Secondary Logo

Journal Logo

Original Article

Therapeutic Drug Monitoring Coupled With Bayesian Forecasting Could Prevent Vancomycin-Associated Nephrotoxicity in Renal Insufficiency Patients: A Prospective Study and Pharmacoeconomic Analysis

Zhang, Ying MPharm; Wang, Taotao PharmD; Zhang, Di PharmD; You, Haisheng PharmD; Dong, Yuzhu MPharm; Liu, Yang MPharm; Du, Qian MPharm; Sun, Dan MPharm; Zhang, Tao PharmD; Dong, Yalin PharmD/PhD

Author Information
doi: 10.1097/FTD.0000000000000750



Vancomycin, a glycopeptide antibiotic, was developed against gram-positive bacteria and has been long considered as the first-line drug for methicillin-resistant Staphylococcus aureus.1 Vancomycin-associated nephrotoxicity, the incidence rate of which is 5%–35%, is correlated with high serum vancomycin concentrations and is a major known side effect of treatment.2,3 Vancomycin-associated nephrotoxicity could increase mortality, hospital stay, and hospitalization expenses.4–6 Because vancomycin has a narrow therapeutic index,7 therapeutic drug monitoring (TDM) is well-accepted for guiding vancomycin therapy to improve the clinical outcome and avoid vancomycin-associated nephrotoxicity,8 especially when other risk factors exist (ie, renal dysfunction). Because of recent cost-control pressures on hospitals, demonstrating that TDM could provide cost-effective medical care is crucial.9 The American Society of Health-System Pharmacists5 and the Japanese Society of Chemotherapy4 recommended vancomycin trough level >10 mg/L for adult patients. The ratio of the 24-hour AUC to the minimum inhibitory concentration (AUC24/MIC) is the pharmacokinetic/pharmacodynamic (PK/PD) parameter closely correlated with effectiveness, which is recommended to be ≥400.5 Because of the practical limitations of monitoring the AUC24 in individual patients, the trough concentration is more routinely applied in clinical practice for drug monitoring. There are some clinical trial data to support the use of either the 10–20 mg/L trough concentration range or AUC24/MIC ≥400 to improve clinical outcome.10,11

In clinical practice, a larger proportion of patients with mild-to-moderate renal insufficiency exists than patients with severe renal insufficiency in China (35.6% versus 3.3%).12 Vancomycin is believed to be primarily excreted by glomerular filtration. For patients with poor initial renal function, the prolonged half-life of vancomycin and reduced creatinine clearance may aggravate nephrotoxicity due to drug accumulation. Therefore, individualized treatment should be determined according to the renal function in this population and performed using TDM.

The application of population pharmacokinetic (PPK) models coupled with Bayesian estimation is the most accurate and fastest way to calculate dose and interval in most therapeutic fields.13,14 In this article, SmartDose (, a vancomycin-based individualized drug decision-making system based on the Chinese population, was selected to promote the rational use of vancomycin in clinical practice.15

The aim of the study was to assess the efficacy, safety, and cost effectiveness of TDM coupled with dose correction based on Bayesian forecasting, at improving the clinical outcome of patients with renal insufficiency.


Study Design and Population

We conducted a prospective, open study at the first affiliated hospital of Xi'an Jiaotong University. The inclusion criteria were (1) diagnosis of gram-positive bacterial infections; (2) age ≥18 years; (3) vancomycin therapy ≥3 days; and (4) patients with mild to moderate renal insufficiency, assuming that the estimated glomerular filtration rate (eGFR) is 60–89 mL/min/1.73 m2 for mild renal insufficiency and eGFR is 30–59 mL/min/1.73 m2 for moderate renal insufficiency using the CKD-EPI creatinine equation.16 Patients with eGFR <30 mL/min/1.73 m2 were not included in the study due to the difficulty encountered when attempting to identify vancomycin-associated nephrotoxicity. Patients were excluded if pregnant and lactating, allergic to the administered medication, were not expected to survive the first 48 hours, or were undergoing kidney dialysis.

