Beta-lactam dosing during continuous renal replacement therapy: a survey of practices in french intensive care units

To the best of our knowledge, the present study is the first to have assessed beta-lactam dosing practices in the context of CRRT. Furthermore, the survey described CRRT practices in France. Physicians practicing intensive care medicine (whatever their initial qualification) were included in the survey. The majority were intensivists or anesthesiologists, which reflects the fact that anesthesiologists are qualified for critical care medicine in France.

In order to review beta-lactam dosing in CRRT, we analyzed compliance with the French guidelines and Kidney Disease: Improving Global Outcomes (KDIGO) guidelines [21, 22]. Our present findings were in line with studies of CRRT practices performed over the last decade [23‐25]. CRRT was preferred to intermittent hemodialysis, although there is no clear evidence of superiority concerning reduced mortality [21, 22]. This technique is considered to provide greater hemodynamic stability. As recommended, citrate was the main anticoagulant used [21, 22]. Multicenter randomized controlled trials and meta-analyses have shown that increasing the CRRT dose intensity above 20-25 ml/kg/h does not increase survival but does lead to more metabolic complications [26‐29]. Therefore, in order to deliver a dose of 20–25 ml/kg/h and to minimize interruptions in the CRRT, the KDIGO guidelines recommend a value of 25–30 ml/kg/h [22]. Although citrate limits filter coagulation, a third of our respondents indicated (as also found in other studies) that they prescribe a higher CRRT dose - leading to greater clearance of beta-lactams [23, 25]. The physicians reported using different reference body weights to prescribe the CRRT dose, which contributed to disparities in RRT practices. Although most studies are based on total weight at the time of randomization, the guidelines do not specify the weight to be used to determine the dialysis dose - resulting in a variety of practices. Our survey highlighted the diversity and lack of harmonization of CRRT techniques.

Physicians may not be sufficiently aware of the need to maintain beta-lactam full doses during the initial phase of treatment. Administration must be optimized in order to maintain effective antibiotic concentrations for a sufficiently long time and thus maximize the chances of therapeutic success. This is especially true in the initial phase of sepsis when cardiac output and capillary permeability increase and protein binding is altered; this leads to increased clearance and a larger volume of distribution, inducing low serum and tissue beta-lactams concentrations [1, 30]. Underexposure increases the likelihood of therapeutic failure and the emergence of resistance. Improving antibiotic exposure is, therefore, a major challenge. Secondly, organ dysfunctions (and especially kidney failure) lead to high serum antibiotic concentrations. CRRT may limit underdosing when using full doses with a potential risk of neurotoxicity in the event of overdosing [4, 12, 15]. The mechanisms by which antibiotics are eliminated by CRRT appear to be poorly understood since the majority of physicians do not reportedly adjust the dosage as a function of the dialysis dose or effluent flow.

Our survey highlighted the broad implementation of extended and continuous infusions and so showed that physicians were well aware of the time-dependent nature of the beta-lactams’ activity. Even though the guidelines recommending the use of extended and continuous infusions for all compounds (to increase the probability of target attainment), these modalities are still mainly used for a few beta-lactams only and adherence to guidelines is suboptimal [5, 20]. This might be due to the recent changes in the French guidelines between 2014 and 2018. The French-speaking Intensive Care Society initially recommended continuous infusions for ceftazidime and extended infusions (over 3-4 h) for a few other beta-lactams [31]. In October 2018, the French Society of Anesthesia and Intensive Care Medicine and the French Society of Pharmacology and Therapeutics suggested the use of prolonged or continuous infusion of beta-lactams (intending to increase the probability of PK/PD target attainment and clinical cure rates) but did not differentiate between the various molecules [5]. These guidelines apply to infections with high-MIC bacteria or with non-fermenting Gram-negative bacilli, in patients in shock or with high severity scores, and lower respiratory tract infections (as described in our clinical vignettes). However, these guidelines were published just a few months before our survey, which may have limited their dissemination. Continuous and extended infusions are mostly used for piperacillin-tazobactam and ceftazidime; these are the compounds for which we have the most literature data, as reported in the ANTIBIOPERF study performed in 2015 [32]. Continuous infusion is mentioned in the French summary of product characteristics for ceftazidime only [33]. In the last decade, several other surveys have focused on these practices. A survey of 34 Belgian hospitals in 2011 showed that four beta-lactams were administered in the ICU by continuous and extended infusion to a varying extent: in 35% of the ICUs for cefepime, 38% for piperacillin-tazobactam, 68% for meropenem, and 81% for ceftazidime [34]. In 2013, an international multicenter survey of 402 physicians in 53 countries reported that piperacillin/tazobactam and carbapenem (meropenem and imipenem) were mainly administered as intermittent infusions (71% and 68%, respectively) [35]. In 2019, a German study reported that meropenem (70%), piperacillin/tazobactam (67%) and imipenem (50%) were the beta-lactams most regularly administered as prolonged and continuous infusions [36]. A survey performed in Australia and New Zealand in 2016 focused on meropenem and piperacillin-tazobactam, which were most frequently administered in intermittent infusions [37]. The administration method may therefore differ from one geographic region to another. The survey of Australia and New Zealand highlighted the fact that the infusion modality could be determined on a case-by-case basis, depending on the presence of bacteria with high MICs, pathological changes (sepsis or sepsis shock), and the severity of the patient’s illness [37].

