At the Health IT Summit in San Francisco, sponsored by the Institute for Health Technology Transformation (iHT2—a sister organization to Healthcare Informatics, under the umbrella of our mutual corporate parent, the Vendome Group LLC), attendees were able to listen to and participate in a vibrant discussion of data analytics on Wednesday morning, March 3, at the Hyatt Fisherman’s Wharf.
Shadaab Kanwal, executive director, Research & Quality, at the Oakland-based Kaiser Permanente, led a panel discussion entitled “Driving Organizational Excellence with Analytics.” Kanwal was joined by Michael Aratow, M.D., CMIO, San Mateo County Health System; Jonathan Palma, M.D., medical director of IS analytics, Stanford Children’s Health (Palo Alto, Calif.); David Kaelber, M.D., Ph.D., CMIO, MetroHealth System (Cleveland, Oh.); Kirk Larson, regional chief information officer, NetApp (Sunnyvale, Calif.); and Todd Bennett, director, healthcare market planning, LexisNexis (Dayton, Oh.).
Broad consensus was seen among panel members as to the length of the journey ahead, the obstacles involved, and the reality that the healthcare industry remains significantly behind other vertical industries in terms of its level of maturity in harnessing data and analytics to drive performance improvement. As MetroHealth’s Kaelber put it, “You have to have data, tools, people and processes, to make all this work. In healthcare, we’re just getting the data now, and maybe the tools. And in my own organization, we don’t have the people and processes to scale in place yet to really do this. So many people think of analytics as an end unto itself,” he added, “but analytics don’t deliver care, they only help people. We’re working on doubling our mammography rate, and to do that, we need the tools.”
Stanford Children’s Health’s Palma urged attendees to think about what kinds of internal partners informaticists can partner with, inside their patient care organizations, for collaborative early wins. “One important partner we have is our quality and safety department,” he noted. “When we partner with them around a defined set of goals, we can really make progress. We’re doing root cause analysis around hospital-acquired infections right now,” he reported. “Hopefully, that will lead to a change in the rates of infections.”
NetApp’s Larson, formerly a CIO, offered that, “A lot of times, people are taking a technology and looking for a problem to solve with it, when it really needs to be the reverse. And sometimes, they don’t know what they’re looking for,” he quickly added. “When I was CIO at Valley Children’s, I said, let’s first look at answering three to five key questions. And people were saying, oh, but we want to save the world! But you’d be amazed—when you focus on answering three to five key questions, you can make tremendous progress” in harnessing data to drive performance improvement and operational change.
Moderator Shadaab noted that “In the past, we’ve dealt with databases that were originally created primarily for research. Meanwhile, healthcare is still grappling with core cultural problems” around achieving consensus for analytics-driven change. How do we move from proprietary infrastructure to disease-specific databases,” for more agile data use? he asked his fellow discussants.
“In San Mateo, we’re beginning to make that transition now,” Aratow said. “The thing is that we all do use data in our daily lives, and even predictive analytics, like weather reports. And we always have our dashboards. Now we have to translate that” familiarity on the part of end-users in healthcare, with simple analytics in their daily lives, “into healthcare operations and clinical decision support.”
Stanford Children’s Health’s Palma opined that, “From where I sit in information services analytics, I see it as both an opportunity and a challenge; in terms of changing the culture around the use of data, that’s one of our responsibilities in informatics. I’d say we have a few pockets of reporting and analytics so far across Stanford Children’s Health,” he added. “And one opportunity is to help facilitate a data-driven culture.”
“Transformation happens at an organizational level. We have predictive, descriptive, and competitive analytics—all types of analytics, really. Have you employed any strategies to create competitive advantage via analytics development?” he asked panelists.
“It starts with a willingness to be totally transparent in sharing data, among healthcare providers, among payers—we’re all going to open our books and share data right now,” Kaelber said. “That’s a cultural transformation. And everyone has to agree on maybe 10 or 20 things to do” with data; it can’t be 100 or 200. We’re in a Medicare Shared Savings ACO [via the MetroHealth Care Partners ACO accountable care organization], and we’re trying to move towards building private payer ACOs as well,” he said. “This year is all about collecting and sharing data and trying to figure out what reasonable metrics we can get to in subsequent years.
