CHAPTER 41: Obstetrical Hemorrhage

As in Williams’ time, obstetrical hemorrhage continues along with hypertension and infection to be one part of the infamous “triad” of maternal death causes. It also is a leading reason for admission of pregnant women to intensive care units (Chantry, 2015; Crozier, 2011; De Greve, 2016; Guntupalli, 2015). Hemorrhage was a direct cause of 11.4 percent of 5367 pregnancy-related maternal deaths from 2006 to 2013 in the United States (Creanga, 2015, 2017). Similarly, 16 percent of 1102 maternal deaths recorded in the Nationwide Inpatient Sample were caused by hemorrhage (Kuriya, 2016). In developing countries, hemorrhage’s contribution is even more striking, and it is the single most important cause of maternal death worldwide (Goffman, 2016; Oladapo, 2016; Thomas, 2016). Despite these numbers, a declining maternal mortality rate from hemorrhage in the United States has been a seminal achievement. But, as discussed in Chapter 1 (Maternal Mortality), it seems unlikely that deaths from hemorrhage have reached an irreducible minimum.

Mechanisms of Normal Hemostasis

A major concept in understanding the pathophysiology and management of obstetrical hemorrhage is the mechanism by which hemostasis is achieved after normal delivery. Recall that near term an incredible amount of blood—at least 600 mL/min—flows through the intervillous space (Pates, 2010). This prodigious flow circulates through the spiral arteries, which average 120 in number. Also, recall that these vessels have no muscular layer because of their remodeling by trophoblasts, which creates a low-pressure system. With placental separation, these vessels at the implantation site are avulsed, and hemostasis is achieved first by myometrial contraction, which compresses this formidable number of large vessels. Compression is followed by clotting and obliteration of vessel lumens.

If, after delivery, the myometrium contracts vigorously, fatal hemorrhage from the placental implantation site is unlikely. Importantly, an intact coagulation system is not necessary for postpartum hemostasis unless there are lacerations in the uterus, birth canal, or perineum. At the same time, however, fatal postpartum hemorrhage can result from uterine atony despite normal coagulation.

Definition and Incidence

Traditionally, postpartum hemorrhage is defined as the loss of ≥500 mL of blood after completion of the third stage of labor. This is problematic because almost half of all women delivered vaginally shed that amount of blood or more when losses are carefully measured (Pritchard, 1962). These results are depicted in Figure 41-1 and show further that approximately 5 percent of women delivering vaginally lose more than 1000 mL of blood. According to the American College of Obstetricians and Gynecologists (2017d), postpartum hemorrhage is defined as cumulative blood loss >1000 mL accompanied by signs and symptoms of hypovolemia. And, almost a third of women undergoing cesarean delivery have blood loss that exceeds 1000 mL. These studies show that estimated blood loss is commonly only approximately half the actual loss. Because of this, estimated blood loss in excess of “average” should alert the obstetrician to possible excessive bleeding. Whether quantification of blood loss improves accuracy is controversial (Hamm, 2017; Toledo, 2007).

The blood volume of a pregnant woman with normal pregnancy-induced hypervolemia usually rises by half, but individual increases range from 30 to 60 percent, that is, 1500 to 2000 mL for an average-sized woman (Pritchard, 1965). The equation to calculate blood volume is shown in Table 41-1. It is axiomatic that a normal pregnant woman tolerates, without any decrease in postpartum hematocrit, blood loss at delivery that approaches the volume of blood that she added during pregnancy. Thus, if blood loss is less than the pregnancy-added volume, the hematocrit remains the same acutely and during the first several days postpartum. It then rises as nonpregnant plasma volume normalizes during the next week or so. Whenever the postpartum hematocrit is lower than one obtained on admission for delivery, blood loss can be estimated as the sum of the calculated pregnancy-added volume plus 500 mL for each 3 volume percent decline of the hematocrit.

Excessive blood loss has been estimated by several methods. Sosa and colleagues (2009) used specially constructed drapes and reported that 10.8 percent of women had hemorrhage in excess of 500 mL with vaginal delivery, whereas 1.9 percent lost >1000 mL. Compared with the findings of Figure 41-1, these estimates likely are too low. Tita and associates (2012) used a 6-volume percent drop in the postpartum hematocrit to define clinically significant blood loss with vaginal delivery. This decline easily signifies a >1000-mL blood loss in the averaged-sized woman. They documented this amount in a fourth of women, which agrees with Figure 41-1.

Another marker used to estimate hemorrhage incidence is the transfusion rate. In the study by Tita just cited, more than 6 percent of women who delivered vaginally underwent blood transfusions. In a study of more than 66,000 women delivered at Parkland Hospital, 2.3 percent overall were given blood transfusions for hypovolemia (Hernandez, 2012). Half of these women had undergone cesarean delivery. Importantly, for those transfused, these investigators calculated blood loss to average approximately 3500 mL! Finally, Green and coworkers (2016) reported that the incidence of massive transfusion for postpartum hemorrhage was 23 per 100,000 births.

From the foregoing, it is apparent that significant blood loss accompanies up to a fourth of vaginal deliveries. The amounts and proportions for cesarean delivery are much greater. And, hemorrhage is underreported. For example, data from the National Hospital Discharge Summary database reported postpartum hemorrhage incidences of only 2.0 and 2.6 percent for two epochs in the United States (Berg, 2009). Similar incidences have been reported by others (Kramer, 2013; Mehrabadi, 2013; Patterson, 2014).

Risks

Numerous clinical circumstances raise the risks for obstetrical hemorrhage. The imposing list shown in Table 41-2 illustrates that hemorrhage can manifest at any time throughout pregnancy, delivery, and the puerperium. Thus, any description of obstetrical hemorrhage should include gestational age. Contributions to maternal death from some of these causes of are shown in Figure 41-2.

Timing

Antepartum Hemorrhage

Obstetrical hemorrhage is traditionally classified as antepartum—such as with placenta previa or placental abruption, or as postpartum—commonly caused by uterine atony or genital tract lacerations. In individual women, however, these terms are nonspecific, and it is reasonable to specify the cause and gestational age as descriptors.

