TranscriptCatalyze: I wanted to thank Chris Belmont for joining us today. Chris is the VP and CIO or chief information officer at University of Texas MD Anderson Cancer Center. He joined MD Anderson in September of 2013, after serving in a similar role at Ochsner Health System, also a very large system in New Orleans, where he was there for seven years.
Chris has more than 30 years of healthcare IT experience from various positions in IT leadership, sales, consulting with Siemens, Healthlink, and IBM. Chris is now leading MD Anderson through a system-wide Epic rollout. And today we're going to be talking about proof-positive impact on healthcare innovation.
Catalyze: I spoke to somebody else at Ochsner as you guys were going through the Epic rollout. I don't know what that was now, five, six, seven years ago? That was a big, at the time, it still is, a pretty big Epic rollout. I'm sure there's a ton that you've learned there that helped you with MD Anderson. And looking specifically at metrics: Peter Drucker famously said, if you can't measure it, you can't manage it. I'd be curious how that affects your work at MD Anderson and the innovation that you guys are bringing forward with Epic and then probably with additional solutions within the enterprise.
Mr. Belmont: From a couple perspectives, so it's the if you can't measure it, you can't improve it. There's a, but probably the more important impact on a project like this is when you install a new tool that's got more access and more availability to information, what it tends to do is put a monster spotlight on your organization from a process perspective, from a behavior perspective.
The eye-opener for us at Ochsner, and we're already starting to see it here, it just puts a lot of focus. For example, we realize that some of the physician behavior was not where it needed to be, things around reviewing their charts, putting their charges in, among other things.
What a tool like this allows you to do, and it's not specific to Epic, it's really any of the more modern EMRs or any system today, it kind of gives you the data to say, wow, this is how we're performing against the way we do business, and it forces you to modify processes. It gives you the opportunity to modify processes, because now you have the data to say here's a baseline. I changed the process. How did it improve on it? So that's what's most exciting.
Just like at Ochsner, here at MD Anderson, we have a legacy system that was homegrown, and the focus was on the learning and functionality and not necessarily on the metrics and the backend. With this tool, we get kind of everything, but we get the benefit of improving the frontline care and then the benefit of the data so that we can modify and improve and tweak our processes, and also assist with the behavioral changes.
Catalyze: Yeah, that makes a ton of sense. With this specific Epic rollout, are there any specific use cases you can point to since you've been at MD Anderson where you've discovered those process issues as part of this rollout?
Mr. Belmont: The one we're working on right now, and the reason I'm not in my office, and I'm sitting in a common area is we're really hashing through the fact we discovered there is a lot of variation in the way we view patient outcomes. From the way patients enter our organization, we discovered that every one of our centers view it slightly different.
At MD Anderson that actually originally was a strength, because we focused on a particular type of cancer. When you go to the, say, the breast center, that's what they do. They wake up every day, and they manage breast cancer. They don’t manage cancer in general. They manage breast cancer. They were a little autonomous, and they did their thing their own way.
When it comes to having an experience at MD Anderson, we realize that there's a lot of variation, and patients sometimes bounce between the centers. Now we have an opportunity to standardize. That was the biggest aha moment. We don't go live with Epic for another year. What's going to happen as we get a little closer in and go to training and start doing the user testing, we'll discover more and more these situations where there's a lot of variability in the amount of processes.
I'm confident we'll get there. It's not just technically hard. It's more politically challenging, and people say, okay, I'm going to give up the way I do it and adopt the MD Anderson way. Defining the MD Anderson way is the most challenging, I would say.
Catalyze: Just to make sure I'm clear, when you talk about MD Anderson facilities, you're talking MD Anderson now? It is much bigger than the University of Texas, the original MD Anderson. When you talk about facilities, you mean sort of all of the branded facilities that MD Anderson is partnering or launching, different places?
Mr. Belmont: Yeah, not so much, but within MD Anderson, our primary center is here in Houston, but we also have some remote centers. Even within the primary center here, we have a breast center, we have a head and neck center, we have a lung center, for example, leukemia. Each one of those has, like I said earlier, been operating somewhat autonomous. It makes them strong, because again, that's what they focus on. But then when we try to look at the institution horizontally and not in more of a vertical fashion, we can see the variation.
