Lessons learned in the IT industry
Chris Harper - CIO, The University of Kansas Health System
CITI sits down with Chris Harper, CIO at The University of Kansas Health System. to discuss how Chris got to where he is today, lessons he's learned since being in the IT industry, standardizing information, and more.
Bill Carter - Regional Director, CITI Healthcare
Elena Bradley - Sales Executive, CITI Healthcare
Bill: Can you share any insights from your experience and lessons learned from your accomplishments, since you've been with the University of Kansas health system?
Chris: In my mind, I always go back to the basics. If you think about information and how it's delivered and how its consumed, you have to go back to some of the fundamentals, which is making sure you have, not only mentally, but the best technology also to support it. The quality of the information that you're capturing and delivering really adds value to whoever's using that. So, what I learned early on is that typical enterprise-wide systems, like an enterprise data warehouse, if they follow the traditional model of understanding what the requirements of the business need are, and then hire a bunch of data architects to model it, and then buy a bunch of servers and storage and computing power to really crunch it, then developing the dashboards or reports, you've already spent anywhere from 12 to 18 months going through that motion. What I realized in my past failure is that's too long. What I brought to KU is, it takes 12 to 18 months to deliver value using data, monies will not start. What we focused on is, we really went and worked with our senior executives, and clinicians who had the best problems that we saw.
We tried to narrow that down into the top five problems that we need to solve using data, and really home in on what that problem is. So before we spend any dollar or buy any hardware, we focus on the problem and really try to focus on what the value delivery could be. Then we recruited our Quality Improvement Team to be able to recruit our clinician champions, whether it's physicians or nurses that really can speak to the problem and work with them to understand what data they need to make that decision. One of the problems that we focused on back then was our readmission rates, with heart failure patients, back then about six years ago, most health providers were under pressure to make sure that we improve that 30-day readmission metric. So we had a strong physician champion who was willing to partner with us to make sure that we're delivering on that. Once we understood who we needed to partner with, and what the problem is, we were able to spin up a small pilot data warehouse, within the first three months of understanding the problem, spitting up the hardware, having the dashboards, having all the things that we need to have in place, and validating the data and dimetric.
Within three months, we had a working solution in place, and then we focused on actually delivering that information in a way that our clinician can make the decision to improve the metric. After about eight to nine months, we started to show improvement. Back then, we were hovering in the mid 22% range for 30-day readmissions for our heart failure patients. Then, within the eight to nine months we were start starting to reduce that number. And after about 12 months, when we did an assessment, we were able to get that down to I believe it was just over 18%. When we equate that to dollar savings, we were able to show that by making sure patients are safer by doing that, but also showing that financial we add about $800,000 in an opportunity avoidance and, and penalty avoidance that, that we can attribute directly to that. So, it was easy for me to go back to our governance group in our senior event and say we need to spend more dollars towards this. But here's a real savings that we were able to show by standing up this technology called enterprise data warehouse and aligning our, our data visualization to that effort. So quickly, we identified many more problems like that, whether it's quality improvements, or improving the revenue or savings opportunities that we were able to quickly stand-up additional sourcing additional data as we needed, additional hardware, servers, and additional license, we needed to make sure that we can keep this thing going. So now it's hard for me to imagine how many savings we have. We have roughly over 40 source systems now flowing into our enterprise data warehouse solving all kinds of problems for the system.
Elena: Can you tell us a little bit about the executive sponsorship of your initiatives and provide some guidance to our audience about how you've engaged your executive sponsors and how you manage their engagement during the span of one of your projects? And how does that work?
Chris: That's probably the most important side of it; making sure you work on the people side of your data analytic strategy. That's really understanding and speaking to those frontliners and leaders who can define a problem. Once you understand that you have to bring that back to your senior executives to make sure they buy off on your approach. After that's done, you’re getting the appropriate approval to stand up a data governance initiative. One thing that was tremendously helpful for me is that our clinical quality leaders and our physician leaders and the CIO really understood that was an important thing. So working together to really build out our governance framework in the process. That's what we focused on first is working directly with those, not only to senior executives, but those leaders who, throughout the health system needed to get engaged and understand and support us. So, that's probably the first thing you need to understand is, if you're anybody that's trying to start a data analytics strategy, you have to speak to those frontline and frontline leaders who, who can support you. You have the data now, to speak to what strategy you want to drive and sell. I think governance is a key component of that.
Elena: How have you worked with your clinicians? Could you discuss an example or two of that?
