PV and Social Media- The Full Article

Reprint from PIPELINE Issue 55; April 2018 © Copyright Pharmaceutical Information and Pharmacovigilance Association. All rights reserved.

The Future of Social Media in Pharmacovigilance

I will start by setting my stall out as being somewhat of a luddite when it comes to social media. I’m not on Twitter, Instagram or Snapchat. Beyond the inexplicable compulsion to check out the lives of people I haven’t seen or spoken to in years, on Facebook (FB), I’ve remained pretty antisocial in my media consumption.

Over my years in the industry, every now and then an article in my morning paper has told the story of an adverse drug reaction (ADR) involving a drug that the company I was working for couldn’t discount as its own. On it went to the safety database as an Individual Case Safety Report (ISCR), just like my pharmacovigilance (PV) training had taught me. But what was the value of this kind of report to the understanding of the safety profile of the product? It would contain the four minimum criteria and one would hope (in the days before a single online comment became known as a ‘Twitter-storm’), basic journalistic standards and competence would mean the events did take place. All in all, I always thought that it was a solid addition to our knowledge of the safety profile of the drug, even if it was inevitably an already well characterised reaction. Anyway, this kind of case was so few and far between, that they weren’t likely to clog up the database with low quality data.

With over a quarter of the global population on FB alone, social media reports have the capacity to swamp PV departments, but there is little understanding of what the merits of this data may be. The area is currently a hotbed of confusion, with limited official guidance from regulators or thought leading opinion from industry.

The WEB-RADR (https://web-radr.eu/) project began in 2014, funded by the Innovative Medicines Initiative (IMI), a large-scale public-private partnership between the European Union (EU) and the European Federation of Pharmaceutical Industries and Associations (EFPIA). Reporting its results at the end of the project, in September 2017, it had had great success with the development of applications (apps) through which patients and healthcare professionals can report ADRs. Along with this, though, there was also a workstream looking at “access to classified social media data via a visualisation platform for signal identification and confirmation”.

The results of this give a strong starting point for any company looking for guidance on how to integrate social media mining into their PV system. As a high-level conclusion, using Twitter and FB may have value as an independent source of data and for hypothesis generation, but they did not add value to ‘business as usual’ PV. In addition, only 17% of identified ADRs could be automatically mapped to MedDRA (although this was doubled if historic VigiBase[i] mappings were used), demonstrating a significant element of manual effort in reviewing the posts. The introduction of Identification of Medicinal Products (IDMP) in the next couple of years will also require greater specification of what product was taken, which doesn’t lend itself to such freeform text, either. Patient forums performed better at identifying signals than Twitter or FB, although identification of these ‘signals’ often occurred later. The EMA representative at the WEB-RADR closing meeting, stated that social media reports should be considered a secondary source. Companies should not have to plan any interaction with users (follow-up etc) and focus on best using the data collected for public health protection. There should be no expectation to report ADRs identified in social media as ICSRs, but companies could consider reporting them in periodic safety update reports (PSURs)/ Risk Management Plans (RMPs). Where used, such data should be clearly defined and documented.

There is no clear, regulatory guidance on defining and documenting how social media data should be used. However, I don’t believe it should be over complicated. Start with deciding exactly how you want to use social media data (if you decide to use it at all), whether it be hypothesis generation or trying to integrate the data into signal validation activities, and work backwards on how to get the data you need. Mining these kinds of data sources is evidently as much a technology problem as a PV one, so may well be out of the reach of most small and even medium sized companies to deal with in-house, whether that be for budgetary reasons, or lack of in-house expertise. There are increasing numbers of PV providers offering to monitor social media, but it’s crucial to know exactly what services they provide : from how the data is validated to what websites are being searched. In my opinion, it is important to clearly identify, up front, how such data will be used, which in turn will guide the optimal methods of collection. It’s certainly not something to jump into first, because it’s an exciting new world, expecting to work out how to use it further down the line.

Social media is here to stay and ultimately patients have a right to expect that they are being listened to, however they have chosen to communicate.

It will certainly be a difficult balancing act for the industry over the coming years. The value of the data will need to be continuously evaluated, perhaps through further collaborative WEB-RADR-like initiatives, so that the findings can inform the industry as a whole.

However, that’s not to say that social media data should be ignored by individual companies until then. Reputationally, that’s not an option. In the event of a signal being detected, which is of such severity, or in such a widely used drug, that it is reported in the popular press, it’s inconceivable that there wouldn’t be attempts to uncover ‘reports’ of occurrences somewhere online. If and when that happens,  there will need to be clearly justification why such a report wasn’t immediately identified and acted upon. Whether or not it is specifically mandated in a Good Pharmcovigilance Practice (GvP) Module, won’t be seen as acceptable outside of the industry. It will be harder still if companies are perceived to be perfectly happy to market and advertise their products online, but not so willing to listen in return.

Social media mining is not going to be the great panacea for pharmacovigilance companies, nor is it completely useless (and even if it were, it can’t just be ignored). The truth, as is always the way, lies somewhere in between. Indeed, the biggest impact social media may have on patient safety, in the short term, may well be the opportunities it affords for companies to develop innovative information delivery methods (Dear Healthcare Professional letters), rather than on the receipt of information. Anything that can improve the relationship between pharmaceutical industry at all stake-holders (patients, HCPs, regulators etc.) must be a good thing.

Drive Phase PV is a pharmacovigilance consultancy and provider of pharmacovigilance services, supporting patient safety throughout the life-cycle of our clients' products. Contact us to discuss developing your PV processes in the social media age. Whether via SOP development or provision of training, we deliver a tailored packaged to seamlessly fit to your company structure and strategy.

[i] VigiBase is the World Health Organisation’s (WHO) ICSR database. Participating states contribute to what is the largest drug safety repository in the world.