This Success Story is also available for download
Opportunity
Vancomycin is widely used in hospitals to treat serious infections, but achieving optimal drug exposure remains challenging. Traditional trough-based monitoring methods have significant limitations, often resulting in suboptimal target attainment and increased risk of acute kidney injury (AKI).1 Three large health systems using conventional trough-based dosing faced these challenges and sought to improve vancomycin dosing precision and patient safety across their networks.
Contemporary studies demonstrate the substantial benefits of model-informed precision dosing (MIPD) for vancomycin:
- Lee et al (2020) demonstrate substantial financial benefits over trough-based dosing, estimating ROI of $2,065 per patient.2
- Alsowaida et al (2022) examine increased efficiencies using MIPD and reveal that integrated Bayesian solutions can save nearly 3 minutes of time per intervention, while secondary endpoints indicate usable laboratory samples also increased by nearly 50% from baseline.3
- Hall et al (2024) highlight several clinical advantages, including a significant reduction in acute kidney injury (AKI) events as well as a decrease in hospital LOS by an average of 2 days per patient.4
Following a successful 2019 implementation at OSF Healthcare — which demonstrated a 50% reduction in severe AKI events, an annualized length of stay (LOS) reduction of 1,428 patient days, and total annualized cost reduction of $3.6M — our team took on the opportunity to replicate and expand upon those results with larger datasets.
Solution
To evaluate impact across these networks, a comprehensive pre-/post-implementation analysis was conducted involving 41,473 treatment courses. Three health systems implemented InsightRX's Nova precision dosing platform for vancomycin, transitioning from traditional trough-based monitoring to AUC-guided dosing using population pharmacokinetic (popPK) models with Bayesian estimation.
The InsightRX Nova platform provides clinicians with individualized dosing recommendations at the bedside by incorporating patient-specific factors into validated popPK models, enabling practitioners to achieve target AUC ranges of 400-600 mg*h/L in alignment with updated IDSA and ASHP guidelines. The Bayesian approach also easily accommodates flexible sampling times (greatly boosting usable samples, as shown in Alsowaida et al), and enabling real-time dose adjustments as the software updates predictions based on observed patient data.
Impact
Across all three participating health systems, implementation of InsightRX for vancomycin dosing led to measurable and meaningful changes in practice. Pharmacists used model-informed precision dosing to more accurately individualize therapy, improving consistency in target attainment while maintaining patient safety.
- System-wide improvement: Each organization demonstrated higher rates of patients achieving target AUC/MIC exposure after InsightRX implementation compared with historical, trough-based dosing.
- Reduced variability in performance: Retrospective data showed more consistent target attainment across hospitals and patient populations following adoption of Bayesian dosing.
- Clinical confidence and optimization: Pharmacists gained greater assurance in dosing decisions through improved predictive accuracy of InsightRX models.
- Alignment with national guidance: The move from trough-based to AUC-guided monitoring positioned all sites in accordance with updated vancomycin dosing recommendations from leading professional societies.
These results highlight the collective impact of adopting model-informed precision dosing at scale — demonstrating that even diverse, multi-hospital systems can achieve consistent improvements in therapeutic precision and patient outcomes.
Results
A total of 41,473 treatment courses were included in the evaluation. Resultant data from the pre-/post-implementation study suggest that Bayesian MIPD provided several benefits to clinical outcomes in patients receiving vancomycin.
Pharmacokinetic (PK) target attainment increased by 44.0% relative to baseline
Updated dosing guidelines (released by IDSA, ASHP and several other professional organizations, released in 2020) recommend a standard target of 400 - 600 mg*h/L in patients receiving vancomycin with confirmed or suspected MRSA infections.6
While specific PK targets per-patient are selected by clinicians at the bedside, aggregate data demonstrated that using Bayesian MIPD significantly increased target attainment by 44.0% — from 41.3% pre- to 59.5% post-implementation (see Figure 1). Results from each individual organization also showed significant increases in target attainment from baseline.
Figure 1 shows target attainment rates pre-InsightRX implementation (navy bars) and post-InsightRX implementation (teal bars); data is shown for overall dataset and stratified by organization.
Rates of severe AKI decreased by a relative 48.4% after InsightRX implementation
AUC dosing to a standard target of 400 - 600 mg*h/L has been associated with decreased rates of AKI compared with trough-based dosing. Data from this analysis underscores benefits of switching to AUC-based dosing using MIPD, with MIPD showing a significantly lower rate of severe AKI events after 48 hours of therapy (3.1% at baseline vs. 1.6% after implementation, Figure 2).
Figure 2 shows severe AKI rates after 48 hours of therapy pre-InsightRX implementation (navy bars) vs. post-InsightRX implementation (teal bars).
Post-InsightRX figures were significantly lower at two of three individual sites, with the third site also trending lower post-deployment. Additionally, Bayesian MIPD showed a relative reduction in overall AKI events after 48 hours of 38.5% (9.1% at baseline vs. 5.6% after implementation, Figure 3).
Figure 3shows overall AKI rates after 48 hours of therapy pre-InsightRX implementation (navy bars) vs. post-InsightRX implementation (teal bars).
Bayesian MIPD helped clinicians achieve target ranges faster by an average of 24 hours
Utilization of population-pharmacokinetic (popPK) models at the bedside demonstrated the ability for clinicians to achieve desired targets faster than trough-based, manual methods. Results indicate that overall time to target attainment decreased significantly by more than 24 hours from baseline. Data also suggest that time to target range was significantly faster at all individual sites, with one site decreasing average time to target by more than 30 hours (see Figure 4).
Figure 4 shows time to target attainment pre-InsightRX implementation (navy bars) and post-InsightRX implementation (teal bars).
Additionally, results indicate that use of Bayesian MIPD resulted in a higher percentage overall of patients on target at any given time vs. pre-InsightRX implementation. This trend followed for each individual site as well (see Figure 5).
Figure 5 shows percent of patient treatment courses on target over time pre-InsightRX implementation (navy lines) vs. post-InsightRX implementation (teal lines).
Conclusions
Bayesian MIPD shows significant clinical benefits vs. traditional, trough-based dosing methodologies. This study serves as a supplement and replication of our first case study (2019) and demonstrates similar trends in benefits and endpoints:
- Bayesian MIPD significantly increased target attainment relative to manual, trough-based methods for calculating drug exposure
- Bayesian MIPD helped significantly reduce AKI events compared to baseline, trough-based dosing for patients receiving vancomycin
- Bayesian MIPD significantly shortened time to target attainment and led to a greater percentage of patients in target range earlier in their treatment courses