Pop health success will hinge on mixing in patient social info

Providers will need more than just clinical data to drive analytics, says Michael Simpson.


As the healthcare industry transitions from fee-for-service to value-based payment models, population health is emerging as a critical component for providers to achieve clinical and financial success. Under this new paradigm designed to reward value, physicians increasingly will be paid by their ability to analyze data and improve patient outcomes.

According to Michael Simpson, CEO of population health vendor Caradigm, moving from episodic, single-patient care to managing the health of populations represents a seismic shift that will require industry focus and commitment, particularly in treating patients with chronic conditions.

For population health to truly have an impact on care delivery reform, Simpson believes collecting the right data—social, economic and physical environments, not just clinical data—is imperative as well as leveraging predictive analytics that not only identifies patients currently costing healthcare organizations the most time and money to treat but those whose treatment is at risk of incurring significant costs in the future.

Managing data for more than 175 million patients, Caradigm boasts more than 200 customers including Greenville Health System, Billings Clinic and Virtua and other large integrated delivery networks, ACOs, academic medical centers, government facilities and community hospitals.

Greg Slabodkin of Health Data Management spoke to Simpson last week at the HIMSS16 conference in Las Vegas.

The CMIO from Kaiser Permanente announced to a HIMSS16 conference session that population health is dead and that the days of a one-size-fits-all approach to healthcare are numbered. Do you agree?
Since Kaiser has been doing it for 30 years, it’s amazing that he would say that.

I think he was saying we will see a day when providers won’t screen populations but will personalize their approach for a whole series of diseases tailored to an individual's genetic profile. So, why do we keep talking about population health in the new world of personalized medicine?
We’ve been talking about population health for years, and it’s just now coming into the reality of the change management process of healthcare. I think personalized medicine will eventually come into that same reality. Think about genomics and all the information that you’ve got to gather. It’s going to take quite a while before personalized care plans are the norm.

Right now, if you’re a hospital and you’ve got a shared savings program, you’ve got to manage that risk otherwise you’re going to go bankrupt. I don’t think population health is anywhere near dead. The question is: how are you going to effectuate population health? Two or three years ago, everyone thought we’re just going to do this pop health thing. And, now, what it’s turned into is you’re doing a bundled payment, a little piece in the ACO, you’re doing some MSP, you’re doing different types of shared savings programs and you’re capitating your risks across the continuum in 10 or 20 different products. People don’t think of those products as pop health, but that’s really what it is.

It seems like many providers don’t necessarily know what they’re looking for when it comes to population health. Ironically, when you ask them if their EHR vendor gives them that capability, they answer no.
It’s the level of maturity of the provider. There’s always the first providers—the Greenville Health Systems of the world—who are on the forefront and know exactly what they are trying to manage. I think the individual doctor practices and smaller clinics of 10 to 20 docs understand that at some point they’re going to have to deal with bundles or some type of capitation. Between now and 2019, when CMS jacks up some more of the regulations around how much of your patient population is going to be at risk, you’re going to have to learn how to manage it. Today, you’re lucky if you’re a hospital and you’ve got 10 percent or maybe 15 percent tops at risk. That’s going to change as it moves to 35 percent. What are you going to do between now and then? You can’t hope. Hope isn’t a strategy. You actually have to put these programs in place.

As we move from fee-for-service to value-based care, what’s the value of data in a real tangible way?
You have to be comfortable with predicting which patients are going to cost you money and which patients are going to hit your ER. If you’re going through a divorce or you’re having some new stressor in your life, what is the chance that your immune system is down and that you’re now going to end up in a hospital? Being able to flag that type of information to a healthcare provider so that they can go take action is prevention and how you save money.

Where do you see predictive analytics going?
Everyday it gets better and better. Our first challenge in healthcare is about getting clean data. Caradigm has spent the last four years trying to codify information as we store it and bring in that clean data. If you look at IBM and what some of the other competitors are doing, they’re just now getting to the point where they’re getting all these data sources but now they’ve got to cleanse it. We’ve already built the algorithms to cleanse data. So, every piece of algorithm and prediction that we do is based on that data that we bring in.

In terms of cleaning the data, is that a valuable undertaking?
It’s not a function of going back and cleaning the last 10 years worth of data. I’d agree with anybody that it’s kind of a waste of time. But, clinically now to take the data that is currently a part of that workflow, that’s what you should spend time on mapping and understanding how you can continually get that good flow of the right data.

How do you handle structured data versus unstructured data?
Unstructured data as well as structured data goes within our Caradigm intelligence platform. So, we’ll bring in an unstructured stream and we’ll use natural language processing to extract the information that we’re looking for. We store the unstructured elements as a kind of a blob file, so to speak, and then we take it and break out the discrete components. If there’re problems, diagnoses or allergies, we’ll pull that in and put it into structured format so you can run real analytics on it. We deal with big data.

The data in our cloud are measured in petabytes, not gigabytes or terabytes. We have tons of data but it’s not about big data. It’s about getting the right data to the provider at the right time. Our job is to screen out that terabyte of data that came in last night and get you the three lines that you need as a doctor or nurse to actually take action for the patients that are coming into your office the next day. It’s all about the analytics and turning the data into actionable insights in a consumable format, whether that’s a simple alert or a reminder injected into an EHR.

Do you get excited about mobile health apps and wearable technology as a source of patient-generated data?
I get excited about it when it comes to home health monitoring and keeping tabs on blood sugar and weight in the case of congestive heart failure patients. Those are the types of home monitoring pieces that are absolutely part of population health. I don’t get excited about it relative to Fitbit and other wearables to make sure people are getting exercise. I wear one, but at the end of the day, there’s yet to be a correlation between your Fitbit data and whether there is a health issue.

More for you

Loading data for hdm_tax_topic #better-outcomes...