What motivates the CIO of a large health system? For Steve Hess, it’s “picturing that day when we save someone’s life through the use of informatics.” Of course, it will take — and has already taken — a great deal of blood, sweat, and tears to get to that point, from getting five hospitals and hundreds of clinics onto an integrated EHR system, to creating standardized workflows, to turning data into “actionable clinical decision support.” In this interview, Hess talks about merger that created UCHealth three years ago, why he’s a big believer in going big bang, the “why not Epic?” philosophy that has helped increase buy-in, and how collaboration is more of an art than a science. Hess also talks about the three tiers of analytics, the “real heavy lifting” when it comes to data, and the exciting direction healthcare IT is taking.
- Big-bang with Epic — “It’s scary. It’s big, and there’s a lot of change management.”
- 80/20 rule with workflow
- Standardized order sets across UCHealth
- 3 tiers of data
- “Analytics is a never-ending journey.”
- Defining data — “It’s not trivial.”
LISTEN NOW USING THE PLAYER BELOW OR CLICK HERE TO SUBSCRIBE TO OUR iTUNES PODCAST FEED
The belief that we have here is let’s rip the Band-Aid off. Let’s go big bang and then let’s maximize the support at the elbow and at the command center, and get to the healing as quickly as possible.
We’re trying to move away from the concept of ‘report writing’ to the point of enabling analytics in such a way that people can answer hundreds of questions with the tools that we provide them, not just answer the one question.
Analytics — or even one metric that you’re trying to monitor — is often a project onto itself, and multiple iterations to the point where everybody accepts how the data’s coming out, how it’s being defined, and how it’s being visualized.
We’ve been treating the metrics, analytics, and dashboard initiatives like projects with very specific scopes, starts, and ends, and steering groups to actually get to the point where the data definition and the data governance is solid and consistent across UCHealth.
Gamble: What was the approximate timeline of the Epic go-live?
Hess: It was about 10 months from kickoff to our first ambulatory go-live, and probably 16 months from kickoff to our inpatient go-live. We first went live in our ambulatory clinics and went live with revenue cycle in our ambulatory clinics before we big-banged inpatient. So from kickoff to first ambulatory, it was 10 and then kickoff to hospital, it was 16. For our subsequent implementations where we took Epic across all the rest of the hospitals, we pretty much big banged those and that was ambulatory and inpatient, and it was about a year from kickoff to go-live for the subsequent hospitals.
Gamble: Judging from a lot of things you said, I can see that the plan was never to do this in phases, and that for all the purposes you talked about, it just made more sense to go big bang and then after that deal with anything that needed to be adjusted.
Hess: Correct. There’s no right or wrong way. We do believe in the big bang and the reason we believe in it is if you don’t have the big bang approach, you have to create a bunch of temporary interfaces, workflows, and processes that then get undone when you actually go live with that. Often, when you have a big bang, you allow everybody to kind of focus on the future and not have to create those bifurcated or temporary workflows. And then you go live, and it’s scary. It’s big. And there’s a lot of change management, but it allows the entire organization’s focus to be on that event. And then there’s healing and getting to a better place after that rather than this multiple waves, multiple pain points bifurcated temporary workflows and interfaces and so on.
Again, there are a lot of people that have been successful with waved go-lives and they’ve made it work, but the belief that we have here at UCHealth is let’s rip the Band-Aid off. Let’s go big bang and then let’s maximize the support at the elbow and at the command center, and get to the healing as quickly as possible.
Gamble: Right. What’s the status at this point across the organization as far as Epic?
Hess: We have Epic deployed every single clinical location across UCHealth. There’s not one single clinical area that’s not on Epic. Again it’s five hospitals, and we have about 425 unique ambulatory departments. That’s not physical buildings, but that’s 425 unique clinics. They’re all on Epic — ED, OR, cancer — everybody’s on Epic.
We do host Epic for 41 independent clinics in the community as well, so we actually are hosting Epic for non-UCHealth ambulatory locations, and we’re beginning the conversation of hosting Epic for independent hospitals in the Colorado area.
Gamble: So you’re looking at being able to not just host but also provide some feedback or best practices to them?
Hess: Absolutely. And again, part of our strategy is to be standardized. Easier said than done — an academic medical center has very different processes than a community hospital, and a community hospital has very different practices and workflows than a critical access hospital. But in essence, it’s trying to standardize as much as you possibly can, so think of the 80/20 role where 80 percent of the workflows are the same across all those different kinds of care settings, 20 percent needs to be configured for that.
