Vancomycin From Trough to AUC: RWJ Barnabas Health Experience | Recorded Webinar

The upcoming changes to the preeminent US vancomycin dosing guidelines have been greatly anticipated by the pharmacy and infectious disease community for some time. The recent draft release of these guidelines has prompted significant discussion on the recommendation supporting a switch from traditional trough serum concentration measurement to an AUC-based dosing strategy.

On July 30th, DoseMe held a webinar “Vancomycin Monitoring Based on AUC: The Implications and Considerations” with Robert Wood Johnson (RWJ) Barnabas Health and parent company Tabula Rasa HealthCare to share RWJ’s experience transitioning from trough-based to AUC-based vancomycin dosing and monitoring.

Introduction by moderator Orsula Knowlton, PharmD, MBA, President Tabula Rasa HealthCare
Precision Pharmacotherapy Research and Development Institute by Dr Jacques Turgeon BPharm, PhD, Chief Scientific Officer Tabula Rasa HealthCare
The Transition from Trough Based to AUC/MIC Based Vancomycin Dosing by Dr Luigi Brunetti, PharmD, A/Prof of Pharmacy Ernest Mario School of Pharmacy, Clinical Pharmacy, Robert Wood Johnson University Hospital Somerset
Step-by-Step: Trough to AUC by Dr Robert McLeay, PhD Chief Scientific Officer, DoseMe
Webinar Q&A

Welcome and Introduction

Orsula Knowlton, PharmD, MBA

Precision Pharmacotherapy Research and Development Institute

Dr Jacques Turgeon BPharm, PhD

The Transition from Trough Based to AUC/MIC Based Vancomycin Dosing

A/Prof Luigi Brunetti, PharmD, Clinical Pharmacy, RWJBarnabas Health

Step-by-Step: Trough to AUC

Dr Robert McLeay, PhD

Webinar Q&A

Learn more

Read more on vancomycin dosing and AUC calculations using DoseMeRx – a bayesian dosing calculator

Dr. Orsula Knowlton: Welcome, everyone. My name is Dr. Orsula Knowlton, and I will be your moderator for today’s webinar. I’m a pharmacist and co-founder, president and chief marketing new business development officer at Tabula Rasa Healthcare. Tabula Rasa Healthcare, or TRHC, is excited to have DoseMe as a part of its suite of medication safety clinical decision support technology tools. We focus on helping people optimize their medication regimens and measure outcomes of care. We appreciate your participation today from all over the United States.

Today’s webinar will focus on the implications and considerations of vancomycin monitoring based on area under the concentration time curve, or what is commonly referred to as AUC. I’d like to go over a few details before we get started. Most of you will want to know if this webinar will be available for viewing on demand, and the answer is yes.

After today’s webinar, you will also receive an email with the link to the video. If you have any questions or comments during the presentation, we ask that you please enter them under the Ask A Question section at the bottom-right of your control panel. We will have a dedicated Q&A session at the end of today’s presentations where we will answer as many questions as time allows. Now, I’d like to introduce our speakers.

Today, your panelists will be Dr. Turgeon, Dr. Brunetti, and Dr. McLeay. Our first panelist will be Dr. Jacques Turgeon. He is the chief scientific officer at Tabula Rasa Healthcare and CEO of our Precision Pharmacotherapy Research & Development Institute in Lake Nona, Orlando, Florida. He was previously associate dean of the Lake Nona Campus at the University of Florida College of Pharmacy and professor in the Department of Pharmaceutics. Prior to that, he was head of the CHUM Hospital in Montreal and dean of the pharmacy school of the University of Montreal.

I’d also like to introduce Dr. Luigi Brunetti. He will be our second panelist. Dr. Brunetti is currently associate professor of pharmacy practice administration at Rutgers, the Ernest Mario School of Pharmacy. His practice site is the Robert Wood Johnson University Hospital in Somerset, New Jersey where he is also the residency program coordinator.

Finally, Dr. Robert McLeay will conclude as our final panelist. He is the founder and chief scientific officer at DoseMe. Throughout his career, Robert has published in the areas of applying machine learning techniques across individualized dosing, glioblastoma, and schizophrenia research. His research focus today is on improving clinical outcomes of translating pharmacokinetics research into clinical practice. So, without any further ado, I’d like to kick things off by welcoming Dr. Turgeon. Jacques, over to you.

Dr. Jacques Turgeon: Thank you very much, Dr. Knowlton. Hopefully, everybody can see my screen right now. I’m going to do it this way, and that should be working. So, thank you very much. As I said, I am the chief scientific officer with Tabula Rasa.

