Material and Methods

We conducted a retrospective observational study at Lille University Hospital, France. All women who received ICS to reduce the need for allogeneic blood transfusion during a scheduled or urgent cesarean section or postoperative PPH were included. Women transferred because of postoperative PPH whose cesarean section was performed in a primary hospital were not included. The inclusion period extended from June 2007 to September 2018.

ICS protocol

Our device was a continuous autotransfusion system (CATS, Fresenius Laboratory, Germany), which performed the blood washing and salvage. Women identified at high risk of PPH, those with a rare blood type, and those who refused the use of allogeneic products in the event of cesarean section were eligible for ICS. In cases of emergency, the anesthesiologist could implement ICS during PPH.
The blood-collecting system was set up as the patient entered the operating theater. Two separate suction
utotransfusion depended on clinical parameters, bleeding flow, and hemoglobin level. In case of contamination of the surgical site with methylene blue or iodine antiseptics or in case of maternal sepsis at the time of the cesarean section, the collected blood was not reinfused. Autotransfusion was performed using a transfusion filter after passing the fluid through a second standard 40 µm filter.
A transfusion traceability sheet was sent to the hospital’s hemovigilance service. Women with rhesus (Rh) incompatibility received an injection of Rh immune globulin, whose dose was adapted to the Kleihauer test result and cell count. An antiglobulin test was recommended for each patient 2 months after autotransfusion.
ICS could be used during a scheduled or urgent cesarean section. The severity of the urgent need for cesarean section was classified according to the color coding used in our maternity ward. Each color is defined by a decision-to-delivery interval: ≤20 min is coded as red; ≤30 min as orange, and no time limit as green11. PPH was defined as a rate of bleeding >500 mL within 24 h and was considered to be severe when blood loss was >1000 mL.

Indications for ICS

The women included in our study could have 1, 2, or 3 of the following indications for ICS: placenta abnormality, such as placenta accreta spectrum or placenta previa; risk factors for uterine atony, such as uterine fibroids, multiple pregnancy or fetal macrosomia; uterine scarring caused by myomectomy or anterior cesarean section scar(s); postoperative bleeding, defined as PPH within 24h after caesarean section delivery; maternal heart disease; pregnancy-related vascular disorders such as preeclampsia, HELLP syndrome, or retroplacental hematoma; transfusion difficulties, such as having a rare blood group or sickle cell disease, or being a Jehovah’s Witness; hemostasis disorder; severe prepartum anemia, defined as hemoglobin concentration <10 g/dL; history of severe postpartum haemorrhage; or other indications, such as uterine rupture, acute fatty liver of pregnancy or a maternal disease.

Assessment criteria

The primary outcome for assessing the effectiveness of ICS was its ability to enable the reinfusion of autologous blood. Data about ICS management were collected prospectively during cesarean section on autologous-transfusion traceability sheets. Anthropometric, obstetric, and anesthetic variables were collected retrospectively from patients’ medical records.

Statistical analysis

The frequency of the different indications of ICS was compared between two groups of women: those who received ICS and autotransfusion and those who received ICS without autotransfusion.
Qualitative variables were described as frequency and percentage, and quantitative variables as median and interquartile interval.
Bivariate analyses were performed using the chi-squared test or exact Fisher test (in case of theoretical numbers <5) for qualitative variables.
Multivariate analysis was performed using a multivariate logistic regression model with a step-by-step backward selection procedure. The logistic regression model was performed for variables with a P value < .1 in the bivariate analysis.
A receiver operating characteristic (ROC) curve was created to study the interrelationship between autotransfusion and the bleeding volume and, predict a bleeding threshold.
Statistical analyses were performed using SAS software, version 9.4. The level of significance was set at P < .05.

Ethics

A statement was submitted to the National Commission on Informatics and Liberties. Patients consented to the collection of their clinical data. Our research received approval number CEROG 2019-OBST-1001 from the Ethical Review Committee “Comité d’éthique de la recherche en obstétrique et gynécologie”.