For Drex DeFord, Seattle Children’s Hospital offered quite a few selling points. The organization, which includes a 250-bed teaching hospital, research institute, and foundation, is among the nation’s top-ranked children’s hospitals, and is located in the scenic Pacific Northwest. But perhaps the biggest boon was the organization’s belief and involvement in lean methodologies. At Seattle Children’s, continuous performance improvement is a core philosophy that bleeds into every area and is guiding the way through Seattle’s five-year IT transformation. In this interview, DeFord talks about the role of change management in a successful implementation; the importance of not just training physicians, but helping them to see the big picture; what policymakers need to keep in mind when it comes to Meaningful Use; and why it’s critical to taking time away and unplug every once in a while.
- Comparing best of breed vs best of suite
- Educating the end users
- Going down the virtualization road (Wyse Technology, Citrix)
Hanging a brand name on a product you’ve acquired and saying that it’s integrated doesn’t make it so. Very often there’s still work that’s going on through an interface engine, even though it’s advertised as ‘integrated’. So there’s a level of complexity that I think, if it’s possible, organizations should try to stay away from.
I think things like bringing your physicians down and letting them walk through the data center and showing them some of the technology from time to time is very valuable, especially if you’ve got individual informal leaders that you can sit down and have conversations with. They get it, and they can definitely help you as an advocate in those tough spots.
With some of our providers, we’ve taken 45 minutes out of their day. When you get here in the morning and you log on with Citrix and Wyse, it takes 40 seconds and you’re ready to go. Your icons are up and running, your desktop’s up and running—you can open your applications and go to town.
What we’ve learned here is how much of a force-multiplier the virtualization-to-desktop layer can be. We’re saving money, we’re supporting the end users with less FTEs and less power, and we’re providing better service. We’ve got better compliance and improved security. It’s hard to argue these things.
This is another thing you can’t just go and slam in. You have to think very carefully about what are the applications that are being used by your end users, and do we do have the right guys to do the packaging to make sure that those applications can run on the servers.
Guerra: You’re largely a best-of-suite enterprise shop. Do you look at organizations that have a whole bunch of different vendors on the clinical side in this pocket and that pocket, and dozens or hundreds of interfaces in their multi-hospital health system and just wonder how they’re going to manage any of this? Does that sound like a much more complex environment than yours?
DeFord: I think it is. When you look at big, integrated delivering networks that have multiple hospitals, many of those hospitals having purchased and installed EMR suites from different vendors, I think it is a challenge. And it certainly relies a lot more on that knowledge management layer to pull data from all those systems to understand where you’re going as an integrated delivery network. Even in a single hospital where there’s lots of sub-components that aren’t from the same vendor—and I would say, even when they are from the same vendors sometimes—hanging a brand name on a product you’ve acquired and saying that it’s integrated doesn’t make it so. Very often there’s still work that’s going on through an interface engine, even though it’s advertised as ‘integrated’. So there’s a level of complexity that I think, if it’s possible, organizations should try to stay away from. Given the reality and politics of many organizations, that’s not always possible. So I think it’s the CIO’s job to point out how additional complexity can certainly compromise the reliability of a system. But in the end, I think you have to team up with your end users and your leadership and deliver the product that is selected. I’m trying to be politically correct.
Guerra: I had an interesting conversation with another CIO who was talking about how she tried to explain to physicians that if they select an application for their practices other than the one on the inpatient side, there were going to be difficulties. And then we talked about how maybe they didn’t care. Some physicians will say, ‘We want this one; you make it all work. We don’t really care if you’ve got interface issues or whatnot.’ So I wonder if that argument even resonates with the physicians who you’re trying to steer into a particular direction with a system selection.
