Received: 12 December 2022 | Revised: 4 May 2023 | Accepted: 11 May 2023

DOI: 10.1002/phar.2858

REVIEW OF THERAPEUTICS

Practical considerations for individualizing drug dosing in

critically ill adults receiving renal replacement therapy

Salmaan Kanji1 | Claire Roger2,3 | Fabio Silvio Taccone4 | Laurent Muller2,3

1
The Ottawa Hospital and Ottawa
Hospital Research Institute, Ottawa, Abstract
Ontario, Canada
Critically ill patients with sepsis admitted to the intensive care unit (ICU) often present
2
Department of Anaesthesiology and
Intensive Care, Pain and Emergency
with or develop renal dysfunction requiring renal replacement therapy (RRT) in addi-
Medicine, Nîmes University Hospital, tion to antimicrobial therapy. While early and appropriate antimicrobials for sepsis
Nîmes, France
3
have been associated with an increased probability of survival, adequate dosing is
UR UM 103 IMAGINE, Faculty of
Medicine, University of Montpellier, also required in these patients. Adequate dosing of antimicrobials refers to dosing
Nîmes, France strategies that achieve serum drug levels at the site of infection that are able to pro-
4
Department of Intensive Care, Hôpital
vide a microbiological and/or clinical response while avoiding toxicity from excessive
Universitaire de Bruxelles (HUB),
Université Libre de Bruxelles (ULB), antibiotic exposure. Therapeutic drug monitoring (TDM) is the recommended strat-
Brussels, Belgium
egy to achieve this goal, however, TDM is not routinely available in all ICUs and for
Correspondence all antimicrobials. In the absence of TDM, clinicians are therefore required to make
Salmaan Kanji, The Ottawa Hospital, 501
dosing decisions based on the clinical condition of the patient, the causative organism,
Smyth Rd, Ottawa K1H 8L6, ON, Canada.
Email: [email protected] the characteristics of RRT, and an understanding of the physicochemical properties
of the antimicrobial. Pharmacokinetics (PK) of antimicrobials can be highly variable
between critically ill patients and also within the same patient over the course of their
ICU stay. The initiation of RRT, which can be in the form of intermittent hemodialysis,
continuous, or prolonged intermittent therapy, further complicates the predictability
of drug disposition. This variability highlights the need for individualized dosing. This
review highlights the practical considerations for the clinician for antimicrobial dosing
in critically ill patients receiving RRT.

KEYWORDS
critical illness, drug, renal replacement therapy, therapeutic drug monitoring

1 | I NTRO D U C TI O N therapies) associated with critical illness alter the pharmacokinetics

(PK) of these drugs. 2 These changes mostly affect the PK of hydro-
Patients suffering from life-­threatening conditions who are admit- philic antibiotics, that is the increase in the volume of distribution
ted to the intensive care unit (ICU) often require antibiotic therapy, (Vd) and clearance, which often lead to lower-­than-­expected drug
either as infection is the main reason for their acute condition or concentrations.3,4 Third spacing and hypo-­albuminemia, which are
because they suffer from a hospital-­acquired infection during their common in critical illness, may influence the proportion of protein-­
hospital stay.1 In this setting, antibiotic regimens are similar to those bound and unbound active drug.5 Renal failure will reduce drug
administered in patients with less severe conditions; however, dy- clearance and can also contribute to increased Vd due to fluid over-
namic pathophysiologic changes (i.e., increased cardiac output, fluid load6; higher than expected drug concentrations might lead to ad-
overload and third spacing, capillary leakage, altered renal and he- verse events, such as renal injury (e.g., aminoglycosides, colistin,
patic function, decreased protein synthesis, and extra-­
corporeal and vancomycin) or neurotoxicity (e.g., colistin and beta-­
lactam

1194 | © 2023 Pharmacotherapy Publications, Inc. wileyonlinelibrary.com/journal/phar Pharmacotherapy. 2023;43:1194–1205.

KANJI et al. | 1195

antibiotics).7,8 Moreover, the introduction of extracorporeal thera- long-­term renal recovery and CKD reduction, although this remains
pies, such as renal replacement therapy (RRT), will add further com- controversial18,22–­24 Nevertheless, no difference of mortality was
9
plexity in predicting antimicrobial PKs. As such, the choice of the reported between the three modalities.25,26 The training of the ICU
optimal empiric antimicrobial regimen in these patients is a complex team for a given technique is more important than the technique it-
clinical challenge. self in terms of safety, stability, and efficiency.23,27,28 Whatever the
Large variability in antibiotic concentration is therefore ob- modality, RRT mainly uses two principles: diffusion and convection.
served within and between critically ill patients on RRT. In septic Diffusion occurs when a semi-­permeable membrane separates
patients, standard doses of broad-­spectrum beta-­lactams result in two aqueous solutions. Small molecules cross the membrane from
a high proportion of patients with insufficient plasma drug levels, in a higher to a lower concentration compartment. When diffusive
10
particular during the early phase of therapy. In a large prospective, principles are applied, water elimination is negligible. For a given
observational, multinational PK study, there was wide variability membrane, diffusion efficacy depends on temperature, mass trans-
(4–­8-­fold) in antibiotic dosing regimens and RRT prescriptions11; me- fer coefficient, membrane surface area, concentration gradient be-
dian trough drug concentrations was largely above recommended tween blood and dialysate compartments, and membrane thickness.
targets for drug efficacy and 25%–­35% of treated patients had ex- The mass transfer coefficient (also named the diffusion coefficient)
cessive concentrations (i.e., at-­risk for potential toxicity). In another depends on the effective radius of the considered molecule, viscos-
study, 53% of samples for broad-­spectrum beta-­lactams given in ity of the medium, and the Boltzmann constant, which represent the
patients on RRT also showed excessive plasma drug levels when un- entropic kinetic energy of the considered molecules. Other factors
adjusted drug regimens (i.e., similar to those used in patients with may influence solute transport, such as protein binding and electri-
normal renal function) were given.12 Therapeutic drug monitoring cal charges of each solute.
(TDM) of beta-­lactams also identified a need for dose modification Convection induces a movement of plasma water from the blood
of beta-­lactam antibiotics in a third of ICU patients, mostly resulting to ultrafiltrate across the membrane, driven by a gradient of hydro-
in decreasing the prescribed drug regimen.13 static pressure, the transmembrane pressure (TMP). The TMP is the
The consequences of inadequate dosing in critically ill patients difference between blood hydrostatic pressure and ultrafiltrate com-
treated with RRT remain to be further evaluated from both an effi- partment hydrostatic pressure, minus the plasma oncotic pressure.
cacy and safety perspective.11,14–­16 Because of the high PK variabil- Convective efficiency also depends on the intrinsic water permeabil-
ity in critically ill patients, an individualized approach to antimicrobial ity of the hemofilter membrane, the membrane surface area, and the
dosing is warranted in this population. TDM is an acceptable strategy sieving coefficient (SC). The SC represents the ability of a convective
to refine drug dosing according to individual patient's needs, but it is therapy to remove solutes, regardless of diffusive transport. The SC
not universally available for most antimicrobials. Furthermore, TDM is the ratio of concentrations of a given solute between ultrafiltrate
will not inform initial empiric drug dosing which has been shown to and the blood compartment. Convection is the only way to remove
influence important outcomes such as mortality in septic shock.17 plasma water during RRT, regardless of the modality used. The fluid
This leaves the clinician with the difficult task of trying to predict transfer across the membrane during convection is called ultrafil-
the outcomes of drug disposition and the probability of adequate trate. The total amount of fluid obtained in the waste bag of the RRT
drug concentrations at the site of infection based on the net balance machine is called effluent. Effluent is the sum of ultrafiltrate and di-
of physiologic characteristics of the patient and their disease along alysate. According to the technique, effluent can be plasma water
with physicochemical properties of the drug and characteristics of (convective therapy alone, continuous veno-­venous hemofiltration
extracorporeal treatments. This review highlights the practical con- [CVVH]), dialysate (continuous veno-­venous hemodialysis [CVVHD],
siderations for the clinician for antimicrobial dosing in critically ill intermittent hemodialysis [IHD], prolonged intermittent renal re-
patients receiving RRT. placement therapy [PIRRT]), or both (IHD with water removal, con-
tinuous veno-­venous hemodiafiltration [CVVHDF]).

