Population Pharmacokinetics and Pharmacodynamics of Meropenem in Critically Ill Patients: How to Achieve Best Dosage Regimen According to the Clinical Situation

Abstract

Background and Objectives
Meropenem is frequently used for the treatment of severe bacterial infections in critically ill patients. Because critically ill patients are more prone to pharmacokinetic variability than other patients, ensuring an effective blood concentration can be complex. Therefore, describing this variability to ensure a proper use of this antibiotic drug limits the rise and dissemination of antimicrobial resistance, and helps preserve the current antibiotic arsenal. The aims of this study were to describe the pharmacokinetics of meropenem in critically ill patients, to identify and quantify the patients’ characteristics responsible for the observed pharmacokinetic variability, and to perform different dosing simulations in order to determine optimal individually adapted dosing regimens.MethodsA total of 58 patients hospitalized in the medical intensive care unit and receiving meropenem were enrolled, including 26 patients with renal replacement therapy. A population pharmacokinetic model was developed (using NONMEM software) and Monte Carlo simulations were performed with different dosing scenarios (bolus-like, extended, and continuous infusion) exploring the impact of clinical categories of residual diuresis (anuria, oliguria, and preserved diuresis) on the probability of target attainment (MIC: 1–45 mg/L).ResultsThe population pharmacokinetic model included five covariates with a significant impact on clearance: glomerular filtration rate, dialysis (continuous and semi-continuous), renal function status, and volume of residual diuresis. The clearance for a typical patient in our population is 4.20 L/h and volume of distribution approximately 44 L. Performed dosing regimen simulations suggested that, for equivalent doses, the continuous infusion mode (with loading dose) allowed the obtaining of the pharmacokinetic/pharmacodynamic target for a larger number of patients (100% for MIC ≤ 20 mg/L). Nevertheless, for the treatment of susceptible bacteria (MIC ≤ 2 mg/L), differences in the probability of target attainment between bolus-like, extended, and continuous infusions were negligible.Conclusions
Identified covariates in the model are easily accessible information in patient health records. The model highlighted the importance of considering the patient’s overall condition (renal function and dialysis) and the pathogen’s characteristics (MIC target) during the establishment of a patient’s dosing regimen.

Full text

Vol.:(0123456789)
European Journal of Drug Metabolism and Pharmacokinetics (2021) 46:695–705
https://doi.org/10.1007/s13318-021-00709-w
ORIGINAL RESEARCH ARTICLE
Population Pharmacokinetics andPharmacodynamics ofMeropenem
inCritically Ill Patients: How toAchieve Best Dosage Regimen
According totheClinical Situation
AmauryO’Jeanson1,2· RomaricLarcher3· CosetteLeSouder4· NassimDjebli5· SoniaKhier1,2
Accepted: 25 July 2021 / Published online: 17 August 2021
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2021
Abstract
Background and Objectives Meropenem is frequently used for the treatment of severe bacterial infections in critically ill
patients. Because critically ill patients are more prone to pharmacokinetic variability than other patients, ensuring an effec-
tive blood concentration can be complex. Therefore, describing this variability to ensure a proper use of this antibiotic drug
limits the rise and dissemination of antimicrobial resistance, and helps preserve the current antibiotic arsenal. The aims of
this study were to describe the pharmacokinetics of meropenem in critically ill patients, to identify and quantify the patients’
characteristics responsible for the observed pharmacokinetic variability, and to perform different dosing simulations in order
to determine optimal individually adapted dosing regimens.
Methods A total of 58 patients hospitalized in the medical intensive care unit and receiving meropenem were enrolled,
including 26 patients with renal replacement therapy. A population pharmacokinetic model was developed (using NONMEM
software) and Monte Carlo simulations were performed with different dosing scenarios (bolus-like, extended, and continu-
ous infusion) exploring the impact of clinical categories of residual diuresis (anuria, oliguria, and preserved diuresis) on the
probability of target attainment (MIC: 1–45 mg/L).
Results The population pharmacokinetic model included five covariates with a significant impact on clearance: glomerular
filtration rate, dialysis (continuous and semi-continuous), renal function status, and volume of residual diuresis. The clearance
for a typical patient in our population is 4.20 L/h and volume of distribution approximately 44L. Performed dosing regimen
simulations suggested that, for equivalent doses, the continuous infusion mode (with loading dose) allowed the obtaining of
the pharmacokinetic/pharmacodynamic target for a larger number of patients (100% for MIC≤20 mg/L). Nevertheless, for
the treatment of susceptible bacteria (MIC ≤ 2mg/L), differences in the probability of target attainment between bolus-like,
extended, and continuous infusions were negligible.
Conclusions Identified covariates in the model are easily accessible information in patient health records. The model high-
lighted the importance of considering the patient’s overall condition (renal function and dialysis) and thepathogen’s char-
acteristics (MIC target) during the establishment of a patient’s dosing regimen.
* Sonia Khier
sonia.khier@umontpellier.fr
Extended author information available on the last page of the article
1 Introduction
Bacterial infections in critically ill patients are still a major
issue in modern medicine, due to their high prevalence
(approximately 50% [1]) and important mortality rate (up to
60% in special situations like septic shock or sepsis [1–4]).
Two main issues arise when critically ill patients have to
be treated with an antibiotic therapy. First, the available
therapeutic arsenal of anti-infection medicines is limited.
Bacterial pathogens responsible for infections in the inten-
sive care unit (ICU) are generally less sensitive than patho-
gens found in other care units [5, 6], and multidrug-resistant
bacteria (MDRB) are on the rise [7, 8].
Second, dosing regimens are rarely adapted to the
patients’ needs. Most dosing guidelines are currently built on
pharmacokinetics studies but without ICU patients, and as a
result exclude all the pharmacokinetic variability observed
in critically ill patients. Due to their critical condition and
related pathophysiological changes, ICU patients are more
prone to pharmacokinetic variability than other patients
[9–13]. To predict the impact of critically ill patients’
pharmacokinetic variability on drug exposure is complex,

696 A.O’Jeanson et al.
Key Points
Meropenem is a broad-spectrum antibiotic used in criti-
cally ill or ICU patients.
Sources of pharmacokinetic variability in ICU patients
are large and ensuring an efficient dose is complex.
Information on pharmacokinetics and pharmacodynam-
ics of meropenem helps to obtain a better antibiotic treat-
ment outcome in patients.
2019 in patients of the ICU of the university hospital of
Montpellier (France).
2.1 Patients
Inclusion criteria were known or suspected infection of a
critically ill patient in the medical ICU and treated with
meropenem. Exclusion criterion was age (< 18 years old).