Chris: Being the non-clinician in the room, that is one of the most important things. Really to launch the product or the data analytics platform, the first person I partner with is our CMIO. I have an amazing CMIO, who is a surgeon by trade. He not only has the role as a CMIO, but he uses the system and understands the data needs. That's the first person that I partner with to be able to tell a good story. The next person, from a clinical perspective, that I partner with is our Quality Improvement Team. And so today, that is our Lean Improvement Initiative with a physician and other executive leaders that are in charge. So, we partner with our Quality Improvement Team to be able to understand the problem that we need to solve. They already have, fortunately, a physician and nurse or clinical leader that's taking initiative on whether it's quality improvement, CAD, Sepsis or CHF, or what have you. So really, the question is what data you are getting today, or not getting today that I can support using technology; I look at a problem that way. That's how I partner with my clinical leadership; they already have the frame and I just piggyback and add support.
Bill: Chris, you mentioned a progression or evolution of your data warehousing. You talked about starting with heart failure data feed into the data warehouse for your work with readmission rates there, but you've also mentioned that it has evolved or grown into quite a substantial number of systems you have today. Will you walk us through the history and evolution of the approaches you've had with data warehousing and where you actually are today?
Chris: With me being in the technology side of the equation that's just my bread and butter. When we first started, I spent about a good six weeks just meeting with the front leaders, the frontline staff, or the executives who are today running their clinics or the business. I try to understand where they're getting their information from. Once I did my analysis, I figured out that there's roughly 32 plus different reporting groups throughout the health system and they were pulling data in from all kinds of different system. But it's true that we had over 60,000 access databases and spreadsheets that people were using to make day to day decisions and really that was our “data warehouse.” So obviously, people hear that, and they say, “how do you manage to sustain that?”; and that was the challenge. Everybody had their own way of doing things or collecting data. So we spent, as an organization, tremendous amounts of resources, dollars, and people at times really hunting and gathering data.
Next, I partner with one of the reporting teams and measure in a three-month timeframe what they were spending their time on when I looked at their intake process and understood how did they turn that request from intake to an output of a report or a product that delivers that information? I single FTE, whether it's a report analyst or BI analyst. Most of their time really was spending, hunting, and gathering data. So what I proposed is for me to partner with really the two critical reporting groups for the health system and try to understand their reporting and data needs and replace their current manual ways or spreadsheets or access database. And we have replaced that with more of a modern enterprise data warehouse effort. So that’s who I really focused on.
Also, the data literacy that you have to make sure you train people on how to consume information, because not everybody knows how to look at a dashboard and get insights out of it. So we really also focused on the education side and then also the skillset of the people that needed to be able to deliver that. So we focused on training those folks that were used to using spreadsheets for access databases. Now we are training them on how to, how to write SQL or how to develop a BI dashboard using different types of tools that are available. So it was an all people, process, technology, and education approach
We started with a handful of source systems like your EMR, your cost counting system, your, your clinical quality measurement systems and so forth. And now we have over 40 different source systems. I don't remember exactly how many people we have now, but we have highly specialized resources such as data scientists and BI statisticians who are really focused on the delivery side of that advanced analytics. On the IT side we stood up the full data management capability. So not only maintaining and managing the data warehouse, but I also have all the data consumption and integration. So things like your EMR interface teams or FTP from file transfers, or even now we have engineers who focused on API or fire development, so that we've been able to slowly grow that nine to 12 individuals to a real robust data management and, and analytics delivery functions.
Elena: With your quality initiatives, are these typically led by your physicians, and if they're not, how would they be chosen and are they engaged in the initiatives and the projects?
Chris: Yes, we have an executive who's a physician, who also partners with another executive that is a nurse leader. Their day-to-day job is really to drive that quality improvement effort. We also recruit other clinical leaders, whether it's physicians, nurses, or others that need to focus on specific metrics. So for example, with CAUTI we have a group that spends time looking at it, we have a unit that we really deploy those improvement efforts, and we try to make sure that from a data and technology perspective, they have what they need. So the entire health systems clinical quality improvement is led by a physician executive and, and a nursing executive who really drive the overall program.
Elena: Are there technical approaches that you all have integrated that you feel have really benefited your organization?
Chris: Speaking to the technical side of innovation, one of the things that we were able to stand up is what's commonly known as a service bus approach, which is to integrate all of your data management from ingestion, although out to output. So we were able to showcase that with a wearable device company where we were able to quickly partner with them to figure out way to ingest wearable data in real time, and then spit out an analytical action based on that. So I think that was possible because we leveraged the cloud infrastructure but had more of a software engineering approach to our data management. I highly recommend those that are just focused on data management to go from managing tables, databases, and structure, but to also adding software engineering to that. That way you're able to move with speed and innovation on all the application programming interfaces that are being stood up right now throughout the health system or a healthcare organization. I would really focus on innovating on that consumption side for those organizations who have not done so already.