What are the three pharmacodynamic models for antibiotic action, and what are the corresponding antibiotics?

Learning objective: understand optimal dosing strategies for different classes of antibiotics based on current pharmacodynamic models, and how these models are replacing the older concepts of bactericidal and bacteriostatic effect.

The three pharmacodynamic models and corresponding antibiotics are:

Time-dependent killing: Expressed as fT>MIC (or T>MIC), representing the percentage of the dosing interval that free (unbound) drug concentration remains above the minimum inhibitory concentration.  Main examples:  beta lactams.

Concentration-dependent killing with peak concentration: Expressed as Cmax/MIC, representing the ratio of maximum drug concentration to MIC.  Main examples:  fluoroquinolones and aminoglycosides.

Area under the curve: Expressed as AUC/MIC or AUC₀₋₂₄/MIC, representing the ratio of the 24-hour area under the concentration-time curve to MIC.  Main examples:  Glycopeptides (vancomycin and congeners) and lipopeptides (daptomycin).

These models are replacing the older concepts of bactericidal and bacteriostatic.

The best explanation of this I was able to find was in Chapter 56 of Goodman and Gilman, available through Access Medicine, UAMS library.

What is the correct dosing regimen for piperacillin-tazobactam when used against Pseudomonas infections?

This question comes up repeatedly on wards. In this post, I take a deep dive to provide clarity. 

Learning objectives:

Apply the principles of evidence-based medicine in evaluating and ranking among secondary sources of information.

Implement optimal dosing strategies for P/T for Pseudomonas coverage.

Disclosures and disclaimers:

The P/T dosages mentioned reflect no renal adjustment. (The threshold for renal adjustment for P/T is a creatinine clearance below 40 ml per minute).

I am a co-author of one of the papers cited in this post:

Rationale and evidence for extended infusion of piperacillin–tazobactam

For this post, I searched PubMed, UptoDate, Dynamed Ex, Open Evidence AI, Mandel’s ID textbook, and guidelines for pneumonia, fever/neutropenia, and sepsis.

Definitions

Internal evidence: locally produced information, often embedded in the EMR, for diagnosis and treatment support.

External evidence:  information from outside the institution found in published literature and secondary source platforms.

Primary source:  original articles including systematic reviews and meta-analyses (things you would do your own PubMed search to find).

Secondary source:  repositories of information in which the authors have drawn upon the primary sources and performed critical appraisal, to provide evidence summaries and recommendations.  (Examples: UpToDate and Dynamed).

Pharmacokinetics:  drug blood levels in relation to dose, time, BMI, renal function, and hepatic function (in other words, how the body handles the drug).

Pharmacodynamics:  drug action on receptors and consequent physiological and clinical responses (in other words, what the drug does to the body, pathogen, or other injurious agent).

The controversy:

Many institutions, including our own, have developed in-house protocols for Pseudomonas coverage with P/T, based on pharmacokinetic and pharmacodynamic models. These protocols differ from guidelines and published recommendations as well as product labeling. In this discussion, we will compare the in-house recommendations with the best available external evidence.

First, we must review the applicable principles of evidence-based medicine (EBM).

The most widely accepted definition of EBM comes from this paper on EBM, considered one of its founding documents:

Evidence based medicine: what it is and what it isn't

From the paper:

Evidence based medicine is the conscientious, explicit, and

judicious use of current best evidence in making decisions

about the care of individual patients. The practice of evidence

based medicine means integrating individual clinical expertise

with the best available external clinical evidence from syste-

matic research.

Note my emphasis on the word external.  The use of external evidence is a requirement for EBM.  (EBM in its original notion opposes the use of internal evidence such as pathways, as pointed out in the BMJ paper linked above).

Has this original notion changed?  No.  What’s changed is the available technology, enabling new and better ways to access the best external evidence.  Brian Haynes, one of the founders of EBM, co-authored this update in 2016:

EBHC pyramid 5.0 for accessing preappraised evidence and guidance

Alper and Hanes constructed a hierarchy of information sources in descending order of accessibility for the clinician, but in ascending order of reliability.  (This hierarchy is not to be confused with the commonly cited EBM evidence hierarchy). 

Here is a graphic based on my interpretation of the secondary source hierarchy:

It is important to note that the article explicitly states that EMR embedded support must be based on all the levels below. Most currently available systems in electronic medical records do not fulfill this requirement. Option number 4 from the figure above strikes the best balance of convenience and scientific rigor for most clinicians at the point of care. 