Main causes of admission in the ICU were severe organ fail-
ure (heart, kidney, liver, respiratory, or hematologic), immu-
nodeficiency, hematologic malignancies, severe metabolic
complications, vital distress situations, and cardiac arrest
victims (acute or prolonged phase).
2.2 Drug Administration
All patients received meropenem (Meronem® IV, intrave-
nous, AstraZeneca, or its generic products) via infusion or
IV bolus. Drug administration information (dose, type, time,
etc.) was collected in the hospital electronic medical record
system. Meropenem dosing regimen was left up to the cli-
nician decision (1 or 2 g Q6H, Q8H, or Q12H ± loading
dose). Pharmacokinetic/pharmacodynamic objective was to
maintain serum concentration: Cmeropenem=5 × MIC during
100% of time.
2.3 Demographic andClinical Data
The patients’ demographic, clinical, and biological data
were recorded. Age, height, weight, sex, volume of resid-
ual diuresis, settings of renal replacement therapy (RRT,
type of RRT—continuous or semi continuous—and session
timetables), co-medications, shock-state associated informa-
tion, standard blood test results, isolated micro-organisms,
and meropenem corresponding MICs and number of days
of antibiotic therapy were collected. Creatinine clearance
values were derived from biological test results (blood and
urine), using Cockcroft–Gault, modification of diet in renal
disease (MDRD), chronic kidney disease-epidemiology col-
laboration (CKD-EPI) formulas or urine/plasma creatinine
ratio. Patient data were included in the pharmacokinetic
analysis if they contained informed dosing history and at
least one adequately documented and quantifiable merope-
nem concentration per patient.
2.4 Sample Analysis
Actual times for data sampling were left at the discretion of
the treating physician, depending on clinical observations.
Meropenem concentrations were determined with pharma-
cokinetic samples (5mL of venous blood in a serum clot
activator tube) collected during routine therapeutic drug
monitoring (TDM). For stability purposes, samples were
suggesting that empirical dosing strategies are not the best
way to achieve effective exposures [14, 15].
Therefore, it is crucial to study and describe this phar-
macokinetic variability to ensure a proper use of antibiotic
drugs, i.e., select the dose that will generate the optimal drug
exposure [16, 17], thus limiting the rise and dissemination
of antimicrobial resistance, and preserving the current thera-
peutic arsenal [18].
Meropenem is a broad-spectrum antibiotic drug pre-
scribed for MDRB infections or empirical treatment of
serious infections, commonly administered in critically ill
patients. Meropenem is a hydrophilic small molecule with
a low-level of protein binding (f < 2%) and a large tissue
distribution. The drug is mainly eliminated by the kidneys
(~70% of the dose excreted in the urine after 12 h) and
partially eliminated as an inactive metabolite (~28%). Mero-
penem is a dialyzable drug with a linear pharmacokinetics
between 500 mg and 2 g. Like all carbapenems, Meropenem
has a time-dependent antibacterial efficiency: to obtain an
optimal bactericidal activity, the blood concentration must
be maintained above the minimum inhibitory concentration
(MIC) during at least 40 % of the dosing interval (40% T >
MIC) [19].
The aims of this study are to describe the pharmacoki-
netics of meropenem in critically ill patients, to identify
and quantify the patients’ characteristics responsible for
the observed pharmacokinetic variability, and to perform
different dosing simulations in order to determine optimal
individually adapted dosing regimens.
2 Methods
This non-interventional study protocol was approved by
the Ethics Committee (study ID: CIMER, 2019_IRB-
MTP_03-01). Blood concentrations of meropenem were
collected retrospectively from November 2017 to February

697
Population Pharmacokinetics and Pharmacodynamics of Meropenem in the ICU
stored at −80°C until assay. After adaptation, the bioana-
lytical method [20] was validated according to the require-
ments of the NF EN ISO 15189 standard (accreditation of
medical biology laboratories) and the recommendations
of the European Medicines Agency for TDM context [21].
Total serum concentrations of meropenem were measured
using ultra-high-precision liquid chromatography (UHPLC)
with a diode array ultraviolet detection (UHPLC Ultima-
teTM 3000 System; ThermoFisher Scientific). The analytical
column used was a HPLC Hypersil™ Gold pentafluorophe-
nyl 100mm, 2.1mm diameter, 1.9µm grain size (Ther-
moFischer Scientific). Two mobile phases were used for the
sample analysis: mobile phase A was composed of milli-Q
water with a 10-mM concentration of phosphoric acid, while
mobile phase B was pure acetonitrile. Analysis data acquisi-
tion and control were possible with Chromeleon™ software.
Limit of quantification for meropenem was 3 ng/mL.
2.5 PopPK Analysis
Pharmacokinetic data from the study were analyzed by a
non-linear mixed-effects modeling approach using NON-
MEM (v7.4.1) and its first-order conditional estimation
method with the INTERACTION option. The development
of the model followed a three-stage strategy: (1) selection
of a structural model, (2) covariate screening and selection,
and (3) evaluation of the final model. Outliers were defined
as data points in the dataset that appear to be outside the
norm for that particular dataset (e.g., data with conditional
weighted residuals > 5) based on inspection of the results
from a preliminary PopPK run. A maximum 5% omission
of these outliers was possible upon report and justification.
After visual inspection of the data and a literature review
[22–35], different structural models have been explored
to describe the pharmacokinetic data (one- and two-
compartment(s) models with a first-order elimination rate).
Additive, exponential, proportional, and combined error
models were tested for residual variability. Both additive and
log-normal models were evaluated for between-subject vari-
ability (BSV). Selection criteria decreased in both objective
function value (OFV) and Akaike’s information criterion
(AIC); acceptance criteria included precision of parameters
estimation and goodness-of-fit plots. The model with the
lowest values of OFV and AIC was selected if data descrip-
tion and parameter estimation precision were adequate (i.e.,
relative standard errors for fixed and random effects <30%
and <50%, respectively).
In a second stage, collected demographic, biologic, and
clinical variables that were considered credible for affecting
the pharmacokinetic of meropenem were tested for inclusion
as covariates in the structural model, following the step-
wise covariate modeling (SCM) method [36]. The impact
of continuous covariates and categorical covariates was,
respectively, considered with six and three different func-
tions (Supplementary Material) during the forward selec-
tion step (ΔOFV=3.84, p<0.05). A covariate was kept
in the model if there was a significant improvement in the
fit over the previous model step (i.e., decrease in BSV of
the parameter and decrease in OFV/AIC). The relevance for
each included covariate was evaluated during the backward
deletion step (ΔOFV=10.8, p<0.001). Criteria for the
final SCM model were: a “minimization successful” result, a
decrease in the between-subject variability of the parameter
on which the covariate was retained, an accurately estimated
covariance matrix, a result of three significant digits for each
fixed parameter, and reaching acceptance criteria.