Patients who signed informed consent and met the eligibility criteria were randomly assigned to one of the 2 treatment arms. An allocation ratio of 1:1 was applied using a MATLAB-generated list of random numbers. Because of the practical nature of this trial, blinding could not be achieved as the clinicians needed to apply the dose adjustment after the incidence of vancomycin-associated nephrotoxicity. The study was approved by the Ethics Committee of the First Affiliated Hospital of Xi'an Jiaotong University (registration number XJTU1AF2018LSK-169).

Patients were randomly assigned to the non-TDM group (administered initial regimen of vancomycin and dosing adjustment based on renal function and clinical response by physicians), or the TDM group (administered initial regimen and dosing adjustment according to TDM practice). The initial dosing regimen of vancomycin was 1 g per 12 hours for patients who had creatinine clearance (CLcr) rates between 50 and 90 mL/min; and 0.5 g per 12–24 hours for patients who had CLcr rates between 30 and 50 mL/min.17 The non-TDM group consisted of 84 patients who were followed by a pharmacist from the start to completion of vancomycin therapy. Management of each patient's therapy was at the discretion of the physician. The pharmacist made no intervention, except to complete the data collection form. The physicians and patients were unaware of the pharmacist's activities. If the patients developed vancomycin-associated nephrotoxicity during the study period, physicians would reduce the dosage of vancomycin or discontinue its administration. The antimicrobial trough concentrations were measured using blood remaining from routine examination items (such as routine blood tests). Only steady-state trough concentrations were stored, and the data were exclusively used for comparison with data in the TDM group. The TDM group consisted of 84 patients who received TDM and had their vancomycin dosages adjusted according to a pharmacokinetic analysis. Dosage adjustments were performed using SmartDose pharmacokinetic software. The therapies of the 84 patients were managed with a one-compartment model (a 2-compartment model might not be suitable as nearly 80% of the available measurement concentrations were trough concentration). A TDM pharmacist was responsible for daily monitoring, adjusting vancomycin dosages, communicating recommendations to the physician, and documenting conditions in the patient's chart. Serum samples were collected 0.5 ± 0.5 hours before the fifth dose of vancomycin to determine the steady-state trough concentrations. Vancomycin dosages were adjusted to maintain trough concentrations of 10–20 mg/L. If patients met the predefined concentration target, physicians continued treatment with the existing dosage regimen. Dose adjustment was calculated according to the embedded subprogram in SmartDose (PPK model coupled with Bayesian forecasting). The patient's demographic data (such as age, weight, serum creatinine, and blood concentrations) were entered to calculate the adjusted dosing regimen. The optimized daily doses were rounded to the nearest 50 mg. Patients were followed up until hospital discharge. This intervention scheme is shown in Figure 1.

Flow chart of TDM and non-TDM practice.

Study Definitions and Outcomes

For each patient, the presence of vancomycin-associated nephrotoxicity was determined using serum creatinine (Scr) data from day 0 of hospitalization to discharge. Both Risk-Injury-Failure-Loss End-stage renal disease (RIFLE) criteria18 (see Table S1, Supplemental Digital Content 1, and an increase in Scr of 0.5 mg/dL or 50% from baseline for 2 consecutive measurements19 were used to define nephrotoxicity. The RIFLE criteria classify nephrotoxicity into 5 groups: (1) risk of renal dysfunction, 1.5× increase in Scr or 25% decrease in GFR; (2) renal injury, 2× increase in Scr or 50% decrease in GFR; (3) failure, 3× increase in Scr or acute increase of ≥0.5 mg/dL or ≥75% in GFR; (4) complete loss of renal function persisting for >4 weeks; and (5) end-stage renal disease. The eradication of bacteria was defined as the incapacity to culture the original gram-positive bacteria at the primary infection site and the absence of the need for vancomycin within 7 days after the end of vancomycin treatment. Infection treatment failure included signs or symptoms of infection persisting or worsening at least 48 hours after the start of vancomycin treatment and within 7 days after the end of vancomycin treatment. Data on the involved pathogens and MIC values are shown in Supplemental Digital Content 1 (see Table S2, The target steady-state trough concentration was set at 10–20 mg/L, which aligns with the TDM guideline for adult patients.5 The AUC24 was calculated using the Bayesian approach (SmartDose software). AUC24 >700 mg·L/h was deemed the nephrotoxicity threshold.20

