Self-reported conflicts of interest of authors, trial sponsorship, and the interpretation of editorials and related phase III trials in oncology

Abstract

Purpose: Growing participation by industry in cancer research has resulted in increased reporting of conflicts of interest (COI). We aimed to test any association between authors' conclusions and self-reported COI or trial sponsorship in cancer studies.

Methods: Editorials and related phase III trials published in six clinical oncology journals in the last 3.5 years were analyzed independently by two investigators who classified study conclusions according to authors' endorsement of the experimental therapy. Logistic regression multivariable models were used to assess predictors of favorable conclusions of editorialists and of phase III authors.

Results: From January 2008 to October 2011, 1,485 articles were retrieved: 150 phase III trials and 150 editorials were eligible. Among the phase III trials, 82 (54.7%) had positive results, and 78 (52.0%) were entirely or partially funded by industry. Any COI were disclosed in 103 phase III trials (68.7%) and in 71 editorials (47.3%). Multivariable analysis showed that phase III trial results were the only significant predictor for a positive conclusion by trial authors (odds ratio [OR], 92.2; 95% CI, 19.7 to 431.6; P < .001). Sponsorship did not predict for positive conclusion by phase III authors (OR, 0.86; 95% CI, 0.3 to 2.5; P = .788). The only factor associated with positive conclusions by editorial authors was a positive conclusion by phase III trial authors (OR, 36.3; 95% CI, 6.8 to 194.2; P < .001).

Conclusion: The interpretation of recently published phase III cancer trials by their authors or by editorialists was not influenced by financial relationships or industry sponsorship. Increased awareness of COI policies may have led to more integrity in cancer research reporting.