Socially vigilant: Does pharmacovigilance need secondary data & social media?
As the world’s drug consumption continues to rise alongside advancements in the medical and pharmaceutical industries, so too do new challenges. And as technology also evolves, new solutions are coming into focus. But they are not without their own challenges, as Matt Fellows discovers.
“Events of the last decade, including more thorough safety documentation and reviews for drug approvals, and increased warnings and awareness about adverse drug reactions, have made drug safety one of the top issues for consumers and regulators,” Dr Olga Carroll explains in her essay Global Pharmacovigilance: Requirements and Trends – the Global Regulatory Perspective. The market for pharmacovigilance was estimated to be worth more than $3 billion in 2015 and analysts project that will more than double to $8 billion by 2024; In 2010 in the US alone, over 6.4 billion drugs were dispensed to patients, and with the rising incidence of specific illness such as cardiovascular, respiratory and oncology-related diseases, in addition to concerns such as ageing populations, the global burden is growing and with it grows demand.
With an unprecedented quantity and range of treatments on the market, potential safety risks are a major concern; never before has pharmacovigilance been so crucial, and it will only continue to become more important as time goes on.
But as the world changes rapidly, current processes across the industry increasingly run the risk of becoming out of step and underperforming. There is a necessity and moral obligation for pharmacovigilance services to deliver not only the highest standards, but also to function in a manner which is fit for purpose in a climate that is ever-shifting.
A new standard
This highlights the need to regularly scrutinise ongoing processes of pharmacovigilance, and indeed all systems surrounding the life sciences and healthcare sectors. And this is just what the European Medicines Agency (EMA) did in 2012, with the launch of new legislation which was, as the EMA describes, “the biggest change to the regulation of human medicines in the European Union (EU) since 1995.” The legislation was introduced based on findings which suggested adverse drug reactions (ADRs) were responsible for 197,000 deaths a year in the EU. Based on reports, it is estimated that ADRs stand as the fifth-highest cause of death in hospitals, accounting for 5% of admissions.
ADRs are not a threat to be underestimated, and the aforementioned statistics only reinforce the necessity of robust pharmacovigilance measures. With this in mind, the EMA’s new legislation called for higher and more consistent standards across the member states of Europe concerning the issues of data collection, regulatory action and transparency levels, all delivered in a timely manner with a view to, primarily, reduce the overall number of ADRs.
Pharmafocus spoke to Mick Foy, group manager, Vigilance, Intelligence & Research Group at the Medicines and Healthcare products Regulatory Agency (MHRA), who was quick to stress the importance of this new legislation early in the conversation, noting that a lack of required guidelines across EU states had led to patchy and inconsistent level of pharmacovigilance efficacy across the continent.
“The European Commission recognised that the member states of Europe were operating at different levels of maturity,” he explains, “so there were a lot of new requirements on us all, such as the broadened scope of what is an adverse drug reaction; the need to operate quality management systems; the different assessment procedures that were brought into play with those new regulations.
“Some countries were more advanced in their ability to operate those new requirements, so we set about a project that looked to identify what practice looked like in a number of these areas, and then turned that into guidance and training material for the member states so we can all raise the bar, if you like, and operate those new requirements. So what that’s shown us is that there’s really differing practice across Europe and what some countries were able to implement was easier than others, so we’ve been learning off of each other.”
The EMA evaluated the effects of the legislation after one full year of application together with representative medicines regulatory authorities in the member states. It discovered, among other findings, that patient reports of suspected adverse drug reactions had risen by more than 9,000 on the previous year, and thousands more staff were adequately trained in pharmacovigilance.
Mick Foy, group manager, Vigilance, Intelligence & Research Group at the Medicines and Healthcare products Regulatory Agency (MHRA)
Foy looks back on the introduction of the legislation as a significant step forward in creating a shared and applied vision of best pharmacovigilance practice:
“I think it’s been successful in a number of areas, not least the more systematic review of emerging signals through the PRAC (Pharmacovigilance Risk Assessment Committee). What we see now is a lot more being dealt with in a timely manner; the process is much better organised for identifying and discussing and agreeing a course of action for the safety issues. Prior to this new system being in place, the timescales were not set down; the process was variable, and I think there’s a much more systematic approach to handling safety issues now. And then beyond that, the communications and the transparency over what’s being discussed and how it’s being handled is much more obvious. So there’s full disclosure and transparency that maybe wasn’t in place prior to that. So I think that’s been a major step forward.
“At a more local level, I think what the legislation has done has engaged much better and fuller with stakeholders,” he continues, “so the introduction of patient reporting across the whole of Europe I think is a major step forward for the inclusion of people who actually take the medicines that we are looking at to understand if there are any safety issues. The wording of the legislation is that we must “take all appropriate measures to encourage the reporting of adverse drug reactions with patients and healthcare professionals”. I think that’s been a bit of a call to arms if you like at a national level to say you must reach out to your stakeholders, tell them about the existence of your systems and encourage them to send you reports of suspected adverse drug reactions.”
