Human research ethics approval was obtained (with approval# MF2058-2021-803 dated 23rd January 2023) for this study. The study is reported in compliance with the STROBE checklist [13].

This is a retrospective descriptive cohort study.

All pregnant women presenting with primary PPH and who were managed with uterine artery embolization (UAE) during the period of February 2012 till March 2020 at Al Ain Hospital, Al-Ain City, United Arab Emirates were included in the study. These patients had 1) persistent bleeding despite using other less invasive methods like uterotonic medications and mechanical interventions (intrauterine balloon tamponade. 2) The hemodynamic status was stabilized after initial resuscitation to allow time for UAE with an aim to preserve the uterus.

UAE was not used in patients with (1) severe hemorrhagic shock judged clinically to require immediate surgical intervention with concern to delay of management if UAE was instead performed, (2) severe allergy to contrast media, (3) pre-existing kidney failure, (4) coagulopathy unresponsive to correction, (5) extra-uterine bleeding including vaginal bleeding and cervical tears, 5) severe pelvic infection, and (6) uterine rupture. There were 80 patients studied in this cohort, one of them had missing important outcome data and was not included in the comparative statistical analysis between emergency and elective UAE.

We used the Seldinger technique to access the common femoral artery either with fluoroscopy or ultrasound guidance followed by retrograde approach to reach the uterine artery of the opposite side [12] (Fig. 1). Unilateral access was used in emergency cases while bilateral access was used in elective cases. When the location of a laceration was unknown, a nonselective aortogram was used to screen the injury. A Roberts Uterine Catheter (RUC) (Cook Inc.) located in the abdominal aorta can select the contralateral or ipsilateral hypogastric artery. Selective catheterization of the uterine artery using the 5-French RUC or Cobra (using the Waltman loop) was preferred. If this could not be achieved, a microcatheter was inserted coaxially through the 5 F Catheter. When focal injury was found, the road map approach was used to selectively cannulate the terminal branches of the uterine or hypogastric artery. Active bleeding foci were defined as: (1) extravasation of the contrast media; (2) pseudoaneurysms; or (3) abrupt cut off an artery.

Bilateral uterine arteries catheterization in Angio-suite before going to operation room for Caesarean section showing the catheters (white arrow), uterine arteries (yellow arrow) and the head of the fetus (yellow arrow heads)

When bleeding was non focal, absorbable gelatin sponge (Gelfoam; Pfizer Inc., New York, NY) was utilized (Fig. 2). It was divided into very tiny cubes, or pledgets that are administered hydrostatically via the catheter. The Gelfoam cubes were mixed to create the slurry. We diluted the mixture 1:1 with iodinated contrast and regular saline to get the “cake frosting,” consistency. Bilateral internal iliac artery arteriograms were used to locate any further bleeding sites. If the patients are suspected to have another source of bleeding (like the ovarian arteries or round ligament arteries), then aortography was performed to locate the source of bleeding.

Angiogram on the C-ARM X-ray machine after a Caesarean section (A) showing a catheter (white arrow), bleeding from the right uterine artery (yellow arrows), abdominal pack (black arrow), and surgical tools (white arrowhead) in the field. Angiogram on the C-ARM X-ray machine in the operating room after embolization (B) showing stagnant flow in both internal iliac arteries after Gelfoam embolization. The Catheters were removed while the surgical tools are still in the field

When bleeding was severe, or the patient was hemodynamically unstable, rapid bilateral nonselective embolization of the anterior divisions of the internal iliac or hypogastric arteries was performed. Speculum examination was always performed on-site for assessment of bleeding cessation following the procedure.

These included demography of the patients (age, BMI, gestational diabetes, number of previous pregnancies and C-sections), predisposing factors for bleeding including the position of the placenta, placenta accreta, previous history of PPH and bleeding tendency, the mode of delivery whether C-section or vaginal delivery, and the urgency of the C-section, the need for general anesthesia and the management of the placenta, the blood product transfusions, stoppage of bleeding and the need for a hysterectomy.

Categorical data were presented as number (valid percentage) excluding missing data. Continuous and ordinal data were presented as median (25-75% interquartile (IQR) range). Fisher’s Exact test was used to compare the categorical data of two independent groups while Mann Whitney U test was used to compare ordinal or continuous data of two independent groups [14]. The patients were divided into two groups, those who had emergency interventional embolization and those who had elective interventional embolization. Variables who had a loose p value of less than 0.1 were entered into a direct logistic regression model [15] to define the factors predicting emergency interventions. Data analysis was performed using the Statistical Package for the Social Sciences (SPSS version 28; Chicago, IL, USA). A p value of less than 0.05 was accepted as statistically significant.