Analysis of user-generated posts on social media of adjuvant analgesics: A machine learning study

Abstract

Background: Antiepileptics and antidepressants are frequently prescribed for chronic pain, but their efficacy and potential adverse effects raise concerns, including dependency issues. Increased prescriptions, sometimes fraudulent, prompted reclassification of antiepileptics in some countries. Our aim is to comprehend opinions, perceptions, beliefs, and attitudes towards co-analgesics from online discussions on X (formerly known as Twitter), offering insights closer to reality than conventional surveys.

Methods: In this cross-sectional study, we collected 77,183 public posts about co-analgesics in English or Spanish from January 1st 2019 to December 31st, 2020. A total of 51,167 post were included, and 2,000 were manually analyzed using a researcher-created codebook. Machine learning classifiers were then applied to the remaining datasets to determine the number of publications for each user type and identify categories through content analysis.

Results: Of the 51,167 posts analyzed, 78% discussed anticonvulsants and 24% discussed analgesic antidepressants (Percentages add up to more than 100% because there were 1,300 posts containing references to both types of medications). Only 13% were authored by healthcare professionals, while 67% were from patients. Medical content predominated, with 70% noting low medication efficacy and almost 50% referencing side effects. Non-medical content included challenges in dispensing (25%), complaints about high costs (15%), and trivialization of medication use (10%).

Conclusions: This study offers valuable insights into public perceptions of co-analgesics. Findings aid in designing public health communications to raise awareness of associated risks, urging both healthcare providers and the public to optimize drug use.