Clinical Analytics

HealthLeaders Media Council Members discuss how they implement clinical analytics in their population health initiatives.

This article first appeared in the September 2016 issue of HealthLeaders magazine.

Kathy Fair
Director of Medical Economics
Blue Cross Blue Shield of Kansas
Topeka, Kansas

As a payer, it's still a challenge to get clinical analytics data right now, with there being so many different EMRs in use; a common extraction system still doesn't seem to exist. We are going to do a pilot with a company that says it can provide an extraction service, but it remains to be seen if this is really possible.

I think that clinical data is the missing piece of the puzzle—we already have the claims, the member information, and the provider information, and this data piece would complete the picture we have of our members. With that said, there are probably many things we need to learn about that data that we don't know yet. It's probably going to be another three to five years before we get that kind of information.

At this point, we've got analytics all over the place, in just about every department of the company. We run analytics for membership information, provider data, everything. We look for anomalies in different data just to see if something is "sticking out" where it shouldn't be.

We used population health data most recently with our PCMH and our HMO activities. We're still new to that arena, but we've been analyzing this data for over a year now, and I think it helped us to identify potential process improvements for physician offices and factors that were outside the norm that they could work on to improve their quality and cost of care.

Matt Ebaugh
Vice President, Chief Strategy and Information Officer
King's Daughters Health System
Ashland, Kentucky

Analytics are a cornerstone, and foundational to whatever we will be doing, whether it's within analysis of clinical process, of integration and transitions of care, of risk-bearing contracts, or of patient population.

Within that, the first step is to understand the population that you serve. Population health basically puts people into logical groups based on their health status. The next step is to isolate the most high-risk patients who need to be managed, focusing on the patients who may be incurring more costs than the average across the nation, and to figure out how to treat those individuals, engage them, or work on behavioral modification.

When you blow population health up to the highest level, it really is working with a certain individual on specific needs; it just happens that you're grouping the patient into like populations. That might change with precision medicine, but people go into groups based on past family history, age, gender, and other factors. We put them in groups from a clinical standpoint. So, it's difficult for me to divorce population health from anything else that we do, because that is essentially how we provide care today.

Until the past one or two years within our market, I hadn't found payers to be cooperative or helpful, but we're slowly seeing better discussions around clinical bundles and taking on risk.

Simon Lin, MD, MBA
Chief Research Information Officer
Nationwide Children's Hospital
The Research Institute
Columbus, Ohio

The way I see it, clinical analytics act as an enabler of the data-driven healthcare enterprise. What I mean by that is three things: first, that clinical analytics make the data accessible; second, that they make it possible to analyze the data and create valuable insights; and third, that clinicians can use those insights to make informed decisions.

I see the clinical metrics as a major connector for the data by which clinicians and patients can make informed decisions.

My organization's situation is somewhat unique, as we are a research institute in addition to being a hospital. Because of this, we've put the effort into developing many analytics capabilities, and analytics are strongly intertwined with our research process—for example, the analysis of a genome, the analysis of clinical utilization, and the analysis of clinical trials results. Some of this might be specific to our research projects, but it also complements population health research and data. Research is an area where analytics really have an opportunity to shine.

I think it's still a little early to call out the benefits of analytics for population health, but we are definitely hoping to see improved outcomes, reduced costs, and stronger patient satisfaction.

Tom Lowry
Vice President of Finance, Physician Integration
Dignity Health
Rancho Cordova, California

The answer to your question depends upon how you define population health. If you want to talk about programs such as CMS' bundled payments, or the Comprehensive Care for Joint Replacement Model, those programs will cause you to manage on the continuum of care, and you are going to need clinical data to do that. Today we are using clinical data for those purposes.

There are other population health programs that you might consider, and in California we've been utilizing some for quite a while. In some of these, you have metrics for things like pay-for-performance measures under capitation. Under these and similar programs, you need to be able to tap the clinical data. To go back to bundled payments, you have to identify patients who are going to fit into a bundle, and you want to do that at the front end as soon as possible, as you are responsible for each of these patients.

For example, under the bundled payments model that we've chosen at Dignity Health, we've elected to carry risk for 90 days postdischarge from the anchor admission, so we're going to want to look at that patient and understand his or her health needs—not just from the anchor admission itself, but also at what his or her health needs are that might affect care in postdischarge. That information is found in the health record—it's found in the patient's clinical data.