SCCM Resource Library-How to Get the Best Curves: Dosing Vancomycin Using Area Under the Curve Monitoring

Video Transcription

Hello, I am Xiuyan Amy Yeung, pharmacist at the Medical Intensive Care Unit at University of Maryland Medical Center in Baltimore, Maryland.  Today I'm going to present the topic, how to get the best curve dosing vancomycin using area under the curve monitoring.  After the presentation today, hopefully you'll be able to compare and contrast the advantages and disadvantages of utilizing first order kinetics versus Bayesian kinetics approaches  for designing vancomycin regimen to achieve the desired AUC targets. In a patient case, describe the approach to optimize vancomycin regimen including designing  the initial dosing scheme and dosage revision based on patient-specific parameters collected. Finally, discuss operation considerations when transitioning from a trial-based dosing  approach to AUC-guided dosing approach. Why is AUC-based approach being recommended over trial-based approach for dosing vancomycin?  First AUC is a better predictor for efficacy for vancomycin in the treatment of MRSA infections.  A meta-analysis of 14 observational cohort studies, including 1,677 patients, which evaluated outcomes in patients with stop aureus bacteremia being treated with vancomycin.  It was observed that AUC of greater than 400 is a better predictor for better clinical outcomes compared to a target trough of greater than 15 compared to less than 15.  In addition, AUC-guided dosing has been associated with less neurotoxicity. Prior observational studies demonstrated trough levels of 15 to 20 is associated with three  times the risk of acute kidney injury compared to those patients with trough levels of less than 15.  Meta-analysis of two studies, including a total of 1,443 patients, demonstrated upon  switching from trough-based strategy to AUC-based strategy, it has resulted in less neurotoxicity.  This is likely due to the fact that AUC-guided dosing approach oftentimes lead to less vancomycin exposure as well as lower trough levels.  It is demonstrated here that there is approximately about 32% decrease in the risk of neurotoxicity upon switching to an AUC-guided approach.  In 2020, ASHP, IDSA, PIDS, and SIDP updated the vancomycin monitoring guidelines for treatment of MRSA infection.  In the guidelines, they recommended to provide a vancomycin loading dose of 20 to 35 milligrams per kilogram actual body weight with a dose cap of 3 grams in patients who are obese.  For treatment of MRSA infection, the target should be AUC to MIC ratio between 400 to 600.  This target is chosen based on clinical studies or observational data of better outcomes associated with AUC to MIC ratio of greater than 400.  In patients who have AUC of 600 or greater, higher risk of neurotoxicity is observed. To achieve these AUC targets, two approaches are recommended.  The first one is the Bayesian-derived AUC monitoring approach, which is preferred based on the guidelines. The second one is utilizing first-order humble kinetics using two levels.  First, I would like to go over the first-order kinetic approach. With this approach, initial dosing regimen is calculated based on population kinetic  estimate with incorporation of patient's renal function and patient's weight. Then, first dose levels may be obtained to provide patient-specific parameters to calculate  dose to achieve AUC target. This is followed by obtainment of steady-state levels to provide the most accurate pharmacokinetic  data to optimize the regimen to achieve the AUC target.  While these calculations can be done by hand, these equations can be embedded into Excel spreadsheets such as the example on the left or in calculators in the electronic medical  record programs such as the example on the right to aid the ease of calculation of vancomycin dosing regimen to achieve the AUC targets.  Advantages of utilizing first-order kinetic approach include low upfront cost. Excel spreadsheet with the equations embedded can be set up with minimal to no cost.  And with support from the IT department, a calculator might be able to be built within some of the EHR system with minimal cost.  Most pharmacists are also familiar with the equations for determining AUC. On the flip side, pharmacist training are needed because this is a shift from a trough-based  dosing approach to AUC dosing approach. Pharmacists might be familiar with nomograms to adjust vancomycin dose to achieve certain  target trough levels, and now they need to be trained to know when to draw levels including first dose level as well as steady state levels and how to utilize these levels to  calculate a new dose of vancomycin for targeted AUC. Two levels are needed for calculation, and these levels need to be timed appropriately.  They need to be drawn at steady state, and two levels need to be at least one half-life apart, usually at least four hours apart.  First-order kinetic equations also cannot adapt to physiological changes as compared to Bayesian modeling approaches.  Finally, initial dosing is based on population assumptions versus Bayesian approach to account for several other patient-specific parameters to be inputted into the program.  