Controlled Clinical Trial of Supplemental Oxygen for the Prevention of Post-Cesarean Infectious Morbidity

We designed a randomized controlled trial of supplemental oxygen for the prevention of post cesarean delivery infectious morbidity at Washington University in St. Louis, MO. Patients who underwent scheduled or intrapartum cesarean delivery with regional anesthesia were eligible for participation. Women were recruited for study participation from the antepartum service and labor and delivery. Exclusion criteria included emergency surgery in which the participant was unable to provide informed consent, human immunodeficiency virus infection, chronic corticosteroid therapy or other immunosuppressive therapy, general anesthesia, and a diagnosis of extra uterine infection (i.e., pyelonephritis or pneumonia) before cesarean delivery. Acute chorioamnionitis was not an exclusion criterion.

After written consent was obtained, patients were randomly assigned in a 1:1 scheme to receive either supplemental oxygen or standard care. Randomization was achieved with opaque envelopes that contained the assigned study group; the envelopes were opened after the patients had agreed to participate in the study but before surgery. Women in the supplemental oxygen group received oxygen at a flow rate of 10 L/min (corresponding to a FiO2 of approximately 80%) by nonrebreather mask (CareFusion, Yorba Linda, CA) during and for 2 hours after cesarean delivery. Compliance with supplemental oxygen by face mask was assessed by the anesthesiologist intraoperatively and by the postpartum nurse at 30, 60, 90, and 120 minutes after surgery. Women who were assigned to the standard care group received oxygen at a flow rate of 2 L/min (corresponding to a FiO2 of 25-30%) by nasal cannula (Salter Labs, Arvin, CA) during the cesarean delivery only. Oxygen saturation was assessed both intraoperatively and postoperatively for both groups; women with oxygen saturations 95% were supplied supplemental oxygen, as needed, to maintain appropriate oxygenation. Women in both groups received standard preoperative skin preparation and prophylactic antibiotics. Subcutaneous depth was measured by the primary operative team with a sterile ruler; demographic, intrapartum, and operative information was abstracted from the medical records. The operative decision to place subcutaneous sutures was left to the surgical team.

The primary outcome for this study was a composite outcome that consisted of endometritis and wound infection. Strict diagnostic criteria were used for infectious outcomes. A patient was diagnosed with endometritis if she had an oral temperature of 38°C after the first 24 hours following the procedure and either (1) fundal or lower abdominal tenderness greater than expected or (2) foul-smelling or purulent lochia.16 Endometritis was diagnosed only if other causes for the patient's signs and symptoms were not identified. Patients had to be treated with intravenous antibiotics for a diagnosis of endometritis to meet our study definition. The diagnosis of wound infection required wound opening >1 cm or other surgical intervention (such as laparotomy or debridement of tissue) plus at least 1 of the following: (1) purulent drainage from the wound, (2) erythema or induration of the surrounding tissues, (3) maternal oral temperature >38°C, or (4) radiographic evidence of infection. Secondary outcomes were defined before the study was initiated and included the need for wound opening >1 cm because of wound hematoma or seroma, hospital readmission, and need for intravenous antibiotics after the first 24 hours after the procedure. We also collected data on immediate neonatal outcomes that included Apgar scores, special care nursery or neonatal intensive care unit (NICU) admission, umbilical artery pH, O2, and CO2, and antibiotic administration after birth. Physicians were provided with educational materials regarding the diagnosis of endometritis and wound infection before study initiation and intermittently throughout the study period. All outcomes were assessed by the primary care team, and the diagnoses were abstracted from the chart by study personnel (research nurse or investigator). The medical record was reviewed at the time of the 2-4 week postoperative visit, and all women who did not return for a postoperative visit within 4 weeks or who had planned follow-up visits at an outside clinic were contacted by the research nurse by telephone to inquire about postoperative complications. The data collection form was used as a prompt during the telephone interviews to ensure standardization. The primary comparison for our study was supplemental oxygen vs standard care with respect to the primary outcome of infectious morbidity. Data were analyzed with the intent-to-treat principle, and a separate analysis was completed that analyzed only those patients who were treated according to protocol.

Analysis included unpaired t tests for normally distributed continuous variables, Mann-Whitney U test for nonnormally distributed continuous variables, and chi-squared or Fisher's exact tests for categorical data. Stratified analyses with the Mantel-Haenszel test and logistic regression were performed to assess for confounding and interaction. Risk ratios with 95% confidence intervals were calculated for outcomes by treatment arms. A probability value of < .05 was considered significant. No interim analyses were planned or conducted. The study protocol included a continuous (daily) process of monitoring and reporting of maternal and neonatal adverse events by the research team. Adverse events were reported to the Human Research Protection Office and the institutional review board as they occurred.

For the initial sample size calculation, we estimated the rate of infectious morbidity, which consisted of endometritis and wound infection, to be 15%. To detect a 50% reduction in surgical site infection with 80% power and an alpha error of 0.05, 278 women per arm were required. After 50% of the subjects had been enrolled, the calculated sample size of 556 women was increased by approximately 10% to account for the observed loss to follow-up rate, which resulted in a final sample size of 606 women.