We would like to thank Drs. Hedenstierna, Belda, Meyhoff and colleagues for their interest in our meta-analysis.1 We attempted to provide a comprehensive quantitative summary on the effects of perioperative high inspired oxygen fraction—definitely an on-going and passionate issue.
The main concern of Drs. Hedenstierna and Edmark is that we considered studies in which nitrous oxide was used as carrier gas, and that the variation in nitrous oxide concentrations among study groups may not have been properly controlled in all trials. Whether studies using nitrous oxide should be considered is, indeed, a relevant question in situations where nitrous oxide has been recognized as a confounding factor. For that reason, we did not consider data on postoperative nausea and vomiting from studies that were using nitrous oxide (because nitrous oxide has emetogenic properties). However, there is no evidence suggesting that nitrous oxide plays any role in the incidence of surgical site infection.2 Nitrous oxide was administrated in one trial only that reported data on pulmonary outcome.3 In that trial, the incidence of atelectasis was significantly higher (P < 0.001) in the group receiving 70% of nitrous oxide (i.e., 30% Fio2 [fraction of inspired oxygen]), suggesting either a detrimental effect of nitrous oxide or a protective effect of high Fio2, or both. In any case, the result tends to support our conclusions. We cannot exclude that, in trials that were using nitrous oxide, some variability in the concentration of nitrous oxide may have led to different oxygen regimens in some patients. This, however, would most probably not have affected our conclusions. If some patients in “high Fio2” groups had actually received a “not so high Fio2,” and some patients in “normal Fio2” groups had received a “higher than normal Fio2,” this would have weakened the beneficial effects of high oxygen fraction. Thus, our conclusions could have indeed been too conservative and the true beneficial effects of high Fio2 would actually be even more pronounced. Drs. Hedenstierna and Edmark are also skeptical about our conclusions on postoperative atelectasis. We fully agree with their view that the occurrence of perioperative atelectasis is of multifactorial etiology. Yet, the question is not so much whether intraoperative atelectases occur in surgical patients, as there is general agreement that this happens, but whether or not intraoperative high oxygen regimens increase the risk of clinically relevant postoperative atelectasis. To date, there is no evidence from randomized controlled trials to suggest that this is the case.
Dr. Belda and colleagues suggest an interesting method to better allow for potential sources of heterogeneity in meta-analyses. Although they agree that high Fio2 should be considered to reduce the risk of surgical site infection, and that this intervention may provide protection throughout a large range of surgeries, they argue that additional trials, with standardized outcome measures and including high-risk patients, will be needed to ensure adequate power and to guarantee wide applicability of these results. We agree that further large trials including patients at high risk of surgical site infection may be warranted. However, it should be highlighted that trials in patients who are not receiving prophylactic antibiotics are probably not ethically acceptable anymore. Thus, the challenge will be to confirm the anti-infective efficacy of high oxygen regimens in surgical patients who are receiving prophylactic antibiotics concomitantly and in whom the baseline risk of infection will be, accordingly, low.
Finally, Dr. Meyhoff and colleagues nicely highlight strengths and weaknesses of meta-analyses. We would like to reassure Dr. Meyhoff that we did not have, as they seem to suggest, any bias in favor, or against high inspired oxygen. One may, or may not, agree with our methodological choices, yet our process was overt; every step of the critical appraisal of included and excluded studies, as well as the rationale behind all quantitative analyses, were transparent, clearly described, and reproducible. Also, we have pointed out for the first time that almost all patients in these trials had received prophylactic antibiotics. This is a serious methodological issue that needs to be addressed when analyzing the anti-infective efficacy of high oxygen regimens and it is surprizing that this problem has not been pointed out in previous similar analyses. We are looking forward to the conclusions of the preannounced Cochrane review on the same subject, and we do hope that Dr Meyhoff and colleagues will take advantage of the methodological considerations depicted in our publication to further our understanding on the clinical relevance of high inspired oxygen fraction during surgery.
The authors declare no competing interests.