Cefepime vs Piperacillin-Tazobactam in Adults Hospitalized
With Acute Infection

Qian ET. JAMA 2023;330:1557-1567. doi:10.1001/jama.2023.20583

Clinical Question

  • In acutely ill adults (ED / ICU) does the administration of cefipime or piperacillin-tazobactam (pip-taz) result in an increased incidence of AKI or death by day 14?


  • Sepsis is one of the most common reasons for admission to ICU and as such antibiotics are frequently used in the critical care environment
  • The use of anti-pseudomonal antibiotics is frequently recommended by guidelines
  • Despite their multiple benefits, antimicrobial usage can cause patient harm especially with altered pharmacokinetics in critical illness
  • In particular there have been concerns regarding both pip-taz and renal dysfunction, and cefepime and neurotoxicity however the evidence for this is mixed and low-grade
  • There are limited randomised trials that seek to evaluate the risks of antibiotic associated harm


  • Pragmatic, open-label, parallel-group, randomized trial
  • Investigator-initiated
  • Randomised without stratification in 1:1 ratio
  • Use of clinical decision tool in EMR to identify patients and subsequently randomise and place appropriate prescription of antibiotic based on eGFR
  • Data predominantly extracted from EMR
  • Clinicans asked to record reason for cessation / commencing unassigned antibiotic for 7 days following randomisation
  • Primary outcome was a novel 5 stage ordinal scale of AKI or death:
    • 0 = no new or worsening AKI
    • 1 = Creatinine (Cr) 1.5-1.9x baseline or increase y ≥0.3 mg/dL [≥26.5 μmol/L])
    • 2 = Cr 2.0 – 2.9x baseline
    • 3 = Cr ≥ 3.0 x baseline, Cr ≥4.0mg/dL [≥353.7 μmol/L] or new RRT
    • 4 = death
  • Pre-specified secondary outcomes
  • Initially, 2050 patients required to provide an OR 0.65 in the primary outcome
    • This was increased to 2500 to account for higher use of vancomycin and powered to show an OR 0.75 in cefipime group compared to pip-taz group
      • An OR of 0.75 would show an absolute difference of 5% in patients with AKI or death (70% no AKI, 6% death)
      • An OR < 1.0 indicates better outcome with cefipime
  • Ethical waiver of consent granted
  • Independent DSMB


  • Single centre, USA
  • November 2021 – October 2022


  • Inclusion:
    •  ≥ 18 years old and located in a participating emergency department or medical intensive care unit
    • < 12 hours from presentation to study hospital
    • Treating clinician initiating an order for an anti-pseudomonal cephalosporin or antipseudomonal penicillin
  • Exclusion:
    • Known receipt of > 1 dose of an anti-pseudomonal cephalosporin or anti-pseudomonal penicillin during the last 7 days
    • Current documented allergy to cephalosporins or penicillin
    • A prisoner
    • Lack of equipoise for need for an anti-pseudomonal antibiotic
  • 3806 assessed –> 2634 randomised –> 1277 cefipime and 1357 pip-taz
    • 123 excluded: 119 did not receive drug as allocated and 4 prisoners
  • Comparing baseline characteristics of cefipime vs. pip-taz group
    • Age: 57 vs 59
    • Male: 57 vs 58%
    • Hours from presentation to enrollment: 1.3 vs 1.1 hours
    • Location of enrollment:
      • ED: 94 vs 96%
    • Suspected site of infection:
      • Lung: 21 vs 23 %
      • Abdominal: 26 vs 23%
      • Skin and soft-tissue: 17 vs 19%
    • Sepsis (as per Sepsis 3): 54 vs 54%
    • SOFA: 2 vs 2
    • Mechanical Ventilation: 9 vs 7%
    • Vasopressors: 13 vs 13%
    • Vancomycin on enrollment: 78 vs 77%
    • CKD: 20 vs 20%
    • AKI at enrolment:
      • Nil: 51 vs 50%
      • Stage III: 12 vs 11%
    • Coma: 7 vs 6%
    • Delirium: 5 vs 4%

Intervention (Cefipime)

