Critical Care Obstetrics part 33 doc

Etiology and Management of Hemorrhage
309
tion of both mother and fetus [11] . If the fetus is alive and of a
viable gestational age at presentation, urgent delivery by cesarean
section is indicated unless vaginal delivery is imminent. If severe
fetal distress is suspected prior to cesarean, presence of fetal heart
activity should be verifi ed before anesthesia is commenced.
Major abruption suffi cient to cause fetal death is life threaten-
ing for the mother. Blood loss is frequently 50% or more of blood
volume [20] ; up to 5 L of blood may extravasate into the myo-
metrium, with little or no revealed bleeding [22] . Thromboplastin
release is a powerful trigger for disseminated intravascular
coagulation and is strongly uterotonic. In a series of 141 cases of
abruption severe enough to kill the fetus, plasma fi brinogen
concentration was below 150 mg/dL in 38%, and below 100 mg/
dL in 28%, and in all cases developed within 8 hours of the onset
of symptoms [20] . Delivery must be expedited, with vaginal deliv-
ery the preferred route unless contraindications exist. Cesarean
delivery in patients with coagulopathy leads to diffi culty in
achieving surgical hemostasis, but is indicated in the presence of
fetal jeopardy, prior classic cesarean delivery, fetal malpresenta-
tion or if the patient is remote from delivery with worsening
coagulopathy. While the fetus is undelivered, correction of hypo-
volemia, blood loss, and coagulopathy must continue. In the
presence of fetal demise, all efforts should be made for a success-
ful vaginal delivery unless a contraindication exists or the mother
becomes increasingly unstable. In the preterm infant, delivery can
be postponed in the absence of profound hemorrhage or coagu-
lopathic abnormalities and steroids for fetal lung maturity can be
administered as long as maternal and fetal well - being are assured.
This may require close follow - up or a prolonged hospital admis-

sion. The use of tocolytics is controversial, with data suggesting
a prolongation of pregnancy with the use of magnesium sulfate
[23] . The β - sympathomimetics cause tachycardia, masking the
clinical signs of hemorrhage, and should not be used. Thus tocol-
ysis should be used with caution in select, stable patients with no
evidence of acute hemorrhage [23] . The major determinant of
maternal outcome is adequate replacement of fl uid and blood,
rather than time to delivery [11] .
Postpartum hemorrhage must be anticipated following a
severe placental abruption, and prophylactic uterotonic drugs
should be considered. Uterine atony may occur since myometrial
contractility is impaired by fi brin degradation products [22] .
Persistent uterine atony despite the administration of uterotonics
may require hysterectomy. Maternal deaths from abruption are
frequently post partum, when ongoing blood loss occurs in
patients with inadequate correction of hemorrhagic shock and
coagulopathy prior to delivery. Acute renal tubular and cortical
necrosis may result from the products of the coagulation cascade
in addition to renal ischemia due to hypovolemia.
Uterine r upture
Uterine rupture may occur in an unscarred uterus, at the site of
a previous uterine scar from a cesarean section or a full - thickness
> 50% separation [16] . Infants born after abruption have increased
rates of cystic periventricular leukomalacia and intraventricular
hemorrhage compared to age - matched controls, most likely a
result of oxygen and nutrient deprivation prior to delivery [17] .
The classic signs and symptoms of placental abruption include
vaginal bleeding accompanied by uterine tenderness, painful
tetanic contractions and non - reassuring fetal heart rate patterns.
However, not all signs may be present simultaneously. Vaginal

bleeding may be concealed, leading to delays in seeking medical
help by the woman as well as in diagnosis by the physician.
Abruption can even present as simple unexplained preterm labor.
Sonographic evaluation of the placenta will fail to reveal over
50% of abruptions. The appearance of hemorrhage changes over
the course of acute hemorrhage to stable hematoma, making the
diagnosis more challenging [18] . In cases where bleeding is visual-
ized sonographically, the likelihood of abruption is very high. A
thickened placenta ( > 5 cm) may also suggest the presence of
abruption. Retroplacental hemorrhage confers a worse prognosis
for the fetus [18] . The Kleihaur – Betke (KB) test has not been
proven to be clinically useful in the evaluation of abruption. In one
study where over 25 placentas with histologic evidence of abrup-
tion were analyzed, there were no positive maternal KB tests. In
the same study, there was a 9% false - positive rate [19] . The clinical
utility of the KB test is mainly to help formulate the appropriate
dosage of Rh immune globulin for the Rh - negative woman.
In a major abruption blood extravasates into the myometrium
and the uterus becomes “ woody hard ” with fetal parts no longer
palpable – the Couvelaire uterus. Hemorrhagic shock and coagu-
lopathy may be present. Blood loss may be over 50% of maternal
blood volume with abruption severe enough to kill the fetus [20] .
Coagulation defects appear to develop rapidly after the occur-
rence of a severe abruption, within a few hours or even in minutes
[11] . Upon presentation, evaluation of hematologic parameters
including coagulation studies should be completed after immedi-
ate intravenous access has been obtained and resuscitation has
begun. The blood bank should be informed about the need for
blood and blood products, keeping several units ahead. If avail-
able, whole blood is preferred in these cases secondary to the

additional benefi t of volume expansion. A Foley catheter should
be placed for hourly urine output measurement. Particular atten-
tion should be given to vital sign determination and coagulation
studies should be performed regularly.
Anesthesia should be involved early with patient care. Co -
ordinated efforts involving blood and product replacement in the
operating room in the event of a cesarean or vaginal delivery will
maximize resuscitative efforts in these as well as other cases
involving hemorrhage.
Invasive monitoring may be necessary. Women with pre -
eclampsia complicated by placental abruption need particular
care in resuscitation. These patients tolerate hypovolemia poorly
because of the contracted intravascular volume and low cardiac
output, and are also highly susceptible to pulmonary edema due
to volume overload [21] . Coagulopathy is uncommon with a
surviving fetus; prompt delivery prevents further decompensa-
Chapter 22
310
section is controversial. Data indicate that induction of labor and
cervical ripening with prostaglandins (PGE - 1 and PGE - 2) may
carry a high risk of uterine rupture [31] . Gynecologic uterine
surgery including laparoscopic myometomy is also considered a
strong risk factor for uterine rupture. Rates of rupture after myo-
mectomy have been found to be as high as 1.5% [32] . Risk of
rupture may have been minimized in this series since > 50% of
these patients underwent cesarean delivery before labor. Although
the posterior fundus is considered the weakest part of the uterus,
most ruptures occur in the lower anterior segment during labor
and at the fundus during prelabor - associated ruptures [33] .
Late complications of uterine hysteroscopy include uterine

