An in-depth look at the patient journey from admission to discharge during identifying key moments where clinical inertia can impede optimal care - and what you can do about it.
Get the Slides!
Hannah Day: Thank you for joining us. My name is Hannah Day. I'm the VP of Clinical Practice at Glytec, and I have the privilege of introducing our next speaker in session. We're so honored to have Dr. Andjela Drincic, who is presenting on overcoming therapeutic inertia during the patient journey. Dr. Drincic is a professor of medicine at University of Nebraska Medical Center, where she also serves as the medical director for the Diabetes Endocrinology Center and Medical Director for Hospital Glycemic Quality and Safety. She also served as a panel expert updating the recent 2022 Endocrine Society Clinical Practice Guidelines for hyperglycemia in hospitalized adult patients in non critical care settings. I will turn it over to you, Dr. Drincic.
Andjela Drincic: Thank you so much, Hannah, and thank you all for having me speak on this important topic.
So I'm going to talk about therapeutic inertia and why does it matter, and I'm going to really take us through some strategies for overcoming therapeutic inertia throughout the hospital stay from admission to discharge. So, therapeutic inertia is defined as a delay or inaction to initiate or intensify therapy when glycemic treatment goals have not been met.
It is a well known entity in outpatient world, and as a matter of fact, what we have noticed is that over many, many years, despite new medications, despite new technologies and new system approaches to care, um, The A1C really kind of hasn't budged. So the American Diabetes Association has taken it to heart and started an outpatient initiative for overcoming therapeutic inertia.
It is providing tools to providers to really deal with this important problem. And in inpatient care, diabetes inertia also exists, it's very prevalent, and really we recognize it as a problem since ever since the studies pointing to the importance of inpatient good glycemic control have emerged.
So this is since Vanderburg study sort of rocked our world, we have been dealing with this. The interest in solutions really parallels the electronic medical record adoption and development of clinical decision support systems. And metrics are mostly pragmatic. Now, when one does a literature search for inpatient therapeutic inertia, it's really not as easy to come up with solutions and sort of recipes the way we have it in the outpatient setting.
And this is just example, a few early studies alerting to this as a problem, but proposed solutions are kind of hidden in a fairly large body of literature, but one needs to kind of know what one is searching for. The inpatient inertia really is being less vigorously measured. Like in outpatient setting, we have definitions.
We say, well, how do we measure therapeutic inertia? We'll take a population. Look, how many patients have an A1C above, say, 9 percent with no recent visit to the doctor or no therapy change, and then calculate. There are even formulas as they are presented here. In the inpatient setting, we measure inertia through outcomes such as glucometrics, are sugars good or bad, right?
Insulinometrics, are we using mostly sliding scale or basal bolus? What is our recurrent hypoglycemia and, of course, what is the change of therapy? Is there evidence for a change in insulin as a response to out of range point of care blood sugar? Insulinometrics, indeed, is a good way to look at it because we evaluate the use of sliding scale and look at the transition, towards more use of a basal bolus, and that is a good way to measure therapeutic inertia.
Now, the reasons for therapeutic inertia in inpatient setting really are the same as the categories that we are well aware of for those in outpatient setting. First of all, there are professional barriers, right? So patients are not here for diabetes and hyperglycemia. They're here for other conditions. CHF, COPD, and competing priorities do get in our way.
There are clearly knowledge barriers. We think by now we should know how to deal with it, but despite guidelines, you know, the translation of this knowledge has been difficult, but also, for many situations, we kind of don't quite know what is the best way to manage certain hyperglycemic conditions. And there is a fear of hypoglycemia.
Insulin has a very narrow therapeutic index range. It's very easy to get a patient low, and we are all scared from that, right? And then the protocols that are given to us to help us sometimes are not really all that useful and can be user unfriendly, right? Then there are system level barriers. Those include complex hospital processes, frequent transitions.
Then, the way the data is presented to us is not always conducive to help us make a change that's needed. Our protocols are complicated, sometimes outdated, maybe ineffective, and many times elective. And also, there is a very limited opportunity for real time intervention. On a patient level, the barriers still exist.