Bleeding during various times in gestation may give a clue to its cause. Many aspects of bleeding during the first half of pregnancy from abortion or ectopic pregnancy are covered in Chapters 18 and 19. Discussions that follow concern pregnancies with a viable-size fetus. In these cases, rapid assessment should always consider the deleterious fetal effects of maternal hemorrhage.

During active labor, slight vaginal bleeding is common. This “bloody show” is the consequence of effacement and dilation of the cervix, with tearing of small vessels. Uterine bleeding above the cervix, however, is concerning. It may follow some separation of a placenta previa implanted in the immediate vicinity of the cervical canal, or it may be from a placental abruption or uterine tear. In some women, especially with a placenta previa, cervical varicosities may bleed (O’Brien, 2013). Rarely, there may be velamentous insertion of the umbilical cord, and the involved placental vessels may overlie the cervix—vasa previa. In this case, serious fetal hemorrhage follows laceration of these vessels at the time of membrane rupture (Swank, 2016).

Near term in many women, the source of uterine bleeding is not identified, bleeding ceases, and no apparent anatomical cause is found at delivery. In most of these cases, bleeding likely originated from a slight marginal placental separation. Despite this, any pregnancy with antepartum bleeding remains at higher risk for an adverse outcome even though bleeding has stopped and placenta previa has been excluded sonographically.

Bleeding after midpregnancy is associated with several adverse outcomes. The Canadian Perinatal Network described 806 women with hemorrhage between 22 and 28 weeks’ gestation (Sabourin, 2012). Placental abruption (32 percent), previa (21 percent), and cervical bleeding (6.6 percent) were the most frequent causes identified. In a third, no cause was found. Of all women, 44 percent were delivered before 29 weeks’ gestation. In more than 68,000 women in Scotland, the incidence of antepartum hemorrhage after the first trimester was 11 percent (Bhandari, 2014). These women were at significantly higher risk for preterm birth, labor induction, and postpartum hemorrhage.

Postpartum Hemorrhage

In most cases, the source of postpartum hemorrhage can and should be determined. Frequent causes are uterine atony with placental site bleeding, genital tract trauma, or both. Postpartum hemorrhage is usually obvious. Important exceptions are unrecognized intrauterine and intravaginal blood accumulation and uterine rupture with intraperitoneal or retroperitoneal bleeding. Another consideration is an expanding vulvar or vaginal hematoma (Puerperal Hematomas). Initial evaluation attempts to differentiate uterine atony from genital tract lacerations. For this, risk factors are sought, the lower genital tract is examined, and uterine tone is assessed. Atony is identified by a boggy, soft uterus during bimanual examination and by expression of clots and hemorrhage during uterine massage.

Persistent bleeding despite a firm, well-contracted uterus suggests that hemorrhage most likely is from lacerations. Bright red blood further suggests arterial bleeding. To confirm that lacerations are a source of bleeding, careful inspection of the vagina, cervix, and uterus is essential. Sometimes bleeding may be caused by both atony and trauma, especially after forceps or vacuum-assisted vaginal delivery. Examination is easier if conduction analgesia was given. If there are no lower genital tract lacerations and the uterus is contracted, yet supracervical bleeding persists, then manual exploration of the uterus is done to exclude a uterine tear (Kaplanoglu, 2016). This also is completed routinely after internal podalic version, breech extraction, or successful vaginal birth after cesarean.

Late postpartum hemorrhage describes bleeding after the first 24 hours. Found in up to 1 percent of women, it may be serious and is discussed in Chapter 37 (American College of Obstetricians and Gynecologists, 2017d).

Blood Loss Estimation

As noted, visual estimates are notoriously inaccurate, especially with excessive bleeding. Instead of sudden massive hemorrhage, postpartum bleeding is frequently steady. If atony persists, bleeding may appear to be only moderate at any given instant but may continue until serious hypovolemia develops. Bleeding from an episiotomy or a vaginal laceration can also appear to be only minimal to moderate. But, constant seepage can lead to enormous blood loss relatively quickly. In some cases, after placental separation, blood may not escape vaginally but instead may collect within the uterine cavity, which can become distended by 1000 mL or more of blood. In others, postpartum uterine massage is applied to a roll of abdominal fat mistaken for the uterus.

All of these factors can lead to an underappreciation of the magnitude of hemorrhage over time. The effects of hemorrhage depend to a considerable degree on the maternal nonpregnant blood volume and the corresponding degree of pregnancy-induced hypervolemia. For this and other reasons, hypovolemia may not be recognized until very late. A treacherous feature of postpartum hemorrhage is the failure of the pulse and blood pressure to undergo more than moderate alterations until large amounts of blood have been lost. The normotensive woman initially may actually become somewhat hypertensive from catecholamine release in response to hemorrhage. And importantly, women with preeclampsia may become “normotensive” despite remarkable hypovolemia.

Some gravidas may be particularly susceptible to hemorrhage because their blood volume expansion is less than expected. This situation is most commonly encountered in small women—even those with normal pregnancy-induced hypervolemia. Women with severe preeclampsia or eclampsia are also more vulnerable to hemorrhage because they frequently do not have a normal blood volume accrual. Specifically, Zeeman and associates (2009) documented a mean increase above nonpregnant volume of only 10 percent in eclamptic women (Chap. 40, Blood Volume). A third example is the moderate-to-severe curtailing of pregnancy-induced volume expansion in women with chronic renal insufficiency (Chap. 53, Chronic Kidney Disease). When excessive hemorrhage is suspected in these high-risk women, crystalloid and blood are promptly administered for suspected hypovolemia.

Third-Stage Labor Management

The most frequent cause of obstetrical hemorrhage is failure of the uterus to contract sufficiently after delivery and to arrest bleeding from vessels at the placental implantation site (General Considerations). That said, some bleeding is inevitable during third-stage labor as the placenta begins to separate. Blood from the implantation site may escape into the vagina immediately—the Duncan mechanism of placental separation, or it remains concealed behind the placenta and membranes until the placenta is delivered—the Schultze mechanism. After signs of placental separation, the uterus should be massaged if it is not contracted firmly, and placental descent is indicated by a slack umbilical cord. Importantly, separation and delivery of the placenta by cord traction, especially when the uterus is atonic, may cause uterine inversion.