The primary endpoints were the probability of trough concentration target attainment and incidence of vancomycin-associated nephrotoxicity defined by the RIFLE criteria (RIFLE criteria are presented in Table S1, Supplemental Digital Content 1, Secondary endpoints included the eradication of bacteria, rate of infection treatment failure, and duration of hospitalization.

Vancomycin TDM Assay

The enzyme-multiplied immunoassay technique, which has a detection interval of 2.0–100.0 mg/L, was applied for vancomycin TDM. A central laboratory tested all the blood samples within 24 hours and performed intrabatch and interbatch quality control according to the China National Accreditation Service for Conformity Assessment standard.

Statistical Analysis

For the primary outcome of vancomycin-associated nephrotoxicity development defined by the RIFLE criteria, a required sample size of 64 patients per study group is needed using G power 3.1, considering a two-sided test with α = 0.05, β = 0.20, and an expected incidence of vancomycin-associated nephrotoxicity in non-TDM (25%) and TDM (7%), as previously reported.21,22 When a dropout rate of 20% is considered, 77 patients per group were projected.

SmartDose ( is an intuitive web-based program. Previously, Gao Yucheng et al15 performed a detailed evaluation of the system function and applicability of the SmartDose software. It provides initial design and the adjustment of dose regimens based on the TDM results and a user-defined module to facilitate optimal vancomycin therapy. SmartDose has a high computational reliability that is validated by NONMEM (version 7.3, Icon, PA).23 Meanwhile, SmartDose is established as a web-based application, and its operational flexibility makes it an efficient tool for vancomycin dose optimization in routine clinical settings. SmartDose is adaptable with 4 PPK models that include adults and elderly patients,24 neurosurgical patients,25 children,26 and newborns.27 Because the population examined herein comprised patients with renal insufficiency, only the PPK model of Chinese adult patients was adopted.24 The predictive performance of this model was previously evaluated and validated. The diagnostic plots of the current data are presented in Supplemental Digital Content 1 (see Figure S1, The analytical approach used for each patient is maximum a posteriori probability Bayesian. An additive error model best described the data in the original population PK analysis, although the present assay error (which is part of the residual error model) follows a different error pattern. The PPK model of Chinese adult patients24 was defined with the following equation:

All statistical analyses were performed with SPSS 22.0 software and R version 3.5.3. Descriptive data for each group were expressed as mean ± SD or median (interquartile range, IQR), according to normal or non-normal distribution, respectively. A value of P ≤ 0.05 was considered to indicate statistical significance.

To adjust our analysis to control for potential bias and balance the observed covariates before comparing the study outcomes, a propensity score-matched analysis was created using a 1:1 ratio. In addition, the caliper length was set to 0.2 SDs of the logit of the propensity score.28 Significant variables between TDM and non-TDM group (Table 1) in the full cohort were assessed as covariates to generate a propensity score. Univariate and multivariate logistic regression analyses were performed in the matched cohort to determine the impact of vancomycin-associated variables on nephrotoxicity. Variables with P ≤ 0.05 in the univariate analysis were combined in the multivariate analysis. A multivariate logistic regression analysis was performed to identify the independent risk or protective factors influencing vancomycin-associated nephrotoxicity, and a Kaplan–Meier survival analysis was performed to estimate the cumulative incidence of nephrotoxicity.

