With widespread industry and government support, population health initiatives continue to gain traction. Organizations have invested millions of dollars into technologies and services to help them improve patient outcomes and reduce costs. However, despite the increased investment in population health, the ability to achieve a return on that investment (ROI) varies greatly from one organization to the next. Why is this?
Based on Evolent Health's experience working with more than 100 risk-bearing provider-led organizations, we have found that population health ROI is tied directly to the ability to successfully identify, engage and intervene with “impactable” patients.
We define “impactability” as “the likelihood that a patient will incur a specific future adverse event that can be prevented through engagement in an optimized, evidence-based intervention.” In practice, this means identifying the right patients at the right time, with the appropriate intervention, leveraging the right resources.
Each dimension—identify, engage, intervene—builds sequentially upon the one before it, compounding the success (or failure) of the previous dimension. This compounding effect means that even above-average execution of each dimension is simply not enough to generate a positive ROI for the entire population and ultimately leaves nearly half of the possible value on the table.
So where do we see organizations go wrong?
Today, many organizations rely on predictive stratification models developed for payers in an attempt to identify high-cost patients. While these models are an improvement on methods that stratify patients based on historical cost or utilization, they still focus on financial risk without incorporating impactability. The extent to which a patient is impactable depends upon multiple factors, including their disease state, level of engagement, income and geolocation. It also depends on the interventions available. Identifying a patient as “high-risk” isn't valuable unless there are interventions available that are likely to succeed.
Once impactable patients are identified, the next step is to engage them in care management. Most organizations use the same patient engagement process for all patients: a series of phone calls and voice messages. Each care manager uses the same script without taking into account any known clinical or social circumstances. This “one size fits all” approach often produces a success rate that likely will not drive substantial clinical improvement for a population, diminishing the program's value.
The last step, selecting and applying interventions, is arguably the most important dimension, yet the most difficult to successfully execute. For many organizations, the selection of interventions for the highest-risk patients is based on a manual clinical review of stratified patients - a time-consuming, costly and subjective process. Additionally, many organizations lack workflow to drive the consistent application of interventions across a team of care managers, instead leaving execution to the discretion of individuals. This not only drives inconsistent care delivery, since it does not control for human error, it also prevents ROI measurement at the intervention level.
Execution of best practice is not an easy task for any organization. Best practice patient identification, or stratification, leverages a suite of models to predict specific avoidable events. These predictive models tap into a broad set of data including clinical, administrative and social determinants. A different set of predictive models is then used to segment patients into groups based on their likelihood to engage with a care manager. Outreach efforts can then be customized to maximize patient engagement, taking into account demographics and communication preferences. Advanced analytics are also leveraged to support the selection and delivery of interventions in three distinct ways: automated intervention selection, standardized workflows supported by technology and stringent evaluation and performance improvement. By tying all of these factors together into an integrated approach, organizations can implement population health at a best practice standard across all three dimensions—identify, engage, intervene.
For most organizations, doing this requires diverse expertise and comes with a great deal of execution risk. Yet as we mentioned, successfully executing on each dimension of impactability is critical to the success of your population health program. While we've only scratched the surface of the current state, our white paper, “Why Do Population Health Results Vary? An Introduction to "Impactability", provides a detailed look at best practice across the dimensions as well as a real-life example of success.