Liberating Data in Wisconsin: How the State is Evolving its All-Payer Claims Database
Many states have mandates to establish all-payer claims databases, but in Wisconsin, the work is being done voluntarily—with several layers of complexity
BY RAJIV LEVENTHAL — DECEMBER 17, 2019
Across the U.S., about one-third of states have established all-payer claims databases (APCDs)—large state databases that include medical claims, pharmacy claims, and other key beneficiary information—with the goal to give healthcare leaders much more insight and transparency into how Americans move through the health insurance system.
This form of data aggregation, directly from payers to the state, could help healthcare stakeholders better manage patient populations while also identifying opportunities to reduce healthcare spending. Most of the states that have created APCDs have been mandated to do so via state legislation, but in a few cases, state healthcare leaders have voluntarily developed them.
One example of this is in Wisconsin, as the Wisconsin Health Information Organization (WHIO) was established over a decade ago, in absence of a state mandate, and today includes claims data on more than 4 million individuals—or about 75 percent of Wisconsin’s population, encompassing more than 314 million claims. Recently, WHIO and its technology partner, SymphonyCare, a Madison, Wis.-based healthcare data management company, announced the launch of WHIO 2.0, designed to take the APCD to the next level in a variety of ways, such as leveraging SymphonyCare’s EHR-agnostic enterprise data warehouse that manages multiple data sources to produce key cost and quality reports, visualizations and intelligence on demand.
Dana Richardson, CEO of WHIO, recently spoke with Healthcare Innovation about the latest growth of WHIO, key initiatives the organization is working on now to make better use of the data, and more.
What’s the latest you can share regarding the growth and membership base of WHIO?
In Wisconsin, our APCD is not mandated by law, meaning that a health plan could choose not to submit data to WHIO. That makes a difference in how much energy and resources we have to spend working with the data contributors, and also if we end up with any ‘holes’ in our data based on having a large data contributor not [sending] their data.
The flip side of that is that the Wisconsin insurance market looks much different from other states, where there are usually are about three to six big insurance companies. Wisconsin’s market, though, is mostly small, regional plans, so we have more data contributors—we have 15 right now. Each one has less impact, though.
So we started back in 2008 on this journey and we have never had 100 percent coverage, but we have had generally between 70 to 75 percent of our population covered, which is where we are at right now. Our 15 data contributors include all the regional health plans that submit both their commercial population and Medicare Advantage data. We receive all the data from the state of Wisconsin for the Medicaid population, and we also get some self-funded employers that submit their data, too.
Looking at the data that’s coming in, what kind of aggregating is happening, and what kind of work is WHIO looking to do with the data?
Based on our longevity and the fact we are not in as a highly-regulated environment as some other states are, we provide de-identified databases to our customers. So we integrate all the data across all these data contributors, we match the data by patient, and each patient gets what we call a WHIO ID, which is essentially a unique, non-intelligent key. And then we do other de-identification of [data] such as date-of-birth, which counties they live, etc., in order to meet HIPAA requirements.
Once we have integrated all that data and made sure there aren’t duplications, we then create two different data files. One is called the standard integrated data file, which is all the analytical data elements of a claim, which then becomes available to our customers to bring into their own technical environment. Then they can use their own business intelligence tools on top of that to do the analytics that they want.
The second data file that we offer is similar in that it has many of those same data elements, but we add some enhancements. We license software from Optum and add episode treatment groups to the database, which are similar to bundles of care both on the procedure side and the chronic disease management side. With that, we offer a risk adjustment that goes along with that episode so that when you are doing analytics you can quickly gather up a whole group of claims that belong to knee surgery, for example, and also use the risk adjustment in there to quickly risk adjust the results you are getting.
The third enhancement we provide is called normalized pricing, which takes each CPT or ICD code, essentially, and puts a price on it that’s standard across all the data. So as you are grouping the data together for analytics you can see the differences in the amount of resources that are used to create an outcome.
What are the biggest advantages of being a WHIO participant?
In states that don’t have any type of statewide or large regional claims databases, it’s very hard for any individual health system, for example, to get access to that information. The reason I say that is because each health system would essentially have to duplicate what we are doing, and that’s complicated and costly.
So in absence of a database, each health system would have to reach out to the health plan to request claims data, but under HIPAA they are only eligible to receive its own claims data. With a statewide or regional APCD, however, we receive everything and then because we de-identify it, we make all the data from the state available, [thus] allowing the health system to not only look at its own performance, but also compare itself to competitors and peers.
You can also use the database for peer comparisons. Our data identifies providers at the system level—hospitals, clinics, nursing homes, etc., all the way down to the individual physician level. So we do a lot of reporting where we take opioid use, as a good example, for any one health system, and we can say that your orthopedic surgeons are prescribing opioids in this practice pattern, and then we can provide a statewide peer comparison of what other orthopedic surgeons are doing across the state. That’s the advantage to having these large geographic databases—you can get access to benchmarks. If you were bringing the claims into your own system, that health plan isn’t allowed to give you everyone else’s work.
How are you evolving to continue to meet the needs of your members?
We have a number of different tracks we are working on in that respect, and we consider our core customers to of course be the health plans who contribute the data and the provider systems who access and use the data. But we also have been working to build more connectivity to the research community. We also have, as one of our priority customers, the state of Wisconsin providing support to various government agencies, and the last [group] we are focusing on is the employer population. So, we need a different strategy for our different market segments in order to be able to work to deliver the type of information that each one needs, since they are unique.
For example, the providers in our state are very interested in combining claims and clinical data. So take, for instance, the ability to evaluate the care that your health system provides compared to other systems on something like diabetic care. In order to do that you need quite a spectrum of information. With the claims data, you can get the results of clinical process measures, you can get some outcomes such as readmissions, ER use, use of other types of providers, and you can also get cost or resource use evaluation to say that you are providing [a given] level of care, and my rates of hemoglobin A1C screening are being done at this cost, at my health system, compared to others.
But what you can’t get out of that is any type of clinical outcome. For example, what is the average hemoglobin A1C of all the patients who are cared for by this health system? So our approach to add value there is actually through a joint collaboration with the Wisconsin Collaborative for Healthcare Quality. They collect clinical data such as hemoglobin A1C levels, blood pressure levels, and other information, from about 70 percent of the state. In the first half of 2020, we will be bringing their data into our technical infrastructure in order to be able to create this more complete picture of what care looks like, specifically focused on chronic disease management. This is something our health systems are working hard to build themselves, but they can’t get that comparison when they are doing it internally.
What core challenges do you still need to overcome?
One core challenge is the lack of a mandate because it means we have to spend more time and use more market pressure to encourage health plans that may not be providing their data to WHIO. Where we have seen that change over the last 10 years is with the national plans that have developed their own data support strategies in a way that they can provide unique benefits to their customer markets, which makes them less interested in participating in something like a statewide database. So that is a struggle that has evolved over time as the national health plan strategies have changed from being more participatory at the local level to having more of a national view of their data use.
Another thing we struggle with is simply helping people understand what the value of this information is. The top question I get asked is, what do organizations do with this information? These large databases in healthcare have not been out there for very long and the analytical capabilities have matured a lot in the last few years, but they are not completely comprehensive. So it’s hard for providers, particularly, who are not accustomed to using claims data—payers have a 20-year track record of using claims data—to truly wrap their heads around the volume of data and what you can do with it.