Can social listening data be used to provide meaningful insights into potential abuse of bupropion?

Trial description: QUESTION: Is it feasible to use social listening data to provide meaningful insights into potential abuse or inappropriate use of bupropion? Purpose: The purpose of this analysis is to determine the feasibility to use social media for collecting meaningful insights into potential abuse or inappropriate use of bupropion.Objectives: 1. To determine if social media can identify cases of potential abuse or inappropriate use of bupropion which can effectively complement existing sources of data currently used for pharmacovigilance activities2. To explore the utility of various internet sites and forums or populations to identify cases of interest3. To describe and characterize the posts of interest (POI) identified during this feasibility analysis (POI is a term coined to denote a user post that may be relevant to drug abuse or inappropriate use of the product in question Methods: This is a preliminary feasibility study for monitoring the potential abuse or inappropriate use of bupropion through social listening. This is a prospective descriptive observational study, analyzing three months of initial data collected on bupropion and comparator drugs (positive controls, negative controls, and opioid controls as noted below). Data, from publically available social media or internet forums posts from individuals who choose to post on a number of sites will be collected by Epidemicotm through the DataSiftTM platform. Data: Collected using Epidemico’s automated classification tools, reading publically available English language data (collected via DataSift) from the following websites: Facebook, Twitter, erowid.org, reddit.com, drugs-forum.com, hipforums.com, shroomery.org, grasscity.org, reddit.com/r/Drugs, reddit.com/r/Nootropics, IOPlist.com, partyvibe.org. All posts mentioning bupropion products (and common misspellings and slang terms) will be acquired. Abuse-related posts will be identified using the automated classifier. Posts will then be divided into categories: Abuse-related (above indicator score thresholds for drug dependence SMQ), Proto-AEs (above indicator score thresholds), Meaningful mention (zero to indicator score thresholds), Unclear/uncodable or spam (negative indicator scores).Manual Curation of data will be done by GSK and conducted by reviewers trained in social media and drug use language. This will be used to remove false positives and to request reclassification of mistakes made by automated classifier.The automated classifier is designed to flag posts where product tampering is mentioned (e.g., for purposes of injecting, snorting, etc.). Data to be collected prospectively include basic demographics, route of administration, categorization of euphoric effect, involvement of the criminal justice/prison system, how the drug was procured, and concomitantly used drugsQuantitative Analysis: Bupropion abuse data available will be compared to positive controls—those with established abuse liability greater than bupropion (methyphenidate, alprazolam), negative controls—non-controlled substances with negligible abuse liability, similar to current international classification of bupropion (venlafaxine, amitriptyline), and opioid controls (buprenorphine, oxycodone).IMS sales data will be referenced to estimate the amount of global drug exposure as denominator.Once data are received from Epidemico, GSK physicians and scientists will determine applicability for drug safety and need for intervention or continued monitoring.Governance: For our social media listening project, we will be purchasing de-identified data from a third party vendor that has been stripped of Personally Identifiable Identification. Therefore, in the absence of an identifiable reporter, we will have no individual case reporting requirements. We will instead report any signals either in an expedited manner or as part of routine aggregate reports in keeping with how we currently treat observational data from other sources. In order to ensure alignment and acceptance both internally and externally, we have already consulted and communicated this approach with the FDA, MHRA and EMA, GSB, the OCMO Leadership Team, Global Digital Risk Board, the patient privacy office, PV compliance, regulatory compliance, legal, IT, the joint GCSP/GRA leadership team, and others.Publication: Results will be handled appropriately by the team in compliance with GSK policiesHas been deemed a PASS study and reviewed by SRT (14 Nov 2014) and PRF (12 Dec 2014)All trademarks identified herein are the intellectual property rights of their respective owners.