 What does our own in-house protocol recommend?  

P/T 4.5 grams Q 8H with extended infusion.

We'll break that down shortly but for now, to cut to the chase: 

The above recommendation is based on computer models and no high-level evidence. Moreover, guidelines, textbooks, and Up-to-Date do not endorse this recommendation.

Where did the controversy originate?

One of the early studies examining the effect of extended infusion of P/T in patients with Pseudomonas infection was this retrospective cohort study by Lodise et al:

Piperacillin-Tazobactam for Pseudomonas aeruginosa Infection: Clinical Implications of an Extended-Infusion Dosing Strategy

This was one of the early papers supporting extended infusion. The study demonstrated that, all things being equal, extended infusion of P/T (over half the dose interval ) resulted in superior outcomes when compared to the standard short infusion times (30 minutes). This general principle is true and is supported by other studies and computer models. At first glance, the paper would appear to support lower doses (the treatment group received 3.375 G Q 8 hours with the infusion extended over 4 hours).

However, there is a fatal flaw in this study which is revealed in the first paragraph under the results section (emphasis mine):

Baseline clinical characteristics of the study population are presented in table 1. Of the 194 patients who satisfied the inclusion criteria, 92 patients received the traditional intermittent infusion and 102 patients received the extended infusion. Only 4 (4.3%) of the patients who received the intermittent infusion received 3.375 g intravenously every 4 h; the overwhelming majority received infusions every 6 h.

In other words, the comparison group received dosages of P/T that were 25% below the minimum recommended dose specified in P/T product labeling, as well as the recommendations of guidelines and secondary sources.  The study did NOT compare dose reduced extended infusion with the standard dose of 4.5 g Q 6H.  

Before we go on to other studies it is important to point out the widespread confusion between infusion extension and interval extension.  Interval extension is a form of dose reduction. Infusion extension is not necessarily dose reduction.   An infusion can be extended regardless of how short the dose interval.  For example, if one wishes to maintain standard P/T dosing for Pseudomonas, one can administer 4.5 g Q 6H and extend the infusion over three hours.  This was the recommendation of our paper.

What does the literature say overall? 

I recently made the following search query in Open Evidence AI:

There has been some controversy about the appropriate use of zosyn to treat pseudomonas infection.  UptoDate, all guidelines I have seen and product labeling call for a regimen that delivers at least 16 grams piperacillin per 24 hour period.   However, there has been some suggestion that a zosyn regimen containing lower amounts of piperacillin, eg 4.5 grams Q 8 hrs may suffice, the rationale being that if one utilizes an extended infusion protocol one can thereby overcome the shortfall in piperacillin dose.  However, this approach seems based only on pharmacokinetic and pharmacodynamic theory, computer simulations and low level research data such as that reported by Lodise, et al.  Where do we currently stand on this?

Here’s what I got:

The current consensus supports using higher doses of piperacillin-tazobactam (at least 16 grams of piperacillin per 24 hours) for Pseudomonas infections, particularly when administered via extended or continuous infusion. While lower-dose regimens like 4.5 grams every 8 hours (13.5 g/day total piperacillin) with extended infusion may achieve adequate pharmacokinetic/pharmacodynamic targets for some infections, they are generally insufficient for Pseudomonas aeruginosa, especially with higher MICs.

FDA labeling and current guidelines recommend 4.5 grams every 6 hours (18 g/day total piperacillin) for nosocomial pneumonia, where Pseudomonas is a common pathogen.[1-2] The Infectious Diseases Society of America notes that recommended doses have been increased over time to overcome rising MICs in common pathogens, and many institutions now employ extended or continuous infusion strategies.

Concerning that last statement, the extended infusion strategies do NOT translate into dose reductions. In that regard, Open Evidence goes on to say (italics mine):

Extended or continuous infusion improves target attainment but does not fully compensate for inadequate total daily dosing when treating Pseudomonas. Monte Carlo simulations demonstrate that for carbapenem-resistant P. aeruginosa susceptible to piperacillin-tazobactam, doses of 4.5 g every 6 hours as a 3-hour infusion are needed to achieve >90% probability of target attainment at MIC 16 mg/L, while lower doses (3.375 g every 6 hours) fail to achieve adequate exposure.[4] Similarly, in critically ill patients with augmented renal clearance or infections with MIC 8-16 mg/L, only continuous infusions of 12-16 g/day (or higher) reliably achieve therapeutic targets.[5-6]

International consensus recommendations endorsed by IDSA, SCCM, and other major societies support prolonged infusion strategies for beta-lactams to optimize time above MIC.[7] However, these recommendations emphasize that prolonged infusion is an administration strategy to optimize exposure from adequate total daily doses, not a substitute for appropriate dosing. 

