OMISSION OF CONFLICTS OF INTEREST DISCLOSURES IN SOCIAL MEDIA DURING SCIENTIFIC MEETINGS

Introduction: Everyday use of social media among health professionals is likely increasing for both personal and professional intents. However, such practice can be accompanied by perilous exposure, as conflicts of interest might bias public perception obtained via fast-paced electronic social networks. Here we address conflict of interest disclosures in social media posts during annual meetings from the American (ASCO) and European (ESMO) oncology societies, which were presumably performed by health professionals. Methods: We applied a Web API request interception approach to retrieve metadata from public posts shared on Instagram and Facebook during ASCOs and ESMOs annual meetings in 2017, 2018, and 2019 (ASCO only). A total of 10 hashtags were used to search for publications in the aforementioned social media: #ASCO17, #ASCO18, #ASCO19, #ASCO2017, #ASCO2018,#ASCO2019,#ESMO17,#ESMO18, #ESMO2017, and #ESMO2018. To completely avoid selection bias associated with the surveyors social media accounts, public data scraping was performed using a JavaScript package without logging into social media. The content of the shared posts was examined using natural language processing (NLP) algorithms in Python 3.7. Results: Overall, 4,522 shared posts were retrieved in their JSON format from Instagram, encompassing 9,959 comments and 340,717 likes. After removing 167 duplicate posts due to the simultaneous use of two or more hashtags, we obtained 1,171 (26.9%), 1,348 (30.9%), and 1,836 (42.2%) unique posts from 2017, 2018, and 2019, respectively. Video posts accounted for 7.9% of all publications over the three-year period. 3,075 posts were published between 7AM and 7PM (70.6%) and 1,280 (29.4%) between 7PM and 7AM. After applying lexical analysis using word tokenization, we recognized a total of 750 Brazilian posts (17.2%). Despite the high volume publications, we did not identify conflict of interest disclosures in any of Brazilian (0%), Spanish written (0%), or English written (0%) posts captions. Facebook data mining will be presented during the meeting. Conclusion: Social media posting has progressively increased during oncology meetings in the US and Europe over the last three years. Our data point towards a remarkable omission of conflict of interest in these venues. Professional guidelines to assist healthcare professionals in ethical communication in electronic social networks are warranted.