Order sets are something that we spend a lot of time standardizing. The nursing documentation across all five of our hospitals is identical, so that a patient and a nurse who actually travels between the hospitals will have a very similar experience, and physicians will have a very similar experience. And the more we do our outreach and population health and telehealth, having that same foundation across all carriers is extremely valuable.
Gamble: Right, especially when you’re dealing with such large volumes.
Gamble: Another area I wanted to get into was data and analytics and what you are looking to do. I imagine that being in academic, that’s another place where your strategy does have to be somewhat different, but can you just talk a little bit about what the organization is looking to be able to do with analytics?
Hess: Sure. So there are a couple of different angles there. Again, analytics is a never-ending journey, and I don’t think we’re perfect or close to where we want to be yet, but we’ve made some good strides.
First of all, we actually look at analytics in tiers. Tier one is the type of data that operational leaders need before their first cup of coffee the morning. That’s the concept. What information do you need to just have visibility into how the clinical or financial operations are running? Tier two is more of that unit or clinic level leadership. They want to be all slice and dice the data a couple of different ways to see what’s going on. And then tier three is essentially access to raw data that allows power users to be able to really almost answer any question.
We’re trying to move away from the concept of ‘report writing’ to the point of enabling analytics in such a way that people can answer hundreds of questions with the tools that we provide them, not just answer the one question that the report answers. We’re still on that journey, making good progress, but that’s something that we’re focusing a lot of attention on. So we’re turning report writing into analytics, and then taking the analytics to the next level, getting more towards predictive analytics and trying to figure out what’s going to happen tomorrow, next week, etc. And that leads to advanced clinical system support — turning the data into information into learning and then into actions back in the EHR to help the caregivers day to day.
A lot of what we’re trying to do is what I call closed loop analytics. Pull the data out of Epic, run it through the various algorithms and learnings, and putting answers or assistance back into Epic to allow the doctors and nurses and ambulatory care folks to be able to take action on the data in their workflows day to day. So again, analytics turning into more of the actionable clinical decision support.
Obviously, a big focus for us is on population health. We’ve implemented some patient management tools within Epic to watch over the risk of various patients for risk of readmission or diabetic follow-up or congestive heart failure, those kinds of things. I won’t call it necessarily population health quite yet, but more of patient management oversight lens follow-up care on the journey toward population health.
Gamble: Okay, so a lot of great information there. I think you said it right away about it being a never-ending journey is something that anybody can relate to, and I can imagine that right now where one of the challenges is that everybody knows what they want to be able to do, but it’s everything that’s required to get there and all the maintenance to be able to set up that infrastructure for closed-loop analytics.
Hess: Absolutely. And getting data out and into Epic or any EHR is hard onto itself, but really the heavy lifting of analytics — and I think a lot times people gloss over this — is actually just defining the data and data governance. Because on the surface, something like length of stay or what is a diabetic patient seems simple and it seems easy to be able to define, but the reality is once you pull back the covers and really see how people define what a diabetic patient is or what truly length of stay is, it takes a lot of work to actually come up with that data definition. And then as a system, trying to get that definition agreed to across three various regions, systems, etc., is not trivial by any stretch.
Analytics, or even one metric that you’re trying to monitor, is often a project onto itself, and multiple iterations to the point where everybody accepts how the data’s coming out, how it’s being defined, and how it’s being visualized. So it’s not trivial by any stretch to get data out in a way that makes sense for everybody. It’s almost like the stages of denial, where the first time we produce data, it’s often just completely refuted — ‘this data’s wrong, there’s no way it could be that bad or that good.’ And often, we actually have to work the data owners or the person who’s actually requesting the metrics, through the process so that they can see, ‘You know what, this is actually right. It looks like this because of how we’re using the system or because how we define the system.’ And then you tweak it, make it better, and then over time, it becomes something that you can really rely on. But it’s not trivial; it’s not like you can just kind of lob a request over the wall, run away and have a perfect metric given to you within hours or days. It just doesn’t work that way.
Gamble: Yeah, I think that you’re right in that it does get glossed over and what you said about governance too. How have you approached the data governance strategy?
Hess: It’s really important, and what we’ve been doing is treating these metric requests and dashboard requests like projects, such that we have a project manager, and we have all the key stakeholders from all the different areas impacted. We have administrative leaders, we have physician leaders, and we use those groups to actually get to the point where they’re signing off on data definition sheets such that the people that actually create the key, create the visualization, and can actually use a standard straightforward definition sheet. Where the workflows of using the EHR such that the data coming out is actually flawed, we then use that same steering group to actually help modify the workflows. So we’ve been treating the metrics, analytics, and dashboard initiatives like projects with very specific scopes, starts, and ends, and steering groups to actually get to the point where the data definition and the data governance is solid and consistent across UCHealth.