  • 2-g intravenous push over 5 minutes every 8 hours
    • Median 3 days duration

Control (Piperacillin-Tazobactam)

  • 3.375-g bolus over 30 minutes for the initial administration followed by an extended infusion of 3.375 g every 8 hours infused over 4 hours for subsequent doses
    • Median 3 days duration

Management common to both groups

  • No other anti-pseudomonal antibiotics allowed
  • Duration of therapy and addition of other antibiotics (e.g. vancomycin) at clinician discretion
  • Dedicated clinical pharmacists in both ED and ICU


  • Primary outcome:
    • AKI (any stage) or death by day 14
    • OR 0.95 (95% CI 0.80 – 1.13)
      • Death: 7.6 vs 6.0% (Risk Difference 1.6%, 95% CI -0.5 to 3.6)
    • No difference with multiple sensitivity analyses (eTable 11, including per protocol, differing durations of anti-pseudomonal coverage, and receipt of vancomycin at baseline)
  • Secondary outcomes:
    • Comparing cefipime vs pip-taz
    • No significant difference in
      • Major Adverse Kidney Events (MAKE) at day 14: 10.2 vs 8.8% (Risk Difference 1.4%, 95% CI -1.0 to 3.8)
        • No differences between all components (death, RRT, final creatinine ≥ 2 x baseline)
    • Significantly worse in cefipime group
      • Delirium and coma free days at day 14: Median 14 vs 14
        • OR 0.79 (95% CI 0.65 – 0.95)
        • Delirium or coma incidence: 20.8 vs 17.3% (Risk Difference 3.4%, 95% CI 0.3 – 6.6)
        • Still significant at day 28 (exploratory analysis)
  • No difference in vasopressor-free days, ventilator-free days, ICU free-days and hospital-free days by day 28 (exploratory analyses)

Authors’ Conclusions

  • In hospitalised adults treatment with pip-taz as opposed to cefipime did not increase the incidence of AKI or death. Cefepime use resulted in more neurological dysfunction


  • Large study addressing an important and frequently used intervention with limited randomised evidence
  • Strong internal validity – randomisation, allocation concealment, well balanced baseline characteristics, detailed screening log and majority received intervention  as randomised (>95%)
  • Leveraging the EMR to recruit, randomize, prescribe intervention and collect data is impressive
    • Given the duration from commencement of randomisation to final follow up was under a year and over 2,500 patients were randomised at a single site, the use of an EMR in this manner will likely be of increasing importance for clinical research
  • Multiple sensitivity analyses for both the primary outcome and days alive and free of delirium and coma within 14 Days enhances robustness


  • Could the primary outcome be more patient centric?
    • Approximately 80% of patients in both groups had no AKI or a Stage I AKI at day 14
    • If a patient has a transient rise in creatinine that then subsequently improves it may be argued this is not a truly patient focused outcome
    • However the counter argument would be that death was included as part of the ordinal scale, secondary outcomes were relevant and meaningful, and the clinical concern around pip-taz use was around renal impairment.
  • Single centre
    • This may mean some groups of patients are under-represented compared to other sites, e.g. stem cell transplants 1.3 vs 1.2% (eTable 1)
  • Unblinded
  • Approximately 20% of patients received alternate antibiotic within the first 14 days (eTable 7), however exposure was similar in both groups
  • Median duration of antibiotic exposure short and only ~50% of the population retrospectively deemed to meet Sepsis III criteria by study investigators
    • This may bias towards the null but is reflective of the undifferentiated nature of ED presentations which made up a large proportion of randomisations
  • Relatively low rates of critically unwell patients

The Bottom Line

  • In my practice I have usually used pip-taz unless there is a specific clinical indication to use cefipime and I will continue to do so following this trial
  • If using cefipime it is important to stay attendant to the potential for neurotoxicity, especially in the critically ill who may have altered pharmacokinetics or blood brain barrier disruption

External Links


Summary author: George Walker @hgmwalker89
Summary date: 17th December 2023
Peer-review editor: David Slessor

Picture by: Petr Ganej / Pexels


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.