DeFord: I would hope the argument would resonate, but there are definitely times where that can be a challenge, I think. To go back to the Maslow’s hierarchy of needs model, if you’re running enterprise architecture really well and you’re running the products that you have really well, and you develop that relationship with physicians where they feel like, ‘Well, they really have got their stuff running pretty well,’ you do create a bit of a situation where they think you are magic and can do almost anything. And that’s really when I think you have to have the intense conversation where you say, ‘Let me explain why this is way more complicated than we make it appear to you as an end user.’
And so I think things like bringing your physicians down and letting them walk through the data center and showing them some of the technology from time to time is very valuable, especially if you’ve got individual informal leaders that you can sit down and have conversations with. They get it, and they can definitely help you as an advocate in those tough spots.
Guerra: So you take them behind the curtain and show them how the sausage is made.
DeFord: Definitely. I think some of them are even interested in that. They want to know why something that they want is difficult to do, when it seems like logically, it should be very simple. And if you can get a few of those, that’s big. For me, sometimes the key is to find the biggest naysayer—the biggest complainer, and try to figure out how you get that person on your side. That can make a world a difference.
Guerra: Right. I know you’ve done some work with virtualization working with Wyse and Citrix. Tell me about the genesis of this project—why something needed to be done, how you came about deciding upon virtualization, and then the project itself.
DeFord: I think a lot of folks are involved in virtualization right now. Several years ago, you’d walk into a data center and look around and there are hundreds of servers and they’re all used at a 3% or 5% capacity, and you scratch your head and wonder why you can’t do better than that. So we, like most organizations, began going through a server virtualization process. We’ve spent a lot of time on that, and I think we’ve been very successful. We’ve eliminated a lot of servers—I don’t have the number at the top of my head, but we run a significant proportion of our servers in a virtualized environment.
And then there’s the Gemba thing. Going back to the Gemba thing really sort of raised its head; spending time with end users and providers and watching them work and really hearing them complain that it takes too long to log on. We’re still largely a PC-based environment, but we’re getting out of the PC business as much as we can. So we were going around talking to end users, and they were saying it takes too long to log on and the machines are entirely too slow. Like many organizations I think that with the economic downturn we had skipped a year in PC replacements and made them last a little bit longer. And all of that really exacerbated the problem that we had with the end users. And again, I think if you don’t make it easy for the end users and then the end users come up with another plan, and that plan might be to log onto one machine and put a sign on it so it’s your machine all day and no one else is allowed to use it. That was never really the intention of how we had designed any of our implementations, but we were finding little pockets of things like that going on.
So I sat down with my really smart guys—my enterprise architecture guy and my chief technology officer—and I said, ‘We have to find a way to do this better, faster, and cheaper.’ That’s a mantra that you’ll hear me say over and over again if you’re on my team—‘how can we do this better, faster, and cheaper?’ And so not long after we got here, we started looking into desktop virtualization, and that let us to Citrix, a product that Wes Wright—my chief technology officer and right hand guy—and I had used at previous locations. We started to explore Citrix, and that led us of course to Wyse, which is really sort of a purpose-built thin client device—the Wyse Xenith, built to work with Citrix.
And over time, we did some initial pilot deployments. We did that inside of the department first to eat our own dog food to make sure that we knew how everything would work, and we were able to make great progress with that pilot. We then took the next step, which was to deploy it to our most senior executives, because our senior executives knew what we were doing, and we felt like if we were going to ask other people to use this technology, that they needed to use it first. So we put it in place there and they used it well and liked it, and then we started deployments, first in the business support areas, and now we’re well into clinical areas and even the inpatient units. And the project has changed from us saying, ‘I’m bringing something to you. Trust me and let me show you how this is going to work; it’s going to be fine,’ to every area of the hospital now stopping me in the hallway saying, ‘When are you bringing virtual desktop infrastructure to us? Hurry up.’ It can’t happen fast enough.
