Optimising Therapeutic Target Attainment in Vancomycin Therapy for Patients Requiring Intermittent Haemodialysis

ABSTRACT

Aims

To describe the pharmacokinetics of vancomycin in patients requiring intermittent haemodialysis (iHD) and provide guidance for optimised dosing for a three‐times‐weekly post‐iHD regimen.

Methods

A population pharmacokinetic study using retrospectively collected data was conducted in a remote Australian dialysis centre. Adult patients receiving post‐iHD vancomycin therapy between 1st January 2017 and 31st July 2023 were screened. Inclusion criteria: ≥ 3 months of iHD, prescribed vancomycin, and recorded ≥ 1 vancomycin concentration. Pharmacokinetic analysis was performed using Monolix. The probability of pharmacokinetic/pharmacodynamic target attainment (PTA) for pre‐iHD trough concentrations of 15–20 mg/L was simulated for various loading and maintenance doses.

Results

A total of 302 vancomycin concentrations were available from 80 courses (58 patients, 45 were female) with a median age of 52 (IQR 45–61) years and weight of 75 (66.5–85.5) kg. A one‐compartment model adequately described the data. Actual body weight was the only covariate retained in the final model. The estimated vancomycin clearance during non‐iHD periods, during iHD periods, and volume of distribution were 0.54 L/h, 4.84 L/h, and 87.02 L, respectively. The simulations supported weight‐based loading doses. The time to the next iHD session also significantly influenced the PTA of loading doses. Subjects with higher weight required lower weight‐based doses, and vice versa. The PTA of maintenance doses was associated with the pre‐iHD trough concentrations and the dosing intervals. Weight had a limited impact on the PTA of maintenance doses.

Conclusion

A comprehensive vancomycin dosing nomogram was suggested for patients requiring iHD. Validation in future prospective studies is recommended.

Summary at a Glance

A population pharmacokinetic study aimed at optimising vancomycin therapeutic target attainment in patients requiring intermittent haemodialysis was conducted in remote Australia. A comprehensive dosing nomogram was devised, addressing gaps unaccounted for by present guidelines: dosing intervals of loading doses, weight‐based doses for weight categories, and dosing intervals of maintenance doses.

Optimal vancomycin dosing in dialysis patients.

1. Introduction

Infection is the leading contributor to morbidity and mortality in patients requiring chronic intermittent haemodialysis (iHD) [1]. A high incidence of Staphylococcus aureus skin and soft tissue and bloodstream infections has been reported in this population [2, 3]. The global emergence of methicillin‐resistant Staphylococcus aureus (MRSA) poses a significant burden on iHD patients, particularly in those residing in rural or remote regions where iHD has been identified as an independent risk factor for MRSA infection [4].

Vancomycin remains the mainstay of treatment for serious MRSA infections in most countries. It is a glycopeptide and exhibits bactericidal properties against common Gram‐positive pathogens [5]. Vancomycin's bactericidal effect is time‐dependent with concentration dependence. It has an efficacy target of an area‐under‐the‐concentration‐time curve to the minimum inhibitory concentration (AUC:MIC) ratio of 400–600:1 for MRSA [5, 6]. However, this target has not been adequately examined in end‐stage renal disease (ESRD) for therapeutic drug monitoring (TDM), where a significant prolongation of vancomycin's elimination half‐life has been observed. As a result, the therapeutic target trough concentration range of 15–20 mg/L is widely used for patients on iHD to accommodate the 2‐, 2‐ and 3‐day dialysis regimens [6]. Nevertheless, current dosing guidelines have not adequately addressed all identified complexities of vancomycin dosing in patients requiring iHD [6, 7]. Examples include: how to account for the timing of the next iHD session for both the loading and maintenance doses, when to employ weight‐based dosing versus set doses, and how to consider vancomycin dosing during iHD versus non‐iHD periods [8].

Given the chronic physiological changes associated with ESRD and iHD and the high incidence of MRSA infections reported in this patient group [4], we aimed to evaluate the pharmacokinetics (PK) of vancomycin in infected patients requiring iHD and develop a comprehensive dosing nomogram that optimises the probability of pharmacokinetic/pharmacodynamic (PK/PD) target attainment.

