Background
A conflict of interest (COI) is defined as “a set of circumstances that creates a risk that professional judgment or actions regarding a primary interest will be unduly influenced by a secondary interest.”1 In academic medicine, this conflict results from an incompatibility between one’s public duties, which involve protecting the integrity of research and clinical care, and one’s personal interests, which involve financial gain or potential professional advancement.2–4 These competing interests can be financial or nonfinancial in nature. Financial conflicts of interest (FCOIs) are easier to recognize than their counterparts because they result from receipt of payments or transfer of valuable items.5 Furthermore, FCOIs have unidirectional outcomes (ie, favorable impression toward the product with which one has financial conflicts).
FCOI has been recognized as an important policy issue in medicine and oncology, and some measures are in place to check the undue influence of such conflicts on research, publication, and policymaking. Previous research has established the prevalence of FCOIs among oncologists,6,7 oncologists on Twitter,8,9 authors of major cancer care guidelines,10–13 and even editorialists of major oncology journals.14,15 Editorials published alongside reports of clinical trials in top journals can potentially have a substantial influence on oncology practice because they contextualize trial results. To our knowledge, no prior study has quantified whether the presence of FCOI is associated with how editorials frame clinical trial results in oncology.
To address this gap in knowledge, we examined editorials accompanying trials of new cancer therapies published in major journals over a calendar year to assess the prevalence of any FCOI, the prevalence of FCOI with the same company whose product was being discussed (direct FCOI), and any relationship between these FCOIs and an unduly favorable interpretation of clinical trial results for new cancer therapies.
Methods
Study Identification
This study did not require ethics approval, because it did not include any patient data and all information was collected from published studies. We identified all editorials published in the print editions of the calendar year 2018 in 5 high-impact journals that publish oncology clinical trials: New England Journal of Medicine (NEJM), Lancet, Lancet Oncology, JAMA Oncology, and Journal of Clinical Oncology (JCO). NEJM and JCO publish their editorial pieces as editorials, whereas Lancet, Lancet Oncology, and JAMA Oncology publish their editorial pieces as comments; both types of report (editorials and comments) were reviewed in this study. We included these journals because they are widely read by clinicians and publish the most important studies in oncology; editorials in these journals therefore have the biggest potential influence on clinical practice. This method of convenience sampling using high–impact factor journals to understand publication patterns of practice changing trials is similar to that used in other previous investigations.16–18
We excluded editorials or comments that were not linked to any study published in the journal (such as commentaries on general topics of interest or editorial position statements) (Figure 1). Of the editorials linked to a particular study published in the same issue of the journal, we included only those that were linked to a drug study; that is, editorials related to studies about radiation therapy, diagnostic studies, or cancer surgery were excluded. Finally, because the intent of this study was to assess the association between financial conflicts of the editorial authors and biased interpretation of study results favoring the new drug, we excluded editorials about trials of older chemotherapeutic agents that are off-patent, trials of repurposed drugs (eg, aspirin or metformin), or de-escalation trials that tested a lower dose or lower frequency of the drug. We included editorials about all oncology drug trials: solid tumors and hematologic or pediatric malignancies.
Selection of editorials for inclusion in the analysis.
Citation: Journal of the National Comprehensive Cancer Network 19, 11; 10.6004/jnccn.2021.7016
Selection of editorials for inclusion in the analysis.
Citation: Journal of the National Comprehensive Cancer Network 19, 11; 10.6004/jnccn.2021.7016
Selection of editorials for inclusion in the analysis.
Citation: Journal of the National Comprehensive Cancer Network 19, 11; 10.6004/jnccn.2021.7016
Data Extraction and Variable Definitions
For all relevant editorials, we recorded the author’s FCOIs as reported in the journal. We also accessed and studied the trial that was the subject of the editorial and extracted data on the drug being tested and the company that manufactured the drug in question. Trials were categorized as “not RCT” for studies that were not randomized controlled trials (RCTs), such as single-arm phase I and II trials; “positive RCTs” for RCTs that met their primary endpoint regardless of trial design (ie, noninferiority or superiority trials); and “negative RCTs” for RCTs that failed to meet their primary endpoint.
