Through the use of de-identified Big Data from online patient forums open to healthcare providers, the pharmaceutical industry may glean useful insights into both the safety of existing products as well as future needs of patients. Post-marketing safety surveillance for pharmaceuticals currently relies on data from adverse event reports to companies or regulatory authorities, medical literature, and observational databases. Together these sources provide some insight into everyday product safety or risk, but the unique insight the patients themselves can offer is also highly desirable.
Using insights from a 2016 research project involving Inspire, GlaxoSmithKline (GSK) Pharmaceuticals, and Epidemico, an innovative informatics company, we are exploring the use of social listening data for pharmacovigilance and other R&D concerns. A core question is, “What valuable insights can we glean from social listening to help improve patients’ lives—whether through improved safety, more relevant clinical trials, or research and development of new treatment options?”
Patient-centric social media for outcomes and pharmacovigilance considerations: The time has come
1. Patient-centric social media for
outcomes and pharmacovigilance
considerations: The time has come
Brian Loew
Founder & CEO
Inspire
Greg Powell
Director, Safety Evaluation and Risk Management
GlaxoSmithKline
Julie Flygare
Founder
Project Sleep
3. Connected Patients
PD-L1 Immunotherapy Clinical Trial
“72 hours post first immunotherapy and feeling great with no side effects that I can feel but a surprising
rather quick response (I hope) is that I have been bleeding for the last several months from the upper
vaginal tumor and today no bleeding. It may start up again but for the last 18hours nothing.”
”My husband is also looking into this trial. Do you know if your tumors tested positive or negative
for PD-L1?”
“Yesterday I got news that my CT after chemo was complete shows increasing metastasis throughout chemo.
The Dr asked me if I had heard of any clinical trials and specifically mentioned immunotherapy and as of
yesterday had not heard of it. Now today, I stumble upon this gem of a conversation and see all of this
wonderful info and inspiring people. Thank you for sharing your experiences, they really do help others.”
4. Patient Reported Outcomes
“Can you please share your Cosentyx side effects? Hair loss? Nausea? I have never been on a biologic
so not sure that insurance will cover it but I'm thinking about Cosentyx....is this a good route?.”
I've been on Cosentyx for a month now and the only side effect that might be attributed to
Cosentyx is a bit of fatigue the day after injection. It's been great so far. My PsO is much less
and it has helped my PsA a very small amount.
I have been taking Cosentyx for almost a year. My psoriasis is gone and the arthritis is almost gone. I have had no
side effects at all. This drug improved my quality of life immensely. Good luck!
5. Through natural language processing of anonymized
data, we are able to uncover insights about adverse
events, medication routine adherence, and many others
related to the patient journey and health experience.
7. Digital World
Project CRAWL
Contextualization of ReAl-World Drug Use Through Social ListeningDIGITAL MEDIA AS A MAINSTREAM COMMUNICATION CHANNEL
38 years for radio to reach 50 million users
13 years for television to reach 50 million users
1.5 years for Facebook to reach 50 million users
85% of adults use the internet
6% of adult internet users have posted comments, questions or
information about health or medical issues on a website of any kind
3-4% of adult internet users have posted their experience with health care
service providers or treatments in the previous 12 months
Social
Media
Adoption
Treatment
Discussion
Patients’
Insights
8. How can we listen to what patients are saying online?
9. Project CRAWL Overview
Contextualization of Real-World Drug Use Through Social Listening
Publically available data is collected from multiple
sources (Facebook, Twitter, Internet chat rooms)
System filters out noise, highlights important information,
and supplemental data is added (spontaneous data, etc.)
Users can interactively explore the information using an easy
to use interface that helps facilitate the review process
10. Key Learning (so far)
Quantity of data
A PEW study found that 3-4% of adult internet users have posted their experience with health care service
providers or treatments in the previous 12 months. A study completed by GSK showed 22 million potential
adverse events for about 1,000 drugs (publically available Facebook and Twitter posts, English language only).
For comparison, the FDA’s FAERS database has ~6 million adverse event reports since 1968.
Timeliness of the data
Traditional data can lag behind (9-12 months for observational data)
Social listening data is usually available within hours from the time the person posts their comments online
Geographic diversity of the data
Traditional data is US and European centric.
Three of the five largest users of the internet are Asia, Latin America/Caribbean, and Africa
11. Key Learning
Drug Specific
Potentially valuable clinical insights
~26% of people discussing drugs will talk about
efficacy/benefits
• Time to onset, complete/partial
• Within context of cost, other treatment
options, access challenges
• How does this improve their quality of
live/average daily living
6% of people discussing drugs will seek
medical advice
• 50% efficacy related
• 25% safety related
OTC (GSK example)
• Real world benefits
• Can patients effectively self-manage
medical condition
• How real-world use differs from the
products label
Drug abuse/misuse (GSK example)
• Encourage/discourage use
• Best route of administration
• Where they get the medication
• Anticipated effects
• How to combine with other drugs
13. Key Learning
Disease Specific
0%
20%
40%
60%
80%
100%
Percent of
patients
represented
per diagnosis
mention
CV Risk
Factors
Mentioned
Diagnostic
Tests
Mentioned
Duration of
Disease
Known
Drug Therapy
Discussed
Type of
Reporter
Known
Age Range
Known
Gender
Known
Country
Known
Forums (N = 231) Twitter (N = 420)
Reddit (N = 9,097) Inspire (N = 2,911)
14. Limitations
Causality & veracity
Patients may not correctly assess causality. Can system be gamed?
Bias
How are people who post online similar/different than general population?
Volume
Volume of reports likely to be large.
Privacy & generalizability
Patient privacy expectations and fear of government oversight.
Regulation unclear
When is there an obligation to monitor or report?
15. Conclusion
Social Media is
an important new data
source that may offer
important clinical insights
that can potentially benefit
the healthcare
delivery system
16. A Patient Perspective:
Patient-centric social media for
outcomes and pharmacovigilance
considerations
Julie Flygare, JD
Founder, Project Sleep
@REMRunner
Pre-diagnosis, post-diagnosis and moments of crisis or changes later on that get one looking for answers, other approaches.
Huge sense of connection (often unexpected benefit) and also sharing invaluable practical information.
As a patient, often feel like no one is listening. For me, I think I might actually share more publicly if I thought someone was listening. “Would you share your experiences publicly IF you thought someone was actually listening?”
Note that social media is not necessarily the "real world”. “Oversharing” or not providing complete story, people in moment of crisis are not always the most articulate in sharing their circumstances, can be hard to fully know what’s going on for someone else. Shouldn't be confused or used in place of consulting patient experts and caregivers directly through panels, patient advisory roles.
Someday could the internet facilitate more REAL-TIME important conversations between different stakeholders in healthcare? If regulations allowed it, could information be more accessible from industry and insurers? Example of “Ask TSA” demystifying rules of TSA.