2. Materials and Methods

2.1. Setting and Study Population

This population PK study was conducted using retrospectively collected data from a renal dialysis unit and the primary referral centre in remote Central Australia. All patients prescribed a vancomycin course between 1 January 2017 and 31 July 2024 were screened for inclusion. Patients were included in the study if they met the following criteria: (i) ≥ 18 years of age, (ii) on iHD ≥ 3 months, and (iii) ≥ 1 vancomycin concentration recorded. Episodes with incomplete documentation were excluded. Data collected included patient age, weight, gender, ethnicity, dialysis vintage, infection type, dose and time of vancomycin administration, time and concentration of vancomycin plasma samples, and initiation and cessation times of dialysis sessions. Ethical approval was obtained from the Northern Territory Human Research Ethics Committee (project code HREC‐2024‐4950).

2.2. Dosing Regimen and Blood Sampling

Vancomycin dosing generally followed a local dosing guideline (see Table S1), with loading doses administered during non‐dialysis periods at an infusion rate of 500 mg/h. Depending on the treatment indication, loading doses of 1 g or 25 mg/kg based on actual body weight (ABW) were recommended. Maintenance doses, guided by a simple dosing nomogram, were administered three times a week to align with dialysis sessions, with intervals of 2, 2, and 3 days. These doses were given during the last hour of a 4‐h iHD session. Most of the vancomycin concentrations recorded were pre‐iHD trough concentrations, defined as samples collected within 30 min before initiating iHD. The dialysers used at the centre were: FX80, FX100, and FX120 (Fresenius Medical Care, Hesse, Germany), with ultrafiltration coefficients of 59, 73, and 87 mL/h/mmHg, and surface areas of 1.8, 2.2, and 2.5m², respectively. Filters were generally selected on the body surface area of each patient.

2.3. Handling of Samples and Drug Assay

Blood samples were collected in accordance with local guidelines. After collection, samples were sent to the hospital pathology laboratory, where they were centrifuged at 5000 rpm for 6 min at 2°C–8°C. Plasma was then aspirated and analysed using the Vitros XT7600 integrated system (QuidelOrtho corporation, California, US).

2.4. Population Pharmacokinetic Analysis and Validation

Population pharmacokinetic (popPK) analysis was conducted using the non‐linear mixed‐effects software Monolix version 2024R1 (Lixoft SAS, a Simulation Plus company, Anthony, France) which implements the stochastic approximation expectation maximisation (SAEM) algorithm. Individual estimates for the PK parameters were assumed to follow a log‐normal distribution. The interindividual variability (IIV) was described using an exponential model, expressed by the equation ${\theta }_j={\theta }_p\times \exp \left({\eta }_j\right)$, where ${\theta }_j$ represents the estimate for a PK parameter in the jth patient as predicted by the model, ${\theta }_p$ is the typical population PK parameter value, and ${\eta }_j$ is a random variable from a normal distribution with zero mean and variance ${\omega }^2$, which was estimated.

One‐ and two‐compartment models with first‐order elimination were compared. Both models accounted for different clearances during dialysis (CL _iHD ) and non‐dialysis periods (CL _noniHD ). Proportional, constant, and combined error models were tested to describe the residual variability $\left(\epsilon \right)$. Model selection was based on the precision of parameter estimates, visual inspection of goodness‐of‐fit (GoF) plots, and numerical assessment of the objective function value (OFV) and the corrected Bayesian Information Criteria (BICc).

From the base model, the effects of various covariates on the PK parameters of vancomycin were assessed. Covariates tested included sex, age, and weight. A stepwise forward inclusion and backward elimination approach was used for covariate model building. A covariate was considered significant if it resulted in a reduction in the OFV > 3.84 (p < 0.05; χ ² distribution) during forward inclusion, and an increase greater than 6.663 (p < 0.01) during backward elimination.

Internal validation of the model was based on GoF plots, which included observations versus population and individual predictions, individual weighted residuals (IWRES) versus individual predictions and time, and normalised prediction distribution error (NPDE) versus population predictions and time. A visual predictive check (VPC) was also performed using 500 simulations with the final model. The stability of the final model was evaluated using a bootstrap method. A 1000‐run bootstrap resampling procedure was performed in Monolix. The median and 95% confidence intervals from the 1000 bootstrap runs for each parameter were compared to the corresponding estimates obtained from the original dataset used for model development.