Catalyze: That's helpful, because I guess it's even within the center, like you said, but it's head, neck, or breast, or in your logic, whatever it might be, they all sort of had their own ways for entering the system.
Mr. Belmont: Absolutely. There was no need to necessarily interact, because again, they stayed within their own little domain, if I can use that term. Now that we're looking at using a monolithic, single system with a lot of standards, that's what we're focused on now.
Catalyze: Actually, specifically looking at that I had a question sort of generically about metric impacts from something like an EHR rollout, in your case, Epic.
Mr. Belmont: It gives us very granular data, and then we can supplement it with other transactions, so it's not purely just in Epic, but more. For example, we're looking at tracking patients using RFID-type technology we call an RTLS technology, so very much like Disney. You can go to Disney, wear the Disney band, and they know where you are at all times. So we're kind of looking at that level of information. Right now, we're looking at something simple around, say, a clinic. How long does it take from the time the patient shows up in the clinic to the time they depart the clinic?
We kind of know by hunch, but now we're looking at specifics. When did the patient actually present to the registration desk? When did they actually go into the room? Who interacted with them in the room? How much time did they spend? When did they actually exit the room, and when did they actually leave the clinic? And then once you get those processes down, you can go bigger.
One of the thoughts is how do I know when the patient left their house, for example? How do I know when they hit the parking garage? If that patient is in a parking garage, I know analytically that it takes about 20 minutes to travel from that parking garage to the clinic, and if the patient is running tight or running a little late, just like air traffic control, I have the ability to bump something, somebody in front of them rather than either give up their spot and reschedule or throw the clinic off. Having that data allows us to get a lot of interaction, and then do the process over.
One of the things we realized at Ochsner, the two biggest resources in the clinic visit are the physician and the room. We found we were doing a lot of non-value-added steps in the room, so things like booking the next appointment, printing the after-visit summary—those are things that could easily be done outside and then open up the room and that physician to see one more patient. But I think that's the level of granularity we can get to. MD Anderson is in such demand, for us to be better at resource utilization is extremely critical, just so we can see those patients that really need to be here.
Catalyze: Who are you guys using for RTLS?
Mr. Belmont: Right now, I'm going with AeroScout, basically, Stanley Healthcare, currently for tracking a lot of our assets like pumps and things like that. But we're hoping that we can go with a single technology, and whether it's a fixed asset or a moveable asset or a human asset, where we can leverage the same technology. That way we don't have to run a lot of different networks and so on.
Catalyze: That makes sense. When you look at something like a Stanley or some sort of RTLS provider, will it eventually be fully integrated into Epic, or will Epic integrate with Stanley? How do you view those two systems?
Mr. Belmont: I just had this conversation with them two days ago that I'm interested in their technology. There are some applications they have. I'm really interested in the transactions they generate. If you want to talk about how we plan on integrating Epic, one of the challenges is room turns on the inpatient side, not just outpatient. But when did the room get vacated? When did the patient actually leave? When did housekeeping get in? When did the room become ready status so we could put somebody in there?
So again, if you use an air traffic control kind of model, if the patient has a tag on their arm, for example, or an arm badge, if we remove, if the patient leaves the room, and we cut the tag off, and we drop it in a box, you know, Stanley knows, or the RTLS system knows that when the tag hits this box, and it could be a very small, cigar-sized like box, the minute that it gets in there, put a transaction in Epic that says patient is discharged. Notify housekeeping. Start the whole process.
In the old days the nurse had to, oh, wait, the patient left, so let me go into the system and turn it to discharge, and then the process starts. You can save potentially hours. This is totally hypothetical. My wife is a nurse, so I'm not going to tell you it's true or not, but nurses, the shift, are not really excited about turning a room and getting a new patient in, because the new admit process is lengthy. Not to say they do that, but just, hypothetically, that could be a situation that occurs.
And then if you look at the process leading up to a transaction, if I needed a wheelchair, instead of me searching and going downstairs, potentially, to the barn where all the wheelchairs are, I can go online and say where is the closest wheelchair, and it could be in the room next to me. And that way I can cut down a lot of these, you know, wasted steps right?
Catalyze: I agree with you, and I think that's one of the really interesting things about what you can do with RTLS and RFID. I think the possibilities are almost endless.
Mr. Belmont: It's just a technology to put in your portfolio that you can leverage in so many ways, just like you said.