In a last stage, the final model was evaluated using
graphical and statistical internal validation methods. Pre-
diction-corrected visual predictive checks were performed,
the bootstrap resampling technique was used to build the
PopPK parameters confidence intervals (CI), and normal-
ized prediction distribution errors (NPDE) were computed
to assess prediction discrepancies. In addition, to evaluate
the predictive quality of the final model, different scores
were computed: root-mean-squared error (RMSE) and mean
prediction error (MPE) to evaluate the precision and bias,
respectively [37].
2.6 Pharmacokinetic/Pharmacodynamic
Simulations
The pharmacokinetic/pharmacodynamic target was the prob-
ability of target attainment (PTA) and defined by the number
of simulated patients who achieved 100% T > MIC. To per-
form the simulations, the population median value of glo-
merular filtration rate (GFR) was used while exploring the
impact of the three categories of residual diuresis (50mL,
250mL, and 750mL), corresponding, respectively, to anuria
state (<100mL/24 h), oliguria state (100–500mL/24h),
and preserved diuresis (>500mL/24h). Each simulation
generated meropenem concentration-time profiles for 1000
subjects per dosing regimen using the parameters from the
PopPK final model. Monte Carlo dosing regimen simula-
tions were performed with bolus-like administration (infu-
sion over 30 min), extended infusion (infusion over 5h),
and continuous infusion. Different dosing scenarios for both
bolus and extended infusion were generated: 1g Q6H or
Q8H and 2 g Q6H or Q8H. Dosing scenarios for continu-
ous infusion were generated with 1g loading dose (unique
30-min infusion) followed by continuous IV infusion Q4H
or Q6H. Choice of exploring these particular dosing regi-
mens falls on Montpellier’s medical ICU clinical habits in
treating patients with meropenem.

698 A.O’Jeanson et al.
3 Results
3.1 Demographic Characteristics
A total of 58 patients were enrolled in the study. From the
133 serum concentrations collected, 13 were below the
limit of quantification and 5 were considered outliers dur-
ing a preliminary PopPK run. As a consequence, 115 serum
concentrations were reported in the total dataset, with an
average of 2 samples per patient. Table1 summarizes the
patients’ demographic and biological characteristics. Of the
total patients, 26 were prescribed RRT, of whom 6 received
continuous RRT, 10 received semi-continuous RRT (either
sustained low-efficiency dialysis or intermittent hemodialy-
sis), and 10 were given a combination of previously cited
RRT techniques and plasmapheresis was used. Residual diu-
resis is routinely monitored in the medical ICU, hence more
than 2000 values of residual diuresis volume were collected
for our study population. Some patients have experienced
the states of anuria, oliguria, and/or preserved diuresis in
the same hospital stay.
3.2 PopPK Model
Pharmacokinetic observations were better described by a
one-compartment model with a linear clearance, using a
log-normal inter-individual variability on both parameters
(clearance and volume of distribution). No omega block
structure was retained between the two BSVs. Residual
variability was best described by a proportional error model.
Significant covariates were GFR (using the MDRD for-
mula), presence of a dialysis session (continuous: DIA_C,
or semi-continuous: DIA_SC), renal function status (RFS)
and volume of residual diuresis (RD), all influencing mero-
penem clearance (Table2).
The volume of residual diuresis was found to be associ-
ated with clearance except during a semi-continuous dialysis
session. GFRMDRD greatly improved the goodness-of-fit and
the OFV in patients with an impaired renal function, but had
no impact when patients were under dialysis. All covariate
relationships are described in Fig.1. A comparison between
PopPK parameters of both the structural model and the final
model is displayed in Table3.The clearance for a typical
patient in our population is 4.20 L/h and the volume of dis-
tribution is approximately of 44L.
3.3 Model Evaluation
Goodness-of-fit plots for the final model were evaluated and
showed no apparent visual bias for the predictions, as shown
in Fig.2. Plots of observations versus individual/popula-
tion predictions (IPRED/PRED) showed a good impression
of data fitting along the identity line. Plots of conditional
weighted residuals (CWRES) versus time after dose also dis-
played a uniformity of distribution along the y=0 line and
contained values between −3 and 3. A prediction-corrected
VPC confirmed the predictive performance of the model,
as shown in Fig.3. CWRES versus PRED, IWRES versus
IPRED, IWRES versus time after dose, quantile-quantile
Table 1 Demographics and
biological characteristics of
patients included in the study
BMI body mass index, CRCLCG creatinine clearance using Cockcroft–Gault formula, GFRMDRD glomerular
filtration rate using 4-variable MDRD formula (
MDRD
=
186.3
×(
creatinine
×
0.0113
)
−1.154
×
age
−0.203 ×A×
B
with creatinine = μmol/L, A = 0.742 if female, B = 1.21 if African), RFS renal function status
Variable n (%) Mean (SD) Median Range Reference values
Female 17 (29)
Male 41 (71)
Age (year) 62 (14.7) 64 18–85
Weight (kg) 72 (17.2) 68 39.5–124
BMI (kg/m2) 25 (6.1) 24 15–42 18.5–24.9
Albumin (g/L) 26 (4.3) 26 15–40 36–45
Lactates (mM) 2 (1.9) 1.4 0.5–22.7 5–22
Serum creatinine (µM) 154 (109) 131 15–737 50–130
Urine creatinine (mM) 3.9 (2.9) 3.2 0.1–17.2 8–18
Serum urea (mM) 13 (9.4) 10 0.5–46 2.5–7.1
Residual diuresis (mL/24h) 1316 (1422) 845 0–7300 >500
CRCLCG (mL/min) 72 (60) 52 10–376 F: 95±20
M: 110 ±20
GFRMDRD (mL/min/1.73 m2) 78 (84) 49 7–567 120
RFS=1 (i.e., GFR > 120 mL/
min/1.73 m2)
7 (12)

699
Population Pharmacokinetics and Pharmacodynamics of Meropenem in the ICU
plots for CWRES and NPDE are presented in Supplementary
material. Statistical distribution (median value and 95% CI)
for the final PopPK parameters obtained from the bootstrap
analysis are shown in Table4. Convergence rate for the boot-
strap analysis was of 93.8% (of 1000 runs), suggesting good
stability of the PopPK model. The 95% CI were reasonably
narrow and median bootstrap values for parameters were
in good agreement with the final PopPK parameters, indi-
cating the robustness of the final model. The final model’s
mean bias (MPE) and precision (RMSE) for IPRED were
−7.03% and 29.4%, respectively, better than those computed
for PRED, which were −8.24% and 53.6%, respectively.