Let’s dive into the actual literature. 

We’ve already mentioned the paper by Lodise.  What else is out there?  Our paper, after reviewing the evidence and rationale, favored traditional dosing (4.5 grams P/T Q 6 hours combined with infusion extension (not interval extension) for Pseudomonas infections based on available recommendations up to 2011.  In the discussion section we acknowledged the lack of high level clinical evidence.  We contended that one should follow established guidelines and published recommendations, absent high level clinical data to the contrary.

What has come out since 2011?  Not a great deal.  A 2023 study utilized a Monte Carlo simulation in Pseudomonas isolates that were carbapenem resistant but P/T sensitive.  (A Monte Carlo simulation is an analysis of sensitivity data from a large number of isolates and PK data from a patient population to predict the probability of target attainment with various dosing regimens).

Piperacillin/Tazobactam Dose Optimization in the Setting of Piperacillin/Tazobactam-susceptible, Carbapenem-resistant Pseudomonas aeruginosa: Time to Reconsider Susceptible Dose Dependent

The investigators concluded that traditional recomended dosing (at least 4.5 G P/T Q 6 h) combined with extended infusion was necessary:

Implications: Although susceptible, piperacillin/ tazobactam has reduced potency in CR-PA. If piperacillin/tazobactam is used for susceptible CR-PA, high-doses (4.5 g q6h) and extended infusion (3 hours or continuous infusion) are needed to optimize exposure.

Another study utilized Monte Carlo simulations in a population of patients with febrile neutropenia.  

Population pharmacokinetics and optimized dosing of piperacillin-tazobactam in hematological patients with febrile neutropenia

The authors concluded that when Pseudomonas was the target, a regimen of no less than 16 grams of piperacillin per day was necessary, combined with extended or continuous infusion.

What about guidelines and secondary sources? 

Here’s a list of the ones I searched:

FDA product labeling:  4.5 G Q 6H for Pseudomonas.

IDSA pneumonia guidelines:  At least 4.5 G Q 6H when targeting risk for Pseudomonas.

Surviving Sepsis guidelines:  Antibiotic doses are not specified.

Up to Date:  4.5 G Q 6H for Pseudomonas

Mandel’s ID text:  At least 4.5 G Q 6H for Pseudomonas.

Dynamed:  Gives a range of P/T doses.   I was unable to pin it down even using its AI functionality.

But there’s more!

Critically ill patients, particularly those who are septic, when their arrival creatinine is normal, tend to have supranormal renal function (GFR above their baseline) and consequently more rapid elimination of antibiotics than would be predicted.   In sepsis, this is due to vasodilation and increased cardiac output early in sepsis.  Cytokine effects may also enhance renal elimination.  Also, early fluid resuscitation likely contributes.  

The problem is illustrated in these references:

Augmented Renal Clearance

From the abstract of the above paper:

Augmented renal clearance (ARC) is a phenomenon in critically ill patients characterized by increased creatinine clearance and elimination of renally eliminated medications. Patients with severe neurologic injury, sepsis, trauma, and burns have been consistently identified as at risk of ARC, with mean creatinine clearances ranging from 170 ml/minute to more than 300 ml/minute. Several potential mechanisms may contribute to the occurrence of ARC including endogenous responses to increased metabolism and solute production, alterations in neurohormonal balance, and therapeutic maneuvers such as fluid resuscitation. Augmented renal clearance is associated with suboptimal exposure to critical medications, including β-lactams and vancomycin, increasing the risk of treatment failure.

Individualised antibiotic dosing for patients who are critically ill: challenges and potential solutions

This paper highlights not only increased GFR in such patients, but also increased volume of distribution, which also contributes to decreased blood levels of antibiotics.

These last two papers illustrate how critical it is NOT to err on the side of underdosing with antibiotics.

Finally, given the need for immediate therapeutic levels of antibiotics, the Surviving Sepsis guidelines emphasize a bolus dose of antibiotic, then an extended infusion to commence immediately, not waiting until the next dose time.  I queried Open Evidence in this manner:

For zosyn dosing in pseudomonas infections, when the patient is critically ill, when extended infusion is to be used, given anticipated supranormal GFR (in a patient whose baseline creatinine is normal), it has been recommended that the first dose of zosyn be given as a bolus followed by the extended infusion regimen.  Does this mean that the subsequent dose, (the first extended infusion dose) after the bolus commence immediately, or wait for the 4 hour interval?