Catalyze: Yeah, exactly. That does relate to terms of strategy for integrating with Epic, or that view of Epic in a broader ecosystem of technology and application and transaction.
Mr. Belmont: One of the things that I talked to the leaders at Epic, and I've been talking to them now for ten years is that from some perspectives I just see them as a transaction generator. Let's say the EHR is not an end-all, be-all. It doesn't do everything. It is the hub. It's where people interact. But you've got to augment it with additional information.
For example, Epic has the ability to send an SMS message, but you've got to put the surrounding technology for Epic. I see RTLS just as a complementary technology, and there will be more to come as we go forward. The tools are all there, and the integration points are all there.
Catalyze: When you look at Epic and integrations, are you looking at more of a traditional HL7 integration, or are you looking at some of the newer stuff?
Mr. Belmont: We will standardize on HL7, but there's a lot of other entry points in Epic that, you know, the, years ago, even when we began the journey with Epic at Ochsner, it was always this rumor that Epic was a closed environment, and you couldn’t interact with it. They've actually flipped that around, and they have quite a few APIs. And already, again, one year before go live, we had a lot of our legacy systems exposing data through the APIs right within the Epic environment. It's, the integration points are very easy.
We'll use a combination of HL7 and other standards, but our goal, really, is to stay away from any custom programming. But you still have to do a little bit. You've got to focus more on the customer and not the challenges related to the integration. It's fairly straightforward. You're never going to move all of the radiology images into Epic, so therefore, you've got to have easy integration. Your PAX viewer has to be available in Epic. So that's not necessarily the HL7. It's through an API.
Catalyze: That makes sense. Also curious if you'll tell the story, whether you're talking about it, but there's a recent story talking about the new Epic app exchange.
Mr. Belmont: I'm not so familiar about that. I like the concept. It's more of a self service type situation, I'm sure. That's one of the things that I've got to get a little more current. They're moving so fast, I've got to get caught up. I can't just hang on my Ochsner experience, which is two years old, because they're moving very rapidly.
Catalyze: They are. It's all speculation at this point. It's something that will resemble an Apple-type app store. Of course, it's healthcare, and it's PHI, and so it's not going to be the same setup, I'm sure. That's what they're talking about for the app exchange. And it will be interesting to have that technology.
Mr. Belmont: You know, it, what I love about them is they are very innovative and they get out in front of things. Sometimes they're, you don't really know until it's there. So it's hard for me to plan for some of this stuff. Like I don't know if it will ever get back to them, but I'm, you know, I always wanted to know were ever they, were they ever going to get into the hosting business. Oh, you know, and I kept hearing, no, we're never going to do it.
And then at UGM here, they had the coolest datacenter I've ever seen. It was like, you know, you've been working on that datacenter for three years at least, so give me a hint, or give me a wink, or give me a nod that says, no, we're not doing it, Chris, and I know that you are thinking about it, at least. So, you know, their goal is not to over commit. You know, I understand that, because I was on the vendor side, and, believe me, I know what it's like to over commit.
Catalyze: We're based in Madison, not far from Verona, so we've been hearing about the datacenter for a while. It is interesting to see the evolution of Epic, especially since it's been the winning EHR, especially on the enterprise side.
Mr. Belmont: Yeah, I usually like that they listen to their customers, and they take the input and move that right into development. They don't develop what they want to. They develop, really, what we need, and that's so refreshing.
Catalyze: You're working on this large Epic rollout, to go live not for another 12 months at MD Anderson. At a high level, are there a few key metrics for this broad-based EHR rollout that you're tracking? Or maybe in your case, the board would say for you, okay, this Epic rollout has been a success?
Mr. Belmont: Yeah, there are. Obviously, we're going to chase the quality metrics and everything else we need. I think the biggest thing is to understand your metrics today and if they will change post-live, though the way you, for example, the way you count the visit today may not be the post-live Epic way of counting a visit. And that was an aha moment for us.
At Ochsner I remember one physician saying, you know, Chris, your system is terrible. I won't use the real world word he used. Your system is terrible. When am I going to get back to seeing 75-patients a day? And I said I'm not sure you really did see them. But he got credit for all these visits, so, for example, if he did an injection, he got a credit through the visit.
