Education Research: Making a Tweetorial Fly: Features of Educational Social Media Posts Associated With High Sharing and Engagement

Background and objectives: Social media platforms such as X (formerly Twitter) are increasingly used in medical education. Characteristics of tweetorials (threaded teaching posts) associated with higher degrees of engagement are unknown. We sought to understand features of neurology-themed tweetorials associated with high sharing and engagement.

Methods: We created a neurology-themed tweetorials data set by searching "tweetorial" AND "neurology" on X that were posted between November 2018 and December 2022. Tweetorial and author characteristics were collected from X and by viewing the author's academic profile. We created and validated a novel formula to determine the tweetorial's "X Factor" (XF), a measure of reader engagement and distribution, reflecting reposts and likes. Each tweetorial was analyzed for basic variables, the author's academic rank, and thematic content. Each first post underwent a language analysis using Linguistic Inquiry and Word Count (LIWC-22) and was hand-coded for style (such as "statement" or "clinical case"). We determined each covariate's impact on XF. The general estimating equation was applied to correct for the author effect.

Results: We identified 392 neurology-themed tweetorials posted by 96 unique authors. XF strongly correlated with impressions (R ² = 0.85) and was validated in a separate data set (R ² = 0.74). The median XF of the tweetorials was 28.5K (interquartile range 12.7K-61.5K). Tweetorials about a "General Neurology Topic" and with a "Clearly Stated Topic" had 48% and 49% higher XF than those without (p = 0.001 and 0.006, 95% CI 17%-88%, 12%-97%, respectively). Having a "creative" first post, including a unique hashtag, and featuring an author-made graphic correlated with 60%, 49%, and 84% higher XF than posts without those elements (p = 0.01,95% CI 13%-125%, p < 0.001, 95% CI 16%-92%, p < 0.001, 95% CI 30%-164%, respectively). Continuing medical education (CME) accreditation and higher scores on "positive tone" negatively affected XF (-80%, p < 0.001, 95% CI 70%-86% and -7%/point of positivity, p < 0.001, 95% CI 2%-10%, respectively).

Discussion: Tweetorial engagement and distribution are determined by multiple factors including authorship, clarity of the topic, and visual appeal of the post. CME accreditation was strongly negatively associated with sharing and may reflect a sharing preference for personal accounts over institutional ones, although further study is needed.