If heavy bleeding persists after delivery of the newborn and while the placenta remains partially or totally attached, then manual placental removal is indicated (Cummings, 2016; Frolova, 2016). For this, adequate analgesia is mandatory, and aseptic surgical technique should be used. As illustrated in Figure 41-3, the fingertips of one hand, with fingers approximated, are insinuated between the uterine wall and placenta. A sweeping forward motion in this plane will peel the placenta off its uterine attachment. After its removal, trailing membranes are carefully teased free from the decidua using ring forceps as needed. Another method to clear membranes is to wipe out the uterine cavity with a gauze-wrapped hand. Most recommend ampicillin or cefazolin antimicrobial prophylaxis after manual removal (World Health Organization, 2015).

The fundus is always palpated following placental delivery to confirm that the uterus is well contracted. If it is not firm, then vigorous fundal massage usually prevents postpartum hemorrhage from atony (Hofmeyr, 2013). Simultaneously, 20 units of oxytocin in 1000 mL of crystalloid solution will often be effective given intravenously at 10 mL/min for a dose of 200 mU/min. Higher concentrations are minimally more effective (Tita, 2012). Oxytocin is never given as an undiluted bolus dose because serious hypotension or cardiac arrhythmias can develop.

Risk Factors

In many women with known risks, uterine atony can at least be anticipated well in advance of delivery. In one study, however, up to half of women with atony after cesarean delivery had no risk factors (Rouse, 2006). The magnitude of risk for atony imposed by each of the factors shown in Table 41-2 varies considerably between reports. Primiparity and high parity are risk factors (Driessen, 2011). In one study, the incidence of postpartum hemorrhage rose from 0.3 percent in women of low parity to 1.9 percent with parity of four or greater. It was 2.7 percent with parity of seven or greater (Babinszki, 1999). The overdistended uterus is prone to hypotonia after delivery, and thus women with a large fetus, multiple fetuses, or hydramnios are at greater risk. Labor abnormalities predispose to atony and include hyper- or hypotonic labor. Similarly, labor induction or augmentation with either prostaglandins or oxytocin is more likely to be followed by atony (Driessen, 2011). The frequency of hemorrhage increases with prolongation of the third stage (Frolova, 2016). Finally, the woman who has had a prior postpartum hemorrhage is at risk for recurrence.

Evaluation and Management

With immediate postpartum hemorrhage, careful inspection is done to exclude birth canal laceration. Because bleeding can be caused by retained placental fragments, inspection of the placenta after delivery should be routine. If a defect is seen, the uterus should be manually explored and the fragment removed. Occasionally, retention of a succenturiate lobe may cause postpartum hemorrhage (Chap. 6, Shape and Size Variants). During examination for lacerations and causes of atony, the uterus is massaged and uterotonic agents are administered.

Uterotonic Agents

Several compounds can prompt the postpartum uterus to contract (Chap. 27, Immediate Postpartum Care). One of these is routinely selected and given to prevent postpartum bleeding by ensuring uterine contractions. Most of these same agents are also used to treat uterine atony with bleeding. Moreover, because many trials combine results from atony prophylaxis and treatment, their evaluation is problematic. For example, oxytocin has been used for more than 70 years, and in most cases, it is infused intravenously or given intramuscularly after placental delivery. Neither route has been shown to be superior (Dagdeviren, 2016). This or other uterotonics given prophylactically will prevent most cases of uterine atony.

To treat uterine atony, ergot alkaloids have been used for centuries. If atony persists despite oxytocin and other preventive measures, ergot derivatives can be used for second-line treatment. Ergot preparations include methylergonovine (Methergine) and ergonovine, however, only methylergonovine is currently manufactured in the United States. Given parenterally, these drugs rapidly stimulate tetanic uterine contractions and act for approximately 45 minutes (Schimmer, 2011). A common regimen is 0.2 mg of either drug given intramuscularly. Methergine can be repeated at 2- to 4-hour intervals as needed. A caveat is that ergot agents, especially given intravenously, may cause dangerous hypertension, especially in women with preeclampsia. Severe hypertension is also seen with concomitant use of protease inhibitors given for human immunodeficiency viral (HIV) infection. These adverse effects notwithstanding, it is speculative whether ergot derivatives offer superior therapeutic effects compared with oxytocin.

In cases of atony refractory to one agent, an agent from a different group can be added. At least two randomized studies have addressed combined ergot-oxytocin regimens. In one, ergometrine plus oxytocin was compared with ergometrine alone to prevent postpartum hemorrhage (Koen, 2016). The overall need for transfusion was significantly lower with the combination regimen. Another comparable study reaffirmed these findings (Şentürk, 2016).

During the past 40 years, other second-line agents for atony have included the E- and F-series prostaglandins. Carboprost tromethamine (Hemabate) is the 15-methyl derivative of prostaglandin F_2α. It is approved for uterine atony treatment in a dose of 250 μg (0.25 mg) given intramuscularly. This dose can be repeated if necessary at 15- to 90-minute intervals up to a maximum of eight doses. Observational data indicate an 88-percent success rate (Oleen, 1990). Carboprost causes side effects in approximately 20 percent of women. These include, in descending order of frequency, diarrhea, hypertension, vomiting, fever, flushing, and tachycardia. Another pharmacological effect is pulmonary airway and vascular constriction. Thus, carboprost should not be used for asthmatic women and those with suspected amnionic fluid embolism (General Management). We have occasionally encountered severe hypertension with carboprost given to women with preeclampsia. It has also been reported to cause arterial oxygen desaturation that averaged 10 percent (Hankins, 1988). Relative contraindications to carboprost include renal, liver, and cardiac disease (American College of Obstetricians and Gynecologists, 2017d).

E-series prostaglandins can also prevent or treat atony. Dinoprostone—prostaglandin E₂—may be used off label and is given as a 20-mg suppository per rectum or per vaginum every 2 hours. It typically causes diarrhea, which is problematic for the rectal route, whereas vigorous vaginal bleeding may preclude its use per vaginum. Hypotension, which is commonly encountered with hemorrhage, is considered a contraindication by some. Intravenous prostaglandin E₂—sulprostone—is used in Europe, but it is not available in the United States (Schmitz, 2011).

