By Erin Drucker, Lead Data Scientist, Cardinal Innovations Healthcare
Putting data to work to ensure members receive behavioral healthcare when it’s needed most.
The goals of value-based care — better population health and member experience at lower cost — are especially relevant for complex Medicaid groups. Yet making a positive impact on the health of our most complex member populations requires innovative new strategies and care models. At Cardinal Innovations Healthcare, that begins by addressing members’ needs using an approach that is uniquely data-driven, person-centric and community-based.
For example, a newer initiative called Project RADAR- Antipsychotic Medication derives value from claims data by integrating key behavioral health indicators into pharmacy workflows. It starts with the Cardinal Innovations data science team working with clinical pharmacists to pinpoint members at risk. Then, upon reviewing the list of members in order of risk for crisis, pharmacists direct outreach to help surround members with appropriate community services.
Information And Insight Spurs Action
Through data analysis, the Project RADAR pharmacists proactively intervene on behalf of members who are more likely to be in danger of crisis. The program relies on behavioral, pharmaceutical and physical health claims information to stratify risk and prevent an individual from experiencing a crisis.
Focusing on members with a mental health diagnosis of schizophrenia, bipolar, or major depression disorder, the data science team reviews and analyzes:
- medication history and usage, including antipsychotic fills, type (including Clozapine usage in last 90 days and long-acting injectable medications), and drug route, as well as patterns that show a lapse in usage of antipsychotic medications
- service utilization patterns that show whether they’ve been involved in high-intensity services or indicate a lack of adherence to care plans or whether they have been involved in crisis prevention services and supports
- history of previous crisis, including suicide attempts, suicide ideation, or self-harm or any self-violence with undetermined intent, as well as Emergency Department, Inpatient Treatment or Mobile Crisis usage
- age (since younger members are more at risk for crisis) and other demographic information
- other physical comorbidities
Behavioral claims information is organized in a data warehouse, where additional data is added each month from Medicaid physical health and pharmacy claims. Then, using SAS, our statistical analysis software, the data science team completes data aggregation, cleaning and analysis using a series of data mining and machine learning techniques. Custom algorithms use factor analysis, ROC curves and a series of logistic regression models to generate a stratified risk score, grouping members in high-, medium- or low-risk groups with corresponding tracked interventions.
By digging deep into data and working with our community partners, our care teams glean insights that help ensure meaningful member engagement.
Creating More Meaningful Interventions
Pharmacists have a number of options to intervene and potentially prevent crisis. They can submit a referral to a care coordinator or, if a member already has a care coordinator, contact that care coordinator directly. They can send the member’s physician a medication review identifying potential medication interactions, pinpointing where problems may be occurring. They can call the member, prescribing physicians or legally responsible persons to see if a barrier exists that is preventing the member from obtaining needed medications or complying with the care plan.
If necessary, pharmacists can even deploy a mobile crisis team. In one especially memorable case, a behavioral health pharmacist called a member who admitted he was just about to end his life. Fortunately, the pharmacist was able to immediately marshal the right resources to help the man and prevent the crisis from escalating.
While such examples prove this program is on the right track, even better is when pharmacists can look for members who may be increasing in risk and reach out to keep them from reaching crisis altogether. That kind of proactive strategy, aimed at keeping members healthier while reducing the utilization of high-cost services, is the hallmark of effective value-based care.
What This Means For Managed Care
As Cardinal Innovations Data Sciences first began developing the parameters for the pharmacists’ reports, thousands of members initially appeared to be at high risk. Of course, with limited resources, it was best to focus the reports on the most at-risk members, as well as those for whom the data science team could provide the appropriate resources.
Getting there required several attempts. However, using different modeling techniques, the team eventually landed on a report that helps pharmacists identify the members considered at high risk of harm. Over time, the data science team has increased the accuracy of the reports even further based on a larger set of historical data and projections. That includes understanding medication usage patterns for other drugs used by high-risk groups as well—such as opioid or anti-depressant use patterns, for example. Each iteration with the medical and pharmacy departments increases the ability to measure true positives.
In tandem with this data analysis, Cardinal Innovations is also growing its array of targeted interventions for this member group, and will adjust over time based on predictive analytics and outcomes. Key to the program’s growth will be adding more data types as they become available, including social determinants of health.
Most importantly, Cardinal Innovations already clearly sees that the analysis of these data sets help to paint a picture of their members — not only when they are currently at risk of harm, but also when they are at high risk with only a limited time for an intervention to occur. The potential for this kind of analysis has broad-reaching implications for any organization that manages similar member populations. Putting data to use to drive collaboration and create positive change for complex populations is an important part of a value-based care program that can have lasting influence.
Today, Cardinal Innovations is impacting the lives of many members living with mental health conditions. Project RADAR has improved the ability to get the right services to members at the right time, surrounding them with the kind of meaningful interventions and person-centric care that really makes a difference.