2 | R E N A L R E PL AC E M E NT TH E R A PY
2.1 | Practical differences between CRRT,
More than 50% of patients hospitalized in ICUs suffer from acute IHD and PIRRT
kidney injury (AKI), and 20%–­25% of these patients will need RRT
during the first week.18 This syndrome is associated with about 50% 2.1.1 | Continuous renal replacement therapy
mortality in one-­year, significantly increased ICU or hospital stay,
and secondary onset of chronic kidney disease (CKD).18–­21 In the Three CRRT techniques are available in the ICU: CVVH (convec-
ICU, three main RRT modalities are used: continuous RRT (CRRT), tive therapy), CVVHD (diffusive therapy), and CVVHDF (using both
intermittent conventional hemodialysis (IHD), and prolonged inter- convective and diffusive therapy at comparable flow rate; Table 1).
mittent renal replacement therapy (PIRRT). In critically ill patients, The recommended CRRT dose in the ICU for AKI management is 25–­
CRRT is used in 60%–­80% of cases because of a better hemodynamic 30 mL/kg/h of the effluent rate. 23 In clinical practice, convection can
tolerance, better fluid balance control, and a potential benefit on be used alone to perform RRT (CVVH 30 mL/kg/h) or in addition to
1196 | KANJI et al.

TA B L E 1 Overview of technical and clinical characteristics of RRT techniques used in the ICU.

Intermittent Continuous Hybrid –­PIRRT

IHD CVVH CVVHDF CVVHD SLED, EDD

Duration (h) 4 >24 >24 >24 5–­16

Blood Flow rate (mL/min) 200–­350 100–­250 100–­250 100–­250 100–­3 00
Diffusion rate ++++ − ++ (50%) ++++ +++
Convection rate + (if fluid loss needed) ++++ ++ (50%) + (if fluid loss + (if fluid loss needed)
needed)
Effluent Flow ratea (Volume 18–­50L/h 30mL/Kg/h 30mL/Kg/hb 30mL/Kg/h 6–­18L/h
b
per hour) 15 mL/Kg/h
diffusion,
15mL/Kg/h
convection
Urea clearance (mL/min) 150–­200 20–­45 20–­45 20–­45 90–­140
Antibiotics clearance, ++ ++++ +++++ ++++ +++ if conventional
combining technique diffusive therapy
and session duration ++++ if online
hemodiafiltration
Filtration ratio (%)b <5 20–­25 15–­20 <5 <5
ICP variations ++++ _ -­ -­ ++
Hemodynamic stability -­ ++++ ++++ ++++ ++
Fluid balance control over + ++++ ++++ ++++ +++
time
Patient's blood pH stability + ++++ ++++ ++++ +
over time
Need for anticoagulation + +++ ++ ++ +
Hemodialyzer membrane Synthetic or modified Synthetic Synthetic Synthetic Synthetic
cellulosic

Note: ++++, Major; +++, Important; ++, Moderate; +, Low; −, Absent.

Abbreviations: CVVH, continuous veno-­venous hemofiltration; CVVHD, continuous veno-­venous hemodialysis; CVVHDF, continuous veno-­venous
hemodiafiltration; EDD, extended daily dialysis; ICP, intra cranial pressure; ICU, intensive care unit; IHD, intermittent hemodialysis; PIRRT, prolonged
intermittent renal replacement therapy; RRT, renal replacement therapy; SLED, sustained low-­efficiency dialysis.
a
Effluent flow rate represents the global volume of treatment/RRT dose.
b
Filtration ratio is the ratio between convection and blood flow rate. It monitors the hemoconcentration rate across filters and the associated risk of
early filter loss. To avoid this risk, the filtration ratio value must be kept below 25%.

diffusion at comparable effluent rates (CVVHDF, 15 mL/kg/h dif- convective techniques has declined over time despite wide practice
fusion + 15 mL/kg/h convection). In the two cases, the amount of variations between countries. 29–­31
plasma water loss exposes the patient to the risk of critical hypov-
olemia and requires total (zero balance) or partial volume compen-
sation by crystalloids before or after filter (pre-­or post-­dilution). 2.1.2 | Conventional intermittent hemodialysis
During CVVHD or IHD (pure diffusive technique), a small amount (IHD)
of convection must be added if fluid depletion is desired (50–­
400 mL/h), as diffusion alone cannot allow water transfer across the IHD mainly uses diffusive techniques, which allow for large solute
membrane. Convection allows the removal of larger molecules than removal. Convective techniques are only used for controlling fluid
with diffusive therapy, with a potential to also remove other sub- overload. The hemodynamic instability and the poor ability to con-
stances such as inflammatory molecules. This theoretical issue has trol fluid balance observed with IHD explain its lower use in ICU,
not been confirmed, even using a high RRT dose.12 Moreover, the use despite the absence of mortality differences in randomized trials.
of convective therapy alone can lead to high filtration ratio values The advantage of IHD is that metabolic control can be achieved
(>20%–­25%), which increases hemoconcentration over filter, risking after a 4-­h session, significantly faster than with CRRT. In trained
early filter loss. The filtration ratio is the ratio of the convective rate ICU teams, IHD can be safely used but requires a higher skill than
divided by blood flow. Adding diffusion to convection significantly with CRRT. 27,28 Typically, ICUs use both CRRT and IHD: CRRT at the
increases filter life by reducing the filtration ratio independently initial hemodynamically unstable phase, and IHD for RRT weaning at
29
from the anticoagulation technique. Therefore, the use of pure the hemodynamically stable recovery phase. 22
KANJI et al. | 1197

2.1.3 | Prolonged intermittent renal 3 | A NTI M I C RO B I A L D OS I N G W ITH

replacement therapy (PIRRT) R E N A L R E PL AC E M E NT TH E R A PY

These techniques use IHD machines with different settings as com- Antimicrobial drug dosing in critically ill patients receiving RRT re-
pared with conventional IHD: lower blood flow rate, lower dialysis mains highly challenging. Variability in drug disposition related to
rate, and longer duration (Table 1). They can be used as intermediate patient and pathophysiologic characteristics, drug physicochemical
techniques between the early unstable phase (where CRRT is used) properties, and RRT practices result in large heterogeneity in RRT-­
and the recovery period (where IHD is used), or during the early adjusted antimicrobial drug dosing.11 Providing dosing guidelines for
unstable phase in place of CRRT. The indications for PIRRT remain individual drugs that can be applied to all patients in all scenarios is
largely dependent on ICU preference. PIRRT is often referred to therefore illusory, but the clinician should understand the general
as SLED (sustained low-­efficiency dialysis), SLEDD (sustained low-­ principles that reduce the risk of improper drug dosing during RRT.
efficiency daily dialysis), or EDD (extended daily dialysis). Similar to The different factors affecting antimicrobial PK in patients receiving
CRRT modalities, PIRRT can employ both diffusive and/or convective RRT are summarized in Figure 1.
techniques in sustained low-­efficiency daily diafiltration (SLEDDf),
extended daily diafiltration (EDDf), and extended daily hemofiltra-
tion (EDH). The influence of PIRRT on prognosis relative to other 3.1 | Factors related to patient and
modes of RRT is not demonstrated. 26,32,33 pathophysiologic characteristics

Several patient-­related factors affect antimicrobial PK. Increased

2.2 | Differences in drug clearance between Vd is one of the key pathophysiological alterations observed in
different RRT modalities and settings critically ill patients and is correlated with illness severity.35,36
Hypoalbuminemia is common in this population and mainly af-
The duration of the RRT session influences antibiotic removal. fects high protein-­bound drugs (e.g., echinocandins, cefazolin, cef-
Continuous or prolonged techniques (CRRT or PIRRT) can lead to triaxone, flucloxacillin, teicoplanin, or antifungal agents such as
higher antibiotic elimination as compared with IHD. Among continu- isavuconazole).5,37 Reduced protein binding leads to higher free
ous or prolonged techniques, if a similar blood flow rate is utilized, drug concentrations, increased clearance, and greater Vd for highly
the amount of antibiotics eliminated by RRT can be summarized as protein-­bound drugs. To illustrate, ceftriaxone (95% bound to al-
follows: CVVHDF > CVVHD > CVVH > PIRRT > IHD.34 bumin) has a 100% increased clearance and 90% larger volume of

