• Type to search or press enter for full results.
Type to search or press enter for full results.



Building and Leading a Data-driven Health Organization with Dr. John Lee, CMIO at Edward Hospital
Episode 93

Dr. John Lee, CMIO at Edward Hospital

Building and Leading a Data-driven Health Organization

” Data is a commodity that we can rely on to make decisions in medicine.”

Building and Leading a Data-driven Health Organization with Dr. John Lee, CMIO at Edward Hospital

Episode 93

Building and Leading a Data-driven Health Organization with Dr. John Lee, CMIO at Edward Hospital

Building and Leading a Data-driven Health Organization with Dr. John Lee, CMIO at Edward Hospital

: [00:00:01] Welcome to the Outcomes Rocket podcast where we inspire collaborative thinking improved outcomes and business success with today’s most successful and inspiring healthcare leaders and influencers. And now your host, Saul Marquez

Saul Marquez: [00:00:20] Outcomes rocket listeners, welcome back once again to the outcomes rocket podcast where we chat with today’s most successful and inspiring healthcare leaders. Really want to thank you for tuning in. And I invite you to go to outcomesrocket.health/reviews where you could leave a rating or review for today’s show. We’d love to hear from our listeners. Can’t wait to hear what you think. So without further ado, I want to introduce our outstanding guest today. His name is Dr. John Lee. He’s a chief medical information officer at Edward Hospital in Naperville Illinois. He’s got a background in emergency medicine and a talent and knack for technology. He’s been a member of the Physicians Committee on HIMMS. He’s been the chair I think he’s currently the chair is John, you’re chair right now yes the chair at Physicians Committee and at HIMMS and he’s just a very curious mind in medicine that’s also at the forefront of what’s happening so I’m so excited to have you on the on the show John. Just want to welcome you to the podcast.

Dr. John Lee: [00:01:22] Thanks for having me.

Saul Marquez: [00:01:23] Absolutely. So what got you started into medicine to begin with John.

Dr. John Lee: [00:01:27] Well it was actually kind of serendipitous when I went to apply for colleges I applied to about half engineering colleges and half of premed colleges. And I got rejected from all the engineering colleges and I got accepted at all the premeds colleges. That was Fate telling me where I was where I needed to go.

Saul Marquez: [00:01:45] I love it I love it. Yeah that’s a loud and clear sign that there’s one. So you’ve been through the front lines as an emergency medical doctor. Now you’re doing a lot of strategic thinking as the chief medical information officer. John what do you think. A hot topic that should be on every medical leaders agenda today what is that topic and how are you guys addressing it over at Edward.

Dr. John Lee: [00:02:09] Well the topic I think truly we’re we’re we’ve been trying to get to and with all the infrastructural changes that we’ve put forth and put all so much effort into over the past 10 years is really trying to get data to be a commodity that we can rely upon to actually make decisions in medicine and church before it really hasn’t been that data really hasn’t been that available. I think the initial iterations and the first wave of meaningful use in the first five or six years was really about just trying to get providers to do sort of the initial obvious steps. So classic story a classic situation that occurred before EMR is when I was going through my training is that the you get a patient in the emergency room one of the first orders that would come forth from the physician was a team patients old church from the charper. And then particularly patients who had been here for in our institution multiple times there would be after probably about I don’t know 30 minutes or so someone would come up with this cart with stacks and stacks and reams and reams of charts. So that’s assuming that that patient was actually even seen in her in her hospital. And so there really was no interim operability whatsoever. And then even if the patient had been in our institution before the charts and the size of them would be just completely and unconsumable and so you look at. Yeah. Yes if you look at the stack of charts and you shrug your shoulders and said there’s no way I’m going to look through all that and you think to yourself I’m just going to wing it. So that’s kind of where we started from. And then the first things that I think we were able to accomplish after high tech was it. Now we don’t have to get that pile of charts. We can actually get the patients old CBC or an old patients note and get it relatively easily and then we also laid the seeds for some interoperability. But really when you when you look at what the data and information that’s available in medical records that’s available right now doing that sort of thing makes sort of intuitive sense to the frontline physician but you’re going deeper and actually using the data to really change care get insights and knowledge into workflows and predictive models that may not necessarily have been apparent before. I think that’s where we’re going to go next. And that’s where a lot of places I think are falling short right now.