Characteristics of Patients With TDM Compared to Those With Non-TDM in the Full Cohort
Characteristics of Patients With TDM Compared to Those With Non-TDM in the Full Cohort

Pharmacoeconomic Analysis

We used a decision-analytic model from a health system in China to assess the cost effectiveness of TDM using TreeAge Pro 2011 [TreeAge Software, MA, (see Figure S2, Supplemental Digital Content 1,]. The probabilities of vancomycin-associated nephrotoxicity data were derived from our study. The costs were obtained from the National Health and Family Planning Commission of the People's Republic of China ( The costs accrued during the monitoring of vancomycin levels were calculated using the following: cost of the serum vancomycin assays; costs of the time spent by nurses; costs of laboratory test; pharmacists performing these monitoring activities; and treatment of vancomycin-associated nephrotoxicity. The outcomes of interest were treating nephrotoxicity costs and the nephrotoxic episodes prevented. The incremental cost effectiveness ratio (ICER) per nephrotoxic episode prevented was calculated using the frequencies of nephrotoxicity found in the 2 groups. Treatment strategies with an ICER of less than CNY 64,644 (ie, Chinese gross domestic product per capita in 2017) per nephrotoxic episode prevented were deemed acceptable (additional details of the cost-effectiveness methodology are presented in the Supplementary Text, Supplemental Digital Content 1,


Characteristics of the Full Cohort

A total of 168 patients were included in the full cohort (84 in the TDM group and 84 in the non-TDM group). The characteristics of patients are summarized in Table 1. Before matching, a total of 119 (70.8%, 119/168) patients were men, and median age was 60 years (IQR, 56–64 years). Most patients were treated for pulmonary infections (41%). Baseline creatinine, the use of concomitant nephrotoxins, and vancomycin treatment duration were comparable. In the 2 groups, 71% (119/168) of patients had mild renal insufficiency at the time of vancomycin initiation, and the remaining 29% (49/168) had moderate renal insufficiency. Patients in the non-TDM group were more likely to have had hypoproteinemia and a higher albumin level than those in the TDM group (P < 0.001). More patients in the TDM group had an adjusted dosing regimen (P < 0.001).

The cohorts were matched according to the propensity scores estimated using age, weight, site of infection, vancomycin duration of therapy, concomitant nephrotoxic agents, albumin, and hypoproteinemia. After matching, albumin and hypoproteinemia were comparable between the 2 groups. There were also no differences between the groups in the number of isolated pathogens, with Enterococcus faecium, methicillin-sensitive S. aureus (MSSA), and S. epidermidis as the top 3 pathogens.

Vancomycin Trough Concentrations for the Matched Cohort

More patients in the TDM group reached the target trough level [53 (80%) versus 33 (42%), P < 0.001] after the dosing regimen was adjusted (Table 2). In the TDM group, the proportion of vancomycin trough concentrations within 10.0–20.0 mg/L (67.8%) was higher than that in the non-TDM group (47.6%, P < 0.05) (Table 2).

Vancomycin Monitoring Parameters for Patients With Culture-Confirmed Gram-Positive Infections in the Matched Cohort

Vancomycin-Associated Nephrotoxicity for the Matched Cohort

The overall rate of vancomycin-associated nephrotoxicity (Table 1) was higher with non-TDM than TDM (19% versus 7%, P = 0.022, respectively). For the nephrotoxicity based on the RIFLE classification (Table 1), all patients who developed AKI met risk, injury, or failure. No patients required renal replacement therapy.

