Effectiveness of supplemental oxygenation to prevent surgical site infections: A systematic review with meta-analysis

Abstract Objective: to assess the effectiveness of supplemental oxygenation with high FiO2 when compared to conventional FiO2 in the prevention of surgical site infection. Method: an effectiveness systematic review with meta-analysis conducted in five international databases and portals. The research was guided by the following question: Which is the effectiveness of supplemental oxygenation with high FiO2 (greater than 80%) when compared to conventional FiO2 (from 30% to 35%) in the prevention of surgical site infections in adults? Results: fifteen randomized clinical trials were included. Although all the subgroups presented a general effect in favor of the intervention, colorectal surgeries had this relationship evidenced with statistical significance (I2=10%;X2 =4.42; p=0.352). Conclusion: inspired oxygen fractions greater than 80% during the perioperative period in colorectal surgeries have proved to be effective to prevent surgical site infections, reducing their incidence by up to 27% (p=0.006). It is suggested to conduct new studies in groups of patients subjected to surgeries from other specialties, such as cardiac and vascular. PROSPERO registration No.: 178,453.


Introduction
Surgical Site Infection (SSI) refers to an infection that occurs after surgery at the incision or in the part of the body where the surgery was performed and can involve the skin, tissues, organs or material implanted within the first 30 days or within 90 days if prostheses are implanted (1)(2) .
The Center for Disease Control (CDC), an American body for disease control, classifies SSIs as follows: superficial, when they involve the skin and subcutaneous tissue; deep, when they reach deeper incision tissues such as fascia and muscle; and organ/space, in cases involving deep regions beyond the fascia, which were exposed after the surgical procedure (1)(2) .
SSI increases the risks for other complications, such as surgical wound dehiscence and sepsis, which can lead to second surgeries, increased hospitalization times and hospital costs, with the possibility of worsening the patients' quality of life, which justifies making every possible effort to prevent this infection (3)(4)(5)(6)(7) .

Eligibility criteria
The studies included were those published from 2000 to September 2021, considering that an increase in the production on the topic is recorded from that period onwards. Studies with adult patients subjected to surgeries from any specialty were included. The intervention (FiO 2 greater than 80%) and the comparator (FiO 2 30%-35%), with each description, considered that the inspired oxygen

Information sources
The following databases were used to select the

Selection process
After the stage where the articles were identified in the databases, the titles and abstracts of each article were analyzed, as well as the keywords/descriptors. Subsequently, the references of all the articles were consulted to identify additional studies.
The studies were selected by two reviewers with experience in review studies, independently and blindly, with consensus for inclusion of the articles. In turn, any and all disagreements were discussed with a third reviewer.

Data extraction process
The first evaluation of the articles took place through their titles and abstracts and, subsequently, the full texts were read to extract the following data: title of the article, name of the journal, authors, country, language, year of publication, type of study, objective, study population, period of study, intervention, evaluation method, statistical analysis, result and conclusion. The web version of the EndNote™ software was used to organize the references found.
The studies selected were imported into the JBI System for the Unified Management, Assessment and Review of Information (JBI SUMARI; JBI, Adelaide, Australia) and evaluated in detail in relation to all the inclusion criteria as designed by the instrument for critical evaluation of studies. SUMARI is a software program developed to support systematic reviews and facilitate the entire review process, from development of the protocol to writing of the final report (21) .

Risk of bias assessment corresponding to the studies
The JBI Data Extraction Form for Experimental/ Observational Studies instrument (22) was used in the final critical evaluation of the articles. In this stage, both evaluators performed the methodological critical evaluation independently and the concepts attributed were considered when in agreement between both.
Subsequently, the articles were included if they presented more than 70% agreement. Finally, the evaluators assessed the risks of bias.

Effect measures
Synthesis of the results occurred in a narrative way and with a meta-analysis. The meta-analysis was prepared with the aid of the SUMARI online software program (21) . The results are summarized in the subgroup analysis (colorectal surgeries, C-sections and abdominal surgeries) by means of the Mantel-Haenszel model.
Considering that the studies are homogeneous in terms of method, population by subgroup, intervention and outcome, the meta-analysis was prepared through the fixed-effects model (23)  were used to define heterogeneity as low, moderate and high, respectively (24) .

Results
The search in the databases selected resulted in 399 articles, of which 160 were excluded for being www.eerp.usp.br/rlae 4 Rev. Latino-Am. Enfermagem 2022;30:e3648.
duplicates. The number of articles excluded for not meeting the criteria after reading the titles and abstracts corresponded to n=216, namely: editorials, errata, responses, opinion, comments and letters to the editor (n=22); abstracts (n=5); literature reviews on the topic (n=14); study protocols (n=1); studies in the animal experimentation phase (n=2); publications in the veterinary field (n=2); articles that dealt with other interventions such as hyperbaric oxygenation, extracorporeal membrane oxygenation (ECMO), hypercarpnia, vacuum therapy and fluid administration, or antibiotic prophylaxis (n=170). One article was excluded because there was a retraction by the authors in the same journal acknowledging errors in the statistical analysis and methods that would preclude recognizing its findings (25) . The articles excluded after reading the full texts (n=16) for not answering the research question evaluated physiological, immunological and hemodynamic aspects, but did not directly or indirectly assess the surgical site infection outcome. Of these, one study was excluded for having being the only one that evaluated the intervention in trauma surgeries (26) , and another for focusing on the administration of nitrous oxide (27) .
Seventeen articles were assessed by independent evaluators, with two exclusions. In the end, fifteen randomized clinical trials were included in the metaanalysis. Figure 1 describes the process corresponding to selection and inclusion of the articles. surgeries, from 6.6% to 31.0%; and C-sections, from 5.3% to 14.5% (Figure 3).