Misoprostol—Cytotec—is a synthetic prostaglandin E₁ analogue that is used for prevention and treatment of atony (Abdel-Aleem, 2001; Ugwu, 2016). Most studies have addressed prevention and have conflicting conclusions. In a Cochrane review, Mousa and associates (2014) reported no added benefits for misoprostol use compared with oxytocin or ergonovine for treatment. Derman and coworkers (2006) compared a 600-μg oral dose given preventively at delivery against placebo. They found that the drug lowered the incidence of hemorrhage from 12 to 6 percent and that of severe hemorrhage from 1.2 to 0.2 percent. In another study, Gerstenfeld and Wing (2001) concluded that 400 μg misoprostol administered rectally was not superior to intravenous oxytocin given to prevent postpartum hemorrhage. From a systematic review, Villar (2002) found that oxytocin and ergot preparations administered after delivery were more effective than misoprostol for prevention of postpartum hemorrhage (Chap. 27, Immediate Postpartum Care). If misoprostol is used to treat atony, the American College of Obstetricians and Gynecologists (2017d) recommends a dose of 600 to 1000 μg rectally, orally, or sublingually.

Bleeding Unresponsive to Uterotonic Agents

If bleeding persists after initial measures for atony have been implemented, then the following management steps are performed immediately and simultaneously:

1. Begin bimanual uterine compression, which is easily done and controls most cases of continuing hemorrhage (Fig. 41-4). This technique is not simply fundal massage. The posterior uterine wall is massaged by one hand on the abdomen, while the other hand is made into a fist and placed into the vagina. This fist kneads the anterior uterine wall through the anterior vaginal wall and the uterus is also compressed between the two hands.

2. Immediately mobilize the emergent-care obstetrical team to the delivery room and call for whole blood or packed red cells.

3. Request urgent help from the anesthesia team.

4. Secure at least two large-bore intravenous catheters so that crystalloid with oxytocin can be continued simultaneously with blood products. Insert an indwelling Foley catheter for continuous urine output monitoring.

5. Begin volume resuscitation with rapid intravenous infusion of crystalloid (Hypovolemic Shock).

6. With sedation, analgesia, or anesthesia established and now with optimal exposure, once again manually explore the uterine cavity for retained placental fragments and for uterine abnormalities, including lacerations or rupture.

7. Thoroughly inspect the cervix and vagina again for lacerations that may have escaped attention.

8. If the woman is still unstable or if there is persistent hemorrhage, then blood transfusions are given (Hypovolemic Shock).

At this juncture, after causes other than atony have been excluded and after hypovolemia is reversed, several other measures are considered if bleeding continues. Their use depends on several factors such as parity, desire for sterilization, and experience with each method.

Balloon Tamponade

Uterine packing to treat refractory uterine atony fell from favor because of concerns regarding concealed bleeding and infection (Gilstrap, 2017). Newer techniques of balloon tamponade help alleviate some of these concerns (Sentilhes, 2016; Zelop, 2011). In one technique, the tip of a 24F to 30F Foley catheter with a 30-mL balloon is guided into the uterine cavity and filled with 60 to 80 mL of saline. The open tip permits continuous drainage of blood from the uterus. We have experienced balloon rupture when more than 50 mL was instilled into the balloon, thus a 34F Foley with a 60-ml balloon can be used. If bleeding subsides, the catheter is typically removed after 12 to 24 hours. Similar devices for tamponade include Segstaken-Blakemore, Rusch, and ebb balloons and condom catheters (Antony, 2017; Georgiou, 2009).

Enthusiasm has developed for specially constructed intrauterine balloons to treat hemorrhage from uterine atony and other causes. A Bakri Postpartum Balloon or BT-Cath may be inserted and inflated to tamponade the endometrial cavity and stop bleeding (Fig. 41-5). Insertion requires two or three team members. The first performs abdominal sonography during the procedure. The second places the deflated balloon into the uterus and stabilizes it. The third member instills fluid to inflate the balloon, rapidly infusing at least 150 mL followed by further instillation over a few minutes for a total of 300 to 500 mL to arrest hemorrhage. It is reasonable to remove the balloon after approximately 12 hours (Einerson, 2017).

In prospective studies, nearly 150 women have been managed for postpartum hemorrhage with these uterine balloons (Grönvall, 2013; Kaya, 2016; Vintejoux, 2015). Perhaps a fourth of cases were caused by uterine atony. For all causes, the success rate was noted to be approximately 85 percent. Combinations of balloon tamponade and uterine compression sutures have also been described (Diemert, 2012; Yoong, 2012). Failures for all of these require various surgical methods including hysterectomy.

Surgical Procedures

These include uterine compression sutures, pelvic vessel ligation, angiographic embolization, and hysterectomy. These are discussed in Adjunctive Surgical Procedures.

Puerperal inversion of the uterus is one of the classic hemorrhagic disasters encountered in obstetrics. Unless promptly recognized and managed appropriately, associated bleeding often is massive. Risk factors include alone or in combination: (1) fundal placental implantation, (2) uterine atony, (3) cord traction applied before placental separation, and (4) abnormally adhered placentation such as with the accrete syndromes (Morbidly Adherent Placenta).

Depending on which of these factors are contributory, the incidence and severity of uterine inversion varies. There is progressive severity of inversion as shown in Figure 41-6. The worst scenario is complete inversion with the uterus protruding from the birth canal (Fig. 41-7).

The incidence of uterine inversion ranges from 1 in 2000 to 1 in 20,000 vaginal deliveries (Coad, 2017; Ogah, 2011; Rana, 2009; Witteveen, 2013). Our experiences at Parkland Hospital comport with the higher 1:2000 incidence. This is despite our policy of discouraging placental delivery by cord traction alone, and before certainty of its separation. It is unknown if active management of third-stage labor with cord traction applied ostensibly after signs of placental separation raises the likelihood of uterine inversion (Deneux-Tharaux, 2013; Gülmezoglu, 2012; Prick, 2013).

Recognition and Management

Immediate recognition of uterine inversion improves the chances of a quick resolution and good outcome (Furukawa, 2015b). If initially unrecognized, continued hemorrhage likely will prompt closer examination of the birth canal. Although complete inversion is usually evident, the partially inverted uterus can be mistaken for a uterine myoma, and sonography can aid differentiation (Pan, 2015; Smulian, 2013). Many cases are associated with immediate life-threatening hemorrhage, and a fourth require blood replacement (Coad, 2017).