Saul Marquez: [00:04:51] Yeah I think that’s a good call John. And just thinking through what is it that we could do these insights if you had to name a couple of them. What would you say some of the insights that the so you could start making on these EMR. You know a lot of people say all right you know it’s a big building machine right. And so how do we make the billing machine into an insight machine. What are your thoughts there.

Dr. John Lee: [00:05:11] Well I would take actually issue with the fact with the contention that these are killing machines. I think that the infrastructure that has been put into place particularly with the regulatory burden. There is an emphasis on that component. For instance the oft used term Nolt bloat which is a direct consequence of 1990s era and encoding which is still persisting and still has no useful purpose in medicine. But it is what it is and it has no logical sense but you still use it to the EMR to create these charts to accommodate that were. I think we can go is number one use the system to smooth out those rough edges. So absolutely no load is an excellent example of that so that you can create these very even m laden notes with admittedly not a lot of noise to signal but you can do it relatively easy and that then frees up the physicians workflow and mindset to do other more productive things. So I think that’s one thing that we can do with the success there. The other is is to actually get unique insights that we may not necessarily have. So as an example one of the big initiatives is is sepsis recently. So the seminal study for this was a surviving SEPTA study I think in 2010 and within that study they identified the three key factors that predicted sepsis at a high level is the patient infected. Does the patient have servs systemic inflammatory response syndrome and does the patient have some evidence of organ damage. Well what they had to do was to kind of dumb it down because that study and the data collection of that abstraction was put together by humans with our puny brains who could only tackle a half dozen or a dozen data points at any given time.But when you actually implement and look at the data unfiltered and when they collect the data they humans actually consciously or unconsciously filter the data so that they know OK while that person may technically be quote unquote an infected but I know that they’re not really infected so I’m not going to count them. So it’s that sort of filtering that occurs that are digital systems really at least at first past didn’t weren’t able to accomplish so when we went live with our initial sepsis score in our digital system. It fired for probably about a third of the patients in the hospital. And so what we did was then we added other points of data to say well if the patient has a certain type of data characteristic well that’s significant because X Y and Z. But that’s not significant because X Y and Z. And so what we need to be able to do is take multiple multiple data points that go beyond what our puny human brain can absorb and create other models to conceptualize insights and knowledge and then have the systems consume that and spit it back out to us in a way that’s actually understandable to us as providers. And that’s actually where a big data cognitive computing. That’s exactly what we’re trying to do with these sorts of systems now. And I think that’s going to be the next big inflection point of what we do with these systems.

Saul Marquez: [00:08:39] Now I think that’s really interesting especially with sepsis patients to big deal. And sounds like you guys took the steps you developed the system and then refined it. You know went from one third of a patients alarming. Now more fine tuned is it in a place now where it’s producing I guess valid alarms.

Dr. John Lee: [00:08:59] Yeah. So right now modulations they. You know the specificity is somewhere on the order of 60 percent which is when you look at the literature that’s actually I think better than average.

Saul Marquez: [00:09:13] Pretty awesome.

Dr. John Lee: [00:09:14] And is actually impacting our care our actual to expected sepsis metrics are have been consistently below what would be expected for quite some time now ever since we put these initiatives in place and that’s the sort of thing that I think that where we can actually make an impact. And the thing is that there is there’s a lot of organizations and mine included that say that their quote unquote data driven but they’re not driven they have some sort of crude dashboards or very and refined data sources that they try to make decisions on or they just guess and they don’t actually make a lot of decisions based on your data. And there are very few organizations that I’ve found that actually are in that category.

Saul Marquez: [00:10:02] Yeah it’s a challenge. And you know kudos to you and your team for putting together this application. It’s a very real problem that clinicians are faced with patients have to deal with them. It’s pretty awesome that you guys made it that specific and actionable and actually polling from data sources that are reliable. So big congrats on that win. Give us an example of a time when things didn’t go so well. And what you learned from it.

Dr. John Lee: [00:10:29] I would say that just generally speaking the issue of alert fatigue is still a very very real and still very very present in our organization and almost every other organization that I know of. And it actually goes back to the issue of data in that the to do like a PTSA cycle. You actually have to study. The key point is studying what I find in both my organization and multiple other organizations. Most of the organizations that are operational leaders crack the whip and say we need this alert or we need this certain metric and then you scramble around put something into place and then it’s like whack a mole you put do that project and now five other projects pop up and there isn’t yet a culture at least in our organization where it’s a regimented. We put something into place. We measure it afterward and then if once we measure it if it’s not doing what we want it to do then we adjust and we act the S and the A in the PTSA cycle are very much missing in our organization just culturally.

