Applying Bayesian Method to Vancomycin Dosing
The concept of Bayesian dosing uses a statistical method derived from the Bayes’ Theorem, which first calculates the initial probability of an event based on prior knowledge and then incorporates new information to recalculate the updated probability of an event. The model is highly adaptive to physiologic change and thus individualized via the incorporation of covariates in the population models (i.e CrCl in critically-ill patients). When applying this principle to vancomycin dosing, the patient's pharmacokinetic parameters (Vd and CL) are estimated before the giving the first dose, using either literature-based population models or pre-existing Bayesian PK data obtained from chart review, also known as Bayesian’s Priori. This allows pharmacists to estimate current pharmacokinetic parameters and develop the initial dosing regimen, commonly seen in the Emergency Department and Intensive Care Unit. After the administration of the first dose, the patient is monitored and the vancomycin serum concentration is measured. The serum level is used to establish the individualized modeling of the patient’s pharmacokinetic parameters, also referred to as Bayesian posterior parameters.
Why is Bayesian Modeling the Method of Choice to Estimate AUC?
New 2020 IDSA vancomycin dosing guidelines published in March recommend monitoring the pharmacodynamic index AUC/MIC to predict the safety and efficacy of vancomycin therapy in patients with serious MRSA infection. This recommendation is based on compelling evidence showing that a vancomycin AUC/MIC ratio of 400 to 600 mg*h/L optimizes clinical efficacy and reduces vancomycin-related nephrotoxicity [1,2]. AUC calculation can be accomplished using either a Bayesian dosing software or a formula-based approach, such as the trapezoidal model. Manual AUC calculation is cumbersome and time-consuming, as it requires more computing efforts and pharmacokinetic sampling (at least two serum levels). The Bayesian approach is the preferred method to estimate AUC, as it provides several advantages compared to traditional PK calculation:
- Dynamic models that employ continuous machine learning
- Requires limited pharmacokinetic sampling (as little as one serum level) for precise Bayesian modeling
- Cost-saving, highly efficient and less biased dosing strategy
.png)
The image above depicts the comparison between typical measurements of predicted vancomycin concentrations using Bayesian method and formula-based methods. From this, we see that Bayesian method results in more accurate predictions.
Why Should You Consider PrecisePK?
The transition from trough-based dosing to area-under-the-curve based dosing protocols can be challenging. Clinical decision support software can facilitate this change by performing the complex Bayesian calculations that aide in the optimal dosing recommendation for each patient. PrecisePK is a user-friendly EHR integrated Therapeutic Drug Monitoring (TDM) Precision Dosing platform driven by Bayesian Analytics. It has been used in numerous hospitals worldwide for over 30 years and was validated by a third-party to give the most accurate and least biased vancomycin AUC results. [3] We use research-based population-specific data and advanced Bayesian analytics to adjust the population values of pharmacokinetic parameters (F, Vd, CL, CF, etc.) and serum level data. Upon entering the patient's demographic information, the program will automatically select the best population model (i.e critically-ill, obese, pediatric, etc) derived from current peer-reviewed literature that is most representative for the patient. Advanced users can also choose to modify the model selection or PK parameters (i.e Vd, CL) that best describe the patient based on their personal experience and clinical judgement. Using our program, users can quickly calculate AUC between two pre-selected time points, analyze dose efficacy, and simulate possible dosage regimen. PrecisePK projects serum level charts that help users predict what the next dose should be using as little as a single serum level.
What Do Pharmacists Like About PrecisePK?
Many of our clinical pharmacist users have come to love the software’s ease of use and the reliable pharmacokinetic monitoring it provides. PrecisePK has become an indispensable component in their practice including daily clinical activities, making dosing recommendations, and research activities related to AUC vancomycin dosing. Jennifer Le, an ID pharmacist specialist, co-author of the new 2020 vancomycin consensus guideline, and UCSD professor of Clinical Pharmacy said “The new version is amazing visually and user-friendly too, requiring minimal training. The unique features of this platform that I find especially helpful for patient care are flexibility in selecting neonatal, pediatric, and adult models to ensure you select the best Bayesian prior for your patient population.” Lauren Dea Pharm. D., used PrecisePK extensively as a PGY2 Infectious Disease resident and commented “The program has helped my research become vastly more efficient due to the ease of inputting patient information and determining dosing recommendations. I would not have been able to retrospectively collect 4 years of vancomycin orders within a condensed residency year without the streamlined efficiency of entering info into their program. They also developed a unique Bayesian model for our unique patient population with a specific set of Bayesian priors.”
Interested in trying out PrecisePK's Bayesian precision dosing software? Book a demo with us today to explore our wide range of features and to see how PrecisePK can help streamline your clinical workflow.
