FIO2 indicates inspiratory oxygen fraction.aUnderwent surgery when research staff were present.bNo outcome data described for these patients.cNo baseline or outcome data described for these patients.dPatients could be excluded for more than 1 reason.
Christian S. Meyhoff, Jørn Wetterslev, Lars N. Jorgensen, Steen W. Henneberg, Claus Høgdall, Lene Lundvall, Poul-Erik Svendsen, Hannah Mollerup, Troels H. Lunn, Inger Simonsen, Kristian R. Martinsen, Therese Pulawska, Lars Bundgaard, Lasse Bugge, Egon G. Hansen, Claus Riber, Peter Gocht-Jensen, Line R. Walker, Asger Bendtsen, Gun Johansson, Nina Skovgaard, Kim Heltø, Andrei Poukinski, André Korshin, Aqil Walli, Mustafa Bulut, Palle S. Carlsson, Svein A. Rodt, Liselotte B. Lundbech, Henrik Rask, Niels Buch, Sharafaden K. Perdawid, Joan Reza, Kirsten V. Jensen, Charlotte G. Carlsen, Frank S. Jensen, Lars S. Rasmussen, for the PROXI Trial Group. Effect of High Perioperative Oxygen Fraction on Surgical Site Infection and Pulmonary Complications After Abdominal SurgeryThe PROXI Randomized Clinical Trial. JAMA. 2009;302(14):1543–1550. doi:10.1001/jama.2009.1452
Author Affiliations: Department of Anaesthesia, Centre of Head and Orthopaedics (Drs Meyhoff and Rasmussen), Copenhagen Trial Unit, Center for Clinical Intervention Research (Dr Wetterslev), and Departments of Anaesthesia (Dr Henneberg) and Gynaecology (Drs Høgdall and Lundvall), The Juliane Marie Center, Copenhagen University Hospital, Rigshospitalet, Copenhagen; Department of Surgery, Bispebjerg Hospital, University of Copenhagen, Copenhagen (Drs Jorgensen and Simonsen); Department of Anaesthesia, Bispebjerg Hospital, University of Copenhagen, Copenhagen (Drs Svendsen, Mollerup, and Lunn); Departments of Anaesthesia (Dr Martinsen) and Surgery (Drs Pulawska, Bundgaard, and Bugge), Vejle Hospital, Vejle; Departments of Anaesthesia (Dr Hansen) and Surgery (Drs Riber, Gocht-Jensen, and Walker), Copenhagen University Hospital, Herlev; Departments of Anaesthesia (Drs Bendtsen and Johansson) and Surgery (Dr Skovgaard), Copenhagen University Hospital, Amager, Copenhagen; Department of Anaesthesia, Nykøbing Falster Hospital, Nykøbing Falster (Drs Heltø and Poukinski); Departments of Anaesthesia (Drs Korshin and Walli) and Surgery (Dr Bulut), Slagelse Hospital, Slagelse; Department of Anaesthesiology, Aarhus University Hospital, Aarhus (Drs Carlsson, Rodt, and Lundbech); Departments of Anaesthesiology (Dr Rask) and Surgery (Dr Buch), Odense Universitetshospital, Svendborg; Department of Surgery, Naestved Hospital, Naestved (Drs Perdawid and Reza); Departments of Anaesthesia (Dr K. V. Jensen) and Surgery (Dr Carlsen), Viborg Hospital, Viborg; and Department of Anaesthesia, Copenhagen University Hospital, Gentofte (Dr F. S. Jensen), Denmark.
Context Use of 80% oxygen during surgery has been suggested to reduce the risk of surgical wound infections, but this effect has not been consistently identified. The effect of 80% oxygen on pulmonary complications has not been well defined.
Objective To assess whether use of 80% oxygen reduces the frequency of surgical site infection without increasing the frequency of pulmonary complications in patients undergoing abdominal surgery.
Design, Setting, and Patients The PROXI trial, a patient- and observer-blinded randomized clinical trial conducted in 14 Danish hospitals between October 2006 and October 2008 among 1400 patients undergoing acute or elective laparotomy.
Interventions Patients were randomly assigned to receive either 80% or 30% oxygen during and for 2 hours after surgery.
Main Outcome Measures Surgical site infection within 14 days, defined according to the Centers for Disease Control and Prevention. Secondary outcomes included atelectasis, pneumonia, respiratory failure, and mortality.