Here’s what I got:

The first extended infusion of piperacillin-tazobactam should begin immediately after the initial bolus dose, not be delayed until the next scheduled dosing interval.

Guidelines and pharmacokinetic studies recommend administering a loading (bolus) dose immediately followed by the first extended infusion to rapidly achieve therapeutic concentrations in critically ill patients, especially those with supranormal GFR.[1-3] The Surviving Sepsis Campaign explicitly states that "administration of a loading dose of antibiotic before prolonged infusion is essential to avoid delays to achieving effective beta-lactam concentrations".[1]

Delaying the first extended infusion until the next scheduled interval risks subtherapeutic drug levels during the critical initial period. A 2025 prospective study evaluating first-dose target attainment used a protocol of 0.5-hour loading dose immediately followed by a 3-hour extended infusion, finding that even with this aggressive approach, many patients (38% receiving piperacillin) failed to achieve 100% fT>MIC during the first dosing interval, particularly those with augmented renal clearance.[2] The ACORN trial protocol similarly specified that piperacillin-tazobactam was administered as a bolus for the initial administration followed by extended infusion for subsequent doses, with the extended infusion regimen beginning after the loading dose.[3]

The rationale for immediate transition is that critically ill patients with augmented renal clearance rapidly eliminate beta-lactams, making any delay in achieving sustained therapeutic concentrations potentially harmful. Meta-analyses demonstrate that prolonged infusion strategies improve mortality and clinical cure rates in severely ill patients, but these benefits depend on maintaining adequate drug exposure from the outset of therapy.[4-5]

Conclusion:

Extended infusion of P/T improves antimicrobial effectiveness but does not enable dose reductions below guideline recommended doses for Pseudomonas infections.

A Bayesian approach to the differential diagnosis of wide complex tachycardia

We use sensitivity, specificity, and likelihood ratios (LR) to interpret lab tests all the time. This is known as Bayesian analysis. It is an essential part of the clinical reasoning we do everyday. We  don't ordinarily apply Bayesian analysis to ECG interpretation but the concept is valid. A review published in 2000 outlines the Bayesian approach to the diagnosis of wide complex tachycardia, linked here:

The Bayesian approach improves the electrocardiographic diagnosis of broad complex tachycardia

Like any other uses of the Bayesian approach, one starts with the pretest probability which can be converted into prior odds. We have data on the likelihood ratios of various ECG findings in wide complex tachycardia. To apply a Bayesian analysis one must first determine the pretest probability (or prior odds). The authors posit prior odds of 4 (80% pretest probability) as appropriate for most cases, based on data from a study of unselected patients. They acknowledge that different prior odds may be applied based on clinical judgment and published data in selected patient groups. For example, in an adolescent with WCT a supraventricular mechanism is more likely with a probability of VT of around 40% in published studies, equated to prior odds of 0.67.

Table 2 from the paper, showing various  ECG findings and their likelihood ratios, is shown here.

Pretest odds are multiplied by the LRs of various findings sequentially. The final product is the post test odds. Post test odds of greater than or equal to 1 lean toward  VT. Odds of less than 1 point to a supraventricular mechanism. The further the number is from 1 the greater the strength of the diagnosis. 

When should a Bayesian approach be used as opposed to a conventional approach using algorithms and scoring systems? In WCT with bizarre QRS morphology measurements may be difficult causing problems in application to algorithms. A Bayesian approach may be advantageous in those situations. Moreover, for teaching and learning purposes, Bayesian  analysis provides a good exercise in clinical reasoning as applied to electrocardiography.

VT or not VT?

A fast heart rate with a wide QRS complex (greater than 120 ms) can be scary. Here’s the crucial question to consider when you encounter wide complex tachycardia (WCT):   Is it VT or not VT?  The distinction is clinically important.

