Arterial embolisation for persistent primary postpartum haemorrhage: before or after hysterectomy?


Dr A. Bloom, Hadassah University Hospital, P.O. Box 12141, Ein Karem, Jerusalem 91120, Israel.


Arterial embolisation is a recognised treatment for postpartum haemorrhage (PPH). In this retrospective study, we evaluate its use in the management of persistent PPH. Records of all births during a 54 month period at a university hospital were analysed. Two sub-groups were identified. Group I (n= 5), underwent embolisation after hysterectomy and Group II (n= 4), had embolisation as a first-line theraphy without hysterectomy. Of 20,215 births, there were 636 cases of PPH (3.1%). Nine required embolisation to control bleeding (1.4%). Group I needed multiple surgical procedures, had a larger pre- and post-operative blood requirement (12–100, median 22 units, vs. 6–12, median 8.5 units), longer embolisation (33–93, median 54 minutes, vs 20–66, 47 minutes) with a larger radiation exposure (5194–9067, median 6301 dGy, vs. 269–3862, median 950 dGy), a longer intensive care stay (3–7, median four days vs. 0–1.5, median one day), and more complications, when compared with Group II. Three of four women from Group II resumed menstrual function. Embolisation prior to hysterectomy may be preferable to embolisation after hysterectomy for the control of PPH.


Postpartum haemorrhage (PPH) is one of the major causes of maternal death worldwide,1 with a reported incidence of approximately 5% of all deliveries. However, difficulties in evaluating blood loss may result in under-reporting of this condition. Traditionally, PPH has been defined as the loss of more than 500 mL after placental delivery. Nonetheless, about 40% of all women who deliver vaginally bleed more than this amount when measured quantitatively and 5% lose more than 1000 mL of blood.2 When PPH is suspected, efforts should be made to immediately identify its cause and to commence appropriate therapy. This includes prompt vigorous crystalloid and blood replacement, repair of lacerations in the birth canal and removal of retained placental remnants. First-line therapy for uterine atony is uterine massage, together with uterotonic agents such as oxytocin, methylergonovine and 15-methyl prostaglandin F2. If significant haemorrhage persists despite appropriate management, there are two options to control bleeding. The first is surgery, which may include the B-Lynch procedure or ligation of the uterine blood supply. If these measures fail, then hysterectomy is necessary. The second option is angiographic arterial embolisation. Despite recent reports in the literature of successful control of PPH using arterial embolisation,3–9 there is no clear consensus as to which point in the management algorithm it should be considered. We present a series of patients with life-threatening primary PPH, who were successfully treated by angiographic embolisation, after the failure of conventional therapies, which included hypogastric artery ligation and hysterectomy in several cases. The appropriate timing of embolisation in the management of PPH is discussed.


This is a retrospective review of the database and patient files of the Departments of Obstetrics and Gynaecology and Interventional Radiology at a university teaching hospital. During a 54-month period up to August 2002, there were 20,215 deliveries with 636 (3.1%) cases of primary PPH. Nine women (1.4%), age range 17–41 years, median 27 years, failed to respond to conventional medical and surgical management. Angiographic embolisation was used to control bleeding in these patients, who are the focus of this study.

Patient data are summarised in Table 1. The decision to perform angiography and its timing were based on the individual preferences of the responsible obstetrician. The patients were divided into two groups, Group I who required embolisation after hysterectomy (and hypogastric artery ligation in two cases) and Group II who did not undergo hysterectomy or hypogastric artery ligation.

Table 1.  Patient data.
PatientAge (years)ParityAetiology of PPHSurgery (prior to embolisation)Transfusion requirement: pre-embolisation and post-embolisation (units of packed cells)Duration of angiography (minutes)Radiation dose (dGy)Intensive care stay (days)ComplicationsMenses after embolisation
Group 1: Patients in whom embolisation was performed after hysterectomy and in two cases after bilateral hypogastric artery ligation.
141G6P3AtonyHysterectomy33 and 39352847DIC, ARDSNo
225G1P1AccretaPacking, laparotomy, bilateral hypogastric artery ligation, hysterectomy100 and 38573194DIC, ARDS, abdominal compartment syndrome, ileus, haematomaNo
339G13P10 CS4PraeviaCaesarean section, packing, hysterectomy22 and 55451945DIC, pelvic abscessNo
426G3P3AccretaCaesarean section, bilateral hypogastric artery ligation, hysterectomy20 and 37490673DIC, abdominal compartment, syndromeNo
540G9P8AtonyCaesarean section, hysterectomy12 and 3333DIC, haematomaNo
Group 2: Patients in whom embolisation was performed without hysterectomy or hypogastric artery ligation.
117G2P1AtonyNone7 and 0202691.5Pulmonary oedemaYes
227G1P1Caesarean sectionCaesarean section6 and 0419500None
325G1P1Caesarean sectionCaesarean section, laparotomy12 and 1531NoneYes
439G6P4 CS4Praevia, lacerationCaesarean section10 and 16638621DICYes