The eating patterns and choices can change and family can bring a meal and really kind of mess up all our whole plans. Patients are studied on steroids, they're made NPO, they are given enteral nutrition or parenteral nutrition, and all of that contributes to therapeutic inertia. So solutions? For inpatient care, again, mimic those that are proposed for the outpatient care.
And for those of you interested, there is a very nice paper from the ADA, how to address the therapeutic inertia three year initiative through education, motivation, support, data, pharmacists, communication, engagement, all of this, right? So, many of the solutions that we talk about for solving inpatient therapeutic inertia are based on clinical decision support systems.
So since we will be talking quite a bit about those, I'm going to just start about a little bit about explaining them. So what are the CDSs, right? These are the algorithmic designs that are in place to help provide support for disease process management. And not all CDSs are as effective. Some are more, some are less effective.
And this is a very nice review from Picardo and group on what are the ways to design a really good clinical decision support system. First, a clear gap needs to be identified that needs to be addressed, right? The problem needs to be defined. Next, we need to define the population that is targeted.
After that, we need to identify discrete EHR data elements. That can be searched, and by the way, automatically, not manually, to address the clinical question that we have. And after that, sure, we can develop a program and decision rules that use these data points. To help us address the clinical problem and providers that need to be faced with a recommendation in real time or the data needs to be displayed in a way that providers can make decisions.
And after all of that is done, we need to see, is the CDS sort of any good? Is it resulting in outcomes that we were hoping for? And for that, we need to have the data. And, basically evaluate the outcomes that were obtained utilizing the CDS. So, some of the examples for clinical decision support include, but those are not limited to computerized order sets, glucose data networks and visualization, case finding tools, insulin calculators, electronic treatment protocols, active surveillance, alert systems, remote management, and many others.
So, one can read some, a really nice review by the authors in current diabetes reports as listed here.
So, now, how do we choose and decide with clinical, which clinical decision system is helpful and which one isn't? So, first of all, having no protocol and relying on clinicians to make good decisions has clear, you know, pitfalls, a state of the nature and it isn't likely to result in a success rate in treatment of certain condition.
Now, decision support can exist, but if it doesn't link to order writing or prompts within orders, again, it'll improve the status quo, but only to a certain degree. It's the well integrated protocols that are integrated within EMR, and specifically, if they have a real-time intervention component to it, that result in the better outcomes.
And finally, really it's when we marry the people and technology, when we add models of care to enforce the clinical decision support system is when we achieve outcomes. So now I'm going to take you through some examples of what this means in patient care. Admission is a busy time. We need to identify patients that need insulin.
We need to monitor them and apply those standardized orders.
You would think that insulin order sets would solve all the problems and I'll take you through our journey, right? We have had so many versions of our insulin order set for admission that I have literally stopped counting. But the first version was something along the lines, well, stop oral medications and start basal bolus.
Sure, easier said than done. The version 2 had a handout. What does that mean? What should be the dose? Right? It was a very nice trifold that was meant to be in a provider's lab code, which, you know, gets lost, right? I was very proud when our version, I don't even know, 37, was actually created. And we have, in the EMR, given clear guidance for the weight-based dosing of insulin based on patient specific criteria, right?
Whether they're elderly, what is their BMI, do they have type 1 diabetes, you know, are they overweight, are they on steroids? Difficult, different basal doses were recommended. But you know, now we're still asking providers to do more math, and you can plug in the wrong data, right? So, our latest version is fully integrated.
Not only do we have a basal insulin recommendation that is displayed for specific patient characteristics in the units for kilo. But also, our system is calculating automatically the dose so that providers can check, sort of like do the emotional check. Does this dose really sound right for my patient, you know?
Does the idea to give them 23 units, does that really sound good? Or maybe I should back off and give them a little bit less. So this way, we have, in a way, math and emotions, working together to assure proper insulin dosing. We have developed something very similar for meal insulin dosing. For instance, we have a weight based dosing recommendation for those patients that are eating set amounts of, consistent carb diet, you know, 0.05 units per kilo as a start.
However, in our institution, we do utilize insulin carb ratios a lot, and because it was very difficult for providers to decide, well, what insulin carb ratio should I use for certain patients? We gave them very clear guidance that is really governed by what is the total daily dose of insulin the patient is using.
