How to Use Identification and Stratification Analytics for Effective Population Health Management
In a previous blog post, we emphasized the three pronged analytic strategy that risk-bearing organizations should employ when implementing chronic care management and other population health programs. In this post, I will review the first component of the strategy we have outlined: the “ID and Strat” analytic, which is industry jargon for identification and stratification.
This analytic is considered to be the “bare minimum” for any program that is aimed at reducing costs associated with high-risk individuals. However, the definition and scope of ID and Strat has grown exponentially. Today, the risk-bearing organizations may decide to deploy data mining, clinical algorithms, and/or predictive modeling based on financial risk, gaps in care, disease stage, likelihood of healthcare high costs, likelihood of non-compliance with quality measures, and more.
How you operationalize your ID and Strat strategy is dependent on several aspects of your organization. Addressing the following series of questions will help you determine what approach is the best.
What are Your Organizational Goals?
- Are you focused on achieving quality measures?
- How much financial risk do you bare for your population?
- Do you have a “special” population that has unique clinical characteristics?
Refine your analytics based on the complexity of the target population and outcomes objectives. In the case of quality measures compliance, you may be able to identify patients using simple clinical algorithms that sift the data to find gaps in utilization associated with quality standards.
In contrast, if you are trying to identify emerging financial risk, you may need to use slightly more sophisticated models to predict this risk. In this case, be careful to know the mechanics of the financial risk model you use (either home-grown or purchased from a vendor). For example, a model that is intended to identify emerging financial risk will be a different model than one that is intended to identify existing financial risk. Existing financial risk can be predicted somewhat easily using readily available data, such as last year’s cost and prior hospitalizations, whereas emerging financial risk, where you are trying to predict “ticking time bombs” (those who have little to no previous medical history), may require untraditional elements such as financial or biometric data.
“Special” populations refers to an organization’s desire to target a specific cohort that may be somewhat arbitrary – for example, you may want to target those who require a mammogram but who are unlikely to get one based on past behavior patterns.
What are Your Internal Capabilities?
- Do you have allocated analysts with the skills to do basic identification and stratification?
- Do you have IT resources that will give these analysts access to the necessary data?
Depending on the goals of your organization, you may be able to use clinical staff members who can use a program (such as Microsoft Office) for simple data management tasks, including sorting and filtering patients from a registry (as long as the registry contains all the data elements required). However, for more complex analyses, your team should include an experienced data manager and analyst who can work with large data sets and build predictive models. Your IT infrastructure should allow these resources to access the data with all the checks in place.
What is Your Intervention Capacity?
- How much money and headcount can you dedicate to this effort?
Think ahead when planning your strategy. Determine how much funding and staff resources the project will need from beginning to end before you get started. Even if you opt to use a vendor’s analytic services, at least one individual within your organization will have to be responsible for implementing and maintaining your ID and Strat efforts long-term.
Our next blog in this series will focus on Evaluation–ensuring you have an unbiased evaluation of your intervention.
Learn how Health Dialog uses a Care Pathways methodology to drive effective population identification and stratification.