Getting comfortable and building that baseline and understanding that dictionary and defining that crosswalk between pre-Epic and post-Epic metrics is really critical. We have a team of about 19 right now here at MD Anderson, because in all my EMR implementation, including the vendor side, I would argue that reporting the metrics piece has been my most disappointing.
This time, which is my last, by the way, I'm throwing a lot of folks at it. And the team is really focused on saying, okay, how do we transition people and organizations in understanding how we measure things today and how we're going to measure them later. If you go on the revenue cycle side, it's going to be things like denials. It's things like recovery rate.
It's obviously, the AR days. Now because Epic will kind of force you to have a cleaner claim coming out, but now it allows us to move further upstream, and instead of working on errors like denials and rework, so now we're going to move our attention more up front and say how do we get a cleaner claim out, faster. We realized at Ochsner it took longer to get a claim out in the early days, but they were paid much quicker, because they were cleaner. So again, how do I shorten that? How do I, instead of paying attention on how will I reduce errors, how do I get the claim out there?
Catalyze: That metrics group, you said there were 19 people? I assume is it like a cross-sectional grouping, so you have maybe people on the financial side, on the clinical side, on the operations side, on the IT side?
Mr. Belmont: Yeah, one of the things that Epic gained in their implementation methodology that I really, really like, at Ochsner we were one of the early, big bang customers. We were in one of the first wave of the first year of the big bang, where you can do all Epic at once, which is what we're doing here at MD Anderson.
But their old methodology, you had an individual report writer, the metrics folks, in each module. You would have one person sitting with the, say, the red cycle, the actual billing team, and you would have one sitting with the clinical documentation team. In the new methodology, you take all your metrics and report writer types and put them all together, which is a much better approach.
One of the things we did, even though we got the metrics out at Ochsner, when we put them all together, they didn't look the same. This way, we have a common look and feel, a common understanding. We have a capacity of 19 versus 1 person in each module. I like this approach a lot better, and it shows a lot of promise. I think we're going to be much more successful in this space, which will, again, make me feel really well, because that's not something that I've done right in the past yet, or I could have done better. I did it right. We just didn't execute right. (Laughter)
Catalyze: Was this group of 19 put together as a part of this Epic implementation or the sort of EHR implementation?
Mr. Belmont: It was. And we went for more data scientist types and not more report writers. So they are more interested in the analytics side. Because, you know, one of the things I've discovered over my career too is good report writers aren't necessarily good analysts or necessarily good data scientist types. These are more heavy on the analytics side and less on the aesthetics of writing a report. Because remember, Epic is not just about generating two-dimensional, paper documents that you print, but there's dashboards, there's drill downs, there's work lists.
Now they have much more analytical tools, which is more of a data warehousing than the analytical approach and less of the recording. I can find people that can write crystal reports. What I can't find is people to make sense out of data and key it up for a user that says, hey, you know, you need to pay some attention here.
Catalyze: Yeah, actually make some meaning for that, from that data, right?
Mr. Belmont: Yeah, instead of, again, just regurgitating data, let me give you some valid information that you can act on.
Catalyze: Do you think about that group of 19 as something that isn't just for this implementation? As things evolve, both at MD Anderson as well as more broadly in the health system, do you think this group of 19 could starts with the EHR implementation but grows to much bigger, because the focus is on how to become a true data-driven organization?
Mr. Belmont: Yeah, you're spot-on, and, by the way, if you're ever interested in leaving Wisconsin and coming to Houston, come work with us, because you get it. The reality of it is you're right. It's not about generating reports. The whole thing about an Epic implementation or any implementation like this, is you need to view the go live as the beginning.
I tell our team all the time, we're not focused on replacing the legacy system. We're focused on changing the way MD Anderson does business. Yes, we have to replace the legacy system. But then after that it's, like you said, about optimization, and how do you take a powerful tool like this that's generating unbelievable, valuable information and getting that back into the business so that we can get on with our mission curing cancer? And that's our primary focus. The go live will happen. We will replace our legacy system, but shame on us if we just turn it on and say we're good.
The optimization piece, if you don't mind touching on that, is critical. I was a little late in getting that team started at Ochsner. What we're doing now is identifying the people that will fit on the optimization team. If you think about where we are today, we're kind of heads down, building the system, getting it ready for go live.
