A new study comparing physician-recorded data regarding adverse drug reactions (ADRs) and social data generated by patients and caregivers is surprising clinical experts. Published in the latest edition of Drug Safety (http://bit.ly/1QoBFxL), the study portrays the increasingly central role social data will play in personalizing treatments, optimizing care, and impacting outcomes.
The aim of the study was to validate social data by evaluating it side-by-side with clinical data stored in electronic health records (EHRs). The results confirmed substantial overlap between the two data sets, as well as the existence on social media of ADRs that are understated in clinical data. The latter phenomenon is further proof that social data and clinical data are both needed to paint a complete picture of an individual’s health profile.
Combining Data Efforts
The study focused on ADRs related to aspirin and Lipitor, chosen because of their high prevalence in both data sets. The EHR database consisted of allergy records shared across a system of hospitals and outpatient clinics, and the social data was collected from online health forums and social media platforms.
In both the clinical and the social data, the most commonly reported ADRs for aspirin were hives and gastrointestinal upset. Patients on social websites, however, referenced tinnitus and a rare condition called Reye’s syndrome much more frequently than in the EHRs. According to the researchers, “Aspirin studies support the social media data, with reports of up to 30% of patients experiencing hearing problems with long-term aspirin use, while our clinical data showed only low tinnitus prevalence.” Conversely, Reye’s syndrome occurs very rarely in the general population. More research should be done to determine whether the appearance of Reye’s syndrome among patients’ top concerns was an anomaly.
Regarding Lipitor, two of the top three concerns (musculoskeletal pain and cramps) in the social data and the allergy records were a match. As with aspirin, other adverse events related to Lipitor, namely fatigue and depression, were more highly represented in the social data than the clinical data. That depression ranked so high among concerns on social websites can potentially be attributed to the reality that some patients feel uncomfortable discussing mental health issues with their physicians.
The Case for Scientific Application
The similarities and discrepancies between physician-reported ADRs and patient concerns described here strengthen the case for the scientific application of social data analytics. This is a preliminary step in deepening the research on online patient discussions and more fully understanding the patient journey in the real world, outside the regimented confines of a clinical environment. Drug adherence, for example, is influenced both by how a drug performs and how patients perceive its effects. Social data allows researchers and clinicians to uncover those factors influencing patient behavior that cannot be captured by other sources.
Along with the ability to identify previously unknown ADRs, pharma companies are increasingly interested in finding ways to predict adverse events among populations that would not meet the standard criteria for clinical trials, which typically do not have a broad enough scope to encompass all of a drug’s possible side effects. With advances in social analytics, the realm of drug safety will experience a veritable revolution by the end of this decade. More effective drugs—and healthier patients—lie on the horizon.