Univariable logistic analysis indicated that the receipt of TDM significantly reduced the incidence of vancomycin-associated nephrotoxicity (P = 0.027). Other variables that might be related to nephrotoxicity included vancomycin plus piperacillin/tazobactam (PTZ) combination therapy (P = 0.002), diuretic (P < 0.001) or nonsteroidal anti-inflammatory drug (P = 0.002), number of nephrotoxins (P < 0.001), vancomycin treatment duration (P < 0.001), and AUC24 >700 mg·L/h (P < 0.001). Based on multivariable logistic analysis, after controlling for the residual differences in vancomycin-associated nephrotoxicity (weight, sex, length of hospital stay, comorbidities, baseline albumin, baseline eGFR, and site of infection), vancomycin plus PTZ combination therapy (P = 0.009), receipt number of nephrotoxins (P = 0.001), treatment duration (P < 0.001), and AUC24 > 700 mgL/h (P = 0.003) were more likely to result in vancomycin-associated nephrotoxicity. Multivariable logistic analysis also revealed that TDM practice was the independent protective factor (P = 0.021) for vancomycin-associated nephrotoxicity (Table 3). Furthermore, Kaplan–Meier survival analysis revealed that the cumulative incidence of nephrotoxicity was significantly higher in patients with non-TDM than TDM (P < 0.001) (see Figure S3, Supplemental Digital Content 1,

Regression Logistic Analysis of Vancomycin-Related Nephrotoxicity in the Matched Cohort

Secondary Outcomes for the Matched Cohort

There was no difference in microbiological outcomes between the TDM and non-TDM groups (P = 0.438, see Table S3, Supplemental Digital Content 1, The median length of hospital stay was longer for patients in the non-TDM group than the TDM group (27 days vs 23 days; P = 0.035) (see Table S3, Supplemental Digital Content 1, There was no difference in the rates of vancomycin treatment failure between the 2 groups (see Table S3, Supplemental Digital Content 1,

Cost Analysis

A total of 138 patients with renal insufficiency were initially enrolled in the pharmacoeconomic analysis. Supplemental Digital Content 1 (see Table S4, shows the costs associated with TDM and nephrotoxicity, whereas Supplemental Digital Content 1 (see Table S5, shows the results of the cost-effectiveness analysis. In the present analysis, the ICER per nephrotoxic episode prevented by using TDM corresponded to CNY 22,638 Supplemental Digital Content 1 (see Table S5, Thus, TDM practice was deemed cost effective and could prevent the occurrence of vancomycin-associated nephrotoxicity within a willingness-to-pay threshold of CNY 64,644.

One-way sensitivity analyses revealed that all variations in a single-model parameter had no substantial impact on the primary analyses performed in the economic analysis. The results of the probabilistic sensitivity analysis revealed a 96.5% chance that TDM practice would be cost effective at a WTP threshold of CNY 64,644 (Fig. 2).

Incremental cost-effectiveness plane and table, with cost-effectiveness acceptability curves (CEAC). A, Cost-effectiveness acceptability curves. B, Monte Carlo simulation. Each blue spot represents one of the 1000 iterations. WTP, willingness-to-pay.


Antimicrobial Stewardship Programs have been proposed to prevent the development of multidrug-resistant bacteria and ensure the appropriate use of antibiotics. As a prospective review plus feedback is one of the Antimicrobial Stewardship Programs recommended by the IDSA,29 we considered vancomycin TDM and Bayesian forecasting as a practical approach to improve vancomycin TDM work and opted to perform this program. The findings suggest that TDM coupled with Bayesian forecasting may be a valuable tool for the handling of real-time vancomycin in patients with renal insufficiency. TDM could reduce the length of hospitalization, increase the probability of trough target attainment, and reduce the incidence of nephrotoxicity.

The Bayesian forecasting methodology combines a priori population-based data with a posteriori individual patient data to derive the most accurate calculation of dose and interval.13 Furthermore, it has a higher predictive ability for achieving a specific AUC24/MIC.15 Clinical studies have shown that the Bayesian method has excellent predictive performance.30 The accuracy of Bayesian estimation depends on the quality of the selected model.31 In our study, we selected the model of the Chinese population.24 Our findings are consistent with a clinical study where TDM coupled with Bayesian forecasting was better than clinician judgement for attaining the PK/PD targets.32 We also found that TDM practice significantly reduced the duration of hospitalization, which would, to a certain extent, lower the medical costs.

