부산대학교 도서관

To evaluate the completeness of conflict-of-interest self-reporting by ophthalmology researchers and to assess factors associated with self-reporting. Cross-sectional observational study. We evaluated articles published between January and June 2017 in Ophthalmology , JAMA Ophthalmology , the American Journal of Ophthalmology , and Investigative Ophthalmology and Visual Science. To assess more accurately the cases in which an author published multiple articles, we defined a unit of analysis, authorship, for which each author of each article is a unique data point. To enable comparison with the Open Payments Database (OPD), we only included United States physician authorships. For each authorship, we defined self-reported relationships as the companies listed in the article's conflict-of-interest disclosures. Based on journal policies, we defined OPD-reported relationships as the list of companies that reported payments to the author within 36 months before submission. For each authorship, we assessed the proportion of OPD-reported relationships that were self-reported. The primary measurement was the proportion of authorships reporting none of their OPD-reported relationships. Of the 660 total authorships (486 unique authors), 413 authorships (63%) reported none of their OPD-reported relationships, 112 (17%) reported some of them, 9 (1%) reported all of them, and 126 (19%) had 0 relationships. The proportion of authorships reporting none of their relationships did not differ significantly between journals that required reporting of all relationships compared with journals that required reporting only of relevant relationships (adjusted percentage, 61.4% vs. 64.3%; P = 0.46). Authorships with more dollars received during the reporting period showed higher rates of self-reporting (P < 0.001). Even among journals that required complete reporting, self-reporting was low compared with an industry-maintained database of financial relationships. Deficiencies in reporting may undermine confidence in self-reporting and may compromise the transparency that is needed to interpret research results fairly. Proprietary or commercial disclosure may be found after the references. [ABSTRACT FROM AUTHOR]