F I G U R E 1 Factors affecting antimicrobial pharmacokinetics in patients receiving renal replacement therapy. AUC, area under the curve;
Cmax, maximal plasma concentration; MIC, minimal inhibitory concentration; PD, pharmacodynamic; RRT, renal replacement therapy.
1198 | KANJI et al.

distribution in critically ill patients compared with non-­critically ill continuous infusion of 11.25 g piperacillin/tazobactam per day, Jamal
38
patients. Similarly, flucloxacillin Vd in critically ill patients exhib- and colleagues demonstrated that in the continuous infusion group,
iting hypoalbuminemia was twofold greater compared with hospi- the proportion of patients on CVVH achieving therapeutic concen-
talized patients, but no increased clearance was observed.39 For trations was higher (87.5% vs. 62.5%) than the other.46
specific drugs (such as aminoglycosides), hypoalbuminemia may re-
quire adjusting antimicrobial drug dosing.40 For patients receiving
RRT, it is commonly accepted that only the free fraction of the drug 3.2.2 | Minimal inhibitory concentration (MIC)
is removed by the hemofilter. As such, highly protein-­bound anti-
microbials may not significantly be affected by RRT.37 However, as MIC of the causative pathogen is a key determinant of antimicro-
wide variations in protein binding occur in critically ill patients, this bial drug dosing adjustments based on PK/PD principles. Data from
can further increase RRT-­related clearance and potentially decrease epidemiological studies confirm that MICs for bacteria isolated from
antibiotic exposure.39 critically ill patients are commonly twofold greater compared with
Residual renal function may significantly contribute to antimicro- patients in wards. Considering that the MIC is the denominator for
bial clearance and requires consideration when determining antimi- each PK/PD targets for the different classes of antibiotics, as this
crobial dosing. Studies assessing the impact of residual renal function denominator increases, the PK exposure must increase accordingly
on antimicrobial exposure in patients undergoing RRT are limited. to ensure an optimal PK/PD ratio. Consequently, the rate of target
Two studies, one on linezolid and one on meropenem, suggest that attainment for comparable dosing regimens of broad-­
spectrum
increased dosing for concentration-­dependent antibiotics and ex- beta-­lactam antibiotics may vary based on local resistance epidemi-
tended infusion for time-­dependent antibiotics should be used in pa- ology and must be considered to determine optimal dosing.
tients with preserved diuresis (>500 mL/day) compared with anuric
patients to achieve adequate PK/PD targets.45,46
RRT affects antimicrobial drug removal but may also alter non-­ 3.3 | Factors related to drug properties
renal clearance of the drugs. As an example, RRT may lead to an
increase in CYP3A4 activity inversely correlated to plasma urea 3.3.1 | Hydrophilic versus lipophilic
concentrations.41 However, given the multiple pathways that are antimicrobial drugs
involved in drug metabolism and clearance, it is difficult to predict
nonrenal clearance changes in critical illness for the purpose of ad- Pharmacological and physicochemical knowledge must be inte-
justing drug doses without measuring serum drug levels. grated to identify drug-­specific properties that can be used to
guide drug dosing on RRT. Hydrophilic drugs (e.g., aminoglyco-
sides, beta-­lactams, vancomycin) do not readily distribute into
3.2 | Pharmacodynamic target tissues and across membranes and exhibit a low volume of dis-
tribution. Their main elimination route is represented by renal
3.2.1 | Concentration versus time-­dependent elimination. As the loading dose of a drug mainly depends on the
antimicrobial drugs volume of distribution, situations, where the volume of distribu-
tion of hydrophilic drugs is increased because of capillary leak,
The impact of RRT on drug dosing depends on the bacterial-­killing massive fluid resuscitation, and hypoalbuminemia, may require
characteristics of antimicrobials. For concentration-­dependent a higher loading dose.42,47 Consequently, the presence of AKI or
antibiotics, optimal killing is associated with maximal plasma con- RRT may not require decreasing the loading dose. However, AKI
centration (Cmax)/minimum inhibitory concentration (MIC) ratio or RRT may significantly affect hydrophilic antibiotics clearance
(e.g., aminoglycosides), the ratio of the area under the plasma whereas lipophilic drugs such as macrolides, quinolones, and echi-
concentration–­time curve during a 24-­h period (AUC24) to MIC nocandins that exhibit large apparent Vd and predominant hepatic
(AUC24/MIC) (e.g., daptomycin), or both (e.g., fluoroquinolones). For elimination are less likely to be affected by RRT because a low
aminoglycosides, high-­loading doses (25–­30 mg/kg per adjusted proportion of the drug is present in the bloodstream from where
body weight for amikacin, 8–­10 mg/kg for gentamicin or tobramycin) clearance occurs. Nonetheless, other mechanisms of elimination
with extended interval (48-­hourly) increases the probability of tar- may still alter lipophilic PK during RRT.
42,43
get attainment while minimizing toxicity.
For time-­dependent antibiotics, killing is related to the time
plasma concentrations remain above a threshold (100%fT > 1-­4MIC 3.3.2 | Molecular weight
for severe infections). To maximize efficacy, continuous or extended
infusion of beta-­lactams may be the optimal mode of administra- The molecular weight (MW) of drugs plays an important role es-
tion.44 In a recent study, Sember and colleagues demonstrated that pecially in diffusive techniques, as the diffusion of a molecule is
a dosing regimen of an extended 4-­h infusion of ceftazidime and inversely proportional to MW. The saturation coefficient (Sd) ex-
cefepime for patients receiving CVVH at 20–­30 mL/kg/h achieved presses the ability of a drug to diffuse through the filter membrane.
45
adequate PK/PD targets. In a study comparing intermittent versus Sd is strictly dependent on MW and tends to decrease progressively
KANJI et al. | 1199

as MW increases. Most antimicrobials have a MW below 1000 Da, flow rate (35 mL/kg/h) using similar polyacrylonitrile filters induced
with few molecules around 1500 Da (e.g., vancomycin and teico- comparable little and middle-­size solute removal.50 On the contrary,
planin). Conversely, as commonly used modern membranes exhibit at equal CRRT dose (30 mL/kg/h), combining convective to diffusive
pore sizes around 10,000–­30,000 Da, convective transport across techniques (CVVHDF) leads to greater removal of linezolid, amika-
modern membranes is independent of molecular weight in hemofil- cin, and piperacillin–­t azobactam compared with convection or diffu-
tration techniques and more effective than diffusive transport for sion alone.42,51,52
drugs with relatively large molecular weights such as vancomycin,
teicoplanin, colistin.
3.4.3 | Renal replacement therapy dose

3.3.3 | Electric charge Different CRRT doses are commonly applied in ICU patients making
it difficult to provide dosing recommendations based on RRT mo-
The electric charges of the hemodialysis membrane may interact dalities alone.11 Several ex vivo and clinical PK studies have demon-
with some drugs. The Gibbs-­Donnan effect refers to the effect of strated the key role of effluent flow rate in adjusting antimicrobial
anionic protein (such as albumin) on the blood side of the filter mem- dosing.53–­55 As an example, in a population PK analysis, Al Shaer
brane tending to retain cationic molecules such as aminoglycosides and colleagues evaluated the impact of three different effluent flow
and levofloxacin while facilitating the transport of anionic drugs rates (20, 30, and 40 mL/kg/h) on achieving cefepime PK/PD tar-
48
such as cephalosporins. However, the clinical significance of this gets.56 They found that the highest effluent flow rates resulted in a
effect must be confirmed before adjusting antibiotic dosing based lower probability of target attainment.
on the electric charge. Recently, a novel siderophore cephalosporin, cefiderocol, was re-
In summary, using PK information, clinicians can predict that leased on the market. Interestingly, based on population PK analysis
drugs with low MW, low Vd, and low protein binding will be signifi- and Monte Carlo simulations, dosing recommendations for different
cantly removed by RRT, and dosing should be adjusted for these an- effluent flow rates in CRRT are now incorporated into the prescrib-
tibiotics on RRT. ing information of this drug (and hopefully subsequent new drugs).57