Next, I would go over the Bayesian guided dosing approach. First empiric dose will be calculated based on a large bank of pharmacokinetic data with  the drug and in this vancomycin in a specific patient population. There are several population models out there utilized by Bayesian software programs, including  ones that have primarily critically ill patient versus some of the other ones are primarily pediatric patients or in a general medicine patient population. Then levels are obtained.  The software will then take into account for the levels, the dosing information, as well as patient characteristics, and then incorporate that into, again, some patient  population pharmacokinetics, and simulate alternating dosing regimen to best estimate the dosing regimen to achieve the AUC target desired.  And this is termed Bayesian conditional posterior. One may ask, how do these Bayesian guided dosing programs perform in the critically ill patient population?  Turner and the group utilized drug level data from critically ill patients to compare the accuracy of AUC estimates by various Bayesian dose optimizing software to AUC calculated  by the linear log trapezoidal rule utilizing the full data set. As you see here, when they utilize only the trough to be put into these software, the  area under the curve estimated by these software programs often underestimate the AUC calculated by the full data set.  The variability are also slightly high in some of these dosing software.  However, when you put in more data into the estimate or into the Bayesian software program, the accuracy of the estimated AUC is improved.  As you see here, if you utilize a one-hour post-infusion level paired with the trough, the estimated AUC by the program is much improved.  The accuracy definitely has been improved significantly when the full data set are being put into the Bayesian software programs.  Bias is also observed to be improved when more levels are being put into these programs.  Another group in Australia, Narayan and colleagues, also utilized PK data in the ICU patients to be simulated in different Bayesian population estimate.  In the study, though they found low bias of these Bayesian population programs, but they also found a low precision, meaning there is a significant difference between the Bayesian  estimate to the true observed trough level in the patient population, especially when the level is at a higher trough level.  However, they also observed when more levels are being put into these Bayesian population estimates program, then the precision is improved.  More recently, University of Kentucky also published their findings looking at the agreement  between Bayesian guided dosing programs to the first order kinetic equations AUC estimates.  As you see here, when they have put in two levels into the Bayesian dosing program, and in this study they utilized the INSIGHT RX program, they observed a pretty high level  of correlation or agreement between the Bayesian program to the linear equation approach with  low variability, versus when they only utilized one level in the Bayesian program, the correlation is less with higher degree of variability.  All these studies have demonstrated that, especially in the critically ill patient population,  rather than using just one level, two levels might be preferred when utilizing the Bayesian guided dosing approach.  There are several advantages in utilizing Bayesian guided dosing approach to optimize AUC. First, model can be modified based on the patient population that this is being used  in to provide a better estimate of vancomycin dosing and AUC estimate. Pre-study state level can be utilized in these models, and therefore we can modify the vancomycin  regimen earlier to achieve the desired AUC targets. One level instead of two levels can be used in these models to provide an AUC estimate.  But keep in mind, as mentioned in previous slides, one level estimates are often less precise in the critically ill patient population, and therefore two levels might be preferred  to provide the AUC estimate, as well as dose optimization. Finally, these models or these dosing programs incorporate real-time patient data into their  simulation, and therefore have improved ability to adapt as well as improve the estimate,  and therefore might be able to provide a more precise dosing strategy to achieve the AUC targets.  On the negative side, definitely there is a significant cost associated with the programs that are commercially available.  One should also consider workflow integration when utilizing these programs because a lot of patient data needs to be inputted into these programs to allow or to provide the  dosing estimates or dosing optimization. There is also lack of familiarity with Bayesian guided dosing approach, and therefore more  training might be needed to ensure they are utilizing the software appropriately. Finally, there is still a lack of information in some of the special population, including  critically ill as well as unstable renal function. What is the uptake of AUC to MIC monitoring for vancomycin currently?  