The tendency to underpredict meropenem concentration
was noticed in both the structural model and the final model
for individual and population predictions. Figure4 presents
final individual clearance values of patients depending on
different categories of GFRMDRD , and stratified by the type
of dialysis technique used. Final individual clearance values
depending on different states of diuresis and stratified by
type of dialysis technique are shown in Fig.5.
3.4 Pharmacokinetic/Pharmacodynamic
Achievement
PTA computed from the results of the simulations performed
to evaluate different dosing regimens (bolus-like, extended,
and continuous infusion) in different states of residual diure-
sis (without dialysis) are shown in Figs.6, 7, and 8. A PTA
of >90% was considered acceptable for a pharmacodynamic
target of 100% T>MIC for meropenem.
Table 2 Covariates/clearance
relationship
ΔOFV difference of objective function value, GFRMDRD glomerular filtration rate using MDRD formula,
(Semi-) continuous dialysis DIAC and DIASC = 1 if continuous or semi-continuous dialysis is “on” and = 0
if dialysis is “off”, RFS renal function status = 1 if GFR≥120mL/min, else = 0
Covariate name Relationship ΔOFV
GFRMDRD CL1 = θCL + θMDRD × GFRMDRD −58.8 Step 1
Continuous dialysis CL2 = CL1 × (1− DIAC) + θDIA_C × DIAC−9.4 Step 2
Renal function status CL3 = CL2 × (1−RFS) + θRFS × RFS −12.6 Step 3
Residual diuresis CL4 = CL3 + θRD × ((RD/ 845) − 1) −9.6 Step 4
Semi-continuous dialysis CL5 = CL4 × (1− DIASC) + θDIA_SC × DIASC −12.9 Step 5
CLfinal = CL5 × exp(ωCL)
Fig. 1 Summary of covariate relationships for the computation of clearance for each type of patient. CL clearance, RFS renal function status, RD
residual diuresis, GFR glomerular filtration rate, θ PopPK parameters

700 A.O’Jeanson et al.
4 Discussion
Our study and specifically its dataset have been used as part
of a complementary work [38], where the predictive ability
of both the literature and the tweaked models on TDM
concentrations of meropenem in critically ill patients was
compared. This approach required the subsetting of our data
and led to insufficient overall results, even when applying
the Bayesian approach with prior information [39], which
comforted us in our strategy of building a new PopPK model
from an unsplit dataset. Our main finding is the identification
and quantification of an important pharmacokinetic variabil-
ity, partly due to a high heterogeneity among the critically ill
patients included in this study, and related to the ICU patient
status. The BSV has been explained, to a relatively large
extent, by five covariates, all included in the relationship
with CL. The inclusion of a renal function marker on mero-
penem’s clearance was expected, and was already identified
in the scientific literature. Unexpectedly, the renal function
marker to have the most mathematical significant impact on
CL was GFRMDRD, rather than the GFR estimated by the
CKD-EPI formula or another marker. The 4-variable MDRD
formula has been controversial because of a less optimal per-
formance in estimating the actual GFR of patients in criti-
cal care settings [40]. Nevertheless, GFRMDRD appeared as
a clinically impactful covariate. A significant relationship
Table 3 PopPK parameters comparison between structural model and
final model
RSE relative standard error,
RSE
(%)=
SE
Estimation
×
100
, CLtotal body
clearance, V apparent volume of distribution, BSV_CL between-sub-
ject variability associated with CL, BSV_V between-subject variabil-
ity associated with V, Prop_Res_Err proportional residual error, CV
coefficient of variation,
CV(%)=√e
𝜔2
−1×100
a The value presented here reflects the clearance for a typical patient
in our population:
CLtypical patient
=
𝜃CL
+
𝜃GFR
×
GFRmedian
+
𝜃RD×
(
RDmedian
845
−1
)
Population parameters Structural model Final model
Estimate RSE (%) Estimate RSE (%)
CL (L/h) 5.48 10.2 4.20aNA
V (L) 60.3 22.6 43.9 17.0
BSV_CL (% CV) 77.4 24.4 30.8 37.0
BSV_V (% CV) 80.5 55.3 87.6 31.2
Prop_Res_Err (% CV) 38.8 27.3 32.1 19.0
AB
Population predictions(mg/L)
0255075
0
50
100
Observations (mg/L)
Individual predictions (mg/L)
0255075
0
50
100
)L/gm(snoitavresbO
CWRES
0
2
-2
010 20 30
Time after dose (h)
1
-1
-3
3
C
Fig. 2 Goodness-of-fit plots for the final model. A Observations versus individual predictions (IPRED), B observations versus population pre-
dictions (PRED), C conditional weighted residuals (CWRES) versus time after dose

701
Population Pharmacokinetics and Pharmacodynamics of Meropenem in the ICU
was also observed between residual diuresis and meropenem
clearance.
Our study is representative of the important population
heterogeneity in ICU. However, normo-renal patients were
poorly represented in the study population (median value
for GFR was 49mL/min) explaining in part the slight dis-
crepancies in CL that could be observed when varying the
GFR value alone around the “RFS” threshold. As a result,
our model is less satisfying in estimating high CL patients.
Time after dose (h)
0510
0
25
50
75
100
Concentration (mg/L)
Fig. 3 Prediction-corrected visual predictive checks plot generated
from 1000 simulations. Blue areas represent the 95% confidence
intervals of the 97.5th percentile (up) and the 2.5th percentile (down)
of the predictions, middle gray area represents the 95% confidence
interval of the 50th percentile (median) of the predictions. Lines
97.5th percentile (up, dotted), 50th (middle solid line) and 2.5th
percentile (down, dotted) of the observations. Black dots represent
observed plasma concentrations
Table 4 Population parameter estimates for the final model and boot-
strap results
RSE relative standard error
RSE
(%)=
SE
Estimation
×
100
, CI confidence
interval, CL total body clearance, V apparent volume of distribution,
θCL, typical value for CL in the population, θGFR GFR covariate effect
on clearance, θDIA_C and θDIA_SC continuous and semi-continuous dial-
ysis covariates effect on CL, θRFS renal function status covariate effect
on CL, θRD residual diuresis covariate effect on CL, θV typical value
for V in the population, BSV_CL between subject variability associ-
ated with CL, BSV_V between subject variability associated with V,
Prop_Res_Err proportional residual error, CV coefficient of variation
CV(%)=√e
𝜔2
−1×100
Population parameters Estimate RSE (%) Bootstrap: median [95%
CI]
CL (L/h)
θCL 1.36 23.9 1.35 [0.55–2.10]
θGFR 0.058 17.8 0.056 [0.032–0.080]
θDIA_C 6.38 12.2 6.49 [5.19–7.83]
θRFS 13.9 15.5 13.9 [7.92–21.5]
θRD 0.60 31.7 0.64 [0.24–1.17]
θDIA_SC 11.0 20.6 10.9 [8.28–27.4]
V (L)
θV43.9 17.0 44.7 [30.6–72.4]
BSV_CL (% CV) 30.8 37.0 26.7 [11.0–40.2]
BSV_V (% CV) 87.6 31.2 81.6 [16.5–146.9]
Prop_Res_Err (%
CV)
32.1 19.0 31.7 [23.5–43.0]
Fig. 4 Individual clearance (CL) values according to their glomerular
filtration rate (GFRMDRD) and stratified by renal replacement therapy
techniques (RRT). Blue dots patients with continuous dialysis, yellow
dots patients with semi-continuous dialysis, black dots patients with-
out RRT
Fig. 5 Individual clearance (CL) values according to their states of
diuresis and stratified by renal replacement therapy techniques (RRT).