Saul Marquez: [00:11:37] I think that’s true across really everywhere not just you guys down.

Dr. John Lee: [00:11:42] Yeah I didn’t there’s this thinking that describe the cracking the whip people panicking about something and they say well we got to do something and then you do something and it gets lost in the weeds. You put an alert in the system and we’ve been reviewing some of these things. And a good alert is something that has an actual response rate probably 10 percent or more which is kind of sad. Well and you circling back to the sepsis. We put a lot of time and effort into that and because we actually did have a PTSA cycle multiple times multiple iterations studying what what worked and where was giving false positives putting the data back into the model our response rate on that particular alert is about 30 or 40 percent. Yeah. So it’s yeah that.

Saul Marquez: [00:12:33] SA part right.

Dr. John Lee: [00:12:35] And that’s that’s actually what’s missing. And the problem is that there are very few organizations that have really industrialized and streamlined that SA part because the processes that produce that sort of data can be very cumbersome and difficult to extract. And then on top of that the people who actually are skilled at extracting that sort of data often don’t have operational context. There’s a CIO who had a saying well that’s a DePWaD project and that DEPWAD is an acronym for designed in programming working as designed. So basically you told them what to build. They built it. It’s not functional but this is exactly what told them the bill and that contact and where. Yeah exactly. Now these people who you know they’re sequal programmers you know that you know they don’t know anything about clinical medicine and they don’t know that a particular piece of data or a set of data may not be necessary relevant but it’s in the data set. So they included.

Saul Marquez: [00:13:32] Yeah that’s a that’s a really good point. And they build you what you ask for but it’s functional that’s where the challenge comes in. That’s a great call out you know you give the listeners, John, an example of I mean this sepsis one was a great example maybe another example of how you and Edward have created results whether it be improved outcomes less alarms by thinking and doing things differently.

Dr. John Lee: [00:13:54] So there is this concept this need your concept that decision support means or particularly interruptive bullets. But we have tried very hard to take the tack that there’s the whole thing about the four or five rites of decisions for right person lifetime right. Right. Workflow so on and so forth. Right. And so what we’ve been able to do is kind of push that alerting forward in the workflow. So a particular example that we had was our compliance with a 1C results on our hospitalized patients for patients who had not had an A1 C within 90 days prior to our electronic system was somewhere in the 80 percent range I think and based on what I know of that metric it’s that’s about average. So what we did was within the Commission order set we have these filters in place that then and reveal a defaulted unchecked A C order if we don’t have anyone see result in our system within the past 90 days. The provider does has to do nothing. It just shows up and it appears for them and it’s that sort of thing I think and then now our numbers are nineteen ninety nine percent as a result of that that particular piece of decisions. And it really is kind of shifting the mindset of what those other decisions actually mean. It’s not something really trying to avoid and get away from this thing that pops up Intel’s you and slaps your hand that you did or did something wrong or you’re about to do something wrong. But actually hand something to you that says this is what we want you to do before you even think about actually doing it.

Saul Marquez: [00:15:42] Yeah I love that. I think that’s more the again going back to your predictive analytics with maybe you didn’t use machine learning or AI for this but it’s getting toward that route.

Dr. John Lee: [00:15:54] Yeah. And to that point it was a very laborious process. If we had a true machine learning tools that probably would taken us about a tenth of the time but most of us don’t have that sort of those sorts of resources. You know we don’t have a data scientist on on staff at our hospital. We’re not Stanford. We’re not Geissinger or not. Oh so we have to do what we can do.

Saul Marquez: [00:16:15] For sure. No and that’s a good example of it. And then once you develop systems like that which you have now you have a really great foundation of hey we did this laboriously check this out and maybe do some research some time with these people. Maybe they’re not there full time to do specific projects that tie into the organizations goals you know.

Dr. John Lee: [00:16:36] Yeah. And I tell our director jokingly threatened I’m in the process of learning our python and it’s of a running joke between us.

Saul Marquez: [00:16:48] Nyice Oh man. But are you?

Dr. John Lee: [00:16:52] I’m dabbling dabbling you just don’t hear it just so that I understand it’s Frasure. It really does help having that sort of background so that you can actually walk the walk and talk the talk. On both sides so that I can speak as a physician to the providers and I can speak as somebody who knows the technology to the analysts and that actually having to be having that ability to bridge that gap really does help prevent that deep wide phenomenon that I was speaking totally.

