RWE’s new dataset: Social media
Pharma is waking up to the potential of social media to enhance existing RWD sources and to offer altogether new insights in rare disease and beyond
Social media is a vast dataset just waiting for RWE teams to mine for a wealth of insights. The pioneers are uncovering granular insights that can massively improve on crudely aggregated patient experience data gathered by other means.
What was very much a niche activity three years ago is becoming more commonplace. In the face of pandemic lockdowns RWE teams, who were unable to conduct primary research interview surveys, turned to social media data, which helped boost interest in social data.
Now they are increasingly publishing the fruits of their labours in peer-reviewed medical journals and sharing insights into everything aspect of the patient experience of disease from symptoms to diagnosis and post-treatment.
Why is this such a rich source of insight? On social media, patients bare their souls, revealing their emotional journey, and sharing in minute detail what day-to-day life is like as they manage their disease and treatment.
This offers pharma the chance to cultivate a wholly new source of valuable data that affords new patient-focused insights that have been hard or impossible to gather prior to the advent of social media.
Getting to ‘why?’
RWD from social media can capture the emotional journey of patients, which can complement RWD from health records, disease registries and so on, says Sathyaraj Aasaithambi, Group Leader for Social Media Listening and Analytics at Novartis Healthcare.
This data can fill knowledge gaps about patient reasoning when it comes to staying with or switching from a treatment option and deeper insight into treatment delays and their impacts, as well as new insights into adverse events. “This is where you are able to understand different parameters not being uncovered in the clinical data,” says Aasaithambi.
Social data offers a powerful complement to and enhancement of existing RWD, says Steve Reeves, SVP, Digital Strategy, Ipsos Healthcare.
Established sources of RWE including claims or health records tell the transactional story of the patient and their healthcare system interactions from which it is possible to derive insights, but it leaves the ‘why’ part blank, the “relational part” says Reeves. “Why do patients switch or stay on a treatment? Why do different patient types have differing unmet needs? Why is there variance in quality of life across therapies?” These are all questions that can be addressed using social data, says Reeves.
The use of social data in pharma to complement RWE data sources goes beyond the patient journey, he adds. “Through advanced text analytics and machine learning we are now able to see signals in social data that sometimes become trends in RWE data. Pharma is using social for early warning detection, especially when monitoring market events like drug launches.”
The enormous volumes of social media data contain signals which can be teased out with advanced text analytics and theme detection, enabling insights to emerge ‘organically’. Alternatively, the dataset can be prompted. Both are valuable modalities. “We can ask the data questions, or we can use advanced text analytics to help understand the “uknown unknowns,” says Reeves.
Being able to mine historic data going back years from dedicated social media created by patients and caregivers is another valuable feature of social data, says Abhinab Bhanja, Head of Insights APAC and Head of Pharma at Convosphere.
And the speed with which it’s possible to gain insight is another appealing feature, says Aasaithambi. “It can be assessed in a short time frame. That is the beauty of social media data, as long as you have a strong taxonomy and strong data sources.”
New insights in rare disease
Some of the most interesting uses of social data as RWD have been in rare diseases, where finding and serving patient populations has always been a challenge. They are small in number, may not yet have been diagnosed, are geographically dispersed and may not know there are clinical trials or therapies out there that could serve them. Social listening offers a useful tool to identify, map, and reach them.
Research into rare disease symptoms, which are often assumed to indicate other more common conditions, can offer highly valuable and specific insights into the diagnostic journey. “We can help understand what patients are experiencing when they are reaching out to HCPs, says Bhanja. “We are able to get a richness and volume from social media that’s not possible in patient interviews. You can get a good volume of data quickly and cost effectively too. That can’t be done through research.”
There is enormous scope to impact rare disease treatment as well as the clinical trials process here. If, for example, patients have friable veins and you know you can’t do daily blood tests or you know they are far from trial centres, it’s possibly to adapt accordingly, says Bhanja. “We think about the amazing treatments we have developed but we may not have thought about the implications of how they meet the needs of patients or how it impacts clinical trials.”
There is also scope to work with patient advocacy groups to help them communicate more effectively with their patients and to help HCPs serve previously unmet needs in terms of information or support.
Novartis has been plotting patient journeys in a range of rare diseases from symptoms to diagnosis to treatment. Forums are insight-rich places to mine for insights into extremely rare and often hard-to-diagnose diseases, says Aasaithambi. “These are places where they are connecting and discussing lots of information despite the rarity of the condition.”
In the case of one hard-to-identify rare autoimmune disease, Immunoglobulin A nephropathy (IgAN), specifics of the patient and diagnostic journey were poorly understood. There was little existing literature and the condition was easy to confuse with other kidney conditions. This made it hard to identify sufferers and reach them in time as the disease often rapidly develops to end stage.
Novartis gathered data from social media, used a clinical taxonomy to categorise the data, cleaned and parsed the data using natural language processing, and was thus able to collect useful volumes of data by patient, caregiver, HCP and region.
These insights enabled Novartis to plot the patient journey in fine detail, segmenting them based on the stage of disease and creating different personas. It gained far more detail on where sufferers were, what age they were and how active caregivers were. A key insight was how often the disease can progress in an individual in their 20s from stage 2 to end-stage renal disease.
The study also gleaned other important insights, such as the terms patients use for the colour of their urine and the wide range of symptoms they experience. This included the way in which the disease progresses. For example, patients reported pain and fatigue, something hardly reported in the literature. Cross checking this with HCPs revealed that these complaints were coming up in consultations.
Quality of life impacts were another valuable dimension, offering new insights into the toll the disease takes on physical ability, emotional state and the burdens on family life and caregivers.
These sorts of insights can be used to change the questions HCPs ask, and to seed fruitful discussions with advocacy groups.
With often scant peer-reviewed literature, little EHR data and hard-to-reach patient groups, rare disease social media offers a goldmine of insights, says Reeves. “Rare disease communities are hyper-active online.”
Executing on the promise of social RWD
It’s an enormous, if messy, dataset but pharma is fast developing the capacity to wrestle the signal from the noise from social data and so do even more good work with it.
Certain key capabilities are needed to create concrete, actionable insights when mining it, says Aasaithambi. These include an interdisciplinary team made up of RWE expertise, epidemiologists, social media expertise, business unit colleagues, as well as machine learning and data science talent.
There’s a strong rationale for developing such strong internal capabilitiers as handing off insight generation to agencies can easily go wrong, says Aasaithambi. “Many agencies say they can do it, but you have to tie all these insights into actions. What is important? What should you do when you find this information? Can you connect these dots to HCA discussions, pricing discussions and market access discussions?
“That is the tricky part and that is why you need an interdisciplinary team not just to analyse this unstructured data and turn it into meaningful insights but to know what action to derive from it.”
Research starts with the key business questions or hypothesis and being clear as to how social media data fits in. The methodology will also need to change based on the nature of the question being asked, adds Aasaithambi.
With the subtleties of the art and science of mining social data mastered, the sky is the limit for RWE teams. “Social media is adding a lot of value to RWE, it has to be taken into consideration,” says Aasaithambi. “This is a new era for RWE.”