Leveraging Data to Craft a Best-in-Class Glycemic Management Program
Kendall M. Rogers, MD,CPE, FACP, SFHM | University of New Mexico Health Sciences
Jordan Messler, MD, SFHM, FACP | Glytec
You'll receive an email with a link to this video in case you want to watch it later.
MAR-0000348 Rev 1.0
MASSON: Hi, everyone, and welcome to today’s webinar, Leveraging Data to Craft a Best-in-Class Glycemic Management Program. On behalf of Becker’s Healthcare, thank you for joining us today.
Before we begin, I’m going to walk through a few quick housekeeping instructions. We will begin today’s webinar with a presentation, and we’ll have time at the end for a question-and-answer session. You can submit any questions you have throughout the session by typing them in to the Q&A box you see on your screen.
Today’s session is being recorded, and it will be available after the event. You can use the same link you used to log in to today’s webinar to access the recording. If at any time you don’t see your slides moving or have trouble with the audio, try refreshing your browser. You can also submit any technical questions into the Q&A box. We are here to help.
With that, I’m pleased to welcome today’s speakers, Dr. Kendall Rogers, division chief of hospital medicine and professor in the department of internal medicine at the University of New Mexico Health Sciences, and Dr. Jordan Messler, executive director for clinical practice at Glytec. Thank you so much for being here today, Dr. Rogers and Dr. Messler. I’ll now turn the floor over to you.
JORDAN MESSLER: Great. Thank you so much. Welcome, everybody. Thanks for joining us this afternoon to talk about a topic that Kendall and I have spent a lot of time talking about through the years.
Today’s learning objectives are to identify strategies to use data to drive change and improve patient outcomes, understanding the metrics and key performance indicators to track in the hospital, particularly around glycemic control, how to think through those categories – the different types of data for measuring outcomes, processes, and beginning to think through return on investment, ROI, reviewing the technologies that can help facilitate data tracking and reporting, and some of the specifics around glucometrics and insulinometrics.
I want to take a second to talk a little bit more of our background around diabetes and glycemic management. In addition to working at Glytec, I still practice hospital medicine at my local community hospital, Morton Plant, here in Clearwater, Florida. I work some at USF in medical humanities and very active still in the Society of Hospital Medicine. The Society of Hospital Medicine, our national hospitalist organization, helped bring to me some of the science of quality improvement and using data to improve.
Kendall and I first met about 10 years ago, and he was a mentor for me around glycemic improvement through the Society of Hospital Medicine’s mentored improvement program – really, where I cut my teeth on data, glycemic management, and through that work, helped bring me to Glytec a few years ago. Kendall?
KENDALL ROGERS: Hi. Thanks for having me. My name is Kendall Rogers. I’m the division chief for hospital medicine at the University of New Mexico. I am also the program director for our clinical informatics fellowship. And I’ve been involved in glycemic management since 2006 at UNM. As Jordan mentioned, I was the lead for the Society of Hospital Medicine’s glycemic control mentored implementation program, which mentored 140 sites through their glycemic control journeys. Certainly, we learned a lot during those processes, and we’re looking forward to sharing some of those insights with you here today.
So what we’re going to start out with is doing a poll, and this is just for Jordan and I to have some idea of who exactly the audience is right now that we’re tailoring some of this. Now, I know that we probably didn’t capture every title that people have, and we apologize for that. But we’re hoping that each of you, on your screens, can select which category you fit best into or select the other. You can start doing that now. I’ll keep describing these so that we have about 30 seconds to get these in.
So if you’re in any type of administrator category, if you’re nursing – front line or nursing managers or nursing leaders, if you’re with pharmacy. Physicians, you can decide if you want to fit under this section if you’re a practicing physician, or if administrator would be a more appropriate title for you. I recognize that we don’t kind of have industry listed on here, which may have captured some. And lastly, for any diabetic educators that we have on the line.
I’m going to give it just a few more seconds before we show these polls. And it is making me wish that we had a text box for other, because I’m recognizing that we may not have fit. But great. OK. And yes, if people want to put what some of those extra titles are that we didn’t capture with other into the question box, we’re able to see those, so we appreciate if people want to let us know. All right, I’m going to switch to the next slide. I think we’ve gotten a decent amount of responses, even though they’re still coming in a little bit.
So we have a bit of a spread here, but overwhelmingly other, kind of showing us that we didn’t capture. I appreciate those who’ve listed – oh, and I am sorry – under physicians was very MD-centric for our advanced practice providers – we really meant clinicians within that, but thanks for those listing kind of the other quality roles that they’re in, the dieticians, and in infection prevention and data analysts and others that are listing those. That’s very helpful for us, so thank you for that.
And then we kind of wanted to know what drew you to this topic. What was it out of the title or out of the description? Was it primarily the focus on data, but maybe less related specifically to glycemic management, whether it’s your interest in your role or – for me, it really was my personal interest in glycemic management for many years, not that it was an institutional priority that was doing – or have you been tasked with improving glycemic control within your organization, and there’s an accountability for why you’re here? Or is it some combination? And recognizing that certainly in many of these, that there’s a combination of these. But if one of those is overwhelmingly true up above, select that as your primary, just so that we have some idea of what drew people here today.
Great. I’ll give that just a few more seconds. Excellent. So kind of as we expected, all of us are driven by multiple motivations, as is this group as well, so we’re excited to touch on each of these.
MESSLER: Terrific. That’s really helpful. It really helps. I think, Kendall, that we’ve sort of planned a talk that’ll hopefully cater to that broad audience and the interests that you’ve expressed.
