Critical Thinking in Critical Care Medicine
Cefepime vs Piperacillin-Tazobactam in adults hospitalized with acute infections
Find the original article!
Summary:
The ACORN Trial randomized 2511 patients in the emergency department or ICU to either zosyn or cefepime in a 1:1 ratio. The objective was to determine if the choice affected the risk of acute kidney injury (AKI) or encephalopathy at 14 days.
PICOTT:
Population: 2634 Randomized patients (2511 analyzed) at Vanderbilt University Medical Center in the US. 54% met criteria for sepsis.
Intervention: Cefepime
Comparison: Piperacillin-Tazobactam
Outcomes: Acute kidney injury (No difference). Encephalopathy / delirium was higher in the Cefepime group OR 0.79 (95%CI, 0.65 to 0.95). Absolute difference 3.4% (0.3%-6.6%). [Mean (SD),11.9 [4.6] days vs 12.2 [4.3] days]
Type of question: HARM
Type of Study: RCT
Overall Risk of bias: Low to moderate due to the patients excluded from analysis.
Blinding of clinicians was not done. Others are not specified.
Co-interventions include the other antipseudomonal antibiotic (same in both groups) and vancomycin co- administration (also similar).
Even though it mentions intention to treat, they didn't consider the patients that did not receive the intervention at all. They did not analyze 4.66% of patients randomized. They are therefore a modified intention to treat / per protocol. In a regular RCT it would mean a high risk of bias. In this case, I did not assign high level of risk of bias because their primary outcome is harm. For harm it makes more sense to analyze per protocol, as the harm is linked to the intervention given.
Clinical Significance:
There is no difference in the incidence of acute kidney injury in both groups. Therefore, the AKI seen is due to the underlying process and not a treatment's side effect, as originally feared.
The encephalopathy risk is statistically significant, even though the difference in days free of delirium is only hours, there was a 3.4% absolute difference, translating to a number needed to harm of 29. For every 29 patients I give cefepime to, I will have 1 extra episode of delirium or coma.
Pearl: Harm should not be the main question in an RCT, it is usually unethical. This case is slightly different, as the benefits are at equipoise and there was a change already made based on very low-quality data. The potential benefit in this case might justify it, but in general we should still be very careful about using RCTs primarily designed to test for harm questions. On the other hand, harm as a side effect can be described in the study and the information would be high quality.
Context: Recently some observational data pointed towards the possibility of the combination of zosyn and vancomycin causing acute kidney injury. There was a big push to switch from zosyn to cefepime in ICU patients, even thought it could be just a coincidence (AKI happens in sepsis and septic shock) and not a causation (Zosyn causes AKI)
Mostly Settled Science - Unlikely to change. Even though it is only 1 RCT, it is fairly well done and has large number of events, making a different outcome in subsequent studies unlikely.
