Refining your approach to risk stratification

Most organizations already have the data they need to identify their most actionable patients—they just need to leverage it a bit differently. The first step aligns with what most organizations do already: leverage a holistic suite of risk factors, both clinical and non-clinical, to create inclusion criteria.

But inclusion criteria will get you only so far. The likely result is a group of patients that is larger than the average organization can fully support with high-touch services—and many patients for whom costs will be difficult to inflect. Identifying actionable patients requires a second step: overlay exclusion criteria to filter out patients whose high cost or utilization is out of the health system’s control. Such patients often require a different care model that necessitates costly, longitudinal support.

Accounting for available resources

Identifying actionable patients cannot be divorced from organizational bandwidth. To truly be “actionable,” the number and type of patients identified must align with an organization’s available services and clinical capacity.

For example, one organization we work with, recognizing their limited number of care managers, used inclusion and exclusion criteria to segment patients by “care coordination burden.” The organization then reallocated RNs to the practices where patients with the greatest care coordination burden were located.

By acknowledging resource constraints early in their risk stratification efforts, organizations can selectively define what constitutes “actionable” and then deploy resources to where they are needed most.

Effective risk stratification requires understanding patients’ psychosocial complexity and non-clinical risk factors. While many organizations screen for some non-clinical risk factors, most leaders are unsure what data they should prioritize in order to prevent patient and clinician survey fatigue.

Focus new data collection efforts on information that is easy to identify and predictive of future health care utilization and poor outcomes. Two risk factors deemed high-priority and predictive (especially in light of the dramatic impact of Covid-19) are:

• Behavioral health: Uncover behavioral health needs through validated screenings such as, a depression screen, and through conversations about stress-related behaviors, such as smoking or changes in routine.
• Financial heath: Uncover through payer status data (particularly dual eligibility) and through nonjudgmental questions about patients’ financial worries such as, “Are you able to refill your prescriptions?”

Encourage patients to fill out screens as pre-visit work, upskill MAs or clinicians to sensitively screen, and/or partner with local community organizations to supplement health system data collection.

Focused data collection driven through the EHR or community partnerships can reveal patient demand and need for certain services writ large—and help prioritize population health interventions that can demonstrate return.

As organizations work to stratify patients by risk, they will quickly uncover a blind spot: high-risk patients who are disconnected from care altogether. And as patients postpone their care due to continued fear surrounding Covid-19, the number of high-risk, non-utilizers will grow—which will lead to an increased risk of costly health care utilization down the line.

There are two main ways organizations can identify these patients. One way is geotargeting, a data-driven method that helps clinicians pinpoint geographic areas of greatest need on which to focus outreach. Such areas can include: areas facing significant job loss, housing instability, high-incidence of Covid-19, and ZIP codes with high levels of high-risk, high-cost patients.

The second, even simpler, way is mining chronic disease registries for patients who haven’t been seen in a while. Be sure to include factors such as ZIP code and clinician judgment, given that clinicians often know which of their patients are and will become increasingly high-risk. Frontline staff can then proactively reach out to these patients to offer disease support through phone consults, video visits, office visits, remote patient monitoring, or home visits.

These efforts can support organizations’ understanding of community drivers of risk—effectively building out their population health enterprise.