I want to touch on why we’re talking about this today, why this matters – I think why you came to the session today. One of the things that we’re going to talk on – that Kendall’s going to talk on next is change. One of the things we recognize over and over when we try to improve is change is hard, and we want to relay how data can help make that change easier.
The other aspect about glycemic management which I’ve seen a lot is many hospitals have no idea where they’re falling around glycemic management, and you can’t improve what you can’t measure. This is a challenging area to measure, and we’re going to highlight some of those challenges and ways to overcome them.
And if you don’t measure – if you’re not measuring now – our hands are going to be forced. There will be some regulatory measures around glycemic management, and we’re going to touch on that at the end – that some of this is coming sooner than I think many had thought. So this matters for a variety of reasons. And I’ll pass back to Kendall to talk about some of the basics of change management.
ROGERS: Yeah, so any time we’re talking about improvement, it’s really important for us all to have a shared foundation of understanding about change management. I’m sure many of you who are taking the time to be on a webinar on this topic today are already familiar with a lot of these concepts, but we wanted to review them just so that we’re all on the same page and because what we’re going to be focusing on is how we use data to impact each of these stages of change management. So bear with us, those who know this topic probably better than Jordan and I, and we’ll do this in sort of a whirlwind fashion here.
A quote that I love that I’m sure many of you have seen is every system is perfectly designed to get exactly the results that it gets. There’s a lot of arguments about who to attribute this quote to that also Jordan and I engaged in as well. I will note that I add the word exactly, which was not part of the original quote, but it is just an important fact to recognize – that it requires change in a system to result in different outcomes. That’s really a main point that we’re going to be making throughout this talk.
So I’m sure everybody can have their quality improvement hat on, and it’s a basic recognition that we cannot use the word improvement without in some way relating that to change. Improvement is change. So we have to be able to manage change to be able to make improvements within our system. And we know that change is difficult, and failure rates are incredibly high. 80% of major initiatives fail to realize intended objectives. We just know that people prefer the status quo, and we are going to need to break them of that if we are going to make improvements within our system.
There are many barriers to change that we’re all aware of – failure to recognize and accept the need for change, failure of leaders to build compelling cases, trying to do too much at once, providing insufficient communication to a whole variety of outcomes – whether what we’re trying to achieve, the benefits, the consequences of failure – failure to alter systems and structures to support that change or failure to manage resistance effectively, failure to begin and maintain in a timely manner.
So there’s a lot of different models related to change management. Here, just for simplicity, we’re going to focus on the Kotter Model of Change. Again, many of you have likely seen this broken into three kind of stages – creating that climate for change, engaging and enabling the organization, and then implementing and sustaining change.
I like this graphic by Jeffrey Greenwald that really focuses on this stage one being setting the stage – so creating that sense of urgency, building a guiding coalition, developing a strategic vision – the second stage being starting to act, and the last stage really being making that change stick. And we’re going to go through each one of these very briefly.
But Kotter was known to say the change process goes through a series of stages that in total usually require a considerable length of time. Skipping stages creates only the illusion of speed and never produces a satisfying result. So I just hope that people can remember that when you run into a barrier related to your change implementation, go back and look at this list, because often it’s a step that you skipped or something you need to go back to. You need to accomplish each of these stages to effectively move on to the next.
So going through them, step one is really creating that sense of urgency, gathering information through interviews, observation, or processes, getting baseline data, including your process measures, and most importantly, using that data to motivate. Often, at this stage, you’re trying to motivate leadership for resources. You’re trying to decide what creates a sense of let’s go – of getting that urgency – using data that you think people will find credible, important, urgent, and then recognizing, how can you use that data to show that the status quo is more dangerous than the change that you’re proposing?
The next step is really building that team – building that guiding coalition. That change team should be leaders – and recognizing that leaders are not just people who have titles. People are influence leaders within the organization. It could be that frontline nurse who’s been there 20 years and who people really depend on. But having those frontline workers as part of this team, assessing stakeholder resistance and possibly bringing one of them in, and then when you’re forming that guiding coalition, asking who has the relevant knowledge, who’s influential and credible, who’s powerful to make changes, and who is a leader?
Then step three is using that team to create the vision – where do you want to go and why – and really having people be able to articulate this very simply and very quickly. Once you have that vision of where you want to go, then it’s developing the strategy for what’s going to get you there, recognizing, how is this future going to be different from the past, and how can we communicate that? Many people will use an AIM statement. So within a one-year period of time, we would like to decrease hypoglycemia rates by 50% in our ICU and floor units. So trying to be very specific, using smart goals for that AIM statement.
So that’s really setting the stage. The next is starting to act. So within starting to act, we’re first communicating the vision. All that work that the change team did – you’re going to try to keep that simple, have each of them out there as ambassadors giving these talks, using different methods for different audiences, using stories, using different media, and then repetition on what you’re trying to do.
What you’re trying to do is create a crisis. So before, you were using your data for that sense of urgency up. Now, you’re using it to create a crisis to those that you’re trying to motivate to change. And often, it’s setting ambitious stretch goals, being sure that you yourself are walking the talk that you are proposing, and really trying to get that buy-in – trying to identify how individuals are going to benefit from this change.
Step five is empowering others to act, so it’s involving those that are affected early in the planning. Recognize that no one hates a plan that they created. So getting them involved early on, so that they are developing the changes that are being promoted. Starting small – most of us are familiar with plan-do-study-act cycles and trying to build up larger, giving authority and accountability to the team that’s designing that.