However, despite this new legislation, the demands of pharmacovigilance have continued to evolve as drug consumption has continued to rise; while these measures do indeed ‘raise the bar’ for pharmacovigilance practices in a much-needed and welcomed fashion, as time passes it becomes increasingly evident that there are still shortfalls in the system which could require a different approach to adequately address. Traditional methods of ADR reporting, such as paper-based reports, are often inconvenient, slow and inefficient. It is a pervading professional opinion that there is a necessity to explore and utilise new avenues of data collection to improve the efficacy of the field, as Itzik Lichtenfeld, CEO of intelligence firm Data2Life outlines in his essay The Past, Present and Future of Pharmacovigilance:
“Today, the industry finds itself in the very early stages of a long-term journey toward a more data-driven and analytic-enabled approach to PV, post-marketing surveillance and drug safety management as a whole,” he says. “The same data and signal detection capabilities can transform innovation and product development, clinical trial design, medical affairs and risk management.
“In practical terms, regulators’ attitudes toward secondary data cannot be overlooked. That one of the world’s biggest and most visible regulatory bodies views secondary data as a valid input to policy making decisions demonstrates its long-term importance in the drug safety realm. Per the FDA’s Sentinel Initiative, EMA’s WEB-RADR program and the SAFER project which is funded by the European commission research executive agency, the PV community is beginning to recognize that social media is not just for marketing anymore. As such, there is reason to believe that today’s scepticism will slowly become tomorrow’s advocacy.”
As mentioned, the industry is moving always closer to engaging with this new avenues; of particular note among these is the EMA’s aforementioned WEB-RADR project. Launched in September 2014, WEB-RADR is an ongoing effort funded by the Innovative Medicines Initiative (IMI) which seeks to utilise the power of social media and other technologies for ADR-reporting purposes, as well as to establish guidelines and regulatory framework on the use of the technology for such reporting. The project’s ultimate goal is to deliver a mobile app which can be used by patients and healthcare professionals as a new method for reporting ADR, one that is convenient, quick and efficient in a way that conventional tools are not.
Foy discussed the significance of the project: “[The IMI] put a call out to understand how social media might help pharmacovigilance. It’s been running now for two and a half years; it finishes in September of this year, it’s a three-year project. It’s been looking at two areas: one is the use of social media and the other half has been looking at the use of mobile technologies to help with reporting cases.
“On the social media side, the reason why this was interesting and warranted research was that there was a growing access to social media data by regulators and the pharma industry. I think it was primarily being looked at by marketing departments to see what people were saying about their medicines and how they might use that data to understand benefits. There was a pharmacovigilance concern there, because if people are talking about their medicines, they could be talking about harms that they’ve experienced, or there might be quality issues or suspected counterfeiting issues, or how people are misusing and abusing their medicines, so pharmacovigilance systems became concerned that if people are accessing this secondary data, they have certain responsibilities there. So good vigilance practice stipulated that if they are in the social media space and they uncover cases of suspected adverse drug reaction and they met the validity criteria for a case, then that needed to be reported.”
According to Foy, the project has been very much a learning experience as the industry ventures into this new medium to assess to what extent it could be useful in the generation of relevant data.
“This was very early days and there was concern that this could generate huge volumes of poor quality, low-value reports, so we started this project to really understand how to best access social media data, how best to analyse it and interpret it,” he continued. “We’ve come a long way since the beginning of the project and we’ve looked at a huge volume of social media data: Facebook, Twitter, forums, Reddit and other forms of social media. The analysis is still underway; data tells you different things depending on what you’re looking at, so there are drugs that are better represented in social media data than others; there are types of use and types of ADR that are more discussed than others. I think at this stage what we’re seeing is not a single best use of social media; there are areas where it might be beneficial and areas where there is an absence of quality data.
“The research findings are not final, but there is a lot of information on how certain products are misused for illicit purposes to get a ‘high’,” he continues. “So people are talking about how to break down their medicines to very creative uses. What that means for pharmacovigilance, we don’t really know yet because it’s not really validating a safety signal, but there are areas that companies and regulators need to be aware of how people might be using products in an unintended way; it’s good to get that understanding.”