3.4 | Factors related to RRT characteristics 3.4.4 | Dilution mode

3.4.1 | Continuous versus intermittent modalities In convective techniques, replacement fluid may be administered
before the hemofilter (pre-­dilution mode) or after the filter (post-­
Because of the continuous clearance induced, drug dosing during dilution mode). The mode of replacement fluid used directly affects
CRRT is easier than during IHD or PIRRT treatments that exhibit the amount of antibiotics removed by RRT as the antibiotic concen-
two distinct PK phases (intra and interdialytic). The time interval tration is diluted by the pre-­filter solution before crossing the mem-
between drug administration and the start of an IHD or PIRRT brane, resulting in decreased drug removal in pre-­dilution mode.
session is crucial in determining the amount of drug removed A study comparing various RRT modalities found that pre-­dilution
during the treatment. In a study comparing different modes of reduces clearances of most solutes by more than 15% at an ultra-
beta-­
lactam administration during PIRRT, the authors found filtration rate of 2 L/h.58 TDM-­assisted dosing or at least decreased
that the probability of target attainment was lower when beta-­ dosing should be considered when a predilution mode is applied.
lactams were administrated concomitantly with PIRRT initiation
rather than started as late as possible (14–­16 h) after drug infu-
sion.49 For cefepime and ceftazidime, extended and continuous 3.4.5 | Type of membranes/adsorption
infusion dosing on PIRRT provided limited improvement in PK/PD
target attainment while increasing the risk of drug-­related toxicity. CRRT membranes vary in permeability, membrane composition, and
Conversely, the administration of cefepime and ceftazidime 16 h contact area, which leads to differences in drug clearance. Over the
after PIRRT decreases the probability of reaching the toxic trough past few years, the surface area of the CRRT filters significantly in-
concentration at 48 h while maintaining the high probability of creased from approximately 0.6–­0.9 to 1.2–­1.5 m2, modulating the
efficacy. magnitude of drug adsorption. In patients receiving CVVHDF with
1.5 m2 polyacrylonitrile membranes, higher piperacillin-­t azobactam
doses were required compared to CVVHDF with 0.9 m2 polyacry-
3.4.2 | Convection versus diffusion lonitrile membranes due to an increased clearance with greater he-
mofilter surface areas.59
In a study assessing the impact of CRRT techniques (diffusive versus The type of hemofilter may influence the clearance induced
convective), both CVVH and CVVHD delivered at a similar effluent by RRT. Today, biocompatible hemofilter membranes are made of
1200 | KANJI et al.

different types of material: polyacrylonitrile, polymethylmethacry- the MIC of the causative bacterial pathogen. MIC determination may
late, and polysulfone. These membranes have different profiles, the not always be available and MIC clinical breakpoints may be used
first two exhibiting adsorption properties. In vitro models suggest as surrogates, although the former is preferred as epidemiologically
that the adsorption of antibiotics on the RRT membrane also affects derived cut-­offs are typically higher than the average MICs for a
drug elimination.33,34 For vancomycin, gentamicin, and tigecycline, particular organism and may result in unnecessarily high drug doses
a significant decrease in drug concentrations in the hemofiltra- and exposures.67 The most appropriate PK/PD indices for a spe-
60,61
tion circuit was observed for the polyacrylonitrile membrane. cific antimicrobial often depend on its bactericidal or bacteriostatic
Additionally, adsorption is inversely correlated to effluent flow rate profile. Time-­dependent antimicrobials (i.e., beta-­lactams) are opti-
and, in the case of amikacin, results in a 17% decrease in adsorp- mized by maximizing the fT > MIC, while concentration-­dependent
tion rate per 1 L/min increase in the flow rate.33,34 For vancomycin, antimicrobials (i.e., aminoglycosides and fluoroquinolones) are op-
adsorption seems lower, ranging from 4% to 12%.62 Consequently, timized by Cmax/MIC or AUC24/MIC. TDM assessments are rec-
clinicians should be aware of the risk of underexposure when poly- ommended upon initiation of empiric therapy and following dose
acrylonitrile filters are used, especially for aminoglycosides. TDM-­ changes, and also after any relevant change in the patient's clinical
based dosing or at least increased dosing should be considered. status that might affect the Vd or clearance of the antimicrobial in
The use of “high cut-­
off” membranes characterized by high question (i.e., aggressive resuscitation or large volume diuresis, ini-
porosity may potentially lead to greater removal of drugs with an tiation or discontinuation of extracorporeal treatments, changes in
MW of >1000 Da in diffusive techniques.56 The available data on end-­organ function). RRT in critically ill patients also adds an element
these membranes remain scarce and the clinical impact is yet to be of complexity with respect to the timing of dosing and sampling,
demonstrated. particularly in PIRRT. The “on/off” nature and frequency of PIRRT
contribute to variability in drug exposure requiring frequent dose
changes over the course of treatment. Patients may receive daily or
3.4.6 | Hemofilter lifetime and RRT downtime even consecutive runs of PIRRT during the acute phase of critical
illness followed by progressively less frequent runs as their severity
Progressive filter clotting and clogging may greatly reduce, over time, of illness improves and innate renal function returns. Even patients
filter performance and solute removal. The decline of solute CRRT with CRRT can have variable drug exposure over time due to tech-
clearance in relation to circuit running time may expose patients nical problems resulting in interruptions in dialysis (i.e., clotting of
to overdosing as observed in a study reporting teicoplanin con- venous access devices) and changing residual renal function as the
centrations for filters older than 24 h.63 As drug removal efficiency clinical condition of patients improves or worsens. Patients receiving
decreases over time for RRT circuits, lowering doses or replacing cir- either PIRRT or CRRT are often switched to traditional IHD when
cuits should be considered for prolonged running time. hemodynamic instability resolves in advance of renal function. TDM
assessments are recommended at all these transition points to guide
dosing when available. PD targets have been determined for most
4 | TH E R A PEU TI C D RU G M O N ITO R I N G antimicrobials although only vancomycin and aminoglycoside targets
A N D R E N A L R E PL AC E M E NT TH E R A PY are derived from well-­designed clinical studies.68,69 It must be ac-
knowledged that the strength of the evidence supporting PD targets
In critically ill patients with life-­threatening bacterial infections, the for other antimicrobials is variable and the relationship between PD
main predictors of survival related to antimicrobial use are early and target attainment and clinical outcomes is weak, largely due to inad-
17
appropriate empiric therapy. Appropriate antimicrobial therapy re- equate studies. Despite this, TDM is encouraged to guide antimicro-
fers to both the appropriate selection of an antimicrobial effective bial dosing, particularly in critically ill patients with life-­threatening
against the causative organisms and appropriate dosing aimed to infections. Given the added variability in drug exposure when further
achieve PD targets safely. Variability in PK and PD target attainment complicated by RRT, TDM is further justified provided the availability
between patients suggests that dosing should be individualized for of assay methods.70
each patient, but these factors are also dynamic suggesting that dos- Recommendations for TDM in critically ill patients receiving
ing needs may change within the same patient over the course of RRT are adapted from Matusik and colleagues70 and are provided
their ICU stay. TDM relies on interpreting serum drug concentra- in Table 2. TDM considerations for PIRRT are provided given the
tions (total or unbound) and individualizing drug dosing to meet pre-­ increasing popularity of this RRT modality in ICUs and the additional
defined PD targets. In critically ill patients, there is growing evidence considerations required with respect to the timing of antimicrobial
that current dosing strategies are insufficient and associated with administration and serum concentration sampling.
both reduced treatment efficacy and toxicity.11,64,65 Advocation for
TDM and dose individualization in critically ill patients receiving an-
timicrobials traditionally associated with minimal toxicity (i.e., beta-­ 4.1 | Aminoglycosides
lactams) is increasing to improve efficacy and safety outcomes.47,66
PK/PD targets correlate either clinical efficacy or safety with Aminoglycosides exhibit concentration-­dependent bactericidal ac-
serum antimicrobial concentrations or PK parameters in relation to tivity but ototoxicity and nephrotoxicity are related to accumulation
TA B L E 2 Therapeutic drug monitoring recommendations for antimicrobials in critically ill adults receiving CRRT or PIRRT (adapted from reference (70)).