And Tabula Rasa is a company where the focus is on preventing adverse events, ER visits, as well as hospitalizations associated with dangerous, complex drug interactions and improving efficacy outcomes and adherence for at risk patients. We are a company of about 1,200 employees and 13 offices throughout the United States, mostly. We have about 50% of our staff which is clinical, mostly pharmacists, and 150 IT individuals. Our offices, as I said, are spread across New Jersey, South Carolina, Arizona, California, Colorado, Florida, Ohio, Texas, and Minnesota.

Our solution for mitigating adverse drug events is relatively simple but at the same time complex. It’s based on research where we have developed a tool that can identify as well as help manage multi-drug interactions at the same time. So, we’re looking at the drug, the entire drug regimens. We’re not doing a one-to-one comparison. And we’re trying to provide information to professionals so that they can manage complex situations at the same time.

Our tools also include a risk stratification strategy where we develop a mitigation risk score which help professionals identify patients at increased risk of drug-related adverse events. This also helps them identify which of the elements that are the constituents of this mitigation risk score to act upon and try to change elements of the drug regimen so that mitigation risk score or the risk for adverse drug events is decreased.

I’m trying to switch my next slide. Okay. As it was mentioned previously, I spent most of my career in academia. My research interests have always been on drug metabolism and cytochrome P450s. I was one of the first to identify several lysoenzymes of the P450. I spent my career also working a lot on pharmacogenomics, on drug-induced Long QT syndrome where we have identified from my lab a lot of drugs associated with this syndrome and the risk of sudden death.

And finally, I spent a lot of time developing clinical decision support systems that could help professionals identify, as I’ve mentioned previously, drug interactions. And finally, one of the elements that we develop around this was a risk stratification strategy for identifying patients at increased risk of drug-related adverse events. So, spending a lot of my time more than 30 years in academia and trying to publish on all of these issues.

The vision of the Tabula Rasa Research and Development Institute in Lake Nona expands with the emerging research initiatives that include bioinformatics. In there, I mean, we’re PK/PD modeling through either pharmacometrics or system pharmacology. We use different tools with biomedical engineering systems; nanoscience, where we use strategies such as human-on-a-chip, patient-specific informations, data-driven technology solutions, and translational research.

Coming back to the clinical decision support systems. As I said, we have spent more than 25 years trying to develop such systems that can not only look at one drug to one drug but looking at multiple drugs at the same time. And we have published, and that’s one example of one of the publications we’ve done on clinical decision support system, which was an editorial, an invited editorial from expert opinion on drug metabolism and toxicology on what it should be looking like, what should be one of the best clinical decision support system.

And this is what we’re trying to apply on a daily basis in the tools that we’re developing throughout Tabula Rasa, and this is how we’re happy to see DoseMe being part of the family, where we’re helping them or trying to assist them and collaborate with them in trying to develop the best tool possible to help health professionals improve drug therapy, preventing adverse events and increasing efficacy.

So, at the end of the day, what we’re spending most of the time is: How can pharmacists or health professionals mitigate risk associated with adverse drug events? That’s the big question that we’re trying to answer and developing tools around this topic. So on that, without further ado, I think you’re all here to hear Dr. Brunetti. So, I will share my screen now with Dr. Brunetti.

Dr. Luigi Brunetti: Okay. Good afternoon, everyone, and thank you for the introduction. Let me just make sure I have the full screen here. There we go. So, the title of my segment is Transition from Trough Based to a AUC Based Vancomycin Dosing. So, I am an associate professor of pharmacy over at the Ernest Mario School of Pharmacy. And as with many pharmacists, I have many different hats. I’m also the co-director of the Center of Excellence in Pharmaceutical Translational Research and Education where the focus of our center is PK/PD modeling, so very fitting for the topic of today.

At the hospital, my practice site, Robert Wood Johnson University Hospital in Somerset, I’m a clinical specialist in internal medicine. I also serve as the antimicrobial stewardship pharmacist. Some brief objectives. We’ll cover some dosing strategies, evaluate some dosing strategies for vancomycin, as well as Identify some challenges and opportunities for transitioning from one type of dosing to another.

At Robert Wood Johnson Somerset, which we’ll go into in a bit, we’ve had the opportunity, really, to spearhead all of vancomycin dosing since 2011. So, we have quite a bit of experience in getting this type of program off the ground. So, the setting is Robert Wood Johnson University Hospital in Somerville, New Jersey. We’re a member of the RWJBarnabas Health system, and our hospital is a 365-bed community teaching hospital.

So, just some very brief background. This is really the basics here. We all know vancomycin is a glycopeptide antibiotic. It has been around for a very long time, typically reserved for serious gram-positive infections. And although we know it’s an AUC/MIC bactericidal profile, historically, we have used trough-based dosing to guide our dosing strategy.

So, typically, the standard of care dosing either involves traditional pharmacokinetics where we measure drug concentrations to determine PK parameters for an individual patient, we can incorporate population kinetics where we have some patient variables and we use population equations to calculate dosing, and of course, the trough concentration as a surrogate marker for area under the curve.