2.5. Monte Carlo Simulations

Monte Carlo dosing simulations were performed using Simulx 2024R1 (Lixoft SAS, a Simulation Plus company, Anthony, France) based on the final popPK model to generate 500 plasma concentration‐time profiles of vancomycin. The probability of target attainment (PTA) was assessed by evaluating simulated plasma vancomycin concentrations resulting from different doses, dosing intervals, and ABW. The PK/PD target was a pre‐iHD trough vancomycin concentration between 15 and 20 mg/L [6]. The 24‐h area‐under‐the‐concentration‐time curve (AUC₂₄) was also simulated for each day of the maintenance dosing interval. The highest PTA results were used to construct a proposed dosing nomogram.

To explore the optimal loading dose, vancomycin doses of 10, 15, 20, 25, 30, and 35 mg/kg were simulated at time points of 12, 18, 24, 36, 48, 60, and 72 h in patients with ABW ranging from 40 to 140 kg (in 10 kg increments). To reflect the clinical setting in which patients are not receiving haemodialysis, CL _noniHD was used in the simulations for loading doses.

Additional simulations were also conducted to evaluate different maintenance doses (ranging from 250 to 2500 mg in 250 mg increments) administered every 48 or 72 h, taking into account pre‐dose trough concentrations from 2.5 to 35 mg/L (in 2.5 mg/L increments). Simulations were performed for ABW of 60, 80, and 100 kg. To mimic the clinical setting, initial vancomycin plasma concentrations were set as the trough concentrations recorded immediately prior to a 4‐h dialysis session. The simulated maintenance dose was administered 3 h into the dialysis session over a 1‐h infusion.

3. Results

3.1. Study Population

A total of 135 patients on iHD received a course of vancomycin during the study period. Of these, 27 patients had incomplete documentation, 26 had incomplete dosing or concentration‐time data, and two had not received dialysis for at least 3 months. Ultimately, 302 vancomycin concentrations from 80 courses (58 patients, 45 of whom were female) were available for popPK analysis. The median age and ABW were 52 (IQR 45–61) years and 75 (66.5–85.5) kg, respectively (see Table 1).

TABLE 1.

Patient characteristics and clinical information.

Parameter	n = 80
Age, years, median (IQR)	52 (45–61)
Female, n (%)	62 (78%)
Actual body weight, kg, median (IQR)	75 (66.5–85.5)
First Nations ethnicity, n (%)	80 (100%)
Number of vancomycin doses per patient, mean	3.66
Time on haemodialysis, years, median (IQR)	3 (1.5–6)
Indication for vancomycin, n (%)	Bacteraemia–28 (35%)
	SSTI–25 (31%)
	DFI–14 (18%)
	Others–13 (16%)

Abbreviations: DFI, diabetic foot infection; IQR, interquartile range; SSTI, skin and soft tissue infection.

3.2. Population Pharmacokinetic Model and Validation

A one‐compartment model, parameterised in terms of clearance during non‐iHD and during iHD periods (CL _noniHD , CL _iHD , respectively), and volume of distribution (V), was selected as the base model to describe the PK of vancomycin. The vancomycin model parameter estimates are presented in Table 2. A proportional error model best described the residual variability. According to the covariate selection criteria, only the effect of ABW on V was retained in the final model, described by the following equations:

$$\mathit{CL}=\left\{\begin{array}{c}{\mathit{CL}}_{\mathit{\text{noniHD}}}\mathit{\text{dialysis}}\mathit{off}\\ {\mathit{CL}}_{\mathit{iHD}}\mathit{\text{dialysis}}\mathit{on}\end{array}\right.$$

$$V=87.02*{\left(\frac{\mathit{\text{weight}}}{75}\right)}^{0.46}$$

GoF plots were deemed acceptable based on visual inspection. Figure 1 presents the GoF plots of observed versus population predictions and observed versus individual predictions for vancomycin concentrations. Bootstrap results are shown in Table 2, demonstrating the stability of the final model. The VPC is presented in Figure 2, showing that the empirical 5th, 50th, and 95th percentiles from the simulated data fell within the corresponding 95% confidence intervals.

TABLE 2.

Parameter estimates for the final population pharmacokinetic model of vancomycin.