Blue dots patients with continuous dialysis, yellow dots patients with
semi-continuous dialysis, black dots patients without RRT

702 A.O’Jeanson et al.
One could ponder the pertinence of keeping in the
final model three covariates of similar information to
the CL: GFR, RFS, and RD. We would argue that each
covariate is bringing complementary additional infor-
mation on the impact of the kidney function to the CL.
Patient kidney function was an element particularly chal-
lenging to capture in our data. We favorably described
it through the three parts of a greater whole: GFR, RFS,
and RD.
In contrast to some published meropenem PopPK models
developed in comparable populations, our final model did
not find body size (weight, BMI, or BSA) to have a signifi-
cant relationship with V. In the literature, the relationship
between diuresis and clearance was highlighted twice in
critically ill patients [22, 26].
In line with meropenem being a dialyzable drug, our anal-
ysis identified dialysis as a significant covariate for mero-
penem’s clearance. For median values of GFR and diuresis,
without dialysis the typical value for clearance was 4.2 L/h,
during a session of semi-continuous dialysis 11 L/h, and
6.4L/h during a session of continuous dialysis. These clear-
ance values are comparable with other clinical studies such
as Braune etal. [22] (CL ~8L/h using a semi-continuous
dialysis) and Ulldemolins etal. [26] (CL ~4.8L/h using
continuous dialysis in patients with preserved diuresis). Typ-
ical values of clearance in patients using semi-continuous
Fig. 6 Probability of target
attainment (%) simulations of
a bolus-like (30 min) of mero-
penem stratified by residual
diuresis. Percentage of patients
for whom the pharmacokinetic
target is met (100% T > MIC):
red < 50% of patients, orange
50–90% of patients, green >
90% of patients. Anuric patient
residual diuresis of 50mL/24h,
oliguric 250mL/24h, preserved
750mL/24h
Dose
regimen
laudiseR)L/gm(CIM
diuresis
≤1 24810 16 20 32 45
99.4% 99.1% 97.8% 94% 90.8% 77.8% 64.6% 26.2%6.3% Anuric
1g q8h 99.3% 98.4% 96.7% 90.4% 86% 69.5% 57.4% 23.4% 5.5% Oliguric
99.1% 98.6% 96.4% 90% 85.1% 65.9% 48.9% 14.8% 2.9% Preserved
99.9% 99.6% 99.1% 98.3% 97.6% 92.7% 87.3% 59.2%28.6% Anuric
1g q6h 99.4% 99.4% 99.1% 96.9% 95.6% 90.5% 83.3% 51.8% 20.7%Oliguric
99.9% 99.8% 99% 96.1% 94.7% 85.2% 77% 41.7%14.8% Preserved
99.8% 99.4% 99.1% 97.8%97% 94% 90.8% 77.8%56.4% Anuric
2g q8h 99.7% 99.3% 98.4% 96.7% 95.6% 90.4% 86% 69.5% 49.6%Oliguric
99.7% 99.1% 98.6% 96.4% 94.8% 90% 85.1% 65.9% 40.8% Preserved
100% 100% 99.9% 99.4% 99.2% 98.1% 97.5% 92.1% 82.1% Anuric
2g q6h 99.8% 99.8% 99.7% 99.1% 98.6% 97.2% 95.5% 87.8% 73.9%Oliguric
99.9% 99.9% 99.7% 98.9% 98.7% 97% 95.4% 87.1%72.6% Preserved
Fig. 7 Probability of target
attainment (%) simulations of an
extended infusion (5 h) of mero-
penem stratified by residual
diuresis. Percentage of patients
for whom the pharmacokinetic
target is met (100% T > MIC):
red < 50% of patients, orange
50–90% of patients, green >
90% of patients. Anuric patient
residual diuresis of 50mL/24h,
oliguric 250mL/24h, preserved
750mL/24h
Dose
regimen
laudiseR)L/gm(CIM
diuresis
≤1 24810 16 20 32 45
100% 100% 99.9% 99% 98.1% 92.2% 83% 41.5%11% Anuric
1g q8h 99.9% 99.8% 99.8% 98.2% 96.7% 86.5% 75.3% 34.6% 9.1% Oliguric
99.9% 99.9% 99.7% 98.5% 96.5% 85.1% 71.6% 26% 5.3% Preserved
100% 100% 100% 100% 100% 100% 99.3% 82.7%46% Anuric
1g q6h 100% 100% 100% 99.9% 99.9% 99.3% 97.1% 75.6% 39.2% Oliguric
100% 100% 100% 100% 100% 99.6% 96.9% 71% 28.9% Preserved
100% 100% 100% 99.9% 99.6% 99% 98.1% 92.2% 75.5% Anuric
2g q8h 100% 99.9% 99.8% 99.8% 99.5% 98.2% 96.7% 86.5% 68.2% Oliguric
100% 99.9% 99.9% 99.7% 99.2% 98.5% 96.6% 85.1% 60.8% Preserved
100% 100% 100% 100% 100% 100% 100% 100% 97.9% Anuric
2g q6h 100% 100% 100% 100% 100% 99.9% 99.9% 99.3% 94.3% Oliguric
100% 100% 100% 100% 100% 100% 100%99.6% 94.2% Preserved
siseruidlaudiseR)L/gm(CIMnemigeresoD
≤1 2481016203245
100%100% 100% 100%100%100%100% 89% 50.9%Anuric
1g q6h 100% 100% 100% 100% 100% 100% 99.6% 87.3% 48.8% Oliguric
100%100% 100% 100%100%100% 99.7% 78.6% 35.6%Preserved
100%100% 100% 100%100%100%100%99.7% 92.3%Anuric
1g q4h 100% 100% 100% 100% 100% 100% 100%99.1% 89.9%Oliguric
100%100% 100% 100%100%100%100%98.4% 83.8%Preserved
Fig. 8 Probability of target attainment (%) simulations of a continu-
ous infusion (with unique 1g loading dose) of meropenem stratified
by residual diuresis. Percentage of patients for whom the pharmacoki-
netic target is met (100% T > MIC): red < 50% of patients, orange
50–90% of patients, green > 90% of patients. Anuric patient residual
diuresis of 50mL/24h, oliguric 250mL/24h, preserved 750mL/24h

703
Population Pharmacokinetics and Pharmacodynamics of Meropenem in the ICU
dialysis are higher than in patients using continuous dialy-
sis, which is consistent with the technical characteristics of
dialysis, the semi-continuous technique being considered to
be more intensive, with a higher dialysate fluid flow.
