OBJECTIVES

Social media are an important way for patients to share experiences, opinions and concerns regarding health issues and treatments. Paracetamol is a well-established OTC medication and has a good safety profile when used as recommended. However, the number of reports of paracetamol misuses has increased these last decades. The goal was to detect within social media discussions paracetamol misuses leading to overdose, dependence or unapproved indications.

METHODS

Posts containing key words related to paracetamol products were retrieved from 18 French forums between 2003 and 2019. Text mining algorithms were developed for the detection of drug misuses. Unapproved indications were detected in posts with drug intake expression and medical concepts identified through Medical Dictionary for Regulatory Activities. The medical concepts were compared with the approved indications found within the products’ Summary of Product Characteristics. Dependence was determined by calculating a similarity score through the BM25 algorithm between a message and a reference pattern composed from a lexical field expressing drug dependence. The overdose algorithm detected expressions composed by digits, dosage and duration terms (e.g. 10 grams per day). The paracetamol overdose was calculated as a percent above 4 grams, the recommended daily dose.

RESULTS

In the 44 283 posts extracted, the method identified fatigue as the most common unapproved indication, with 190 posts, followed by dependence with 148 posts. Reference to overdose was made by 245 users across 291 posts. Posts with reference to dependence were made by 136 users, accounting for 171 posts.

CONCLUSIONS

The use of social media provides valuable data on patient’s behavior and allows the monitoring of drugs use. These findings highlight the necessity to raise awareness of the toxicity risks of paracetamol and that accessible healthcare information for the general public would be beneficial.