Parameter	Estimate (RSE%) (Shr%)	Bootstrap median (95% CI)
Fixed effects
CL noniHD (L/h)	0.54 (4.23)	0.54 (0.46–0.61)
CL iHD (L/h)	4.84 (7.95)	4.86 (3.32–6.62)
V (L)	87.02 (3.62)	86.89 (80.47–93.29)
ABW effect on V	0.46 (32.9)	0.47 (0.17–0.77)

Random effects
Interindividual variability (IIV)
IIV_CL noniHD	0.22 (12.0) [27.0] (CV% 22.52)	0.23 (0.17–0.31)
IIV_V	0.28 (10.4) [10.2] (CV% 28.99)	0.27 (0.22–0.32)
Residual variability
Proportional	0.14 (4.45)	0.14 (0.13–0.16)

Abbreviations: ABW, actual body weight; CI, confidence interval; CL_iHD, vancomycin clearance when dialysis is on; CL_noniHD, vancomycin clearance when dialysis is off; CV, coefficient of variation; IIV, interindividual variability; RSE, relative standard error; Shr, shrinkage; V, volume of distribution.

FIGURE 1.

Goodness‐of‐fit plots. (A) observations vs population predictions; (B) observations vs individual predictions. The solid black line represents the line of identity, while the dashed red line corresponds to the regression line.

FIGURE 2.

Visual predictive check. The dots represent the observed concentrations. The lines show the empirical 5th, 50th, and 95th percentiles from the simulated data, while the shaded areas depict the 95% confidence intervals for these percentiles, derived from 500 simulations.

3.3. Dosing Simulations and Probability of Target Attainment

Table 3 and Table S2 present the PTA results for loading and maintenance doses evaluated separately. The selection of optimal loading doses depended on patient ABW and the time until the next iHD session. In general, patients with higher body weights required lower ABW‐based doses, while those with lower weights required higher ABW‐based doses. The PTA of maintenance doses was influenced by pre‐iHD trough concentrations and the length of the dosing interval, whereas ABW had a relatively limited impact. Based on the PTA simulations for maintenance dosing, a trough concentration between 15 and 20 mg/L closely corresponded to an AUC₂₄ of 400 mg h/L in the final 24 h of the dosing interval (see Table S2). Doses with the highest PTA were incorporated into a comprehensive dosing nomogram, providing recommendations for both loading and maintenance doses (see Tables 4 and 5).

TABLE 3.

Probability of target attainment (%) for weight‐based loading doses based on interval to next dialysis session.

