At the HIT Summit in Raleigh, a Probing Discussion of the Value of Data Analytics
On Thursday at the Health IT Summit in Raleigh, sponsored by Healthcare Informatics, a probing discussion took place among industry leaders of some of the challenges and opportunities facing patient care organization leaders around leveraging data analytics to achieve strategic goals.
The Health IT Summit is being held at the Washington Duke Inn & Golf Club, on the Duke University campus, in Durham. On Thursday morning, following an opening address by Stephen Blackwelder, Ph.D., chief analytics officer at the Duke University Health System, based in Durham, Dr. Blackwelder led a panel entitled “Advancing Analytics and Data Best Practices in Your Health System.” He was joined by Tricia Nguyen, M.D., CEO of the Commonwealth Health Network, part of the Falls Church, Va.-based Inova Health System; Walter Kwiatek, chief academic information officer at Duke Health Technology Services; and Mark Pitts, a healthcare analytics executive at the Cary, N.C.-based SAS.
Panelists discuss analytics issues on Thursday at the Summit (l. to r.): Blackwelder, Nguyen, Kwiatek, Pitts
Early on in the discussion, Blackwelder, who had just introduced the audience to some of the advances he’s been helping to lead at Duke University Health, asked his panelists, “What sorts of challenges are you struggling with, top of mind challenges in your organizations?”
“I’ve been in multiple organizations from payers to delivery systems,” Dr. Nguyen responded, “and what I find is, the business understands what they need to do their transactional, daily business. When I meet the network team, they’re only concerned with the transactional activities of recruiting physicians into the network. But what they fail to understand is, who are the physicians we need? Their goal is to recruit as many PCPs and specialists as they can, without data and analytics to identify high-value providers. When I arrived at Inova almost two years ago, that was my first question. It turns out that the team was relying on Aetna to tell them who the high-value providers were in the network. Yet even Aetna didn’t know, because they didn’t have real data miners working in their business. In health plans, network is separate from the clinical division, and care management, and finance. And they do not coordinate. Now, they collaborate much more, because everybody is talking about high-value networks. So, one of the biggest challenges is getting the business to think about data and make high-value decisions.”
Duke HTS’s Kwiatek said, “One of the interesting things to me is teaching the business of research to researchers here at Duke. Typically, data assets are tied to a specific brand or type of user. Figuring out how to address our assets differently or uniquely, is driving our culture change.”
“I have a unique perspective because my role at SAS is to support the largest payers, providers, and life sciences in the world,” Pitts said. “I’m also married to a provider, who’s seeing patients right now, so I hear about the challenges of using data in patient care every day. And the core problems are not technical; the challenges are really human. The challenges I’ve had throughout this industry are almost entirely human; it’s human resistance to change, a lack of understanding among folks about how to get things done. It’s about, don’t ask me for another report, let’s figure out how to change processes. So for me, it’s a human challenge: it’s how you educate people, get them to collaborate across different elements of an organization, and drive collaboration across IT, business, and clinical folks, and pursue change.”
Blackwelder continued the discussion, saying, “I had been in a health plan and spent some time in a multispecialty ACO, done some work for a state government. Duke was the first organization I’d been in that had really unleashed the idea of empowering people to create change. That’s partly because it’s a research university. And those of you who’ve been in vertically integrated organizations that are top-down, this may sound a little strange or that it involves in political backbiting. But at Duke, I really have the opportunity: someone could propose an idea for a project, go to a bunch of folks and develop a proof of concept, and if that proof of concept is accepted, it can become a part of the organization, and we can see what happens. In terms of that sort of “Shark Tank”-y type of activity, if I’m able to be helpful to that person, we’ll collaborate; if not, he’ll go to Tricia, because she can be helpful. And my job is about making sure there’s a solid foundation under everybody’s feet; and that means that I’m really motivated to collaborate with that person. I’d rather that we created tools that played into a scaling fashion. We talk a lot about collaboration in an enterprise organization.”
Kwiatek noted that “There are things we do that create value today, and things that create value tomorrow. There’s this hurdle in the middle, this chasm, around scaling up. Creating a partnership around the scale-up function is critical to us: how do you take something from a model that might work in an isolated area, and scale it up to create value organization-wide?”
“In a large integrated health system that’s not just a provider but also an ACO [accountable care organization]—it started with the CMO and CIO, agreeing that there needed to be some form of structure around data management and governance, so they put together a structure; and they were relatively mature when I arrived, in that they had governance structure around the EHR,” Nguyen said. “But the claims and lab data weren’t integrated. So we asked, could it be integrated; so they expanded the governance to include the team outside the clinical data we’d been governing. And what worked well was they had data storers—what you called data janitors or custodians. And then they could explain what the business needs were and explain it in technical terms that the data storers and data miners could understand, so that if there was a question that came up, they could explain it to them. But it all revolves around data governance and structure. But it has to come from the top down.”