Now, while we have used trough concentration as a surrogate, and this was even recommended in the 2009 vancomycin guidelines, there is a potential for over-aggressive dosing. And I’ll try and touch upon some of that data that really identifies the pitfall or problem that you can encounter by using trough concentrations as your PK target.

So, just a little bit of history and background on Somerset. Our vancomycin protocol was established in 2011. And at that time point, all of the vancomycin was to be dosed by pharmacy. So, that was quite a responsibility that we undertook, and we did a very good job. We had some data where we looked at prior to 2011 where the initial trough concentration above 10 was very low. It was about 35 to 40%. When pharmacy took over in 2011 and we re-evaluated the effectiveness of our program, we were in the 65 to 70% range in terms of target attainment at the first trough concentration.

So, that was pretty darn good. And really, I think that this experience set us up to move onto the next step which we’ll talk about in a few slides. But essentially, the way it works is the prescriber orders therapy, and the pharmacist adjusts, orders the trough concentration, relevant labs as needed. The initial dose was based on a nomogram using popPK, targeting a trough concentration of approximately 15. The trough concentration would be obtained at steady state. So, we say at max before the fifth dose, ideally before the fourth dose.

And our targets were very typical, I think. The target was somewhere between 10 and 20 depending on the severity of infection that you encounter. The creatinine clearance was also calculated by the standard Cockcroft-Gault equation. We did have some stipulations for body weight depending if someone was obese or non-obese. In addition, we did round serum creatinine to 1 for those aged 65 years or older. So, this was kind of our standard from 2011 till about maybe six months to a year ago.

So, really, what has happened since our protocol in 2011 — or let’s go back even further from the consensus guidelines in 2009, to present where we have a 2019 draft consensus for vancomycin. And essentially, what we see is that the shift has been away or moving away from trough serum vancomycin concentrations.

So, if you look at the table on the following slide, in 2009, trough serum vancomycin concentrations were deemed to be the most accurate and practical method for monitoring efficacy. And that was really based on all of the evidence that we had at that time, and collectively, that was the decision of the group.

The minimum serum vancomycin trough concentration should always be maintained above 10 milligrams per liter. There were also some recommendations. If you have more severe infections, vancomycin serum trough concentration should be in the 15 to 20 range. And of course, trough should be obtained just before the next dose at steady state. So, if we look at 2019, really, what has changed is now what’s being advocated in the draft statement is that Bayesian-derived AUC/MIC targets should be used, and we should target about 400 to 600.

In addition, the most accurate and optimal way to monitor vancomycin is through AUC-guided dosing. The preferred approach is to use Bayesian software programs. And one of the reasons for that is, it’s pretty complex to do Bayesian kinetics.

So, software programs tend to make it a bit easier for clinicians to absorb. And of course, one of the advantages of Bayesian dosing is that you don’t necessarily need to have a trough drawn at steady state. So, really, any concentration that is obtained can be entered into the model and it can be used to predict a better dose for your patient.

So, what is the risk of targeting of vancomycin trough concentration above 15? Well, the risk is that there is a greater association with nephrotoxicity. And just from our experience, to achieve a trough concentration of greater than 15, we would often have to exceed doses greater than 4 grams. And that has also been reported to be a risk factor for nephrotoxicity.

So, why do we target a trough concentration of greater than 15 is the question I posed on this slide. We have data showing that more than 60%, or I should say up to 60% of patients, with a therapeutic AUC/MIC greater than 400 actually have a trough concentration of less than 15, meaning there is a large number of patients that we may be targeting a concentration of greater than 15 unnecessarily.

And the repercussions of this are really that, if nephrotoxicity does occur, we know that in-hospital mortality is higher in this patient population. And we also know that acute kidney injury in hospitalized patients is associated with a longer length of stay by about 2.8 days as well as increased costs.

So really, all of this information here is kind of what we use to make the determination that we probably need to relook at our protocol which was successful in achieving what was deemed to be the standard of care targets previously.

But now, with more data emerging on AUC and the draft guidelines suggesting AUC/MIC as a better target, we really have a strong case for switching. Now, one of the other steps that we did is we said, «Well, if we were to simulate in the background alongside of our protocol the Bayesian-recommended dose, how would that differ compared to what our protocol is recommending?» So, this slide simply shows the results of one of our residency projects.

So, one of our residents did this last year. And what she found is essentially the mean vancomycin dosage required was about 1 gram difference if we would use vancomycin — the previous protocol versus the Bayesian dosing run in the background. And similarly, when we looked at the difference in terms of median, it was also about 1 gram.

So essentially, we found that dosing would be lower if we were to use the Bayesian dose and potentially reducing the risk of nephrotoxicity. We also found that greater than 4 grams daily, that occurred less often in individuals where the Bayesian dose was recommended in the background, 3% versus 16.4% which was our standard protocol.

