For diabetes and hyperlipidaemia, we found that conflicts of interest (COI) were present for the vast majority of guideline panels reviewed. We also found that among panels that had chairs, half of these had COI. In addition, among panellists with an opportunity to publicly report COI, one out of nine reported no COI but were found to have undeclared COI. Our data illustrate the pervasiveness of COI among members of guideline panels and may raise questions about the independence and objectivity of the guideline development process in the United States and Canada.
A recent study of COI for members of cardiovascular guideline panels also found that COI were common.21
However, this study was limited to guidelines produced by the American College of Cardiology/American Heart Association, an organisation whose guideline development process has previously been scrutinised for influence by industry.22
Our study includes a wide range of guideline producing organisations and can be applied to the general guideline development process in the United States and Canada. Furthermore, our study exposes the problem of incomplete disclosure and highlights the important relation between sponsorship of guidelines and presence of COI. Our results are similar to those of a study done more than a decade ago, which found that 59% of authors of guidelines admitted to having received funding from manufacturers whose products were included or considered in the guideline. However, only two of 44 guidelines studied at that time had publicly available disclosure documents, and COI were documented only through self report.23
Our study shows that in spite of increasing requirements for disclosure of COI by guideline panellists the rate of COI remains similar, suggesting that mandating transparency may not translate into decreased COI on guideline panels.
Among guideline sponsoring organisations with public disclosure of COI, we found that government sponsored panels had significantly fewer panel members with COI, suggesting that expert panels can be convened from members predominantly without COI. However, although our sample included six government sponsored guidelines, only two of them (one Canadian and one US) included a public statement on the COI of panel members. The remaining four were produced by the Veterans Administration and the US Preventive Services Task Force (USPSTF). The USPSTF is the main government sponsored body in the United States that issues guidelines on screening and has a strict policy of managing its members’ conflicts of interest.24
However, task force panel members’ COI disclosures can be viewed only after a request under the Freedom of Information Act is filed with the government. This process is unnecessarily cumbersome and does not lend itself to easy scrutiny of this information by providers or the public. Similarly, the Veterans Administration does not disclose its panel members’ COI. In contrast to the USPSTF, the Veterans Administration presents no apparent official policy on the management of COI among panel members. This is in contrast to NICE, which provides detailed policies and procedures on disclosure as well as COI declarations on its website.16
Although this lack of full transparency by government sponsored panels in the United States is troubling, our search identified only three members of USPSTF and Veterans Administration guideline panels with undeclared financial COI, suggesting that COI are managed well on government sponsored panels.
In contrast to government sponsored panels, we found that COI were very common among panel members for guidelines produced by specialty societies. Guidelines produced by non-government sponsored organisations have been shown to be of poor methodological quality25
; however, they contribute substantially to the guideline pool in the United States and Canada, with specialty societies alone accounting for almost 40% of guidelines in the National Guideline Clearinghouse.17
Furthermore, several of the specialty societies included in our review have international prominence outside of North America, and their guidelines may have broad international influence. The high prevalence of COI among panel members of guidelines sponsored by specialty societies combined with the less rigorous development process may adversely affect the independence and the evidence base of the recommendations issued.26
Our study has several strengths and limitations that deserve comment. Five panel members’ conflicts were counted twice because our reporting of COI was based on panel membership and was not limited to unique individuals. If we had reported COI on the basis of the 265 unique individuals who participated in the process, our declared COI rate would have been 50% (n=133). This alternative analysis would not change the implications of our findings. We compared the prevalence of COI among government sponsored guidelines with that of non-government sponsored guidelines, and many government sponsored guidelines did not have accompanying declarations of COI. However, to identify potential conflicts we did a comprehensive Medline and Google search for each panel member. In addition, to account for potential bias related to opportunity for disclosure, we did a sensitivity analysis and limited the sample to guidelines with disclosed COI. This analysis confirmed our finding that COI were less prevalent among guidelines sponsored by the government.
We also note that our methods were deliberately conservative and may have underestimated COI among guideline panellists. We considered a financial conflict of interest to be present only when clear documentation of compensation of the panellist by a manufacturer of a product mentioned in a guideline was present. For the two guidelines in our study that were directly sponsored by drug companies, we searched each panel member individually for COI, rather than assuming they had received compensation for their work on the guideline. We also found that many members of guideline panels were authors of papers reporting on industry sponsored clinical trials or had chaired industry sponsored conferences. We did not count these relationships as financial conflicts; however, these panel members probably received funding from industry. We also did not include financial conflicts that did not relate directly to a guideline’s content or that did not directly involve the panel member of concern. For example, if a panel member’s spouse was an employee of the drug industry, we did not consider it a conflict. We were also unable to account for relationships with industry that were not publicly available and did not search for additional conflicts if one was already declared. Both of these factors probably limited the number and scope of COI found in our study.
Finally, our sample size was limited to only two disease conditions. However, together, these two conditions are predicted to be among the top five drug treatment categories in the world by 2015, with combined consumer spending of up to $82bn (£52bn; €59bn).27
Furthermore, our results are more generalisable than are those of other recent studies of its kind. In contrast to the recent paper by Mendelson et al,21
which focused on one guideline producing organisation, our study sample included guidelines from many different organisations, which provides a more realistic estimate of the prevalence of COI on guideline panels.
The finding that most current members of guideline panels and half of chairs of panels have COI is concerning and suggests that a risk of considerable influence of industry on guideline recommendations exists. The Institute of Medicine recently recommended that guideline panels should be as conflict-free as possible,17
The minimisation of COI on guideline panels is likely to do far more to mitigate bias than would mere disclosure. The guideline development process needs to be reformed to minimise conflicts of interest among panel members to ensure the credibility and evidence based nature of the clinical practice guidelines issued in the United States and Canada. The limited COI we identified among panel members who participated in developing government sponsored guidelines suggests that expert panels without many COI can be convened. Future research should examine whether the presence of COI diminishes in the light of the international trend, articulated by the Institute of Medicine, towards both transparency and exclusion of panel members with COI. Conflict-free guideline panels are feasible and would help to improve the quality of the guideline development process.
What is already known on this topic
- Conflicts of interest (COI) among panel members are common in guidelines issued by certain specialty organisations
What this study adds
- The prevalence and under-reporting of COI are high and transparency is incomplete among a wide range of guideline producing organisations
- An association exists between the source of sponsorship of guidelines and the presence of COI