So, say, if people are on less than 40 units a day, we can start with the insulin carb ratio 1 to 15, and so forth. And this is guidance that's based on a total daily dose of insulin mirrors our correction factor that's listed here for level 1, correction because our providers are used to thinking about total daily dose and they're used to making choices based on that total daily dose of insulin.
So this has really worked for us quite good. I've been very happy, but now I'm so excited that we are in a process of adopting, actually, Glucommander, which is based on very similar principles. So we'll see, more to come on that. Now, when we talk about overcoming therapeutic inertia, and we can give sophisticated clinical decision support systems that will solve some problems, but the question is...
Is keeping it simple a solution to therapeutic inertia? You know, when we overcomplicate things, then it's hard to do them. So, do we have to put everybody on basal bolus? And actually, our Endocrine Society guideline 2022 has addressed this issue in question number seven, which said, should non insulin therapies versus scheduled insulin therapy be used for adults with hyperglycemia with and without type 2 diabetes that are hospitalized for non critical illness.
And based on the evidence, we have concluded that for most adult patients with hyperglycemia, sure, scheduled insulin is the way to go, right? Basal bolus correction. Prior to discharge, it is really okay to consider starting the oral medications. However, based on the new evidence, we also have concluded that for select adult patients with mild hyperglycemia, we suggest using either DPP 4 inhibitors with correction insulin or scheduled basal bolus insulin therapy.
Right? So there is a role to keep it simple, and that by itself these may help overcome therapeutic inertia, but it's very important to understand who are these select patients. And it is clearly listed in this table for you. You know, those are people with recent A1C less than 7.5%, the blood glucose in admission less than 180, total daily dose of insulin needed less than 0.6 units per kilo per day.
And of course, if this strategy doesn't work, then they should be switched to scheduled insulin therapy. So, this pendulum swung from basal bolus and sliding scale to it's okay to do sliding scale and oral diabetes, oral agents for selected people, as I have outlined. And this is from Pascal and Umppierez, a very nice review in Annals a few years ago.
So now we're done with admission. We have written for that initial order set, you know, keeping our fingers crossed that it works, but the chances are it'll need to be adjusted. So, how should that be done? At what level? And you would think it's a simple thing to do. You'll see a blood sugar, you adjust insulin.
However, it's really not simple at all. And lots of ways, including patient specific characteristics, come in the way. And this is again where we need a little bit of help. Truly. Addressing therapeutic inertia during the patient's stay is testing the limits for most order sets. It's just hard. And I will talk about utilization of CDSs in helping us pay attention to hyperglycemia via alerts, glycemic tabs, and dashboards.
But even more so, I'm going to talk about what is the role of systems and models of care to help us overcome therapeutic inertia. So, alerts as part of clinical decision support system, you know, can be dreaded by all of us, right? We all have, you know, alert fatigue, however, it turns out that alerts improve glucometrics, reduce insulin sliding scale use, increase basal bolus use, and even reduce length of stay.
One has to be careful and realize the limitations of this Clinical Decision Support System, but for instance, even the one that I have pictured here, Melbourne Glucose Alert Pathway, that was published by the authors here in Diabetic Medicine 2018, is very rudimentary and really uses color coding both in a meter and for provider alert.
And all it does, it tells them, hey, pay attention. Once something needs to be done. And even sort of rudimentary CDS like that has been shown to improve glucometrics. A really interesting one. CDS has been developed by the group at Penn, by again, Picardo Lewin and, so Penn and Emory group, and they have developed this fairly sophisticated CDS alert system that has multiple components, one, basically, looks for patients who may have stress hyperglycemia, identifies stress hyperglycemia, and in real-time educates providers as to what is the glucose target.
And that gives specific insulin recommendations where it gives them a range of basal insulin dosing to consider and also a prandial insulin dosing to consider. It's not as specific as that I showed you that we have developed, but still fairly in depth. They also are alerting physicians to persistent hyperglycemia and educating them how to make changes to the insulin regimen.
But in addition, they are also alerting providers to the presence of hypoglycemia, recurrent hypoglycemia. And then to that, giving very specific instructions. Okay, so what do I do if my patient has a low fasting blood sugar? Well, then you should change basal insulin and it's given percent change. What if the patient is low before the meal?