Some data that I don’t show on this slide is also the association with the AUC/MIC of greater than 600, and we did find in the simulation that many of the patients that develop nephrotoxicity in fact were above the threshold for AUC. So, collectively, our team said, «We need to update our approach.» And we know that our protocol is effective in achieving a trough of greater than 10. We know that our nephrotoxicity rates are there, and it varied in terms of the occurrence of nephrotoxicity depending on the patient type.

So, we found ranges anywhere between 5 and 20% depending on the sub-population that we looked at. Of course, the sicker patients are patients perhaps in critical care where we were being more aggressive. They tended to have higher nephrotoxicity rates. And now, we also have the updated guidelines, and of course, we always want to enhance efficacy while reducing toxicity. Now, we were very fortunate to have a supportive executive pharmacy team on board.

RWJBarnabas Health has an executive pharmacy team, and we first approached them with the idea of using AUC vancomycin dosing, and they were on board with this. And fortunately, we also have an infectious diseases collaborative which has representation from all of the healthcare facilities throughout the system.

And this collaborative actually really spearheaded putting together all the materials, notices, protocols, policies that each institution might need to switch over to AUC-based dosing. Fortunately, at Robert Wood Somerset, we had much of this in place and we were able to share this with the members of our team. And in turn, they tweaked it and really updated it to represent current-day evidence and practice.

Of course, all the team members needed for success have to be at the table. And this is more than just pharmacists and physicians. This also includes nursing. They need to know who to contact, what to do when they see a specific concentration. Of course, the lab should also be involved. We need to also allow individual sites to modify the policy and procedures to accommodate their practice. So, knowing that every hospital has a little bit of a different nuance or a little bit different population, they should be able to modify the policy and procedure to fit. And of course, what’s the best method for AUC/MIC calculation and dosing selection?

So, I won’t go into this slide in too much detail, but there are several methods out there for choosing in terms of calculating the area under the curve. The Rodvold method is one, the trapezoidal approach, and of course the Bayesian-based approach. We elected to move forward with a Bayesian-dosed approach using Bayesian software, PK software. And really, this was based on the ease of this type of an approach as well as the guideline or draft guidelines suggesting that this is what we should be using.

Some ongoing considerations that we have, of course, are our special populations. And this tends to be one of my big areas of interest. We often have these subsets of patients that become difficult to manage. So, one of the first things that we did at Robert Wood Somerset is that we stratify patients according to creatinine clearance.

So, as an example, if we have anyone with a creatinine clearance of less than 40, or hemodialysis, or CVVH, those patients are followed by our clinical specialists or residents. Anyone with a creatinine clearance of above 40, those individuals are followed by our clinical or staff pharmacists.

And we really did that because we understand that there is a time commitment and a workload consideration with the staff. So, we kind of wanted to alleviate some of those more complex patients and allow some of our specialists that have more expertise in those populations to manage that subset.

So, of course, in the obese population, how do we manage those patients? Hemodialysis; what is the best strategy for our hemodialysis patients? We tend to just do pre-vancomycin, random vancomycin concentrations and then dose post-HD.

How do we dose our CVVH? Low muscle mass and low-weight patients, below-weight patients we also have our clinical specialist manage. So, that would be adults weighing less than 40 kilos. And on the flip side, for our obese individuals, especially those greater than 200 kilos, the clinical specialist would manage that patient as well.

How do we document it into the electronic health record, always something we’re trying to improve so other clinicians are aware where to find our information, integration within the electronic health record, and the big one here is management of workload.

And of course, when you’re kicking off the program, there is a tremendous amount of workload involved. So, one of the things we try and stress to all of our staff: early discontinuation of vancomycin whenever possible. So, we really need to discontinue vancomycin when it’s not indicated or when susceptibilities come back suggesting that you can de-escalate therapy.

And finally, share your data. I think sharing your data with your institution is immensely important and does help get buy-in from all the key players. In fact, when we made the decision to switch to AUC/MIC-based dosing. We really had the support of all the clinicians in the hospital because they are had already seen our previous data when we switched in 2011 to pharmacy-based dosing. And I really think that being transparent and sharing your data is absolutely important and getting buy-in from key players in the institution.

Just to conclude and stay on time here, I strongly feel AUC/MIC-dosing is optimal to ensure efficacy and decrease toxicity. There is a time investment for transitioning. However, the benefits in transitioning to AUC-based dosing are too significant to ignore, and it should be something we strongly consider. So with that, I’ll conclude my segment of the presentation and I’ll hand off to Robert McLeay who is the chief scientific officer at DoseMe. Thank you.

