At Vitreos, we know there is a lot of opportunity in reducing costs in multiple areas of healthcare spend. We have experience in identifying ‘Rising Risk’ members for avoidable emergent events for primary care, oncology-related post-chemo avoidable ER visits, and medical adherence among other areas. In fact, most of the healthcare market has been discussing cost avoidance in these same areas. However, our research indicates that behavioral health accounts for more than $200 billion in annual spend nationally (find more info on that here). That number astounded us, so we decided to partner with some of our clients and take a hard look at the impact of mental and behavioral health conditions on their population. We specifically wanted to see if there were significant behavioral health cost avoidance opportunities for their exchange populations.

Every new client relationship begins with the standard Vitreos™ State of Health Analysis (SOHA) to identify the ‘Rising Risk’ members for each population. We then group members with similar clinical characteristics, social determinants, and behavioral risk into cohorts (as seen in Figure 1). For this specific health exchange population, we performed the SOHA with multiple years of historical data. We call groups of people that move from the left (low-cost members) to the right (high-cost members) the ‘Rising Risk’ cohorts. 

Figure 1: Members are mapped onto this grid according to their cost (x-axis) and clinical risk (y-axis). Members with similar clinical and non-clinical characteristics are grouped into cohorts (denoted here by white bubbles). Some cohorts move from year to year depending on the health and cost of the members of that cohort. These sorts of movements become very clear after a SOHA.

Our objective is to understand the characteristics of these ‘Rising Risk’ cohorts and leverage AI-driven predictive models to identify who is most at risk and intervene with proactive care management and member engagement programs.

Upon analyzing these exchange population through our Vitreos™ predictive and prescriptive analytics framework, our data scientists found that, on average, 35% – 38% of people in exchange populations had behavioral health conditions. However, these people contribute to almost 50% of total spend for the exchange population. That’s how we realized that the cost burden lies more heavily on those that have behavioral health conditions. After this realization, our data scientists started exploring where this additional cost came from. They found that members with behavioral health conditions had 34% more ER visits and 60% more ER admissions as compared to the rest of the population.

We then broke down the ‘Rising Risk’ members into two applicable cohorts: those that move from Hidden to Critical (Cohort A) and those that move from Healthy to High Utilizers (Cohort B). We leveraged our AI-driven predictive models to provide prescriptive insights into these two cohorts (see Figure 2).

Figure 2: Two cohorts arise when assessing how many ‘Rising Risk’ members have behavioral health conditions. Cohort A comprises of members that are typically higher in chronic risk conditions than Cohort B. Cohort A has members that predominantly suffer from depression, while Cohort B has members that predominantly suffer from anxiety.

Cohort A is made up of sicker and older people with depression. Because of mental health conditions, they are typically non-compliant to best evidence-based care practices and don’t have a record of strong medical adherence to drugs. There is also a prevalence of opioid abuse related to pain management in these members.

Cohort B are younger people with high socioeconomic, access-to-care, and well-being risks. They struggle with social wellbeing, obesity, hypertension, along with anxiety. The members in this cohort tend to have high-risk addictive and substance abuse behaviors including recreational substance abuse.

Many of your ‘Rising Risk’ members are struggling with mental and behavioral health conditions. Understanding these two cohorts’ clinical chronic conditions coupled with mental illness and behavioral risks can help design high-touch care management and member engagement programs. The reason that most current population health programs are not effective is because of a lack of predictive and prescriptive analytics to drive decision-making.

Predictive Insights from a State of Mental Health Analysis of a Typical Exchange Population

Leave a Reply

Your email address will not be published. Required fields are marked *