Many studies have evaluated the relationship between vancomycin trough concentration, a measure of exposure, and nephrotoxicity. In this article, we found that vancomycin-associated nephrotoxicity was significantly correlated with trough concentration (P < 0.001). Different guidelines recommend trough levels as low as 10 to 20 mg/L; however, there is little supportive evidence of efficacy.33 By contrast, many studies evaluated the safety of this recommendation by comparing the incidence rates of nephrotoxicity above and below 15 mg/L.5 Therefore, a potential hazard is more likely to occur with trough levels ≥15 mg/L than a potential benefit. Most studies suggest that a trough concentration >15 mg/L is an independent risk factor for nephrotoxicity.3,34 Moreover, a prospective study reported the applicable cutoff concentration for the Chinese population and may support the viewpoint that the target range of 10–15 mg/L is still applicable to adults infected with gram-positive bacteria in China.17

In the 2016 HAP/VAP guideline issued by IDSA/ATS, antibiotic dosing regimens using PK/PD instead of the manufacturer's prescribing information were emphasized.35 Currently, the use of trough level for vancomycin TDM has its limitations.36,37 In addition, many researchers have suggested that the AUC-guided vancomycin dosing is more relevant to clinical outcomes than trough level.38,39 Therefore, with rapid changes in bacterial resistance, the clinical application of the PK/PD theory is one of the reliable strategies for fulfilling the potential of existing antibiotics. However, in our study, correlating the PK/PD index with the clinical effectiveness could not be achieved because of missing MIC values for 71% of patients (tested by VITEK 2 Compact). The correlation between PK/PD targets and clinical outcomes will be evaluated in a further study using the exact MIC values and a PPK model generated from a larger population.

By using a decision analysis instead of different probabilities and costs, we found that the pharmacokinetic dosage adjustment of vancomycin is cost effective for preventing nephrotoxicity. Darko et al40 revealed that the cost of preventing one vancomycin-related nephrotoxicity episode by TDM was $5564 for patients receiving concomitant nephrotoxins. Similarly, Fernandez de Gatta et al41 revealed that the cost to prevent one mild or moderate vancomycin-related nephrotoxicity was $435 and $1,307, respectively. Both studies indicated that vancomycin TDM may be cost effective for preventing AKI. Although some studies have investigated the pharmacoeconomics of TDM, prospective research is still warranted.

This study had limitations. First, it was monocentric. Second, we could not intervene in the initial vancomycin dosage regimen. Third, we measured vancomycin trough concentration using blood samples; however, blood concentration may differ from the concentrations at the site of infection. Nonetheless, blood samples are easy to analyze, and the unbound concentration of vancomycin might correlate with concentrations at the site of infection in patients. Fourth, although patients with eGFR <30 mL/min/1.73 m2 were excluded from the study according to the research design, the trough concentration of vancomycin was closely monitored in severe renal insufficiency patients in the clinic. Finally, we did not evaluate the correlation between PK/PD index and clinical outcomes. Nevertheless, the American Society of Health-System Pharmacists5 and the Japanese Society of Chemotherapy4 support the use of either the 10–20 mg/L trough concentration range or AUC24/MIC ≥400 to improve clinical outcome.


TDM coupled with Bayesian forecasting is a valuable tool for the appropriate handling of vancomycin by clinicians. Based on our findings, TDM could safely and effectively decrease the length of hospitalization and increase the probability of attaining the vancomycin trough target. The TDM of vancomycin could also reduce the incidence of vancomycin-associated nephrotoxicity and be cost effective for preventing nephrotoxicity. Collectively, to some extent, our results shed light on the benefits of vancomycin TDM.