And then really focus on these early short-term wins. Generate and publicize those wins early and often. You’re trying to create energy. You’re promoting credibility. You’re maintaining morale. And even those small wins need big celebrations.
And this is kind of you’re starting to act. You’ve completed that starting to act. You’re getting some early wins. But making it stick really requires that we’re going to consolidate that improvement, and we’re going to try to create more change. Those early wins are not likely to have achieved your AIM statement, and the most important thing is to continue driving until you have achieved what your goal is. So build on those early successes. Gain momentum. But be relentless until that vision is achieved.
Deal with resistors. Sometimes your resistor is giving you feedback that’s going to make your process better. So it’s not quelling them. It’s engaging them, trying to improve. And avoiding complacency – don’t allow the small wins to be your mission-accomplished flag.
And then, finally, you want to anchor that change within the culture, using increased credibility to change the policies, structures, and the systems that don’t support that vision, articulating the connection between the change that you’ve already gone through and organizational success, and hopefully, building the data management – the data collection and management – into the process and never stop moving forward.
So hopefully, you can already see, as we’ve gone through each of those, there is a role for data in every one of those steps of change management. There are many things that we can use data for to overcome barriers. Using data can help us achieve situational awareness, which we’re going to talk about in more detail, create that sense of urgency, fostering transparency, accountability. It allows for real-time changes and improvement over time. And when you have transparency of data, you can see the variation that exists from the standardization that you’re trying to implement.
There are many workshops, books that focus on how to effectively use data. A great one is Storytelling with Data, which really focuses on, once you’re formulating your data, you first need to understand the context of it, choose whatever visual you’re going to use for it, try to eliminate any of the clutter, focus on what was most important out of that, and most importantly, find a way to tell a story with that data that hopefully is going to motivate others to make the changes that you’re promoting.
But there are some caveats to using lots of data. The most important one that all of us recognize – and glycemic control is no different – is we can get buried in data. There are so many different things that we can measure, there are outcomes and processes, and it’s amazing what just one glucometer is capturing on each of these. And our job is really to identify the signal versus the noise in that. We want to create real-time situational awareness. Who has hyperglycemia right now? We want real-time reviews that hopefully could intervene. That’s a deeper dive. And then the reports are often looking backwards. What do we need to change in the system?
But as Jordan already mentioned, sadly, 60% of hospitals out there really have no meaningful glycemic outcomes data as of now. And because it’s not a regulatory metric right now, most people have no idea where they are on the spectrum of glycemic control, and those are some of the things that we hope to hit during this.
But any time we’re talking about data, we have to at least highlight kind of the spotlight effect that can occur. That’s just a warning that whatever you are focused on can affect things that you are not measuring. So if you are only focused on hyperglycemia, you may not realize that you’ve driven down hyperglycemia, but at the same time, you were driving up hypoglycemia. It was in one of your dark areas. And if you’re not measuring it, you’re not going to see it. So recognize that effect that can occur.
And again, if you’re choosing the right things – if you’re putting the spotlight on the right areas – then of course you can be having the outcomes. That’s really our job as leaders to select those. And hopefully you’ll come away with more of those tools and which metrics from this talk.
MESSLER: So I think it’s critical hearing from Kendall reminding us, remembering the fundamentals of change management, and really beginning to think how data is so essential to those change management pieces.
I want to take a second here to talk about the fundamentals of data. I think both these aspects – this aspect of reminding ourselves of the foundation and fundamentals when we’re certainly overcome with so much information, particularly around glycemic management, not really knowing where to start – let’s remember our fundamentals.
So thinking through the types of measures and any type of project or process you’re trying to improve, there are structure measures, outcomes. We tend to see a lot in that outcomes area – we’re just sort of seeing the outcomes – and we really want to highlight, when we talk through glycemic management, to begin thinking through some of those process metrics as well. That’s where you can start seeing those short-term wins and start really understanding your processes and what needs to improve. And as Kendall just highlighted in that spotlight slide, those balancing measures are essential, too – not getting lost in one area of outcomes and having unintended consequences somewhere else.
There’s a myriad of pictures I can show on the right here for forms of data. I’m going to highlight three here that are really effective visuals and can certainly overcome, when we’re often just presenting simple tables in our dashboard – so scatter plots, run charts and Pareto charts.
Because thinking of a lot of our dashboards, often they may look like this. Maybe they have some red, yellow, green in there that can help translate the information a little better. Sometimes around glycemia, it’s only one of these columns – less than 70. And maybe it’s just your one row – just your health system. So really trying – here, we’re even a little more enhanced. We got our system. We got each hospital, and then potentially a benchmark. We already have a balancing measure in greater than 300.
But a much more effective visual is putting this simply on a scatter plot. We’re not going to go through some of the fancy visualizations that are out there. Many of your systems have really top-notch software to get visualizations, and certainly recommend learning how to use those. But there are some simple visualizations that can really garner a lot of effective information.
So simply a scatter plot here, where the Y axis is that hyperglycemia, that X axis is that hypoglycemia, and that arrow in that left lower quadrant is the goal here. So you can see the spread, you can see the variation, and quickly begin to see, all right, here are my top performers that I can look at best practice, here are my lower performers that I need to really understand what’s happening there when I seek out an improvement.
