Cannabinoid Use in Epilepsy: A Qualitative and Quantitative Study of Public Perception via Social Media
Justine Ker1, Lauren N. Cooper2, Alexander P. Radunsky2, Christoph U. Lehmann2, Marisara Dieppa1
1University of Texas Southwestern Medical Center, 2Clinical Informatics Center, University of Texas Southwestern Medical Center
Objective:
Leveraging social media data, we aim to understand public perceptions on cannabinoid use in epilepsy patients.
Background:
Evidence supporting the benefit of cannabinoids in epilepsy patients remains limited except for one FDA approved prescription, cannabidiol.
Design/Methods:
We collected English-language posts containing selected epilepsy- and cannabinoid-associated terms from X and analyzed posts using Python and natural language processing library spaCy. We used the M3-Inference and Ethnicolr libraries to infer age, gender, and ethnicity and assigned posts to an ‘epilepsy’ versus ‘non-epilepsy’ cohort based on hashtags. We used the SentiStrength library to assign values based on posts’ sentiment (positive or negative). The Text2Emotion library used natural language processing to categorize posts into happy, angry, sad, surprise, or fear emotions. Topic modeling was completed using Latent Dirichlet Allocation from the Gensim Library and ChatGPT.
Results:
We analyzed 2,756 posts. 1,391 unique accounts were individuals and 645 were organizations. 41% of users were 18 or under and 64.6% were male. Mean sentiment score for all posts was –0.57 on a –4 to +4 scale. Posts were mainly negative (45.6%) or neutral (34.3%). Most common emotions included 'fear' (42%), 'sad' (20.5%), and 'surprise' (19%).
Epilepsy posts were statistically more likely to express neutral to positive sentiment compared to non-epilepsy posts which were more likely to express negative sentiment (p < 0.001). Non-epilepsy posts were statistically more likely to express the ‘sad’ emotion, without a Bonferroni correction, but not with the correction of αadj= 0.01 (0.05/5). Most common themes for epilepsy posts related to cannabinoid use focused on personal experiences and community support, emotional support for families and individuals, and challenges and legal implications faced by those with epilepsy using cannabis.
Conclusions:
Tracking trends in public perception regarding cannabinoid use in epilepsy patients may allow providers to counteract disinformation or misinformation through early counseling and timely public health interventions.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.