Back in 2019, a survey is done within the FECEN group, and majority of the institution who  responded to the survey have pharmacist-driven dosing protocols or pharmacists are the ones responsible for vancomycin monitoring.  However, only 23.1% of these institutions utilize the AUC to MIC monitoring approach. Among those who used trial-based dosing, majority did not plan to or unsure about transitioning  from trial-based dosing to AUC-MIC ratio-based monitoring. A more recent survey is conducted among ACC PPRN listserv members and ASHP members.  Majority of the respondents, meaning 70.3%, have not implemented AUC to MIC dosing. However, majority of the respondents do plan to do so.  Now I would utilize a patient case to demonstrate how one can optimize a vancomycin regimen utilizing the AUC to MIC approach.  This is a 52-year-old patient admitted to your ICU for sepsis from necrotizing skin and soft tissue infection.  And pre-antibiotics, including pericillin, tasopectin, and vancomycin were ordered to be started. So what dose of vancomycin would you recommend for this patient?  Here are his specific parameters. Based on the guidelines, a loading dose should be utilized. So I have recommended a loading dose of 2.5 grams based on the 25 milligrams per kilo  actual body weight load. Based on first-order kinetic equations, I have calculated initial dose based on patient population estimates.  A dose of 1.75 grams IVQ 12 hours will result in an estimated AUC of 525 with a trough of 13.4 micrograms per milliliters.  And one would ask, should we utilize first-dose promovable kinetics? There are two studies that looked into this question.  And they both demonstrated this approach will result in a greater target attainment at steady state in critically ill patient.  In the first study where they look at AUC target, 58.6% of patients utilizing the first-dose  kinetic approach have achieved target attainment versus only 32.4% achieve target attainment when they only utilize population estimates.  Similar findings are also observed in a study look at trough target attainment. 84% of patients have achieved a target when they use first-dose kinetics versus only 29.4% among  patients who just use population estimates. Some considerations include whether to utilize this approach in patients with very dynamic renal function or dynamic volume status.  In these previous studies, they excluded patients with acute kidney injury because it's very hard to  estimate or very hard to calculate the patient-specific kinetic parameters because the renal function is fluctuating so much.  Dynamic volume status, which is occurring in a lot of our subject shop patients, might also result in overestimation of volume distribution.  And therefore, it might lead to more aggressive dosing or overaggressive dosing leading to supertherapeutic levels.  Also, if it is only for empiric therapy or short treatment duration, meaning less than 48 hours, obtaining two levels might be not necessary.  For this patient, since we know he might be on vancomycin for a good period of time due to his  infection, which is a skin and soft tissue infection, we decided to go ahead to obtain first-dose levels to provide a better estimate of vancomycin dosing.  So vancomycin dose of 2.5 grams was given at 2 p.m., and then we have obtained levels about one hour  after the end of infusion, as well as a trough level obtained at 1 a.m., and these are the levels.  Based on first-order equations, we calculated the patient-specific elimination constant as well as  volume distribution, and therefore, we are able to optimize the regimen to 1.75 grams IVQ8 hours, which result in an estimated AUC of 496.5 and a trough of 14.6.  After you have implemented the new regimen, oftentimes, we need to confirm at steady state whether this regimen is still appropriate for the patient.  So the steady state level are often obtained after the fourth dose on the regimen, and here are the levels that are obtained.  Then we can, again, plug in these numbers into the steady state level equations, and based on these,  we calculated the 1.75 grams IVQ8 hours will result in an AUC of 523 with trough of 11.5.  One might also argue you can decrease the dose to 1.5 grams IVQ8, which result in AUC of 449 and a slightly lower trough of 9.6 milligrams per milliliters.  One may ask whether it's worth the cost and the effort to transition from a trough-based dosing to AUC-based dosing method.  Bakeri and colleagues compared the cost of AUC to MIC-based monitoring, which they utilized a two-level first-order kinetic approach to trough-based dosing monitoring.  They did not find any difference between the two methods in terms of cost. Interestingly, there is also no difference in terms of the number of levels being monitored between the  two periods. We and colleagues also conducted a cost-benefit analysis comparing the AUC approach to trough-based approach.  For the two-sample AUC approach, for each patient, about $846 can be saved compared to the trough-based dosing, considering the risk of AKI with the trough-based dosing approach.  For a single-sample basin monitoring approach, for each patient, $2,065 can be saved.  Assuming 1,000 patients are being treated per year, the annual cost savings for the two-sample approach is about $846,000.  