If you're striving to make better decisions at your healthcare organization, data analysis and visualization solutions are critical. You can drive continual process improvement by empowering business analysts to visualize, manipulate, and take action on governed data in real time. Thanks to self-service analytics tools like Qlik and Tableau, you're now able to integrate analytical solutions into your EHR, getting the right information to the right people, at the right time in your workflow.
To introduce you to this topic and dive a little deeper, Nordic Director of Optimization Solutions Joel Martin sat down with Manager of Data & Analytics Tim Grilley. You can play the episode below or download the episode here. If you prefer the written word, the transcript is below.
If you’re interested in learning more after you’ve listened, visit our Data and Analytics page where we have some additional information.
We’d love to chat about how Nordic is helping our clients build a framework for internal data-driven decision making.
Show Notes
- [00:37] Introduction
- [01:23] Defining self-service analytics
- [02:52] Identifying governance and users
- [06:14] Learning the four tools
- [08:38] The benefits of applications versus reports
- [10:19] Customer successes
- [14:59] Determining feasibility
Transcript
Joel: Hi, I'm Joel Martin. I'm the Director of Optimization here at Nordic. In my spare time, I'm also a data scientist. I'm really excited here to talk today with Tim about data science. Tim, can you introduce yourself?
Tim: Sure. I'm Tim Grilley with Nordic Consulting. I'm the Manager of Data Analytics. I, also, am a data scientist. I have a Master's in Statistics. I think about it as we've laid this ten-billion-dollar pipe framework, and my master's is getting data in to EHRs. My master's has given me that ability to direct exactly how we're getting data out of the EHRs in a really good way. Joel, go for it.
Joel: Today, we want to talk about self-service analytics, so that goes to what you were just talking about of how we get the data out. Tell me a little bit about what you think of when we say self-service analytics.
Tim: Yeah. It's a good question. I think people have been thinking about self-service analytics for a while, and we now have really good tools to start to realize it. To put it in a very specific framework, self-service analytics is where the operational users and healthcare organization are able to access and manipulate the data that is important to their business themselves, at their own workstation, at their own pace when they need it. They're really getting their answers today instead of getting their answers after requesting a specific view in the future, like six months down the line.
Going in a little bit deeper, and I think this is where self-service analytics today is looking different. It's also governed by a centralized body. Those users are able to get those answers today, but they're working with a data set that someone has already put together and curated. Enabling the executive sponsors to really know that the data they're accessing is accurate, it is validated, and the answers they're getting out of it are appropriate. That's kind of that next layer in self-service analytics.
Joel: Let's start with that part. The groundwork. Organizationally, who's typically setting up that foundation and doing the governance?
Tim: Yeah. In an analytics project or a self-service project, you're going to meet and make some sort of groundwork decision with the responsible person. A director at both the operational side or VP on the operational side with the director/VP on the IT data side, and come to some consensus of exactly what is in the data set, what those governed data pieces are, and how they're defined. Then, effort. Research and development happens on the IT side. Validation happens on the operational side. Everyone agrees that the agreed upon metrics were actually implemented in the data set as you are able to start using it.
Joel: Sure. Then, once those are in place, you said there's operational users that are going to actually implement that and dig into the data. Can you tell me a little bit about who the ideal users are? Where does this actually play out in an organization?
Tim: Yeah. The ideal users of one of these self-service analytics is a business analyst who has some background in data. One of the things that I find successful in lots of projects, and that's not in general, is a business user who has background in data is not extremely frequent in healthcare organizations. There's some departments that have a lot of them, and there's some departments that there's not as many of those types of people. One of the things that I found very successful in this type of project is enabling the business users. Enabling them and doing some education around how you use data. That's really going through, "Well, this is what," going back to statistics and really introductory statistics, and going through, "Well, this is what a mean is. This is what it means to be making decisions from the mean. This is how correlation works." Go through really the whole progression of how you're thinking about data, and making decisions from that data.
To take that clinical user, take them from being just a clinical person and really growing their skill set, educating their skill set, and making them a data user, too.
Joel: They understand the operational processes, clinical processes, then you can educate them kind of on the basics of data.
Tim: Right.
Joel: Is there anything else they need to know or should the tool be intuitive enough?
Tim: In general, we're really talking about three, maybe four different tools inside of healthcare that are in the healthcare scope right now. In general, the tools and the feedback I'm hearing is they're intuitive enough for people to understand without a lot of training in the specific tool.