Interval	Dose	Actual body weight (kg)
		40	50	60	70	80	90	100	110	120	130	140
12 h	15 mg/kg	0.9	2.9	6.9	11.7	21.2	24.6	31.9	35.6	37.2	38	39.7
	20 mg/kg	11.8	22.1	32.4	41.7	40.5	40.2	41.2	35.5	30.2	30.4	23.6
	25 mg/kg	31.5	39.9	43.5	38.4	31.6	30	21.2	15.3	12	9	7.8
	30 mg/kg	42	39	32.2	21.8	18.8	12.8	8.1	5.5	3.4	2.4	1.6
	35 mg/kg	38.5	26.8	16	11.3	7.5	4.9	2.4	1.4	1.3	0.9	0.2
18 h	15 mg/kg	0.4	1.6	4.4	8.7	16.4	23.9	30.1	33.6	36.8	38	40.6
	20 mg/kg	7	18.4	27.4	38.7	39.6	40.7	42.6	39.9	37.8	32.7	26.1
	25 mg/kg	28.3	37.1	43.8	42.9	35	32.6	23.7	17.2	14.6	10.1	8.8
	30 mg/kg	41.3	43	35.7	26.6	21.6	14.7	9.2	6.2	3.9	3	2
	35 mg/kg	41.5	31.5	19.1	12.9	9.2	5.8	2.6	1.9	1.6	1.1	0.2
24 h	15 mg/kg	0	0.7	2.8	6.2	13.1	19.9	26.5	32.3	35.9	38.3	42.3
	20 mg/kg	3.9	12.6	22.3	35	37.5	40.8	45.6	43.1	39.5	36.7	28.6
	25 mg/kg	21.8	36.6	45.8	47.2	39.4	37.4	27.8	20.9	17	11.8	9.9
	30 mg/kg	39.5	47.1	39.7	31.7	25	17.7	10.2	7.7	5	3.5	2.3
	35 mg/kg	47.7	38.2	23.3	15.8	10.7	6.4	3.3	2.2	1.6	1.3	0.2
36 h	15 mg/kg	0	0.1	0.5	2.4	5.8	12	20	25.3	32.1	35.5	42
	20 mg/kg	0.4	4.9	12.1	23.5	34	39.8	47.3	48.9	44.7	41.8	34.5
	25 mg/kg	8.5	25	41.7	48.9	46.9	41.7	35.4	26.9	22.7	17.1	14.4
	30 mg/kg	29.6	46.6	49.6	39.5	33.1	24.2	14.1	9.6	6.4	4.5	3.1
	35 mg/kg	48.1	46.5	33.6	20.9	15.6	9.1	4.7	2.6	2.1	1.6	0.4
48 h	15 mg/kg	0	0	0	0.4	1.4	4.6	10.3	18.2	24.9	31.2	37.8
	20 mg/kg	0	0.6	4.4	12.9	24.4	35.1	45	48.4	47.8	47.2	40.5
	25 mg/kg	1.4	12.1	28.2	45.3	47.9	50.1	45.7	36.4	31.4	22.7	18.8
	30 mg/kg	13.3	36.8	52.9	50.3	43.1	33	19.1	13.6	9	6.1	4.3
	35 mg/kg	35.8	53.3	47.1	31.5	22.2	14.4	7.1	4.4	2.6	1.9	0.6
60 h	15 mg/kg	0	0	0	0	0.5	1.5	4.3	8.5	16.6	23.4	31
	20 mg/kg	0	0.2	1.4	5.5	13.9	24.9	36.7	47.2	47.7	49.8	47.4
	25 mg/kg	0.4	3.6	15.3	32	45.2	51	52.7	47.7	40.2	31.4	23.9
	30 mg/kg	4	21.7	44.9	55.2	51.2	41.5	29.1	19.1	13.9	8.5	7.2
	35 mg/kg	19.2	47.1	55.7	44.5	32	20.4	10.5	6.5	3.8	2.4	1.1
72 h	15 mg/kg	0	0	0	0	0.1	0.2	1.6	3.4	8.9	12.9	19.8
	20 mg/kg	0	0	0.1	1.7	5.4	14.2	26.3	36.8	43.8	49.7	53.9
	25 mg/kg	0	0.9	6	18.6	34.1	47.8	56.1	55	49	41.3	30.8
	30 mg/kg	1.1	9.2	28.3	50.2	55	52.4	41.5	29	22.1	13.9	10.5
	35 mg/kg	6.9	30.1	53.8	56	44.6	30.4	16.5	9.4	6.3	3.4	1.9

Note: The bold values are the highest PTA for their corresponding body weight and interval.

TABLE 4.

Recommended loading doses based on the highest PTA values.

Interval	40–59 kg	60–79 kg	80–99 kg	100–119 kg	120–139 kg	140–150 kg
12 h	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg	15 mg/kg	15 mg/kg
18 h	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg	15 mg/kg	15 mg/kg
24 h	35 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg	20 mg/kg	15 mg/kg
36 h	35 mg/kg	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg	15 mg/kg
48 h	35 mg/kg	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg	20 mg/kg
60 h	35 mg/kg	35 mg/kg	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg
72 h	35 mg/kg	35 mg/kg	30 mg/kg	25 mg/kg	20 mg/kg	20 mg/kg

Abbreviation: PTA, probability of target attainment.

TABLE 5.

Recommended maintenance doses based on the highest PTA values.

Pre‐iHD vancomycin conc (mg/L)	Dose for 48 h iHD interval	Dose for 72 h iHD interval
0 to < 5	2000 mg	2250 mg
5 to < 7.5	1750 mg	2000 mg
7.5 to < 10	1500 mg	1750 mg
10 to < 12.5	1250 mg	1500 mg
12.5 to < 15	1250 mg	1250 mg
15 to < 17.5	1000 mg	1250 mg
17.5 to < 20	750 mg	1000 mg
20 to < 22.5	750 mg	750 mg
22.5 to < 25	500 mg	500 mg
25 to < 27.5	250 mg	250 mg

Abbreviations: pre‐iHD conc, pre‐dialysis trough vancomycin concentration; PTA, probability of target attainment.