rupture [34] .
Spontaneous uterine rupture has been documented in mul-
tiparous women with placenta accreta. Although rare, this has
also been found in primiparous women with placenta accreta
[35,36] . Congenital uterine anomalies are also a risk factor for
uterine rupture, including those related to DES exposure. In one
series, fi ve cases of rupture were discovered in primiparous
women with bicornuate uteri [32] . Literature suggests that an
interdelivery interval of < 24 months of gestation was associated
wth a 2 – 3 - fold increase in the risk of uterine rupture compared
with an interval of greater than 24 months [37] . The same study
demonstrated a twofold increase in rupture rates when single -
layer closure was used to reapproximate the uterine incision in
the previous cesarean delivery. Conversely, there are other studies
demonstrating no difference in maternal or neonatal outcomes
with single - or double - layer closure techniques [38 – 40] .
The role of postpartum fever has been invested as a risk factor
for rupture as well. In one study, the odds of rupture were four
times greater than controls in women with postpartum febrile
morbidity [41] . Although fetal macrosomia decreases the likeli-
hood of successful vaginal delivery after cesarean delivery, mul-
tiple studies report no difference in rates of uterine rupture
[42 – 44] .
Management options consist of surgical repair and hysterec-
tomy. Published case series, many spanning several decades, vary
widely in reported use of each technique, with hysterectomy rates
of 26 – 83% [24] . Most authors consider hysterectomy to be the
procedure of choice for uterine rupture [25,45] . Subtotal hyster-
ectomy may be performed if the rupture is confi ned to the uterine
corpus. Evidence has shown that subtotal hysterectomy is associ-

ated with decreased operating time, lower morbidity and mortal-
ity and shorter hospital stay when compared to surgical repair
[46] . Suture repair may be considered when technically feasible
and there is a desire for future fertility. However, there is an
increased risk of recurrence, which may be fatal. A meta - analysis
from 1971 provides the most comprehensive data [47] . This
analysis includes 194 women, with a total of 253 pregnancies fol-
lowing uterine rupture; two maternal deaths occurred. Overall,
repeat rupture occurred in 6% with a previous lower segment
rupture, 32% with a previous upper segment rupture, and 14%
where the site of previous rupture was unknown. Of note, three
women in this series had repeated rupture in two or three subse-
incision secondary to gynecologic surgery. The overall rate varies
from 2 to 8 per 10,000 deliveries [24] . Asymptomatic and blood-
less dehiscence of a previous cesarean section scar may occur
during subsequent vaginal delivery, and may also be seen at
repeat cesarean section in women who have not labored. Uterine
rupture is generally considered to include only cases with com-
plete separation of the wall of the pregnant uterus, with or
without expulsion of the fetus, that may endanger the life of the
mother and/or fetus [25] .
Rupture may occur antenatally or intrapartum; however, it is
commonly fi rst suspected post partum. The most common clini-
cal sign in labor is the sudden onset of fetal distress, reported in
81% of cases [24] . The most common fetal heart rate disturbance
is prolonged bradycardia [26] . Abdominal pain, cessation of con-
tractions, and recession of the presenting part are less common.
Data regarding the placement of an intrauterine catheter monitor
to predict impending uterine rupture have revealed no correla-
tion between uterine contractility patterns and rupture [27,28] .

Bleeding may be intraperitoneal and into the broad ligament
rather than revealed vaginally. Over 50% of cases are fi rst diag-
nosed after delivery, when intractable hemorrhage follows
precipitate, spontaneous or instrumental vaginal delivery.
Alternatively, if bleeding is concealed, profound shock may occur
before rupture is suspected. Uterine rupture should be consid-
ered in every obstetric patient with hemorrhagic shock in whom
the cause is not immediately apparent.
Rupture of an unscarred uterus is frequently related to obstet-
ric intervention. This includes use of uterotonic drugs for induc-
tion or augmentation of labor, mid - cavity forceps delivery or
breech extraction with internal podalic version [25] . Prolonged
labor in the presence of cephalopelvic disproportion or malpre-
sentation may also cause uterine rupture; this is most common
in underdeveloped countries with poor access to medical care,
but also occurs due to inappropriate management of labor in
industrialized nations. External trauma may cause uterine rupture
at any gestation. Grand multiparity also increases the risk.
Increased blood loss, transfusion rates, and fetal mortality are
clinically signifi cant with rupture of an intact uterus.
Rupture of a previously scarred uterus is more common than
rupture of the intact uterus. Of 23 cases of uterine rupture
reported from New Orleans between 1975 and 1983, 61.3%
(14/23) involved rupture of a previous cesarean section scar, with
six cases occurring prior to labor, fi ve during labor, and the
remaining three unknown [25] . The overall risk of uterine
rupture for women attempting a trial of labor following lower
segment cesarean section is 1%, but higher if the trial of labor is
unsuccessful [29] . A previous classic cesarean section has a risk
of rupture of 3 – 6%, increased to 12% if a trial of labor takes place.

Spontaneous rupture of a classic cesarean section scar has been
reported as early as 15 weeks of gestation [30] . Use of uterotonic
drugs (prostaglandins, oxytocin, and misoprostol) in the pres-
ence of a cesarean section scar is associated with an increased risk
of rupture. The risks are diffi cult to quantify, and use of these
drugs in patients undergoing vaginal delivery after cesarean
Etiology and Management of Hemorrhage
311
minute. Rapid cessation of blood fl ow at the placental site is
therefore essential, occurring via myometrial contraction causing
compression of uterine vasculature. If this fails to occur, life -
threatening hemorrhage may rapidly ensue. Many risk factors for
uterine atony have been identifi ed. These include high parity,
chorioamnionitis, uterine fi broids, overdistension of the uterus
(multiple gestation, fetal macrosomia, polyhydramnios), labor -
related factors (precipitate labor, prolonged labor, oxytocin use)
and uterine - relaxing drugs (magnesium sulfate, halogenated
anesthetic agents, nitroglycerin) [49] . Antepartum hemorrhage
due to both abruption and placenta previa carries an increased
risk for postpartum hemorrhage. Previous postpartum hemor-
rhage confers a 10% risk of recurrence. Management of uterine
atony is outlined in the next section.
Uterine atony unresponsive to medical treatment may be due
to retained placental fragments. Visual examination of the pla-
centa following removal cannot always exclude this diagnosis; a
placenta appearing complete may be missing an entire or partial
cotyledon or there may be a retained succinturiate lobe. In cases
where the placenta is not easily removed or appears fragmented
or incomplete upon manual extraction, placenta accreta should
be suspected. This condition is not uncommonly associated with

