For decades a leader in internal healthcare system reform, Intermountain Healthcare has blazed trails in every area of clinical and operational performance, as well as in care management, disease management, and early electronic health record (EHR) development. What’s more, leaders at the 22-hospital, 185-clinic integrated health system based in Salt Lake City have spent decades looking at patient care delivery from the patient safety, care delivery quality, operational efficiency, and utilization standpoints.
In that context, Intermountain leaders have spent decades thinking about and analyzing patterns around physician delivery of a variety of procedures, and utilization trends, both internally and in relation to nationwide patterns and trends.
One Intermountain executive who has been instrumental in all these activities is Greg Poulsen, who has been with the Intermountain organization for 31 years, and who has been senior vice president and chief strategy officer there for 20 years.
In that context, Poulsen will be delivering the opening keynote address, “Eliminating Unnecessary Care with IT,” at the Health IT Summit in San Francisco, to be held March 25-26, and sponsored by the Institute for Health Technology Transformation (iHT2). The Institute became a part of Vendome Group, LLC, Healthcare Informatics’ parent company, in December 2013. Poulsen spoke recently with HCI Editor-in-Chief Mark Hagland about the ongoing activity taking place at Intermountain Healthcare around unnecessary care. Below are excerpts from that interview.
What is the background in terms of your organization’s work on unnecessary care?
It actually started about 20 years ago. This is something that’s been on our minds for a long time. And I think that the indications were that we became aware, as were others, of clear cases of abuse, where case was being provided that was unnecessary, for financial reasons. And you’ve seen the New York Times articles on this topic, for example; it’s really hit the headlines in the past few years, but thoughtful people have been aware of it for decades. And maybe I can take some responsibility for this—about 25 years ago, everybody in the country was putting together strategic plans for joint ventures and surgical centers, and such; and I said, I think this is a horrible idea, and a dead-wrong incentive, because it gives people the opportunity to try to be paid doubly. And we made a policy decision in our organization to prevent that here. And there were a lot of unhappy physicians back then, and many ended up creating surgical centers on their own anyway. But it led Brent James [M.D., executive director of Intermountain’s Institute for Health Care Delivery Research] and myself and others to engage in discussions about care that was clearly unnecessary. We became aware, for example, of radiation oncology centers in Texas that were providing two to three times as much radiation as called for.
And we came to the conclusion that these are not victimless crimes; first of all, the taxpayer or somebody is getting ripped off. But we seriously couldn’t think of an example where people weren’t put in harm’s way while getting care that was unnecessary; they were going under anesthesia or being exposed to radiation, or receiving meds, that were unnecessary. So we came to conclusion that unnecessary care can be not only problematic, but also harmful. And we were becoming more and more through our research that this was a systemic problem.
And there were incentives throughout the healthcare system until recently, for unnecessary care.
Yes, including with regard to the practice of defensive medicine. And we started out trying to avert abuse situations. And then over the next 15 years or so, Brent and I came to the conclusion that it’s probably more pernicious than that—not that unnecessary care is being regularly provided in a fraudulent or malicious way, but rather, that it’s just osmotic—it just creeps forward. And then the Dartmouth Atlas was created and published. And one of the reasons it hasn’t had as much impact as it might have, is that they’ve tended to do it in an academic way and also one at a time, in terms of back surgery or hip replacements, for example. And they’ll come out with statements such as, for example, Buffalo, New York has the highest rate of such-and-such surgery in the country. I do remember that Provo, Utah, where I am, had the highest rate of back surgeries in the country. And were neurosurgeons paying more attention to back surgery, perhaps? But it was very atomized in approach.
So about seven years ago, we approached CMS [the federal Centers for Medicare and Medicaid Services] and asked for the entire data file, the MedPar data file, essentially the entire discharge database—a nationwide database. We had the information for ourselves, but they gave it to us for the entire country; and that’s what Dartmouth had been using. And we did it for hysterectomies, gall bladder removal, every DRG, which is how Medicare looks at this. And it was the same database that Dartmouth had been using. But we used it for every area, and asked, what’s the mean for each of this? We first looked at it in the form of standard deviation; but most people aren’t statisticians, but they do understand quartiles. So we said, OK, for each and every DRG type, the top quartile does about 2.4 times as many of a procedure as the bottom quartile. These are the top and bottom quartiles. These are huge, huge differences. And it isn’t just limited to things like back surgery; it’s true for bypass surgery, hysterectomies, and even shockingly, for appendectomies. And everybody assumes that the indications for appendectomy are clear-cut; but actually, when I talked with surgeons, they said, no, there’s a lot of individual judgment involved in that area.
When we were then able to come out and say, look, guys, this is not limited to one area, it’s true in every area. And oh by the way, they’re different for every different case type. For example, Salt Lake City is the second-lowest area in the country for carotid endarterectomies, but the tenth-highest in the country for knee replacements. And that’s true everywhere. Now, some places tend to be high more or less across the board… Utah tends to be low generally, but high in joint replacements.
And if you’ve talked at all to Brent, you know that one of his articles of faith is that wherever there’s variation, there’s opportunity.