And again, it's given very specific instructions. How to make a modification to Prandial Insulin Dose. So, definitely the system has been shown to improve outcomes.
Now, clinical decision support systems come in, in various shapes, right? And, and aside from alerts that I've shown you, dashboards can be very, very useful and I'm illustrating here in this slide because, you see, when we're seeing patients and we have to make changes to insulin based on blood sugar readings, we need to have all the information needed at our site at that very moment to make a good decision.
So, just looking at the isolated blood sugar doesn't mean anything. But! If we have the blood sugar and the trends over time and then when we throw in admission A1C, so we kind of have a sense who is the patient in front of us, and then we have the creatinine and the GFR, and whether the patient is on steroids or not and what is the dose, and what is the insulin that they needed in a previous day, and if they're on tube feeds, what is the rate of tube feeds?
And we have in a nutrition note also the details on carbohydrates. Now it's much easier to make a decision. So dashboards are instrumental as a part of a clinical decision support system to help providers make a good decision. And some of the references are listed here for you. And lastly, electronic glucose management systems definitely have an exciting and new role, right?
This is sort of the future we are hoping for and I am listing, this is Christy Colasso's group has written a very nice review on various eGMS systems from Glucommander to EndoTool to Glucostabilizer, and GlucoCare and they kind of talk about which one's SubQ, which one's IV, what are the perks that they offer, how they deal with hypoglycemia, how, you know, what is the time to target, and they list hypoglycemia incidents for every system.
This is not meant to be a comparative chart. This is just to kind of, suddenly these systems haven't been compared against each other, but it's showing. And overall that we are entering a new and exciting era where we as providers will have some help and the systems are, you know, pretty good at controlling hyper and hypoglycemia.
But as I said, it is only when people, processes, and technology are collectively and harmoniously leveraged to coordinate care that we get the true outcomes that we are hoping for. So, models of care, how can we harness them? There are a few of them reported and I'm sort of listing and just briefly going over a few very interesting ones.
So, one is the glycemic management program by Donahue et al. where they have essentially developed a program that targeted provider response to inpatient hyperglycemia. And, indeed, it resulted in improvement in glucose metrics. Our Diabetes Research Nurse Program, by providing extra education to nurses made them a more effective allies to providers and again, have improved outcomes, including readmissions, in truly visionary virtual glucose management services developed by Dr. Roushikov at UCSF, before the Zoom days, before we all lived in a virtual world, he developed a system where they have a report that identifies people with hyper and hypoglycemia is created. And an endocrinologist is reviewing those records and providing the actual, um, uh, recommendation in a chart for the team to make changes in the insulin.
And again, overcoming therapeutic inertia by directly working with a team and providing timely advice. And that has resulted in decreases of hyper and hypoglycemia. What we have done in Nebraska Medicine is that we have taken this concept that Dr. Rushakov developed and then we have implemented it, but this time by the pharmacy service and our diabetes physician.
Pharmacy stewardship has done something very similar where we have a dashboard and we are identifying patients at risk based on their glycemic control, the hypoglycemia in the previous 48 hours, whether they have type 1 diabetes. And it's, basically alerting the diabetes stewardship pharmacist that attention needs to be paid to their blood sugars.
He reviews the record and places a note in a chart suggesting a change in therapy and then contacts the team based pharmacist so that indeed these recommendations are actually adopted. And this also has been shown to improve glucometrics as we have published. And we are finally coming to the discharge time.
And when patients are about to be discharged, a lot of things need to be done. We need to choose a med regimen, we need to order meds and supplies, and assure follow up, and communicate with primary care doctors, and you know, the easiest thing to do at that time is just to click the button that says, Resume Home Meds.
And, you know, that's not necessarily the best thing to do, right? So, the body of literature that deals with this is listed here for those of you that are interested to read more. But, Dr. Umpierrez has developed a discharge insulin algorithm, published in 2014 and revised it since then, that is suggesting that based on the admission A1c, one can make decisions on how to modify the discharge treatment.
In order to overcome therapeutic inertia, right? So if the A1C is less than 8%, one can restart outpatient treatment. If the A1C is anywhere between eight to 10%, then one can restart outpatient oral agents and still add a basal insulin at a 50% of the hospital dose. And if the A1C is above 10%, then discharging on a basal bolus regimen that worked in a hospital or restarting oral agents, if appropriate, and adding insulin at 80 percent of the hospital dose is a suggested, approach.