Saul Marquez: [00:17:21] Yeah and it gives you the strength and ability to pressure test both sides of the aisle right. I think that’s a really good point yeah. And so you know we’re walking through a lot of these really cool things that you guys are up to really appreciate you sharing that. I know that you know the listeners you all maybe you will take some ideas from what John has shared today and maybe you’re curious about how he got there. We’ll share you know just maybe the best place to get in touch with John here at the end. But it’s all about improving outcomes collaborating. That’s why we’re here today so think about some of the things that you’re doing at your institution or at your company and how you can make things better based off this conversation with John which has been really really great. John let’s pretend you and I are building a medical leadership course on what it takes to be successful in medicine today. It’s the 101 course or the ABC of Dr John Lee we’re gonna write a syllabus. I got four or you share. And these are lightning round so give me a snippet as you can and then we’re going to finish with a book that we’re going to add to the syllabus. You ready. Ok sure. Awesome. So what is the best way to improve health care outcomes.

Dr. John Lee: [00:18:29] Use data use data well.

Saul Marquez: [00:18:30] Love it. What is the biggest mistake or pitfall to avoid.

Dr. John Lee: [00:18:34] Using data poorly.

Saul Marquez: [00:18:37] How do you stay relevant as an organisation. Despite constant change.

Dr. John Lee: [00:18:41] Innovate be at the leading edge.

Saul Marquez: [00:18:43] What is one area of focus that drives all else in your organization.

Dr. John Lee: [00:18:48] Really is developing that DI and analytics infrastructure that’s agile and can get the sorts of data information and knowledge that. Our position needs quickly.

Saul Marquez: [00:18:57] And finally John what’s the book that you recommend for the listeners here on the syllabus.

Dr. John Lee: [00:19:02] Probably can I say two.

Saul Marquez: [00:19:04] Yeah you could you could give us new All right listeners you got you got it you got work ahead of you on the syllabus John’s. John’s a teacher here.

Dr. John Lee: [00:19:10] Drive and outliers. I think those I refer to quite frequently both in interactions with other people and just internally when I think about things as a manager and leader in our in my her position.

Saul Marquez: [00:19:27] Love it John and Dr who’s the author of Drive.

Dr. John Lee: [00:19:29] I think that’s Dan Pink is the author.

Saul Marquez: [00:19:32] I got it Dan Pink and then outliers as Malcolm Gladwell right.

Dr. John Lee: [00:19:36] That’s correct. Yeah.

Saul Marquez: [00:19:37] Awesome awesome. So there you have it listeners you got this awesome syllabus. These two books don’t worry about writing them down. Just go to outcomesrocket.health/JLee That’s J L E E. And you’re going to be able to pop up all of the show notes for today’s episode as well as the links to John’s organization and any other things that he’s up to. He does some really great shares on LinkedIn writes articles that I think you’ll find very intriguing. He’ll share the best way for you to tap into what he’s putting out there. John before we conclude we’d love to just you know hear you share a closing thought with the listeners and in the best place where they could get in touch with you or follow you.

Dr. John Lee: [00:20:17] I think really harped on it before and it really is about the data. And if we can unlock the data and actually make it meaningful for our organizations we can actually change our organizations and actually save lives and actually save the healthcare system which is collapsing really under its own weight. Probably the best way to look at what I’m up to you mentioned before is is my linked in profile and will share some contact information on the Website as well.

Saul Marquez: [00:20:46] Outstanding so listeners go ahead and check them out. You’ll see the link to his LinkedIn profile in the show node’s just go to outcomesrocket.health/jlee. And again just want to say thank you so much John for spending time with us today. It’s been a really really interesting and fun time.

Dr. John Lee: [00:21:03] My pleasure thank you.

: [00:21:08] Thanks for listening to the Outcomes Rocket podcast. Be sure to visit us on the web at www.outcomesrocket.health for the show notes resources inspiration and so much more.

Recommended Book/s:



The Best Way To Contact Dr. John:

Linkedin – John Lee

Mentioned Link/s:

Articles in Linkedin

Episode Sponsors:

Healthcare Podcast

Building and Leading a Data-driven Health Organization with Dr. John Lee, CMIO at Edward Hospital