Discussion
To prevent surgical site infections, it is essential to optimize the perioperative conditions, as the first hours after exposure of the surgical site to bacterial contamination are fundamental to avoid infection (43) . The findings on perioperative oxygen supplementation have a potential to contribute by bringing diverse evidence to the adoption of this practice in the prevention of surgical site infection.
Tissue hypoxia reduced production of collagen and revascularization, which are necessary for tissue repair (18,32,(43)(44)(45) . Perioperative and wound arterial oxygen pressure (PaO 2 ) can be increased by a higher inspiratory oxygen fraction (14,32,45) , and hyperoxygenation may also be related to the optimization of the effect of some antibiotics (32) . However, the inspired fraction is not always related to better oxygenation in the surgical wound, due to dependence on other clinical factors of the patient and related to anesthesia.
Although national studies have not yet evaluated high FiO 2 as a risk factor for SSI (3)(4)42) , a research study included the supplemental oxygenation strategy in a care bundle for obese patients subjected to bariatric surgery, which was related to the lower incidence of SSI (48) . However, this study did not describe how the intervention was performed in the postoperative period or how long it was maintained (48) .
Regarding the methodological issue, it was observed that for this type of study it was possible, in all cases, to blind of the patient and the SSI evaluator in the postoperative period; however, it was not possible to blind the anesthesiologist, as mentioned by some authors (28)(29)31,33,37,42,45,50) . In addition to that, it was evident that the multicenter studies presented larger samples, which significantly impacted on heterogeneity and on the results of the meta-analysis (33,35,(37)(38)(41)(42) . It is also noted that, after the evaluation of a partial, initial sample, four studies were terminated by the futility criterion because it was considered by statistical analysis that they would not find different results if they were continued (29,33,(41)(42)50) .
In the meta-analysis there was low heterogeneity prophylaxis, reducing this sample selection bias (12) . In addition to that, most of the study protocols provided for blinding of the patient and of the evaluator of the wounds in the postoperative period, although not blinding the anesthesiologist, in order to ensure maintenance of oxygen supply according to the randomized group (28)(29)31,33,37,(41)(42)45) .
The studies with C-sections presented a limitation due to the use of epidural or rachidian anesthesia.
These surgeries are generally performed with the use of masks or nasal catheters, which hinder maintenance of a standard and constant FiO 2 , as is the case in general anesthesia with orotracheal intubation (16,45,50) .
Oxygenation with a mask or nasal catheter is a limiting factor for these studies due to the variability of mask models and fits, differences in tidal volume per patient, failures in equipment and accessories, conversion to general anesthesia (not considered in the studies) and, perhaps the most important, deficient fit of the mask to the face with significant oxygen leakage (16,45,50) . In the studies referring to abdominal and colorectal surgeries, the patients undergo general anesthesia and maintain greater sedation in the immediate postoperative period, ensuring better adherence to the use of face masks.
In addition to the type of anesthesia associated, the effect of hyperoxygenation may have been better observed in colorectal surgeries because they are contaminated, when compared to cesarean surgeries. The surgeries presented surgical site infection rates proportional to the contamination degree (3)(4) .
The assessment of hyperoxygenation was concentrated on two large groups of procedures, namely: gastrointestinal tract surgeries and gynecological surgeries. A publication that did not comprise the sample due to its methodological design (a series of ten cases) evaluated vascular surgeries and observed that high anesthesia with endotracheal intubation, an increased inspired oxygen fraction (FiO 2 ) should be administered intraoperatively and post-extubation in the immediate postoperative period (1,(8)(9)(10) . Only one guideline indicates a time of supplemental oxygen postoperative administration from 2 to 6 hours (10) and none of them guides the form of administration, as they only leave the inspired oxygen fraction greater than 80% as a goal (1,(8)(9)(10) . Two guidelines reinforce that, in order to optimize oxygen delivery to the tissues, perioperative normothermia and adequate volume replacement must be maintained (1,9) . Only the APIC guideline emphasizes that the data are stronger for the colorectal surgeries, as can be found both in this meta-analysis and in others (11,13,15,19) .
To change the practice, due to the potential risks that have not yet been well clarified in the studies considered in this review and in the current guidelines for SSI prevention, it should first be considered that normovolemia, normotension, normothermia, normoglycemia and normoventilation can be effective in the prevention of SSI and safely applied in these cases (56) .
Finally, it can be asserted that the perspective about the current SSI prevention guidelines has been expanded after this discussion. The limitation of this review is the fact that segmentation into subgroups, although necessary to increase validity of the findings, reduces heterogeneity and the total sample size.

Conclusion
Providing inspired oxygen fractions greater than 80% during the perioperative period in colorectal surgeries can be effective to prevent SSI, reducing its incidence by up to 27% (p=0.006). It is suggested to conduct new studies in groups of patients subjected to surgeries from other specialties, such as cardiac and vascular.