Once any degree of uterine inversion is recognized, several steps must be implemented urgently and simultaneously:

1. Immediate assistance is summoned, including obstetrical and anesthesia personnel.

2. Blood is brought to the delivery suite for potential use.

3. The woman is evaluated for emergency general anesthesia. Large-bore intravenous infusion systems are secured to begin rapid crystalloid infusion to treat hypovolemia while awaiting arrival of blood products.

4. If the recently inverted uterus has not contracted and retracted completely and if the placenta has already separated, then the uterus may often be replaced simply by pushing up on the inverted fundus with the palm of the hand and fingers in the direction of the long axis of the vagina (Fig. 41-8). Some use two fingers rigidly extended to push the center of the fundus upward. Care is taken not to apply so much pressure as to perforate the uterus with the fingertips.

5. If the placenta is still attached, then attempts are made to reposition the uterus with the placenta in situ. Many recommend a trial of an intravenously administered tocolytic drug such as terbutaline, magnesium sulfate, or nitroglycerin for uterine relaxation and repositioning (You, 2006). If these fail to provide sufficient relaxation, then a rapidly acting halogenated inhalational agent is administered. After the uterus is replaced, the placenta is carefully manually removed.

6. If uterine repositioning fails with the placenta attached, then it is peeled off and steady pressure with the fist, palm, or fingers is applied to the inverted fundus in an attempt to push it up into and through the dilated cervix as described in Step 4.

7. Once the uterus is restored to its normal configuration, tocolysis is stopped. Oxytocin is then infused, and other uterotonics may be given as described for atony (Risk Factors). Meanwhile, the operator maintains the fundus in its normal anatomical position while applying bimanual compression to control further hemorrhage until the uterus is well contracted (see Fig. 41-4). The operator continues to monitor the uterus transvaginally for evidence of subsequent inversion. A Bakri balloon has been used to maintain the repositioned uterus (Haeri, 2015; Ida, 2015).

Surgical Intervention

In most cases, the inverted uterus can be restored to its normal position by the techniques just described. Occasionally, manual replacement fails. One cause is a dense myometrial constriction ring. At this point, laparotomy is imperative. The anatomical configuration found at surgery can be confusing as shown in Figure 41-9. With agents given for tocolysis, a combined effort is made to reposition the uterus by simultaneously pushing upward from below and pulling upward from above. Application of atraumatic clamps to each round ligament and upward traction may be helpful—the Huntington procedure. In some cases, placing a deep traction suture in the inverted fundus or grasping it with tissue forceps may be of aid. Either or both of these may be technically difficult. If a constriction ring still prohibits repositioning, a sagittal surgical cut—Haultain incision—is made posteriorly through the muscular ring to release it. The exposed fundus can then be reinverted (Sangwan, 2009). After uterine replacement, tocolytics are stopped, oxytocin and other uterotonics are given, and the uterine incision is repaired. Risks of separation of this posterior hysterotomy incision during subsequent pregnancy, labor, and delivery are unknown. Further illustration and discussion is found in Cunningham and Gilstrap’s Operative Obstetrics, 3rd edition (Zahn, 2017).

In some cases, the uterus will again invert almost immediately after repositioning. With this problem, uterine compression sutures can be used to prevent another inversion (Matsubara, 2009; Mondal, 2012). Occasionally, chronic puerperal uterine inversion may become apparent weeks after delivery.

Childbirth is invariably associated with trauma to the birth canal, which includes the uterus and cervix, vagina, and perineum. Injuries sustained during labor and delivery range from minor mucosal tears to lacerations that create life-threatening hemorrhage or hematomas.

Vulvovaginal Lacerations

According to the American College of Obstetricians and Gynecologists (2016b), up to 80 percent of women sustain some type of laceration at vaginal delivery. These may lie proximally or distally along the lower genital tract.

First, small tears of the anterior vaginal wall near the urethra are relatively common. They are often superficial with little to no bleeding, but they occasionally require sutures for hemostasis. Those large enough to require extensive repair are typically associated with short-term voiding difficulty, and an indwelling bladder catheter will obviate this.

Deeper perineal lacerations are usually accompanied by varying degrees of injury to the outer third of the vaginal vault. Some extend to involve the anal sphincter or varying depths of the vaginal walls. Repair of these perineal lacerations is detailed in Chapter 27 (Laceration and Episiotomy Repairs).

Lacerations involving the middle or upper third of the vaginal vault usually are comorbid with injuries of the perineum or cervix. These sometimes are missed unless inspection is thorough. Those that extend upward usually are longitudinal. They may follow spontaneous delivery but frequently result from injuries sustained during operative vaginal delivery. Most involve deeper underlying tissues and thus usually cause significant hemorrhage, which is controlled by suture repair. For this, effective analgesia or anesthesia, clear visualization, capable assistance, and sufficient resuscitation of hypovolemia are mandatory.

Extensive vaginal or cervical tears should prompt a careful search for evidence of retroperitoneal hemorrhage or of peritoneal perforation with hemorrhage. Also, intrauterine exploration is considered to exclude uterine tears or rupture (Conrad, 2015). If peritoneal perforation or uterine rupture is strongly suspected, laparotomy is considered (Rafi, 2010). As discussed later (Angiographic Embolization), imaging and potential embolization may be suitable for large retroperitoneal hematomas.

Cervical Lacerations

Superficial lacerations of the cervix can be seen on close inspection in more than half of all vaginal deliveries. Most of these measure <0.5 cm and seldom require repair. Deeper lacerations are less frequent, but even these may be unnoticed. Due to ascertainment bias, variable incidences are described. For example, with close inspection, the incidence of cervical lacerations in the Consortium on Safe Labor database was 1 percent in nulliparas and 0.5 percent in multiparas (Landy, 2011). But, the overall incidence in a study of more than 81,000 Israeli women was only 0.16 percent (Melamed, 2009). Such lacerations are more likely to be associated with vacuum- or forceps-assisted vaginal delivery (Fong, 2014).

