Precision at Scale: A Multi-Health System Outcome Study on Vancomycin Dosing with InsightRX

44.0%
Increase in PK target attainment from baseline
48.4%
Relative reduction in severe AKI events
24hr
Faster time to target range

About the study design

A retrospective study was done on three large health systems in the United States to evaluate vancomycin dosing performance before and after implementing InsightRX’s Bayesian model-informed precision dosing platform. Each site transitioned from traditional trough-based monitoring to a data-driven, AUC-guided dosing approach. Collectively, the participating institutions represented a broad and diverse care network, encompassing 36 hospitals and more than 7,000 staffed beds.

Organization Type
Health System
Highlighted Solution
Bayesian MIPD
Therapeutic Area
Infectious Diseases

This Success Story is also available for download

 

Opportunity

Contemporary studies continue to show the benefit of using model informed precision dosing (MIPD) applications at the bedside:

  • Lee et al (2020) demonstrate substantial financial benefits over trough-based dosing, estimating ROI of $2,065 per patient.1
  • Alsowaida et al (2022) examine increased efficiencies using MIPD and reveal that integrated Bayesian solutions can save nearly 3 minutes of time per intervention. Secondary endpoints indicate usable laboratory samples also increased by nearly 50% from baseline.2
  • 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.3

 

Solution

In 2019 — in conjunction with OSF Healthcare — InsightRX set out to evaluate impact of adoption of our precision dosing solution across the network. Data from the initial study showed several benefits, including:

  • Reduction in severe AKI events of 50% from baseline
  • Annualized length of stay (LOS) reduction of 1,428 patient days
  • Total annualized cost reduction of $3.6M

This evaluation serves to both replicate and supplement our previous case study, providing additional context with a larger data set as to the benefits of a properly-implemented Bayesian MIPD solution.

 

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.4

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.

multi-system-figure-1-p-auc-target-attainment

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).

multi-system-figure-2-p_severe_aki

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).

multi-system-figure-3_p_overall_aki

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

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).

multi-system-figure-4-p_tta

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).

multi-system-figure-5-p_tta_cdf

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