“You’ve seen the CVS-Aetna news,” Blackwelder continued. “Some of us are struck by how familiar some of these new ideas sound. The way in which the CVS-Aetna partnership plays out could be interesting; they each own some direct-to-consumer elements that wasn’t the case in the 1990s. And outside healthcare, you look at Lyft and Uber and Facebook, etc. And I’m going to throw out to you the question, is this something we should embrace, or be concerned about? Is it inevitable that someone will figure it out? I remember everyone was concerned when Google and HealthVault got involved. And then Apple was going to get involved in healthcare. Maybe it would be something that would be a flash in the pan, or maybe something that we should help to bring into the business?”
“Google is one of the greatest machine-learning organizations out there,” Kwiatek noted. “I think that technology related to providers is ultimately going to happen, but there will be a period of transition of many years; I don’t think that Apple is going to be providing healthcare anytime soon.
“I truly believe in the retail-ization of healthcare,” Nguyen testified. “You look at CVS and Walmart, they have large investments in HC. Ultimately, HC will be put in the hands of individuals. There are startups doing amazing things. There’s a company called Dermpath, and you put a patch on your skin and send it off to a lab and they tell you if it’s malignant or not. That requires data and analytics. And if you can diagnose yourself—that will require a lot of machine learning and analytics. But that requires education of consumers. Right now, all that knowledge is housed in the minds of physicians and clinicians, but the data will be freed to create more predictive and descriptive information that will help facilitate clinical team workflow, and will be translated down to the consumer level. Consumers will eventually be able to take care of themselves. And you can’t do CPR or resuscitation at home, and that’s when you go into the ED. But much care will happen at home.”
“And what will things look like in 10 or 20 years?” asked SAS’s Pitts. “We’ve gone about it the wrong way in healthcare, compared to how other industries have done it. In the Blues plans, we’ve tried to do things creating databases, to see who’s a four-star or five-star provider, who’s costly and not? But the problem is, we don’t use them [those quality rankings systems]. And as a consumer, my challenge is getting an appointment. So I think we need to use some of the same types of data strategies that Amazon and others have used outside healthcare. For example: when do I get my haircut at GreatClips? The app tells me, here are the three GreatClips places near me, here’s the wait. And you know what? I didn’t check to see quality ratings or price ratings, I went to a particular place because of the convenience. Now imagine if consumers had that information about providers? My challenge has always been, if I know the highest-quality provider, as a health plan, how do I get the patient to go to that provider? You need to provide analytics to consumers that speak to convenience; you embed that technology into a process in a way that consumers actually find it valuable.”
An audience member stated, “The one thing we’re missing here is the patient experience. While they’re waiting in the waiting room, in the doctor’s office, there’s no reason they can’t participate in their own care. And getting access to that data could take place while patients are waiting.”
“Do you have somebody who wakes up in the morning every day in your organization, who’s thinking about those problems?” Blackwelder asked. “And you don’t have such a person, do you want one? Or don’t you want one?”
Pitts offered, “Jeff Hammerbacher, founder of Cloudera, said, the best innovations involve getting consumers to click on apps. We can apply some of those same learnings to healthcare.”
“We’re looking at how we integrate our marketing and clinical teams,” Nguyen noted. “Marketing has traditionally looked at promoting services. Now we’re looking at integrating the data they’re using, to segment and to introduce messaging. And we’re looking to see how many clicks patients need to go through, to see the messaging we’re trying to send. All of those end up as data points that we can use to better define messaging, for marketing or for clinical engagement across the population.”
And, in response to an audience question, “What are we competing over” as patient care organizations? Nguyen replied that “Competition really depends on the seat you’re in. If you’re the CFO, you’re competing for the revenues. If you’re the CMO, you’re competing on quality. For me, as the CEO of population health, we’re looking to be successful in managing the costs of care while trying to achieve the Triple Aim. In northern Virginia, we have probably the three top zip codes, income-wise, in the country. And 70 percent of our population is covered by commercial, employer-based health insurance. And we’ve got 70 percent of our market. And we’re so big that we command a higher payment from payers. So some of the health plans are trying to create narrow networks to steer patients away from us. And while these are rich people, we thought some would pay out of pocket to stay with us, and even with this particular payer with 35,000 lives in a commercial ACO, we started to see some fall-off. So competition around effectiveness—delivering care at an efficient, effective cost, is beginning to heat up. There are people who will pay for convenience, but that’s not everyone.”