Antimicrobial administration
Antibiotic PD profile PK/PD Target Timing of TDM assessments Sampling times strategies for PIRRT
KANJI et al.

Aminoglycosides Concentration-­dependent • Cmax/MIC ≥8–­10 After the first dose and after • Cmax: 30 min after the end of the Administer immediately before
bactericidal activity • AUC: 80–­120 mg/h/L each dose change infusion PIRRT session on dialysis day
(nephrotoxicity • Cmin (amikacin) ≤2.5 mg/L (safety) • Cmin: immediately before the next
and ototoxicity are • Cmin (gentamicin/tobramycin/ infusion
time-­dependent and netilmicin) ≤0.5 mg/L • AUC: Cmax and an additional sample
cumulative) between 6 and 22 h post-­infusion
Beta-­lactams Time-­dependent bactericidal • ƒCmin or ƒCss ≥ MIC (i.e., 100% 24–­4 8 h after treatment onset • Cmin: immediately before the next Continuous or prolonged infusions
(penicillins, activity ƒT > MIC) or dose change infusion after a loading dose. Administer
cephalosporins, • Safety target (piperacillin) Cmin • Css: any time during a continuous after a PIRRT session (if using)
carbapenems) <361 mg/L or Css < 157 mg/L infusion and before the PIRRT on dialysis days for intermittent
• Safety target (cefepime) < 20–­22 mg/L session if using infusions
or Css < 35 mg/L
• Safety target (meropenem) Cmin
<64 mg/L
Colistin Concentration-­dependent • Cmin ~2 mg/L 48–­72 h after treatment onset • Cmin immediately before the next PIRRT sessions may be initiated
bactericidal activity or dose change infusion ideally before the next after the dose is administered
PIRRT session if using and followed by an additional
dose at the end of the session
Daptomycin Concentration-­dependent • AUC/MIC ≥666 48–­72 h after treatment onset • Ideally 2 samples for AUC including Administration after PIRRT on
bactericidal activity • Safety target: Cmin <24 mg/L or dose change a Cmax 30 min after the infusion and dialysis days
a second sample 6–­22 h later but
before the next dose
• Cmin just before the next dose
Fluoroquinolones Predominantly • AUC/MIC ≥100–­125 48–­72 h after treatment onset • Cmax: 30 min after the end of the Administer dose post-­PIRRT
concentration-­ • Cmax/MIC ≥8–­12 or dose change infusion session on dialysis days
dependent bactericidal • An additional sample during the
activity dosing interval for AUC calculations
Linezolid Time-­dependent • Cmin 2–­7 mg/L 48–­72 h after treatment onset • Cmin just before the next dose Administer dose post-­PIRRT
bacteriostatic activity or dose change session on dialysis days
against most organisms
Teicoplanin Predominantly time-­ • Cmin ≥15–­4 0 mg/L 48–­72 h after treatment onset • Cmin just before the next dose Administer after the end of PIRRT
dependent bactericidal or dose change on dialysis days
activity
Tigecycline Time-­dependent bactericidal • AUC/MIC ≥7–­18 96 h after treatment onset or • Ideally 2 samples for AUC including Low rates of removal by dialysis
activity dose change a Cmax 30 min after the infusion and and can be dosed regardless of
a second sample 6–­22 h later but PIRRT timing
before the next dose
Vancomycin Predominantly time-­ • Cmin 10–­20 mg/L 24–­4 8 h after treatment onset • Cmin just before the next dose Administer after the end of PIRRT
dependent bactericidal • Css 20–­25 mg/L or dose change • Css anytime during the continuous on dialysis days Consider a
activity • AUC/MIC ≥400 mg/h/L infusion continuous infusion after
• • AUC: ideally two levels including the a loading dose for severe
|

Safety target: AUC < 600 mg/h/L

Cmin and the Cmax 30–­6 0 min after infections
the end of infusion
1201
1202 | KANJI et al.

over time. PK/PD targets associated with efficacy are related to the beta-­lactams.15 The DOLPHIN trial, a recent multi-­centered trial
maximal concentrations (Cmax) or maximal exposure (AUC), and comparing beta-­lactam TDM in adult ICUs using model-­informed
once-­daily dosing strategies are preferred to obtain high maximal precision dosing and pharmacometric modeling to TDM according
concentrations. With CRRT modalities, the rates of drug clearance to usual care did not improve outcomes such as length of stay or
are relatively constant within and between days provided the CRRT mortality.74 This study highlights the importance of early, appropri-
settings are not changed. Timing of drug administration and TDM ate, and adequately dosed empiric antimicrobials as TDM can only
are more challenging when using IHD or PIRRT due to large differ- be applied after empiric antimicrobial and dose selection and initial
ences in clearance when dialysis is on versus off. When PIRRT mo- dosing may be more predictive of meaningful outcomes than TDM-­
dalities are used for RRT dosing, the aminoglycoside just before the assisted maintenance dosing. Continuous or prolonged infusions of
initiation of a PIRRT session allows for high concentrations and maxi- beta-­lactams (particularly those that are typically dosed more than
mal clearance.71 In severe infections caused by multi-­drug resistant once per day such as piperacillin, cefepime, and meropenem) appear
(MDR) gram-­negative organisms, the administration of high-­dose to be the most reliable strategy to achieve 100% ƒT > MIC, particu-
aminoglycoside therapy coupled with continuous RRT could allow larly if TDM is not available. This is a cost-­neutral intervention that
the achievement of adequate peak drug levels with acceptable drug is associated with improved outcomes including mortality.44 The
elimination. In a proof-­of-­concept cohort study, 15 patients (five with TARGET trial which randomized 256 patients with septic shock to
normal renal function, four with evidence of renal dysfunction but TDM-­guided continuous infusions compared with fixed-­dose con-
not being dialyzed, and six already on CRRT) with multidrug-­resistant tinuous infusions of piperacillin/tazobactam showed that although
gram-­negative sepsis deemed to have failed previous combinations TDM-­guided continuous infusions were more likely to achieve
of antimicrobial therapy were treated with daily, high dose aminogly- PD targets, there was no difference in clinical outcomes between
cosides; CVVHDF was initiated for all subjects and administered at groups.75 This further suggests that continuous infusion of piperacil-
40–­45 mg/kg/h an hour after the end of the aminoglycoside loading lin/tazobactam is a reasonable strategy to achieve PD targets when
dose and after a serum drug level measurement (Cmax) and contin- TDM is not available.
ued for the entire duration of therapy.71 The Cmax/MIC target was
8 and trough targets were <2 mg/L for tobramycin and gentamicin
and <5 mg/L for amikacin. Drug dosing was adjusted to reach Cmax 4.3 | Vancomycin
targets while dosing intervals and CVVHDF dose were adjusted for
supratherapeutic trough levels. A favorable clinical response was TDM for vancomycin has been the standard of care for decades, but
observed in 53% of patients and no renal toxicity was observed recently a shift from trough level assessment to AUC24/MIC meas-
among survivors. The effectiveness of this approach (initiating CRRT urements has been proposed by clinical practice guidelines for both
to facilitate aminoglycoside clearance) should be evaluated in larger efficacy and toxicity, particularly when vancomycin is used to treat
cohorts and compared with other existing therapeutic approaches methicillin-­resistant Staphylococcus aureus (MRSA) infections.69 For
but until then, this strategy coupled with TDM to determine the pro- AUC calculations, Bayesian methods using PK models and special-
portional clearance while on CRRT may be used to guide dose and ized software are recommended and two samples will increase the
interval determination when aminoglycosides are indicated. A similar precision of AUC estimation and are associated with better target at-
strategy in PIRRT has not yet been studied, but given the “on/off” tainment than dosing guided by population PK models.76 Vancomycin
nature of dialysis in PIRRT trials, is needed to determine the utility treatment failure rates in MRSA infections are notoriously as high
of this approach. as 50%. Comparisons between TDM methods (trough level versus
AUC24/MIC monitoring) do not predict differences in treatment fail-
ure rates but rather show high rates of concordance. Antimicrobial
4.2 | Beta-­lactams stewardship-­related variables related to early treatment initiation
and rapid PD target attainment using loading doses (regardless of
For beta-­lactam antibiotics including penicillins, cephalosporins, and TDM method) are greater predictors of treatment success.77
carbapenems, time-­dependent bactericidal activity is optimized by While a detailed discussion of TDM for each antimicrobial is
maintaining serum concentrations above the MIC. Although main- beyond the scope of this review, readers are referred to Table 2
taining serum concentrations above the MIC for at least 40–­60% for TDM recommendations regarding other antimicrobials not dis-
of the dosing interval has been associated with clinical efficacy, cussed above.
maintaining serum concentrations above the MIC for 100% of the
dosing interval may be more appropriate for critically ill patients
with life-­threatening infections.72 Even achieving 100% above 4x 5 | FU T U R E D I R EC TI O N S A N D A DV I C E
the MIC has been proposed to prevent microbiological failure and FO R C LI N I C I A N S
resistance development,73 however, achieving such high concentra-
tions may increase the risk of neurotoxicity particularly with cefa- This review has highlighted the challenges clinicians face when dos-
zolin which appears to have the lowest threshold for toxicity of all ing antimicrobials in critically ill patients receiving RRT. As a rational
KANJI et al. | 1203