Results Surgical site infection occurred in 131 of 685 patients (19.1%) assigned to receive 80% oxygen vs 141 of 701 (20.1%) assigned to receive 30% oxygen (odds ratio [OR], 0.94; 95% confidence interval [CI], 0.72-1.22; P = .64). Atelectasis occurred in 54 of 685 patients (7.9%) assigned to receive 80% oxygen vs 50 of 701 (7.1%) assigned to receive 30% oxygen (OR, 1.11; 95% CI, 0.75-1.66; P = .60), pneumonia in 41 (6.0%) vs 44 (6.3%) (OR, 0.95; 95% CI, 0.61-1.48; P = .82), respiratory failure in 38 (5.5%) vs 31 (4.4%) (OR, 1.27; 95% CI, 0.78-2.07; P = .34), and mortality within 30 days in 30 (4.4%) vs 20 (2.9%) (OR, 1.56; 95% CI, 0.88-2.77; P = .13).
Conclusion Administration of 80% oxygen compared with 30% oxygen did not result in a difference in risk of surgical site infection after abdominal surgery.
Trial Registration clinicaltrials.gov Identifier: NCT00364741
Surgical site infection is a common and serious complication following abdominal surgery.1 To prevent surgical site infection, it is essential to optimize perioperative conditions in the first hours following bacterial contamination.2 Tissue oxygen tension is often low in wounds and colorectal anastomoses, and this may reduce tissue healing via oxidative killing by neutrophils and also reduce induction of collagen formation, neovascularization, and epithelialization.3- 7 Perioperative arterial and wound oxygen tension can be increased by a higher inspiratory oxygen fraction.7
Trials by Greif et al8 and Belda et al9 have suggested a significant reduction in the frequency of surgical wound infections when 80% rather than 30% oxygen was administered during surgery and in the first postoperative hours. Conversely, a trial by Mayzler et al10 found no significant difference, a trial by Pryor et al11 was stopped prematurely because the frequency of wound infection was more than doubled with high oxygen fraction, and a trial by Gardella et al12 was stopped for futility. High inspiratory oxygen concentrations throughout the perioperative period also may result in pulmonary complications, but this important aspect has been studied in only 30 patients.13 In a subgroup of patients from the trial by Greif et al,8 there was a nonsignificant increase in the size of computed tomography (CT) scan–detected areas of atelectasis in the group receiving 80% oxygen.13
Furthermore, a high inspiratory oxygen fraction has been related to detrimental effects such as an increased risk of airway inflammation,14 poor regulation of blood glucose levels,15 and changes in the cardiac index16 but has also been related to benefits such as improved healing of colorectal anastomosis17 and reduced frequency of postoperative nausea and vomiting.18,19
The aim of this trial was to assess the benefits and harms of a high perioperative oxygen fraction in a general surgical population of patients undergoing laparotomy, the primary outcome measure being surgical site infection. We hypothesized that 80% oxygen would reduce the frequency of surgical site infections without increasing the frequency of pulmonary complications.
The Danish Medicines Agency and the regional ethics committee approved the trial, which adhered to the International Conference on Harmonisation Good Clinical Practice standards. Written informed consent was obtained from all patients participating in this multicenter trial in 14 Danish hospitals between October 8, 2006, and October 6, 2008. Patients were eligible if they were 18 years or older and scheduled to undergo acute or elective laparotomy. When laparotomy was indicated for a gynecological disease, only patients with suspected malignancy (defined as risk of ovarian malignancy index >200 or a specimen showing atypical or neoplastic cells20) were included. Exclusion criteria were operations performed under general anesthesia within 30 days, chemotherapy for malignancy within 3 months, inability to provide informed consent, and preoperative arterial hemoglobin oxygen saturation below 90% without supplemental oxygen assessed by pulse oximetry.
Patients were randomized according to a computer-generated allocation list by a central interactive voice-response system at the Copenhagen Trial Unit using study center, diabetes mellitus, acute or elective operations, and body mass index (<30 or ≥30, calculated as weight in kilograms divided by height in meters squared) as stratification variables.
