When it comes to sharing data, information, and insights with physicians around the principles and practices of evidence-based medicine, studies are finding that physician education can only do so much to improve patient care (and only for so long). John Kontor, M.D. was one of the speakers to address the subject at the chief clinical executive summit, held last month in Orlando, and sponsored by The Advisory Board Company, Washington, D.C. Dr. Kontor, executive vice president for consulting at The Advisory Board Company, spoke with Healthcare Informatics Editor-in-Chief Mark Hagland in October, just prior to his scheduled speech at that event.
In his interview with Hagland, Dr. Kontor referenced one large integrated health system in particular that has obtained exceptional results from reducing variation in care. For example, that health system’s leaders, by reducing variation in the care of percutaneous coronary intervention (PCI), have achieved $2.5 million in avoidable costs, and avoided 247 days per year of inpatient stays, and averted 9.27 deaths per year, all compared with other organizations that are in the top quartile nationwide of most expensive cases in the PCI area. Those kinds of results, Kontor argues, show that, when deployed and managed well, evidence-based medicine initiatives can document firm results. Below are excerpts from the interview that Dr. Kontor gave Hagland last month.
Tell me a bit about your current research and consulting in this area?
Let me start at a high level, in terms of what we’re seeing evolving across the industry. Much of what has driven our focus has been similar to what we’re seeing on the provider side of the industry, particularly with the evolution towards value-based purchasing models. There’s increasing recognition that the care that organizations are giving is still not achieving the quality outcomes that folks are hoping for, and certainly, as seen through a value lens, not worth the money that payers are spending. And that’s why evidence-based care delivery is emerging as such an important phenomenon.
We look at both broad and specific measures. One example is interventions around percutaneous coronary interventions (PCIs), or stents. In that focused area alone, most hospitals or small health systems even, have multi-million-dollar opportunities to reduce inappropriate spending as well as to improve outcomes.
Essentially, stents are overly broadly used? Or is there inadequate scrutiny of their use?
Yes, over-utilization generally, and the main issues more specifically are around drug-eluting versus bare metal stents. And there are pretty good guidelines available.
Drug-eluting stents are not justified over bare-metal stents in many situations, then, correct?
Yes, that’s correct. And as organizations move into value-based models, they begin to look at these areas—both in terms of resource utilization and quality outcomes. So clearly, we’re seeing an acceleration of organizational attention to this, both because of value-based purchasing, as well as bundles coming along, as well as PAMA [the Protecting Access to Medicare Act of 2014], the outpatient imaging regulations that are requiring—there are a couple of components of that—but it’s requiring organizations to essentially use decision support to pre-authorize decisions around high-cost imaging studies. There are specific requirements around decision support, and this is one of the areas where health systems are saying, aha! I see how decision support in real time can help drive out some of the inappropriate ordering.
So that’s influencing thinking around decision-making in areas such as stents?
That’s correct. What we’re seeing is sort of a J-curve around interest in decision support, as organizations go into EMR implementation. EMR vendors focus on this, and there’s some evidence in the literature stating that decision support can influence decision-making. So they’re doing excessive turning on of alerts. And then folks start shutting down decision support like mad—in some cases, they’re turning off all the alerts that providers see, which is not the right answer, either. But what we’re starting to see now as most health systems have done at least an initial implementation of an EMR, as providers get more comfortable with using an EMR, the health system leaders are starting to look for value, and so we’re seeing interest in evidence-based clinical decision support.
Tell me a bit about the different areas of interest you’re seeing?
There are three areas that we’re seeing. One is quality measures. Again, with star measures, ACO quality measures, opioid control, etc., folks are increasingly looking towards decision support to drive performance improvement; the second is around risk capture. HCC and similar risk-adjustment methodologies—HCC stands for a risk adjustment methodology that CMS [the federal Centers for Medicare and Medicaid Services] uses for all their programs, including Medicare Advantage and all the MSSP, NextGen, CPC+… Hierarchical condition categories—it assigns a risk adjustment multiplier to future payments around populations. They make the calculations for each individual patient, roll that up into entire populations, and put a multiplier on it. This is one of the areas that is top of mind for any leaders of any organizations involving in any of these risk models. Right now, they’re not doing a good job of capturing the risk profiles of their populations. So they’re looking at HCCs to improve their performance. And the third area is around driving out inappropriate utilization, looking at high-cost, high-risk procedures and diagnostics, and looking to use CDS for evidence-based guidelines, and consensus-based guidelines when evidence isn’t available.
Can you offer a few examples of how this is playing out in specific organizations?
