Background
Author: Nick Scott, MD
Editor: Naillid Felipe, MD
- Empiric antibiotic treatment for infections is a staple of emergency medicine. Standard broad-spectrum regimens for patients include vancomycin and cefepime and vancomycin and piperacillin-tazobactam.
- Both regimens have their risks; observational trials have observed an association between cefepime exposure and neurologic dysfunction, such as agitation and coma, while piperacillin-tazobactam, particularly in combination with vancomycin, has been associated with acute kidney injury.
- The two combinations cover a similar range of pathogens, with the exception being cefepime’s lack of coverage for anaerobic bacteria. Additionally, piperacillin-tazobactam has limited ability to cross the blood-brain barrier, which restricts its efficacy in central nervous system infections.
- Two studies published within the last year, the Antibiotic Choice on Renal Outcomes (ACORN) trial and Mortality of Patients with Sepsis Administered Piperacillin-Tazobactam vs. Cefepime (written by Chanderraj et al.) sought to determine the safest regimen, but the two papers produced conflicting answers.
Clinical Question
- ACORN – In patients presenting to the emergency room with infection, does treatment with piperacillin-tazobactam result in higher rates of acute kidney injury when compared to treatment with cefepime?
- Chanderraj – In patients presenting to the emergency room with infection, which empiric antibiotic regimen – vancomycin and cefepime or vancomycin and piperacillin-tazobactam – results in lower mortality?
Population
- ACORN – 2511 adults (age 18 or older) in the ED or the medical ICU at Vanderbilt University Medical Center, for whom a clinician ordered an anti-pseudomonal antibiotic within 12 hours of presentation to the hospital.
- Chanderraj – 7569 adults (age 18 or older) who presented to the ED at the University of Michigan had evidence of acute organ dysfunction and received vancomycin and either cefepime or piperacillin-tazobactam within 24 hours of ED presentation.
Intervention
- Both studies studied the same intervention – the addition of an empiric antipseudomonal antibiotic, cefepime or piperacillin-tazobactam.
Outcomes
ACORN
- Primary outcome: development of an acute kidney injury, which was defined by the KDIGO cutoffs, or death, within 14 days of enrollment, measured on a 5-point ordinal scale (with 0 being no AKI, 1-3 being the three stages of AKI, and 4 representing death)
- Two secondary outcomes: (1) a composite of death, need for kidney replacement therapy, and persistent kidney dysfunction at 14 days, and (2) number of days alive and free of coma or delirium.
Chanderraj
- Primary outcome: all-cause mortality within 90 days.
- Secondary outcome: organ-failure-free days, defined as the number of days within the first 28 days following initial presentation without the receipt of vasopressors, mechanical ventilation, or dialysis.
Design
- The ACORN trial was a single-center, open-label, randomized trial.
- The Chanderraj study was a single-center retrospective cohort study, which included a period of piperacillin-tazobactam shortage used as an instrumental variable. The instrumental variable analysis aimed to predict the effect of selecting cefepime over piperacillin-tazobactam simply because cefepime was not available in an attempt to emulate a randomized study.
Excluded
- ACORN – Patients were excluded if they had a direct indication for treatment with one antibiotic over the other, received an antipseudomonal antibiotic within 7 days of enrollment, had an allergy to either antibiotic or were incarcerated.
- Chanderraj – Patients were excluded from the analysis if they had necrotizing, intra-abdominal, head, and neck infections (indications for anti-anaerobic coverage) or CNS infections (indications for CNS penetration). In order to be excluded, this indication needed to be identified within 24 hours of presentation.
Primary Results
- ACORN – No differences in AKI, death, or the composite secondary outcome at 14 days. Patients randomized to cefepime were more likely to develop coma or delirium (20.8% vs. 17.3%) and had fewer delirium-free days (11.9 vs 12.2) during the first 14 days. Both of these differences were statistically significant.
- Chanderraj – Treatment with piperacillin-tazobactam resulted in a 2.6% absolute increase in 90-day mortality (p=0.01) in the unadjusted analysis and a 5% absolute increase in mortality in the instrumental variable analysis (p=0.002). They also analyzed outcomes at 14 days (to compare directly to the ACORN trial results) and found no significant differences in any of their primary and secondary outcomes.
