Introduction

Therapeutic discovery is an expensive process. As clinical research becomes increasingly reliant on industry sponsorship,1 oncology in particular emerges as a specialty with heavy ties to private sources of research funding, especially when compared to other specialties.2 Physicians in oncology specialties in 2014 were 1.09 to 1.75 times more likely than those in non-oncology specialties to receive industry-related payments that could be considered potential financial conflicts of interest (COIs).2 However, it should be noted that the presence of financial COIs is not indicative of biased or dishonest research. Recent publications have demonstrated the lack of significant correlations between the presence of financial COIs and research outcomes.3,4 Yet a prevalent body of literature raises concerns to the contrary5,6 and should not be ignored in the characterization of COIs in the current landscape of cancer research.

With the introduction of the Centers for Medicare & Medicaid Services (CMS) Open Payments program in 2013, federally mandated reporting of private industry involvement in research made the subject of COIs in research both more public and more uniform than it had been when COI disclosure was dictated by individual journal policy. The need to standardize disclosure policy has been demonstrated,7 making current self-disclosure information obtained solely through articles both inconsistent and incomplete. Open Payments, on the other hand, offers a free-access, centralized database of payment information, sortable for type, source, and nature of payment. The service is undoubtedly an invaluable tool in maintaining researcher accountability but suffers from serious issues of its own.8 This makes the need for an accurate self-disclosure protocol even more important to ensure that neither source of information prevails with incomplete or inaccurate data. Though Open Payments added a comparable source of disclosure information to article level COI statements, self-disclosed financial relationships in literature are still arguably the more accessible COI information available to readers, who may not choose, or know how, to look beyond the article to find potential COIs.

In this study, we investigated whether discrepancies between self-reported and industry-reported COI existed among physician authors (MD or DO) of cancer research journals. We also sought to characterize the extent of this discrepancy between male and female physician authors to better understand the current landscape of available disclosure transparency.

Methods

Original research articles published in The Journal of Clinical Oncology, Cancer Research, and Nature Reviews Cancer from January 1, 2017, to December 31, 2018, were identified (N= 525). These three general oncology journals were selected based on high impact factors (33.0, 9.1, and 53.0, respectively), and an appreciable portion of the last authors were practicing physicians in the United States, which was determined both by author institutional affiliations published in the articles and Google search. Only last authors who were US-based physician (MD or DO) were selected for investigation. Articles written by last authors who were physicians based outside the US or who did not have a record in Open Payments were excluded. A total of 67 eligible authors were identified based on the selection criteria.

Data collection for the 67 authors included sex, degree (MD or DO), and general payment information in CMS Open Payments. General payments include fees related to consulting, travel and lodging, food and beverage, services other than consulting, education, honoraria, licenses, and speaking at accredited educational or academic programs. Associated research funding, research payment, ownership and investment were excluded. Self-reported financial disclosures included in the articles were compared to data reported to the CMS Open Payment Program to determine the accuracy of COI disclosures among authors. A match between the two criteria was documented as proper COI disclosure. A mismatch between the two criteria was documented as failure to properly disclose COI.

Authors that received general payment were stratified into monetary percentile ranks and two groups were formed: <50th percentile and >50th percentile of the received payment. The COI disclosure rate among <50 percentile authors were compared with the >50 percentile authors using chi-square analysis. Discrepancies between male and female authors were investigated using chi-square analysis.

Results

Based on the exclusion criteria, 67 last authors were eligible for the study. Among them, 43 authors (64.2%) received general payment according to CMS Open Payments. Together, these authors received $566 937.61. Of these 43 authors, 28 authors (65.2%) properly disclosed the COI in the COI statement of their published articles, whereas 15 authors (34.8%) failed to disclose properly (Figure 1). In addition, we also examined the percentage of authors who properly disclosed the COI by journal among Journal of Clinical Oncology, Cancer Research, and Nature Reviews Cancer. We found that the COI disclosure rate was 77.8%, 63.6% and 50.0%, respectively. The difference among the three journals was statistically non-significant (Figure 2).

Figure 1.COI disclosure distribution of last author.

Of the 67 authors who met the eligibility criteria for this study, 43 (64.2%) received a general payment according to CMS Open Payments. Although 65.2% of these authors disclosed this information, 34.8% failed to do so.

Figure 2.COI disclosure rate of journals.

Journal of Clinical Oncology (JCO), Cancer Research (CR), and Nature Reviews Cancer (NRC) had rates of disclosure of 77.8%, 63.6%, and 50.0%, respectively (JCO vs CR, P = .4084; JCO vs NRC, P = .1008; CR vs NRC, P = .4954).