approach, the following dosing strategy during RRT in ICU patients 2. Heffernan AJ, Mohd Sazlly Lim S, Lipman J, Roberts JA. A per-
could be proposed: (1) high-­loading doses to avoid low concentra- sonalised approach to antibiotic pharmacokinetics and pharma-
codynamics in critically ill patients. Anaesth Crit Care Pain Med.
tions associated with increased apparent Vd at the early phase of
2021;40(6):100970.
sepsis (first 24–­48 h); (2) followed by adjustments of maintenance 3. Jacobs A, Taccone FS, Roberts JA, et al. Beta-­lactam dosage regi-
dose based on the effluent flow rate and residual renal function, mens in septic patients with augmented renal clearance. Antimicrob
and (3) subsequent adjustment guided by TDM and Bayesian dosing Agents Chemother. 2018;62(9):e02534-­17.
4. Taccone FS, Laterre PF, Dugernier T, et al. Insufficient beta-­lactam
software (if available) to avoid toxicity. However, generalized dosing
concentrations in the early phase of severe sepsis and septic shock.
recommendations intended to be applied to any and all critically ill Crit Care. 2010;14(4):R126.
adults receiving RRT in an ICU should be considered with caution. 5. Ulldemolins M, Roberts JA, Rello J, Paterson DL, Lipman J. The
Although individualized dosing strategies that consider local epide- effects of hypoalbuminaemia on optimizing antibacterial dosing in
critically ill patients. Clin Pharmacokinet. 2011;50(2):99-­110.
miology and resistance patterns, RRT modalities, and patient and
6. Roberts DM, Roberts JA, Roberts MS, et al. Variability of antibiotic
drug characteristics are preferred, the practical application of this concentrations in critically ill patients receiving continuous renal
guidance in the absence of TDM for most antimicrobials used in the replacement therapy: a multicentre pharmacokinetic study. Crit
ICU is challenging. Not only is this information difficult to obtain in Care Med. 2012;40(5):1523-­1528.
7. Beumier M, Casu GS, Hites M, et al. Elevated beta-­lactam concen-
clinical settings, but even when it is available the net effect on drug
trations associated with neurological deterioration in ICU septic
disposition is difficult to determine. The mode and characteristics patients. Minerva Anestesiol. 2015;81(5):497-­506.
of RRT highly influence drug disposition and should be considered 8. Sadeghi K, Shahrami B, Hosseini Fani F, Hamishehkar H,
when dosing and monitoring antimicrobials, as this information is Mojtahedzadeh M. Detection of subclinical nephrotoxicity induced
by aminoglycosides in critically ill elderly patients using trough
typically readily available at the bedside. There is an urgent need to
levels and urinary neutrophil gelatinase-­associated lipocalin. Eur J
make TDM available and practical at the bedside to assist clinicians
Hosp Pharm. 2022;29(e1):e63-­e66.
in individualizing antimicrobial dosing. Innovation in the form of in- 9. Fiore M, Peluso L, Taccone FS, Hites M. The impact of con-
expensive and precise point-­of-­care testing is one possible solution. tinuous renal replacement therapy on antibiotic pharmacoki-
There is also a need to consider alternative and innovative methods netics in critically ill patients. Expert Opin Drug Metab Toxicol.
2021;17(5):543-­554.
for studying PK/PD of antimicrobials in this setting. Population PK
10. Seyler L, Cotton F, Taccone FS, et al. Recommended beta-­lactam
methods are superior to traditional PK studies in that they allow for regimens are inadequate in septic patients treated with continuous
covariate analysis and are more feasible by allowing for less frequent renal replacement therapy. Crit Care. 2011;15(3):R137.
and variable sampling strategies but still do not address the variabil- 11. Roberts JA, Joynt GM, Lee A, et al. The effect of renal replace-
ment therapy and antibiotic dose on antibiotic concentrations
ity in RRT modalities, characteristics, and dose when done in single
in critically ill patients: data from the multinational sampling
centers. One example of innovation in this research field is the inte- antibiotics in renal replacement therapy study. Clin Infect Dis.
grated dialysis pharmacometrics model. This approach incorporates 2021;72(8):1369-­1378.
all RRT samples (pre-­and post-­filter plasma and effluent), RRT types 12. Beumier M, Casu GS, Hites M, et al. Beta-­lactam antibiotic con-
centrations during continuous renal replacement therapy. Crit Care.
and settings, and drug physiochemical properties to provide quanti-
2014;18(3):R105.
tative insight into drug clearance in patients receiving RRT.78 13. Economou CJP, Wong G, McWhinney B, Ungerer JPJ, Lipman J,
In conclusion, we highlighted the complexity and sources of Roberts JA. Impact of beta-­lactam antibiotic therapeutic drug
variability with respect to antimicrobial disposition in critically ill pa- monitoring on dose adjustments in critically ill patients undergo-
ing continuous renal replacement therapy. Int J Antimicrob Agents.
tients receiving RRT. Thus, variability between patients warrants an
2017;49(5):589-­594.
individualized approach to antimicrobial dosing using clinical data, 14. Dhaese SAM, De Kezel M, Callant M, et al. Emergence of antimi-
RRT characteristics, and TDM for guidance in this highly vulnerable crobial resistance to piperacillin/tazobactam or meropenem in the
population. ICU: intermittent versus continuous infusion. A retrospective co-
hort study. J Crit Care. 2018;47:164-­168.
15. Roger C, Louart B. Beta-­ lactams toxicity in the intensive care
C O N FL I C T O F I N T E R E S T S TAT E M E N T
unit: an underestimated collateral damage? Microorganisms.
The authors declare no conflicts of interest. 2021;9(7):1505.
16. Udy AA, Roberts JA, Lipman J. Clinical implications of antibiotic
DATA AVA I L A B I L I T Y S TAT E M E N T pharmacokinetic principles in the critically ill. Intensive Care Med.
2013;39(12):2070-­2082.
Data sharing not applicable to this article as no datasets were gener-
17. Kumar A, Roberts D, Wood KE, et al. Duration of hypotension
ated or analysed during the current study before initiation of effective antimicrobial therapy is the criti-
cal determinant of survival in human septic shock. Crit Care Med.
ORCID 2006;34(6):1589-­1596.
18. Hoste EA, Bagshaw SM, Bellomo R, et al. Epidemiology of acute
Salmaan Kanji https://orcid.org/0000-0003-0594-0360
kidney injury in critically ill patients: the multinational AKI-­EPI
study. Intensive Care Med. 2015;41(8):1411-­1423.
REFERENCES 19. Coca SG, Singanamala S, Parikh CR. Chronic kidney disease after
1. Vincent JL, Bassetti M, Francois B, et al. Advances in antibiotic acute kidney injury: a systematic review and meta-­analysis. Kidney
therapy in the critically ill. Crit Care. 2016;20(1):133. Int. 2012;81(5):442-­4 48.
1204 | KANJI et al.