4. Discussion

In response to the high community and hospital prevalence of MRSA infection in remote Australia, we developed vancomycin dosing nomograms, based on the PTA, for both loading and maintenance doses in adult patients requiring iHD. We found that optimal loading doses were strongly associated with a patient's ABW and the time until the next iHD session. Maintenance doses were primarily influenced by pre‐iHD plasma vancomycin concentrations and the length of the dosing interval. These factors were incorporated into the dosing simulations and accounted for in the proposed nomograms. To the best of our knowledge, this study presents the most comprehensive set of vancomycin dosing simulations to date, incorporating multiple clinically relevant parameters that significantly influence the PTA in patients requiring iHD.

In this study, we report a V of 87.02 L (1.16 L/kg), which is consistent with previously published data in patients with ESRD (0.9 ± 0.21 L/kg and 1.00 ± 0.12 L/kg) [9, 10]. ABW was the only covariate retained in the final model, and simulation results indicated that PTA was influenced by dose, ABW and dosing interval. This supports the use of weight‐based dosing for loading doses. Conversely, ABW had less impact on the PTA of maintenance doses, particularly when pre‐iHD vancomycin concentrations were higher (see Table S2). A slightly stronger association between ABW and PTA was observed at lower pre‐iHD vancomycin concentrations (i.e., when a ‘reloading’ component was required). Overall, we propose a maintenance dosing nomogram that does not incorporate ABW.

We report a low typical vancomycin CL _noniHD of 0.54 L/h in our cohort with ESRD, which is consistent with values reported in other studies (0.29, 0.35, and 0.41 L/h) [9, 10, 11]. This is markedly lower than the CL reported in individuals with normal renal function (CL 3.48–5.90 L/h in those with creatinine clearance > 60 mL/min) [9, 12]. Despite the low CL _noniHD observed, the dosing interval remained strongly associated with PTA. Accordingly, we incorporated the dosing interval as a parameter in our simulations, and it is reflected in both the loading and maintenance dosing nomograms (see Tables 4 and 5). This has been identified as a gap for loading doses in previous studies [8]. By selecting a loading dose based on the time until the next dialysis session, optimal PTA can be anticipated. For maintenance doses, dosing intervals were fixed at 48 and 72 h to simulate conventional dialysis schedules.

Administering vancomycin during the final hour of iHD, rather than post‐iHD, is a common practice in dialysis units [13]. This approach benefits both staff and patients by reducing the overall time burden associated with facility‐based dialysis care [14]. It helps avoid delays, maintains treatment continuity, and prevents disruptions to workflow in dialysis units [13]. Vancomycin clearance during iHD must be considered when developing a PK model. This study used an advanced polysulfone dialyser membrane (Helixone), a high‐flux dialyser membrane with ultrafiltration coefficients ranging from 59 to 87 mL/h/mmHg. We report a CL _iHD of 4.84 L/h, consistent with previous findings for high‐flux dialysers (4.56–7.84 L/h) [15, 16, 17]. In contrast, low‐flux dialyser membranes yield significantly lower vancomycin CL _iHD values, ranging from 0.24–0.9 L/h for cuprophane and cellulose triacetate [17, 18, 19, 20], and 2.58–3.48 L/h for cellulose acetate and polyacrylonitrile [19, 21]. Currently, synthetic high‐flux and medium‐cutoff membranes dominate clinical practice and show minimal differences in dialysis efficacy for small‐ to medium‐sized molecules, [22] such as vancomycin. Since iHD sessions are typically limited to 4 h every 48 or 72 h, the amount of vancomycin cleared is unlikely to differ significantly across high‐flux membranes. This suggests that our proposed maintenance doses can be generalised to other high‐flux dialysers. It is important to note that simulations for loading doses were performed during non‐dialysis periods (i.e., with CL _noniHD ) to reflect real‐world clinical practice, where initial vancomycin doses for patients with signs of systemic infection were usually given in a timely manner rather than waiting for the next dialysis session [8]. Therefore, the consideration of enhanced drug clearance by the dialysis modality does not apply to our loading dose nomogram.