1. Levine DP. Vancomycin: a history. Clin Infect Dis. 2006;42(suppl 1):S5–S12.
2. Moellering RC Jr. Vancomycin: a 50-year reassessment. Clin Infect Dis. 2006;42(suppl 1):S3–S4.
3. Lodise TP, Patel N, Lomaestro BM, et al. Relationship between initial vancomycin concentration-time profile and nephrotoxicity among hospitalized patients. Clin Infect Dis. 2009;49:507–514.
4. Matsumoto K, Takesue Y, Ohmagari N, et al. Practice guidelines for therapeutic drug monitoring of vancomycin: a consensus review of the Japanese Society of Chemotherapy and the Japanese Society of Therapeutic Drug Monitoring. J Infect Chemother. 2013;19:365–380.
5. Rybak M, Lomaestro B, Rotschafer JC, et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am J Health-Syst Pharm. 2009;66:82–98.
6. Ye ZK, Chen YL, Chen K, et al. Therapeutic drug monitoring of vancomycin: a guideline of the division of therapeutic drug monitoring, Chinese Pharmacological Society. J Antimicrob Chemother. 2016;71:3020–3025.
7. Duffull SB, Begg EJ. Vancomycin toxicity. What is the evidence for dose dependency? Adverse Drug React Toxicol Rev. 1994;13:103–114.
8. Leader WG, Chandler MH, Castiglia M. Pharmacokinetic optimisation of vancomycin therapy. Clin Pharmacokinet. 1995;28:327–342.
9. Tonkin AL, Bochner F. Therapeutic drug monitoring and patient outcome. A review of the issues. Clin Pharmacokinet. 1994;27:169–174.
10. Álvarez R, López Cortés LE1, Molina J, et al. Optimizing the clinical use of vancomycin. Antimicrob Agents Chemother. 2016;60:2601–2609.
11. Steinmetz T, Eliakim-Raz N, Goldberg E, et al. Association of vancomycin serum concentrations with efficacy in patients with MRSA infections: a systematic review and meta-analysis. Clin Microbiol Infect. 2015;21:665–673.
12. Zhang L, Wang F, Wang L, et al. Prevalence of chronic kidney disease in China: a cross-sectional survey. Lancet. 2012;379:815–822.
13. Avent ML, Vaska VL, Rogers BA, et al. Vancomycin therapeutics and monitoring: a contemporary approach. Intern Med J. 2013;43:110–119.
14. Lodise TP, Drusano GL, Zasowski E, et al. Vancomycin exposure in patients with methicillin-resistant Staphylococcus aureus bloodstream infections: how much is enough? Clin Infect Dis. 2014;59:666–675.
15. Gao YC, Jiao Z, Huang H, et al. Development of decision system for individualization of vancomycin dosage. Acta Pharm Sin. 2018;53:104–110.
16. Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612.
17. Liang X, Fan Y, Yang M, et al. A prospective multicenter clinical observational study on vancomycin efficiency and safety with therapeutic drug monitoring. Clin Infect Dis. 2018;67(suppl_2):S249–S255.
18. Bellomo R, Ronco C, Kellum JA, et al. Acute Dialysis Quality Initiative workgroup. Acute renal failure—definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care. 2004;8:R204–R212.
19. Neely MN, Kato L, Youn G, et al. Prospective trial on the use of trough concentration versus area under the curve to determine therapeutic vancomycin dosing. Antimicrob Agents Chemother. 2018;62:e02042–e020417.
20. Suzuki Y, Kawasaki K, Sato Y, et al. Is peak concentration needed in therapeutic drug monitoring of vancomycin? A pharmacokinetic-pharmacodynamic analysis in patients with methicillin-resistant staphylococcus aureus pneumonia. Chemotherapy. 2012;58:308–312.
21. Huang YZ, Mao MY, Yuan KX. Clinical application of drug monitoring of vancomycin. Chin J Infec Chemother. 2011;11:295–298.
22. Welty TE, Copa AK. Impact of vancomycin therapeutic drug monitoring on patient care. Ann Pharmacother. 1994;28:1335–1339.
23. Sheiner LB, Beal SL. NONMEM Users Guide—Part II Users Supplemental Guide. San Francisco, CA: Regents of the University of California; 1988.