20. Hsu CY, Chinchilli VM, Coca S, et al. Post-­acute kidney injury pro- 39. Ulldemolins M, Roberts JA, Wallis SC, Rello J, Lipman J. Flucloxacillin
teinuria and subsequent kidney disease progression: the assess- dosing in critically ill patients with hypoalbuminaemia: special em-
ment, serial evaluation, and subsequent sequelae in acute kidney phasis on unbound pharmacokinetics. J Antimicrob Chemother.
injury (ASSESS-­AKI) study. JAMA Intern Med. 2020;180(3):402-­410. 2010;65(8):1771-­1778.
21. Rimes-­Stigare C, Frumento P, Bottai M, Martensson J, Martling CR, 40. Bukkems LH, Roger C, Hodiamont CJ, et al. Predictive perfor-
Bell M. Long-­term mortality and risk factors for development of mance of a gentamicin population pharmacokinetic model in two
end-­stage renal disease in critically ill patients with and without western populations of critically ill patients. Int J Antimicrob Agents.
chronic kidney disease. Crit Care. 2015;19:383. 2018;52(2):218-­225.
22. Bonnassieux M, Duclos A, Schneider AG, et al. Renal replacement 41. Nolin TD, Appiah K, Kendrick SA, Le P, McMonagle E, Himmelfarb J.
therapy modality in the ICU and renal recovery at hospital dis- Hemodialysis acutely improves hepatic CYP3A4 metabolic activity.
charge. Crit Care Med. 2018;46(2):e102-­e110. J Am Soc Nephrol. 2006;17(9):2363-­2367.
23. Khwaja A. KDIGO clinical practice guidelines for acute kidney in- 42. Roger C, Wallis SC, Muller L, et al. Influence of renal replacement
jury. Nephron Clin Pract. 2012;120(4):c179-­c184. modalities on amikacin population pharmacokinetics in critically
24. Schneider AG, Bellomo R, Bagshaw SM, et al. Choice of renal re- ill patients on continuous renal replacement therapy. Antimicrob
placement therapy modality and dialysis dependence after acute Agents Chemother. 2016;60(8):4901-­4909.
kidney injury: a systematic review and meta-­analysis. Intensive Care 43. He S, Cheng Z, Xie F. Population pharmacokinetics and dosing op-
Med. 2013;39(6):987-­997. timization of gentamicin in critically ill patients undergoing continu-
25. Harvey AK, Burns KEA, McArthur E, et al. Short-­and long-­term out- ous renal replacement therapy. Drug des Devel Ther. 2022;16:13-­22.
comes of sustained low efficiency dialysis vs continuous renal re- 44. Thabet P, Joshi A, MacDonald E, et al. Clinical and pharmacoki-
placement therapy in critically ill patients with acute kidney injury. netic/dynamic outcomes of prolonged infusions of beta-­ lactam
J Crit Care. 2021;62:76-­81. antimicrobials: an overview of systematic reviews. PloS One.
26. Nash DM, Przech S, Wald R, O'Reilly D. Systematic review and 2021;16(1):e0244966.
meta-­analysis of renal replacement therapy modalities for acute 45. Sember AM, LoFaso ME, Lewis SJ. An optimal extended-­infusion
kidney injury in the intensive care unit. J Crit Care. 2017;41:138-­144. dosing of cefepime and ceftazidime in critically ill patients with con-
27. Douvris A, Zeid K, Hiremath S, et al. Mechanisms for hemodynamic tinuous renal replacement therapy. J Crit Care. 2022;69:154011.
instability related to renal replacement therapy: a narrative review. 46. Jamal JA, Roberts DM, Udy AA, et al. Pharmacokinetics of pipera-
Intensive Care Med. 2019;45(10):1333-­1346. cillin in critically ill patients receiving continuous venovenous hae-
28. Joannes-­Boyau O, Velly L, Ichai C. Optimizing continuous renal re- mofiltration: a randomised controlled trial of continuous infusion
placement therapy in the ICU: a team strategy. Curr Opin Crit Care. versus intermittent bolus administration. Int J Antimicrob Agents.
2018;24(6):476-­482. 2015;46(1):39-­4 4.
29. Brain M, Winson E, Roodenburg O, McNeil J. Non anti-­coagulant 47. Roberts JA, Abdul-­A ziz MH, Lipman J, et al. Individualised antibi-
factors associated with filter life in continuous renal replacement otic dosing for patients who are critically ill: challenges and poten-
therapy (CRRT): a systematic review and meta-­ analysis. BMC tial solutions. Lancet Infect Dis. 2014;14(6):498-­509.
Nephrol. 2017;18(1):69. 48. Tian Q, Gomersall CD, Ip M, Tan PE, Joynt GM, Choi GY. Adsorption
30. AlEnezi F, Alhazzani W, Ma J, et al. Continuous venovenous hemo- of amikacin, a significant mechanism of elimination by hemofiltra-
filtration versus continuous venovenous hemodiafiltration in crit- tion. Antimicrob Agents Chemother. 2008;52(3):1009-­1013.
ically ill patients: a retrospective cohort study from a Canadian 49. Jang SM, Gharibian KN, Lewis SJ, Fissell WH, Tolwani AJ, Mueller
tertiary centre. Can Respir J. 2014;21(3):176-­180. BA. A Monte Carlo simulation approach for beta-­lactam dosing in
31. Friedrich JO, Wald R, Bagshaw SM, Burns KE, Adhikari NK. critically ill patients receiving prolonged intermittent renal replace-
Hemofiltration compared to hemodialysis for acute kidney injury: ment therapy. J Clin Pharmacol. 2018;58(10):1254-­1265.
systematic review and meta-­analysis. Crit Care. 2012;16(4):R146. 50. Ricci Z, Ronco C, Bachetoni A, et al. Solute removal during contin-
32. Dalbhi SA, Alorf R, Alotaibi M, et al. Sustained low efficiency di- uous renal replacement therapy in critically ill patients: convection
alysis is non-­inferior to continuous renal replacement therapy in versus diffusion. Crit Care. 2006;10(2):R67.
critically ill patients with acute kidney injury: a comparative meta-­ 51. Roger C, Wallis SC, Louart B, et al. Comparison of equal doses of
analysis. Medicine. 2021;100(51):e28118. continuous venovenous haemofiltration and haemodiafiltration on
33. Zhang L, Yang J, Eastwood GM, Zhu G, Tanaka A, Bellomo R. ciprofloxacin population pharmacokinetics in critically ill patients. J
Extended daily dialysis versus continuous renal replacement Antimicrob Chemother. 2016;71(6):1643-­1650.
therapy for acute kidney injury: a meta-­analysis. Am J Kidney Dis. 52. Roger C, Cotta MO, Muller L, et al. Impact of renal replacement
2015;66(2):322-­330. modalities on the clearance of piperacillin-­ t azobactam adminis-
34. Hoff BM, Maker JH, Dager WE, Heintz BH. Antibiotic dosing for crit- tered via continuous infusion in critically ill patients. Int J Antimicrob
ically ill adult patients receiving intermittent hemodialysis, prolonged Agents. 2017;50(2):227-­231.
intermittent renal replacement therapy, and continuous renal re- 53. Carlier M, Taccone FS, Beumier M, et al. Population pharmacoki-
placement therapy: an update. Ann Pharmacother. 2020;54(1):43-­55. netics and dosing simulations of cefepime in septic shock patients
35. Goncalves-­Pereira J, Povoa P. Antibiotics in critically ill patients: receiving continuous renal replacement therapy. Int J Antimicrob
a systematic review of the pharmacokinetics of beta-­lactams. Crit Agents. 2015;46(4):413-­419.
Care. 2011;15(5):R206. 54. Chaijamorn W, Charoensareerat T, Srisawat N, Pattharachayakul
36. Marik PE. Aminoglycoside volume of distribution and illness S, Boonpeng A. Cefepime dosing regimens in critically ill patients
severity in critically ill septic patients. Anaesth Intensive Care. receiving continuous renal replacement therapy: a Monte Carlo
1993;21(2):172-­173. simulation study. J Intensive Care. 2018;6:61.
37. Ulldemolins M, Bastida C, Llaurado-­Serra M, et al. Once-­daily 1 g 55. Chaijamorn W, Shaw AR, Lewis SJ, Mueller BA. Ex vivo Ceftolozane/
ceftriaxone optimizes exposure in patients with septic shock and tazobactam clearance during continuous renal replacement ther-
hypoalbuminemia receiving continuous veno-­venous hemodiafil- apy. Blood Purif. 2017;44(1):16-­23.
tration. Eur J Clin Pharmacol. 2021;77(8):1169-­1180. 56. Al-­Shaer MH, Maguigan K, Ashton J, et al. Applying cefepime pop-
38. Joynt GM, Lipman J, Gomersall CD, Young RJ, Wong EL, Gin T. The ulation pharmacokinetics to critically ill patients receiving con-
pharmacokinetics of once-­daily dosing of ceftriaxone in critically ill tinuous renal replacement therapy. Antimicrob Agents Chemother.
patients. J Antimicrob Chemother. 2001;47(4):421-­429. 2022;66(1):e0161121.
KANJI et al. | 1205