And for the single-sample basin approach, the saving is about $2 million. They also estimated that you will treat at least 41 patients for 48 hours with vancomycin to break even,  assuming the cost of a software for the basin modeling is about $10,000. Finally, just want to share some tips for implementing AUC guided dosing approach.  First, you need to define which patient population to include, keeping in mind some of the limitations for AUC guided dosing.  Currently, the AUC targets are mainly validated for MRSA infections. And the targets for osteomyelitis, meningitis, as well as non-staph aureus infections are not well-defined or  have very limited data. Therefore, in those patient population, you need to consider whether to include those into your AUC guided dosing.  In our institution, however, to minimize confusion, we do include those patients into the AUC guided dosing approach.  Also, you need to set some exclusion criteria for AUC guided dosing. For example, patients who are on vancomycin for short duration of time, such as for simple cellulitis, or in  patients who are on vancomycin for surgical prophylaxis, then you might want to exclude them from AUC guided dosing.  Also, for patients with renal impairment or fluctuating renal function, the AUC guided dosing approach might not  be appropriate because it's very hard to estimate AUC or optimize the dosing in those patient population.  One exception to the rule might be for patients who are on stable continuous renal replacement therapy settings.  Next, you need to consider how to calculate the dosing regimen for the AUC targets, whether to utilize first-order kinetics versus Bayesian modeling.  To help you decide that, you need to compare the different approach to find the one that best suits the need for your institution.  With consideration of cost for the program of cost for implementation for either approach, as well as the need  for drawing levels, keeping in mind two levels are needed for first-order for all patients versus one level can be utilized for the Bayesian approach in some of the patients.  Which calculator or program to use based on the program or like how comfortable with each of the programs. Workflow considerations are also important.  For example, whether to build in some extra tools for the pharmacist so that it's easier for them to input data  in whatever program that you will be utilizing for calculating the dosing regimen.  For example, for our institution, we do have a flow sheet that contain most of the information needed for the pharmacist.  Finally, to see if any of these programs can be integrated into your EHR. Some of the current available Bayesian programs do have the capability to be integrated in some of the very  commonly utilized EHR systems, such as Epic or Cerner.  Finally, need to educate your pharmacist as well as nursing on the appropriate timing of these levels.  This is especially true for first-order kinetic approach when the levels are needed to be drawn at steady state  and adequate time needed between the two levels to ensure accurate calculation.  Also, these levels need to be labeled appropriately so that we know the exact time of the levels being drawn.  Education is very important for successful implementation of AUC guided dosing.  Pharmacists need to be educated on how to perform calculations or utilizing the Bayesian dose optimization  software. Also, when is the appropriate time to draw levels, including first dose level, which paper population  should we utilize first dose level, and when to draw steady state levels.  Prescribers also need to be educated on the correct interpretation of levels. Oftentimes, they might not be used  to seeing peak levels being drawn and therefore might mistakenly interpret those levels to be super therapeutic levels, leading to incorrect adjustments of vancomycin doses.  Finally, nurses also need to be educated on when to draw the levels, as well as importance of appropriate timing of these levels.  Finally, evaluate your implementation of AUC guided dosing, including in the initial phase, evaluate any errors that  have occurred, including missed levels or errors in drawing levels or errors in providing a dose recommendation.  After a few months, look at the target attainment with your AUC guided dosing method, as well as renal outcomes,  meaning if there's a decrease in nephrotoxicity upon transitioning to AUC guided dosing.  There are several valuable resources out there to help you for this process. The Society of Infectious Disease  Pharmacists vancomycin AUC toolkit provides a lot of useful information, including calculators, as well as slide sets  for education, as well as template for implementation of AUC guided dosing.  The MedID website also have a lot of valuable resources, including comparison of different available software for  base in dose optimization program, as well as first dose kinetic equations.  Finally, just ask the question, do you really need vancomycin? As we know, not all patients that started on  vancomycin need vancomycin, or vancomycin might not be the best treatment option for some of the patients with MRSA infection, so keep that in mind.  Thank you for your time.