Over the past decades, studies have reported high vancomycin AUC values (e.g., above 600 mg h/L) when a target trough concentration range of 15–20 mg/L was used, leading to updated guidance from clinical guidelines recommending AUC‐based monitoring [23]. However, there is a paucity of data associating vancomycin trough concentrations with AUC in patients with ESRD requiring iHD. Therefore, guidelines continue to recommend a trough therapeutic target range of 15–20 mg/L for this patient population [6]. Our dosing simulations demonstrate that the pre‐dialysis trough concentrations of 15–20 mg/L align with an AUC₂₄ of 400 mg h/L on Day 2 and Day 3 of the dosing intervals, indicating that this dosing strategy effectively maintains therapeutic exposure without significant underdosing. This differs from observations in patients without ESRD and can be explained by the plateaued concentration‐time curve associated with the low CL in ESRD. Our findings support the validity of the therapeutic target trough concentrations of 15–20 mg/L in patients requiring iHD, and show that administering vancomycin during the final hour of iHD ensures sustained therapeutic drug concentrations until the next session.

First Nations populations have a greater risk for chronic kidney disease, and this risk is compounded by the remoteness of their residency [24, 25]. First Nations people requiring iHD in remote Australia have reported a higher incidence of infection requiring hospitalisation, as well as a higher risk of contracting MRSA and infection‐associated mortality compared with non‐First Nations Australians [26, 27]. The current study was conducted in a dialysis unit in remote Australia, where MRSA infections were frequently diagnosed, and optimised vancomycin therapy would be essential. The dosing recommendations from this study can increase the incidence of target attainment, thereby improving patient clinical outcomes. Although all participants included were First Nations people, based on the physicochemical characteristics of vancomycin, significant interethnic differences in vancomycin PK are unlikely. Hence, the results obtained from this study can justifiably be applied to patients of different ethnic backgrounds [28].

We have identified some limitations in our study. First, the popPK model was based on retrospective TDM data. Data accuracy may be compromised, and standard clinical TDM usually provides only one to two concentration‐time data points per dose interval. Nonetheless, we attempted to compensate for this limitation with a relatively large sample size (n = 80), with multiple doses of vancomycin administered per patient. Second, this study site only used high‐flux dialysers, the most commonly used dialyser types among patients requiring iHD. While our recommendations for loading doses can still be applied to patients using other dialyser types, this is not the case for maintenance doses due to the significant differences in CL _iHD between high‐flux and low‐flux dialysers. Lastly, due to the narrow therapeutic window used in our dosing simulations (15–20 mg/L), the highest PTA results for a given set of parameters can be as low as 35%. Nonetheless, this is not an uncommon observation for dosing simulations targeting a narrow concentration range [29, 30]. The population setting was predominantly based on three‐times‐weekly iHD; however, the value of the nomogram may also be extended to patients receiving incremental or twice weekly iHD, as part of a comprehensive personalised‐medicine approach. Lastly, clinical outcome data were beyond the scope of this study and were not collected. Due to the retrospective nature of this study, we recommend that future prospective studies be conducted to validate the nomogram proposed.

This study provides a comprehensive vancomycin dosing strategy for patients receiving iHD, incorporating weight‐based loading doses and interval‐based maintenance dosing. While highly applicable to high‐flux dialysis settings, the retrospective design and lack of prospective validation are limitations. Future prospective studies comparing AUC‐based vs. trough‐based dosing could refine these recommendations and further improve MRSA treatment outcomes in patients with ESRD.

Funding

This work was supported by The Hospital Research Foundation, CF‐EMCR‐106‐83100; Australian National Health and Medical Research Council, APP2007007, APP2009736, GNT1174758.

Conflicts of Interest

D.T. is a member of the Australian National Antimicrobial Stewardship Advisory Committee. A.L. is a member of the Australian Prescriber editorial advisory committee and has received a travel grant from AstraZeneca. W.N. is a member of the Royal Australasian College of Physicians—Aboriginal and Torres Strait Islander Health Committee; received a writing fee from Australian Doctor; received travel support to attend the College of Intensive Care Medicine Rural and Remote Health Conference 2024. J.A.R. received consulting fees from Sandoz, Wolters Kuwer, Gliead, Qpex, and Advanz Pharma; received speaking fees from Sandoz, Pfizer, MSD, and Gilead; received grants from Biomerieux and Pfizer. M.P.H., J.M., C.S., S.K., S.J., J.T.H., and X.L. have none to declare.

Supporting information

Table S1: Haemodialysis vancomycin order for Central Australia.

Table S2: Probability of target attainment of maintaining trough concentration between 15 and 20 mg/L (PTAtrough15‐20, in %) and AUC24‐48h, AUC48‐72h (in mg h/L) of maintenance doses for various weights, initial concentrations and dose intervals.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.