The trial protocol21 included several important aspects of perioperative care, including epidural analgesia, control of temperature and glucose level, absence of preoperative oral bowel preparation, and standardized anesthesia without nitrous oxide. The protocol recommended cefuroxime (1.5 g) and metronidazole (1 g) given intravenously as standard antibiotic choice, but ampicillin (2 g) or benzylpenicillin (2 million IU) in combination with gentamicin (0.240 g) and metronidazole (1 g) were also allowed.21 Fewer antibiotics were required in the case of elective cholecystectomy or laparotomies with no potential contamination. We considered “timely” administration of antibiotics as administration of the first and second antibiotic within 60 minutes prior to skin incision. Perioperative fluids were given only to replace measured or calculated deficits (no third-space loss), aiming at a postoperative body weight increase of less than 1 kg. Blood loss was replaced 1:1 with colloids, and blood transfusion was initiated if blood loss exceeded 20 mL/kg.21
Anesthesia was either inhalational or total intravenous anesthesia, determined entirely by the attending anesthetist. An inspiratory oxygen fraction (FIO2) of 1.0 was used at induction of anesthesia until tracheal intubation and again immediately before tracheal extubation. After induction of anesthesia and tracheal intubation, patients randomized to the 80% oxygen group were given an FIO2 of 0.80 until the end of surgery. The first 2 hours following extubation, patients breathed an FIO2 of 0.80 administered by means of a nonrebreathing face mask with a reservoir (High Concentration Oxygen Mask; Intersurgical Ltd, Wokingham, United Kingdom) and a flow of 14 L of oxygen and 2 L of air per minute. The patients randomized to the 30% oxygen group were given an FIO2 of 0.30 after tracheal intubation and received a mixture of 2 L of oxygen and 14 L of air per minute through an identical nonrebreathing face mask after extubation.21 The patients were ventilated to ensure normocapnia.21
In both groups, FIO2 was increased if hypoxia was detected or suspected to ensure arterial oxygen saturation greater than 94% and arterial oxygen tension greater than 9 kPa. Positive end expiratory pressure was used at a level chosen by the attending anesthetist. Two hours after surgery, supplemental oxygen was administered only at the physician's discretion and according to clinical practice.
The risk of infection was assessed with the NNISS (National Nosocomial Infections Surveillance System) and the SENIC (Study on the Efficacy of Nosocomial Infection Control) scales.22,23
Cardboard shields were placed on the side of the anesthesia machine to keep the surgical team blinded to group allocation. In the postanesthesia care unit, opaque bags covered the flow meters. Information about perioperative FIO2, arterial oxygen tension, and flow of oxygen and air was noted on a separate paper form and placed in a sealed opaque envelope in the medical record when patients were discharged from the postanesthesia care unit. The staff on the wards and the patients were not informed during follow-up about the intervention received.
To evaluate participant blinding, the patients were asked which of the groups they believed they belonged to and to give a reason for their choice. No investigator had access to patient allocation and outcome data at the same time. All statistical analyses were performed with the intervention groups named A and B, and the statistician was blinded to allocation. This manuscript, including the discussion and conclusion, was written in 2 versions, one based on the assumption that treatment A was 80% oxygen and treatment B was 30% oxygen and the other based on the reverse assumption.21,24 All authors approved both versions before unmasking the allocation groups.21
The primary outcome was surgical site infection within 14 days of surgery, defined according to criteria from the Centers for Disease Control and Prevention (CDC) as superficial or deep wound infection or as intra-abdominal organ/space infection.25 The secondary outcomes were pneumonia within 14 days (according to CDC criteria26), atelectasis within 14 days, respiratory failure within 14 days (defined as the need for controlled ventilation or arterial oxygen saturation below 90% despite supplemental oxygen), mortality within 30 days, duration of postoperative hospitalization, admission to the intensive care unit within 14 days (if not part of postoperative care), and abdominal reoperation for any reason within 14 days.
