The amount of drug absorbed from the site of administration (e.g., enteral, subcutaneous, and intramuscular) to the systemic circulation is influenced by physicochemical properties of beta-lactam antibiotics (e.g., solubility and molecular size), as well as the properties of the organ/tissue through which the beta-lactam antibiotics are absorbed11. In sepsis and septic shock, reduced gut motility, diminished regional blood flow, and delayed gastric emptying have been suggested to reduce drug absorption12. Intravenous (IV) administration of antibiotics is preferred to account for impaired drug absorption in critically ill patients in the ICU.

Abnormal fluid balance following aggressive fluid resuscitation and capillary leak syndrome can result in the “third spacing” phenomenon and cumulative fluid accumulation13. This can lead to an increase in the volume of distribution, especially for hydrophilic antibiotics, including beta-lactam antibiotics, glycopeptides, aminoglycosides, linezolid, and lipopeptides14-18. The impact is more pronounced for hydrophilic antibiotics compared to lipophilic antibiotics, as the latter already has a larger volume of distribution compared to the former17. A systematic review of clinical studies that evaluated the PK of beta-lactam antibiotics in critically ill patients reported that large volume of distribution differences were commonly observed and most studies reported a 2-fold variation in this PK parameter when compared with the non-critically ill population19. This phenomenon is likely to decrease concentrations of beta-lactam antibiotics, particularly in the earlier phase of the disease. Therefore, higher initial loading doses should be applied in critically ill patients with sepsis or septic shock to compensate for the enlarged volume of distribution. Numerous studies have shown that higher initial loading doses of beta-lactam antibiotics and other antibiotics (e.g., amikacin, colistin, gentamicin, teicoplanin, and vancomycin) are required to rapidly attain effective concentrations in patients with sepsis or septic shock20-25.

Hypoalbuminemia (serum albumin < 25 g/L) is also common in critically ill patients, and this can lead to increased antibiotic distribution and clearance especially for moderately to highly-protein bound beta-lactam antibiotics (e.g., flucloxacillin, ceftriaxone, and ertapenem)26-29. The volume of distribution for highly-protein bound beta-lactam antibiotics, such as ceftriaxone and flucloxacillin, are found to be increased (as much as 90%) in critically ill patients with hypoalbuminemia. However, tissue concentrations remain low due to the “third spacing” phenomenon and cumulative fluid accumulation associated with this patient population. Furthermore, as these beta-lactam antibiotics are also cleared renally, the increase in the free fraction of drugs will also result in rapid drug clearance. The altered volume of distribution and clearance for beta-lactam antibiotics may lead to low antibiotic concentrations particularly at the end of the dosing interval. Maintenance doses for these antibiotics should be increased to compensate for this phenomenon and this is particularly relevant for time-dependent antibiotics.

Several medical interventions in the ICU, such as aggressive fluid resuscitation30, mechanical ventilation31, extracorporeal circuits32, the presence of post-surgical drains33, and total parenteral nutrition34, have also been reported to be associated with enlarged volume of distribution and consequently decreased concentrations of hydrophilic antimicrobials.

Enhanced renal clearance of antibiotics due to elevated glomerular filtration and tubular secretion/reabsorption has been increasingly described in critically ill patients, leading to subtherapeutic concentrations of antibiotics and treatment failure35. This phenomenon is known as augmented renal clearance (ARC), which is defined as glomerular filtration rate (GFR) of above 130 mL/min/1.73 m2 (preferably based on urinary creatinine clearance)36. The phenomenon is likely observed in polytrauma, postoperative and head trauma patients, as well as younger patients, especially in those with lower disease severity37,38. Augmented renal clearance has been strongly associated with suboptimal beta-lactam antibiotic39,40 and vancomycin41-43 exposures, which may partly explain the poor clinical outcomes associated with critically ill patients receiving these antibiotics. Therefore, for these antibiotics, which display time-dependent properties and predominantly cleared by the kidneys, applying altered dosing strategies, such as extended or continuous infusion, may likely maintain effective drug concentrations for a longer duration in critically ill patients with ARC.