Strengths
- ACORN has the advantage of being a randomized trial. They met their pre-specified enrollment goal and had an excellent follow-up rate of over 95%.
- Chanderraj et al. rigorously examined the internal validity of their instrumental variable analysis, and their instrumental variable (the period of piperacillin-tazobactam shortage) was well above their pre-specified thresholds for relevance, exclusion restriction, and independence. In other words, it simulated a randomized controlled trial very well.
Limitations
- ACORN had considerable crossover (nearly 1 in 5 patients crossed over during the study period). It is also possible (and suggested by the findings of the Chanderraj study) that the follow-up period of 14 days needed to be longer to capture differences in important clinical outcomes between the two groups.
- ACORN also allowed clinicians to opt patients out of randomization by stating that there was a specific indication for one drug over the other without identifying the indication. For this reason, many patients (over a quarter of those otherwise eligible) were excluded.
- Chanderraj excluded patients with CNS infections and/or indications for anaerobic therapy that were identified within 24 hours of presentation. It is unclear how many of these indications for coverage were discovered after the empiric antibiotic regimen was already decided. This exclusion criterion tossed out nearly a quarter of patients who received empiric antibiotics during the study period.
- Both studies have limited generalizability as they were both performed at single sites. It’s possible that some of the differences between the two sets of results could be explained by differences in resistance patterns to the two antibiotics between the two sites.
Author's Conclusions
- The ACORN authors concluded that treatment with piperacillin-tazobactam did not increase the incidence of acute kidney injury and that treatment with cefepime resulted in more neurologic dysfunction.
- Chanderraj et al. concluded that administration of piperacillin-tazobactam to patients with sepsis without a clear indication for anti-anaerobic therapy results in higher mortality and a greater degree of organ dysfunction compared to cefepime, quantified at one additional death for every twenty patients treated.
Our Conclusions
- A key factor in reconciling the differences in results between these two studies is that the populations studied in these two trials were far from identical. The Chanderraj study included a much sicker group of patients, with more patients requiring ICU admission and all patients receiving vancomycin. Only about half of the patients in the ACORN trial even met the criteria for sepsis at the time of randomization.
- It is very difficult to draw any practice-changing conclusions from the ACORN trial, primarily because of its nebulous exclusion criteria. I’m not reassured by the lack of difference in AKI rates between the two groups – since exclusion was at physician discretion, it’s possible some patients were excluded because the physician was concerned they may develop an AKI and didn’t want them randomized to receive piperacillin-tazobactam. Additionally, not everyone in the study received vancomycin. These factors could’ve muffled the signal for increased AKI in patients who received piperacillin-tazobactam. Additionally, the magnitude of association for neurotoxicity, which translates to a few extra hours of delirium during the hospital stay, is relatively slim, so much so that it’s unclear how clinically significant and/or patient-centered this result is.
- On the other hand, the magnitude of the mortality association with piperacillin-tazobactam in the Chanderraj trial is astronomical. With the amount of patients we see with sepsis in the ED, a 5% absolute increase in mortality could translate to thousands of additional deaths per year. My largest reservation with this paper was also related to their exclusion criteria, particularly the way they retrospectively excluded patients with indications for anti-anaerobic coverage that may or may not have been discovered after an empiric antibiotic regimen was already started. I think it is safe to say that based on the Chanderraj study, patients who have zero chance of having any indication for anti-anaerobic coverage (perhaps those with a clear individual source identified before antibiotics are started) are probably better off receiving cefepime.
- The question still remains…why did the patients who received piperacillin-tazobactam in the Chanderraj study do worse? Could it have been related to the increased incidence of AKIs that the ACORN trial didn’t pick up on because of its limitations outlined above? Is it simply related to local resistance patterns and, therefore, not generalizable? Or is there something inherently harmful about empiric anti-anaerobic coverage? Teasing out this mechanism could give us greater confidence in choosing the safest empiric antibiotic regimen for the patient in front of us.