Joel: You're teasing us here. What are those tools?
Tim: Yeah. Yeah. Those four tools ...
Joel: Everyone wants to know.
Tim: Yeah. Power BI is a great tool. The one caveat there is, right now, it's in the cloud only. If you want your data “on-prem”, that's a big thing.
Joel: On-premise you mean?
Tim: On premise, yes. “On-prem” is what people say in industry. If you want your data on premise, then it doesn't work for you. Qlik and Tableau, both, very good tools. Nordic, we have a partnership with Qlik. Tableau is another really great tool as well. Both of those have great visualization components. Then lastly, SAP, the Universe and Webi is a tool that Epic has a pretty nice, robust set of already built universes. You can use those to get your first steps in, but it's not as powerful as Qlik, Tableau, or Power BI. Those are really the four tools that healthcare is focused on right now.
Joel: Okay, so you get your governance set up. You get your tools set up. You get some people educated. People start doing this kind of self-service analytics. What happens? Who benefits? What's some of the potential that you think about and gets you excited?
Tim: Yeah. There's two parts that get me excited. I'm going to separate it into big camps. I think a lot about managing a BI Team, and managing the reports for our customer. One of the things that gets me excited about these projects is you can take thousands of reports and replace them with hundreds of applications. You're talking about an order of magnitude shift in the amount of maintenance that you're needing to do as a reporting manager or as a data manager on the customer side. That's very exciting. It's a very promising project. ThedaCare, as an example, replaced all of their Crystal reports, which was numbered in the five thousand area, with Qlik applications numbered in the two hundred range, maybe three hundred.
Joel: Wow. What is it about these applications rather than reports that allows you to have that kind of compression?
Tim: Yeah. Reports, a lot of them are the same question asked slightly differently. What an application gives you is the ability or a tool, we can generalize to say tool. What one of these tools gives you is slicing and dicing inside of that information, so you're able to take many reports that are asking very similar questions around specific KPI for an organizational area. You're able to replace them with one tool or application that shows those KPI, shows the relevant dimensions, which allow you to slice and dice into that KPI, and then allow users to schedule and interact with those tools at their own leisure.
Joel: Wow. Okay. That's benefit one. Benefit two?
Tim: That second benefit is really those operational users are able to realize gains instead of ... When they're asking a question, it's pertinent right then. When they get an answer in a traditional reporting sense, it's usually three to six months. I've heard of a year, down the line, if they actually even get an answer. For the operational areas that have the aptitude and the bandwidth to be able to use a tool that one of these optimization analytics, they can realize big gains. As an example, I'll use...
Joel: You could tell I was going to ask for an example.
Tim: Yeah. Yeah. As an example, I'll again use ThedaCare. They used, on one of their first applications of Qlik in the OR, they realized a million dollar save on eliminating one supply from their OR that they would have never thought to even look at until they invested in one of these Agile Projects.
Joel: Right, and that's because normally with a report you have to ask a specific question, and say "I want this data filtered this way with these columns." Right?
Tim: Right.
Joel: When they created an application that just had everything that they could explore, then it allowed them to look beyond the thing that was most obvious.
Tim: Exactly.
Joel: Ask questions from the data. That's pretty revolutionary.
Tim: It totally is. Thank you. That's a great summary. I can think of a couple more examples that are pretty pertinent.
Joel: Well, maybe if you could just lay out really quickly the picture of that, "Here are a bunch of reports. Here's the data. Then, it turned into an app that did A, B, and C all in one piece."
Tim: Yeah. Mercy Health, I'll give that as an example, made an application specific around RAF and HCC scoring, which is ...
Joel: What do those mean?
Tim: Basically, what the acuity is of a patient has an influence on how much reimbursement that patient will receive from Medicare and Medicaid. Scoring those patients appropriately can lead to significant gains for an organization. For example, if you identify someone as obese, but you don't identify how obese they are, you can raise your level of acuity for that patient appropriately if you just identify the exact level that the patient is for the acuity.
Joel: It goes beyond that, right? Then, if your case mix index is if you have more acute patients, then that raises your case mix index and that raises your reimbursement across the board.
Tim: Exactly.
Joel: Right.