Dr. Robert McLeay: Thank you very much, and good afternoon, everyone. Thank you for joining us on today’s webinar. What I’m going to walk through now is an approach to how you can transition from trough to AUC-based dosing step by step. We’ve heard from Luigi about some of the challenges that they went through at Robert Wood Johnson in this transition process. And I’m going to talk today about how DoseMeRx in particular can help as a Bayesian dosing tool.

But before I touch on that aspect of the talk, I just really want to go back to the fundamental rationale behind using AUC targets for vancomycin therapy. And really, the summary, as you’ve heard, is that AUC directly relates to vancomycin efficacy while toxicity is trough-linked. And this paper from Mizukami et al from 2013 I think really shows this very, very clearly. The plot on the left-hand side shows mortality on the y-axis against vancomycin trough concentrations and different ranges for each bar.

And you can see that there isn’t really very much relationship between a particular trough concentration and the mortality rate. On the right-hand side, however, you can see AUC bins for each bar against mortality. And there’s this really clear area where a particular AUC range in this patient cohort had significantly reduced mortality.a

And for me, that really is the motivation. Unfortunately, there are a couple of challenges with AUC-based dosing. The first one is that targeting an area under the curve or an area under the curve to an MIC ratio, we’re looking at targeting towards a derived number. And what I mean by this is it’s not a number that’s directly measurable off a lab. In order to calculate an AUC, you either need multiple labs or you need pharmacokinetic software with a single lab to calculate that individualized AUC.

Secondly though, and perhaps more challenging, is that transitioning clinical protocols is simply hard. There are a large number of stakeholders — and this is the case whether you’re transitioning from trough-based dosing to area under the curve or otherwise — there are many stakeholders. And you’ve heard Dr. Brunetti go through what this looked like at Robert Wood Johnson. And so, it’s really important to do this process.

As a Bayesian dosing tool, our role really is around supporting that change. We provide training, and resources, and 24/7 support in order to assist making the move, but also increase the motivation by giving instant clinical decision support. And really, our goal is to ease that change from trough to AUC-based dosing where you get the instant evidence for a decision, and that can automatically be saved and edited as a progress note in your electronic health record. And you can start immediately with trough-based without requiring physical workflow changes.

So, just before I get into some data from DoseMeRx at two groups of hospitals, I just mentioned that DoseMeRx really is a Bayesian dosing tool that’s built for the clinical environment, that regulated environment. We currently have more than 130 sites, more than 2,000 active clinicians, and are calculating more than 20,000 doses a month.

DoseMeRx can be accessed via the web, via mobile applications, or your electronic health care record. And on the lower right-hand side of this slide, you can see a screenshot taken from within Cerner, and DoseMeRx is live inside of both Epic and Cerner today.

We support many drugs. And while we’re talking about vancomycin today, we also support many other antimicrobials that you might wish to monitor in a therapeutic drug monitoring program but other classes as well such as immunosuppressants, anti-epileptics, and more. And of course, depending on the drug in question, you of course will wish to target different pharmacokinetic targets. And DoseMe supports a wide range of these, including AUC to MIC, peak, trough, time above an MIC, and many more.

As Dr. Brunetti touched on earlier, there’s a challenge with some of the particular patient groups. And so, for vancomycin, this is something that we really have focused on and have grown inside of DoseMeRx. We have both one and two compartment models for vancomycin as well as models for specific patient cohorts. We have an obese model, a hemodialysis model, the pediatric and adolescent model, as well as a neonate and preterm model, so that for these more challenging and perhaps somewhat different patient cohorts, we can continue to provide clinical decision support as well.

So, let’s go through the idea of a step-by-step approach to transitioning. As you’ve heard today, the motivation behind this change in the guidelines is to give an efficacious dose but avoid trough-based toxicity. And trough and AUC are related but not really closely enough. And if we have a look at the data presented on the right-hand side, we have the area under the curve on the y-axis and the trough on the x-axis. And while you can see there’s a correlation, there is still quite a spread. And as Dr. Brunetti said, up to 60% of patients can achieve a therapeutic AUC while not reach your traditional trough targets.

And as trough matters for toxicity but AUC matters for efficacy, both really do matter. And that’s really the motivation behind the design of DoseMeRx when dosing patients. From the ground up, it’s designed to display both. So on the left-hand side, we can see a dose calculated for a particular patient where we’re targeting an area under the curve of 400. We’ve predicted a dose, in this case 1,000 milligrams twice daily, that will have a predicted area under the curve of 409.

But we’re also presenting the trough as well, so in this case 12.8. And whether you’re targeting the AUC or continuing to target a trough before switching to AUC targets, DoseMe will still display both and then place that in context of a guideline dose, either from the industry such as UpToDate or from your own hospitals guideline.

