Digital Drug Safety Surveillance: Monitoring Pharmaceutical Products in Twitter

• DOI: 10.1007/s40264-014-0155-x
• Publication Date: 2014-04-28T05:01:27Z
• Authors: Clark C. Freifeld; John S. Brownstein; Christopher M. Me...; Wenjie Bao; Ross Filice; Taha Kass-Hout; Nabarun Dasgupta
• Keywords: MedDRA; Concordance; Postmarketing surveillance

Traditional adverse event (AE) reporting systems have been slow in adapting to online AE from patients, relying instead on gatekeepers, such as clinicians and drug safety groups, verify each potential event. In the meantime, increasing numbers of patients turned social media share their experiences with drugs, medical devices, vaccines.The aim study was evaluate level concordance between Twitter posts mentioning AE-like reactions spontaneous reports received by a regulatory agency.We collected public English-language 23 products 1 November 2012 through 31 May 2013. Data were filtered using semi-automated process identify resemblance AEs (Proto-AEs). A dictionary developed translate Internet vernacular standardized ontology for analysis (MedDRA(®)). Aggregated frequency identified product-event pairs then compared data FDA Adverse Event Reporting System (FAERS) Organ Class (SOC).Of 6.9 million collected, 4,401 Proto-AEs out 60,000 examined. Automated, dictionary-based symptom classification had 86 % recall 72 precision [corrected]. Similar overall distribution profiles observed, Spearman rank correlation rho 0.75 (p < 0.0001) reported FAERS SOC.Patients showed range sophistication when describing experience. Despite availability these data, appropriate role pharmacovigilance has not established. Additional work is needed improve acquisition automation.