I want to illustrate my point by discussing the story of a CEO of a Medicare Advantage Plan with 10,000 members. This CEO is well known in the industry for his forward-thinking vision, capacity for risk-taking, and commitment to innovation. He has built an exceptional care management team to reach out to rising-risk members proactively and reduce avoidable emergent events. As such, his organization has been rewarded by the market with year-over-year organic growth and a 10% decrease in medical loss ratio over the last three years.

This year, the CEO set out to further reduce their overall costs by 6% relative to last year. He does not believe in beating the market; he wants to beat their own internal cost performance. The operations team under the leadership of their COO took up the challenge and set out the develop the plan to accomplish this goal.

Their Total Medicare Advantage spend in 2017 was $100 MM. A 6% reduction translated to $6 MM savings to be accomplished in 2018 with no risk adjustment. Upon analyzing the data along with the operations team, VitreosHealth found 20% of members made up 85% of total costs, which was a consistent trend over the last three years. VitreosHealth team also found that 10% (1,000) of the MA population contributed to 85% of Avoidable ER costs. Avoidable ER costs were $15MM for 2017.

If the operations team had access to predictive models with 100% predictive accuracy (yellow line in the figure below), to target $6 MM of avoidable ER costs, the care management team must proactively reach out and avoid ER events for 350 members. Adjusting for member effectiveness of 2:1, that is if we reach out to 2 members, we can have an impact on 1 member, target outreach needs to be 700 members. On average, a care manager can consistently reach out and actively manage 250 chronic care members annually. So, we need 3 care managers to target this rising-risk population with a predictive model with 100% accuracy (“The Divine Model”).

By using CMS-HCC risk scores as a tool for risk stratification (blue line in the figure below), we found that the care management team must proactively reach out to 4,200 members to achieve the 6% cost reduction. They would need 17 care managers to target this rising-risk population.

However, by using the 2017 VitreosHealth AI-based predictive models, we found the care management team must proactively reach out to only 1,500 members. This means that they need approximately 6 care managers to target this high-risk population.

This translates to $1.1 MM savings in people costs to accomplish the same operational goals of $6 million in avoidable emergent event costs.
Exhibit 3: Care management productivity is directly tied to predictive accuracy. In this graph, I am trying to illustrate how four different models with different predictive accuracies can give you different results. Suppose you only have the resources to target 20% of your population for proactive outreach. You would want to choose a model that helps you pick the top 20% of members that are most at risk for avoidable emergent visits. If you choose which members to focus outreach efforts on by using CMS HCC, you will most likely only prevent 15% of avoidable ER events, while if you used Vitreos Regression models, you could prevent around 38% of emergent events. Our new AI models will give you the ability to prevent 65% of ER events. A perfect model (powered by whatever deity you choose to believe in) will probably help you prevent 85% of emergent visits (in our experience, the top 20% of members result in 85% of avoidable cost).
How do you measure the ROI of your population health AI-driven predictive models?