Like many pharmacy benefit management companies, OmedaRx has relied on population health analytics to help it ensure patients are taking the medications they need. But often, the information gleaned from the software is too general. Now, the company is implementing a new cloud-based system that uses big data, analytics and machine learning to create more precise, cost-effective care management plans aimed at producing better patient outcomes.
OmedaRx is the pharmacy benefit management company of Cambia Health Solutions, parent company of Regence Blue Cross Blue Shield, and manages the prescription plans to Regence health plan members in Oregon, Washington, Utah and Idaho.
A year ago, OmedaRx began piloting Max for Medication Adherence from analytics provider GNS Healthcare, a company in which Cambia Health Solutions is a primary investor. The system uses big data to identify individuals at risk of costly drug-related events because they aren’t adhering to the recommended timing and dosage of their medications.
The results to date have been impressive and this month OmedaRx is beginning a full-scale evaluation of the platform’s effectiveness, phasing in 500 patients every month for six months. The plan is to ultimately include OmedaRx’s entire Medicare population of about 120,000.
Poor medication adherence is relatively common, according to the U.S. Department of Health and Human Services’ Agency for Healthcare Research and Quality. Studies have shown that 20 to 30 percent of medication prescriptions are never filled and that, on average, 50 percent of medications for chronic disease are not taken as prescribed. Poor medication adherence drives $290 billion in avoidable health care costs, or about 13 percent of total U.S. healthcare expenditures, according to the New England Healthcare Institute, an independent, not-for-profit organization.
Medication management has evolved from intuition-based treatments to population health, which emerged in the mid-1990s with a top-down, rules-based approach that treats individuals as hypotheticals, says Colin Hill, chairman and CEO of GNS Healthcare.
But traditional methods to determine which patients to reach out to don’t target people who are at high risk of negative health consequences. For example, OmedaRx might now determine that 5,000 people need intervention. With big data and more advanced analytics, it’s possible to determine that only 700 are really at risk.
“If we can increase medical adherence in a high-risk patient, we expect that will reduce the negative outcomes and costs associated with that patient,” says OmedaRx VP Jim Carlson, who is a pharmacist by training. “When we are able to, by intervention, reduce the risk of negative health outcomes, this is the Holy Grail.”
The growth, and availability, of data is key to creating analytics applications that can refine and advance medication adherence programs. Another key is machine learning, which advances the type of questions that can be asked, and answered. Consider analysis that determines, in a specific population, that the cost to treat a patient’s diabetes has increased 50 percent over the last year. The next questions are why, and what are the correlations? After that, forecasting needs to be done to determine what happens if that trend continues.
With machine learning, an inference of cause and effect, known as causal inference, can be determined: What happens to the cost if a particular course of action is taken? What happens if another course is taken? This lets companies consider many different future paths and pick the best one. “Big Data is really the breakthrough,” says Hill. “We now have large enough and rich enough data to power up causal inference and decision optimization.”
OmedaRx began testing Max for Medication Adherence in a pilot with about 100 patients.
The program starts with OmedaRx sharing electronic feeds of its pharmacy claims data with GNS Healthcare, which combines it with electronic feeds of pharmacy and medical claims data from Cambia (and its insurance companies such as Regence BCBS) as well as consumer and demographics from third-party vendors including information information clearinghouses. Much of the data OmedaRx sends to GNS Healthcare is structured, but GNS Healthcare also can work with unstructured data.
The data is fed into GNS Healthcare’s Max Solution Platform using a number of coding applications capable of handling the unique features of individual data sets. From there, the mixed data is used to develop computational models.
The data sets and analytics platform are housed in a GNS Healthcare data warehouse, MeasureBase. Analysts can query the data warehouse using GNS Healthcare’s Measure Language, which uses its own straight-forward language that the company says makes it easy to specify what they want to measure, on which people, over what time periods in simple terms – without writing SQL or database queries. The data warehouse is housed in a highly secure virtual private cloud operated by GNS Healthcare. The cloud employs a massively parallel cloud-based architecture made of multiple web-based servers, according to GNS Healthcare. All services have been approved under the Federal Risk and Authorization Management Program (FedRAMP) for Federal Information Security Management Act of 2002 (FISMA).
GNS Healthcare’s platform can synthesize trillions of data points coming from claims history, electronic medical records, socioeconomic and geographic data, consumer behavior data, genomics data, bioinformatics data and more, the company says.
The data is run through GNS Healthcare’s patented Reverse Engineering and Forward Simulation (REFS) machine learning and simulation engine within the Max Solution Platform, which analyzes and models the data sets as multidimensional observations about people over time. The engine learns by reverse-engineering collections of models and then simulates representations to generate predictions, including risks of negative outcomes such as adverse events. It also quantifies the effect of behavior changes, stratifies populations based on individuals’ likelihood to engage, measures the power of interventions to change behavior and matches individuals to the most cost-efficient and effective intervention.
GNS Healthcare routinely generates analytics reports and models for OmedaRx. The data can be presented in dashboards as well as in stratified lists of individuals and can include risks, efficacy that quantifies risk reduction resulting from behavior and clinical behavior changes.
Another reporting tool is the MAX for Medication Adherence Planner that allows for the easy planning of intervention portfolios and for reviews of projected population-level impact. GNS Healthcare’s analytics can compute and return each member’s ‘best’ intervention (i.e. the one that maximizes their individualized impact) and projected ROI.
The final product is a GNS IEScore that identifies people not only at high risk of an adverse event, but also those high-risk patients that are likely to respond positively to an intervention, reports that quantify how interventions drive changes in behavior and ROI which determines the best intervention match for each individual.
OmedaRx’s care management team uses the data to create daily, weekly and monthly outreach programs that can include phone calls, in-person consultations, emails, etc., to talk with the individuals, better understand their concerns, and make recommendations about their prescriptions so they can improve adherence. Pharmacists do the outreach.
“Medication adherence has been an issue for a long time,” says OmedaRx’s Carlson. There are a variety of reasons for this – personal beliefs, distrust in the healthcare system or in pharmaceuticals, the patient, or a relative or friend, has had a negative outcome, the patient is confused about the medication, the patient doesn’t like the side effects, or the medication is too expensive. When designing outreach programs that involve intervening with patients and include talking with patients and trying to understand and address the barriers to taking their medications, it is important to understand the patient’s perspective, he says.
“Using the big data analysis, we can consider all of the medical and pharmacy claims, the history of office visits with physicians, the ER visits, hospitalizations, medications prescribed and medication adherence and our own claims data, and per patient, predict what their risk is of being admitted into the hospital or visiting the ER,” Carlson says. “Out of those people, we can better determine which ones are not taking medications as prescribed.”
From there, it’s all about fine-tuning the outreach so interventions can be tailored by the pharmacist to increase medication adherence that ultimately lower healthcare costs and improve patients’ health.
Register or login for access to this item and much more
All Health Data Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access