After you’ve had initial doses calculated, you can of course simulate what the most likely area under the curve, and trough, and peak concentrations are of a given selected dose, or you can adjust the target and calculate a dose to reach that particular target for your individual patient. And once that’s done, you can generate a progress note that can contain information about both the selected dose as well as clinical notes of your choice and save that back to the patient’s chart.

So, these progress notes can be saved back automatically to the EHR. And for many of our sites, they find that this alone gives a really significant time-saving. So, finally, I’d like to go through a comparison of five hospitals without DoseMeRx compared with five hospitals with DoseMeRx. And this data is from pre-draft guideline publication. So, the five hospitals not using DoseMeRx were targeting trough targets of 10 to 20 micrograms per mil.

We then compared results without DoseMe as well as the five hospitals with DoseMe from their standard clinical practice. So, I want to make it clear that this is real-world data and not trial data. In addition to looking at target attainment, we also compared the number of drug concentrations or drug levels that were taken as an indicator of both cost, a direct cost, and the time required to monitor these patients.

So, what I’ve got here is a violin plot, and on the y-axis is the trough concentration, and then we’ve got five different hospitals each represented by the different violins. For those of you who may not have seen a violin plot before, it’s essentially showing the distribution of the trough achieved by every dose. And this includes doses prior to the fifth dose, so these patients are not necessarily at steady state. So, in this case, we’re looking at: Where do the trough concentrations fall from all of the dosing at a particular hospital?

And for example, for the leftmost hospital there, you can see that the trough concentrations achieved tend to be lower with many of them being below 10 milligrams per liter. Conversely, the rightmost hospital there is tending to dose super therapeutically and is often achieving trough concentrations above 20 milligrams per liter. And this picture that we see here with significant variability from site to site is quite common. And this is one of the things that we really notice when we analyze data from hospitals who are not using DoseMe.

When we look at the results of a group of five hospitals using DoseMe, it’s quite different. Firstly, we’ve gone from only one in two doses in range to three and four, but we’ve also got much more consistency and less variability where those doses are ending up. All five of these hospitals are much more often ending up with in that therapeutic window and doing so consistently. Now, this is looking at troughs. When we look at AUCs, we see very, very similar outcomes with, again, approximately three and four doses ending up in that AUC target range and reduced variation of both sub and super therapeutic dosing.

So, moving on to the cost part of the equation. When you’re looking at switching to AUC-based dosing, there’s the potential for needing to do significantly more levels if you’re needing to calculate an AUC from multiple levels. So, if we have a look at existing hospitals that are not using DoseMeRx and are targeting a trough we measured that they’re taking one level per every 1.34 days of therapy, which is one level every 34 hours on average.

Were they to switch to AUC dosing using the approach published by Meng et. al this year. That would require additional levels, and they’d need to do at least one drug level every 21 hours of vancomycin therapy. When we looked at the hospitals actually using DoseMeRx in practice to manage their vancomycin dosing, they were only taking one level every 2.14 days of therapy or only one level every 51 hours, suggesting that patients were being placed into range more stably, as well as there was additional confidence from pharmacy that those patients were getting into and staying in the desired target therapeutic range.

And this corresponds to a decrease in lab count of approximately 37% while still increasing target attainment compared to our non-DoseMeRx hospitals for both trough and area under the curve. Finally, there’s one benefit to using electronic clinical decision support software that I’d like to touch on before I wrap up. And that’s that when you have this data available inside electronic clinical decision support, it can be used not just to individualize the dose for a single patient but also to manage risk across your site or multiple sites if you have a multiple-hospital network.

And on the right-hand side, you can see a screenshot of a clinical risk dashboard, where not only are we looking at levels that have already been done, but we’re also using charted orders to predict future trough concentrations, which we can relate to adverse drug event risk, as well as predict patients who will have super therapeutic dosing.

So, with this clinical decision support, you can not only avoid potential problems, but highlight issues in a given unit and therefore improve your resource allocation. So, in conclusion, using DoseMeRx can deliver an increase in target attainment for both area under the curve and trough-based dosing. And whether you’re ready to start with AUC-based dosing now, or wish to start with trough and then transition over, we can configure our clinical decision support software to be switched to AUC24 targets or AUC to MIC targets when you’re ready.

And finally, one of the benefits of Bayesian dosing is not only do you calculate an accurate dose for an individual patient, but that data can also be used for predictive reporting and analytics on outcomes, whether you’re seeking these in the aggregate such as the violin plots presented earlier or for risk management to identify individual patients.

So with that, I’ll wrap up. And thank you very much for listening.

Dr. Orsula Knowlton: Well, thank you. Dr. McLeay, Brunetti, and Turgeon. Your presentations were excellent. We have quite a few questions, which is exciting. So at this time, we’ll begin to address those questions with the few minutes that we have left. While we may have more questions than time to accommodate all of them, we will certainly follow up with each individual to answer those questions.