I’m presenting tomorrow at a conference in Orlando with representatives from a large health system in Ohio that we’ve been working with. There are internal drivers for them around this, and they’re putting a really focused effort around decision support. And there are a couple of areas they’re focused on; one is developing internal governance and management processes around decision support. In many organizations, someone might be assigned to certain tasks, but there isn’t much governance or management around these processes, so that takes specific work. The second thing they did was to leverage analytics to drive the optimization of a performance improvement strategy. And the third piece is to use analyze provider performance and organizational performance around specific alerts or disease categories.
An example of leveraging analytics in that area has been around stress ulcer prophylaxis. We know that stress ulcer prophylaxis is over-used in the hospital, and it has direct adverse clinical impact. Many hospitalized are at increased risk for developing ulcers related to the stress of illness and medical care. And there are standardized protocols that have been developed to reduce the risk of certain populations that are at high risk for stress ulcers, that have been around for years. Now, though, we’ve seen a swing towards stress ulcer prophylaxis overuse. And there’s a high risk of acquiring c diff—clostridium difficile—through prophylaxis. And many organizations overuse stress ulcer prophylaxis. They’re spending money unnecessarily and are putting patients at risk. So this hospital used analytics to understand which providers were inappropriately ordering stress ulcer prophylaxis, and to understand where in the workflow that was happening. So they were able to target their work so that 27 percent of those inappropriate orders were coming from their low-risk chest pain rule-out-MI order set—they were able to see that that was the most common culprit for where these inappropriate orders were coming from. So they removed that order from the standard chest-pain order set, and overnight, literally, the numbers improved. There were 80-100 inappropriate orders per day across a large health system (18 hospitals or something), that dropped to 20 with this one change.
What kinds of process, cultural, and organizational change issues are there around this? And those would include governance issues, correct?
Yes, so the governance piece—the organizations most successful with their CDS strategy, have an intentional governance process that’s also very efficient and effective in managing process, to really drive results. One, organizations have to prioritize decision support and devote prime resources and attention to it. Two, there need to be groups within the organization that have as at least part of their primary role, to manage decision support. And the third piece we see in highly effective organizations is that they also empower their teams to be nimble. Because some organizations get too laborious around this, and the whole system grinds to a halt. And then you also need to engage the right people around the effort—clinicians, pharmacists, finance people, risk contracting groups, quality people, your IT team, your data and analytics folks—all of those folks are key stakeholders and need to be involved. And tactically, organizations need to understand the opportunities where they can focus their decision support efforts, and they need to understand the impact of implementing certain alerts. Cedars-Sinai has done a great job of developing a large library of decision support rules that they essentially switched on silently in the background. They develop rules, switch them on silently in the background, and then develop an organizational understanding of whom they’re developing these alerts for, how those clinicians will respond, and how it will affect care processes and outcomes.
What should CIOs and CMIOs be doing around all of this?
Decision support is one of those areas where, if health systems don’t spend more effort around governance, process, and technology, we’re going to start to see increasing external mandates that will force organizations forward, via governmental regulations and so forth. So CIOs need to create a plan around decision support, make sure they engage senior executive and clinical leadership around this, and make they’ve got the technical capabilities to use data and analytics to drive prioritization and performance around decision support.
The good news is that PAMA, that imaging regulation—is one of those things that make people realize they have to move forward. Similarly, the total joint replacement bundle. So—take one of those situations that’s going to force you to move forward, and exploit that to figure out how to get good at this.
This area seems to speak to strong CMIO involvement in helping to lead the leverage of data, information, and processes around data and information, to improve clinical performance, correct?
You’re exactly right; this is a natural sweet spot for CMIOs, to be able to do two things: one, to drive improved clinical and financial performance in the organization; and two, it’s an opportunity for CMIOs who have not yet been in the lead in this area. I’m a former CMIO; and I can say that one of the challenges CMIOs face now is that many have not yet made the leap to become recognized, empowered, strategic leaders of organizations, and they’re still stuck in the EMR implementation shoebox. And this is an area where they can become more empowered in their leadership in their organization.
How quickly will all of this work accelerate across U.S. healthcare?
We think this will accelerate pretty quickly. I’ll get you a quote around this. We’ve got some projections around what the decision support market will do over the next 3-5 years. But we think there will pretty quickly be a dramatic J-curve over the next few years, as health systems put more emphasis on their decision support capabilities. And related to that, you’ll see the growth in a whole cottage industry of decision support vendors. You’ll see continued growth of that industry, of companies providing decision support tools; and the next big thing, which is probably five or more years away, we’ll start seeing more targeted or personalized decision support that is customized to individual providers and individual patients, based on their genetics and so forth.