A run chart – I’m sure for those doing quality improvement are frequently using a run chart, an essential aspect of quality improvement to see what’s happening over time. But I really highly recommend that you’re annotating those run charts. That’s really one of the key takeaways when you’re seeing these over time – remembering those steps that you put in place for improvement, putting those annotations in those run charts, and reminding the team what you’ve been working on.
This run chart’s even more advanced. It’s a control chart here as well, with some statistics around it. You may not have that capability at your institution. That’s certainly really important when you’re doing some Six Sigma or lean projects, but not always essential. Simply annotating a run chart provides a high level of storytelling to your data.
And then a simple Pareto chart – when you’re trying to do an improvement, some of your first steps are going to be walking the process. Let’s see what’s going on. Let’s audit some charts, brainstorm some ideas of what’s contributing to the outcome you’re looking at, whether it’s hypo- or hyperglycemia or doing a root-cause analysis, and then try to count up the times that you’re seeing that causal factor.
And often, you’ll see this 80:20 rule – the Pareto principle – that really 20% of your causal factors are contributing to 80% of your opportunities, and that’s where you focus. You have a list here of 20 different potential items contributing, but you can focus on a small number that can lead to a greater amount of improvement.
Let’s dive in a little more into glucometrics and insulinometrics. I see some questions coming in specifically about glucometrics. We can have an exhaustive list here of the type of metrics around glycemic management in the hospital. I’m going to highlight a few in those four categories. Structurally, going to the basics – do you have that committee, do you have those champions, do you have those diabetes educators that you need?
The left lower quadrant – the outcomes – we’ll highlight some more about glucometrics, but also those clinical outcomes, which certainly take longer to see but could be important to monitor as well. And return on investment – financial impacts – are also important to measure outcomes.
But a really core piece, again, that we’ll talk about are those processes – those sort of shorter-term wins – really understanding what’s happening. Are you seeing how you’re doing around ordering, around specific workflows, simply checking an A1c, which is essential for deciding plans in the hospital, recognizing stress hyperglycemia or discharge plans, and then again those balancing measures – hypoglycemia to hyperglycemia? Or if you’re working on length-of-stay improvements, is that impacting other things unintended, such as readmissions?
So a deeper dive – many of you may know this around outcomes for glucometrics, but we thought this is essential to review if you’re not familiar with glucometrics, because how you present data at your facility may vary from the left side here, event – all blood sugars are the denominator – patient day metric, which I think many of us in this space hear about and try to use, or patient stay metric, where the denominator is the number of patients.
The reason I really want to highlight this is, again, not many hospitals are reporting data. If they have data around glycemic control, it’s often in that left side, the event, the all blood sugars, where they’re looking at all denominators. And if you’re looking at something like hypoglycemia, which is probably one of the top glycemic metrics you’re looking at, your executive team may just see 0.1% and think that, oh, severe hypoglycemia is very low, we’re doing well, and then sort of move on. When in essence, that 0.1%, with such a large denominator, could be 10,000 blood sugars at your institution in the past year with less than 40.
And the patient day number could be certainly more significant. It accounts for length of stay – really one of the most clinically useful. Patient stay is another good one, where the denominator is patients, but can certainly overinterpret data. It’s easy to interpret. You’re just saying, hey, I had five patients –10% of my patients with less than 40. But sometimes, it could really fall on the other extreme, where it seems like too much.
So patient day is the one that’s most clinically useful. We’ll talk about that. But you’ll see, if you try to work on patient day, it is a very complicated measure, and often many institutions can’t calculate that.
This Slide – largely for reference, I’ll highlight it quickly here – on those variabilities, again, between patient stay, patient day, and patient events. On the left here is highlighting 10 patients in the hospital for five days. They’re having four blood sugars a day. So that’s 10 patients, four blood sugars a day, five days, 200 blood sugars. 10 times five – 50 patient days. Patient stay is 10. And you can see how the results can vary here dramatically between 8% for all blood sugars, patient day – 22%, patient stay – 40%. We at Glytec, when we’re reporting data out to our sites, use the patient day level. SHM really also has tried to focus on the patient-day level. But that can also be a very complicated level to measure.
And this is on the simple end. There are certainly more metrics that you might be looking at. Somebody asked about patient-day-weighted mean, which is another important measure as well, but I’m trying to even wrap my hands around this for many sites.
And then we know in the future, the next few years – this past year has introduced the opportunity for continuous glucose monitors in the hospital. They’re not approved yet for inpatient use. They’re allowed during the pandemic. What could this bring in the next few years, once we have continuous glucose monitors? We’re barely wrapping our hands around four blood sugars a day, just trying to think – the AGP report on the left here is the ambulatory glucose profile for just one patient. What is the future going to bring as we’re thinking in all sorts of other metrics – time and range, glucose variability, average glucose?
So I’ll spend a few more minutes diving a little bit deeper into process metrics as well. At the basic level, thinking about what you’re looking at – IV insulin – we have different protocols than subQ insulin. I could have added more bullets here if you’re looking at your bypass patients, DKA patients, which also will have different processes. Well, let’s make it simple. If you’re looking IV insulin for process metrics, one easy one is IV blood sugar delay. When you’re managing IV insulin, you’re expecting blood sugars to be usually hourly.
I show an example here. We had one institution that had high rates of hypoglycemia while using IV insulin, trying to understand what happened. And as we were looking at the IV blood sugar delay and really trying to drill it down, it was recognizing that one nurse was accounting for a large portion of the delays. And just working on one simple improvement, we were able to reduce a majority of the hypoglycemia events where the data really began to tell the story.