So, the first question: I believe it’s addressed to Dr. Brunetti, however, you may decide if it should be addressed by someone else. The question is: Do you know the software or do you use the software to target AUC for other pathogens such as enterococcus strep or coagulase-negative staph?

Dr. Luigi Brunetti: Sorry, I just had to unmute myself. That’s a great question. So, we use this software both for empiric dosing and continued dosing. And at this time, we don’t differentiate between the organism that grows out. So in other words, just for ease and for logistics purposes, our target is 400 to 600. Now, we do have a carve-out. If we are managing or treating meningitis or suspected meningitis, our AUC target is actually 500 to 700, but we don’t have any difference depending on the organism that comes out.

Dr. Orsula Knowlton: Great. Thank you. Did anyone want to add to that? Okay, I think this next question is directed for Robert. Robert, you may comment on the previous question as well if you were planning to. Are the 130 sites and 20,000 doses per month just for vancomycin?

Dr. Robert McLeay: No. To be clear, not all of those 20,000 doses are for vancomycin. That said, vancomycin is our largest drug by use on the platform. So, a significant proportion of those are for vancomycin. Other drugs that are available on the platform, aminoglycosides, for example for antimicrobials, as well as even drugs such as [INAUDIBLE 00:45:29]. So, we do have a wide range, but vancomycin is the single, most heavily-used drug.

Secondly, I guess to follow on from the earlier question asked to Dr. Brunetti, as I showed in the presentation, you can either simulate the AUC that you’re most likely to achieve from a given dose or you can actually change what your target is. So, there’s a large amount of flexibility there should you have, for example, a separate protocol for meningitis.

Dr. Orsula Knowlton: Thanks very much for your response. The next question has to do with pregnant women. Has this protocol been used for pregnant patients? And if so, what weight is used? Is it pre-pregnancy adjusted body weight if obese or total current pregnancy body weight? Dr. Brunetti, would you like to take a stab at that one?

Dr. Luigi Brunetti: Sure, so we, for our protocol, we actually exclude pregnant females. So, there’s a couple of carve-outs. Pregnancy is one that we exclude, and we also exclude surgical prophylaxis because we have just a separate protocol for that.

Dr. Orsula Knowlton: Great. Thank you so much for taking the time to answer that. The next question. Could you briefly explain advantages and applications of the vancomycin two-compartment model? If it more closely approximates vancomycin pharmacokinetics, why not use it for everyone? What is the downside?

Dr. Robert McLeay: Yeah, I might take this one. So, Robert here. The vancomycin two-compartment model has the advantage that it can, of course, more closely fit both the distribution and elimination phases of vancomycin. So, there is definitely an advantage in how closely you can fit your data. That said, the more parameters that you have in any given model, the more data you need in order to have a really, really high quality fit.

So, ultimately, it comes down to a choice at your particular site. And for many of our sites who are transitioning from perhaps using a hand-calculated trough-based approach, a one-compartment model is more consistent with their historical way of dosing. So for them, that’s an easier change management project. That’s it. We have both models in there for a reason, and we do find that many sites do appreciate the ability to model that distribution phase of vancomycin more accurately than using a one-compartment model.

Dr. Orsula Knowlton: Thank you. For our next question: Are you aware of any evidence that suggests an outcome benefit associated with targeting higher AUC24 in patients with more severe infections? For example, 550 micrograms per mil per hour.

Dr. Luigi Brunetti: So I think perhaps I can try and take this question here. So, there’s several studies that have been performed looking at: What is the lower threshold of AUC/MIC? If you actually take a peek at the draft vancomycin guideline, they do a pretty good job, or they do a great job, actually, of summarizing all the literature that they use to come up with their target of 400 to 600. But it’s important to note that that 400 to 600 was based on the individuals of the guidelines interpreting all of the evidence that’s out there.

But if you look at individual studies, they may provide kind of a different threshold from when vancomycin appears to be the most efficacious in terms of AUC/MIC. But from what I understand is the majority of evidence points towards a 400 to 600 range as being a target.

Dr. Orsula Knowlton: Great. Thank you so much. Our next question: For the pediatric model of DoseMe, are these otherwise healthy pediatric patients, or can we use the dosing model in pediatric patients who also have renal dysfunction?

Dr. Robert McLeay: So, that’s an excellent question. My understanding of that particular model is that it does include serum creatinine as one of the covariates. In other words, you can enter in the serum creatinine of your patient and have that vary over time, and the model will adjust the clearance over time. So, you don’t need to do additional drug levels to say that there’s a change in your patient’s pharmacokinetics if you have a serum creatinine that shows that their rate of clearance has changed.

Now, this is from the top of my head. I don’t have the models right in front of me now. My understanding is that for the neonate models, however, that isn’t the case. There are some complications with measuring serum creatinine in the immediate post-birth period.