The patients were seen daily in the postoperative period by a surgical investigator blinded to allocation. A follow-up visit was conducted between the 13th and 30th postoperative day. Surgical site infection, pulmonary complications, and other adverse events were evaluated at each visit, and additional information about wound characteristics in the postoperative period were collected to calculate the ASEPSIS (additional treatment, serous exudate, erythema, purulent exudate, separation of deep tissues, isolation of bacteria, and duration of inpatient stay) score.27
Patients with symptoms of pulmonary complications were examined according to clinical practice by the attending physician, including chest radiographs or CTs when relevant. All chest radiographs and CTs were evaluated for infiltrate and atelectasis by the attending radiologist, who was unaware of the intervention applied. Atelectasis was considered present if described in the radiologist's evaluation of chest radiograph or CT. Adverse events were collected prospectively according to the CONSORT (Consolidated Standards of Reporting Trials) Statement28 and specifically addressed at the follow-up visit.21
If patients did not attend the follow-up visit, we contacted hospital outpatient clinics, emergency departments, and the patient's family physician for information about outcome and adverse events. Wound evaluation carried out in accordance with the CDC criteria was considered adequate. The patients were interviewed by telephone in the remaining cases. The complete statistical analysis plan is described in the protocol.21
Patients were excluded from the per-protocol analysis if they did not meet the inclusion criteria, fulfilled an exclusion criterion, had an FIO2 greater than 0.60 for more than 1 hour (30% oxygen group) or an FIO2 less than 0.60 for more than 1 hour (80% oxygen group), used an oxygen mask less than 1 hour, had no in-hospital evaluation of the outcomes for 4 or more consecutive days, or had no follow-up visit between the 13th and 30th postoperative day; patients were also excluded if assessors were unblinded.
The intervention effect was assessed both with and without adjustment for the stratification variables as well as for chronic obstructive pulmonary disease, daily smoking, and surgical incision extending above the umbilicus. All intervention-effect estimates were reported with 95% confidence intervals (CIs), and a 2-sided P < .05 was considered statistically significant. Analyses were performed using R version 2.8.0 (http://www.r-project.org).
In the protocol we estimated that the frequency of surgical site infection would be 16% in the 30% oxygen group.21 This was based on the previously reported frequencies8- 11 and the inclusion of acute laparotomies in our trial. A meta-analysis with a fixed-effects model showed a relative risk reduction of 25% if all 4 completed trials are included and of 48% if all but the Pryor trial11 are considered. We thus expected a relative risk reduction of 33%. We calculated that a total sample size of 1400 patients would allow us to detect or reject a difference in surgical site infection of between 11% and 16% with 5% type I error risk, 80% power, and 10% dropout. In accordance, we identified a lack of information of more than 1400 randomized patients in a trial sequential analysis29 of the meta-analysis of all trials conducted prior to the PROXI trial to detect or reject a 33% relative risk reduction in a future meta-analysis.21
An independent data monitoring committee recommended continuing the trial after the interim analysis, based on 40% of the records, was performed on January 24, 2008.21
A total of 1400 patients were randomized and 1386 included in the modified intention-to-treat analysis (Figure). Fourteen patients were excluded because of withdrawn informed consent or cancellation of surgery after randomization. Demographic and perioperative characteristics were similar in the 2 groups (Table 1 and Table 2).
Surgical site infection occurred in 131 of 685 (19.1%) vs 141 of 701 (20.1%) patients in the 80% and 30% oxygen groups, respectively (odds ratio [OR], 0.94; 95% confidence interval [CI], 0.72-1.22; P = .64) (Table 3). The incidence of pulmonary complications was not significantly different, with atelectasis occurring in 54 of 685 (7.9%) vs 50 of 701 (7.1%) patients (OR, 1.11; 95% CI, 0.75-1.66), pneumonia in 41 (6.0%) vs 44 (6.3%) (OR, 0.95; 95% CI, 0.61-1.48), and respiratory failure in 38 (5.5%) vs 31 (4.4%) (OR, 1.27; 95% CI, 0.78-2.07) in the 80% and 30% oxygen groups, respectively. Thirty patients (4.4%) died in the 30-day follow-up period in the 80% oxygen group, vs 20 patients (2.9%) in the 30% oxygen group (OR, 1.56; 95% CI, 0.88-2.77; P = .13).
Rupture of the abdominal fascia occurred in 23 of 685 (3.6%) vs 17 of 701 (2.4%) patients in the 2 groups. Among patients undergoing colorectal surgery, surgical site infection occurred in 72 of 303 (23.7%) vs 83 of 330 (25.2%) patients (P = .68) and mortality within 30 days in 14 of 303 (4.6%) vs 5 of 330 (1.5%) patients in the 80% and 30% oxygen groups, respectively (P = .03). Anastomotic leak occurred in 0 of 53 vs 2 of 49 (4.1%) after small bowel resections, 10 of 175 (5.7%) vs 10 of 174 (5.7%) after colonic resections, and 2 of 42 (4.8%) vs 6 of 49 (12.2%) after rectal resections in the 80% and 30% oxygen groups, respectively. Other adverse events are reported in Table 4.