In fact, without a dedicated pharmacokinetic study, evalu-
ating and anticipating the impact of a dialysis session on a
drug’s concentration is a difficult task. The drug’s extracor-
poreal excretion is the sum of the following clearances: dial-
ysis, renal CL, and non-renal CL (hepatic and other ways).
One of these clearances becomes clinically significant once
its contribution to the total CL represents ≥25%. This means
that, if dialysis corresponds to more than 25% of total CL,
the dosing regimen needs to be adapted. Hence, one question
arises: how to adapt the dosing of the drug? One frequently
used possibility is to adapt the dose using creatinine total
clearance, which is a surrogate for the sum of extracorpor-
eal and renal (residual) clearances. Other more complex
methods exist, using either the anuric dose or the anuric
dosing interval [41]. Notwithstanding the method used, the
pharmacokinetic impact of a RRT session on an antibacte-
rial therapy should be adequately estimated to avoid under-
or over-exposure, sources of treatment failure. The PopPK
model presented here accounts for that impact and stresses
to clinicians the different criteria they should consider when
determining the optimal dosing regimen for a patient: (1) for
patients without dialysis, we propose to consider GFRMDRD
and residual diuresis when adapting the dose, and (2) for
patients using dialysis, the type of dialysis used will impact
on the choice of a dosing regimen.
The high number of significant covariates retained in the
final model strongly suggest a possible over-parameteriza-
tion. It would be of much interest to evaluate the predictive
performance of this PopPK model with new subjects. Cli-
nicians considering using our PopPK model in Bayesian-
feedback TDM settings should first go through this external
validation process. To that effect, we are sharing the NON-
MEM control file of our final model (Supplementary Mate-
rial). However, the important number of retained covariates
in the model highlight the necessity to take into account the
multiple clinical aspects of the critically ill patients, beside
creatinine clearance, when administering meropenem.
Above all, it again highlights the extensive pharmacokinetic
variability that exists among patients treated in one mutual
care unit. Because of the limited size population (notably for
RRT patients), we did not collect technical aspects of RRT
sessions (intensity, blood flow, type of membrane, etc.) in
order to estimate the Sieving coefficient.
We performed simulations to evaluate the impact of
residual diuresis on the type of administration of fixed
doses of meropenem, aiming for a pharmacokinetic/phar-
macodynamic target of 100% T > MIC more appropriate
in ICU patients [42, 43]. To equivalent doses, the continu-
ous infusion mode (with loading dose) allowed to attain the
pharmacokinetic/pharmacodynamic target for a larger num-
ber of patients. Nevertheless, for the treatment of susceptible
bacteria (MIC ≤ 2mg/L), differences of PTA between bolus-
like, extended, and continuous infusions are negligible. In
such cases, the patient’s need for liquid intake could overrule
the mode of administration (bolus or infusion). The largest
differences in PTA are found for the treatment of bacteria
with a higher resistance breakpoint (MIC > 10mg/L) in
patients with preserved diuresis. For example, with a dosing
regimen of 1g/6h and a MIC=32mg/L, PTAs of bolus-
like, extended, and continuous infusion were, respectively,
42%, 71%, and 79%, showing the capability of the extended
and continuous infusions to reach for very high pharmaco-
dynamic targets. As a whole, continuous infusion granted
more PTA when treating high MIC bacteria. However, the
use of the continuous infusion mode to attain high pharma-
codynamic objectives should be considered with the utmost
caution. For example, with a continuous infusion of 1g/4h
in anuric patients, half of the simulated population had a
meropenem concentration at steady-state maintained over
67.4mg/L. These are very high concentrations and could
lead to patients developing toxicity. Therefore, we strongly
advocate the use of Bayesian-feedback TDM to avoid too
low or too high exposures.
5 Conclusion
The introduction of effective antibacterial therapy in critical
ill patients is complex. It must consider the patient’s renal
function and also the global clinical state of the patient, as
the half-life time of meropenem in normo-renal ICU patients
is twice longer than in healthy volunteers [19]. In patient
monitoring, it is interesting for clinicians to dispose of
clinical elements and landmarks for adjusting the dosing.
These elements should be easily accessible in patient health
records, which is the case for the covariates found in our
model (GFR, residual diuresis, and dialysis information are
routinely gathered data in the ICU). While slightly over-
predicting meropenem concentrations, this model should be
evaluated with a more sizable dataset. Nevertheless, it once
again highlights the importance of considering the patient’s
overall condition (renal clearance, renal function, and dial-
ysis) and the pathogen’s characteristics (MIC target) dur-
ing the establishment of a patient’s dosing regimen (bolus,
extended infusion, loading dose, dosing interval reduction).
Supplementary Information The online version contains supplemen-
tary material available at https:// doi. org/ 10. 1007/ s13318- 021- 00709-w .
Acknowledgements We thank David Fabre from the department of
Modelisation & Simulation at Sanofi R&D (Montpellier) for the provi-
sion of NONMEM licence.

704 A.O’Jeanson et al.
Declarations
Funding No funding was received for either this study or for the prepa-
ration of this manuscript.
Conflict of interest None to declare
Ethical approval The study protocol was approved by the Ethics Com-
mittee 2019_IRB-MTP_03-01.
Consent to participate Not applicableas this was a non-interventional
retrospective study; patients were informed of their right of opposition
to data collection.
Consent for publication Not applicable.
Availability of data and material On demand (email: sonia.khier@
umontpellier.fr).
Code availability In supplementary file.
