Collecting Adverse Events with Social Media
Social media as a tool for improving drug safety is now an old idea (“old” in the world of social media means more than a year ago). We have seen several sites, mainly patient oriented that deal now with drug safety. See such sites as www.patientslikeme.com and www.adverseevents.com. One also sees AEs collected on “old” social media sites like Facebook and LinkedIn.
What these sites seem to be doing is collecting AEs from consumers, patients and healthcare professionals. There are forums, groups, chats, blogs, discussion groups in which all sorts of things are seen including advice from one patient to another (rarely from a healthcare professional to a patient for all the usual reasons: liability, hard to know a patient and what to do from a short paragraph on a social media site etc.), lamentations, fury at the drug companies/doctors/hospitals/insurance companies but not so much with nurses or pharmacists.
In other words these sites seem to be a very useful forum for venting and exchanging ideas, anger and possible solutions.
The question that folks are only now starting to ask is whether there is a good way, using social media, to collect meaningful medical and scientific information from which sound, evidence-based advice can be given. If the answer to this is yes, then social media will be providing more than just an outlet to vent spleen but perhaps a mechanism to cure spleen (sorry for stretching the cliché).
A very interesting proposal has just come out of Brussels from an organization known as the “Innovative Medicines Institute”. This is a public – private venture funded by the European Union (EU) and the pharmaceutical industry in Europe via their trade organization (European Federation of Pharmaceutical Industries and Associations (EFPIA)). The EU and EFPIA have each contributed € 1 billion ($1.3 billion). From their website (https://www.imi.europa.eu/content/mission):
“With a €2 billion euro budget, IMI supports collaborative research projects and builds networks of industrial and academic experts in Europe that will boost innovation in healthcare. Acting as a neutral third party in creating innovative partnerships, IMI aims to build a more collaborative ecosystem for pharmaceutical research and development (R&D).”
On July 9, 2013 the IMI announced a new initiative-the WEBAE Project (Web Adverse Events):
Tapping into social media to monitor drug safety
“Recent years have seen an explosion in the use of social media, applications (‘apps’) and online platforms that allow people to share news and information and connect with one another in unprecedented ways. At the same time, legislation requires the relevant authorities to monitor the safety of medicines by gathering information on adverse drug reactions (ADR) experienced by patients. Currently this takes place via official forms. Nevertheless, for a variety of reasons, many ADRs go unreported.
One of the main objectives of the proposed WEBAE (‘Web Adverse Events’) project is to develop tools to tap into the wealth of content that is openly available online to both detect ADRs, and provide patients with the most up-to-date drug safety information.”
The budget for this is approximately € 4.5 million ($6 million) and will run 3 years. Full details are available at https://www.imi.europa.eu/sites/default/files/uploads/documents/9th_Call/Calll_9_Text.pdf
We’ll discuss them briefly here.
The ideas underlying this project include:
- The use of digital media and Web 2.0 (e.g. Facebook, Twitter, LinkedIn) with the ensuing changes in the way people communicate and connect.
- The rapid adoption of new hardware including smartphones and tablets producing an interactive exchange rather than a passive receipt of information.
- The use of apps.
From industry, the participants in the project are Novartis, Janssen, AZ, Sanofi and UCB. There are many other stakeholders: biotechnology companies, small & medium enterprises, regulators (EMA and member states), policy makers, government agencies, payers, academics, and non-profit organizations (e.g.: patient, disease advocacy and privacy groups). (Author’s note: Nobody from Facebook, Google, Microsoft etc?!).
The problems they wish to resolve include:
- How reliable signals can be picked up from social media,
- Developing tools and methods to capture spontaneous reports from social media or mobile with real time data mining to find emerging safety signals
- Deal with the noisy nature of social media data. The credibility and source of such reports are key issues. For example, there is concern that social media with no appropriate checks can open the avenue to unscrupulous attacks from “pseudo-reporters”. (Translation: Bad guys will cheat and game the system).
The major goal is to develop methodologies and data mining algorithms for social media content (forums, blogs, tweets, public posting, etc.) in order to find emerging, medical signals. The system will be outside of and outside of the control of pharmaceutical companies. (Author’s comment: Oh my. Does this mean that they don’t trust industry?).