blood loss often greater than 2500 mL [55] . Examination under
anesthesia and evacuation of retained placental tissue are neces-
sary. Care is required with curettage since a postpartum uterus is
easily perforated [56] .
Genital tract trauma is commonly associated with instrumental
delivery. Other risk factors include shoulder dystocia and precipi-
tate delivery. Lacerations may occur throughout the urogenital
system, including perineum, vagina, bladder, cervix, uterus, and
anorectal tissues. Hemodynamic compromise may occur if diag-
nosis and repair are not carried out promptly.
Genital tract hematomas may result in postpartum cardiovas-
cular collapse due to concealed blood loss. Vulvovaginal hema-
tomas that lie below the levator ani may contain 1.5 – 2.0 L of
blood [57] . They probably result from contusion or avulsion of
the vascular supply due to radial stretching of vaginal tissues
quent pregnancies. Other women had an uneventful pregnancy
following uterine rupture, but with a repeat (even fatal) rupture
in a subsequent pregnancy.
If suture repair is performed, elective cesarean section has been
advocated as soon as evidence of fetal lung maturity is obtained
in a future pregnancy. Repair has also been advocated if successful
control of hemorrhage can be attained in hemodynamically
unstable patients, avoiding further blood loss and prolonged
surgery during hysterectomy. Bilateral tubal ligation should be
considered in these cases. The need for massive transfusion
usually accompanies operative management of uterine rupture.
In a study evaluating over 25 peripartum hysterectomies, 98% of
cases required multiple units of blood and blood products. Co -
ordinated efforts with anesthesia and blood bank are of vital
importance when the diagnosis of uterine rupture is suspected.

Primary p ostpartum h emorrhage
Primary postpartum hemorrhage is traditionally defi ned as blood
loss of more than 500 mL within the fi rst 24 hours of delivery [5] .
However, in the 1960s, studies by Pritchard involving
51
Cr - labeled
red cells demonstrated an average blood loss of 505 mL at vaginal
delivery and 930 mL at elective repeat cesarean section [48] . Some
clinicians therefore consider as clinically signifi cant only blood
loss greater than 500 mL at vaginal and 1000 mL at cesarean deliv-
ery, as major obstetric hemorrhage, with a reported incidence of
1.3% [5] . However, visual estimates of blood loss are commonly
inaccurate, with frequent underestimation by at least 30 – 50%
[49 – 51] . Delay in recognizing signifi cant postpartum hemor-
rhage and therefore in instituting management contributes to
maternal mortality from this condition; additionally, the pres-
ence of concealed hemorrhage may not be appreciated. A more
objective, although retrospective, defi nition of postpartum hem-
orrhage is of a 10% change in hematocrit or a need for red blood
cell transfusion. On this basis, the incidence of postpartum hem-
orrhage is 3.9% following vaginal delivery and 6.4% following
cesarean section [52,53] .
Most major hemorrhage occurs within the fi rst hour post
partum. The blood volume expansion of 1.5 – 2.0 L in healthy
pregnant women provides a physiologic reserve for blood loss at
delivery [54] . However, women with a below average increase in
blood volume (pre - eclampsia, low prepregnancy Body Mass
Index) tolerate postpartum hemorrhage less readily and are
therefore more at risk of hemorrhagic shock.
Etiology of p rimary p ostpartum h emorrhage

Causes of primary postpartum hemorrhage can be divided into
four major categories (Table 22.1 ). Prolonged or severe hemor-
rhage is the most common cause of coagulopathy post partum
and exacerbates bleeding due to other causes [22] .
Uterine atony is the most common cause of primary postpar-
tum hemorrhage, accounting for 80% of all cases. At term, the
uteroplacental circulation has a blood supply of 600 – 800 mL per
Table 22.1 Causes of primary postpartum hemorrhage.
Uterine atony
Retained placental tissue
Genital tract trauma
Vaginal lacerations
Cervical lacerations
Vulval hematoma
Broad ligament hematoma
Placenta accreta and other abnormal placentation
Uterine inversion
Uterine rupture
Coagulopathy
Secondary to hemorrhage or obstetric causes
Inherited or acquired bleeding disorders
Anticoagulant drugs
Chapter 22
312
placenta and uterine atony, together with cord traction or fundal
pressure.
The placenta should not be removed prior to uterine replace-
ment; this exacerbates blood loss [64] . Manual replacement
should take place without delay, by placing a hand vaginally with
the fi ngers placed circumferentially and the fundus cupped in the

palm. Replacement is such that the region of the uterus that
inverted last is the fi rst to be replaced, so avoiding multiple layers
of uterine wall within the cervical ring. Uterine relaxation may be
necessary, with β - sympathomimetic agents, magnesium sulfate
or low - dose nitroglycerine [65] . Caution should be exercised with
the use of nitroglycerin secondary to its vasodilatory effect,
further potentiating hypotension and tachycardia. Recruitment
of anesthesia services for rapid intubation provides full uterine
relaxation along with the benefi ts of a controlled operating room
environment, and patient, so that resuscitation/transfusion can
be accomplished expeditiously. Intravaginal hydrostatic replace-
ment is an alternative technique [66] . The vaginal introitus is
occluded and warm saline infused into the posterior fornix from
a meter or more above the patient. Ensuring an adequate vaginal
seal may be diffi cult; a silastic Ventouse cup connected to the
infusion and then inserted into the vagina has been successfully
employed [67] . In the presence of a cervical ring prohibiting
vaginal replacement of the fundus, options include incising the
ring through a vaginal approach. An anterior or posterior vaginal
incision has been described with subsequent repair once the
fundus has been replaced [63] . If these measures fail, laparotomy
is required. Two procedures are described. The fi rst involves step-
wise traction on the funnel of the inverted uterus or the round
ligaments, using ring or Allis forceps reapplied progressively as
the fundus emerges (Huntingdon procedure). If this fails, a
longitudinal incision is made posteriorly through the cervix,
relieving cervical constriction and allowing stepwise replacement
(Haultain procedure). This can also be accomplished vaginally as
described above. Once the uterus is replaced, all relaxants should
be stopped and manual removal of the placenta should follow.