Certainly, another simple process around IV insulin is just your utilization. You have processes and protocols you put in place – why we’re starting IV insulin for DKA, which patients critically ill should be on IV insulin, which patients on perioperative. You’ve laid those out. So then monitor that utilization. Are people following what you have laid out for the indications?
SubQ insulin – you could see a variety of ordering practices that could be helpful for process metrics. If you wanted to focus on one area, there’s a lot of processes around the mealtime triad that can have measurements to understand if you’re hitting what you’re educating to your nurses on targeting that blood sugar check, the insulin delivery, and the meal delivery. And then the timeliness of hypoglycemia rechecks – if you’re having one low blood sugar, you can manage and check those rechecks for hypoglycemia.
So this highlights a visual for that simple ordering practice around non-ICU management. Basal-bolus ordering on the left – we unfortunately still see a lot of our colleagues using sliding-scale insulin, which is not the recommended practice, so you see initial audit of charts will have a large amount of sliding-scale insulin. Put some improvements in place, start seeing driving towards more basal insulin, and over time, watching those improvements to see basal-bolus.
And that’s just the first few steps. I sort of lay out the outline here on the usual path for physicians. Providers are moving from sliding-scale insulin only, start adding that basal to basal-bolus I just mentioned. And then if you’re achieving success there, you’re going to dive in deeper. Are they getting the correct starting doses? OK, they’re getting the correct starting doses. Are we making the right edits if we’re having glycemic excursions? And you can see how process metrics can really give you a lot of great information.
Now, this can all get very exhaustive. I mean, the list of metrics – we can go on and on. I’m sure you’re thinking of many others that you can utilize. So again, back to that point of going to the fundamentals – thinking through your stages of change, what you’re trying to improve, and the most important question at the beginning, what’s your aim? What are you trying to improve here? Develop a clear aim and think through your audience. You’re working on a quality-improvement project. Is this data needed for that team? Is this data for an ROI that you want to get to your financial team and your executive team? Think through your audience – who’s going to be seeing this dashboard, who’s using the data – and think through that user journey on how they would manage this data, remembering those basics that you have these various metrics to help drive that change.
So the user journey – the patient journey is part of this to think through, right? So we start IV insulin in the emergency room. We might have a different set of metrics and processes you’re looking there. Transferring them to the intensive care unit – while in the ICU, they’re moving from IV insulin, transition to subQ insulin. And then on the floor – the non-ICU setting – subcutaneous insulin management. And then there’s a whole other process and metrics you could think through for the transition to home and certainly outpatient. Here at Glytec, where I work and have Glucommander insulin management software, we have modules that try to target each of these areas, because each of these do require different processes and management of insulin between IV, transition to subQ, subQ insulin.
And then the journey for your institution – so what we’re talking through around data is trying to understand the problem. What’s the diagnosis for what we’re trying to improve? And then that data can help clarify what the treatment plan is going to be. If you’re using largely sliding-scale insulin only, then we know the standards of care. If we can move towards basal-bolus and get those right starts and get those edits, then you understand what the treatment plan will be on your improvement journey.
So I think we’ve gone through a core bit of foundation information. We thought we’d spend the last half of the session really thinking through examples that we’ve seen in our experience – some real-world examples and some test cases that help illustrate the points that we’re making here. So I’ll pass it back to you, Kendall.
ROGERS: We’re going to start by talking about situational awareness – the real-time visualizations. And I know we just covered a lot of information. I want to encourage people to jot your questions in while you have them, because we’ll get to those at the end of the lecture, because now we’re going to kind of dive into the weeds on just some specific examples.
So with situational awareness, we first all have to recognize – and likely we are aware of the features that need to be in place for us to create high reliability within a system. We know that we have to make it easy to do the right thing – having the default action as the right thing to do, having redundancy that ensures that things don’t get missed, having alerts but not too many alerts that are hitting the right person at the right time to effect change in whatever they are doing. That we’re focusing on workflows to make them simpler, not always just adding another step to try to prevent errors. And obviously, trying to standardize care across a spectrum.
But each of these are needed as a bundle to try and achieve targets of very high reliability. Greg Maynard and Jason Stein put out this graphic almost 10 years ago related to VTE. And really, it was through our mentored implementation programs – what we were seeing in many institutions that we were mentoring is when they had no protocol, but they knew what the best practice was, they were in general getting about a 40% success rate of every step of whatever that practice being done correctly. Once they started adding some nudges to get that, they could get up to 50%. Once they got to an order set and that order set was well integrated in the workflow, they could get anywhere from 65-85%. And when they enhanced that protocol and made sure that it was meeting all of our goals for a really well-designed order set with clinical decision support built in, we were getting around 90%. But to get that last 5-10%, it required a different strategy.
Really, that strategy is kind of coined to be the term measure-vention, which really is being able to measure and intervene in real time. And that requires some form of active surveillance, identifying and measuring whatever the process is that you’re following – either outcomes or processes – but that being transparent and able to intervene on that in real time, and these improvement interventions directly coming from the performance measurement that you’re doing, both in real time – but then it also guides the system changes you need to make to improve the system.
An example of this that they’ve published – this is out of Jason Stein’s data – was this VTE measure-vention showing that they had had, on the left side, an enhanced protocol. And this is exactly what we were seeing. They went from kind of that 60% range up to around 90%. But it was only when implementing some form of real-time measure-vention that they broke that 90% and got closer to 100%. And on the right side, it shows the same thing in three different units – one in a different hospital – but showing that real-time surveillance as being effective.