Angiography was performed in the angiography suite using a right transfemoral approach in all cases. In all but one patient (patient 1, Group II), a diagnostic pelvic arteriogram was initially performed to identify the source of bleeding. This was followed by either selective embolisation of a single bleeding vessel (n= 5) or by embolisation of the anterior divisions of the internal iliac arteries if the source was bilateral (n= 4). Embolisation materials used included Gelfoam pledgets (Upjohn, Kalamazoo, Michigan) and/or metallic platinum tufted coils (Cook, Bloomington, Indiana), at the discretion of the angiographer performing the procedure. Embolisation was defined as completed when cessation of flow, in the affected vessel(s) and distal branches, was identified. In the presence of an abnormal coagulation profile, an access sheath was left in the common femoral artery until coagulation was normalised (<24 hours). Patients were managed in the intensive care unit, subsequently in the ward and then followed in the outpatient clinic. The following outcome measures were recorded and analysed for statistical significance using the Student's t test: the number and type of surgical procedures required prior to embolisation, pre-embolisation and post-embolisation blood transfusion requirement, duration of arteriography, radiation dose, intensive care stay, complications and restoration of menstruation.


The causes of PPH in the nine patients who underwent embolisation were uterine atony (n= 3), placenta accreta (n= 3, one with praevia), placenta praevia with chorioamnionitis (n= 1) and injury related to caesarean hysterectomy (n= 2).

Table 1 summarises the outcome data for both groups of patients. Embolisation without hysterectomy, irrespective of the cause of PPH was associated with less blood transfusion requirement (6–12, median 8.5 units, vs 12–100, median 22 units), less radiation exposure (dosage 269–3862, median 950 dGy, vs 5194–9067, median 6301 dGy), shorter intensive care stay (0–1.5, median 1 day, vs 3–7, median 4 days), fewer complications and preservation of uterine function, than embolisation after hysterectomy. These differences were not statistically significant.

The bleeding vessel(s) was identified angiographically in eight patients prior to embolisation. In one patient with uterine atony, bilateral internal iliac artery embolisation was performed empirically without a prior diagnostic pelvic arteriogram. Bleeding vessels included uterine, lumbar, ovarian and inferior epigastric arteries. Embolisation was successful in all cases and was confirmed by cessation of flow in the vessel and its branches. It was accompanied by an immediate improvement in haemodynamic parameters and by correction of haematocrit during the subsequent 24–48 hours.

There were no complications due to angioraphy and embolisation. Major complications related to PPH and surgical interventions included disseminated intravascular coagulation (DIC, n= 6), abdominal compartment syndrome (n= 2), adult respiratory distress syndrome (ARDS, n= 2), pulmonary oedema (n= 1), pelvic abscess (n= 1), small bowel ileus (n= 1) and an abdominal wall haematoma (n= 2). Combined medical and surgical measures led to resolution of these complications in all cases.


PPH is a leading cause of maternal mortality and morbidity accounting for approximately 4% of maternal deaths in the United States.10 Primary PPH is most commonly caused by uterine atony, birth canal lacerations or retained placenta. Less common causes are abnormalities of placental implantation (mainly placenta accreta, and rarely placenta increta or precreta), uterine rupture and uterine inversion. Secondary PPH is most often due to retained gestational products.

The treatment of primary PPH is staged and is directed to the cause of haemorrhage.11 Initially, physical examination will suggest the diagnosis, and after fluid and blood product administration, more directed therapy is commenced. This may include the use of pharmaceutical agents to promote uterine contraction, uterine massage, repair of lacerations, removal of retained products of conception and tamponade. If these measures are unsuccessful, the next stage is usually laparotomy with haemostatic suturing, including the B-Lynch procedure for uterine atony, hypogastric artery ligation or hysterectomy. However, by the time surgery is performed, the patient has usually received multiple transfusions of blood and blood products. This may lead to the development of systemic complications, such as DIC, and ARDS, thereby increasing the morbidity and mortality of any subsequent surgical procedure.

