Perioperative oxygen fraction - effect on surgical site infection and pulmonary complications after abdominal surgery: a randomized clinical trial. Rationale and design of the PROXI-Trial

Abstract

Background: A high perioperative inspiratory oxygen fraction may reduce the risk of surgical site infections, as bacterial eradication by neutrophils depends on wound oxygen tension. Two trials have shown that a high perioperative inspiratory oxygen fraction (FiO(2) = 0.80) significantly reduced risk of surgical site infections after elective colorectal surgery, but a third trial was stopped early because the frequency of surgical site infections was more than doubled in the group receiving FiO(2) = 0.80. It has not been settled if a high inspiratory oxygen fraction increases the risk of pulmonary complications, such as atelectasis, pneumonia and respiratory failure. The aim of our trial is to assess the potential benefits and harms of a high perioperative oxygen fraction in patients undergoing abdominal surgery.

Methods and design: The PROXI-Trial is a randomized, patient- and assessor blinded trial of perioperative supplemental oxygen in 1400 patients undergoing acute or elective laparotomy in 14 Danish hospitals. Patients are randomized to receive either 80% oxygen (FiO(2) = 0.80) or 30% oxygen (FiO(2) = 0.30) during surgery and for the first 2 postoperative hours. The primary outcome is surgical site infection within 14 days. The secondary outcomes are: atelectasis, pneumonia, respiratory failure, re-operation, mortality, duration of postoperative hospitalization, and admission to intensive care unit. The sample size allows detection of a 33% relative risk reduction in the primary outcome with 80% power.

Discussion: This trial assesses benefits and harms of a high inspiratory oxygen fraction, and the trial may be generalizable to a general surgical population undergoing laparotomy.

Trial registration: ClinicalTrials.gov identifier: NCT00364741.

Figure 1

Meta-analysis comparing perioperative inspiratory oxygen fractions of 0.80 and 0.30/0.35 on surgical site infection.

Figure 2

Trial sequential analysis with a required information size of 5051. A priori heterogeneity adjusted information size (APHIS) based on an a priori relative risk reduction (RRR) of 33% with a type I error risk of 5% and a power of 80%. The cumulative z-curve constructed for a random effects model as heterogeneity is 74% crosses the traditional boundary (P = 0.05) once and return to non-significant values. The cumulative z-curve never crosses the trial sequential monitoring boundary. Despite 989 patients randomized we may still need more than 4000 randomized participants to close the information gap considering repeated analyses of accumulating data.

Figure 3

Trial sequential analysis excluding the trial of Pryor. Meta-analysis of the trials by Greif [13], Belda [12] and Mayzler [14], excluding the trial of Pryor [15] with a required information size of 1304 (APIS, a priori information size) based on an a priori relative risk reduction (RRR) of 33% and a type I error risk of 5% and a power of 80%. The cumulative z-curve constructed for a fixed-effect model as heterogeneity is 0% crosses both the traditional boundary (P = 0.05) after the first trial and the trial sequential monitoring boundary during the second trial. So there may be evidence for an effect of at least 33% RRR in a cumulative meta-analysis of trials investigating a high oxygen fraction when the Pryor trial is excluded when adjusting for repeated analyses of accumulating data.

Figure 4

Trial sequential analysis of all trials irrespective of adjuvant inhaled gases. The effect of 80% oxygen vs. 30% oxygen on surgical site infections calculated in cumulative meta-analysis of all trials irrespective adjuvant inhaled gases (the trials by Greif [13], Pryor [15], Belda [12], Mayzler [14] and Myles [42]). The low-bias heterogeneity adjusted information size (LBHIS) is 4500 based on a relative risk reduction (RRR) suggested by the low-bias trials of 33% and a meta-analytic estimate of the frequency of surgical site infection in the control group (30% oxygen) on 14% with a type I error risk of 5% and a power of 80%. No crossing of the trial sequential monitoring boundary at any time despite P < 0.05 after the first trial [13]. The gap of information to reject an intervention effect of 33% relative risk reduction is approximately 1500 patients.