Follow two simple rules to ensure you are getting data insights that actually lead to better outcomes

About two decades ago I served as the president of a physician management business owned by a publicly traded company. The company had chosen to invest in this business because they wanted to better understand global capitation (when a provider agrees to take on the total risk of a member by accepting a pre-established PMPM amount from a payer) and how their organization would fit into such a structure. Global capitation was like a bug hitting a windshield at 90 mph, but that is a story for another day.

Because our division was struggling to make it under the capitation rates—we were losing money and costs were running higher than global capitation—the powers that be (executives in the publicly traded company) decided to bring in a consulting firm. The firm was a premier international consulting firm, one of the alphabet houses. They conducted a performance analysis, comparing our costs in different slices and by specialty to normative data.

After spending $150,000 (in 1999 dollars) for the analysis, I received a thick binder that ultimately ended up sitting on the shelf. The one finding I recall was that it said our cardiology costs were way too high. “No kidding,” I thought. “I knew that.” What I really wanted to know was which cardiology patients were driving the costs and could have been treated more effectively. Which doctors were providing ineffective treatments and what were the alternatives? In other words, I needed the information broken down to the member- or claim-line level detail. Without this level of information, the findings were not actionable.  

To clarify, had I not already been aware that cardiology was a problem, one could argue it constituted an actionable finding. However, it still stands that just this high-level insight alone was not enough to help me take action and was ultimately inactionable. In order to implement an effective strategy, a deeper, more meaningful analysis would need to be conducted.  

This brings us to the analytics world of today. I work in healthcare so I'll speak about that area although I am sure this exists in other industries as well.

Rule #1: Ensure the analytics you are using in your decision-making are based on a foundation you can trust. This means the vendor should be fully transparent about the logic they used to arrive at their conclusions.

Rule #2: Get analytics that offer a 360-degree view of the member and allow you to drill down to the member- or claim line-level detail.

I often observe vendors in our space presenting findings that say something like “There are 49 diabetics with open gaps in care and they are costing you $250,000. You need disease management.” There are potential discrepancies with such a finding:

• What gaps in care are being analyzed and how can you verify that those gaps really exist? Not all gaps in care are equal (or as life-threatening)—a member who is barely diabetic with a lipid profile headed in the wrong direction is not as concerning as a diabetic with renal manifestations who is spiraling towards kidney failure and not seeing their doctor. You must be able to dive into each individual member’s profile.
• It’s also essential to understand how the vendor has determined that a member is a diabetic. What definition was used?
• How exactly did they calculate the $250,000 in expenses?

“Avoidable emergency room visits are running $500,000 so you should hire a telemedicine vendor” is another example of a finding a vendor might share. Before engaging a telemedicine vendor, you should ask:

• Who is accounting for the emergency room visits? Is it a few people or a few hundred people causing the problem?
• How was a particular emergency room visit classified as “avoidable?” What definition was used?

If you can’t look at the actual individual records that produced these findings, you're just blowing in the wind. Without being able to get down to the individual member record or claim-level detail, you are taking a huge leap of faith in simply trusting the conclusion provided by the vendor. It is important to have access to all data points underlying any recommendations for action.

Let’s say the analytics are impeccable. This is a big “IF” because one reason analytics vendors don’t allow analysis down to the claim-level detail is because such transparency would expose flaws in their logic. Even if the logic is perfectly sound and you want to take action by engaging a medical management vendor to help close care gaps or initiate a disease management program, how on earth would you know where to get started? Unless your analytics vendor has an open API, the medical management vendor will not be able to access the data.  

For example for the diabetes scenario, you can’t simply tell the vendor you have 49 problem diabetics. They are going to need to know which specific members are driving the issue. They will want to know any co-morbidities, the drugs the members are taking, and all sorts of other clinical indicators. In essence, before the vendor can begin implementing an effective program, you are going to have to pay that new vendor to grab the raw data and run their own analysis to identify the diabetics (which will most likely produce a different result than yours). Talk about a redundant process!

If all you're interested in are some summary dashboards for a couple of talking points, then none of this matters. However, if your objective is to be truly proactive, then the ability to capture and generate information down to the member- and claim line-level detail is paramount. Make sure you are purchasing analytics that are truly actionable. Ask yourself, “How would I execute a plan based solely on the results provided by this analytics vendor?” If you can’t answer that, it’s best to move on and look into using a different vendor.

Deerwalk Plan Analytics is based on a foundation of data integrity for data insights you can trust. We provide complete visibility into the health of a population with 360-degree views of each individual member and full transparency into the underlying logic. Below are some examples of the types of data transparency we provide:

• From our Avoidable ER Admissions report, you can view the entire list of avoidable admissions and drill down to the visit records for each to see the avoidable diagnosis. You can take it a step further to view all the claims that roll up to a specific ER visit, allowing you to conduct an audit of the all-in cost and/or grouping logic.
• All procedure/diagnosis groupers are available directly within the application so you can view the codes that Deerwalk rolls up into designated procedure and diagnosis sub-groupers and understand how we calculate gaps in care.
• Concurrent and prospective member risk scores (MARA - Milliman Advanced Risk Adjusters) go beyond a simple scoring system to provide a breakdown of the weighted clinical conditions that are driving the individual's risk score.
• Documentation that gives clients visibility into how every single metric is calculated, down to the level of individual procedure, diagnosis, national drug codes (NDC), and more.

Schedule a demo to learn more about how to put Deerwalk Plan Analytics to work for you.

Subscribe to our blog or follow us on Facebook, Twitter or LinkedIn to join the conversation.