We then categorized the FCOI of the editorialist as “yes” or “no” based on whether any industry COI was disclosed. Furthermore, whenever FCOI was present, we also checked whether the editorialist had an FCOI with the same company that manufactured the drug(s) being discussed in the editorial. We defined this as direct FCOI and categorized it as “yes” or “no.”
Editorials were classified as “yes” or “no” with regard to whether they were unduly favorable to the cancer drug in question. We defined an unduly favorable editorial as the following: a positive spin or clinical recommendation without discussing limitations for a non-RCT study, a positive spin for a negative RCT, and an excessively positive interpretation with no discussion of limitations for a positive RCT. Each editorial (and the accompanying clinical trial report) was reviewed independently first by one oncologist author (Oncologist 1) who categorized the editorials as “No” (ie, not unduly favorable), “Yes,” “No, but needs review,” or “Yes, but needs review.” All the “Yes” and “needs review” editorials (and the accompanying trial reports) were then reviewed by a second oncologist author (Oncologist 2 or 3) without any knowledge of the outcomes of review by Oncologist 1. Any discordance in assessment between Oncologist 1 and Oncologist 2/3 was resolved by discussion or consultation with the third oncologist author (Oncologist 2/3) if disagreement remained. The second and third reviewers (Oncologists 2 and 3) were unaware of the purpose of the study (ie, investigate association with FCOI or direct FCOI) until after they had classified the studies. However, the second and third reviewers were not blinded to the FCOI of the editorialists, because this information did appear by default at the end of each article PDF or on the journal site when accessed online. Any attempt to blind the COI info might have had the unintentional effect of bringing more attention to it.
For editorials with multiple authors, if any one author had an FCOI or a direct FCOI, the editorial was classified as such. We relied on authors’ self-disclosures for assessing the presence or absence of FCOIs, rather than checking the FCOI in the Open Payments database, for 2 reasons. First, not all editorialists are American, and the Open Payments database would not capture the data for non-American editorialists. Second, our aim was to assess whether the journals allowed editorials by authors who self-disclosed their relevant FCOIs rather than to investigate the veracity of disclosures.
Statistical Analysis
The association between presence of an FCOI or a direct FCOI and a favorable editorial was examined using the Fisher exact test or chi-square test as appropriate. All analyses were performed using Microsoft Excel and STATA, version 15 (StataCorp LLC). A 2-sided P value <.05 was considered statistically significant.
Results
Of 665 editorials and comments published in the top journals, 243 were related to oncology. The final study cohort included 90 editorials based on the inclusion and exclusion criteria (Figure 1). Characteristics of the editorials and their accompanying clinical trials are shown in Table 1. Of the studies about which the editorials were written, 41 (45%) were non-RCT studies, 17 (19%) were negative RCTs, and 32 (36%) were positive RCTs. Twenty-one studies (23%) met criteria for double review by the authors. There was 71% agreement (15/21) between pairs of authors regarding the presence or absence of bias (favorable editorial), with a Cohen’s kappa statistic of 0.44, suggesting moderate agreement. The agreement increased to 95% (20/21) (Cohen’s κ statistic of 0.88, suggesting almost perfect agreement) after discussion, with only 1 editorial requiring the opinion of a third reviewer.
Characteristics of the Editorials and Their Accompanying Clinical Studies
Seventy-four percent (67/90) of editorials were written by author(s) with at least one declared FCOI with the pharmaceutical industry. In 39% (35/90) of editorials, the authors had FCOIs with the same company whose product was being discussed (direct FCOI). Overall, 11 editorials (12%) were deemed to be unduly favorable to the product being discussed, of which 8 (73%) had editorialist(s) with an FCOI, all of whom also had a direct FCOI with the same industry.
There was no significant association between the editorialist(s) having any FCOI and the editorial being unduly favorable to the drug being discussed. Editorials were classified as being unduly favorable for the study drug among 12% (8/67) and 13% (3/23) of those with and without any FCOI, respectively (Fisher exact test, P=1.0) (Table 2).
Association of FCOI and Unduly Favorable Editorials
However, the association between having a direct FCOI and writing an unduly favorable editorial for the drug of the same company was significant. Editorials were classified as being unduly favorable for the study drug among 23% (8/35) and 5% (3/55) of those with and without direct FCOIs, respectively (Fisher exact test, P=.009) (Table 2).