Greg Maynard had a similar example, but in glycemic control. This was active surveillance with measure-vention. It was using a nursing super-user who was getting daily reports of new patients on hypoglycemic agents and had regular measurement of patients who were off protocol and able to intervene. In the case study that they wrote, this person was getting a report of all patients who had readings over 180 or less than 70. They were collaborating with the hospitalist that they were working with. They were able to look at the blood glucose trends, identify factors that were associated. They were reviewing powerplan usage to make sure people were ordering things appropriately. And they were looking at providers who were off protocol and, again, intervening in real time on these.
An example of this from the University of New Mexico – this is our homegrown system that we built – real-time dashboard that we call Cachet (sp?). You likely can’t read all the data on here. But what’s most important is this was my team last week. It had the patient information, which I’ve deleted off of here. But in one view, I can get situational awareness for if my patients are meeting the quality indicators that I would like.
So this red-green-yellow system – this is for have they received their COVID test, or are they still pending? Their VTE prophylaxis – are they on appropriate – green would mean that they are receiving low molecular-weight heparin or a heparin equivalent. And then you can see our glycemic control column – whether the patient is controlled over the past 24 hours or uncontrolled. And this is just based on how setting the parameters – two readings over 180 in a 24-hour period or a hypoglycemic event would both turn these red. Being less than 100 would turn these yellow. But also, we have it for central line, Foleys, code status, cardiac monitoring, SCD orders, etc. So in one view, you can have a very good assessment of what’s going on with your patients. This can occur for the team, but also the nurses can pull this up for individual floors, and they can know that the patient in room five is not meeting one of these metrics and be able to intervene.
That’s talking on an institution or a floor level, but I want to point out that this also can occur at the patient level. The patient level is an insulin single view. And it is a goal of bringing in all of the data that’s needed to make decisions. On this particular slide, the line represents the CBGs, and each of the dots on it represents a different type of insulin. But it also brings in the labs that you have that are pertinent – the A1c, their creatinine. It shows the other medications that might be affecting this. And this is a similar situational awareness for a clinician to be able, in a very quick view, to have an idea of exactly where their patient stands at this time.
MESSLER: I want to highlight a practical tool that we work with sites to try to implement – these visual management dashboards. These kind of dashboards – you may be using them for similar quality-improvement areas at the unit level – I’m thinking unit visual management dashboards. Or you might be using them around (inaudible), falls, other aspects that you’re trying to regularly improve. But when done well, they can really provide regular feedback that require real-time data – meaningful data – often put run charts on there. And then they allow for this goal, which we’ll talk about again – is transparency – really being transparent about transparent about information with your team in as close to real time as you can.
When thinking about a unit-level visual management dashboard, you’re going to want to think about measuring what matters, measuring something that facilitates process change, again, that transparency, highlighting some wins, and sharing and spreading that change, and auditing and giving effective feedback.
And a true visual management dashboard that you might have at the unit level around glycemic management would be something that’s incorporated into workflow, something that you might use at a huddle. Even if it’s just once a week, you stop by the board and talk about glycemic management, maybe revealing some recent data, reviewing some patients in real time and really highlighting and making this part of the workflow.
So what is an example of this that could really highlight a lot of the areas that you’ve heard from Kendall about reliability, about change management? Having a dashboard that incorporates many of these qualities can be difficult to build, but when instituted well can really drive change in itself – can really raise a lot of awareness for your unit around a topic that’s important. And here, we’re talking about diabetes and glycemic management.
Have some providers – a nurse of the month – that are rewarded and recognized for their good work. Put your mission statement – your practice vision around glycemic management. Some easy resources to provide education – cases, tips of the week – current cases that highlight things that you’re talking about. Or maybe those real-time awareness – if you have that access to that data that Kendall just mentioned – showing the team and highlighting monthly wins around those various metrics – structures, processes, outcomes that you’re working on, and highlighting some key aspects of the data.
So you could see in one visual that you’re sharing a whole story on your journey of glycemic management. Have it placed in a prominent spot. Some sites do this digitally. Some sites do it with simply paper and putting up the information regularly. But if you have an owner of this, if it becomes part of a workflow, this could really drive change in and of itself.
ROGERS: Yeah, so that leads right into kind of the art of storytelling. Much of what we’ve talked about here is focused on analytics. It’s focused on how you use this data and how you present this data. But as all of us know, some people respond very well to data and analytics and can look at a graph and they will be motivated to change, but others really require linking that data to an individual. And that’s where it’s important to really focus on stories – stories of individual patients.
Robert Campbell stated, when behavior is fueled by emotion, it’s more likely to last longer than when fueled by analysis. And all of us know this. If we can generate some type of emotion, we are more likely to sustain some type of change. But also recognize not all emotions are the same. Negative emotions are motivators, but if we can actually start driving positive emotions like trust and optimism and pride, that’s where we’re really motivating people to do the right thing when no one is watching. Those are the types of emotions that are going to drive that change.
So the question to you is how can you use the data that you currently have or what’s happening in your hospitals to motivate that type of change? And the most important thing that I would recommend for you to do is to cultivate those stories. We are not normally taught on storytelling. But behind every one of those graphs, there is a patient. Our tendency is we got a patient safety report of a hypoglycemic event, so we got all the data, and we’re going to present what the percent of patients that were impacted.
