Analytic Pathways: Four Steps to Increase the Yield of your Analytic Projects

With the rollout of the Affordable Care Act (ACA) nearly complete, our healthcare system is in the midst of a historic transition to a value-based system driven by proactive population health management. This shift creates a tremendous opportunity to apply analytics to improve outcomes and reduce costs.

Despite this promise, it is important to remember that analytics initiatives in other industries are known to have a high failure rate. What’s needed is not more data or reports, but meaningful and actionable insights to drive change. Consider this four-step approach, called Analytic Pathways, which grounds analytical work in a strategic perspective, and provides structure to an inherently exploratory and iterative process.

Step 1. Frame a Powerful Question: The first, and most important, step is to clarify the problem you are trying to solve. In order to leverage the power of analytics, broad business issues must be translated to a set of specific questions that can be explored quantitatively.

This is easier said than done. To help get started, it’s useful to develop a high-level Key Performance Indicator scorecard of metrics related to a high-level strategic question. For example, if you are evaluating strategies to refine benefit design for spouses, your scorecard could compare risk, cost, and utilization metrics of spouses compared to employees.

Step 2. Identify Key Drivers: Once you’ve framed the key question for the analysis, the next step is to understand where the problem lies. Is it driven by a particular conditions, or types of procedures? Is it concentrated in certain locations or plan types? Step two centers on slicing the data in different ways to reveal a small set of factors that have a disproportionate influence on the question.

Focus your analysis on understanding the causes of absolute differences in “rated metrics,” like per-member-per-month (PMPM) costs, or admission rates. For example, consider the breakdown of inpatient admissions below.  Many analysts would gravitate towards the 23 percent increase in behavioral health admissions. However, looking at the absolute differences shows that the 13 percent increase in medical admissions is the most important driver, accounting for just over half of the total increase in the population’s admission rate.

Step 3. Pinpoint Root Causes: Now that you understand where the problem lies, the next step is to drill down and determine the underlying causes of the cost or quality issues, like disease prevalence, provider contracting, or plan design.

When exploring differences in PMPM costs, applying the accounting technique of cost variance analysis can be a useful tool to further focus your efforts. Cost variance analysis breaks down a difference (variance) in costs into three specific factors: unit cost, mix, and volume. For example, say you were analyzing differences in imaging PMPM costs by procedure. If the largest driver was unit cost, the root cause could be the presence of high-cost facilities in your network. On the other hand, a large mix variance could indicate that certain providers tend to provide higher-cost services; if the issue was driven by volume, it could indicate that the prevalence of a condition was driving higher overall utilization.

In identifying root causes, remember that specificity drives action. For example, it is not enough to know that the admission rate for diabetics is too high; it is far more useful to pinpoint the hospitals with the most avoidable admissions for diabetes, and discover the PCPs that could help better manage these patients. 

Step 4. Develop the Action Plan: After working through the first three steps, the focus shifts from mining the data to applying experience and judgment to develop the right solutions.

An effective action plan hinges on assembling the right team and deploying a thoughtful process. A robust, cross-functional perspective ensures that you are applying the right data-driven insights to the key strategic issues. Recognize that what is actionable is different for each organization, and depends on the characteristics of your population and the mission and goals of your organization.

Analytic Pathways is a general process that leads to unique, targeted results. Each analysis is different. There is no “algorithm.” But by beginning the process with a strategic goal in mind, investing time up front to frame the right questions, and applying a structured and flexible process, organizations can increase the yield on their analytic efforts—and ultimately, develop better insights to improve the health of the populations served.

Sam Stearns leads Verisk Health’s consulting practice, which provides analytic consulting services that complement Verisk Health’s population health solutions. He has led numerous engagements to help employers, payers, and providers increase the value of care delivered to their populations, including population health diagnostics, health program evaluation, provider cost and quality analytics, and predictive modeling.