Recently, the B-Lynch compression suture has been proposed as a simple first surgical step to control atonic uterine bleeding.9 It reduces blood flow to the uterus from its lateral margins and from the placental bed vessels by the apposition of the anterior and posterior uterine walls. Although promising, experience with this technique and its potential complications is limited. Surgical ligation of the hypogastric (internal iliac) arteries is a relatively ineffective method for controlling PPH. The reported success of this procedure is 42%.12 It was also noted that patients requiring hysterectomy after hypogastric artery ligation had an increased blood loss, morbidity and mortality when compared with hysterectomy alone. Although more contemporary series report higher success rates for hypogastric artery ligation, the outcome remains poor in patients with uterine atony.4 The reason for these failures is that ligation is performed close to the origin of the hypogastric artery. Although this results in decreased flow proximally, distal collateral inflow to the gravid uterus continues. It has been reported that bilateral hypogastric artery ligation results in a 49% reduction in blood flow and an 85% reduction in pulse pressure.13 However, this study was performed on non-pregnant women. The uterus and pelvis at term have a much increased blood supply and probably a more developed system of collateral flow. Improved results for the surgical control of primary PPH have been reported with the use of uterine artery ligation, but this procedure has not been extensively studied.5 Hysterectomy of the gravid uterus in the setting of major PPH uncontrolled by other means is associated with significant morbidity,3 in addition to the loss of reproductive capability.

Transcatheter embolisation offers an effective means of controlling PPH while obviating much of the morbidity and mortality of open surgical procedures. In addition to avoiding general anaesthesia and open surgery, embolisation has a number of other advantages. The use of particles allows rapid and more distal vessel occlusion, thereby decreasing collateral inflow to the uterus and pelvis, and thus providing more effective haemostasis. Sources of bleeding other than the hypogastric arteries may be identified and embolised.7 In the current series, unexpected sources of haemorrhage were encountered in three patients, two of whom had already undergone ligation of the hypogastric arteries and subsequent hysterectomy. These would have been difficult to identify and control with a further operation, which would have been likely to result in significant morbidity.

An initial diagnostic pelvic angiogram is performed via the common femoral artery. In cases of massive bleeding where the aetiology is clear prior to angiography (e.g. uterine atony without surgery), both hypogastric arteries are rapidly occluded with small pledgets of Gelfoam. Gelfoam is a temporary occluding agent, which dissolves in one to three weeks. The pledgets are carried distally by the blood flow, occluding small arteries and preventing continued bleeding from collateral inflow. Frequently, the diagnostic arteriography identifies a specific arterial bleeding point, which can be selectively embolised avoiding the need for more widespread occlusion. The extent and number of vessels that need to be occluded determine the embolic agent(s) selected.

Two recent reviews in the English literature reported 163 cases where transcatheter embolisation was used to control PPH.5,6 The success rate was 95% or greater, requiring repeat embolisation to control persistent haemorrhage in some cases.6 Hysterectomy was performed in patients where embolisation did not ultimately control bleeding. Major complications associated with uterine artery embolisation in obstetric patients are rare, with an overall complication rate of approximately 9%. These include transient fever, transient buttock ischaemia, transient foot ischaemia, iliac artery perforation and abscess. We believe that many of these reported complications could be avoided by the use of appropriate sized embolisation particles by scrupulous technique that avoids non-target embolisation and by the use of modern catheters and guide wires, which are flexible, trackable and non-traumatic. Our experience confirms that there is significant morbidity associated with major surgical interventions in this patient population and no significant complications from embolisation.

Despite reproducible data for the success of embolisation in controlling PPH, it is unclear what may be the optimal timing of embolisation? In the present series, patients from Group I underwent multiple surgical procedures and had larger pre-and post-embobilisation transfusion requirements than Group II. They all developed DIC, defined by a prolonged prothrombin time, increased International Normalised Ratio (INR) and thrombocytopaenia, prior to angiography. Only one of four patients from Group II developed DIC. Patients from Group I had a larger radiation exposure, more complications and longer ICU stay than Group II. Since embolisation has a high success rate and a relatively low complication rate, we think that in selected cases it should be incorporated into the treatment algorithm at an earlier stage. This approach might avoid the morbidity from multiple transfusions and from the surgical procedures themselves, which were the cause of ongoing haemorrhage in several cases in this series. An additional advantage of embolisation is uterine preservation with the possibility of future pregnancy.6 Based on our experience, we believe that hysterectomy should be performed early in the management of PPH, if initial measures have failed. However, if bleeding continues, if an unusual source of bleeding is suspected or if uterine preservation is desired, then embolisation should be performed immediately. If angiographic embolisation is not available in a hospital, then elective referral of patients at increased risk of PPH to a centre where it is available should be considered. A prospective comparative study to evaluate the role of embolisation in the management of PPH would be desirable but not necessarily feasible.


Transcatheter embolisation is an effective minimally invasive technique for treating primary PPH. It has a low complication rate and appears to be associated with reduced blood transfusion requirement, shorter ICU admission and fewer complications, together with preservation of reproductive potential, when performed prior to hysterectomy. In selected cases it might be introduced into the treatment algorithm before proceeding to open surgical procedures, in centres where trained interventional radiologists are available.

Accepted 26 March 2004