A challenging nature
The immediate and universal finding when exploring the functionality of secondary data, particularly social media, for pharmacovigilance uses is that applying the technology in a manner so much in its relative infancy is the challenges it presents. WEB-RADR has run into plenty of obstacles in its almost-three year history, and the fleeting, mercurial nature of social media is foremost among them, as Foy explains:
“The social media environment is very changeable and new social media platforms come on stream quite quickly while others disappear; the predominant data at the beginning of this project was Facebook data – it had a lot more quality to it because when you compare it to Twitter – with its limited character set, you got fuller stories form Facebook posts. However, Facebook changed our access policy, so we no longer have an ongoing access to Facebook data. So we brought in Reddit, and that’s a different community again. So I think where we’re heading with this is not to say “this is how you should use social media data, as a rule”; it’s likely that we will be giving case studies on where it’s been beneficial, where it hasn’t been beneficial, the sorts of areas where it might be put to good use. I think the GDP may well change in light of the findings from the research, but we haven’t come to any conclusions as yet.”
Beyond their fast-moving nature, another trait makes social media platforms difficult to bend to pharmacovigilance utility: namely, the fact that they function as a social media platform in the first place, and not a specialised medical tool.
As Foy states, the platforms can be a great source of information, but this is often dependent upon user input and there is still a propensity for “an absence of high-quality data”. A possible solution would be the use of computer algorithms and tools to trawl the huge amounts of data available, but the platforms’ very nature often makes it very difficult to harvest data from them in efficient, automated ways, as outlined by Richard Sloane, Orod Osanlou,, David Lewis, Danushka Bollegala, Simon Maskell and Munir Pirmohamed in their essay Social media and pharmacovigilance: A review of the opportunities and challenges:
“Current technological challenges include the difficulty for algorithms to interpret layperson technology. Inter-annotator disagreement during the annotation process on interpreting ADRs in social media makes it extremely challenging, if not impossible, for algorithms to interpret postings with certainty.”
As the use of social media in pharmacovigilance practices continues to evolve, this will be a consistent challenge that will need to be solved to properly unlock its full usefulness.
Where the ultimate goal of pharmacovigilance is to keep patients safe from ADRs to the best possible degree, it can be argued that the use of social media as a tool in this regard becomes a necessity. Sloane, Osanlou, Lewis, Bollegala, Maskell and Pirmohamed echo this sentiment, but again, it proves to be a complex issue:
“It could be argued that there is an ethical obligation to attempt to harvest social media data in order to enhance pharmacovigilance and ultimately improve patient safety. However, this raises important and controversial ethical issues. The boundaries of what constitutes public and private data are not comprehensively defined.”
Thus, there is a fine line to be walked in application of the technology; as the essay states: “The balance of privacy, duty of care and the greater good is complex.”
But do we need it?
But with all these apparent complications, is the payoff worth it? While it definitely has its place in the pharmacovigilance system, Foy is not entirely convinced of its necessity, particularly when compared to other, more directly applicable secondary data sources such as electronic medical records, but the need for it to be held up to regulatory scrutiny is still a priority:
“I wouldn’t say at this stage that we ‘need’ [social media] for pharmacovigilance purposes. I think people are using it, and if you’re going to use it, you need a good set of policy guidelines on where it’s useful and where it might not be useful, and some of the pitfalls when you get involved in this space. There is no requirement to use social media data. I don’t see as yet that it is a significant tool in the pharmacovigilance armoury; I don’t think social media is replacing spontaneous reporting and the Yellow Card scheme. It’s certainly not replacing things like CPRD (Clinical Practice Research Datalink), electronic healthcare records which have a trusted, high-quality and longitudinal history to them. Proper scientific evaluation of drug safety signals, I think there are good tools in place for that. But social media is still a new tool and a new possibility, and it’s probably maturing all the time, but saying that, it’s changing all the time as well. I think it might have a place to help understand some real-world use and what real users are saying, but I don’t think it’s at the stage where we can say it’s an authoritative resource for pharmacovigilance.”
It appears industry professionals are still somewhat divided. Lichtenfeld argues that when it comes to wider secondary data, including social media, it is important to stress that it is complementary to standard practice:
“There is lingering skepticism among industry veterans regarding secondary data. Some consider secondary data inferior to the primary research data that has been the gold standard of the industry for decades. But secondary data is not meant to replace primary research data. Nor is it meant to replace skilled and experienced researchers. Rather, secondary data will complement and enrich primary data, enabling deeper and more sophisticated analysis and serving as another source for detecting and validating adverse event signals.”
Ultimately, it is of course early days and experts are divided, but the value of the possibility cannot be denied. Foy’s cautious patience at this stage could prove to be prudent, but there are compelling practical and ethical arguments for social media and secondary data as an important, perhaps crucial, source of timely, real-world data collection for pharmacovigilance. As Sloane, Osanlou, Lewis, Bollegala, Maskell and Pirmohamed state: “The rapid development of technology brings with it the opportunity to harvest vast amounts of data, both from social media and emerging biomedical technologies, to allow earlier signal detection, develop early warning systems, and enhance pharmacovigilance and patient safety. If data is harnessed correctly, the field of pharmacovigilance could potentially undergo a revolution.”