Talking with Paula Rausch, Ph.D., RN, Associate Director of Research and Risk Communications in CDER’s Office of Communications
Proactive pharmacovigilance is a broad, systematic approach to monitoring drug use and abuse trends in society. It employs various tools to mine multiple sources of epidemiological data. CDER has begun to use nontraditional sources to explore the social contexts in which substances are being used, as well as to identify potential drugs of concern that may be emerging. These data can provide clues about the dynamics of use, misuse and abuse, and potentially identify changing patterns.
One of the first steps in this nontraditional monitoring process is to identify and analyze content found in conversations occurring online and on social media sites. Paula Rausch, Ph.D., RN, who is the associate director for research and risk communications in CDER’s Office of Communications, describes how we use online and social media data to inform our pharmacovigilance efforts, and what it can and cannot tell us.
What can social media monitoring tell us in general about prescription drug use or abuse?
Social media monitoring is just one part of a broad research process CDER undertakes to understand how people may be using or abusing opioids and/or related substances as alternatives or adjuncts to prescription drugs. Because people often talk about these topics with others in online forums and on social media, exploring these conversations is one of the preliminary steps in our broader formative social science research strategy and is intended to help us gain the most expansive understanding possible. This data can help us identify the areas we want to explore in greater depth through more detailed and rigorous primary data collection efforts. When we monitor social media platforms, we are trying to get the broadest swath of information that we can about how people are talking about a particular issue.
How does CDER use the data gleaned from social media monitoring?
For our broader social science research projects, we often conduct a social media analysis as part of our environmental scanning process, which might include reviews of peer-reviewed articles in the literature, and of non-social media websites and campaigns. We also conduct social media research on narrower topics, such as on a specific opioid, or for specific time periods, such as over a five-year period or the last 30 days. These sources give us a basic framework of information.
We can use this data in several ways. First, we may use it to inform our communications activities. For example, what we uncover about people’s perceptions or attitudes about a specific drug or issue can help us develop targeted messaging and enhance our efforts to reach intended audiences. It can also help us answer the questions and concerns they may have.
We may also use the data to identify potential new trends in the use of drugs, including trends that could suggest issues we need to act on. With this early warning, we can turn to other tools, including more rigorous qualitative and quantitative social research efforts, such as focus groups and interviews, surveys and experimental studies. These can help us decide how to proceed. We also share the data with CDER leadership and others who may find the information valuable for their own work, including the Office of Surveillance and Epidemiology (OSE).
What methods do we use to monitor social media platforms, and which types of social media platforms do people frequently use to discuss prescription drug use?
We use a commercially available social media “listening” platform that aggregates publicly available data daily across more than 95 million online and social media sites. These sites include popular social media services like Twitter, Facebook, and Instagram, but it also includes forums, blogs, discussion groups, news sites, and other nontraditional sites. We also supplement that data with searches conducted directly on Reddit, through Google, and on specific forums that we’ve identified through the social media listening platform as popular places for people to discuss the topic of interest. We download these online conversations and then analyze them manually using a detailed and systematic process of qualitative data collection, analysis and reporting. When possible, we will also do qualitative analysis using computer software.
It is crucial to note that we only draw from and look at publicly available data. The aggregator does not breach any firewalls, and the 95 million sites that it pulls from, as well as all the other sites and forums we review, require no registration, subscription or membership. We see only what people have agreed to share with others and do not see data from any social media account that is set to private. We do not collect any demographic or personal data.
Can you describe an example of how you have used social media monitoring to broaden our understanding of an issue of concern?
One example is gabapentinoids (e.g., gabapentin and pregabalin), which are approved for the treatment of seizures and certain types of neuropathic pain. Epidemiologic data showed that between 2002 and 2015, U.S. rates of gabapentinoid use tripledi. By 2017, pregabalin was ranked as one of the 10 best-selling drugs in the United Statesii. This striking increase in usage raised concerns, including about the possible misuse and abuse of these drugs.
As part of CDER’s efforts to understand what was happening, we identified and analyzed public conversations about how these drugs were being used, for what purposes, and what experiences people were having while taking them. Our findings suggested a shift between 2013 and 2017 away from discussing the drugs for legitimate treatment of pain and seizures, and toward a focus on misuse and abuseiii. People were discussing taking dosages up to three times greater than the maximum recommended doses, taking them with numerous other substances, and taking them to feel euphoric and to manage withdrawal symptoms.
What are the benefits of monitoring social media as part of our pharmacovigilance efforts? What are the limitations?
A majority of Americans who have internet access use social media. Online forums and social media are important venues for diverse groups of people to share information and opinions, ask questions, and get advice about health and other issues. In addition, due to online anonymity, people may be more honest and open to sharing information online that they might be less likely to disclose in person. They can express themselves in their own ways, on their own time, and on the platforms of their choice. This can provide a broader, more detailed and potentially more authentic picture of a topic or issue than we might get through other forms of research. In addition, we can essentially follow conversations in real time as they unfold, or over certain periods of time.
On the other hand, there are significant limitations to this type of data. For example, we cannot know for sure if the people participating in discussions are real people, and we do not know if their accounts are truthful. In addition, we can only obtain a snapshot of what people are talking about at the specific time the data is collected. If we conduct a search today, and then conducted an identical search using the same search parameters a week from now, we would likely get different results. Moreover, a large chunk of social media material is not publicly available. For instance, only about a quarter of Facebook content is public.
The findings are not reproducible, generalizable or even representative, so we cannot use them conclusively. Social media research must be used in conjunction with other data and methods. However, for our purposes, social media monitoring is the fastest and cheapest way to gather a lot of information quickly, and to help us formulate steps for subsequent research. It gives us a big picture view of a very heterogeneous group across the country and potentially globally, as well.
i Johansen ME. Gabapentinoid use in the United States 2002 through 2015. JAMA Intern Med 2018; 178:292-4.
ii Urquhart L. Top drugs and companies by sales in 2017. Nat Rev Drug Discov 2018;17:232.
iii Throckmorton D, Gottlieb S, Woodcock, J. The FDA and the Next Wave of Drug Abuse—Proactive Pharmacovigilance. NEJM Jul 19;379(3):205-207. Doi: 10.1056/NEJMp1806486. Epub 2018 May 30.