And so we also used an entirely different database—we used some from commercial insurance, some from within our own state, and some from Medicare, to do further analysis. We used traditional analytics tools, to try to identify how much variation there was in the ways cases were treated, once a case was decided on. In this case, we were looking at hospitals and how much variation there was in how hospitals do a gallbladder removal, for example. We were not looking for the efficiency or cost of inputs; we were looking at volume, lengths of stay, related lab tests. We were trying to get to the question, if we were looking for ways to improve—if we believe the premise that variation equals opportunity for improvement, we were looking for variation within care patterns. And we found that there was one-third as much variation in the way a case was done, compared to whether a case was taken on in the first place. And while most organizations are trying to become as efficient as possible and then getting their doctors to do their hip replacements as efficiently as possible; but the data pushed us upstream, to where the greatest opportunity for improvement was. And basically, we realized that doing cases as efficiently as possible was necessary not sufficient. So we realized that we needed to put incentive sin place to make sure we’re not putting people in harm’s way. And we’re not asking physicians to change their behaviors in ways that harm them economically.
Did you make use of dashboards for the physicians?
Yes. We’ve provided dashboards, and a whole series of care practice model development steps. So we do more knee replacement here than nearly anywhere else in the country.
But Utah is a healthy state, with lots of runners and skiers, right?
Exactly. And maybe you don’t give a person a knee replacement that procedure, and they end up sedentary, with chronic disease. So being high is not bad, being low is not bad. What it shows us is where there’s big variation—and especially where one place tends to be out of the norm…
Tell me about areas where you found intense variation?
Here’s one example. Our core service area is along what’s called the Wasatch Front in Utah, which includes the communities of Ogden, Provo, and Salt Lake City. Those communities lie in three valleys about 100 miles long. And in one of those communities, we were seeing a very high cardiac stent rate, compared to the other two. So we went to the doctors in the community with the exceptionally high stent rate, and shared information with them; and they had had no idea that their utilization level was so high. So we simply shared the data with them, without telling them they were over-treating. And then, over the next year, the use rate went down to match that of the other two communities.
Did they just talk to each other?
Yes, they would see an artery that was 60-percent occluded, and they would stent that; but then after the data was shared with them, they began to shift to watchful waiting.
In healthcare, it’s fascinating, because we’re going through our Industrial revolution and our Information Age revolution at the same time.
Yes, exactly, and we don’t want anyone to perceive us as engaging in jack-booted behavior in any way. And we didn’t go in and say, you guys are wrong, and you need to correct it; we just shared with the data. And when they argued the reasons they were different, we didn’t argue back; but they looked at the data, and they changed.
Moving towards standardization and the elimination of unnecessary variation is inherently challenging. Do the physicians at Intermountain realize that eliminating variation in care, moving towards standardization, is inherently valuable, at least in some cases?
To some extent, yes; and we never want to frame this in terms of cost benefit, only in terms of patient safety or care quality benefit. And in that stent case, there was no evidence of bad care. Meanwhile, we were the first ones to take pre-39-week elective induction of labor seriously. I think we were the first place in the country that validated the difference in outcomes; we started to look at the pre-39-week elective induction issue back in September 2000. The data became clear—we had about 500 kids on ventilators each year who wouldn’t have been, had their delivery not been induced prematurely. So we sat down with our physicians and nurses and began the process of creating a protocol., in which a physician needs to obtain the approval of the chair of labor and delivery to induce electively without medical necessity; and now, that happens less than half a percent.
Has the physician culture at Intermountain shifted because of this?
Yes, I think the professional accreditation societies have helped a lot; the idea of lifetime tenure without scrutiny have changed, and that has helped. And providing information in a supportive rather than a confrontational way here, has helped, too. We’ve been trying to provide data to the physicians for years. And we now have put in place about 110 care process models; and those are areas where we and physicians, particularly the leaders in a clinical area, believe there’s enough evidence to present to create a protocol, as in the 39-week standard. For instance, there is a protocol in the way we treat sepsis. And if you don’t follow that guideline, you’re asked why. There are ways we manage ventilator care to prevent ventilator-acquired pneumonia. So there are roughly 110 of those, and are areas where we think the evidence is clear on best practices.
And the physicians have responded positively?
Mostly positively; there are always a few rough riders, but they tend to be in the extreme minority, because we’ve been careful in the areas we pick, so that the evidence is extremely clear, so that you’ll be isolated—this is airline pilots who don’t put their flaps down when they land.
What is the role of IT people and leadership in facilitating this kind of work?
It’s huge, absolutely. It’s great to have people like Marc Probst [Intermountain’s CIO], who, while technically very capable, really understands at a visceral level the importance of high-quality clinical care. He “gets” the service he’s supporting, which is really a big deal. Someone who is simply a good technical leader would be much less effective, if they didn’t bring an understanding of the underlying clinical role and mission. And we see it go in the other direction, too; you’ll have Marc Probst and Mark Briesacher [M.D., senior administrative medical director, Intermountain Medical Group], and Dr. Briesacher is a clinical person who also understands some of the technical aspects.
Any advice for our audience ofCIOs and CMIOs?
Particularly from an IT perspective, the area where people have fallen short, is where they have not kept in mind the totality of what IT can bring to the clinical setting—not only automation, but also decision support and data analytics; and if it doesn’t bring all three of those elements into the discussion, it won’t be optimally helpful.