Now, when we are writing for discharge orders, we at Nebraska Medicine had individual order entry for every single medication. No access to ADA guidelines, no recommendations for adjustment based on GFR. We had individual orders for every pen and needle and syringe and lancet and strip and a meter.
Absolutely overwhelming options and... You know, it's not surprising that 70 percent of our patients discharged on insulin had none or only some of their supplies. This was mind boggling list. So we basically went into action and created a new discharge diabetes order set where with one click of a button will assure that outpatient management follow up is scheduled, that the guide for resuming oral meds is there for physicians with a click in case they need a reference.
And also, all our insulins and non injectable therapies, non insulin injectable therapies are linked to appropriate pens and needles so that no mistakes will be made and no omissions will be made. And we are excited to report the paper coming out soon that this indeed, has resulted in people going home ready to take the medicines that are actually prescribed.
So, at the end of the day, therapeutic inertia in the inpatient setting is a thing, it's not just an outpatient issue. It occurs throughout the patient journey, and just having the admission order set is not enough. One needs to follow the patient until discharge. That a pathway to optimize the treatment and overcome therapeutic inertia is to harness people and technology and that requires leadership, innovation, and constant vigilance. So, thank you.
Hannah Day: Thank you, Dr. Drincic. Your presentation highlighted the many barriers clinicians face throughout the patient's hospital journey, contributing to therapeutic inertia, and how implementing clinical decision support tools can help simplify already complicated processes and protocols.
Ideally, clinical decision support tools will address the many pieces and processes of proper glycemic management in the hospital. And I'll highlight in the next few slides how Glucommander can be leveraged as a tool to help simplify and standardize best practices as it aligns with guidelines, provides a structure to standardized care processes, simplifies dosing calculations, provides automated dosing adjustments, provides tools for glycemic management teams and leverages integrations within the EMR.
The ADA standard of care in diabetes updated yearly, and the Endocrine Society clinical practice guidelines recently updated in 2022 really give us clinical practice recommendations around inpatient glycemic management.
I'll highlight a few of these recommendations that are key in avoiding therapeutic inertia in glycemic management and how Glucommander provides clinical decision support around these guidelines. The three recommendations I'll highlight are ordering basal bolus insulin therapy for non critically ill patients, the importance of daily adjustments to prevent hypoglycemia, and the importance of specialized diabetes or glucose management teams when caring for hospitalized patients with diabetes.
So why is basal bolus insulin therapy recommended? It closely mimics physiologic insulin secretion. It's required for patients with type 1 diabetes and it's proven to improve glycemic outcomes and reduce hospital complications compared to sliding scale insulin alone. So how does Glucommander support the recommendations for basal bolus insulin therapy?
It provides a structured process for ordering either basal plus correction or basal bolus plus correction insulin regimens using either a weight based dosing approach or a custom dosing approach, and it includes three weight based multipliers based on the literature for determining that total daily dose.
It includes a calculator for converting home regimens to appropriate inpatient insulin regimens, which is helpful to adjust basal heavy insulin home regimens to avoid potential hypoglycemia when a hospitalized patient is either temporarily NPO or has a poor appetite. And why are daily insulin adjustments necessary?
First, insulin needs often change during the course of an acute illness, so even with our best estimate of a patient's current insulin needs, changes are often necessary to reach or maintain glucoses in a target range. But most importantly, insulin dose adjustments are needed to prevent hypoglycemia.
Hypoglycemia, as we all know, is associated with increased mortality, and many episodes are preventable. Glucommander will make daily automatic insulin dose adjustments to reach and maintain the target range. An integration within the EMR is also leveraged to display the projected insulin doses and glucose trends for clinicians.
So let's look at two examples of basal bolus insulin therapy over a three day period in the hospital. First, on the left, here's a scenario we're all familiar with, unfortunately. Basal bolus insulin is ordered, Glargine 18 units nightly at bedtime, and Lispro 6 units with each meal plus a correction scale.
