Understanding the "Social" in "Social Media" An Analysis of Twitter Engagement of Pulmonary and Critical Care Fellowship Programs

Background: Social media is ubiquitous as a tool for collaboration, networking, and dissemination. However, little is known about use of social media platforms by pulmonary and critical care medicine fellowship programs. Objective: We identify and characterize pulmonary and critical care fellowship programs using Twitter and Instagram, as well as the posting behaviors of their social media accounts. Methods: We identified all adult and pediatric pulmonary, critical care medicine (CCM), and combined pulmonary and critical care medicine (PCCM) programs in the United States using the Electronic Residency Application Service. We searched for Twitter profiles for each program between January 1, 2018, and September 30, 2018. Tweets and Twitter interactions were classified into the following three types: social, clinical, or medical education (MedEd) related. We collected data about content enhancements of tweets, including the use of pictures, graphics interchange format or videos, hashtags, links, and tagging other accounts. The types of tweets, content enhancement characteristics, and measures of engagement were analyzed for association with number of followers. Results: We assessed 341 programs, including 163 PCCM, 36 adult CCM, 20 adult pulmonary, 67 pediatric CCM, and 55 pediatric pulmonary programs. Thirty-three (10%) programs had Twitter accounts. Of 1,903 tweets by 33 of the 341 programs with Twitter accounts, 476 (25%) were MedEd related, 733 (39%) were clinical, and 694 (36%) were social. The median rate of tweets per month was 1.65 (interquartile range [IQR], 0.4-6.65), with 55% programs tweeting more than monthly. Accounts tweeting more often had significantly more followers than those tweeting less frequently (median, 240 followers; 25-75% IQR, 164-388 vs. median, 107 followers; 25-75% IQR, 13-188; P = 0.006). Higher engagement with clinical and social Twitter interactions (tweets, retweets, likes, and comments) was associated with more followers but not for the MedEd-related Twitter interactions. All types of content enhancements (pictures, graphics interchange format/videos, links, and tagging) were associated with a higher number of followers, except for hashtags. Conclusion: Despite the steadily increasing use of social media in medicine, only 10% of the pulmonary and critical care fellowship programs in the United States have Twitter accounts. Social and clinical content appears to gain traction online; however, additional evaluation is needed on how to effectively engage audiences with MedEd content.