One thousand eighty-one patients were included in the per-protocol analysis (Figure). Surgical site infection occurred in 122 of 555 (22.0%) vs 116 of 526 (22.1%) patients in the 80% and 30% oxygen groups, respectively (P = .98), and there were no significant differences regarding pulmonary complications.
When patients in the 80% oxygen group were asked about suspected treatment allocation, 16% of the patients reported assuming that they had received 80% oxygen and 5% that they had received 30% oxygen; 79% answered “do not know.” The corresponding values for the 30% oxygen group were 15%, 5%, and 80%. Less than 1% of the reasons provided suggested unblinding.
Our trial did not find a 33% relative risk reduction in the frequency of surgical site infection when a high inspiratory oxygen fraction of 80% vs 30% was given in the perioperative period of acute or elective laparotomy. The high oxygen fraction was not associated with a significant increase in the frequency of pulmonary complications or other adverse events. Mortality differences were not statistically significant.
The primary strength of this trial is the inclusion of a large number of patients at risk of surgical site infection undergoing a wide range of laparotomies, including 28% acute procedures. Surgical site infection was defined according to the CDC criteria,25 allowing for comparison with other trials. Follow-up was thorough, with assessment of all adverse events and vital status in all patients.
Our trial has some limitations. First, we were unable to apply the elements of the intended optimized regimen, such as timely administration of adequate antibiotics and prevention of hypothermia, to all patients. However, the distribution of the quality measures was similar in our 2 groups. Second, 51 of 701 patients (7.3%) in the 30% oxygen group required an FIO2 of 0.60 or greater for more than 1 hour to maintain arterial oxygen saturation above 94%, in accordance with safety measures in clinical practice. Third, a total of 157 of 1386 patients (11.3%) did not have a follow-up visit between the 13th and 30th postoperative day, and some minor wound infections could thus have been missed during that period. However, follow-up status was obtained in all patients, because we reviewed all hospital admissions and visits to primary physicians when follow-up occurred beyond the 30th postoperative day. Our per-protocol analysis, in which we excluded these patients, showed a result similar to the intention-to-treat analysis. Fourth, the inclusion of different surgical procedures may be associated with the risk of overlooking a beneficial effect related to specific surgical procedures, such as colorectal resections. However, the effect on surgical site infection was similar in this subgroup, and in clinical practice the type of surgery is not always known at the time the FIO2 is selected.
Two trials have suggested that high (80%) FIO2 is effective in preventing surgical wound infections, with relative risk ratios of 39%9 and 54%.8 In contrast, 3 other trials did not report a significant reduction; one trial was stopped early because the frequency of surgical wound infections doubled,11 one was statistically insignificant,10 and recently, a large trial investigating a high oxygen fraction delivered via nonrebreathing face mask to prevent postcesarean surgical site infection was stopped for futility.12 The relative risk reduction in our trial was only 5% in favor of the 80% oxygen group. Some important differences exist between our trial and the 2 trials showing benefit of high FIO2,8,9 because, according to current recommendations,21 the patients in our trial received epidural analgesia, did not receive oral bowel preparation, and were not aggressively hydrated; however, they may have received more vasopressors during anesthesia.
The risk of surgical site infection, as evaluated by the level of the NNISS and SENIC scores, was similar to that in the previous trials,8,9,11 although there were fewer medium-risk and more low-risk and high-risk patients in our trial. This was related to the inclusion of gynecological and emergency procedures.
We found a frequency of surgical site infections comparable to that in the trial by Belda et al,9 which also adhered to the CDC definition, but our event rate was higher than those previously reported.8,10,11 These 3 trials, however, did not report the frequency of intra-abdominal organ/space surgical site infection. With an event rate of 19%, we had nearly 80% power to detect a 20% relative risk reduction. On the other hand, this trial has only 15% power to detect a 10% relative risk reduction, so we cannot exclude that a high perioperative oxygen fraction has a minor effect on surgical site infection.