Author contributions AOJ collected majority of raw data, performed
the literature review and analyses, and wrote the original manuscript,
tables and figures. RL checked clinical raw data. CLS collected part
of clinical raw data. ND challenged pharmacokinetic analyses and
reviewed and revised the manuscript. SK supervised the work, designed
and planned the work, reviewed the analyses, and wrote the manuscript,
tables and figures.
References
1. Vincent J-L, Rello J, Marshall J, Silva E, Anzueto A, Martin CD,
etal. International study of the prevalence and outcomes of infec-
tion in intensive care units. JAMA. 2009;302:2323–9.
2. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo
J, Pinsky MR. Epidemiology of severe sepsis in the United States:
analysis of incidence, outcome, and associated costs of care. Crit
Care Med. 2001;29:1303–10.
3. Shankar-Hari M, Harrison DA, Rubenfeld GD, Rowan K. Epide-
miology of sepsis and septic shock in critical care units: compari-
son between sepsis-2 and sepsis-3 populations using a national
critical care database. Br J Anaesth. 2017;119:626–36.
4. Sakr Y, Jaschinski U, Wittebole X, Szakmany T, Lipman J,
Ñamendys-Silva SA, etal. Sepsis in intensive care unit patients:
worldwide data from the intensive care over nations audit. Open
Forum Infect Dis [Internet]. 2018 [cited 2020 Feb 10];5. https://
acade mic. oup. com/ ofid/ artic le/5/ 12/ ofy313/ 51931 71.
5. World Health Organization. Antimicrobial resistance: global
report on surveillance. Geneva: World Health Organization; 2014.
6. Kinch MS, Patridge E, Plummer M, Hoyer D. An analysis of FDA-
approved drugs for infectious disease: antibacterial agents. Drug
Discov Today. 2014;19:1283–7.
7. Cassini A, Högberg LD, Plachouras D, Quattrocchi A, Hoxha A,
Simonsen GS, etal. Attributable deaths and disability-adjusted
life-years caused by infections with antibiotic-resistant bacteria in
the EU and the European Economic Area in 2015: a population-
level modelling analysis. Lancet Infect Dis. 2019;19:56–66.
8. Stemming the Superbug Tide—Just A Few Dollars More—en—
OECD [Internet]. [cited 2019 Nov 22]. https:// www. oecd. org/
health/ stemm ing- the- super bug- tide- 97892 64307 599- en. htm.
9. Roberts JA, Abdul-Aziz MH, Lipman J, Mouton JW, Vinks AA,
Felton TW, etal. Individualised antibiotic dosing for patients who
are critically ill: challenges and potential solutions. Lancet Infect
Dis. 2014;14:498–509.
10. Blot SI, Pea F, Lipman J. The effect of pathophysiology on phar-
macokinetics in the critically ill patient — Concepts appraised
by the example of antimicrobial agents. Adv Drug Deliv Rev.
2014;77:3–11.
11. Udy AA, Roberts JA, Lipman J. Implications of augmented
renal clearance in critically ill patients. Nat Rev Nephrol.
2011;7:539–43.
12. Udy AA, Baptista JP, Lim NL, Joynt GM, Jarrett P, Wockner L,
etal. Augmented renal clearance in the ICU: results of a multi-
center observational study of renal function in critically ill patients
with normal plasma creatinine concentrations*. Crit Care Med.
2014;42:520–7.
13. Wong G, Brinkman A, Benefield RJ, Carlier M, De Waele JJ, El
Helali N, etal. An international, multicentre survey of β-lactam
antibiotic therapeutic drug monitoring practice in intensive care
units. J Antimicrob Chemother. 2014;69:1416–23.
14. Mohr JF, Wanger A, Rex JH. Pharmacokinetic/pharmacodynamic
modeling can help guide targeted antimicrobial therapy for noso-
comial gram-negative infections in critically ill patients. Diagn
Microbiol Infect Dis. 2004;48:125–30.
15. Gonçalves-Pereira J, Póvoa P. Antibiotics in critically ill patients:
a systematic review of the pharmacokinetics of β-lactams. Crit
Care. 2011;15:R206.
16. Levy Hara G, Kanj SS, Pagani L, Abbo L, Endimiani A, Wertheim
HFL, etal. Ten key points for the appropriate use of antibiotics in
hospitalised patients: a consensus from the Antimicrobial Stew-
ardship and Resistance Working Groups of the International Soci-
ety of Chemotherapy. Int J Antimicrob Agents. 2016;48:239–46.
17. de With K, Allerberger F, Amann S, Apfalter P, Brodt H-R, Eck-
manns T, etal. Strategies to enhance rational use of antibiotics in
hospital: a guideline by the German Society for Infectious Dis-
eases. Infection. 2016;44:395–439.
18. WHO | Global action plan on antimicrobial resistance [Internet].
WHO. [cited 2020 Feb 7]. http:// www . who. int / antim icr ob ial- r esis
tance/ publi catio ns/ global- action- plan/ en/.
19. Summary of products characteristics: meropenem STADA powder
for solution for injection or infusion. https:// docetp. mpa. se/ LMF/
Merop enem% 20STA DA% 20pow der% 20for% 20sol ution% 20for%
20inj ection% 20or% 20inf usion ,% 20ENG% 20SmPC_ 09001 be682
a8f1bd. pdf. Accessed 3 July 2021.
20. Legrand T, Vodovar D, Tournier N, Khoudour N, Hulin A. Simul-
taneous determination of eight β-lactam antibiotics, amoxicillin,
cefazolin, cefepime, cefotaxime, ceftazidime, cloxacillin, oxa-
cillin, and piperacillin, in human plasma by using ultra-high-
performance liquid chromatography with ultraviolet detection.
Antimicrob Agents Chemother. 2016;60:4734–42.
21. Guideline on bioanalytical method validation. EMEA/CHMP/
EWP/192217/2009 Rev. 1 Corr. 2. https:// www. ema. europa. eu/ en/
docum ents/ scien tific- guide line/ guide line- bioan alyti cal- method-
valid ation_ en. pdf.
22. Braune S, König C, Roberts JA, Nierhaus A, Steinmetz O, Baehr
M, etal. Pharmacokinetics of meropenem in septic patients on
sustained low-efficiency dialysis: a population pharmacokinetic
study. Crit Care. 2018;22:25.
23. Jamal J-A, Mat-Nor M-B, Mohamad-Nor F-S, Udy AA, Wallis
SC, Lipman J, etal. Pharmacokinetics of meropenem in critically
ill patients receiving continuous venovenous haemofiltration: a
randomised controlled trial of continuous infusion versus intermit-
tent bolus administration. Int J Antimicrob Agents. 2015;45:41–5.