Video Summary

In this video, pharmacist Xiuyan Amy Yeung discusses the topic of optimizing vancomycin dosing using area under the curve (AUC) monitoring. AUC is a better predictor of efficacy and has been associated with less neurotoxicity compared to trough-based dosing. The updated guidelines recommend a loading dose of 20-35 mg/kg with a dose cap of 3 grams for obese patients, and a target AUC-to-MIC ratio between 400 to 600 for MRSA infections. Two approaches are recommended for achieving these targets: first-order kinetic and Bayesian modeling. The first-order kinetic approach involves calculating initial dosing based on population kinetic estimates and adjusting based on patient-specific parameters. Bayesian modeling uses an empiric dose calculated from a large bank of pharmacokinetic data and incorporates patient characteristics to estimate the dosing regimen. Studies have shown that using two levels in Bayesian modeling improves accuracy. Implementing AUC-guided dosing requires defining patient populations, selecting the appropriate dosing method, considering costs and workflow integration, and training staff. Education is crucial for pharmacists, prescribers, and nurses to understand correct interpretation of levels and timing of measurements. Evaluation of the implementation should include target attainment and renal outcomes. Resources like the Society of Infectious Disease Pharmacists Vancomycin AUC Toolkit and MedID website provide valuable information for this process. It is also important to consider the appropriateness of vancomycin treatment for each patient.

Asset Subtitle

Pharmacology, 2022

Asset Caption

Guidelines for vancomycin area under the curve (AUC)/minimum inhibitory concentration (MIC) monitoring were published by the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the Society of Infectious Diseases Pharmacists in March 2020. Many institutions are developing new protocols for monitoring vancomycin, including the updated recommendation from the guidelines to use AUC/MIC monitoring in place of traditional trough monitoring. In addition, recent literature has promoted the use of continuous infusion beta-lactam dosing. In this flipped classroom session, attendees will bring local protocols and dosing schemes to launch a comparison with the newly published guidance and compare strategies by using audience polling. This session will also review the advantages and disadvantages of 2 level AUC pharmacokinetics and more advanced Bayesian AUC pharmacokinetics in an interactive session using simulated patient data, including initial dosing schemes and dosage revisions. A similar approach will be taken to determine time > MIC dosing for beta-lactam antibiotics using therapeutic drug monitoring techniques and extended/continuous infusion dosing approaches.

Meta Tag

Content Type Presentation

Knowledge Area Pharmacology

Knowledge Level Intermediate

Knowledge Level Advanced

Membership Level Select

Tag Antibiotics

Tag Pharmacokinetics Pharmacodynamics

Year 2022

Keywords

vancomycin dosing

area under the curve

AUC monitoring

neurotoxicity

loading dose

MRSA infections

Bayesian modeling

	Society of Critical Care Medicine 500 Midway Drive Mount Prospect, IL 60056 USA Phone: +1 847 827-6888 Fax: +1 847 439-7226 Email: support@sccm.org	Contact Us About SCCM Newsroom Advertising & Sponsorship DONATE	MySCCM LearnICU Patients & Families Surviving Sepsis Campaign Critical Care Societies Collaborative	GET OUR NEWSLETTER
© Society of Critical Care Medicine. All rights reserved. \| Privacy Statement \| Terms & Conditions The Society of Critical Care Medicine, SCCM, and Critical Care Congress are registered trademarks of the Society of Critical Care Medicine.