Aratow said, “At San Mateo, we’re a Lean [management] organization, and that’s a foundation” for data-driven performance improvement. “You go through a rigorous methodology [with Lean management], and we use that to springboard off for our strategy in business intelligence. And specifically, in the outpatient sector, they’re focusing on one metric that they can use that they can figure out, to help them with their operations. They focus on that metric and go through a process on how that metric affects their performance. Then they request that metric from our BI [business intelligence] team, and then they use that metric to rate their performance. They’ve got about five or six different metrics they’re employing now,” he reported. “I agree, it has to progress over time; otherwise, you’ll drown in data.”
With regard to making priorities clear, Stanford Children’s Health’s Palma said, “Our guiding principle in our informatics group is to enable efficient and effective decision-making through user-friendly processes. And people will say, we need predictive analytics; and we’ll say, well, you don’t even know what your benchmarks have been yet.”
“So when we are driving analytics initiatives in organizations,” Shadaab said, “the one thing that really helps is to look at the strategic imperatives of the organization. You’re trying to look at what is more important from a business and clinical standpoint, and trying to tie your efforts to those priorities. You have a variety of data sources and have to channel in to create real information that creates that intelligence, whether business or clinical intelligence. Is there a process you follow?
Kaelber responded by saying, “In 2010, the CEO of our healthcare system put out a statement on our inter-e-mail saying, hospital volumes are down, therefore, we are not performing to budget. And everyone recognized that that was bad. And I said, as CMIO, what can I do to help patient volume? What I knew was that we were a huge referrer to ourselves. So I could look at our own data and see how many of our internal referrals were completed. In a 30-day period, out of all internal referrals, what percentage of the time were patients actually following through on what the doctor ordered? The answer at MetroHealth System, and nobody had ever asked that before, turned out to be 48 percent.”
Further, Kaelber said, “As a primary care physician myself, I was rather shocked at how low the compliance was. So we thought we could improve that measure. We had relied on the patient. It was like, good luck patient, hope you can schedule with us. So,” he said, “we shifted to a very active process where if you didn’t have an active referral within 24 hours, we would call you. And within one month, we went from 48 percent to 63 percent, and that was a big win for patient satisfaction, because it was like, gosh, you guys are calling me? And it was a clinical win, too, of course. What’s more, every one percent increase in our referral rate led to $100,000 a month in additional revenues, so that change led to $1 million a month in new net revenues.”
Meanwhile, Larson said, “At Valley Children’s, we had 200 projects and 300 systems. So we took a step back and shared with the executive team what data we had. And when organizations realize just how much information they already have—the challenge is often not, what do we have, give me more? But let’s parse what we do have. You have to leverage what you have. And that’s a real opportunity for a lot of health systems right now.”
Asked by Shadaab how strong the executive support has been in the various patient care organizations, Aratow said, “We’ve had complete executive sponsorship; and all senior managers have been Lean-trained, to understand a completely effective process, being patient-centered, and being data-driven. And one of the reasons that I think we have had as much traction as we’ve had so far, is because we’ve applied rigor to what would otherwise be a Wild West of data management.”
Looking more broadly at the U.S. healthcare system’s overall shift towards data- and analytics-driven change and performance improvement, Kaelber said, “I think people will say that the decade 2010-2020 was just a decade of installing systems and getting data. Probably, 2020-2030 will be people figuring out the tools, processes, and killer apps that really work. And,” he added, “it’s probably not going to be until 2030-2040 that the culture of healthcare will be like retail or financial services, where everybody is data-driven; and it will not be until then that the healthcare culture will have changed fundamentally, and everybody will be performing at a much higher level.”
When Shadaab asked what specific technologies have helped to move their organizations forward, Aratow said, “You have to have a good foundation, like a master patient index. You also need to be standards-based, so you’re not locked into any one vendor. And starting from there, you can build into the products in the marketplace that are also standards-based. Also, you need data scientists to do this. We’re trying to enable a self-service environment, like an Amazon of data analytics, so that everyone becomes a data scientist,” he added. “We’ve got to translate to operations. And by having that foundation, everyone becomes a data scientist.”