24. Giles LJ, Jennings AC, Thomson AH, Creed G, Beale RJ,
McLuckie A. Pharmacokinetics of meropenem in intensive care
unit patients receiving continuous veno-venous hemofiltration or
hemodiafiltration. Crit Care Med. 2000;28:632–7.

705
Population Pharmacokinetics and Pharmacodynamics of Meropenem in the ICU
25. Krueger WA, Schroeder TH, Hutchison M, Hoffmann E, Diet-
erich HJ, Heininger A, etal. Pharmacokinetics of meropenem
in critically ill patients with acute renal failure treated by con-
tinuous hemodiafiltration. Antimicrob Agents Chemother.
1998;42:2421–4.
26. Ulldemolins M, Soy D, Llaurado-Serra M, Vaquer S, Castro P,
Rodríguez AH, etal. Meropenem population pharmacokinetics
in critically ill patients with septic shock and continuous renal
replacement therapy: influence of residual diuresis on dose
requirements. Antimicrob Agents Chemother. 2015;59:5520–8.
27. Dhaese SAM, Farkas A, Colin P, Lipman J, Stove V, Verstraete
AG, etal. Population pharmacokinetics and evaluation of the pre-
dictive performance of pharmacokinetic models in critically ill
patients receiving continuous infusion meropenem: a compari-
son of eight pharmacokinetic models. J Antimicrob Chemother.
2019;74:432–41.
28. Tsai D, Stewart P, Goud R, Gourley S, Hewagama S, Krishnas-
wamy S, etal. Optimising meropenem dosing in critically ill Aus-
tralian indigenous patients with severe sepsis. Int J Antimicrob
Agents. 2016;48:542–6.
29. Mathew SK, Mathew BS, Neely MN, Naik GS, Prabha R, Jacob
GG, etal. A nonparametric pharmacokinetic approach to deter-
mine the optimal dosing regimen for 30-minute and 3-hour
meropenem infusions in critically ill patients. Ther Drug Monit.
2016;38:593–9.
30. Roberts JA, Kirkpatrick CMJ, Roberts MS, Robertson TA, Dalley
AJ, Lipman J. Meropenem dosing in critically ill patients with
sepsis and without renal dysfunction: intermittent bolus versus
continuous administration? Monte Carlo dosing simulations
and subcutaneous tissue distribution. J Antimicrob Chemother.
2009;64:142–50.
31. Mattioli F, Fucile C, Del Bono V, Marini V, Parisini A, Molin A,
etal. Population pharmacokinetics and probability of target attain-
ment of meropenem in critically ill patients. Eur J Clin Pharmacol.
2016;72:839–48.
32. Alobaid AS, Wallis SC, Jarrett P, Starr T, Stuart J, Lassig-Smith
M, etal. Effect of obesity on the population pharmacokinetics of
meropenem in critically ill patients. Antimicrob Agents Chem-
other. 2016;60:4577–84.
33. Crandon JL, Ariano RE, Zelenitsky SA, Nicasio AM, Kuti JL,
Nicolau DP. Optimization of meropenem dosage in the criti-
cally ill population based on renal function. Intensive Care Med.
2011;37:632–8.
34. Jaruratanasirikul S, Thengyai S, Wongpoowarak W, Wattanavijit-
kul T, Tangkitwanitjaroen K, Sukarnjanaset W, etal. Population
pharmacokinetics and Monte Carlo dosing simulations of mero-
penem during the early phase of severe sepsis and septic shock in
critically ill patients in intensive care units. Antimicrob Agents
Chemother. 2015;59:2995–3001.
35. Isla A, Rodríguez-Gascón A, Trocóniz IF, Bueno L, Solinís MA,
Maynar J, etal. Population pharmacokinetics of meropenem in
critically ill patients undergoing continuous renal replacement
therapy. Clin Pharmacokinet. 2008;47:173–80.
36. Jonsson EN, Karlsson MO. Automated covariate model building
within NONMEM. Pharm Res. 1998;15:1463–8.
37. Sheiner LB, Beal SL. Some suggestions for measuring predictive
performance. J Pharmacokinet Biopharm. 1981;9:503–12.
38. Chan Kwong A, O’Jeanson A, Khier S. Model-informed thera-
peutic drug monitoring of meropenem in critically ill patients:
improvement of the predictive ability of literature models
with the PRIOR approach. Eur J Drug Metab Pharmacokinet.
2021;46:415–26.
39. Chan-Kwong AH-XP, Calvier EAM, Fabre D, Gattacceca F, Khier
S. Prior information for population pharmacokinetic and pharma-
cokinetic/pharmacodynamic analysis: overview and guidance with
a focus on the NONMEM PRIOR subroutine. J Pharmacokinet
Pharmacodyn. 2020;47:431–46.
40. Sunder S, Jayaraman R, Mahapatra HS, Sathi S, Ramanan V,
Kanchi P, etal. Estimation of renal function in the intensive care
unit: the covert concepts brought to light. J Intensive Care [Inter-
net]. 2014 [cited 2020 Oct 26];2. https:// www. ncbi. nlm. nih. gov/
pmc/ artic les/ PMC42 67588/.
41. Reetze-Bonorden P, Böhler J, Keller E. Drug dosage in patients
during continuous renal replacement therapy. Clin Pharmacokinet.
1993;24:362–79.
42. Li C, Du X, Kuti JL, Nicolau DP. Clinical pharmacodynamics
of meropenem in patients with lower respiratory tract infections.
Antimicrob Agents Chemother. 2007;51:1725–30.
43. Roberts JA, Paul SK, Akova M, Bassetti M, De Waele JJ, Dimo-
poulos G, etal. DALI: defining antibiotic levels in intensive care
unit patients: are current β-lactam antibiotic doses sufficient for
critically ill patients? Clin Infect Dis. 2014;58:1072–83.
Authors and Aliations
AmauryO’Jeanson1,2· RomaricLarcher3· CosetteLeSouder4· NassimDjebli5· SoniaKhier1,2
1 Pharmacokinetic Modeling Department, UFR Pharmacie,
Montpellier University (School ofPharmacy), 15 Avenue
Charles Flahault, 34000Montpellier, France
2 Probabilities andStatistics Department, Institut
Montpelliérain Alexander Grothendieck (IMAG), CNRS
UMR 5149, Montpellier University, Montpellier, France
3 Intensive Care Unit Department, Montpellier University
Hospital (CHU Lapeyronie), Montpellier, France
4 Toxicology andTarget Drug Monitoring Department,
Montpellier University Hospital (CHU Lapeyronie),
Montpellier, France
5 Roche Innovation Center Basel, Roche Pharma Research
andEarly Development, Basel, Switzerland