Identifying Individuals

Now that we have chosen our population segment, how will we find individual patients who are a good fit for our program?

clemens-hong-pictureDr. Hong shares guidance on how to begin identifying individual patients. This process of identifying potential patients for enhanced programming is referred to as ‘patient identification’.

“A first step is to pull together the data that you have. One thing to be aware of is to not let perfect be the enemy of good. There is no perfect risk prediction tool, specifically for your population or just in general. There’s no perfect risk prediction tool for costs. There’s no perfect risk prediction tool for readmissions. My advice is to comb the literature, talk to people who have done this successfully, figure out what they’re using, and then look at your data and see how much of it you have. Start with what you have and refine it as you learn more about your chosen population.” Clemens Hong, MD, a primary care physician, and Medical Director of Community Health and Improvement for Los Angeles County, California

Why is this important? And what do we know?

  • Enhanced care programs serving patients with complex needs and high costs carry a higher cost than usual care, with smaller caseloads and team-based care. If a program is going to survive financially, the enhanced care needs to be offered only to those people who are likely to benefit from it and whose health care costs are likely to be decreased through enhanced care. If a program were to engage people in care who are not a good fit for the care interventions, or for whom there is little opportunity to decrease costs, the team’s time, effort, and the cost of care would be wasted. To build a sustainable program that has the potential to serve the entire chosen population segment, it is important to determine the specific characteristics that indicate an individual is a good candidate to be a patient in enhanced care.
  • There is no magic bullet. All patient identification methods are imperfect: any one method will screen in people who are not the best fit for enhanced care; and will screen out, or miss, people who are a good fit. Given that, the best approach is to combine quantitative and qualitative identification methods, offering both redundancy and a wider net.
  • The more time-sensitive your data is, the more rapidly your team can find potential patients. It is crucial to develop nimble real-time identification methods that allow your team to find and meet patients when they most need your service, and when their motivation to participate in enhanced programming is higher than usual. Real-time identification methods enable teams to meet potential patients as soon as knowledge of patient need is documented, streamlining access to care and aiding in patient engagement.
  • Teams that are committed to quality improvement refine their identification methods as they serve more patients and learn more about their program, patient needs, and identification methods. This discipline around learning allows teams to develop the best possible identification methods.

Core recommendations

Identify specific individuals within your chosen population using multiple approaches.
Develop real-time patient identification methods.
  • Develop an event notification in the electronic health record to flag that a patient that meets the criteria for the enhanced care program is present in care.
  • Test using a triage coordinator to manage the inflow of potential patients.
  • Develop clear referral methods to elicit patient referrals from primary care or other providers.
  • Detail to develop real-time patient identification methods.

How to get started identifying patients for your enhanced care

In the first step, “Choose your population and learn about its needs and assets,” you chose a population segment and defined its characteristics. You may have chosen, for example: patients with four or more chronic conditions; or patients with a combination of three or more hospitalizations in the past year and three or more emergency department visits; or patients experiencing homelessness; or patients over age 65 with frail health.

To get started on identifying people who fit the description of your chosen population, there are a few different things you can try:

Test ways to find people in your chosen population
  • Primary care physicians can share a list of potential patients who are a good fit for your program.
  • Compare utilization or clinical data with the list of potential patients created by primary care physicians.
  • Deploy a care manager directly to the hospital emergency department, perhaps one day a week for 2-3 weeks, to identify and then try to meet patients when they present for care. This step moves from identifying people for enhanced care to trying to engage people in care.
Refine your identification approaches using iterative learning cycles.
  • Study what works and what does not in each of the patient identification methods you test.
  • Be realistic about the time and effort it takes to accomplish each identification method. What method is likely to work at scale?
Meet 5-10 of the people identified through your identification methods
  • Is the first cohort of people a good fit for your program? Have the identified people had high needs and high costs in the past 2-3 years, and do they have a risk of high costs in the future?
    • If yes, use your identification methods to identify another cohort of potential patients.
    • If no, rework your identification methods to try to achieve a more effective identification approach.

Note that assessing the unique strengths and needs that each person presents with is part of engaging patients in care. For more on patient engagement, visit the Revolutionize Patient Engagement section.

Overall tips and guidance

  • Commit to continually improve the work processes that the team uses to identify patients.
  • Be aware that, over time, the characteristics of your chosen population segment might change – which means that you may need to revisit steps 1 (choose your population) and 2 (identify potential patients).
  • Develop inclusion criteria – determine the characteristics or information that will screen potential patients into care.
  • You may find it helpful to stratify the population segment with complex needs and high costs, meaning assessing relative levels of need or relative risk for high utilization in the future. This can be a way of prioritizing high-need subgroups for more immediate identification.
  • Read further on how Denver Health and the Camden Coalition approach stratifying their population segment.
Here is guidance from leaders in the field, particularly around what to do when your utilization data is not robust:

“What we’ve learned in our system is that the data is so out of date that it just frustrates our providers when you present them with high risk patients. They’re reviewing lists of which half aren’t their patients because we are still developing our empanelment process in primary care; a number of patients have died; others aren’t high risk at all; and other high risk patients are missing, making the lists not useful. It feels like a waste of time, which they do not have. So, we’ve moved, in our program, to strictly a provider referral approach. Eventually, we would like to go back to the quantitative identification approach. We are also doing chart reviews to find the folks who should be in the program but are not engaged by primary care. It’s intensive and time consuming, but can be more effective. We are taking our lists and saying, ‘Are there high-risk people that aren’t being seen in primary care that the provider might not know are high risk?’ It’s much more tedious, but we have had to do that because that is the state of our data for identification.” Clemens Hong, MD, a primary care physician, and Medical Director of Community Health and Improvement for Los Angeles County, California


“I would emphasize here that this issue of how best to identify patients for the program, it’s very important that the process includes matching the individuals to the interventions that you have available. It sounds kind of obvious but we’ve seen groups miss this point over and over when they are starting out. A classic example is, let’s go to the highest-cost people, let’s run the financial numbers and who’s the highest cost. Well, a lot of groups have shown that obviously for commercial population, but even for the Medicare groups, many of those patients may have very severe illness that’s already very established that they need extremely intensive costly care for. It’s not that you might not be able to find strategies to help mitigate their cost by contracting means or a narrow network means and maybe help them with quality-of-life or end-of-life or palliative care decisions. But if your program is trying to be more upstream in its preventive interventions, for example, weight management counseling and a lot of education on self-management of chronic diseases, then that approach is probably not well suited for that chosen population, right? Whatever criteria you use to better identify individuals make sure they are well matched for the interventions of your enhanced care model.” Ken Coburn, MD, DrPH, FACP, President, CEO, Medical Director, Health Quality Partners