24. He XR, Liu ZG, Ji SM, et al. Population pharmacokinetics of vancomycin and prediction of pharmacodynamics in the Chinese people. Acta Pharm Sin. 2014;49:1528–1535.
25. Lin WW, Wu W, Jiao Z, et al. Population pharmacokinetics of vancomycin in adult Chinese patients with post-craniotomy meningitis and its application in individualised dosage regimens. Eur J Clin Pharmacol. 2016;72:29–37.
26. Liu TT, Deng CH, Cheng DH, et al. Population pharmacokinetics of vancomycin in Chinese pediatric patients. Int J Clin Pharmacol Ther. 2017;55:509–516.
27. Li JJ, Liu YX, Tang L, et al. Population pharmacokinetics of vancomycin in Chinese neonates. Chin Pharm J. 2017;52:1434–1441.
28. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25:1–21.
29. Dellit TH, Owens RC, McGowan JE Jr, et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis. 2007;44:159–177.
30. Jelliffe RW, Schumitzky A, Van Guilder M, et al. Individualizing drug dosage regimens: roles of population pharmacokinetic and dynamic models, Bayesian fitting, and adaptive control. Ther Drug Monit. 1993;15:380–393.
31. Broeker A, Nardecchia M, Klinker KP, et al. Towards precision dosing of vancomycin: a systematic evaluation of pharmacometric models for Bayesian forecasting. Clin Microbiol Infect. 2019;25:1286.e1–1286.e7.
32. Hughes DM, Goswami S, Keizer RJ, et al. Bayesian clinical decision support-guided versus clinician-guided vancomycin dosing in attainment of targeted pharmacokinetic parameters in a paediatric population. J Antimicrob Chemother. 2020;75:434–437.
33. Kullar R, Davis SL, Levine D, et al. Impact of vancomycin exposure on outcomes in patients with methicillin-resistant Staphylococcus aureus bacteremia: support for consensus guidelines suggested targets. Clin Infect Dis. 2011;52:975–981.
34. Horey A, Mergenhagen KA, Mattappallil A. The relationship of nephrotoxicity to vancomycin trough serum concentrations in a veteran's population: a retrospective analysis. Ann Pharmacother. 2012;46:1477–1483.
35. Kalil AC, Metersky ML, Klompas M, et al. Palmer LB6 management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 clinical practice guidelines by the infectious diseases Society of America and the American Thoracic Society. Clin Infect Dis. 2016;63:e61–e111.
36. Morrison AP, Melanson SE, Carty MG, et al. What proportion of vancomycin trough levels are drawn too early?: frequency and impact on clinical actions. Am J Clin Pathol. 2012;137:472–478.
37. Martin JH, Norris R, Barras M, et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Clin Biochem Rev. 2010;31:21–24.
38. Finch NA, Zasowski EJ, Murray KP, et al. A quasi-experiment to study the impact of vancomycin area under the concentration-time curve-guided dosing on vancomycin-associated nephrotoxicity. Antimicrob Agents Chemother. 2017;61:e01293–17.
39. Fukumori S, Tsuji Y, Mizoguchi A, et al. Association of the clinical efficacy of vancomycin with the novel pharmacokinetic parameter area under the trough level (AUTL) in elderly patients with hospital-acquired pneumonia. J Clin Phar Ther. 2016;41:399–402.
40. Darko W, Medicis JJ, Smith A, et al. Mississippi mud no more: cost-effectiveness of pharmacokinetic dosage adjustment of vancomycin to prevent nephrotoxicity. Pharmacotherapy. 2003;23:643–650.
41. Fernández de Gatta MD, Calvo MV, Hernández JM, et al. Cost-effectiveness analysis of serum vancomycin concentration monitoring in patients with hematologic malignancies. Clin Pharmacol Ther. 1996;60:332–340.

therapeutic drug monitoring; Bayesian forecasting; vancomycin-associated nephrotoxicity; renal insufficiency patients; pharmacoeconomic analysis

Supplemental Digital Content

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the International Association of Therapeutic Drug Monitoring and Clinical Toxicology.