With early diagnosis and prompt replacement of the fundus,
most often laparotomy and hysterectomy can be avoided [62,68] .
It is delay which leads to increased edema, blood loss and associ-
ated morbidities.
Treatment of u terine a tony
Emergency p rocedures
Fundal massage is the simplest treatment for uterine atony, is
effective and can be performed while initial resuscitation and
administration of uterotonic drugs are in progress. If this fails to
rapidly control hemorrhage, bimanual compression may be suc-
cessful. A fi st or hand is placed within the vagina such that the
uterus elevated; stretching of the uterine arteries reduces blood
fl ow. The abdominal hand continues fundal massage, whilst also
compressing the uterus. A urinary catheter may be inserted; this
not only aids assessment of fl uid status, but a distended bladder
during delivery. Spread is usually limited by Colles ’ fascia and
fascia lata. They do not cross the mid - line secondary to the central
tendon. Supralevator hematomas can dissect paravaginally and
extend into the retroperitoneal space [58] . Supralevator hemato-
mas may present with hemorrhagic shock and can be associated
with uterine dehiscense in women with prior cesarean delivery.
Usually, laparotomy is required. Infralevator hematomas present
with pain although other signs and symptoms include fever, ileus,
leg edema and thigh pain. Hematoma formation may also occur
in association with inadequate hemostasis during repair of episi-
otomy or vaginal tears. Evacuation is required for larger hemato-
mas, ( > 3 cm) [59] , best performed by incision through the vaginal
wall to minimize scarring. In vulvar hematomas, bleeding vessels
usually arise from the pudendal artery. In vaginal hematomas, the
descending branch of the uterine artery may be involved. When

attempting evacuation, the bleeding vessels should be identifi ed
and ligated. Frequently vessels may retract and the source of
bleeding cannot be identifi ed. Figure - of - eight sutures may be
applied; alternatively, tight packing of the hematoma cavity may
be necessary. If bleeding continues despite these measures, arte-
rial ligation or angoigraphic embolization may be necessary (see
below).
Broad ligament hematomas may result from uterine tears due
to rupture or traumatic extension of a lower segment cesarean
hysterotomy. Alternatively, deep cervical tears during spontane-
ous or operative vaginal delivery may involve the uterine artery
at the base of the broad ligament [60] . Conservative management
is possible if the patient is hemodynamically stable after vaginal
delivery; however, bleeding may be ongoing, and diagnosis may
only occur following postpartum collapse. Broad ligament hema-
tomas may be apparent clinically by the presence of a tender,
boggy mass suprapubically, with a fi rmly contracted uterus devi-
ated past the mid - line. By the time a mass is clinically apparent,
it may contain several liters of blood, and continued arterial
bleeding may result in broad ligament rupture. Conservative
surgery may be possible, but hysterectomy may be necessary (see
below).
Leiomyoma may increase the risk of postpartum hemorrhage.
A recent study demonstrated a 2.5 - fold increased risk of postpar-
tum hemorrhage in women with at least one leiomyoma diag-
nosed during pregnancy [61] .
Uterine i nversion
Uterine inversion often presents with profound shock, both
neurogenic (due to traction on the uterine ligaments) and
hemorrhagic (if the placenta is separated and uterus is atonic)

in origin. Greater than 90% of patients will present with hemor-
rhage with typical blood loss approaching 2 L [62] . In complete
inversion, clinical diagnosis may be obvious, with the uterus
not palpable abdominally and the fundus visible as a mass
protruding through the introitus [63] . Partial inversion may not
be apparent without vaginal examination, leading to delayed
diagnosis. Risk factors relate to the management of the third
stage of labor. Predisposing factors include fundal insertion of the
Etiology and Management of Hemorrhage
313
causes. Rapid administration of an intravenous bolus of oxytocin
results in relaxation of vascular smooth muscle. Hypotension
with a refl ex tachycardia may occur, followed by a small but
sustained increase in blood pressure. Oxytocin is stable at tem-
peratures up to 25 ° C, but refrigeration may prolong shelf - life.
Methylergonovine/ e rgometrine
Methylergonovine (methylergometrine) and its parent com-
pound ergometrine result in a sustained tonic contraction
of uterine smooth muscle via stimulation of α - adrenergic myo-
metrial receptors [72] . The dose of methylergonovine is 0.2 mg
and of ergometrine is 0.2 – 0.5 mg, repeated after 2 – 4 hours if
necessary. Time to onset of action is 2 – 5 minutes when given
intramuscularly. These agents are extensively metabolized in
the liver and the mean plasma half - life is approximately 30
minutes. However, plasma levels do not seem to correlate with
uterine effect, since the clinical action of ergometrine is sustained
for 3 hours or more. When oxytocin and ergometrine derivatives
are used simultaneously, postpartum hemorrhage is therefore
controlled by two different mechanisms, oxytocin producing
an immediate response, and ergometrine a more sustained

action. In a recent large meta - analysis comparing ergometrine -
oxytocin with oxytocin alone, a small but statistically signifi cant
reduction in postpartum hemorrhage was found with blood loss
greater than 500 mL. However, there were no differences between
the two groups with greater degrees ( > 1000 mL) of blood loss
[74] .
Nausea and vomiting are common side effects. Vasoconstriction
of vascular smooth muscle also occurs as a consequence of their
α - adrenergic action. This can result in elevation of central venous
pressure and systemic blood pressure and therefore pulmonary
edema, stroke, and myocardial infarction. Contraindications
include heart disease, autoimmune conditions associated with
Raynaud ’ s phenomenon, peripheral vascular disease, arteriove-
nous shunts even if surgically corrected, and hypertension.
Women with pre - eclampsia/eclampsia are particularly at risk of
severe and sustained hypertension.
Intravenous administration is associated with more severe side
effects, but onset of action is almost immediate. This route may
be indicated for patients in whom delayed intramuscular absorp-
tion may occur (e.g. shock patients). The drug should be given
over at least 60 seconds with careful monitoring of blood pressure
and pulse. Initial reports suggested that methylergonovine
resulted in hypertension less frequently than ergometrine, but no
difference has since been reported in randomized controlled
trials. Ergometrine and its derivatives are both heat and light
sensitive, and should be stored at temperatures below 8 ° C and
away from light.
Prostaglandins
Prostaglandin F - 2 α results in contraction of smooth muscle cells
[72] . Carboprost/hemabate (15 - methyl prostaglandin F - 2 α ) is an

established second - line treatment for postpartum hemorrhage
unresponsive to oxytocic agents. It is available in single - dose vials
may interfere with uterine contractility. Controlled cord traction,
early cord clamping and prophylactic oxytocic administration
reduce postpartum hemorrhage by 500 – 1000 mL.
Aortic compression is a temporizing procedure that can be
used in life - threatening hemorrhage, particularly at cesarean
section. A closed fi st compresses the aorta against the vertebral
column just above the umbilicus [69] . Suffi cient force is required
to exceed systolic blood pressure; this can be assessed by absence
of the femoral pulses. Intermittent release of pressure to allow
peripheral perfusion then enables bleeding intra - abdominal
vessels to be identifi ed.
Following vaginal delivery, external aortic compression may be
possible, due to lax abdominal musculature [70] . A study of the
hemodynamic effects of aortic compression on healthy non -
bleeding women within 4 hours of vaginal delivery found that leg
blood pressure was obliterated in 55%, with a substantial reduc-
tion in a further 10%. No signifi cant elevation in systemic blood
pressure was noted, and the authors concluded that this proce-
dure is safe, and a potentially useful maneuver for patient stabi-
lization and transport. However, there have been no studies
addressing the feasibility and effi cacy of external aortic compres-
sion in patients with uterine atony following vaginal delivery; a
high fundus may mean that adequate compression is impossible
in this situation.
Medical t reatment of u terine a tony
The prophylactic use of uterotonic drugs is an effective means
of preventing postpartum hemorrhage from uterine atony.
Either oxytocin alone (5 IU or 10 IU intramuscularly) or with