A second goal is to try to develop applications for direct reporting of suspected ADRs to the health authorities electronically from social media via EudraVigilance (the EU safety database).
So far this is reasonable enough.
The deliverables, alas, are primarily rather soft and are, in fact, not really deliverables but rather are objectives and goals. Some “hard” deliverables are in blue:
- Establishment of regulatory framework for social media mining for ADRs
- Creation of algorithms to enable the extraction and identification of ADRs and to “enable the provenance of the data to be established” (Author’s note: seems to mean figuring out where the data came from).
- Increase public access to enable direct reporting
- Develop guidance on how to identify reliable signals from social media (Author’s note: companies and governments can’t even do that now even when they own, control and generate the data themselves).
- Provide direct acknowledgement to the source on receipt of data
- Engage patients in awareness and reporting of safety concerns
- Strengthen the monitoring of potential ADRs in the paediatric as well as the elderly population, the latter specifically affected by poly-pharmacy
- Further engage healthcare professionals (HCPs) in reporting suspected ADRs
- Integration of apps with existing EudraVigilance and workflows & tools as well as with existing global safety databases used by companies
- Develop quality metrics for evaluating ADRs from social or crowd-sourced platforms taking into account the multilingual aspects of reporting.
- Creation of a sense of community around pharmacovigilance (Author’s note: ??????)
- Improved methods for mining social media for pharmacovigilance and pharmacoepidemiological data
- Enable hypothesis generation for researchers
- Provide a platform for monitoring the effects of risk minimization
- Provision of safety messages and alerts to patients and health-care professionals (including video and interactive features)
- Registration/certification for prescribers (Author’s note: What does this mean?)
- Enhanced surveillance
- Potential for enhanced monitoring
There are many potential issues to resolve not the least of which is language. There are dozens of languages and dialects used in the EU. If one throws in abbreviations and slang as well as incorrect, ungrammatical, misspelled words, well…..lol.
There is an aggressive three year plan with more detail. Based on these plans, the primary outcomes seem to be the development of apps, databases and visualization interfaces (do you know what that is?).
This is a lovely and laudable idea in theory. The practice looks very very messy.
There are too many stakeholders and they include some of the wrong ones and exclude some of the right ones. As with any large and varied group of dozens (or more) players, usually a handful of people end up dominating the group and determining the outcome. How the operational aspects of this will play out remain to be seen.
It’s not clear what “success” is. Will success be simply uploading AEs from many languages into EudraVigilance (without viruses and malware)? Or will it be more subtle: eliminating cheaters, getting ADRs rather than just AEs, avoiding duplicate reports etc. etc. In many situations like this, the lofty goals are rarely met but some minimal deliverables ultimately do get prepared and everyone will declare a victory.
Will they actually test it? It seems they will since this will be an iterative procedure. If the software developers have any say in the matter they will force the project to actually test the software using standard software life cycle development procedures. That may not be easy.
What is that “registration/certification for prescribers” deliverable mean? It is not explained but does not sound good on its face.
Don’t tackle they language issues till they figure out the science.
Lots of other issues in here make this whole project way too ambitious and over-arching. What is needed is simplification. Here’s my revised proposal:
- · Limit the size of the committee to ten members and include people who know about social media(!) such as from Google, Facebook, Twitter etc.
- · Start only in English (the major language of the internet – at least right now).
- · Limit the scope to one or at most two social media sites: Facebook and Twitter (say).
- · Limit the goals to developing two apps:
- One to pick up real ADRs
- One to pick up signals from these ADRs (with or without additional data already in EudraVigilance).
- · That’s it. Stop there. If they can do this they should be heartily congratulated!
This seems possible (perhaps) in three years with $6 million.
Undoubtedly my proposal will not be adopted. Rather, a large project will occur with lots of meetings (in lovely venues no doubt), lots of talk, lots of delays, cost-overruns, a rethinking (18 months into the project) of the goals and on and on and on. Some deliverables will be produced but the final outcome will be that another project, this time with more funding and more time, is needed to achieve these goals.
Hope I’m wrong but I doubt it.