Reply: Antimicrobial Prophylaxis in Extracorporeal Membrane Oxygenation: Is the Debate Still Open?

From the Authors:

We would like to thank Dr. Montrucchio and colleagues for reading and providing valuable comments on our study, in which we investigated the efficacy of prophylactic antibiotics during extracorporeal membrane oxygenation (ECMO) in a Japanese nationwide database (1). Our results showed that prophylactic antibiotic treatment during ECMO was associated with reduced in-hospital mortality and a lower incidence of nosocomial pneumonia. We herein provide our responses to the concerns raised by Dr. Montrucchio and colleagues.

First, we defined the prophylactic antibiotics group as patients who received cephalosporin or glycopeptides within 2 days after starting ECMO. We agree that the timing of antibiotics may alter the efficacy of prophylaxis. However, our definition of prophylaxis (antibiotic administration within 2 days after the intervention) is in line with previous studies (2, 3).

We also agree with their second comment that prophylactic and empirical antibiotic therapies were very difficult to distinguish. In our study, some patients with infection may have been included in the prophylactic group, and we acknowledge that this may have caused misclassification bias. Although this bias may work toward reducing the observed effect, our findings showed a significant difference in mortality between the groups. That is, proper adjustment for this bias may not alter the conclusion.

We also agree with their third comment that there is an unclear correlation between the use of the selected antimicrobials and the prevention of nosocomial pneumonia. We only hypothesized that prophylactic antibiotics reduce nosocomial pneumonia by potentially reducing oral bacterial colonization. The mechanism should be investigated in future studies.

We have performed two sensitivity analyses using propensity score matching to clarify the above points. We first defined the prophylactic group as patients who received cephalosporin or glycopeptides within the first 24 hours after starting ECMO. We then defined the prophylactic group as patients who received all kinds of antibiotics. The results are shown in Table 1; the trends in the data are similar to those in the original results. These additional analyses support the conclusions in our original paper.

Table 1.

Outcomes in the propensity score–matched analyses

	Prophylaxis	Control	Risk Differences	95% CI	P Value
Sensitivity analysis 1	(n = 3,560)	(n = 3,560)
In-hospital mortality	1,932 (54.3)	2,089 (58.7)	−5.2	−7.9 to −2.6	<0.001
Nosocomial pneumonia	456 (12.8)	527 (14.8)	−2.3	−4.0 to −0.5	0.01
Sensitivity analysis 2	(n = 3,994)	(n = 3,994)
In-hospital mortality	2,280 (57.1)	2,399 (60.1)	−3.6	−5.7 to −1.4	0.001
Nosocomial pneumonia	569 (14.2)	632 (15.8)	−1.7	−3.5 to 0.03	0.054

Definition of abbreviation: CI = confidence interval.

Data are presented as n (%). Sensitivity analysis 1: We set the patients who received cephalosporin or glycopeptides within the 24 hours after starting extracorporeal membrane oxygenation treatment as prophylactic group. Sensitivity analysis 2: We analyzed the patients who received all kinds of antibiotics as prophylactic group.

The debate regarding antimicrobial prophylaxis in ECMO is still open. Future prospective studies may resolve most of our concerns.