With some of our providers, we’ve taken 45 minutes out of their day. When you get here in the morning and you log on with Citrix and Wyse, it takes 40 seconds and you’re ready to go. Your icons are up and running, your desktop’s up and running—you can open your applications and go to town. You can open four or five applications, and let’s say I have a meeting that’s not at the hospital but at the research institute. I can click my disconnect button and go to the research institute. I can find a drop-in station there; I can reconnect and log on, which takes about 12 to 15 seconds, and my desktop is exactly where I left it. All of my windows open all exactly in the same place where I was before.
And then at the end of the day, I go home and log on to my personal machine at home, which is an Apple MacBook. And using two-factor authentication, I log in through a desktop’s portal, and I get a Windows 7 desktop that is exactly the same desktop that I was just using at work, with all of my applications open exactly where I left them. It’s a wonderful system; it’s incredibly fast. For some physicians, we know we’ve taken 45 minutes out of their day from a log-on perspective. The Wyse Xenith devices last for years—literally five, six or seven years, and they’re super easy to replace when they break. I don’t know that I’ve had one break yet, but they’re super easy to replace when they do break. And from a power consumption standpoint, they use about a tenth of the power consumption of a PC, so they’re super quiet and there are no moving parts. It’s just been a terrific deployment for us. We’ve got about 1,400 or 1,500 hundred in the field, and we will continue the deployment through next year. I’d like to say we’re going to replace virtually all of the PCs in the building, although there are probably still reasons to have PCs in some locations.
And this has been a ton of work by the team. They go through every individual department; they talk to that department and see what applications they’re using on their PCs. If they haven’t already packaged those up for Citrix, they have to take some time and package those up and make sure they operate well through Citrix. And then they move the person or department over to virtual desktops. Everyone saves everything on their O-drive, so from a security perspective, there’s no PC for somebody to walk off with. There’s no PC to crash and then have people say, ‘Oh I lost everything that was on my C-drive.’ A lot of those problems go away, so we’re able to do it better, faster, and cheaper.
Support is much more simple, and security patches or application patches become much more simple. It’s a great success story. I think we’re one of the few in health care that are using it like this. We were a finalist for the Citrix innovation award and AMDIS has recognized this. We’ll get an informatics award later this month from AMDIS. And so I think what we’ve learned here is how much of a force-multiplier the virtualization-to-desktop layer can be. We’re saving money, we’re supporting the end users with less FTEs and less power, and we’re providing better service. We’ve got better compliance and improved security. It’s hard to argue these things. And in the end, going back to Maslow’s hierarchy of needs, it creates an enterprise architecture layer that people can rely on to access their applications more easily and more completely.
Guerra: Are there any common mistakes people might make going down this virtualization road that you can help them avoid?
DeFord: I’ve gotten a lot of calls from guys saying, ‘How are you doing that? We’re having problems with our deployment.’ For us, a lot of it has been the way that we’ve packaged the applications to run on the servers and not try to create a situation where every time you open something up, it streams the application down to the device. That really does slow things down. And the only thing I would say is that if you’re going to do desktop virtualization with Citrix—and that’s the one I’m familiar with and use—do it the way Citrix advertises it. Do it the way Citrix says you should do it, and you’ll do fine.
But it does take a lot of time and a lot of planning. Again, this is another thing you can’t just go and slam in. You have to think very carefully about what are the applications that are being used by your end users, and do we do have the right guys to do the packaging to make sure that those applications can run on the servers. And you’ll meet resistance sometimes from vendors who say, ‘We won’t support that if you run it in a virtualized environment,’ or ‘We won’t support that if you try to do it through Citrix.’ And we reassure them that if we have a problem, we’ll put it on a server or we’ll put it on a PC, and you’ll be able to troubleshoot it. But we usually have done it anyway, and we’ve only found maybe a few that don’t really work well with the infrastructure we’ve created. And in most cases, vendors come back to us and say, ‘Can you show us how you did that,’ because of course that’s something they would like to sell. So it’s been a really fun project and it’s been very successful. I’m very happy with the way it’s gone.