On the other hand, reduced antibiotic metabolism and clearance might occur following organ hypoperfusion leading to renal and/or hepatic dysfunction. As the disease progresses in a critically ill patient, myocardial depression may occur leading to decreased organ perfusion and microcirculatory failure. These could cause end-organ damage or in extreme cases, multi-organ dysfunction syndrome32. This syndrome often includes renal and/or hepatic dysfunction that consequently results in decreased antibiotic clearance. Renal dysfunction or acute kidney injury (AKI) leads to reduced GFR and clearance of renally eliminated antibiotics. Nonetheless, dose reduction of antibiotics is not straight forward in patients with AKI, as the decision should consider patient's residual kidney function and fluid status, use of renal replacement therapy (RRT), and the antibiotic PK/PD target44. Beta-lactam antibiotic dosing requirements for critically ill patients are highly dynamic, and regular dosing reviews and modifications are likely required to prevent both suboptimal dosing and development of adverse events.

Mechanical intervention for organ support has been shown to alter antibiotic PK45-49. Mechanical ventilation can alter antibiotic PK by increasing the intrathoracic pressure, leading to a decreased venous return to the heart50. This can lead to an increase in the antibiotic volume of distribution, as well as reduced clearance due to decreased GFR47,51,52. In a study performed by Medellín-Garibay et al., mechanical ventilation was shown to reduce the clearance of vancomycin by 20%46. The decrease in drug clearance was linked to hemodynamic changes in mechanically ventilated patients, which reduced the renal blood flow leading to a reduction in the glomerular function and urine output53. Renal replacement therapy may alter the PK of beta-lactam antibiotics by augmenting antibiotic clearance and volume of distribution54. This is dependent on a few factors including dialysate flow rate, mode of dialysis, type of dialysis membrane, dialysis duration, as well as antibiotic physiochemical properties, and the degree of protein binding55. Patients with AKI receive various forms of RRT, but continuous renal replacement therapy (CRRT) remains the common mode of RRT for critically ill patients in the ICU56,57. CRRT is commonly not applied in a uniform way and therefore, antibiotic clearance may greatly vary and be lower than what has been initially prescribed48,57. No conclusive dosing recommendations can be made currently for critically ill patients receiving CRRT but as a general rule, antibiotics with a high volume of distribution (1 L/kg or greater) and/or that are highly protein bound (80% or greater) are generally poorly eliminated by CRRT58. Current data suggest that a significant proportion of CRRT patients are at an increased risk for either antibiotic underexposure or overexposure48.

Another form of mechanical intervention in the ICU is extracorporeal membrane oxygenation (ECMO). It is an artificial and temporary respiratory and/or cardiac support that is carried out extracorporeally (i.e., cardio-pulmonary bypass) in patients with cardiorespiratory failure refractory to conventional medical therapies59. Its use has increased steadily, especially during the current COVID-19 pandemic, whereby the World Health Organization (WHO) recommends its use in COVID-19 patients with profound hypoxemia (with or without hypercapnia) not amenable to the lung-protective ventilation60. Earlier neonatal and paediatric data suggest that ECMO has the potential to alter the PK of many important antibiotics including beta-lactam antibiotics61. ECMO extracorporeal circuits consisting of conduit tubing provide an additional “compartment” through which beta-lactam antibiotics can distribute62. The circuit tubes, as well as the oxygenator membrane, introduce additional surface areas that the beta-lactam antibiotics can adhere and sequester onto. This is particularly problematic to antibiotics that are lipophilic63, highly-protein bound (e.g., ceftriaxone64), or chemically unstable (e.g., meropenem65). The priming of the ECMO circuit can also dilute and sequester the drug further66. All the above are theorized to lead to an increase in the beta-lactams volume of distribution and possibly treatment failure due to subtherapeutic concentrations65. In addition, ECMO patients have lower drug clearance when compared to patients not undergoing ECMO67. However, based on current clinical evidence: (1) modern ECMO circuits have minimal impact on the PK of most antibiotics, including beta-lactam antibiotics; (2) PK changes in patients receiving ECMO are more reflective of critical illness rather than ECMO therapy itself; and (3) apart from lipophilic and highly-protein bound antibiotics, the impact of ECMO on the PK and dosing requirements is likely to be minimal68-73.

In conclusion, profound alteration of beta-lactam antibiotic PK is common in critically ill patients, especially in those receiving mechanical organ support. Nonetheless, contemporary antibiotic dosing is largely based on dose-finding studies which mostly included healthy participants and patients who are not critically ill. Such a dosing approach has been increasingly shown to increase risks of suboptimal beta-lactam antibiotic exposure in a large proportion of critically ill patients74,75.