Dr. Orsula Knowlton: Thank you, Robert. Another question: What is the time commitment that the healthcare system’s IT department requires for adding the DoseMeRx to the EMR of various systems?

Dr. Robert McLeay: Yeah, so that is a question that is a little bit hard to answer in general. But we’re talking about, typically, a small number of weeks to organize the process, to go through it. So, DoseMeRx is delivered as what’s called a SMART on FHIR application. The big benefit of that is that it’s already been validated by both Cerner and Epic as being a safe product in how it integrates and works with your EHR. So, that makes it significantly easier than traditional EHR Integrations to deploy.

We typically find that most of the time in deployment isn’t actually spent with the IT department. We probably spend slightly more time working with pharmacy to ensure that we’ve set the default behavior of DoseMe to match your protocols as closely as possible. Realistically, from an IT department point of view, there’s quite a low barrier to entry compared to traditional approaches.

Dr. Orsula Knowlton: So, generally quick and easy compared to other systems, for sure. Our next question is: Can you speak to monitoring frequency after you switch to AUC-based dosing? Perhaps for Dr. Brunetti?

Dr. Luigi Brunetti: Sure. So, monitoring frequency is something that we really struggle with. Even when we switched over, there’s a challenge, because many clinicians often want to see that concentration recorded in the patient’s medical record. And they get a little uneasy sometimes when you try and push the trough concentration a bit too far.

So, I know the guideline essentially recommends you would get one initial, and if within target, you should be good unless it’s a prolonged course. But in clinical practice, I have to say that seldom happens. Even with the AUC-based dosing, I still think we share that struggle that we had even with our old protocol from 2011. So, that happens to be one of our things that we’re trying to shift. We’re trying to get clinicians more comfortable with less frequent vancomycin concentration monitoring.

Just to elaborate on that, Robert had presented some data on how often vancomycin concentrations were drawn in the hospitals that they evaluated. And I have to say, that’s probably spot-on for how often we’re measuring concentrations here at our hospital.

Dr. Orsula Knowlton: Thank you. And perhaps a follow-on question. AUC-based dosing appears not to pay much attention to patient’s weight. How best would you go about educating physicians that are used to using 15 to 20 milligrams per kilogram dosing and to not paying attention to the kilogram-based dosing for vancomycin without having them take a pharmacokinetics class?

Dr. Luigi Brunetti: Yeah, that’s a great question. And you know, I just had this happen yesterday where a clinician ordered a dose of 1.5 Q12, and I change the dose to 750 Q12. And they called me and said, «Are you sure you want to lower the dose? I don’t think it’s enough.» So, what I did to kind of appease the clinician, because we’re using the Bayesian software, I just got a drug concentration fairly early so I didn’t have to wait till steady state. And once the concentration came back and I calculated the AUC and the trough, they were comfortable with it. So, I actually use the tool to convince the clinician that we should be okay.

Dr. Robert McLeay: I might jump in there as well. So, this is certainly something that we’ve seen with some of the hospitals that we work with. So, we very recently just assisted a hospital in conveying this change to some of their clinical staff. And ultimately for them, it came down to pharmacy-led education. We’ve provided some resources around what that looked like in the sense of, you know, what would your maintenance dose at steady state be for different combinations of weight and renal function?

And using that, they have been able to run education both inside and outside of pharmacy. You’re definitely right. It’s a challenge, but it is quite addressable. And as Dr. Brunetti said, being able to actually show, «Well, what if we gave a different dose using a Bayesian dosing tool?» is incredibly powerful.

Dr. Orsula Knowlton: Those are great recommendations and strategies. Thank you. One last question: Do you still round serum creatinine when using DoseMeRx, and what creatinine clearance estimates do you use? Ideal body weight, actual adjusted, or combination of these depending on the patient characteristics?

Dr. Robert McLeay: That’s a fantastic question. The answer is that does depend on the particular model. DoseMe has the position that what we’re providing is a tool based on validated research, so we implement the particular model. That said, these models nearly always do round serum creatinine. Typically, they have a minimum serum creatinine of approximately .075.

Some models have that set to 1 instead, particularly for the obese. And for the adult models, nearly always Cockcroft-Gault is used as the method of calculating eGFR. To talk about weight a little bit for vancomycin, it’s typically total body weight for most of these models. Although, the obese model does have a correction in there as well.

Dr. Orsula Knowlton: Well, thank you so much for the answers to these very interesting and important questions. We’re so pleased to have everyone on this call today. A follow-up communication will be sent that will include information about this presentation and DoseMeRx. So, thank you so much and have a wonderful rest of your day. Thank you all.

Sources

Mark J. Biagi, David A. Butler, Eric Wenzler. AUC-Based Monitoring of Vancomycin: Closing the Therapeutic Window

Here are some more vancomycin resources for you: