P665 Social Media Listening to better Understand the Lived Experience of Individuals With Ulcerative Colitis

Auriat, O.(1);Lucats, L.(1)*;Badalamenti, S.(2);Kruger, A.(2);Wiekowski, M.(2);Glotfelty, L.(2);Van Hoogstraten, H.(3);Coulouvrat, C.(4);

(1)Sanofi Recherche & Développement - 1 avenue Pierre Brossolette, Sanofi Recherche & Développement - 1 avenue Pierre Brossolette, Chilly-Mazarin, France;(2)Sanofi Research and Development- 55 Corporate Drive- Bridgewater New Jersey 08807- United States of America, Sanofi Research and Development- 55 Corporate Drive- Bridgewater New Jersey 08807- United States of America, Bridgewater, United States;(3)Sanofi Global Medical Affairs- 55 Corporate Drive- Bridgewater New Jersey 08807- United States of America, Sanofi Global Medical Affairs- 55 Corporate Drive- Bridgewater New Jersey 08807- United States of America, Bridgewater, United States;(4)Sanofi Research and Development- 1 avenue Pierre Brossolette- CHILLY-MAZARIN 91385- France, Sanofi Research and Development- 1 avenue Pierre Brossolette- CHILLY-MAZARIN 91385- France, Chilly-Mazarin, France;

Background

Lived experience is a depiction of a person's experiences and decisions, as well as the knowledge gained from these experiences and choices. Despite a wealth of clinical studies and real-world data on ulcerative colitis, large-scale and comprehensive insights into the lived experience are lacking. This study aimed to explore how the lived experience of UC is described on social media.

Methods

All English-language posts from the UC open-access forums of Reddit.com, Inspire.com and Everyday Health.com were extracted and anonymized with Synthesio. Concept occurrence and sentiment analysis were performed using Luminoso, a natural language processing (NLP) platform. NLP is the ability of a computer program to interpret human language as it is spoken and written.

Results

A total of 29,250 comments posted between July 01, 2018 and January 01, 2021 were extracted and analyzed. The first 40 concepts were manually reviewed to remove the synonyms of UC and high level health-related concepts (e.g., illness, doctor, treat…) and several concepts were manually merged (eg, stomach and gut). The most common topic expressed in the posts was feeling (n=4,749, 16%), two thirds reporting negative sentiment (68%). Symptom analysis revealed the highest occurrence in pain (n = 4,085, 14%), rectal bleeding (n = 3,385, 12%), toilet frequency (n = 2,003, 7%) and fatigue (n = 626, 2%). These symptoms were associated with negative sentiments (58% to 89%). Among pain, a quarter of the posts were related to stomach or intestinal pain (n = 1,097, 27%). Further sub-concept analysis related to pain revealed topics on pain related to disease flare (n = 711, 17%), joint pain (n = 394, 10%) and bone pain (n = 124, 3%). Notably, patients did not post frequently about their medication (<1%), but surgery was mentioned in 1,494 posts (5%). Sentiment analysis again outlined the dominance of negative feelings whether the comment was related to fear of this definitive procedure or to life post-surgery. Lastly, anxiety (n = 1,019, 3%) and concerns about sleep (n = 857, 3%)  were less frequently identified, but still meaningful, concepts.

Conclusion

NLP analysis of posts on UC in unmonitored social media forums demonstrates that patients are firstly concerned by pain and its impact that is strikingly not part of the Mayo score and not routinely assessed in large clinical trials. While social media-based retrospective data collection inherently limits the amount of demographic or clinical description, social media listening studies offer the potential to sample spontaneous patient concerns from large numbers of individuals. Analysis of these data may lead to the development of patient-centered endpoints to be incorporated into clinical studies in the future.