The people that are great at building may not be great at things like customer facing activities. People that can go into it and say, you know, I've been looking at your metrics, and I look at your physician behavior, and and this guy could really use some help. I'm coming to you proactively versus us discovering that they have a problem, and I'd like to focus in on that. Or I noticed that your clinic group that might be a little out of the norm from the other clinics, because now I have my hands on all the data.
Instead of saying this clinic is in the ditch, let's go in there prior to them going in the ditch, and let's help them optimize their business. But it's a different skill set and probably a different person. It's more of an attitude than it is a technical expertise.
A builder may not port over to being a good optimization person. Right now we're doing is looking at this group of—we have about 200 folks in the Epic project—looking at that group of 200 and saying how many of those will move over to being more in the optimization role? Because we still need builders, and we still need technical types, but we probably, the benefit to the institution is the optimization team that you mentioned.
Catalyze: You have a lot of experience in this, so you know. It's definitely interesting talking, as you are still a year away, already thinking about optimization and what that's going to look like. I think that is different than a lot of groups that we've talked to, so that's very impressive.
Mr. Belmont: Well, the other thing is, when you're driving transformation, one of the things is continuity. You don't want to build a relationship in the build phase and say, you know, here, let me understand how you do business, and then hand that off to someone else.
If we can build this continuity that builds a relationship pre-live during training, during the go live, and then optimization, and it's like, okay, where is Joe, or where is Chris that can come in, because they helped me in the build, they helped me in the design, and they helped me in the go-live, and now they can help me in optimization. You build a relationship, and then you more or less embed those resources into the department or the divisions.
Catalyze: Shifting gears a little bit. We work with a lot of startups that are building applications or platforms for bundled payments or telemedicine or patient engagement, etc. As you look at newer solutions and Stanley is one, but Stanley is an operational tool.
But if you look at sort of extending the EHR, tabling the discussion about the Epic app exchange, how do you guys think about working with additional third-party vendors that are generating, like you said, just additional data points, and I assume, additional transactions, and have better analysis, whether it's how long it takes for them to get from the parking lot to inside or post-discharge or whatever it might be?
Mr. Belmont: You're touching on a challenge that we have in IT. While we're going through this transformation part of you wants to say don't give me any more information. My plate is full. But things are changing so quickly, and the benefit that can be derived in the rapid development cycles today, you can't ignore what's going on out there.
MD Anderson has got a huge research arm. We do a tremendous amount of research. In fact, we get more grants than any other cancer institution, for example. How do we get the MD Anderson experience into other's hands so you don't have to come to MD Anderson to have the MD Anderson experience?
Let's say we are this location for people that really have to be here, either they need a particular surgeon or piece of equipment, or they're in a situation where they really need to be in Houston. But how do we allow others to have the MD Anderson knowledge and experience?
We're using all kind of technologies. We're partnering with AT&T. We're partnering with IBM. You know, we're one of the early adopters of Watson here, and we're taking advantage of that to try to teach Watson to think like an oncologist with a lot of good success. It's very interesting, but it, obviously, requires a lot of data. Well, so we're open to a lot of technology. Epic is just a part of the journey. We partner with so many different organizations.
We're looking to drive telemedicine, and again, get outside of the four walls, so it's not about coming to Houston. It's about MD Anderson experience. One of the researchers told me, and I'm new to the research world, I didn't really appreciate it until I got here, but they said the knowledge we have today on curing certain cancers or treating certain cancers doesn't become common knowledge in the world for about ten years.
It has to hit a journal. It has to go through a trial. You know, it has to be distributed. Somebody has to read it, and you have to see that type of case or a particular number of cases before it becomes common knowledge. So how do we compress that ten years? What we know today that's very effective, how do we get that into somebody's hands immediately so that a patient can have that knowledge and experience without waiting for that traditional process to work its way through?
Catalyze: I think that I've read some of the numbers on how long it takes for evidence to get actually into practice, and there are all sorts of issues related to publishing in a journal, reading it, and then actually integrating it into a training program where people come to learn, you know, how they practice medicine.
Mr. Belmont: About a third of our patients come from the Houston area. About a third of our patients come from Texas. But then a third of our patients come from worldwide, and our worldwide reach is growing every day. So how can we start interacting and doing things differently?
We've got our partnership now. We work our cancer network that we have, so we have a partnership with close to, I think, 20 organizations now, whether they're just physicians exchanging consult information or a patient having the MD Anderson experience like in Phoenix and in New Jersey.