Discussion
In this analysis of editorials published in major oncology journals, we found that 74% of editorials were authored by at least one author with industry-related FCOIs. Of even greater concern is the finding that 39% of editorials were authored by an individual who disclosed an FCOI with the same company that produced the drug evaluated in the trial about which the editorial was written. Our data demonstrate that this is not just a problem with optics; editorialists with direct FCOIs were more likely to author an unduly favorable editorial for the drug. These findings have important implications for journal editorial policies and consumers of journal articles, such as physicians, patients, and policymakers.
Studies have consistently shown that FCOIs have important implications in medicine. FCOI is associated with research productivity,19 favorable stances on controversial topics,20–22 and physician prescribing behavior.23 A prior analysis of 3 high-impact general medicine journals reported a direct FCOI rate of 18.2%.15 We report a direct FCOI rate that is 2-fold greater than the prior results (39%), suggesting this may be an even larger problem within oncology. A recently reported study of FCOI among leaders of US medical organizations also suggested that FCOIs may be more prevalent and more numerous among the leadership positions in oncology than in other medical disciplines.24
The impact of FCOI is especially important in oncology, where there is a large presence of the pharmaceutical industry, and treatments are expensive and often associated with significant toxicity with sometimes modest benefits.5,19 Many clinical decisions in oncology need to be made in the absence of level I evidence; the voices of editorialists can therefore have a major influence on practice and policy. Prior studies have reported that among the authors of major international cancer care guidelines, up to 86% had FCOIs,25 and a significant proportion also had undisclosed FCOIs.12 Similarly, the editorialists of major cancer journals also represent key voices in oncology that help contextualize clinical trials in practice, and, for this reason, it has been suggested that editorialists be free from industry conflict.5 Our study demonstrates that not only is FCOI pervasive among the authors of oncology editorials but also these authors are more likely to present the results of clinical trials in a manner that may be unduly favorable for the drug.
An important means to address FCOI in medicine has been disclosure, and we also relied on authors’ self-disclosure in our study. Some experts argue that promoting transparency by disclosing COIs is key, so that reviewers and readers have enough information to judge for themselves whether the work is biased.3,4,26–28 Other experts argue that declaring COIs is not enough, because that does little more than convince suspicious readers that their suspicions were warranted.29,30 However, there have been studies suggesting that disclosures are frequently biased toward underreporting of FCOIs,31,32 and some high-profile examples in oncology of failure to disclose FCOIs in publications have recently come to light.33 There have been several instances of incomplete COI disclosure that compromised the credibility of research findings.34–36 However, the purpose of our study was not to verify the truthfulness of disclosure. Rather, we wanted to assess the prevalence of self-disclosed FCOI and direct FCOI and their effect on the content of the editorial from the perspective of the reader.
Cumulative evidence suggests that journals should take important steps to address the issue of FCOI, especially direct FCOI, among the authors of editorials. In the past, NEJM took a commendable step of preventing any authors with FCOIs from writing editorials37; however, the journal has since then backtracked on the stance, citing a lack of experts without FCOIs.38 Other journals, such as The BMJ, have restricted publication of editorials and educational materials to authors without FCOIs.39 The incorrect perception that the presence of FCOI is equivalent to the presence of expertise needs to change.38,40 However, the type, extent, and relevance of FCOIs is also important to consider. For example, active principal investigators of clinical trials may have some FCOI as a result of being on the steering committee for trials, which should be viewed differently from FCOI due to receiving honoraria for serving on advisory boards, stock ownership, and other general payments (eg, food, travel, entertainment). A way to address this would be to consider a minimum threshold of FCOI over which the conflict would be considered substantial enough not to allow the researcher to write an editorial in the journal.
The purpose of this study was not to point fingers at particular studies but to reveal the state of affairs in publication practices for editorials accompanying top oncology trials. So, our concern was not whether a certain editorial was biased, but whether a journal should knowingly ask for an editorial regarding a cancer drug from an expert with substantial financial ties to the same industry. Our study does not establish causality, and we do not wish to imply any malfeasance on the part of editorialists with direct FCOIs.