57. Wenzler E, Butler D, Tan X, Katsube T, Wajima T. Pharmacokinetics, 69. Rybak MJ, Le J, Lodise TP, et al. Therapeutic monitoring of van-
pharmacodynamics, and dose optimization of Cefiderocol during comycin for serious methicillin-­resistant Staphylococcus aureus in-
continuous renal replacement therapy. Clin Pharmacokinet. fections: a revised consensus guideline and review by the American
2022;61(4):539-­552. Society of Health-­ System Pharmacists, the Infectious Diseases
58. Brunet S, Leblanc M, Geadah D, Parent D, Courteau S, Cardinal J. Society of America, the Pediatric Infectious Diseases Society, and
Diffusive and convective solute clearances during continuous renal the Society of Infectious Diseases Pharmacists. Am J Health Syst
replacement therapy at various dialysate and ultrafiltration flow Pharm. 2020;77(11):835-­864.
rates. Am J Kidney Dis. 1999;34(3):486-­492. 70. Matusik E, Boidin C, Friggeri A, et al. Therapeutic drug monitoring
59. Ulldemolins M, Martin-­Loeches I, Llaurado-­Serra M, et al. Piperacillin of antibiotic drugs in patients receiving continuous renal replace-
population pharmacokinetics in critically ill patients with multiple ment therapy or intermittent hemodialysis: a critical review. Ther
organ dysfunction syndrome receiving continuous venovenous Drug Monit. 2022;44(1):86-­102.
haemodiafiltration: effect of type of dialysis membrane on dosing 71. Brasseur A, Hites M, Roisin S, et al. A high-­dose aminoglycoside
requirements. J Antimicrob Chemother. 2016;71(6):1651-­1659. regimen combined with renal replacement therapy for the treat-
60. Onichimowski D, Ziolkowski H, Nosek K, Jaroszewski J, Rypulak E, ment of MDR pathogens: a proof-­of-­concept study. J Antimicrob
Czuczwar M. Comparison of adsorption of selected antibiotics on Chemother. 2016;71(5):1386-­1394.
the filters in continuous renal replacement therapy circuits: in vitro 72. Maguigan KL, Al-­Shaer MH, Peloquin CA. Beta-­lactams dosing in
studies. J Artif Organs. 2020;23(2):163-­170. critically ill patients with gram-­negative bacterial infections: a PK/
61. Vossen MG, Pferschy S, Milacek C, et al. In vivo / in vitro correlation PD approach. Antibiotics. 2021;10(10):1154.
of pharmacokinetics of gentamicin, vancomycin, teicoplanin and 73. Gatti M, Cojutti PG, Pascale R, et al. Assessment of a PK/PD tar-
Doripenem in a bovine blood hemodialysis model. Front Pharmacol. get of continuous infusion Beta-­lactams useful for preventing mi-
2021;12:702455. crobiological failure and/or resistance development in critically
62. Uchida M, Sawada M, Yamazaki S, Suzuki T, Suzuki T, Ishii I. ill patients affected by documented gram-­ negative infections.
Contribution of diafiltration and adsorption to vancomycin clear- Antibiotics. 2021;10(11):1311.
ance in a continuous hemodiafiltration circuit model in vitro. Artif 74. Ewoldt TMJ, Abdulla A, Rietdijk WJR, et al. Model-­informed preci-
Organs. 2022;46(6):1086-­1096. sion dosing of beta-­lactam antibiotics and ciprofloxacin in critically
63. Meyer B, Traunmuller F, Hamwi A, et al. Pharmacokinetics of te- ill patients: a multicentre randomised clinical trial. Intensive Care
icoplanin during continuous hemofiltration with a new and a Med. 2022;48(12):1760-­1771.
24-­h used highly permeable membrane: rationale for therapeu- 75. Hagel S, Bach F, Brenner T, et al. Effect of therapeutic drug
tic drug monitoring-­ guided dosage. Int J Clin Pharmacol Ther. monitoring-­based dose optimization of piperacillin/tazobactam on
2004;42(10):556-­560. sepsis-­related organ dysfunction in patients with sepsis: a random-
64. Abdulla A, Dijkstra A, Hunfeld NGM, et al. Failure of target attain- ized controlled trial. Intensive Care Med. 2022;48(3):311-­321.
ment of beta-­lactam antibiotics in critically ill patients and associ- 76. Oda K, Jono H, Kamohara H, Nishi K, Tanoue N, Saito H.
ated risk factors: a two-­center prospective study (EXPAT). Crit Care. Development of vancomycin dose individualization strategy by
2020;24(1):558. Bayesian prediction in patients receiving continuous renal replace-
65. Roberts JA, Paul SK, Akova M, et al. DALI: defining antibiotic ment therapy. Pharm Res. 2020;37(6):108.
levels in intensive care unit patients: are current beta-­lactam an- 77. Marko R, Hajjar J, Nzeribe V, et al. Therapeutic drug monitor-
tibiotic doses sufficient for critically ill patients? Clin Infect Dis. ing of vancomycin in adult patients with methicillin-­ resistant
2014;58(8):1072-­1083. Staphylococcus aureus bacteremia or pneumonia. Can J Hosp
66. Guilhaumou R, Benaboud S, Bennis Y, et al. Optimization of the treat- Pharm. 2021;74(4):334-­3 43.
ment with beta-­lactam antibiotics in critically ill patients-­guidelines 78. Broeker A, Vossen MG, Thalhammer F, et al. An integrated dialy-
from the French Society of Pharmacology and Therapeutics sis Pharmacometric (IDP) model to evaluate the pharmacokinet-
(Societe Francaise de Pharmacologie et Therapeutique-­SFPT) and ics in patients undergoing renal replacement therapy. Pharm Res.
the French Society of Anaesthesia and Intensive Care Medicine 2020;37(6):96.
(Societe Francaise d'Anesthesie et reanimation-­SFAR). Crit Care.
2019;23(1):104.
67. Woksepp H, Hallgren A, Borgstrom S, et al. High target attainment
for beta-­lactam antibiotics in intensive care unit patients when
How to cite this article: Kanji S, Roger C, Taccone FS, Muller L.
actual minimum inhibitory concentrations are applied. Eur J Clin
Microbiol Infect Dis. 2017;36(3):553-­563. Practical considerations for individualizing drug dosing in
68. Kashuba AD, Nafziger AN, Drusano GL, Bertino JS Jr. Optimizing critically ill adults receiving renal replacement therapy.
aminoglycoside therapy for nosocomial pneumonia caused Pharmacotherapy. 2023;43:1194-1205. doi:10.1002/phar.2858
by gram-­ negative bacteria. Antimicrob Agents Chemother.
1999;43(3):623-­629.