Cervical lacerations are not usually problematic unless they cause hemorrhage or extend to the vagina. Rarely, the cervix may be entirely or partially avulsed from the vagina in the anterior, posterior, or lateral fornices, an injury termed colporrhexis. Another rare injury is when the entire vaginal portion of the cervix is avulsed—annular or circular detachment. These injuries sometimes follow forceps deliveries performed through an incompletely dilated cervix with the blades applied over the cervix. In some women, cervical tears reach into the lower uterine segment and involve the uterine artery and its major branches. They occasionally extend into the peritoneal cavity. More severe lacerations usually manifest as external hemorrhage or as a hematoma, however, they may occasionally be unsuspected. In the Israeli study just cited, almost 11 percent of women with a cervical laceration required blood transfusions (Melamed, 2009).

At times, the edematous anterior cervical lip is compressed between the fetal head and maternal symphysis pubis. This usually is of little consequence and resolves spontaneously. Rarely, this causes severe ischemia, and the anterior lip may undergo necrosis and subsequently separate from the rest of the cervix.

As with vulvovaginal lacerations, cervical tears can be more fully appreciated with adequate exposure, which may be best attained with transfer to an operating room. An assistant applies firm downward pressure on the uterus, while the operator exerts gentle traction on the lips of the cervix with ring forceps. A second assistant can provide even better exposure with right-angle vaginal wall retractors or Breisky vaginal retractors. Use of suction devices can also aid viewing.

In general, cervical lacerations of 1 and even 2 cm are not repaired unless they are bleeding. Such tears heal rapidly and ultimately create an irregular, sometimes stellate appearing, external cervical os that indicates previous delivery.

Deep cervical tears usually require surgical repair. When the laceration is limited to the cervix or even when it extends somewhat into the vaginal fornix, satisfactory results are obtained by suturing the cervix after bringing it into view as depicted in Figure 41-10. While cervical lacerations are repaired, any associated vaginal lacerations or an episiotomy may be tamponaded with gauze packs to arrest their bleeding. Because hemorrhage usually comes from the upper angle of the wound, the first suture using 2–0 chromic or polyglactin is placed in tissue above the angle. Subsequently, either interrupted or continuous locking sutures are serially placed outward toward the operator. If the uterus is involved and hemorrhage persists, some of the methods described later (Adjunctive Surgical Procedures) may be necessary to obtain hemostasis.

Puerperal Hematomas

Classification and Risks

Pelvic hematomas can have several anatomical manifestations following childbirth. One classification is anatomical and describes vulvar, vulvovaginal, paravaginal, and retroperitoneal hematomas. Vulvar hematomas may involve the vestibular bulb or branches of the pudendal artery, which are the inferior rectal, perineal, and clitoral arteries (Fig. 41-11). Paravaginal hematomas may involve the descending branch of the uterine artery. In some cases, a torn vessel lies above the pelvic fascia, and a supralevator hematoma develops. These can extend into the upper portion of the vaginal canal and may almost occlude its lumen. Continued bleeding may dissect retroperitoneally to form a mass palpable above the inguinal ligament. In some cases, it may even dissect up behind the ascending colon to the hepatic flexure (Rafi, 2010).

Risks for puerperal hematomas include vaginal or perineal laceration, episiotomy, or an operative delivery (Iskender, 2016). Any hematoma can also develop following stretch and rupture of a blood vessel without an associated laceration (Nelson, 2012). This may be especially true with forceps delivery. Occasionally, they are associated with an underlying coagulopathy (Obstetrical Coagulopathies).

Diagnosis

Perineal, vulvar, and paravaginal hematomas can develop rapidly and frequently cause excruciating pain (Fig. 41-12). A tense, tender swelling of varying size rapidly develops, encroaches on the vaginal lumen, and causes overlying skin or epithelium to become ecchymotic. A paravaginal hematoma may escape detection initially. However, symptoms of pelvic pressure, pain, or inability to void should prompt evaluation. Others may go undetected until other measures of hypovolemia become evident. When there is a supralevator extension, the hematoma extends upward in the paravaginal space and between the leaves of the broad ligament. The hematoma may escape detection until it can be felt on abdominal palpation or until hypovolemia develops. Imaging with sonography or computed tomographic scanning may be useful (Cichowski, 2017; Kawamura, 2014; Takeda, 2014).

Clinical Course and Management

Small hematomas often remained contained and show minimal expansion. In others, the tissues overlying an expanding hematoma may rupture from pressure necrosis. In some, profuse hemorrhage may follow, but in other cases, the hematoma drains in the form of large clots and old blood. In those that involve the paravaginal space and extend above the levator sling, retroperitoneal bleeding may be massive and occasionally fatal. Finally, we have encountered a few that rebled up to 2 weeks postpartum (Cunningham, 2017a).

Vulvovaginal hematomas are managed according to their size, location, duration since delivery, and expansion. If bleeding ceases, then small- to moderate-sized hematomas may be treated expectantly until absorbed. But, if pain is severe or if the hematoma continues to enlarge, surgical exploration is preferable. Blood loss with large puerperal hematomas is nearly always considerably more than the clinical estimate. Hypovolemia is common, and transfusions are frequently required when surgical repair is necessary.

For repair, an incision is made at the point of maximal distention, blood and clots are evacuated, and bleeding points ligated. The cavity may then be obliterated with absorbable sutures. Often, no sites of bleeding are identified. Nonetheless, the evacuated hematoma cavity is surgically closed, and the vagina is packed for 12 to 24 hours. Supralevator hematomas are more difficult to treat. Although some can be evacuated by vulvar or vaginal incisions, laparotomy or interventional embolization, described next, is a consideration if bleeding continues.

Angiographic embolization has become popular for management of some puerperal hematomas. This is especially true for supralevator or retroperitoneal hematomas. Embolization can be used primarily, or more likely secondarily, if surgical attempts at hemostasis have failed or if the hematoma is difficult to access surgically (Distefano, 2013; Lee, 2012; Poujade, 2012). The use of a Bakri balloon for a paracervical hematoma has also been described (Gizzo, 2013; Grönvall, 2013). Finally, ultrasound-guided drainage of a recurrent supralevator hematoma has been reported (Mukhopadhyay, 2015).