Day 1, the fasting blood glucose is 95. This is below target, but in this scenario, no changes are made to the insulin doses, even though the fasting blood glucose is less than 100, and that's a predictor of hypoglycemia in the next 24 hours. Day 2, the first preventable hypoglycemic event occurs. Still, there are no insulin dose changes leading to a recurrent hypo event on day three.
Research shows that many severe hypoglycemic events were preceded by a glucose less than 70 that same hospital stay. With Glucommander SubQ, let's look how automated dose reductions in the basal dose can prevent hypoglycemia. So, same insulin regimen to start with. Day 1 fasting blood glucose 95. The Glargine dose is decreased from 18 units to 14 units.
Day 2 fasting blood glucose is 115, still below target, so the Glargine dose is further decreased to 13 units. And day three, blood glucose is 125. Basal dose adjustments are important, but so are meal dose adjustments based on carb intake and pre meal blood glucose. The meal triad of checking the pre meal blood glucose, ensuring timely alignment of the insulin injection, with the meal and the pre meal glucose, and then adjusting the prandial insulin dose based on the carb intake.
This is a process that requires standardization. Glucommander will support a structured meal triad process that aligns with ADA guidelines, simplifies meal dosing calculations to prevent errors and automatically adjust the meal bolus dose based on the carb intake. So there's no need to stop and restart meal insulin when a patient's temporarily NPO or maybe experiencing a poor appetite.
Let's look at an actual Glucommander SubQ case review that involves several daily insulin dose adjustments. The initial orders were for basal bolus plus correction using a weight based total daily dose multiplier of 0.5 units per kilogram per day, and this equated to a total daily dose of 102 units, in which 50 percent was the glargine dose, so 51 units, and the remaining 50 percent was further divided by the three meals was 17 units of Lispro for a 60 gram carb diet.
After several days in the hospital and ongoing dose adjustments based on glucose trends, by the end of the stay, the total daily dose had decreased significantly to only 57 units. The chart on the right shows the daily basal dose adjustments based on that fasting blood glucose. So you can see the basal dose started at 51 units based on the physician's initial order.
This was increased up to 73 units by day 3 when insulin needs were the highest. And then down to 23 units by day 12, as insulin needs declined all while avoiding hypoglycemia in the same case. Meal dose adjustments were also made based on carbon intake. The first day due to the patient eating less than 60 grams of carbs, the 17 units of Lispro was decreased to seven units for 25 grams of carbs, 13 units for 45 and nine units for 30 grams of carbs at each of the meals.
Carb intake lessened even further on day 2 to only 0 to 15 grams of carb at each meal. Therefore, the meal insulin doses were 0 to 4 units. And then finally, Glucommander will also calculate a negative correction dose when a pre meal blood glucose is less than the target range. You can see on this day, due to pre meal blood glucoses of 95, 97, and 135, Glucommander subtracted insulin from the total meal dose to avoid hypoglycemia.
And let's not forget the importance of our people, our diabetes specialists. ADA standards recommend consulting with specialized diabetes or glucose management teams when possible. And such teams are proven to result in reduced length of stay, 30 day readmissions, reduced cost of care, and improved patient outcomes as they provide support and expertise for process development and quality improvement.
Glucommander provides a few tools for these care teams to support real time glycemic management. The Glucommander dashboard, which can provide visual cues when action is needed, such as an overdue basal dose or when hypoglycemia rechecks are due. Glucose surveillance is a tool to identify patients with hyperglycemia who may benefit from treatment progression.
And then GlucoMetrics for ongoing data monitoring and support for quality improvement initiatives. So let's face it, implementing best practice guidelines can mean creating complex processes which are bound to fail. We must simplify these processes where and when we can in order to standardize and hardwire best practices.
Rigid or too perfect guidelines can be impossible to achieve. Clinical decision support tools like Glucommander are key in providing a foundational structure and guidance around essential inpatient glycemic management processes. We must not avoid doing what is right just because doing what is wrong is easier.
We must leverage tools to overcome therapeutic inertia and ensure our patients are receiving quality inpatient diabetes care.
Thank you everyone for attending this presentation and thank you again Dr. Drincic for leading us through the patient journey and reminding us how important it is to take steps to overcome therapeutic inertia. If you have any questions for myself or Dr. Drincic, please feel free to reach out and thank you again.