Syntometrine (5 IU of oxytocin plus 0.5 mg ergometrine: not
available in the USA) may be used. The combination drug is more
effective, but has more side effects [71] . These drugs are also fi rst -
line treatment for postpartum hemorrhage due to atony.
Oxytocin
Oxytocin binds to specifi c uterine receptors and intravenous
administration (dose 5 – 10 IU) has an almost immediate onset of
action [72] . The mean plasma half - life is 3 minutes, so to ensure
a sustained contraction, a continuous intravenous infusion is
necessary. The usual dose is 20 – 40 units per liter of crystalloid,
with the dose rate adjusted according to response. Plateau con-
centration is reached after 30 minutes. Intramuscular injection
has a time of onset of 3 – 7 minutes, and the clinical effect is longer
lasting, at 30 – 60 minutes. Most studies fi nd oxytocin alone
reduces the need for further medication and is associated with
fewer adverse side effects [73] . Compared with other agents, oxy-
tocin has been found to reduce the need for manual placenta
removal in some studies, regardless of route of administration
(intramuscular versus dilute solution), and is safe [73] .
Oxytocin is metabolized by both the liver and kidneys. It has
approximately 5% of the antidiuretic effect of vasopressin, and if
given in large volumes of electrolyte - free solution, can cause
water overload (headache, vomiting, drowsiness, and convul-
sions), symptoms that may be mistakenly attributed to other
Chapter 22
314
gemeprost pessaries, a prostaglandin E - 1 analog, but with actions
resembling PGF - 2 α rather than its parent compound. Both rectal
and intrauterine administration have been reported [80,81] .
Misoprostol

Misoprostol is a synthetic analog of prostaglandin E - 1 and is
metabolized in the liver. The tablet(s) can be given orally, vagi-
nally or rectally. As prophylaxis for postpartum hemorrhage, an
international multicenter randomized trial reported that oral
misoprostol is less successful than parenteral oxytocin adminis-
tration [82] . Misoprostol may, however, be of benefi t in treating
postpartum hemorrhage. In a recent meta - analysis, oral or sub-
lingual misoprostol at a dose of 600 µ g was found to be useful in
postpartum hemorrhage but did not demonstrate a benefi t over
other uterotonics [83] .
Two small case series have reported an apparently rapid
response in postpartum hemorrhage refractory to oxytocin and
syntometrine, with rectal doses of 600 – 1000 µ g. Sustained uterine
contraction was reported in almost all women within 3 minutes
of its administration [84,85] . A single - blinded randomized trial
of misoprostol 800 µ g rectally versus syntometrine intramuscu-
larly plus oxytocin by intravenous infusion found that misopro-
stol resulted in cessation of bleeding within 20 minutes in 30/32
cases (93%) compared to 21/32 (66%) [86] . There was no differ-
ence in need for blood transfusion or onset of coagulopathy. In
a recent meta - analysis comparing the evidence for rectal mispro-
stol, no difference was found with interventions between rectal
misoprostol and placebo or combinations of ergometrine and
oxytocin, although there was a small decrease in blood loss greater
than 500 mL [73,87] . Adverse effects include maternal pyrexia
and shivering. Of note, misoprostol is cheap, heat and light stable,
and does not require sterile needles and syringes for administra-
tion. It may therefore be of particular benefi t in developing
countries.
Surgical m anagement of p ostpartum

h emorrhage
The majority of the surgical techniques described here aim to
arrest hemorrhage due to uterine atony. Many have been utilized
for bleeding resulting from placenta accreta and placenta previa
or for severe genital tract trauma when simple repair is unable to
control hemorrhage.
Surgical intervention for uterine atony is necessary when
uterotonic agents have failed to control bleeding, and there is no
evidence of retained products of conception or concurrent genital
tract trauma. An examination under anesthesia is generally neces-
sary to exclude the latter. An extensive range of surgical tech-
niques has been advocated. Case reports and audit studies
constitute the major clinical evidence. Comparison between pub-
lished reports is diffi cult – factors such as the severity of hemor-
rhage, time lapse from delivery to effective surgery, hemodynamic
and coagulation status, available surgical expertise, and the
of 0.25 mg. It may be given by deep intramuscular injection or by
direct injection into the myometrium, either under direct vision
at cesarean section or transabdominally/transvaginally after
vaginal delivery. It is not licensed for the latter route and there is
concern about direct injection into a uterine sinus, although it is
commonly used in this way [75] . Additionally, it may be more
effi cacious in shock patients, when tissue hypoperfusion may
compromise absorption following intramuscular injection [76] .
A second dose may be given after 90 minutes or, if atony and
hemorrhage continue, repeat doses may be given every 15 – 20
minutes to a maximum of eight doses (2 mg), with ongoing
bimanual compression and fundal massage.
Small case series have reported an effi cacy of 85% or more in
refractory postpartum hemorrhage [76,77] . The largest case series

to date has involved a multicenter surveillance study of 237 cases
of postpartum hemorrhage refractory to oxytocics and found that
it was effective in 88% [78] . The majority of women received a
single dose. When further oxytocics were given to treatment fail-
ures, the overall success rate was 95%. The remaining patients
required surgery and many of these had a cause for postpartum
hemorrhage other than atony, including laceration and retained
products of conception.
F - class prostaglandins cause bronchoconstriction, venocon-
striction and constriction of gastrointestinal smooth muscle.
Associated side effects include nausea, vomiting, diarrhea,
pyrexia, and bronchospasm. There are case reports of hypoten-
sion and intrapulmonary shunting with arterial oxygen desatura-
tion, so it is therefore contraindicated in patients with cardiac or
pulmonary disease. Studies have demonstrated no signifi cant
difference between injectable carboprost compared to ergot
compound injections in rates of postpartum hemorrhage [73] .
Carboprost is expensive and therefore unaffordable in many
developing countries. Dinoprost (prostaglandin F - 2 α ) is more
readily available; intramyometrial injection of 0.5 – 1.0 mg is effec-
tive for uterine atony. In randomized controlled trials comparing
intramuscular prostaglandin F - 2a with ergometrine and combi-
nations of oxytocin and ergometrine, no difference between
interventions in measures of blood loss or need for transfusion
was found. Low - dose intrauterine infusion via a Foley catheter
has also been described, consisting of 20 mg dinoprost in 500 mL
saline at 3 – 4 mL/min for 10 minutes, then 1 mL/min [79] .
Intravenous infusion of dinoprost has not been shown to be
effective.
Prostaglandin E - 2 (dinoprostone) is generally a vasodilatory

