inadequately comprehensive and often difficult to locate. Like all studies, our work has limitations. First, we requested a maximum of 3 policies from each site; some may speculate that this truncated our results if sites had more than 3 policies relating to the core fCOI items. (For example, the Pennsylvania School of Medicine reports having more than 90 policies regarding conflicts of interest.19 However, given that 33 of 74 (45%) sites reported no relevant policies and only 10 (13%) sites identified 3, we think this is unlikely. Second, we used a stringent definition of “policy” and recognize that this would have excluded other terms used to label policies (e.g., “protocols,” “statements” or ”standards”). Moreover, a strategic position taken by the institution or mechanisms to deal with institutional fCOI could have been in place without being articulated in a policy and thus would have been missed by our research method. Third, our research focused on the adoption of policies and gave particular attention to their comprehensiveness. This does not capture the measures established at each institution to manage fCOI. Some AHSCs may have very detailed policies that are ineffective because they are not implemented or enforced, while others may have a poorly developed formal policy while still having effective mechanisms to address and manage institutional fCOI. Even detailed policies may not be sufficient to anticipate all issues related to institutional fCOI that could arise within an academic institution.20 While simply having a comprehensive policy is not sufficient, it is a usual means of communicating expected standards in academic institutions.18 Fourth, the list of the 16 items we used to evaluate the comprehensiveness of a policy was based on information from the AAMC2,11 and the AAU 10 that was available at the time of our survey. The 2008 AAMC template policy 1 contains the 16 core items we identified as being central to an institutional fCOI policy and adds further clarification. We expect that relevant policy items will continue to evolve. Further, we evaluated only whether the policy mentioned the core fCOI items. An evaluation of the quality of information provided about each core item may reveal further deficiencies. Finally, this study was conducted on policies in place in 2006. Given the requirement for Canadian institutions holding federal funds to put conflict-of-interest policies in place by January 2009,21we expect that many Canadian AHSCs are actively developing and implementing their institutional fCOI policies. Attention will need to be paid to having university-wide fCOI policies that are sensitive enough to capture issues specific to medical schools, or allowances will need to be made for medical schools to have a supplemental fCOI policy. In summary, over half of the Canadian AHSCs lacked institutional fCOI policies at the time of our survey. Where policies existed, they were not comprehensive and were frequently difficult to access. The 2008 Report of the AAMC-AAU Advisory Committee on Financial Conflicts of Interest in Human Subjects Research1 offers a thoughtful discussion on the complex institutional fCOI issues and provides a useful template for institutional policy. Other hospitals and universities that are not affiliated with AHSCs will also need to develop fCOI policies if they receive Canadian Tri-Council federal funds. Professional societies and those involved in clinical practice guideline development may also wish to develop these policies.22 We trust that our results related to core policy items will support appropriate policy development in this area.