Our study has important methodologic limitations. We did not verify the accuracy of disclosed COIs; prior work has suggested that disclosed FCOIs may underestimate actual FCOIs.27,36,41 As a result, our estimates may therefore underestimate the true rate. We also did not quantify the FCOI amount. However, simply relying on dollar values to gauge FCOIs is challenging, too, because the monetary value may not directly relate to level of change; even small gifts can be enough to sway decisions.42,43 Furthermore, FCOI with the same company and FCOI with the same drug may have different impact, although for non–US-based authors, it is impossible to track these in the absence of a database. Finally, there is inherently some element of subjectivity in our classification of whether an editorial presented an unduly favorable impression of study results. We attempted to mitigate this by first having 2 independent investigators assess all editorials that the first investigator classified as potentially favorable, and not informing the other 2 investigators about the purpose of the exercise until the exercise was over. Providing formal training to the reviewers on determining spin in the editorials would have made the study more robust. However, we wanted to assess the impression of the editorial on a typical consumer of the editorial (in our case, oncologists) who does not have any training in assessing bias or “spin.”
Conclusions
Our findings suggest that the presence of direct FCOI is frequent among the editorialists of top journals that publish practice-changing cancer drug trials. The presence of direct FCOI was associated with an unduly favorable editorial. Journals should take proactive steps to exclude experts with substantial direct FCOIs from authoring editorials.
References
- 1.↑
Institute of Medicine; Board on Health Sciences Policy; Committee on Conflict of Interest in Medical Research, Education, and Practice; Lo B, Field MJ, eds. Conflict of Interest in Medical Research, Education, and Practice. Washington, DC: National Academies Press; 2009.
- 2.↑
Dowsett SA, Van Campen LE, Bednar LA. Developing good scientific publishing practices: one pharmaceutical company’s perspective. Curr Med Res Opin 2010;26:1249–1254.
- 3.↑
Edwards IR. Conflicts of interest in medicines safety and regulation: how much conflict and how much interest should we allow? Drug Saf 2011;34:617–621.
- 4.↑
Miller DR. Disclosure of conflicts of interest in biomedical publications. Can J Anaesth 2008;55:265–269.
- 5.↑
Prasad V, Rajkumar SV. Conflict of interest in academic oncology: moving beyond the blame game and forging a path forward. Blood Cancer J 2016;6:e489.
- 6.↑
Wayant C, Turner E, Meyer C, et al. Financial conflicts of interest among oncologist authors of reports of clinical drug trials. JAMA Oncol 2018;4:1426–1428.
- 7.↑
Kesselheim AS, Lee JL, Avorn J, et al. Conflict of interest in oncology publications: a survey of disclosure policies and statements. Cancer 2012;118:188–195.
- 8.↑
Tao DL, Boothby A, McLouth J, et al. Financial conflicts of interest among hematologist-oncologists on Twitter. JAMA Intern Med 2017;177:425–427.
- 9.↑
Kaestner V, Brown A, Tao D, et al. Conflicts of interest in Twitter. Lancet Haematol 2017;4:e408–409.
- 10.↑
Khan R, Scaffidi MA, Rumman A, et al. Prevalence of financial conflicts of interest among authors of clinical guidelines related to high-revenue medications. JAMA Intern Med 2018;178:1712–1715.
- 11.↑
Neuman J, Korenstein D, Ross JS, et al. Prevalence of financial conflicts of interest among panel members producing clinical practice guidelines in Canada and United States: cross sectional study. BMJ 2011;343:d5621.
- 12.↑
Saleh RR, Majeed H, Tibau A, et al. Undisclosed financial conflicts of interest among authors of American Society of Clinical Oncology clinical practice guidelines. Cancer 2019;125:4069–4075.
- 13.↑
Weir IR, Marshall GD, Schneider JI, et al. Interpretation of time-to-event outcomes in randomized trials: an online randomized experiment. Ann Oncol 2019;30:96–102.
- 14.↑
Lerner TG, Miranda MC, Lera AT, et al. The prevalence and influence of self-reported conflicts of interest by editorial authors of phase III cancer trials. Contemp Clin Trials 2012;33:1019–1022.
- 15.↑
Kaestner V, Prasad V. Financial conflicts of interest among editorialists in high-impact journals. Blood Cancer J 2017;7:e611.