The prescription tools used by the majority of physicians are not designed to recommend personalized dosages, and the use of pharmacokinetic software for finer dosage adjustment is rare. However, beta-lactam TDM is increasingly being used in CRRT, in line with the guidelines [5, 20]. The 2018 French guidelines suggested performing TDM for ICU patients with expected variability in beta-lactam pharmacokinetics and/or in patients with clinical signs potentially related to beta-lactam toxicity [5]. The 2020 European guidelines recommended TDM of beta-lactams as a standard of care in critically ill adult patients [20]. The fact that 56% of the physicians in our survey reported using full beta-lactam doses during CRRT and 62% mentioned using TDM suggests that beta-lactam TDM is often used only to determine the etiology of the neurotoxicity and that poor availability of TDM technique and the long wait for results from subcontractors limits the use of TDM in practice. This limitation has also been highlighted by other surveys [32, 35, 36, 38]. TDM requires complex analytical systems (such as high-performance liquid chromatography or liquid chromatography/mass spectrometry) associated with high equipment and personnel costs. TDM is mostly performed for meropenem (22%) and piperacillin (17%) but is available for other beta-lactams in less than 5% of hospitals [36]. A lack of access to TDM at the weekend, the perception of a wide therapeutic range, and the relatively recent awareness of the clinical implications of beta-lactam underdosing were major hindrances to implementation [35, 36]. Moreover, MICs for beta-lactams are not measured on a routine basis, which prevents antibiotic therapy from being adapted accordingly. Indeed, European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints are reported by laboratories in categories (i.e. susceptible, intermediate, or resistant) [36, 38]. An international survey of 328 hospitals in 53 countries performed in 2013 showed that TDM of piperacillin/tazobactam and meropenem was rare [35]. A French survey performed in 2015 found that beta-lactam TDM was available to 21% of the 507 respondents [32]. In 2019, a German survey showed that piperacillin, meropenem and ceftazidime were the beta-lactams most commonly dosed with TDM - especially in patients undergoing RRT [38]. Overall, these surveys highlighted differences in beta-lactam TDM practices, patient selection, PK/PD targets, drug assay methods, and dose adjustment strategies.

The diversity of PK/PD targets emphasizes the uncertainties in the literature data, even though most physicians are well aware that a beta-lactam concentration over several times the MIC is required throughout the dosing interval (as already reported by Wong et al. in 2014 [38]). However, few of our respondents answered this question, showing that the concept of dosage adjustment based on PK/PD indices is poorly known (as already described in the ONTAI study [36]). In 2018, the French Society of Pharmacology and Therapeutics and the French Society of Anesthesia and Intensive Care Medicine suggested that targeting a free plasma beta-lactam concentration over four times the MIC of the causative bacteria (or the EUCAST epidemiological cut-off, when the MIC of the isolated strain is not available) for 100% of the dosing interval would maximize the bacteriological and clinical responses in critical care patients, whereas the European guidelines recommend a PK/PD index of between two and five times the MIC [5].

The duration of the beta-lactam infusion was not defined in the questionnaire’s clinical vignettes; this limited the interpretation of our respondent’s practices but highlighted the diversity of dosing regimens used. Although meropenem is the best-studied beta-lactam in CRRT, its dosing regimens differed most significantly among our respondents (Table 1, Supplementary material). This variability was also evidenced in a Belgian study of ICUs and non-ICU wards [34].

Moreover, our survey results underlined the need for procedures and the importance of a multidisciplinary approach for providing stable infusions, since most physicians are not aware of the stability data [32, 34]. The fact that respondents using TDM were more likely to call other specialists and more likely to use prolonged/continuous infusions shows that TDM is part of a comprehensive, complex approach to PK/PD optimization, which also requires determination of the MIC and knowledge of the PK/PD target. The TDM included in Bayesian software represents the best option for personalizing antimicrobial dosing by taking account of various parameters (the MIC, site of infection, weight, renal function, severity, etc.). However, not all the clinical scenarios are available, and the high level of sophistication of these pharmacokinetic tools limits their implementation. Further studies are required to validate these tools and their potential clinical impact. Moreover, the variety of selected PK/PD targets raises the question of whether TDM is useful, given the resulting differences in dose adjustments. The different perceptions of neurotoxicity in patients on CRRT are probably related to the diversity of doses used, as illustrated in the replies to the clinical vignettes.

Our survey had some limitations. Firstly, the questionnaire did not fully reflect the complexity of having to decide on the dosing regimen at the bedside. The clinical vignettes and questions were simple and standardized. Secondly, the low response rate (11%) was explained by recruitment bias, since the survey was conducted during the summer vacation. Thirdly, we did not have an exhaustive list of intensivists in France. However, all types of hospitals (university or not, public or private sector, etc.) were represented. Despite these limitations, we undertook the largest yet study of this type in French ICUs. The diversity of replies to our questionnaire highlights the uncertainties regarding dosage adjustments required in CRRT and the lack of harmonization of PK/PD targets and emphasized the need for further research on a topic that is crucial in critically ill patients.