Other additional trials have suggested that increased perioperative oxygen supply results in fewer wound infections. Patients undergoing nitrous oxide–free anesthesia with 80% oxygen had fewer wound infections than patients receiving nitrous oxide–based (30% oxygen, 70% nitrous oxide) anesthesia.30 It is unclear whether nitrous oxide or the higher oxygen concentration caused this difference.31 In a meta-analysis of 5 trials based on 3001 patients, including the nitrous oxide trial,30 Qadan et al found that perioperative administration of 80% oxygen was associated with a reduced risk of surgical site infections.32 However, when the data from our large trial are added to this existing evidence, it seems likely that the overall benefit of 80% oxygen observed by Qadan et al will be attenuated and perhaps even no longer statistically significant. Another large trial investigating the treatment of hypoxia via continuous positive airway pressure in the postoperative period also demonstrated a reduction in wound infections.33
Ventilation for only 5 minutes with 100% oxygen results in significantly larger areas of atelectasis than ventilation with lower oxygen concentrations,34 and atelectasis on CT tended to be more frequent in patients receiving 80% oxygen compared with patients receiving 30% oxygen (94% vs 64%, P = .12) in a subgroup13 of the trial by Greif et al.8 We could not demonstrate significantly increased frequencies of pulmonary complications when 80% oxygen was given for a median of approximately 5 ½ hours. Our method of detecting atelectasis is likely to have resulted in fewer events than if a chest radiograph or CT was obtained routinely in all patients. On the other hand, we consider the reporting of asymptomatic atelectases to have less clinical relevance, and the frequency of atelectasis in our trial (7.5%) was the same as in the nitrous oxide–free group in the trial by Myles et al.30 In the nitrous oxide–free group, the frequency of pneumonia (1.5%) was lower than in our trial (6.5%), which is more comparable to that reported by Squadrone et al,33 in which 5.7% of patients with postoperative hypoxemia developed pneumonia. The difference may be caused by our thorough postoperative follow-up, because pneumonia was diagnosed within a median of 5 days after surgery.
The proportion of patients with respiratory failure and death within 30 days was higher in the 80% oxygen group; although those differences were not statistically different, these events were uncommon and the study did not have adequate power to assess this outcome. This finding is in contrast to previous trials comparing 80% and 30% oxygen.8- 11 The unadjusted OR for mortality in our trial was 1.56 (95% CI, 0.88-2.77) in favor of the 30% oxygen group. Thus, there was no evidence of a benefit in reducing mortality with a high perioperative FIO2 (80%), although we recognize that a lack of important information must be addressed before this can be established beyond reasonable doubt, because our trial was not powered to document lethality differences.
Gynecological cancer surgery and acute laparotomies were included in this trial, but conversions of laparoscopic procedures to laparotomy were not eligible. Apart from conversion of laparoscopy, we believe the results of this trial may be generalizable to a general surgical population undergoing laparotomy.
In conclusion, administration of 80% oxygen compared with 30% oxygen during and for 2 hours after abdominal surgery did not result in a difference in risk of surgical site infection. The frequencies of pulmonary complications and mortality were not significantly different.
Corresponding Author: Lars S. Rasmussen, MD, PhD, DMSc, Department of Anaesthesia, Section 4231, Centre of Head and Orthopaedics, Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Denmark (firstname.lastname@example.org).
Author Contributions: Dr Rasmussen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Meyhoff, Wetterslev, Jorgensen, Henneberg, Høgdall, Lundvall, Rasmussen.
Acquisition of data: Meyhoff, Wetterslev, Jorgensen, Henneberg, Høgdall, Svendsen, Mollerup, Lunn, Simonsen, Martinsen, Pulawska, Bundgaard, Bugge, Hansen, Riber, Gocht-Jensen, Walker, Bendtsen, Johansson, Skovgaard, Heltø, Poukinski, Korshin, Walli, Bulut, Carlsson, Rodt, Lundbech, Rask, Buch, Perdawid, Reza, K. V. Jensen, Carlsen, F. S. Jensen.
Analysis and interpretation of data: Meyhoff, Wetterslev, Jorgensen, Henneberg, Bendtsen, Poukinski, Rasmussen.
Drafting of the manuscript: Meyhoff, Wetterslev, Jorgensen, Rasmussen.