prostaglandin; however, it causes contraction of smooth muscle
in the pregnant uterus [72] . Dinoprostone is widely available on
labor wards as an intravaginal pessary for cervical ripening. Rectal
administration (2 mg given 2 hourly) has been successful as a
treatment for uterine atony, vaginal administration probably
being ineffective in the presence of ongoing uterine hemorrhage.
Due to its vasodilatory effect, this drug should be avoided in
hypotensive and hypovolemic patients. However, it may be useful
in women with heart or lung disease in whom carboprost is
contraindicated [49] . Case reports also document the use of
Etiology and Management of Hemorrhage
315
infusion and prophylactic antibiotic cover are advised for these
procedures.
Uterine b race s uture
The B - Lynch suture is a uterine brace suture designed to vertically
compress the uterine body in cases of diffuse bleeding due to
uterine atony [100] . In order to assess whether the suture will be
effective, bimanual compression is applied to the uterus. If bleed-
ing stops, compression with a brace suture should be equally
successful. Single or multiple stitches may be inserted at the same
time and, according to the shape, they may be called brace sutures
[100] , simple brace [101] or square sutures [102] . The patient is
placed in the Lloyd - Davies position on the operating table to
enable assessment of vaginal bleeding. If delivery occurred via
lower segment cesarean section, the incision is reopened. If deliv-
ery was vaginal and retained products have been excluded via
manual exploration, hysterotomy is not necessary. The uterus is
exteriorized and response to bimanual compression assessed. If
vaginal bleeding is controlled, the “ pair of braces ” suture is

inserted using a 70 mm round - bodied needle with number 2
chromic catgut suture (Figure 22.1 ). The two ends are tied while
an assistant performs bimanual compression and the lower
segment incision is closed as normal. The authors describe fi ve
cases in which the procedure was attempted, with success in all
cases. They included hemorrhage due to uterine atony, coagu-
lopathy and placenta previa. The authors state that the advantages
of this method are its surgical simplicity and that adequate hemo-
stasis can be assessed immediately after its completion. Multiple
case reports have described similar success with this procedure
with and without other interventions, including radiologic pro-
cedures or uterotonics [103 – 105] . Normal uterine anatomy has
been demonstrated on follow - up [106] . Resumption of normal
menses along with uncomplicated pregnancies following the
B - Lynch procedure for postpartum hemorrhage has also been
described [107] . Unexpected occlusion of the uterine cavity with
subsequent development of infection (pyometra) has been
reported with the occlusive square - stitch [108] .
A modifi cation of the B - Lynch suture has been described
[109] . A less complex procedure is involved, consisting of two
individual sutures, tied at the fundus. A lower segment incision
is not necessary, and the authors suggest that more tension may
be applied with individual sutures than with one continuous
suture. They also describe tying the loose ends of the sutures
together, to prevent slippage laterally. A summary of published
studies is the subject of a review article [110] .
Uterine d evascularization
Uterine devascularization is a long - practiced technique for post-
partum hemorrhage due to atony, placenta, previa, and trauma.
These techniques can also be used prophylactically in women

with pregnancies complicated with placenta accreta in the operat-
ing room at the time of delivery. Ligation of the uterine arteries
and internal iliac arteries is described; ovarian artery ligation may
also be performed, generally as an adjunctive procedure. Evidence
presence of other obstetric and medical problems all contribute
to differences in outcome.
Uterine t amponade
Uterine packing is a procedure long abandoned by many units
but more recently revived, with case reports detailing new tech-
niques for tamponade of the bleeding placental bed. Historically,
uterine packing was performed using sterile gauze, with up to 5 m
of 5 – 10 cm gauze introduced into the uterus, either using a spe-
cifi c packing instrument or long forceps [88] . Gauze is applied in
layers from side to side, to give maximum pressure on the uterine
wall, with the lower segment packed as tightly as possible.
Indications for uterine packing include atony, placenta previa,
and placenta accreta. Packs are generally left in situ for 24 – 36
hours, and prophylactic antibiotics given.
Uterine packing fell out of use due to concerns about concealed
bleeding, infection, trauma, and problems in performing ade-
quate packing. However, there is little documented evidence to
support these concerns, and it has been suggested that the risks
have been overstated [88,89] . Small studies have demonstrated
that uterine packing is effective for controlling hemorrhage
refractory to other medical treatment [90,91] . In a case series
involving 20 women with postpartum hemorrhage, failure of the
uterine packing to control bleeding was demonstrated in three
women [92] .
The pelvic pressure pack, also known as the “ mushroom, ”
“ umbrella ” or “ Logothetopoulos ” pack, has been successfully

used for control of posthysterectomy hemorrhage in both gyne-
cologic and obstetric patients. Although studies are limited, the
success rate of the pelvic pressure pack in controlling posthyster-
ectomy bleeding in obstetrics has approached 86% after other
therapies were attempted [93] . Several infl atable mechanical
devices have more recently been employed as alternative means
of uterine tamponade. Proponents of these devices state that their
advantages are that they are rapid and easy procedures to perform,
and that their effi cacy can readily be evaluated.
A Sengstaken – Blakemore tube has been utilized in this context
[89,94] . The fi rst report infl ated the gastric balloon with normal
saline, and the second infl ated only the esophageal balloon.
Balloon tamponade has also been performed with a Rusch uro-
logic hydrostatic balloon catheter infl ated with 400 – 500 mL of
saline. This was effective in two women with hemorrhage due to
morbidly adherent placentae [95] . In a more recent study, the
Rusch hydrostatic balloon was effective in controlling postpar-
tum hemorrhage in seven out of eight women when infl ated with
1000 mL of normal saline [96] . Balloon tamponade has also been
accomplished with the use of a sterile condom infl ated with up
to 500 mL of solution tied to a Foley catheter [97] . Several case
reports have demonstrated similar results with a Foley catheter
infl ated with 300 mL [98,99] . These temporizing agents may
allow for correction of coagulopathy in anticipation of surgical
intervention. Often they lead to cessation of hemorrhage alto-
gether and should be attempted in cases where future fertility is
a consideration or in low - resource areas. A continuous oxytocin
Chapter 22
316
increase the risk of arteriovenous sinus formation. If vaginal

