Pneumonia after cardiac arrest is problematic for several reasons:
- Many patients aspirate during cardiac arrest.
- Patients are intubated and mechanically ventilated, usually with poor mental status – conditions which do not promote secretion clearance.
- Early diagnosis of pneumonia is impossible for many reasons (inability to measure a fever due to therapeutic temperature monitoring, inability to report symptoms due to intubated/sedated status, masking of mild hypoxemia because patients are on mechanical ventilation already).
- Hypothermia to 33C may impair neutrophil function, thereby increasing the risk of infection.
Consequently, one persistent controversy regarding post-arrest management is whether patients should receive prophylactic antibiotics to prevent pneumonia. This has gone in and out of vogue over time. A while ago, there was a bit more enthusiasm for it. One very small single-center study did find benefit from a three-day prophylactic course of ampicillin-sulbactam.1 Currently, most literature argues against prophylactic antibiotics. Overall, this topic is ripe for a RCT.
ANTHARTIC study: design
This is a multicenter, double-blind, placebo-controlled trial of patients resuscitated from shockable out-of-hospital cardiac arrest.2 Patients were randomized to placebo versus intravenous amoxicillin-clavulanate for two days. The primary endpoint was the incidence of ventilator-associated pneumonia within a week (as adjudicated by an independent, blinded committee).
Exclusion criteria are as shown below:
Based on these criteria, about two hundred patients were included (below). Patients were well matched at baseline:
The primary endpoint (early ventilator-associated pneumonia) was less common in patients treated with antibiotics. I recalculated the statistics using a Fisher Exact test, and the numbers check out:
- The Fisher exact test p-value is 0.023 (a smidge lower than the value reported in the study, which used a slightly different statistical test).
- The fragility index is 2, which is fairly good for a study this size (yes, I know, fragility index is controversial, but I still like to consider it along with the multitude of other data in the paper).
Antibiotic treatment also reduced the overall incidence of ventilator-associated pneumonia at any timepoint (23% vs 39%, p = 0.02, fragility index of 3). This suggests that a two-day course of antibiotics didn’t merely temporarily suppress infection (with a rebound effect later on in the patient’s hospital course).
So it’s fair to say that this is a “positive” trial – which is uncommon in critical care research these days.
Given the relatively small size of this trial, antibiotics didn’t affect hard endpoints (such as mortality). There was a very weak trend towards more ventilator-free days in patients treated with antibiotics, suggesting that a lower risk of ventilator-associated pneumonia might potentially translate into patient-centered outcomes.
There was no sign of harm due to antibiotics. For example, no patients in the antibiotic arm developed C. Difficile infection. Fecal cultures didn’t show any differences in the carriage of multidrug-resistant bacteria. The study is under-powered to evaluate for rare harms due to antibiotics, but this is nonetheless reassuring.
overall take on the study
The study’s strengths and weaknesses might be summarized as follows:
- Multi-center design with appropriate blinding.
- Addresses a meaningful clinical question about which equipoise exists.
- Represents the highest level of evidence available to answer this question.
- Primary endpoint shows benefit.
- Excludes patients with gross aspiration identified during laryngoscopy (inclusion of these patients might have further increased the benefit seen from antibiotics).
- Primary endpoint isn’t a patient-centered or particularly “hard” endpoint (committee-adjudicated diagnosis of ventilator-associated pneumonia).
- Statistics don’t entirely exclude the possibility that results could have occurred due to chance (e.g. see discussion of whether the p-value cutoff should be <0.05 or <0.005 here).
- Inclusion was restricted to shockable out-of-hospital arrests, which may limit generalizability to other situations.
Overall, this was a well-designed study which may guide practice. The study doesn’t prove that antibiotics reduce ventilator-associated pneumonia with 100% certainty. As such, further research into the topic would certainly be reasonable. However, at this juncture it seems fair to say that the potential benefits of a two-day antibiotic course are very likely to out-weigh any potential risks from this treatment.
some comments on applying these study results
The investigators used a two-day course of intravenous amoxacillin-clavulanate. This antibiotic isn’t available in IV form in the United States, but ampicillin-sulbactam could easily be substituted for it. Ampicillin-sulbactam is a good choice here because it covers community-acquired gram-positive and gram-negative organisms (without causing much C. difficile).
For patients with an allergy to ampicillin-sulbactam, a cephalosporin monotherapy would probably be fine (e.g. ceftriaxone 1 gram IV daily for two doses). Please note that anaerobic coverage is not needed for the treatment of pneumonia (unless there is empyema or lung abscess).
generalization to other arrest contexts?
The study investigated only outpatient with out-of-hospital shockable arrest. This raises some questions regarding generalizability to other clinical contexts:
- Generalization to out-of-hospital non-shockable arrest seems fairly reasonable. A similar bacterial flora is likely involved. Most principles of post-arrest care seem generalizable between shockable and non-shockable arrest patients.
- Generalization to in-hospital arrest is more tenuous. Patients with in-hospital arrest are likely to be colonized with more resistant organisms. This raises a question of whether more powerful antibiotics would be needed (antibiotics with potentially greater side-effects).
generalization to Therapeutic Temperature Monitoring at 36C?
All patients treated in this study received therapeutic hypothermia (with an initial target temperature of 33C). Maintaining patients at 33C may impair neutrophil function and thereby increase the risk of infection. Thus, it’s possible that prophylactic antibiotics could be more beneficial among patients maintained at 33C (compared to patients treated at 36C).
Ideally, this study should be replicated among patients who are being treated at 36C. However, currently it seems reasonable to apply the study results to patients treated at 36C.
generalization to extracorporeal life support (ECMO)?
Patients resuscitated using ECMO were excluded from this study. Whether these results could be generalized to ECMO is a complex question with no clear answer. If prophylactic antibiotics were used for a patient on ECMO, ampicillin might not be a great choice (due to sequestration of this medication within the circuit). Ceftriaxone might be a better selection.3
- Among patients revived from shockable out-of-hospital arrest, a two-day course of amoxicillin-clavulanate reduced the rate of ventilator-associated pneumonia.
- This is a multi-center, placebo-controlled RCT with no major methodological flaws. It’s not statistically unassailable, but it represents the best available evidence on the topic.
- It is reasonable to change practice based on this study (while simultaneously maintaining an open mind on the topic and awaiting additional data).
- IBCC chapter on post-arrest care.
- 1.Acquarolo A, Urli T, Perone G, Giannotti C, Candiani A, Latronico N. Antibiotic prophylaxis of early onset pneumonia in critically ill comatose patients. A randomized study. Intensive Care Med. 2005;31(4):510-516. https://www.ncbi.nlm.nih.gov/pubmed/15754197.
- 2.François B, Cariou A, Clere-Jehl R, et al. Prevention of Early Ventilator-Associated Pneumonia after Cardiac Arrest. N Engl J Med. November 2019:1831-1842. doi:10.1056/nejmoa1812379
- 3.Cheng V, Abdul-Aziz M, Roberts J, Shekar K. Optimising drug dosing in patients receiving extracorporeal membrane oxygenation. J Thorac Dis. 2018;10(Suppl 5):S629-S641. https://www.ncbi.nlm.nih.gov/pubmed/29732181.