Format

Send to

Choose Destination
Pharmacoepidemiol Drug Saf. 2016 Dec;25(12):1425-1433. doi: 10.1002/pds.4090. Epub 2016 Sep 7.

Can social media data lead to earlier detection of drug-related adverse events?

Author information

1
Analysis Group, Inc., Boston, MA, USA.
2
Groupe d'analyse, Ltée, Montréal, QC, Canada.

Abstract

PURPOSE:

To compare the patient characteristics and the inter-temporal reporting patterns of adverse events (AEs) for atorvastatin (Lipitor® ) and sibutramine (Meridia® ) in social media (AskaPatient.com) versus the FDA Adverse Event Reporting System (FAERS).

METHODS:

We identified clinically important AEs associated with atorvastatin (muscle pain) and sibutramine (cardiovascular AEs), compared their patterns in social media postings versus FAERS and used Granger causality tests to assess whether social media postings were useful in forecasting FAERS reports.

RESULTS:

We analyzed 998 and 270 social media postings between 2001 and 2014, 69 003 and 7383 FAERS reports between 1997 and 2014 for atorvastatin and sibutramine, respectively. Social media reporters were younger (atorvastatin: 53.9 vs. 64.0 years, p < 0.001; sibutramine: 36.8 vs. 43.8 years, p < 0.001). Social media reviews contained fewer serious AEs (atorvastatin, pain: 2.5% vs. 38.2%; sibutramine, cardiovascular issues: 7.9% vs. 63.0%; p < 0.001 for both) and concentrated on fewer types of AEs (proportion comprising the top 20 AEs: atorvastatin, 88.7% vs. 55.4%; sibutramine, 86.3% vs. 65.4%) compared with FAERS. While social media sibutramine reviews mentioning cardiac issues helped predict those in FAERS 11 months later (p < 0.001), social media atorvastatin reviews did not help predict FAERS reports.

CONCLUSIONS:

Social media AE reporters were younger and focused on less-serious and fewer types of AEs than FAERS reporters. The potential for social media to provide earlier indications of AEs compared with FAERS is uncertain. Our findings highlight some of the promises and limitations of online social media versus conventional pharmacovigilance sources and the need for careful interpretation of the results. © 2016 The Authors. Pharmacoepidemiology and Drug Safety published by John Wiley & Sons Ltd.

KEYWORDS:

Granger causality; Internet; adverse event; pharmacoepidemiology; pharmacovigilance; social media

PMID:
27601271
PMCID:
PMC5157765
DOI:
10.1002/pds.4090
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Wiley Icon for PubMed Central
Loading ...
Support Center