delivery has occurred, the bladder may need to be adequately
mobilized prior to suture insertion to avoid ureteric injury.
The largest case series was published in 1995 [112] . This was a
30 - year study involving 265 patients with postcesarean postpar-
tum hemorrhage of > 1000 mL, refractory to oxytocics, methyler-
gonovine and carboprost. Bilateral uterine artery ligation failed
to control hemorrhage in only 10 women, giving a 96% success
rate. An immediate effect was reported, with visible uterine
blanching; myometrial contractions sometimes occurred, but
even if the uterus remained atonic, hemorrhage was usually con-
trolled [111] . No long - term effects on menstrual patterns or
fertility have been reported [112,113] . In women who have
subsequently undergone repeat caesaran section, the uterine
vessels appeared to have recanalized.
Failure of this procedure is most commonly associated with
placenta previa with or without accreta. More recently, low bilat-
eral uterine artery ligation has been described for ongoing bleed-
ing from the lower segment in these cases. A series of 103 patients
involving stepwise uterine devascularization reported a 75%
success rate with conventional uterine artery ligation [113] .
for the effi cacy of these techniques is based on published case
series. The expertise and experience of individual units are
important determinants of the surgical approach to postpartum
hemorrhage.
Bilateral u terine a rtery l igation
The pregnant uterus receives 90% of its blood supply from the
uterine arteries. Bilateral ligation of the ascending branches of the
uterine artery is considered by its practitioners to be a simple,
safe, and effi cacious alternative to hysterectomy [111] . This pro-
cedure was originally utilized to control postpartum hemorrhage

at cesarean section. Mass ligation of the uterine artery branches
and veins is performed 2 – 3 cm below the lower segment incision.
The suture is placed laterally through an avascular window in the
broad ligament, and medially through almost the full thickness
of the uterine wall, to include the uterine vessels and 2 – 3 cm of
myometrium. The vessels are not divided, and inclusion of myo-
metrium avoids vascular damage and obliterates intramyometrial
ascending arterial branches. An absorbable suture such as number
1 chromic catgut on an atraumatic needle is used. Non - absorbable
and fi gure - of - eight sutures are avoided as they are considered to
Round ligament
Round ligament
Fallopian
tube
Ovarian ligament
Fallopian
tube
Broad
ligament
(b) Posterior view
(a) Anterior view
(c) Anterior view
Figure 22.1 The B - Lynch uterine brace suture. (a) Anterior and (b) posterior views are shown during insertion of the suture. (c) After successful insertion. (Reprinted
from B - Lynch et al. The B - Lynch surgical technique for the control of massive postpartum haemorrhage: an alternative to hysterectomy? Five cases reported.
Br J Obstet
Gynaecol
1997; 104: 372 – 375. © 1997 with permission from Elsevier Science.)
Etiology and Management of Hemorrhage
317
A second study involved intra - arterial pressure recordings

before and after ligation [116] . Following bilateral ligation, distal
arterial pulse pressure decreased by 85%, with a 24% reduction
in mean arterial pressure. In addition, a 48% reduction in blood
fl ow resulted following ipsilateral ligation. The authors concluded
that internal iliac ligation controls pelvic hemorrhage mainly by
decreasing arterial pulse pressure. The smaller diameter of the
anastomoses of the collateral circulation was proposed to explain
this phenomenon. The arterial system was considered trans-
formed into a venous - like circulation, with clot formation able
to arrest bleeding at the site of injury. These studies have been
extensively quoted, but similar studies have not been performed
in postpartum women. A single case report found no change in
uterine artery Doppler waveform velocity before and 2 days after
bilateral internal artery ligation performed to control hemor-
rhage due to uterine atony [118] .
Internal iliac artery ligation is a more complex procedure than
uterine artery ligation. The bifurcation of the common iliac artery
is identifi ed at the pelvic brim, and the peritoneum opened and
Success was highest with uterine atony and abruption. Of seven
cases of placenta previa with/without accreta, hemorrhage con-
tinued in four women. A further bilateral ligation was performed
3 – 5 cm below the fi rst sutures, following further mobilization of
the bladder. Ligation therefore includes the ascending branches
of the cervicovaginal artery and the uterine artery branches sup-
plying the lower segment and upper cervix. This procedure was
effective in all cases. A vaginal route for uterine artery ligation has
also been described, with moderate success [114] . This interven-
tion includes incising the anterior cervix near the cervicovaginal
fold with the bladder retracted. The uterus is then gently pulled
to the contralateral side of the intended suture placement. A

single absorbable suture is then placed around the vessels while
including myometrial tissue. Although this technique may be
quick and minimally invasive, more studies are required to prove
its utility in postpartum hemorrhage.
Unilateral or bilateral ligation of the ovarian artery may be
performed as an adjunct to ligation of uterine arteries. The liga-
ture is tied medial to the ovary to preserve ovarian blood supply.
This was the fi nal phase of the stepwise uterine devascularization
approach described above [113] . Following uterine artery liga-
tion, 13/96 cases that did not involve placenta previa/accreta had
ongoing bleeding. Of these, seven responded to unilateral ovarian
artery and six to bilateral ovarian artery ligation. All patients in
this case series therefore avoided hysterectomy.
Bilateral i nternal i liac a rtery l igation
Internal iliac artery ligation was fi rst performed as a gynecologic
procedure by Kelly in 1894 [115] . He termed this “ the boldest
procedure possible for checking bleeding ” and assumed that
the blood supply to the pelvis would be completely arrested.
From the 1950s, internal iliac ligation was increasingly performed
for gynecologic indications, mostly for carcinoma of the
cervix. Ligation was still considered to shut off arterial fl ow,
despite the fact that necrosis of pelvic tissues had not been
observed.
In the 1960s, Burchell reported cutting a uterine artery follow-
ing bilateral internal iliac ligation in order to demonstrate the
absence of fl ow. However, to the surprise of those present, blood
still fl owed freely. This observation led to extensive studies of the
hemodynamic effects of internal iliac ligation. These were per-
formed on gynecologic patients, but are quoted widely in the
obstetric literature [116,117] . Aortograms performed between 5

minutes and 37 months post ligation demonstrated an extensive
collateral circulation, with blood fl ow throughout the internal
iliac artery and its branches. Three collateral circulations were
identifi ed: the lumbar and iliolumbar arteries; the middle sacral
and lateral sacral arteries; and the superior rectal and middle
rectal arteries. Ligation above the posterior division resulted in
collateral and therefore reversed fl ow in its iliolumbar and middle
sacral branches (Figure 22.2 ). Ligation below the posterior divi-
sion caused collateral fl ow only in the middle hemorrhoidal
artery, again in a retrograde direction. Flow to more distal
branches of the internal iliac artery was normal.
Common iliac
Int. iliac
Iliolumbar
Lateral
sacral
Middle
hemorrhoidal
Ext.
iliac
Reversed
flow
Normal flow
Common iliac
Int. iliac
Iliolumbar
Lateral
sacral
Middle
hemorrhoidal