But I want to implore you to focus – to go back to the patient who that was reported on – knowing something about that patient and being able to cultivate that story, even to start your presentations with those stories. So I had an 88-year-old female who came in for pneumonia and suffered a hypoglycemic event, and these are the factors that led to that. Especially if you’re sharing the factors that those people are under control of, that is going to be more moving than any amount of data or any amount of recommendations that you’re making. That’s how we can get those to stick. So please use that and always have a combination of individual patient stories along with your data presentations.
MESSLER: One of the questions I get repeatedly is, well, what’s the benchmark? What’s the benchmark for glycemic management? I wish I had an easy answer for you. Some of this is going to be answered for us over time if we do have a national measurement like the CMS measure that’s being proposed.
But if we’re looking for benchmarks, there’s a variety of ways to begin thinking. You have glucometers that have specific software solutions to compare within that glucometer group. You can certainly talk to your glucometer software company. But often, that’s at that basic level – the percent BGs. The Society of Hospital Medicine has benchmarking tools. We try to use an eQUIPS system that you can sign up for. Certainly talk to sites about using yourselves over time – and not just yourself, but you can see from what we’ve showed before there’s often spread, so find those best practices. At Glucommander, we have data that we share with all the sites to provide benchmarking. And certainly, you can look at national studies, although those are obviously controlled practices, and use those as a guide about what can be achieved.
One of the areas that I share with sites around benchmarking is looking at the spread. And you can see within a hospital, maybe you have a variety of units within a system, a variety of hospitals. And your benchmark – you might have an average in that middle there that might fall around the 4% hypoglycemia, 25% hyperglycemia, and that may – you pick as your benchmark. But that’s your basic average. If you start seeing that spread and look at your top performers on the bottom left, that may need to be your new benchmark. That’s where you should be striving for.
ROGERS: Now, we’re going to focus on kind of sustainability. How can we use metrics to maintain that change? And we could have led with this, but we thought it would be a good time – try to improve my credibility within this. Jordan just went through some of those benchmarking sites. This particular one is from the Society of Hospital Medicine that we developed during our mentored implementation program, and that sites can upload their data and it will generate graphs for them out of many of the metrics that Jordan – Glucommander does something similar to this as well. It’s just all of my data is in this one.
So the blue graph here is showing hypoglycemia rates less than 70. The middle line is less than 54. And the bottom line is less than 40. And if you haven’t noticed, the X axis – if you can’t read that – is 13 years of data, so this is back to 2008. Actually, I wish I could get 2006 to 2008, because it’s 2006 when we moved away from sliding scale and started our powerplans. So I am confident we had much worse outcomes even before then.
But you can see, over a 13-year period of time, we’ve not only sustained improvements, but we have continued to change over time. And these are utilizing all of the strategies that we just talked about. This started with trying to move from sliding scale to basal. Then it’s recognizing that we had to focus on basal – basal was better than just sliding scale – but we really had to focus on getting people to give the basal at night. But the patient isn’t eating. Then we had to have a focus on basal plus bolus. We had to focus on meal delivery with insulin timing. Physician – their ordering – what’s their initial ordering? How do we prevent that they’re leaving the orders unchanged?
At each of these points that we have continued to improve, our glycemic management team has chosen a different target to start focusing on. But we’ve never stopped. We could have stopped when we went from 9% to 4%, and we could have said we’ve had a 50% reduction in hypoglycemia, as if that was enough. But it’s not enough.
We need to continue driving change throughout, and that’s really what this graph shows. Your goal should be continuous improvement, avoiding complacency, again, celebrating those successes, but at the same time, raising the bar for what it is that you’re trying to accomplish right after that. Continue to push that target and recognizing that your work is never done.
I will admit that when I started this in 2006, I thought I’m going to knock this one out in the next year or two, and I’m going to move on to all these other things that I’m interested in. But this work is never done, and there are opportunities for us to continue to improve.
Another example of this was out of Grady. And this is just showing, after they implemented an electronic glycemic management system, which was Glucommander in this example, they were able to drive. The blue line is showing what their pre-implementation hypoglycemia rate was. They were almost able to immediately eliminate that hypoglycemia rate. But showing that over time, they were able to sustain that. And again, success may be sustaining very excellent outcomes as well.
MESSLER: In these last few minutes, I want to go through a case that really highlights the importance of understanding standardization and really being able to use the data to recognize the variation to implement an improvement plan – and again, iterating the importance of sharing that data – not just keeping it to the glycemic team, but sharing it.
So we know our basics. We know the standards of care that we want to strive to – in the ICU setting, basal-bolus is the preferred method for managing, with a goal of 140 to 180. From this first randomized control trial looking at non-ICU setting of insulin management, basal-bolus insulin clearly outperforms sliding-scale insulin and is the standard for how we want to manage the non-ICU setting for patients needing insulin that are eating.
The other standard that we know is that that split of basal-bolus should be 50% basal, 50% bolus. And from that RABBIT 2 trial and others, the starting dose is generally around 0.4 to 0.5 units per kilogram per day, particularly if they don’t have hypoglycemia risk factors, which would require a lower dose.
So we would look at – when we’re working on our improvement plan, we look at that spread. We’re concerned perhaps about hypo- or hyperglycemia. Maybe here, we pick out one of these outlier hospitals on the top right to focus on. And that’s what we did for this sort of analysis. There was a site that was concerned about high rates of hypoglycemia, but they were saying they weren’t using that much insulin, or they thought they were using it per the standard.
So we would expect, if they were following the standard of basal-bolus, that they would have starts around 0.5 units per kg per day. That’s the green line. But you can see, visually, all the dots below the green line. 68% were less than 0.5 units per day, and even almost half were less than 0.3 units a day, so clearly underdosing. Yet they were having hypoglycemia.
