Antibiotic dosing as continuous infusion

Looking at the kill characteristics of antibiotics, one naturally starts to wonder: what if, for the ones that exhibit time-dependent killing, we just give them as an infusion instead of intermittent boluses? Surely their concentration would just stay above MIC, and you would kill more bugs? Considering that several trials of this exact idea had completed in the last decade, it was only a matter of time before the college asked people to critically evaluate them. Question 23 from the second paper of 2020 asked this question more or less exactly, except instead of "critically evaluate", the examiners actually spelled out the structure they wanted (rationale, advantages, disadvantages, evidence). This is refreshing break from having to guess what they want, and will be used here because it makes excellent sense to break the discussion up into these points.

Rationale

• Clinical efficacy of β-lactam antibiotics (and others which kill in a time-dependent manner) depends on time spent at a site concentration above MIC
• Specifically, MIC must be exceeded 40-70% of the time for optimal efficacy (Nicolau, 2008), though it does vary somewhat depending of which class of drug you look at (MacVane et al, 2014, give 40% for carbapenems, 50% for penicillins and 70% for cephalosporins)
• This may be difficult to achieve in septic patients who have variable erratic pharmacokinetics (clearance rate may be much higher than predicted by models)
• Inadequate dosing is an important predictor of treatment failure and can lead to the development of resistance (Martinez et al, 2012)
• High peaks seen with intermittent dosing can increase the risk of drug toxicity
• Increasing the frequency and decreasing the dose to avoid these peaks increases the nursing workload and can paradoxically increase the risk of bloodstream infection because of increased line/cannula access handling
• In the ICU specifically, continuous infusion dosing seems to make more sense, because:
  • The microbes are often more resistant, and therefore
  • The MIC is higher, and thus
  • The toxicity of the necessary drug doses would be higher, but
  • The patients have more fragile organ systems already damaged by the disease process, more susceptible to toxicity, with altered clearance kinetics

Advantages

• Should improve outcomes
  • Time above MIC will be longer, which would lead to more rapid rates of clinical improvement and decreased morbidity
  • As sepsis has a substantial mortality associated with it, this should translate into decreased mortality from sepsis
  • Because drug peak levels are associated with drug toxicity, continuous infusion should prevent drug-related adverse effects
• Logistical/economic advantages
  • There may be an advantage in terms of nursing/pharmacy workload, as setting up a 24 hour infusion pump is less labour-intensive then making and administering four or six intermittent doses every day. 
  • The decreased staff requirements make it possible to complete these infusions in an outpatient setting, decreasing healthcare costs (or, at least shifting them to the patient and the community)
  • The total dose of the drug may be lower, as clinical resolution may occur sooner, decreasing costs
• Advantages for antibiotic stewardship
  • Less underdosing and sub-MIC concentrations should produce a reduction in  the rate of antimicrobial resistance
  • This decreases the pressure on the antibiotic pipeline, as we should still be able to use existing agents at higher (continuous) doses

Disadvantages

• Argument from pharmacokinetics:
  • The intermittent boluses may achieve a higher concentration temporarily which would increase the concentration gradient between the blood and the infected site. Often, these sites can be devitalised or deep (abscess, infected bone, etc) and the concentration gradient drives diffusion. Craig & Ebert (1992) list several animal studies where drug tissue penetration was examined, and intermittent dosing appears to have the advantage.
• Argument from pharmacodynamics
  • "Time-dependent killing" is an oversimplification of the drug effect; many antibiotics exert their effect by binding to and disabling microbial enzymes, and the binding is with sufficiently high affinity that it leads to a sustained "post-antibiotic effect". 
  • The extent of this post antibiotic effect would therefore be dependent on the maximum concentration you can achieve, i.e. the more drug molecules you give, the more bacterial ribosomes you disable, i.e. this is concentration-dependent killing. 
  • Many drug classes exhibit these effects; notably penicillins and carbapenems -especially for gram-positive bugs (Rodvold, 2012). 
• Argument from logic: We say that administration by infusion should aim to achieve effect site concentrations above MIC; however:
  • Nobody seems to agree how far above MIC your concentration should be, and published data uses inconsistent endpoints (usually around 4 × MIC)
  • We rarely know what the MIC is in the early stages of treatment, and this information may never become available (particularly if nothing is ever cultured)
  • Even if we had MIC information available, the use of this data would require for us to measure effect site antibiotic level to make sure we are above MIC. That may not be possible; in fact even just serum β-lactam levels are not widely available.
  • To just administer a higher dose in hope that it is above MIC would be to waive the theoretical advantages of reduced toxicity with infusions
• Logistical disdvantages
  • Not all drugs are suitable for a 24-hour infusion because of their stability in room temperature solution, and so there may be no benefit in terms of nursing/pharmacy workload 
  • The need for 24 hour infusion usually requires secure IV access, which usually means central access, whereas intermittent dosing can usually be achieved through peripheral cannulae
  • The 24 hour infusion usually takes out one lumen of a line, as the antibiotics are often not compatible with anything

Evidence

This is obviously a hot topic, or at least a topic where people think they can sneak an easy publication. Thabet (2021) performed an review of reviews, hoovering up twenty-one systematic review papers from all over the place. 

Let's narrow things down slightly, and look at just β-lactams, and just sepsis or septic shock. This was the subject of a meta-analysis by Kondo et al (2020).Thirteen RCTs have been performed since 2000 in this field. Some of the more influential ones are listed here, mainly because some of the participants and co-authors were college examiners:

• Dulhunty et al, 2013: multicentre Australian trial, n=60. There was no difference in any of the outcomes, but the study was not powered to detect a difference in mortality.
• BLING II (Dulhunty et al, 2015): multicentre Australian trial, n=432. No difference in any of the outcomes.
• BLING III
• BLISS (Abdul-Aziz, 2016): unblinded trial with short follow-up; n=140. The continuous infusion group had more ventilator-free days and improved biomarkers, but there was no difference in mortality or length of ICU stay. The main outcome of this study was to demonstrate that the pharmacokinetics of beta lactams were clearly superior when given by infusion. 
• BLING III - 2024 - continuous vs intermittent β-lactams, n=7202 in 7 countries. No difference in mortality, but significally higher rates of clinical cure (antibiotic-free 48 h after the end of the course).

In short?

• The evidence for mortality benefit is conflicting 
• The larger trials with robust methodology have not produced any mortality benefit or ICU length of stay improvement
• The overall literature is heterogeneous, with everybody using different antibiotic regimens, infusion standards, MIC targets, patient populations and follow up strategies. 
• One thing which has emerged is a consistent signal that there is no harm with continuous infusion
• Of the (admittedly, low quality and largely observational) studies that report better outcomes with continuous infusions, the populations which seem to benefit the most are critically ill patients, i.e. low acuity patients do not seem to benefit.