Uterine Rupture

Predisposing Factors

Uterine rupture frequently is catastrophic. It may be primary, defined as occurring in a previously intact or unscarred uterus, or may be secondary and associated with a preexisting incision, injury, or anomaly of the myometrium. Some of the etiologies associated with uterine rupture are presented in Table 41-3. Importantly, the contribution of each of these underlying causes has changed remarkably during the past 50 years. Specifically, before 1960, when the cesarean delivery rate was much lower and women of great parity were numerous, primary uterine rupture predominated. As the incidence of cesarean delivery rose and especially as a subsequent trial of labor in these women became prevalent through the 1990s, uterine rupture through the cesarean hysterotomy scar became the preeminent cause (Gibbins, 2015; Mone, 2016). However, concurrent with the diminished enthusiasm for a trial of labor in women with a prior cesarean delivery, incidence trends for the two types of rupture have again changed. In a study of 3942 cases of uterine rupture in more than 15 million women, approximately half were in women with a prior cesarean delivery (Yao, 2017). In 40 cases of rupture at Parkland Hospital from 2009 to 2016, 15 events (37 percent) were primary, and 25 (63 percent) were secondary (Happe, 2017).

Additional risks for rupture include other previous operations or manipulations that traumatize the myometrium. Examples are uterine curettage or perforation, endometrial ablation, myomectomy, or operative hysteroscopy (Kieser, 2002; Pelosi, 1997). In a study by Porreco and colleagues (2009), seven of 21 women without a prior cesarean delivery had undergone prior uterine surgery.

In developed countries, the incidence of rupture is 1 in 4800 deliveries (Getahun, 2012). During a 40-year period in Norway, the uterine rupture rate rose significantly to about 1 in 1560 deliveries (Al-Zirqi, 2016). The frequency of primary rupture, however, approximates 1 in 10,000 to 15,000 births (Porreco, 2009). As discussed, one reason is a decreased incidence of women of great parity. Another is that excessive or inappropriate uterine stimulation with oxytocin—previously a frequent cause—has mostly disappeared. Maggio and associates (2014) found no association between the number of Montevideo units and secondary uterine rupture. In addition, in a recent analysis of three trials comparing high- versus low-dose oxytocin regimens, the rate of uterine rupture did not differ between groups (Budden, 2014). The rate of rupture is elevated with sequential induction of labor with prostaglandins and oxytocin (Al-Zirqi, 2017). At Parkland Hospital, we too have encountered primary uterine rupture in a disparate number of women in whom labor was induced with prostaglandin E₁.

Blunt abdominal trauma can precipitate uterine rupture. Although the distended pregnant uterus is surprisingly resistant, pregnant women sustaining such trauma should be watched carefully for signs of a ruptured uterus (Chap. 47, Other Blunt Trauma). In one study of 13 cases of primary uterine rupture, trauma accounted for three cases (Miller, 1996). Other causes of traumatic rupture that are uncommon today are those due to internal podalic version and extraction, difficult forceps delivery, breech extraction, and unusual fetal enlargement such as with hydrocephaly.

Uncommon associations of rupture are uterine anomalies or multifetal pregnancy (Bankada, 2015; Tarney, 2013; Tola, 2014). Occasionally, focal inherent weakness in the myometrium predisposes to rupture. Examples include anatomical anomalies, leiomyomas, adenomyosis, choriocarcinoma, and connective-tissue defects such as Ehlers-Danlos syndrome (Arici, 2013; Nikolaou, 2013; Noh, 2013; Ramskill, 2014; Sun, 2016).

Pathogenesis

Rupture of the previously intact uterus during labor most often involves the thinned-out lower uterine segment. When the rent is in the immediate vicinity of the cervix, it frequently extends transversely or obliquely. When the rent forms in the portion of the uterus adjacent to the broad ligament, the tear is usually longitudinal. Although these tears develop primarily in the lower uterine segment, they can extend upward into the active segment or downward through the cervix and into the vagina (Fig. 41-13). In some cases, the bladder may also be lacerated. If the rupture is of sufficient size, the uterine contents will usually escape into the peritoneal cavity. If the presenting fetal part is firmly engaged, however, then only a portion of the fetus may be extruded from the uterus. Fetal prognosis is largely dependent on the degree of placental separation and magnitude of maternal hemorrhage and hypovolemia. In some cases, the overlying peritoneum remains intact, and this usually is accompanied by hemorrhage that extends into the broad ligament to cause a large retroperitoneal hematoma.

Following vaginal delivery in an unscarred uterus, we and others have occasionally encountered cases of an incomplete tear on the inside of the uterus that extends vertically into the active segment and is a source of profuse hemorrhage (Conrad, 2015). These tears are usually not visible from below but are found at the time of hysterectomy for intractable bleeding despite a contracted uterus. Hemorrhage with this type of tear can be torrential, and bleeding is usually not slowed until the uterine artery pedicles are clamped bilaterally.

Management and Outcomes

The varied clinical presentations of uterine rupture and its management are discussed in detail in Chapter 31 (Uterine Scar Rupture). In the most recent maternal mortality statistics from the Centers for Disease Control and Prevention, uterine rupture accounted for almost 10 percent of deaths caused by hemorrhage (Creanga, 2015, 2017). Maternal morbidity includes hysterectomy that may be necessary to control hemorrhage. Rates of perinatal mortality and morbidity, which may include severe neurological impairment, are also high (Gibbins, 2015; Porreco, 2009). Maternal obesity comorbid with uterine rupture is associated with increased rates of adverse neonatal outcomes (Yao, 2017).

Etiopathogenesis

Separation of the placenta—either partially or totally—from its implantation site before delivery is described by the Latin term abruptio placentae. Literally translated, this refers to “rending asunder of the placenta,” which denotes a sudden accident that is a clinical characteristic of most cases. In the purest sense, the cumbersome—and thus seldom used—term premature separation of the normally implanted placenta is most descriptive because it excludes separation of a placenta previa.

Placental abruption is initiated by hemorrhage into the decidua basalis. The decidua then splits, leaving a thin layer adhered to the myometrium. Consequently, the process begins as a decidual hematoma and expands to cause separation and compression of the adjacent placenta. Inciting causes of many cases have been posited. The phenomenon of impaired trophoblastic invasion with subsequent atherosis is related in some cases of preeclampsia complicated by abruption (Brosens, 2011). Inflammation or infection may be contributory (Mhatre, 2016; Nath, 2007). Histological findings cannot be used to determine timing of the abruption (Chen, 2017).