Catalyze: We touched on something else which I wanted to ask you about. You mentioned hosting in Epic. Traditionally healthcare has had everything on premise, and so they run big datacenters, and they host a lot of their own applications. I remember years ago I was working at Catholic Health Initiatives, and they had like 900-something applications to get in their datacenter. Is MD Anderson looking at other options beyond on premise, so a mix of private Cloud or public Cloud options for technology and applications, either you guys host or even third parties that you work with?
Mr. Belmont: Yeah, we are. In fact, that democratization project that I was just talking about is how we distribute our knowledge elsewhere. It's going to be purely Cloud-based. We're not necessarily, I'm not necessarily a fan of picking the technology or the delivery mechanism first. I would prefer not to have to host everything. And again, if you host it, therefore, you have to divide, provide continuity, so it doubles and triples, not to mention, you know, I think we have 19 terabytes of data to back up every night. It adds challenges to it.
The flip side of that, and I think this is where CIOs need to think about it, what is your core competency? Our core competency should be curing cancer. If there's somebody else out there that does datacenters, massive amounts of storage better, we're in. We'll obviously have to balance that with the vendors and what they can offer. Some of them are not rapidly moving to the Cloud. Others are moving extremely rapidly.
For example, we have a relationship with Box. Since I came here, and Box was available when I arrived, I don't have anything onsite. All my storage is on my laptop and Box. I can get to it through any device. It's a great technology. We just have to introduce it appropriately. Then you've got the security concerns and initiatives, but our security department works extremely well with us on that.
But, yeah, I'm not a fan of leading with the technology. I have some preferences, and as the industry moves, I'll go there with them. It's, again, seeing that core competency is the most important.
Catalyze: You put it really well that your core competency at MD Anderson is curing cancer, right, not necessarily running data sets and doing disaster recovery and...
Mr. Belmont: Yeah, I'm not a software developer. There's others that can develop software better than us. I mean, our EMR, which is homegrown, is really well done, but, you know, we don't develop as fast as Epic. We don't develop it as fast as Google. We don't do massive amounts of high-performance computing. Actually, we do. You know, Amazon is pretty good at high-performance computing as well. Then with the swipe of a credit card, you can spin up a high-performance environment where, I had to go out and buy a bunch of cores and go through our purchasing department, it might be nine months. So if a researcher has a grant and needed and HPC environment, you know, they can start it up almost tomorrow with some of the new services that are out there.
Catalyze: Yeah, we're seeing a lot of that. There were people who were talking about Epic, who is investing for it. But it seems to be that a lot of people are really thinking about, you know, those research projects or pilots or some of those sort of third-party SaaS vendors.
But a lot of it would seem to be on the research side, the ability to staff. And we have actually had a couple conversations with academic centers where some of the timelines related to IRB and getting through security, they're starting to sort of butt heads when it comes to security and compliance and timelines for IRB. So I think research, in this case, is actually a really interesting area, and you guys are probably seeing it as much as anybody.
Mr. Belmont: You're spot-on, and, there's a speed component, which is very critical, especially when you're in trials and so on. But then there's the value in the data. Once that research data becomes more public domain or the trial is over, having that information spread out throughout everywhere, you don't really, as an institution, get the benefit of looking at the data more horizontally.
My concern is once it goes somewhere, and maybe it occurs today, because these scientists keep the data to themselves, can I really leverage it as an MD Anderson institution? Can I use that data and slice and dice it differently? That's the biggest concern about having it completely distributed and kind of in the Wild, Wild West.
Catalyze: Yeah, all right. Well, actually, that's really all that I had. I mean, this has been really interesting and really helpful to understand. And I think that, you know, you're definitely a CIO that has a ton of experience in the EHR space, or EHR implementation space.
But it's really interesting to talk to you as you think about not just EHR implementation, but how that's going to drive and then kind of, sort of, you know, push the organization towards more of a data driven process improvement organization as opposed to just sort of, well Wild West, you weren't using it for that, but just sort of not anything worth measurable and managed. Is there anything we didn't cover that you'd like to touch on?
Mr. Belmont: I really enjoyed the conversation. I think you hit spot-on. It's not about software. It's not about any particular application. It's about taking the portfolio of things that you have and using it for whatever your business is about.