Medication discontinuation and non-adherence are big problems for both pharmaceutical manufacturers and patients. Annually, pharmaceutical companies lose an estimated $250 billion due to medication non-adherence and premature discontinuation. Approximately 50% of patients stop taking their medication for their chronic condition within the first year — with an estimated 125,000 U.S. deaths each year attributed to medication non-adherence.
To improve outcomes, reduce healthcare costs, and improve commercial success of medications, a better understanding of why patients stop taking and providers stop prescribing a particular medication is needed. Currently, manufacturers primarily use claims data to analyze who is or isn’t taking a medication and focus groups with either patients or prescribing physicians to understand the drivers as to why.
While focus groups can provide valuable direct feedback, they also present challenges and limitations. A reliance on small sample sizes with significant recall bias limits their value. For example, a provider who recently had an immunologic patient discontinue a TNF-alpha inhibitor is more likely to talk about the recency of that stoppage in a focus group because it is most top of mind. While offering some insight into discontinuation, that example may also be telling a different story than other similar patients and certainly doesn’t get at why patients with the same disease but different characteristics or disease severity may be continuing or discontinuing the same treatment.
Claims data provide their own set of challenges. Getting to that “why” is much larger than just the medication claim and ICD 10 code. Claims data provides information in the aggregate of who is prescribing a medication for a particular condition, but it doesn’t describe the patient who may or may not continue on a medication. Clinical data, such as from physician notes in electronic medical records (EMRs), provide an opportunity to dig deeper into the story behind discontinuation.
Large datasets that pull in clinical data, such as outcomes and physician notes from electronic health records, reveal more than just the volume of patients who are on a particular therapy and prescribing habits. With the appropriate application of medical language processing and text extraction, these records can show the clinical narrative at scale and offer more precise reasons why patients are continuing, discontinuing, or switching a medication.
Claims data may show patient X stopped taking the medication, but the clinical narrative will show that a patient stopped because of injection site pain, weight gain, headache or an upper respiratory tract infection, for example. Clinical narratives can also reveal patterns behind how patient journeys, characteristics and disease severity all combine to influence patients’ continuation or discontinuation of a treatment. For instance, how patients with different ethnicities respond to the same medication or showing how insurance or medication price is among the primary reasons for drop-off. A computerized dataset of hundreds of thousands of patients inclusive of a clinical narrative is much more valuable for identifying trends than a five-physician focus group or even a 1,000-case extraction from notes.
Understanding specifics in discontinuation among certain patient populations, the drivers behind why physicians decide to prescribe or switch a patient’s medication, and how a certain drug may be performing against its competitor, all provide actionable opportunities to improve adherence and reduce discontinuation. With this type of data analysis, manufacturers have the opportunity to address issues with their own medication or to leverage insights on a competitive treatment.
The end result is more informative dialogues with stakeholders about why one medication should be prescribed over another for certain patients. Physicians have the opportunity to translate those insights into evidence-based clinical decision support. The goal is to help retain patients on medications for the long term, supporting both positive outcomes and minimal side effects, and ultimately direct the right patient, to the right medication, at the right time.
> View a webinar recording with Drs. Marci and Weiss on using real-world data and analytics to understand medication discontinuation, comparative outcomes and prescribing trends (http://om1-2628076.hs-sites.com/rwa). Learn more about OM1 Real-World Analytics at OM1.com.
