ACO Success Requires Precise Patient Population Definitions
How to make an accountable care organization (ACO) successful is a big topic in our industry today. The overwhelming nature of establishing and operating an ACO makes it easy to miss fundamental aspects. Often overlooked in discussions on this topic: an accurately defined patient population is crucial to a successful ACO.
Without precise definition of its populations, everything else in the ACO strategy suffers. Simply put, ACOs taking on risk for a patient population must be able to pinpoint who belongs in that population and who does not.
It sounds obvious, right? But it’s far from simple. In fact, few solutions on the market can define ACO populations really well.
How ACOs Use Population Definitions
ACOs define populations for four main reasons:
- To identify their members and attribute those patients to the correct physician(s)
- To attribute the right patient to the right physician for pay-for-performance incentives
- To accurately report CMS and other quality measures
- To perform population health analytics that enable providers to effectively manage each patient’s health
To further explain the last bullet, ACOs define populations so that they can identify high-risk members and then manage their health more carefully. This management includes:
- Performing analytics to predict the likelihood of a specific event (such as a readmission or the need for a surgical intervention).
- Tracking utilization trends across the full care continuum. This enables ACOs to define the costs of managing both the high-risk patient pool and the population as a whole.
- Monitoring quality of care by facility and by provider. This enables the ACO to steer the right patients to the highest-quality, lowest-cost provider.
Challenges to Defining an ACO Population
As I mentioned before, none of this work is easy. And it is only as effective as the population definitions are accurate. Here are some of the key challenges in defining ACO populations.
Multiple providers per member
The truth is that very few people—especially those with chronic health conditions and comorbidities—visit one doctor. Most, in fact, visit several. Getting data from all these providers, attributing members to the provider that has the most impact on the patient’s care, and establishing a medical home involve complex and difficult processes.
Multiple data sources
One of the biggest challenges to creating an accurate registry is aggregating data from across the continuum. When I say “continuum” I’m not just paying lip service to the concept of bridging the gap between hospitals and physician practices. Well-defined populations require data from all of the following and more:
- Physician practices and clinics with their various EMRs
- Claims or billing data
- Skilled nursing facilities
- Rehab facilities
- Dialysis facilities
- Surgery centers
- Home health
Multiple identifiers for each member
Once you’ve aggregated data from all of these sources, matching each encounter from multiple systems to the correct member is no easy task. Reconciling all of the medical record numbers and other internally assigned identifiers from different source systems requires robust master patient indexing capabilities.
Successfully Defining ACO Populations
Health Catalyst has been successfully and accurately defining patient populations for many years. Here are some of the elements that make our definitions so effective for managing populations of patients.
Using a Clinical Integration Hierarchy
As those of us in the business of population health management know, the majority of strategies and interventions to improve health won’t be applied to the population as a whole but to sub-populations that share one or more clinical characteristics. Health Catalyst recommends using a clinical hierarchy system for grouping these common clinical characteristics and identifying individuals who share these characteristics. The hierarchy organizes and classifies care around key clinical work processes. This classification reflects how care is really delivered—it spans department and organizational boundaries across the continuum. The following is an example of the Clinical Integration hierarchy system that we have developed.
Starting at the most-general and moving to the most-granular level, this hierarchy is as follows:
1. Clinical program: Twelve clinical programs make up a comprehensive healthcare delivery system. Primary Care, Cardiovascular, and Mental Health are examples of clinical programs.
2. Care process family: Each clinical program consists of multiple care process families. For example, the Cardiovascular clinical program includes these care process families:
- Ischemic Heart Disease
- Vascular Disorders
- Heart Failure
- Heart Rhythm Disorders
3. Care process: Care processes represent the most granular level of the hierarchy and may exist anywhere along the continuum of care. The following care processes are part of the Ischemic Heart Disease care process family:
- Coronary Atherosclerosis
- Acute Myocardial Infarction
- Percutaneous Intervention (PCI)
- Coronary Artery Bypass Graft (CABG)
- Cardiac Rehab
We initially developed this hierarchy for inpatient care process analysis. Today, we are expanding the mapping process throughout the continuum of care.
Using Technology to Build and Refine Registries
Health Catalyst has linked this Clinical Integration hierarchy to the population definition process. We have developed two main tools that enable our clients to easily, efficiently, and accurately define their populations. These tools—or analytics applications—run on top of our enterprise data warehouse (EDW) platform.
The first tool is our Population Explorer, which delivers a “starter set” of more than 1,000 different registries. It begins with standardized rules for creating the registry (for example, all the ICD-9 diagnosis codes associated with congestive heart failure). Then, it adds care mappings that bring additional data elements (like laboratory results) into the registry to create a more comprehensive rule for defining the population.
The next tool is our Cohort Builder. This analytics application enables more sophisticated definitions of a population. Teams of clinicians, analysts, and others work together using this tool to adjust and refine the population definition to suit the ACO’s specific needs.
The level of refinement these tools enable is essential to an ACO’s effectiveness. Imagine you’re trying to better manage your diabetes population. You build your diabetes registry—or, rather, you define your diabetes population—based on ICD-9 codes. Most standard registries are based on these diagnosis codes. What many people don’t know is that registries built solely on diagnosis codes miss 30 to 40 percent of the people who should be included in the population. The tools go beyond those diagnosis codes and pick up those missed patients through additional data elements such as laboratory results.
In a value-based, fixed-price contracting model, that level of inaccuracy would be financially devastating to the ACO. Using these tools, which are then mapped to a deep clinical integration hierarchy, allows ACOs to develop the precise, detailed population definitions needed to keep from falling into that trap.
Has your organization created population registries? What factors went into defining those registries? How confident are you in the accuracy of your population definitions?
Would you like to use or share these concepts? Download this ACO Success presentation highlighting the key main points.