Association Between Conflicts of Interest and Authors' Positions on Harms of Varenicline: a Cross-Sectional Analysis.

Background: Few studies have investigated the relationship between industry funding/conflicts of interest and authors' positions in opinion pieces on drug safety. Harmful effects of varenicline, a treatment for smoking cessation, have been highly contested.

Objective: To examine the association between pharmaceutical industry funding/authors' financial conflicts of interest and position on varenicline in opinion articles, especially in relation to the minimization of harms; to assess whether opinion pieces on drug safety issues written by authors with conflicts of interest are more frequently cited in the news or social media.

Design: Cross-sectional analysis.

Participants: English language opinion pieces and narrative reviews about varenicline published between May 2006 and February 2019.

Main Measures: Odds ratios and 95% confidence intervals; the Mann-Whitney two-sample statistic was used to test for differences in Altmetric scores, a measure of media attention.

Key Results: Of the 221 included articles, 30.3% (67) disclosed the funding source and 62.9% (139) disclosed authors' conflicts of interest. Authors of opinion pieces on varenicline who reported financial ties to the pharmaceutical industry (as a conflict of interest or funding source) were more likely to minimise the cardiovascular and psychiatric risk of varenicline compared to those without conflicts of interest or industry funding (OR: 4.00; 95% CI: 1.32 to 12.16 for cardiovascular risk; OR: 8.51; 95% CI: 3.79 to 19.11 for psychiatric risk). These associations persisted in sensitivity analyses. No statistically significant difference in Altmetric score was found between articles with (mean 15.83, median 3) and without (mean 11.90, median 1) conflicts of interest, indicating similar media attention (p-value=0.11).

Conclusions: We found that authors with financial ties to drug companies were more likely to publish opinion pieces that minimised harms of varenicline. These results raise questions about journals' editorial policies to accept reviews of treatments from authors with financial relationships with manufacturers.