Ext.
iliac
Normal
flow
Reversed flow
(a)
(b)
Figure 22.2 Internal iliac artery ligation. (a) Ligation above the posterior
diversion; collateral pathways result in reversed fl ow in the iliolumbar and lateral
sacral arteries. (b) Ligation below the posterior diversion; collateral pathways
result in reversed fl ow in the middle hemorrhoidal (middle rectal) artery.
(Reprinted from Burchell RC. Arterial physiology of the human pelvis.
Obstet
Gynecol
1968; 31: 855 – 860, with permission from the American College of
Obstetricians and Gynecologists.)
Chapter 22
318
Compared to surgical devascularization, embolization has
several advantages. It is less invasive and generally results in visu-
alization of the bleeding vessel. Occlusion of distal arteries close
to the bleeding site is possible, thereby reducing the risk of
ongoing bleeding from a collateral circulation [120] . The effi cacy
of embolization can immediately be assessed, and repeated embo-
lization of the same or different arteries can be performed.
Disadvantages are the necessity of rapid availability of specialist
equipment and personnel, and the need for transfer of a hemor-
rhaging patient to the radiology suite. Embolization may also be
a time - consuming procedure, generally requiring between 1 and
3 hours, but with hemostasis of the major bleeding vessel fre-

quently established in 30 – 60 minutes. Pelage and colleagues
evaluated the role of selective arterial embolization in 35 patients
with unanticipated postpartum hemorrhage [123] . Bleeding was
controlled in all except one who required hysterectomy for
rebleeding 5 days later. All women in this series who had success-
ful embolization resumed normal menstruation. These fi ndings
have been reported in other studies [124,125] . Patients with life -
threatening hemorrhage have also been successfully treated with
arterial embolization. A 1998 case series included 27 women with
life - threatening bleeding, 12 of whom were intubated and venti-
lated, and four were successfully resuscitated following cardiac
arrest [122] . Fever, contrast media renal toxicity, and leg ischemia
are rare but reported complications of this procedure.
A variation on this theme is the prophylactic placement of
infl atable balloon catheters in internal iliac arteries of patients
who are expected to bleed excessively at the time of surgery, for
example elective cesarean delivery in a patient with placenta per-
creta. In this situation the patient is taken to the interventional
radiology suite prior to surgery and the balloon catheters are
placed but not infl ated. Following delivery of the baby, the cath-
eters can be immediately infl ated. Such catheters can be defl ated
at the completion of surgery and left in situ during the next 24 – 48
hours, to be reinfl ated if required. The use of prophylactic occlu-
sion balloons in the internal iliac arteries before selective embo-
lization has shown a greater than 80% success rate for control of
postpartum hemorrhage [121,124] . Various reports have con-
fi rmed these fi ndings with normal resumption of menses within
3 – 6 months and subsequent uncomplicated pregnancies [126 –
129] . In a recent study comparing outcomes of over 65 women
with placenta accreta who underwent cesarean hysterectomy with

and without prophylactic balloon catheters, no differences were
found in operative time, blood loss, number of hospital days or
transfused products [130] .
Hysterectomy
Peripartum hysterectomy is frequently considered the defi nitive
procedure for obstetric hemorrhage, but is not without complica-
tions. In the long term, the loss of fertility may be devastating to
the patient. In the emergency situation, the major concern is that
peripartum hysterectomy can be a complex procedure, due to
ongoing blood loss and grossly distorted pelvic anatomy due to
edema, hematoma formation, and trauma. Pritchard showed an
refl ected medially along with the ureter [109] . The internal iliac
artery is identifi ed, freed of areolar tissue, and a right - angled
clamp passed under the artery. Two ligatures are tied 1 – 2 cm
apart. The artery is not divided. Both the uterine and vaginal
arteries are branches of the anterior division, and ligation should
if possible be distal to the origin of the posterior division. This is
more effi cacious and does not compromise blood supply to the
buttocks and gluteal muscles. A retroperitoneal approach may
be used when hemorrhage has followed vaginal delivery.
Complications of this procedure include damage to the internal
iliac vein and ureter. Tissue edema, ongoing hemorrhage, and the
presence of a large atonic uterus may make identifi cation of
anatomy diffi cult and prolong operating time. Incorrect identifi -
cation of the internal iliac artery may result in accidental ligation
of the external or common iliac artery, resulting in lower limb
and pelvic ischemia. Femoral pulses should therefore be checked
after the procedure. Recanalization of ligated vessels may occur,
and successful pregnancy has been reported whether or not
recanalization has taken place.

Demonstration of the extensive collateral circulation explains
why the effi cacy of internal iliac ligation is less than for uterine
artery ligation. Success rates are generally reported to be approxi-
mately 40% [109] . A 1985 study reported a success rate of 42%
in a series of 19 patients, with hysterectomy necessary in the
remainder [119] . Morbidity was higher than for a group of
patients in whom hysterectomy was performed as a primary pro-
cedure; mean blood loss was 5125 mL for patients with unsuc-
cessful internal iliac artery ligation followed by hysterectomy, and
3209 mL for those undergoing hysterectomy alone. Complications
associated with unsuccessful arterial ligation in this series were
associated with delay in instituting defi nitive treatment (hyster-
ectomy) rather than as a consequence of arterial ligation. These
authors consider that there is only a limited role for this proce-
dure in the treatment of postpartum hemorrhage, being restricted
to hemodynamically stable patients of low parity in whom future
fertility is of paramount concern.
Arterial e mbolization
Uterine devascularization by selective arterial embolization has
recently gained popularity in centers with expertise in interven-
tional radiology. Access is via the femoral artery and the site of
arterial bleeding is located by injection of contrast into the aorta.
The bleeding vessel is selectively catheterized, and pledgets of
absorbable gelatine sponge injected [120] . These effect only a
temporary blockade and are resorbed within approximately 10
days. If the site of bleeding cannot be identifi ed, embolization of
the anterior branch of the internal iliac artery or the uterine artery
is performed.
In published studies, uterine atony and pelvic trauma are the
major indications for embolization, and overall success rates of

85 – 100% are reported [121] . Higher failure rates are associated
with placenta accreta and procedures performed following failed
bilateral internal iliac artery ligation [122] . Subsequent successful
pregnancies have been documented.