Predictive Modeling and Benefit Plan Changes

Predictive Modeling and Benefit Plan Changes
BRIEFING
September 14, 2017
Vice President, Enterprise Analytics, CVS Health

Payors today need ways to improve benefit management and reduce health care costs. Yet many hesitate to take action due to concerns about how plan design changes may impact plan member satisfaction. A proactive member engagement strategy can ease transitions, help members make cost-effective choices, and maintain their prescription therapy.

At CVS Health, we use predictive modeling to help facilitate member transitions when a plan is undergoing formulary or network changes. We develop algorithms that will help us predict future behavior by looking at members’ past behavior. Over the years, we’ve engaged with millions of members going through plan changes. Data on how they’ve responded is the foundation of our model to predict how similar members will respond to a potential change.

Identifying Members Most Likely to Change

We examine more than 500 baseline characteristic variables to identify which are most predictive of member behavior. Many of these characteristics are ones you’d expect — age, gender, and number and type of prescriptions. In addition, prescription history helps us identify possible comorbidities and tells us about the member’s current adherence. To these, we add other data points such as response to previous benefit plan changes, whether a member is a registered web user, and how many prescriptions a member has in various channels. ZIP code and census data can also help identify probable income and education levels.

Analyzing this data yields insights on the most predictive factors about a member’s response to change. For example, a plan may want to encourage members to move from a 30-day to a 90-day prescription or from a retail channel to mail service pharmacy. Such changes can help reduce costs and improve adherence. Our predictive models have helped us identify the key predictors for how members will respond to such a change. The algorithms work to evaluate the a member's propensity to make a change by scoring them on predictors such as:

Eldery

Age

Each additional year increases the likelihood of change.

Gender

Gender

Overall, men have a lower likelihood of making such changes than women.

Smart Phone

Use of online tools

Website registration increases the likelihood.

Pill Bottles

Comorbidities

The number of conditions being treated and the number of prescriptions filled in the previous month affect the likelihood of change.

Pharmacy

Rxs at retail

Beyond a certain point, each additional prescription currently at retail decreases the likelihood of a change to mail service.

This knowledge helps us use resources wisely and focus outreach on those members most likely to make a change. Data also helps us predict the best channel to communicate with a member about the option, thereby increasing the likelihood of a positive response to the plan sponsor’s desired change.

Predicting a Possibly Problematic Response

When there is a formulary or network change, some members will be more affected than others. Most members adjust to the change with few, if any, problems, but some may benefit from extra support.

By reviewing prescription claims, we can identify which members will need to transition to an alternate prescription therapy. Using data on changes implemented by other clients, a member’s baseline characteristics, and how they have responded to change in the past, we can also predict which members are most likely to benefit from extra support. Our analysts can identify those members who are likely to:

Checkmark

Try to fill at an out-of-network pharmacy, resulting in a rejected claim

Checkmark

Call into Customer Care

Checkmark

Discontinue therapy

Tiered Communication Strategy

To address each member’s unique needs and potential challenges, we have developed a tiered communication strategy. Typically, all members receive a notification 30 to 45 days in advance of the change. Those who will need to take additional action, such as getting a new prescription from their doctor or moving their prescription to a new pharmacy, receive a more detailed communication.

A second pre-implementation communication targets those at higher risk for a negative response. Depending on the level of risk, the second wave could involve automated interactive voice response or live outbound calls. Once the change is implemented, we monitor members for at least six months. If a member has a claim rejected or drops off therapy, we intervene with communications designed to help get them back on track.

Implementation Chart
Implementation Chart

Every member’s journey is unique, and our communications and member engagement programs emphasize personalized outreach. Our predictive modeling tools help us understand and anticipate each member’s journey so that we can offer the appropriate support in a transition, helping the member and the plan make the most of their pharmacy benefit.

Do you have any questions about using predictive modeling to facilitate formulary or network transitions? Ask Us
BRIEFING
September 14, 2017
Vice President, Enterprise Analytics, CVS Health

Explore Programs

Benefit Plan Changes

Intelligent targeting and tailored communications about formulary or pharmacy network changes can ease the transition for affected members.

Formulary Management

A comprehensive formulary strategy is foundational to mitigating the impact of escalating drug prices, and the introduction of new high-cost therapies.

Network Strategies

Managed networks can help reduce costs as well as help improve clinical outcomes by supporting health and wellness initiatives.

CVS Health uses and shares data as allowed by applicable law, our agreements and our information firewall.