Critical revision of the manuscript for important intellectual content: Meyhoff, Wetterslev, Jorgensen, Henneberg, Høgdall, Lundvall, Svendsen, Mollerup, Lunn, Simonsen, Martinsen, Pulawska, Bundgaard, Bugge, Hansen, Riber, Gocht-Jensen, Walker, Bendtsen, Johansson, Skovgaard, Heltø, Poukinski, Korshin, Walli, Bulut, Carlsson, Rodt, Lundbech, Rask, Buch, Perdawid, Reza, K. V. Jensen, Carlsen, F. S. Jensen, Rasmussen.
Statistical analysis: Meyhoff, Wetterslev, Rasmussen.
Obtained funding: Meyhoff, Rasmussen.
Administrative, technical, or material support: Meyhoff, Wetterslev, Jorgensen, Henneberg, Høgdall, Svendsen, Mollerup, Lunn, Simonsen, Martinsen, Pulawska, Bugge, Gocht-Jensen, Johansson, Heltø, Poukinski, Walli, Bulut, Carlsson, Rodt, Lundbech, Rask, Buch, Perdawid, Reza, K. V. Jensen, F. S. Jensen, Rasmussen.
Study supervision: Wetterslev, Henneberg, Lundvall, Rasmussen.
Financial Disclosures: None reported.
Funding/Support: This study was supported by the Danish Medical Research Council (271-05-0206), the Lundbeck Foundation (402/06), Rigshospitalet's Research Council, the Novo Nordisk Foundation, the Aase and Ejnar Danielsens Foundation (105728), the A. P. Møller Foundation for the Advancement of Medical Science, the Danish Society of Anaesthesiology and Intensive Care Medicine's Research Initiative, the Beckett-Foundation, the Brødrene Hartmanns Foundation, and the Etly and Jørgen Stjerngrens Foundation.
Role of the Sponsors: The funding sources had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript.
PROXI Trial Group Investigators: Peter R. Kirkegaard, MD (Department of Anaesthesia, Bispebjerg Hospital, University of Copenhagen, Copenhagen, Denmark), Vagn Berg, MD (Department of Surgery, Vejle Hospital, Vejle, Denmark), Henrik D. Petersen, MD (Department of Surgery, Nykøbing Falster Hospital, Nykøbing Falster, Denmark), Ole Schiødt, MD (Department of Anaesthesia, Slagelse Hospital, Slagelse, Denmark), Henrik Stougaard, MD (Department of Anaesthesiology, Odense Universitetshospital, Svendborg, Denmark), Helle S. Pedersen, MD (Department of Anaesthesia, Naestved Hospital, Naestved, Denmark), Susan K. Pedersen, MD (Department of Anaesthesia, Viborg Hospital, Viborg, Denmark), Morten Gaarden, MD, PhD (Department of Surgery, Viborg Hospital, Viborg, Denmark), Jacob Rosenberg, MD, DMSc (Department of Surgery, Copenhagen University Hospital, Gentofte, Denmark), Søren Bøgevig, MD (Department of Anaesthesia, Holbaek Hospital, Holbaek, Denmark), Claus Juul, MD (Department of Surgery, Holbaek Hospital, Holbaek, Denmark), Jan M. Krzak, MD (Department of Surgery, Kolding Hospital, Kolding, Denmark). All participating centers received compensation per included patient. Data Monitoring Committee: Jørgen Hilden, MD (Department of Biostatistics, University of Copenhagen, Copenhagen, Denmark); Jørgen B. Dahl, MD, DMSc (Department of Anaesthesia, Glostrup Hospital, Copenhagen University Hospital, Glostrup, Denmark); Peer Wille-Jørgensen, MD, DMSc (Department of Surgery, Bispebjerg Hospital, University of Copenhagen, Copenhagen, Denmark). The Data Monitoring Committee received no compensation. Steering Committee: Drs Meyhoff, Wetterslev, Jorgensen, and Rasmussen. Statistical Support: Klaus K. Holst, Cand Scient (Department of Biostatistics, University of Copenhagen, Copenhagen, Denmark); Jeremy Silver, Cand Scient (Department of Biostatistics, University of Copenhagen, Copenhagen, Denmark). Compensation consisted of an hourly salary. Language Correction: Kate Whitfield, CRA (European Clinical Research Infrastructures Network, Copenhagen Trial Unit, Copenhagen, Denmark). No compensation.