Financial Disclosure, Industry Sponsorship, and Integrity in Cancer Research Reporting

In the United States and elsewhere in the world, the pharmaceutical industry provides a substantial and an ever increasing proportion of clinical trials funding. This trend has prompted considerable anxiety in some quarters because several studies have shown that there is a clear association between pharmaceutical industry funding and proindustry results, calling into question the integrity of data reporting. It is also well known that financial relationships between physicians and the pharmaceutical industry (eg, payments for consulting, for giving lectures, or enrolling patients onto trials, and so on) are common and that these interactions potentially influence a physician’s attitudes and practices in ways that may be imprudent or even unethical. This issue was highlighted in an Institute of Medicine report that concluded that physician-industry financial relationships “present the risk of undue influence on professional judgments and thereby may jeopardize the integrity of scientific investigations, the objectivity of medical education, the quality of patient care, and the public’s trust in medicine.” Financial disclosure, enhanced reporting standards, and trial registries are common strategies to mitigate sponsorship bias and related investigator potential conflicts of interest (COI). However, to what extent disclosure policies and other mitigation strategies have actually lessened author bias and industry influence is the topic of frequent debate and ongoing scientific investigation. In the article that accompanies this editorial, Bariani et al report on a study designed to “identify whether there was any association between conclusions of authors of editorials and related phase III trials and self-reported COI or sponsorship” in recently published cancer studies. The investigators reviewed nearly 1,500 articles published between January 2008 and October 2011 in five widely read oncology journals and New England Journal of Medicine. COI was defined as any self-reported financial relationship between an author and a pharmaceutical company (eg, consultancy, honoraria, employment, stock ownership, research grants, and so on). One hundred fifty trials were ultimately included in the analysis. Two investigators, blinded to author COI disclosures and the trial sponsorship, independently analyzed the outcome of each of the phase III trials and its related editorial with respect to the authors’ and editorialist’s conclusions. Results were then categorized on a positivity scale as previously described by Djulbegovi et al—ie, how favorable authors were toward the experimental arm. Agreement between investigators with respect to grade assignment was excellent; however, if a disagreement occurred, a third investigator also reviewed the publication. Discrepancies were adjudicated by consensus of all three reviewers. The final results were then correlated with the presence or absence of a COI disclosure statement and with trial sponsorship. Similar to their earlier report, the survey revealed at least one COI was disclosed in 103 (68.7%) of the phase III trials, most commonly in the form of a consultancy, honoraria, or research funds, and in 71 (47.3%) of the editorials. The trial outcomes were considered “highly positive” or “positive” by the authors in approximately 70% of the studies whereas the editorialists judged the trials as “highly positive” or “positive” in half. In a multivariable analysis, the only predictor of a “favorable” author conclusion was a positive result of the phase III trial. In other words, the trial results and the authors’ conclusions were congruent and not seemingly influenced by the presence or absence of a disclosed financial COI, a gratifying result to be sure. Interestingly, the only predictor of a positive conclusion by an editorialist was a favorable conclusion by the authors of the phase III trial. Neither the outcome of the phase III trial nor the status of a financial COI was statistically associated with the editorialist’s conclusion. Notably, Bariani et al also observed a few reports wherein negative results were discussed in a positive manner by the authors and/or the editorialist. In general, these discordant conclusions were based on the authors’ or editorialists’ assessment of secondary end points, improved toxicity profiles compared with standard care, benefit in subgroups, and the like. Nevertheless, Bariani et al conclude that the interpretation of the data by the authors of the trials assessed, and that of the corresponding editorialists, “was not influenced by financial relationships or industry sponsorship ” and that “increased awareness of COI policies may have led to more integrity in cancer research reporting.” Do the aforementioned findings merit the authors’ hopeful conclusions? Has increased awareness of COI policies reduced bias and improved the validity of trial reporting as suggested by these investigators? No doubt awareness of COI policies has increased in recent years, at least in part because most creditable scientific journals now require authors to provide financial COI data. The existence of thirdparty trial registries (eg, ClinicalTrials.gov), drug company physician payment registries, and public watchdog Web sites (eg, ProPublica.org) also may have prompted some investigators to be more forthcoming vis-a-vis financial relationships with industry. However, compliance with existing COI policies remains suboptimal and failure to disclose financial relationships persists as one of the top ten misbehaviors of scientists. Even more problematic, disclosure actually has the potential to worsen author bias through processes JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 31 NUMBER 18 JUNE 2