- 16.↑
Gyawali B, Shimokata T, Honda K, et al. Reporting harms more transparently in trials of cancer drugs. BMJ 2018;363:k4383.
- 17.↑
Haslam A, Herrera-Perez D, Gill J, et al. Patient experience captured by quality-of-life measurement in oncology clinical trials. JAMA Netw Open 2020;3:e200363.
- 18.↑
Kim MS, Prasad V. Assessment of accuracy of waterfall plot representations of response rates in cancer treatment published in medical journals. JAMA Netw Open 2019;2:e193981.
- 19.↑
Kaestner V, Edmiston JB, Prasad V. The relation between publication rate and financial conflict of interest among physician authors of high-impact oncology publications: an observational study. CMAJ Open 2018;6: E57–62.
- 20.↑
Wang AT, McCoy CP, Murad MH, et al. Association between industry affiliation and position on cardiovascular risk with rosiglitazone: cross sectional systematic review. BMJ 2010;340:c1344.
- 21.↑
Fugh-Berman A, McDonald CP, Bell AM, et al. Promotional tone in reviews of menopausal hormone therapy after the Women’s Health Initiative: an analysis of published articles. PLoS Med 2011;8:e1000425.
- 22.↑
Stelfox HT, Chua G, O’Rourke K, et al. Conflict of interest in the debate over calcium-channel antagonists. N Engl J Med 1998;338:101–106.
- 23.↑
Abud-Mendoza C. Medical ethics, research and the pharmaceutical industry [in Spanish]. Reumatol Clin 2012;8:233–235.
- 24.↑
Moynihan R, Albarqouni L, Nangla C, et al. Financial ties between leaders of influential US professional medical associations and industry: cross sectional study. BMJ 2020;369:m1505.
- 25.↑
Mitchell AP, Basch EM, Dusetzina SB. Financial relationships with industry among National Comprehensive Cancer Network guideline authors. JAMA Oncol 2016;2:1628–1631.
- 27.↑
Wong SL, Stain SC. Resolve the conflict rather than dissolve the relationship. Ann Surg Oncol 2019;26:2325–2326.
- 29.↑
Spelsberg A, Martiny A, Schoenhoefer PS. Is disclosure of potential conflicts of interest in medicine and public health sufficient to increase transparency and decrease corruption? J Epidemiol Community Health 2009;63:603–605.
- 30.↑
Klein DF, Glick ID. Conflict of interest, journal review, and publication policy. Neuropsychopharmacology 2008;33:3023–3026.
- 31.↑
Torjesen I. Oncology trial authors don’t fully disclose financial conflicts of interests, analysis finds. BMJ 2018;362:k3749.
- 32.↑
Bauchner H, Fontanarosa PB, Flanagin A. Conflicts of interests, authors, and journals: new challenges for a persistent problem [editorial]. JAMA 2018;320:2315–2318.
- 33.↑
Orenstein C, Thomas K. Top cancer researcher fails to disclose corporate financial ties in major research journals. New York Times. September 8, 2018. Accessed April 22, 2020. Available at: https://www.nytimes.com/2018/09/08/health/jose-baselga-cancer-memorial-sloan-kettering.html
- 34.↑
Ozaki A. Conflict of interest and the CREATE-X trial in the New England Journal of Medicine. Sci Eng Ethics 2018;24:1809–1811.
- 36.↑
DeAngelis CDF, Fontanarosa PB. Impugning the integrity of medical science: the adverse effects of industry influence. JAMA 2008; 299:1833–1835.
- 39.↑
The BMJ. Competing interest policy. Accessed April 22, 2020. Available at: https://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists/declaration-competing-interests
- 41.↑
Norris SL, Holmer HK, Ogden LA, et al. Characteristics of physicians receiving large payments from pharmaceutical companies and the accuracy of their disclosures in publications: an observational study. BMC Med Ethics 2012;13:24.
- 42.↑
Wiersma M, Kerridge I, Lipworth W. Dangers of neglecting non-financial conflicts of interest in health and medicine. J Med Ethics 2018;44:319–322.
- 43.↑
Booth CM, Detsky AS. From the $80 hamburger to managing conflicts of interest with the pharmaceutical industry. BMJ 2019;365:1939.