So we looked a little further. One, those low rates, the less than 0.3 – that blue line was a very similar line to that sliding-scale insulin graph on the RABBIT 2. They just were not getting patients in control. Even using less than half of recommended, 0.25, they were beginning to at least get into control.
All right, so they’re using underdosing – less than 0.5. How are they using their basal-bolus? Well, we looked at the split of patients. They should be on 50% basal, 50% bolus. The green bar in the middle would be all those patients that are around 50%. And you can see way too many dots outside the green bar. About 50% were outside of that 50/50 split. So they weren’t dosing it correctly, and a lot of them were really up on that top line – 100%.
Well, they were underdosing, yet they were having hypoglycemia, but they were using too much basal insulin. So we thought, well, if you’re using too much basal insulin, even though your total daily dose is underdose, maybe you’re going to see hypoglycemia in the morning. Sure enough, that’s what we show from this visualization. Almost 50% of the hypoglycemias were morning. And the treatment plan, after this diagnosis, was going to be get back to those standards of care. Monitor this over time, so you’re getting closer to that 50/50.
So that was a whirlwind tour of a lot of examples that we’re using in practice, a lot of ways to think through intrinsic motivators of data to help drive success, help create that sense of urgency – that burning platform – measure improvement.
Well, if you’re not doing all that, these extrinsic motivators are going to come. This is proposed by CMS. They have two medication-related adverse events that are currently in proposal that are open for comment until June 28th – a final decision at the end of the summer. And they could begin to take place by October. These will be elective measures in the IQR system. And there’s one around patient stay for hypoglycemia and one around patient days for hyperglycemia. I won’t go to the details. Recommend that you look that up, begin to understand it. And this is more so than ever – begin raising that awareness in your organization why glycemic management is so critical.
ROGERS: So as Jordan said, we’ve really gone through quite a bit here. I want to encourage you to get your questions in. Any questions that we don’t get to, we’ll try to reach out to you directly to get those answers, but we should have time for a few of those.
We went through some change management, data basics. We talked about real-time visualization, storytelling with data, sustainability, transparency, how to use standardization and transparency to find variation, and we talked about benchmarking in this. So we want to thank you all for your time, letting us talk on these topics. And I’m going to turn it over to the moderator for some questions.
MASSON: Thanks so much. Thank you, Dr. Rogers and Dr. Messler, for a great presentation. We’ll begin today’s question-and-answer session. So audience members, you can submit any questions you have by typing them into the chat box on your webinar console.
Let’s get started with – one audience member would like to know what EMR do you use in your New Mexico study, and was your EMR team able to set that up for you?
ROGERS: At UNM, we’re a Cerner site. I will say most of the major EMRs you are able to modify. I would recommend to people, don’t look for all your solutions within your primary EHR. Best-of-breed software solutions, third-party software often can do these processes better than what’s available in the big-box programs. Our major EMRs want to be our end-all, answer-all-of-our-problems thing, but I would encourage you to look outside of that.
So we were able to build a lot of this with our designers at UNM, though I am sure that many of you, if you have robust enough of a IT department, should be able to do the same. But even if you can’t, there are products out there – Glucommander, the SHM, the slide that Jordan had that listed to you others that can help with that.
MASSON: Thanks so much for that answer. Another audience member would like to know – they find real-time data collection very consuming, so do you have any recommendations on how to implement the measure-vention for a busy team?
ROGERS: Yeah, I wanted to answer this one. Just one, I’m so glad that you mentioned this, because I do just want to state that data collection is the bane of quality improvement. Unfortunately, if you are spending 80% of your time trying to get the data, you barely have energy left to act on that data. So you must push your institution to develop automated ways to get this data.
And I will say, especially for glucometrics, coming up with a good graphical display, knowing which data to throw out, which data to keep is incredibly complex. And the likelihood that you’re able to do that at a local level and to create these glucometrics effectively that you’re going to be able to measure over time – not when that quality consultant left your organization and someone else started measuring it a different way – I would strongly encourage you to use one of those benchmarking metrics.
Obviously, if CMS ends up accepting one of these, hopefully, we will have a consistent metric to use. But most commonly, you can go to your glucometer company. Often they have some type. I would make sure that they’re giving you the information the way that you need it. Or explore one of these other third-party ones for being able to manage your data collection. And then if you have the time and energy within the organization, try to build some of those single views or reach out to organizations that have done that.
MASSON: Great. Thank you much. In the time we have left, we’ll grab one more question. Are there any tangible tools or templates available to highlight organization-specific metrics?
MESSLER: So I think we tried to share a few – something like the visual dashboards that you could create. As Kendall just mentioned, there are third-party softwares that are trying to establish some standards for metrics. When I talk to sites, it’s really understanding your specific aims and objectives, so it’s hard to really come out with a template for this. I think the regulatory measure is going to sort of force us to have some standard metrics that we’re looking at. I hope that this type of talk begins to help frame where you want to look at process metrics and what outcomes you want to look at and think of how that impacts your user journey at your site.
MASSON: Thank you so much. And that is, unfortunately, all we have time for today. But I do want to thank Dr. Rogers and Dr. Messler for an excellent presentation and Glytec for sponsoring today’s webinar. To learn more about the content presented today, please check out the resources section on your webinar console. Thank you so much for joining us. We hope you have a wonderful day.
END OF FILE