Reactions 1635, p10 - 21 Jan 2017
1
Social media: pharmacovigilance early warning system? Social media monitoring "may provide earlier insights into certain adverse events," according to study results reported in Drug Safety, although more work is needed to elaborate additional uses. The retrospective study used data from posts in English to Facebook or Twitter between March 2009 and October 2014 regarding 10 recent US FDA postmarketing safety signals for product-adverse event pairs,* starting 2 years before the signal was first identified. Time of reporting was compared with the first report to the FDA Adverse Event Reporting System (FAERS) and the time the safety signal was identified. Posts regarding the selected products which were identified by an automated classifier as containing information with a resemblance to an adverse event were reviewed by a human classifier and causality was evaluated. Negative controls comprised six different product-event pairs. Of the 935 246 automatically harvested posts, there were 98 252 posts with resemblance to an adverse event. Causality assessment was performed for 13 product-event pairs; no product-event associations were found among the negative controls. Clinical assessment revealed sufficient information to warrant further investigation for two product-event pairs; specific symptomology did not appear to be discussed for the other combinations evaluated. The first report to FAERS regarding dronedarone and vasculitis was received on 7 July 2010; signal detection was reported on 15 April 2012 resulting from 10 unique Individual Case Safety Reports (ICSR). The first social media case was identified on 5 December 2010. Causality for the four social media reports was evaluated as possible (n=3) or probable (n=1). The first report to FAERS regarding Banana Boat Sunscreen Spray and flammability resulting in serious skin burns was received on 19 June 2012; signal detection followed later in June from one unique ICSR. The first social media case was identified on 2 June 2012. Causality for the two social media reports was evaluated as certain. "Social media can be an adjunct to traditional safety reporting systems to possibly uncover postmarketing safety signals more rapidly," note the authors, "although considerations of noise and volume need to be further characterized" and "further research is needed to develop best practices and methods for determining what constitutes a ’safety signal’ in social media". They add that many social media posts "do not describe symptoms using the medical terminology needed to explicitly confirm a certain condition or diagnosis," but that nevertheless, the most important contribution from social media may be "that an early serious concern is expressed about a drug harm before it is medicalized". * lisdexamfetamine and Raynaud’s phenomenon; olmesartan and spru-like enteropathy; methylphenidate and priaprism; topical econazole and increased INR; dabigatran etexilate [Pradaxa] and angioedema; dronedarone and vasculitis; hydroxyzine and serious injection site reactions; ziprasidone and DRESS; dimethyl fumarate [Tecfidera] and progressive multifocal leukoencephalopathy; Banana Boat Sunscreen Spray and flammability resulting in serious skin burns Pierce CE, et al. Evaluation of Facebook and Twitter Monitoring to Detect Safety Signals for Medical Products: An Analysis of Recent FDA Safety Alerts. Drug Safety : 2 Jan 2017. Available from: URL: http://dx.doi.org/10.1007/ 803217368 s40264-016-0491-0
0114-9954/17/1635-0001/$14.95 Adis © 2017 Springer International Publishing AG. All rights reserved
Reactions 21 Jan 2017 No. 1635