Using Advanced Analytics to Smooth Member Transitions

Advances in data science are spurring change in every industry, including health care, and the pace of change is accelerating. Some of the fastest growing applications are in the areas of artificial intelligence (AI) and machine learning.

AI refers to technology that mimics cognitive functions typically associated with humans such as learning and problem solving. Machines “learn” when they recognize patterns and adjust behavior and improve performance accordingly.

CVS Health has a long-standing commitment to analytic research, and we have invested in the teams and resources to fuel in-depth analysis, and develop and deploy programs and initiatives based on analytic insights. Most recently, we’ve accelerated this initiative by bringing together a team of experienced data scientists across CVS Health to explore and develop new solutions that use AI and machine learning to improve efficiencies, enhance service, and favorably impact outcomes for plans and their members as well as patients and customers of our retail and Specialty pharmacies, MinuteClinic locations and other businesses.

Seamless Transitions

One important area of focus for the team is pharmacy benefit changes such as transitioning to a new formulary or pharmacy network or becoming a new CVS Caremark member. Even though only a very small percentage of members experience disruption in most transitions, payors are often concerned about the impact on their members. Among the millions of members making a transition on January 1 of this year, careful planning and communications helped ensure the change was seamless for  close to 99 percent.

How Analytics Provides Value

It helps identify patterns

Predict what’s likely to happen

Prescribe what to do

To better support the small number of members who may experience disruption, we used machine learning to identify those most likely to find the transition challenging. We evaluated data from previous transitions to identify factors that characterize high-risk members, baseline knowledge about a member population, and an individual member’s prescription history, and response to previous plan changes. We used these insights to develop models to predict risk of an adverse response, and these models have proven highly accurate. After a formulary change, members predicted to have a high probability of dropping off therapy were up to 10 times more likely to do so compared to members with low probability. During a formulary or network transition, we focus on helping these high-risk members transition successfully, and introduce solutions that can reduce risk, such as 90-day prescriptions. In addition to being more convenient, 90-day prescriptions reduce the possibility of a member dropping off therapy, thereby also improving adherence.

After implementing an enhanced communication program for high-risk members during a transition to a new pharmacy network, we saw an 11-percent reduction in out-of-network claims and a 6 percent improvement in early adherence indicators. Similarly, after a formulary change, there was a 3 percent improvement in early adherence indicators.

Welcoming New Members

For a payor, a transition to a new PBM can seem especially challenging with possibly hundreds of thousands of new members and prescriptions to process. A successful onboarding experience depends on accurate benefit set-up and adjudication of new claims. If a member experiences difficulty or delay in receiving a prescription it can have a negative impact on their satisfaction with the PBM and the plan itself.

For the 2018 benefit year onboarding, our data scientists partnered with the PBM’s client services team and built a claims surveillance platform that incorporates machine learning to identify outliers and potential issues enabling faster corrective action to reduce member impact. This claims surveillance detected issues such as formulary or prior authorization mismatches and copay errors in near real-time. By combining faster detection of issues with a 41 percent reduction in the mean time to resolve these issues we helped 99.82 percent of members make a seamless transition, without experiencing any disruption. We are further expanding our system capabilities to include a broader array of potential claims errors and enhanced reporting functionality for even greater performance improvement in the future.

Expanding the Use of AI, Machine Learning

We’re applying advanced analytics across CVS Health to further improve efficiencies and enhance experience and outcomes for members and patients.

	We use data science to improve adherence among specialty pharmacy patients, customers at our retail pharmacies, and Medicare beneficiaries.
	At MinuteClinic, machine learning is enabling us to optimize scheduling with better visit forecasting. The goal is to minimize wait times and improve our already-high patient satisfaction.
	For CVS Pharmacy patients and PBM members, we personalize outreach by member preference and receptivity to behavior change, and use machine learning to customize interventions about the individual’s “next best action.”

Advanced analytics are already producing real-world results for our members and payor clients. As we continue to expand these capabilities, we anticipate even greater personalization, higher satisfaction and a more seamless experience, as well as better outcomes and lower costs.