If it's not VT it is usually a supraventricular mechanism such as atrial fibrillation, atrial flutter or SVT with aberrant conduction (bundle branches not working properly at tachycardic rates). So it's mainly a binary differential diagnosis between VT and supraventricular wide complex tachycardia (SVWCT). However, a couple of very infrequent exceptions should be noted for which the diagnostic rules to be described herein may not apply:  Some WCTs in Wolff Parkinson White syndrome and suraventricular mechanisms with severe metabolic disturbances, eg TCA overdose. Otherwise for most purposes, in the binary differential diagnosis between VT and SVWCT certain rules are helpful. Through the years many proposed rules and algorithms have been put forth. This has led to some confusion. However, finally they have all been compiled in one place in the review linked below: 

Wide Complex Tachycardia Differentiation: A Reappraisal of the State‐of‐the‐Art

Here is a graphic from the review that highlights some of the key findings, any of which, if present, favor VT. Most of these findings have high specificity but low sensitivity. 

Figure 2 from the review summarizes the various algorithms and other approaches:

Some key points to be gleaned from figures one and two are summarized here:

AV dissociation

This finding has high specificity but low sensitivity. This is when P waves can be seen independently of the QRS complexes. If this is seen it is strong evidence for VT. There are some caveats. First, many cases of VT, (in some series up to half ) exhibit retrograde conduction to the atria (in other words, AV association ). Whan AV dissociation is present, particularly at faster VT rates, it can be difficult to see because of interference from the wide QRS complexes and the ST segments and T waves. 

Morphologic criteria

Does the tachycardia have a typical bundle branch block appearance? If not, it is more likely VT. This test is difficult to apply unless one knows the ins and outs of bundle branch block morphology. (The details of BBB morphology are covered in the archived lecture series).

QRS duration

If the QRS is a greater than 160 milliseconds it favors VT. (If the QRS in V1 is upright, greater than 140 milliseconds favors VT). In general, the wider the QRS the more it favors VT. 

Chest lead concordance 

In the chest leads, (V1-6) if the QRS complexes are all upright (monolithic R waves) or are all pointing down (QS complexes) VT is the diagnosis.

Frontal plane (limb lead) axis

Right upper quadrant axis ( -90  to -180) strongly favors VT. Left upper quadrant and left lower quadrant axis (normal ) less so.

Ventricular activation velocity

If the rate of voltage change of the first 40 milliseconds is less than or equal to that of the last 40 milliseconds of the QRS complex, VT is favored. In addition, if the time from the beginning of an R wave to the nadir of an S wave greater than 60 milliseconds it favors VT. (In the Brugada algorithm that cut off is 100 milliseconds). 

Algorithms

Several algorithms have been proposed and have good test characteristics. In the paper the Brugada algorithm, the Vereckei algorithm, the limb lead algorithm, the Griffith algorithm, and the RWPT algorithm are described.

There are also point scoring systems that perform well. These are discussed in the paper. Another, someone novel method, is the Bayesian approach which will be discussed in a separate post.

You suspect your patient’s chest pain is cardiac. Should you activate the cath lab?

Below is an algorithm that has been discussed on wards several times. It incorporates the new acute coronary syndrome guidelines with a modification based on the OMI (occlusive MI) model.

The new guidelines, which have not adopted the OMI model, are liked here:

2025 ACC/AHA/ACEP/NAEMSP/SCAI Guideline for the Management of Patients With Acute Coronary Syndromes: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines

NSTEMI, the catch-all diagnosis

The diagnostic label NSTEMI (non ST segment elevation MI) tends to be inappropriately applied to many cases of acute troponin elevation in the absence of electrocardiographic ST segment elevation. The result has been that NSTEMI has become a catch all diagnosis, to the point of being devoid of meaning.  Actually, troponin elevations without ST segment elevation have a differential diagnosis including NSTEMI, type 2 MI, takotsubo cardiomyopathy,  PE, and acute nonischemic myocardial injury. This last one is frequently seen in sepsis and other non cardiac critical illnesses. Labeling such instances as NSTEMI is problematic because it can lead to inappropriate and potentially harmful treatments such as systemic anticoagulation.

This problem has been addressed in the literature. The paper linked below from JAMA is somewhat dated but useful.

Increasingly Sensitive Assays for Cardiac Troponins

Figure 1  from the paper is a proposed algorithm.

A good general resource, one with which all internists should be familiar, is the following,

Fourth Universal Definition of Myocardial Infarction (2018)

The JAMA article was published before the most recent addition of the universal definition. It contains references to the third universal definition, now out of date. I should also point out that although not explicitly addressed in the fourth universal definition one should not conflate type 2 MI with NSTEMI, as the latter term implies ACS. Thus terms like “ type 2 NSTEMI”  should be discouraged.