The mean general payment received by the authors was $8 590.7. Of authors in the <50th percentile earner group, 36.4% did not properly disclose COI, whereas 33.3% of the authors in the >50th percentile earner group failed to disclose COI properly. Upon chi-square analysis, the difference was not statistically significant. Finally, we also investigated if there was a gender disparity in terms of proper disclosure of COI. Among the 43 authors who received general payments, 12 (27.9%) were female and 31 (62.1%) were male. The gender of the authors was obtained through Google search. Proper disclosure of COI occurred with 74.2% of male authors and 41.6% of female authors (P = 0.045) (Figure 3).

Figure 3.COI disclosure rate between genders.

Of those who received general payments, 41.6% of females and 74.2% of males were found to disclose COI properly (P = 0.045). The * indicates a statistically significant difference between male and female authors in disclosing COI properly via chi-square analysis.

Discussion

We found that more than one-third of the authors failed to disclose COI information properly. We did not find any significant differences in COI disclosure rates between the Journal of Clinical Oncology, Cancer Research, and Nature Reviews Cancer, but we did find that female physicians were less likely to accurately disclose potential financial COIs than their male counterparts. While it is not clear why a disparity was observed between self-reported vs CMS-reported COI disclosures, variable journal disclosure policy7 and non-standardized COI definitions8 may contribute to difficulties experienced by both authors and journals in determining what financial relationships and payment types are relevant for self-disclosure. Furthermore, general payments not directly related to the studies in question may never be explicitly stated.

Universal disclosure policy and the establishment of standardized COI language offer a solution to these and other issues plaguing disclosure currently. These would alleviate the difficulties of deciding which financial information is relevant for physician authors, increase the accuracy and wealth of accessible reader information, and improve the transparency of all COI information available by providing a practical way to compare information between self- and industry-disclosed information for accuracy and completeness. We believe complete disclosure has the potential to improve the credibility of all works by authors with financial ties to industry, allowing readers to develop informed opinions about the information shared in scientific literature.

When performing data collection, we aimed to identify the same number of US-based authors holding MD or DO degrees across the three journals. However, the percentages of last authors who were US-based doctors in Cancer Research and Nature Reviews Cancer are much lower than Journal of Clinical Oncology. Therefore, a significantly larger number of articles from more volumes over a longer timeframe from these two journals were reviewed compared to Journal of Clinical Oncology. Thus, one of the potential shortcomings of the study is that journals weren’t strictly selected over the same period due to the difference in the makeup of the last authors of the three journals. In addition, we were not able to compare discrepancies in COI disclosure practices by race or ethnicity due to the lack of self-reported information of the last authors.

Further research with a larger sample of authors or journals is recommended to explore disparities in COI disclosure to better understand how, if at all, the lack of proper self-disclosure in the presence of potential financial COIs impacts the results of research. Additionally, to begin formulating universal policy and standard COI terminology, research aiming to characterize the variation within current disclosure policy and COI language may be of great benefit to the process as well.


Acknowledgement

We would like to thank Darina L. Lazarova, PhD, for guidance and support throughout our research and Iris Johnson for her help in obtaining interlibrary loans for literature used in this project.

Disclosures

BJP is part of an osteoarthritis research team supported by Pfizer and Eli Lilly. The rest of the authors have no financial disclosures. BJP is also a member of the Editorial Board of The Guthrie Journal, for which he receives no financial compensation.

CREDIT Contributor Roles

Shuyi Chen: Conceptualization, data curation, software, writing—original draft

Alivia Roberts: Data curation, visualization, writing—original draft

Kevin Zhao: Investigation, writing—original draft

Abigail Burke: Data curation, writing—original draft

Jesse Ritter: Data curation, writing—original draft

Katherine Musto: Project administration, writing—reviewing and editing

Brian Piper: Funding acquisition, supervision, writing—reviewing and editing

Publisher’s Note

A preliminary version of this article appeared in Scholarly Research in Progress (SCRIP), which is “aimed at promoting and disseminating student scholarly activity at Geisinger Commonwealth School of Medicine.” See Chen S, Roberts AL, Zhao K, Burke AC, Ritter JE, Musto KM, and Piper BJ. Conflict of interest disclosure accuracy among physician authors of cancer research journals. Scholarly Research in Progress. 2021;5:89-91. This version underwent full peer review by The Guthrie Journal prior to acceptance. Published with permission.