Requirements of health policy and services journals for authors to disclose financial and non-financial conflicts of interest: a cross-sectional study

Ninety-three percent of health policy and services journals had a COI policy. A minority of policies described how the disclosed COIs of authors would impact the editorial process; none had clear-cut criteria for rejection based on the content of the disclosure. About a fifth of policies explicitly stated that inaccurate or incomplete disclosures might lead to manuscript rejection or retraction. No policy described whether the journal would verify the accuracy or completeness of authors’ disclosed COIs. For financial COIs, the majority asked for specification of the source of payment; a minority asked for the amount (Table 3). Conversely, 81% of policies explicitly required disclosure of non-financial COIs (Table 2).

This is the first study to investigate the COI disclosure requirements in health policy and services journals, and only the second study investigating how health journals address the potential impact of COI disclosures and incomplete or inaccurate COI disclosures on the editorial process [18]. We used systematic methods for data abstraction (i.e. in duplicate), and evaluated the actual implementation of the COI disclosure policy during the submission process. We do not believe that any of the authors had biases against any of the included journals. Specifically, the team members who abstracted data were naïve to the list of included journals and had no personal experience with any of them. One limitation of this study is our inability to capture the COI policy details that might emerge during the later stages of the editorial process. Another limitation of our study is that we included journals listed under the category of ‘Health Policy and Services’ by the Web of Science. This list might not have included journals listed under similar categories in other databases such as PubMed. However, we do not have a reason to suspect that this might have affected the representativeness of our sample.

We have recently conducted a very similar survey addressing policies for COI disclosure but focused on the core clinical journals. All but one of the 117 clinical journals (99%) had a policy for the disclosure of COIs [18], compared to 93% of health policy journals. Two other studies found that 87% of oncology journals and 95% of high-impact biomedical journals adopted COI policies [19, 20].

Eighty-one percent of health policy journals’ policies asked authors to disclose their non-financial COIs, a percentage higher than what previous studies have found. Shawwa et al. [18] recently found that 57% of the core clinical journals asked authors to disclose at least one form of non-financial COI. Earlier, Kesselheim et al. [20] found that 42% of oncology journals asked for non-financial COI disclosure. Health policy journals seem to appreciate the potential impact of non-financial COIs in this field, as compared to other fields (e.g. clinical) where the focus has been on financial COIs.

Table 5 compares the findings of this study to those of a similar study focused on clinical journals [18]. Similar percentages of health policy and clinical journals mentioned a potential impact of inaccurate or incomplete disclosure of COIs on the editorial process (21% and 23%, respectively) [18]. However, medical journals described more severe implications than health policy journals, including manuscript rejection or retraction, ‘sanctions’ and prohibition from future submissions. Moreover, health policy journals were less likely than medical journals to report verifying the disclosed COI (0% vs. 17%).

Box 1 provides suggestions for items that journals could include in their policies for authors’ disclosure of COIs. Indeed, the editor of a German journal reported that the introduction in 2005 of a policy requiring all authors to disclose their COI increased disclosures from none in 2002 to 30% of the published articles in 2006 [21]. Given our findings, health policy journals could improve on specific items such as definitions and instructions for disclosing non-financial COIs.

Although there is little evidence regarding the amount of money that is likely to bias decisions and judgments, it may nevertheless be advisable for journals to ask for more specific details about financial COIs, including the amount of money involved. Lewison et al. [22] suggested a COI statements registry, where authors could declare their competing interests, that could be effortlessly followed over time to verify the authors’ declared interest. Requirement for reviewers and the editorial team to disclose their own COIs, and how these are managed, represents another major area where policies could be improved (e.g. editors withdrawing from handling manuscripts on which they could be conflicted).

Further insights into conflicts of interest issues will require a framework that provides clear typology and operational definitions to facilitate both the disclosure of COIs by authors and reviewers and their assessment and management by the editorial team. Optimal progress will also necessitate development of a valid and reliable approach to verify the different types of disclosed COI. Future research should also build the evidence base for the different interventions and policies intended to manage COIs.