Tim: Mercy made an application where they had identified specific areas by queries, where they had identified specific areas where they were miscoding or misdiagnosing patients. Worked that application into a workflow for the providers to get feedback about the potential that they misdiagnosed the patient. Then, the provider would correct that if it was appropriate. Doing that, they replaced lots of reports, because they integrated that Qlik application into all the appropriate areas including Epic. That allows them to see fairly substantial gains as you referenced. Their case mix index plus their individual reimbursement for those patients. Fairly substantial gains monetarily. It is not disclosed the exact public number.
Similarly, Nationwide Children's created an access application where they took a whole bunch of access reports, including a thirty-day new patient arrival, next available appointment, third next available appointment. They collated all those reports into one application and were able to reduce their no-show rate by twenty percent for new patients. Which is huge, because that's twenty percent new reimbursement in specific areas considering those patients.
Joel: I suppose their physicians missed those empty slots?
Tim: Yeah. My wife is a physician, and it's interesting to hear her say things like, "Yeah. I was having a terrible day, then I had two no-shows." I'm sure those physicians probably did if you make that more efficient. Then, you just need to adjust things accordingly if you're having less no-shows.
Joel: Right. Yeah. These are really great stories. What does it take to get something like this off the ground? That's probably a daunting prospect for some organizations.
Tim: Yeah. In general, you want to focus on these as being projects. Thinking about this inside your PMO, you want to involve your PMO in the project, get that scoping discussion set up. Followed by, at Nordic we like to do Agile Methodology, so we do a couple of build sprints where at the end of two weeks you'd be looking at having at least your data model defined. People being able to start to interact with the data, start getting validated at that detail level. At four weeks, have a pretty good framework for the full application completed. Then, a couple of validation sprints where you're really involving operational owners in the validation and giving them two week intervals to validate key parts of that application.
In total, you're talking about six to sixteen weeks depending on complexity. Some of the big add-ons that we discuss a lot, and you need to scope in to your project right away would be thinking about row level security for different areas. This is particularly key for organizations with affiliates for professional billing and access statistics.
Joel: Sorry. What do you mean by row level security? Can you elaborate on that?
Tim: For organizations that have affiliates where you want to separate the data between the two organizations, these analytical applications, so Qlick, Power BI, Tableau, all of those allow row level security. I open it as Nordic affiliate. I only see Nordic affiliate data using the same exact application, so using the same exact maintenance, data governance, everything as when you open it from Nordic Consulting non-affiliate from the parent organization. Perfect separation of data allowing less maintenance for IT, and a great sell to potential affiliates is you get this product that will allow you to optimize your care instead of needing to go through the reporting process. However, it adds about a month, so one build sprint and one validation sprint any project.
Joel: Got it.
Tim: That's a key thing to keep in mind as you're going through. We like to think of it as there's an enablement period afterwards, and you want to schedule between 40 and 160 hours of training time and education time for those users. This is a complete change in the way that they think about data. You need to give them, you, enough time to train them and educate them, and them, enough time to come iteratively and grow their skills.
Joel: You mean 40-120 hours from whoever is building the application, then working on enabling it?
Tim: Yes. Exactly. Enabling for a business user is probably between four and 16 hours depending on how many sessions they need to attend to enable them to really understand what's going on.
Joel: I'm sure there's a lot of difference depending on their experience with doing analysis and that sort of thing.
Tim: Exactly. Exactly. Yup.
Joel: Absolutely. This is all pretty exciting stuff. If people want to know more, are there places they can go to check out what these kind of applications look like, or what's the next step if people are interested in?
Tim: I mean, certainly, I think they should contact either of us. That's pretty clear. That's a good laugh. Certainly, they can contact either of us. We're working on these projects continually right now. I had discussions about vendor selection, as well as thinking about scoping and helping organization scope how long it's going to take and what they need to be thinking about during that install or during that project. A through-to, we build apps, and we have ready projects for specific areas already created. Besides that, checking out Qlik.com, Tableau's website, Power BI's website, all of those have pretty good information. Tableau and Qlik both have good healthcare specific information that you should check out. I'm not sure about Power BI. It's from Microsoft, so they're just bigger in general. They maybe don't have a healthcare vertical set up. Yeah. Those are really good resources as well. There's lots of great free training material online.
Joel: Great. Okay. Well, thanks for chatting Tim, and look forward to talking again soon on data and other analytics topics.
Tim: Yeah. Me, too.
Joel: All right.