Poor beta-lactam antibiotic PK/PD target attainment in critically ill patients has been illustrated in two recently published multicentre clinical studies48,74. In a large point prevalence PK study of beta-lactam antibiotics (i.e., the DALI study) involving 384 patients, up to 500-fold variations were seen in the unbound concentrations of the studied beta-lactam antibiotics74. Of these, 248 (64.5%) patients were treated for infection and 40 (16.0%) of them did not achieve the predefined PK/PD target (50%fT>MIC) and they were 32.0% less likely to have a positive clinical outcome. Similarly, poor target attainment was seen in the SMARRT study, a large prospective, multinational PK study, involving 381 patients on RRT receiving either meropenem, piperacillin-tazobactam or vancomycin48. Up to 55% of the concentrations failed to achieve the low target trough concentrations, with higher failure rates (up to 72.0%) seen with the high target trough concentrations. The trough concentrations were inversely associated with the estimated total renal clearance of the prescribed RRT and residual renal clearance. In addition, highly variable trough concentrations (up to 8-fold) were seen in this study.

Numerous studies have shown better PK/PD target attainment with prolonged (PI) or continuous infusion (CI) of beta-lactam antibiotic compared to intermittent bolus (IB) dosing78,79. The altered dosing strategy increases the percentage of time that free beta-lactams concentrations remains above the target MIC for a given dosing interval, allowing enhanced bacterial killing80. Roberts et al., demonstrated median steady-state concentrations of 16.6 mg/L with CI of piperacillin when compared to median Cmin concentrations of 4.9 mg/L following IB dosing, despite a lower total CI daily dose (25% lower than the IB regimens)79. Noteworthy, systemic review and meta-analyses comparing between CI/PI and IB dosing of beta-lactam antibiotics in terms of survival benefit have shown mixed results81-87. However, when the studies are limited to patients with severe sepsis who received equivalent doses of antibiotics in both arms (IB vs. PI/CI), lower in-hospital mortality was shown in patients receiving CI of beta-lactam antibiotics81.

Prior to the rollout of CI or PI dosing of beta-lactam antibiotics in the ICU, physicians should consider the following: 1) the use of loading dose; 2) antibiotic stability; 3) residual volume or dead space; 4) compatibility with other drugs; 5) drug accumulation in patients with renal dysfunction; and 6) target population and knowledge of susceptibility data for antibiotics aimed for PI program. Application of loading dose would shorten the time to therapeutic exposure88,89. A loading dose is a short-term dose given as an intermittent bolus (30-60 minutes) followed by the total recommended daily dose given as PI or CI90. Infusion duration, beta-lactam antibiotic concentrations (with lower final concentrations post reconstitution being more stable compared to higher concentrations), types of diluents, and container used for reconstitution can affect the beta-lactam stability, and therefore, these need to be considered prior to the beta-lactam antibiotic CI/PI dosing rollout in the ICU. Of the beta-lactam antibiotics that have been studied for PI, both meropenem and imipenem show the shortest duration of stability (4 to 9 hours after reconstitution with water for injection)91,92. Other beta-lactam antibiotics including amoxycillin, benzylpenicillin, and ceftazidime have also been reported to be stable less than 24 hours93-95. Therefore, multiple infusions need to be given over 24 hours when these agents are used. Next, residual volume following beta-lactams infusion can reduce the total amount of beta-lactams given to patients. Bolla et al., showed that more than 10% of antimicrobial would be lost for 26 of 39 studied antibiotics if the residual volume is not infused back to patients at the end of the infusion96. Noteworthy, rapid flushing of the intravenous (IV) line to deliver this “infusion line dead space” to patients is contra to the principle of PI as the residual volume will be delivered in a bolus form rather than prolonged infusion97. Therefore, the residual volume should instead be infused at an appropriate rate with clear administration instructions given to the nurses. Another strategy that can be used to overcome this is by giving higher doses to compensate for the loss from the residual volume98. To the best of our knowledge, there is yet a published PK study assessing the effectiveness of these strategies in improving beta-lactam antibiotic exposures in patients with the infusion line dead space. Another consideration is the beta-lactam antibiotics compatibility with other IV drugs when the beta-lactams cannot be given through a dedicated line. The issue can be handled through three possible ways: 1) co-administration of a drug/drugs that is proven to be compatible with the beta-lactam antibiotics; 2) the placement of additional IV line; and 3) shortening the infusion time of the beta-lactams (i.e., less than 24 hours) with the adjustment of medication administration time to allow the incompatible drug(s) to be administered at different occasions99.