Abruption likely begins with rupture of a decidual spiral artery and then an expanding retroplacental hematoma. In the early stages of placental abruption, clinical symptoms may be absent. Even with continued bleeding and placental separation, placental abruption can still be either total or partial (Fig. 41-14). With either, bleeding typically insinuates itself between the membranes and uterus, ultimately escaping through the cervix to cause external hemorrhage. Less often, the blood is retained between the detached placenta and the uterus, leading to concealed hemorrhage and delayed diagnosis. The delay translates into greater maternal and fetal hazards. Also with concealed hemorrhage, the likelihood of consumptive coagulopathy is elevated. This is because increased pressure within the intervillous space, caused by the expanding retroplacental clot, forces more placental thromboplastin into the maternal circulation (Diagnosis).

Most blood in the retroplacental hematoma in a nontraumatic placental abruption is maternal. This is because hemorrhage derives from separation within the maternal decidua, and placental villi are usually initially intact. In 78 women at Parkland Hospital with a nontraumatic placental abruption, fetal-to-maternal hemorrhage was documented in only 20 percent—and all of these had <10 mL fetal blood loss (Stettler, 1992). Atkinson and colleagues (2015) identified fetal cells in peripheral blood in only 4 percent of 68 women with a placental abruption.

When clinically suspected, an abruption is seen on a freshly delivered placenta as a circumscribed depression on the maternal surface. These usually measure a few centimeters in diameter and are covered by dark, clotted blood. Because several minutes are required for these anatomical changes to materialize, a very recently separated placenta may appear totally normal at delivery. Our experiences are like those of Benirschke and associates (2012) in that the “age” of the retroplacental clot cannot be determined exactly. In the example shown in Figure 41-15, a large dark clot is well formed, it has depressed the placental bulk, and it likely is at least several hours old.

Defining severity of placental abruption is problematic. We have considered abruption severe when the fetus dies, however, maternal and fetal complications can be serious even with a liveborn fetus. Ananth and coworkers (2016) have defined severe abruption as displaying one or more of the following: (1) maternal sequelae that include disseminated intravascular coagulation, shock, transfusion, hysterectomy, renal failure, or death; (2) fetal complications such as nonreassuring fetal status, growth restriction, or death; or (3) neonatal outcomes that include death, preterm delivery, or growth restriction.

Traumatic Abruption

External trauma—usually from motor vehicle accidents or aggravated assault—can cause placental separation. The frequency of abruption originating from trauma varies. Kettel (1988) and Stafford (1988) and their associates have appropriately stressed that abruption can stem from relatively minor trauma. The clinical presentation and consequences of these abruptions differ somewhat from spontaneous cases. For example, associated fetomaternal hemorrhage, while seldom clinically significant with most spontaneous abruptions, is more common with trauma because of concomitant placental tears or “fractures” (Chap. 47, Placental Injuries). Fetal bleeding that averaged 12 mL was noted in a third of women with a traumatic abruption reported by Pearlman (1990). In eight women cared for at Parkland Hospital, we found fetal-to-maternal hemorrhage of 80 to 100 mL in three of eight cases of traumatic placental abruption (Stettler, 1992). Importantly, in some cases of trauma, a nonreassuring fetal heart rate tracing may not be accompanied by other evidence of placental separation. A sinusoidal tracing is one example. Traumatic abruption is considered in more detail in Chapter 47 (Placental Injuries).

Chronic Abruption

Some cases of chronic placental separation begin early in pregnancy. Dugoff and coworkers (2004) observed an association between some abnormally elevated maternal serum aneuploidy markers and subsequent abruption. Other have correlated first- and second-trimester bleeding with third-trimester placental abruption (Ananth, 2006; Weiss, 2004). In some cases of a chronic abruption, subsequent oligohydramnios develops—chronic abruption-oligohydramnios sequence—CAOS (Elliott, 1998). Even later in pregnancy, hemorrhage with retroplacental hematoma formation is occasionally arrested completely without delivery. These women may have abnormally elevated serum levels of alpha-fetoprotein or placenta-specific RNAs as markers of the event (Miura, 2016; Ngai, 2012).

Frequency

The reported incidence of placental abruption varies because of different criteria used for diagnosis. That said, its frequency averages 0.5 percent or 1 in 200 deliveries. From one database of almost 28 million births from 2006 through 2012, the incidence of placental abruption was nearly 1 percent (Ananth, 2016). From a cohort of more than 1.57 million births in the Netherlands, Ruiter and coworkers (2015) found the frequency was 0.22 percent—1 in 450. In more than 250,000 deliveries at Parkland Hospital from 2000 through 2015, the incidence of placental abruption averaged 0.35 percent or 1 in 290 (Fig. 41-16).

The frequency of placental abruption has risen in this country, and most of this increase is in black women (Ananth, 2005, 2016). At Parkland Hospital, however, the frequency of severe abruption has declined. This discrepancy may be explained in part by the variations in management of early-onset preeclampsia (Chap. 40, Preeclampsia). Specifically, with placental abruption so extensive as to kill the fetus, the incidence was 0.24 percent or 1 in 420 births from 1956 through 1967 (Pritchard, 1967). As the number of high-parity women giving birth declined along with improved availability of prenatal care and emergency transportation, the frequency of abruption causing fetal death dropped to 0.12 percent through 1989 in our obstetrical population. And, most recently through 2015, it declined to 0.05 percent or 1 in 2060.

Perinatal Morbidity and Mortality

Overall, perinatal outcomes are influenced by gestational age, and the frequency of placental abruption rises across the third trimester. As seen in Figure 41-17, more than half of the placental abruptions at Parkland Hospital developed at gestational ages ≥37 weeks. Perinatal mortality and morbidity, however, are more common with earlier abruptions (Furukawa, 2015a). Of other related factors, major fetal congenital anomalies have greater association with placental abruption (Riihimäki, 2013).

Although the rates of fetal death have declined, the contribution of abruption as a cause of stillbirth remains prominent because other causes have also decreased. For example, since the early 1990s, 10 to 12 percent of all third-trimester stillbirths at Parkland Hospital have been the consequence of placental abruption. Others have documented high perinatal mortality rates caused by placental abruption. Salihu and colleagues (2005) analyzed more than 15 million singleton births between 1995 and 1998. The perinatal mortality rate associated with placental abruption was 119 per 1000 births compared with 8 per 1000 for the general obstetrical population.

Neonatal deaths are common following placental abruption. At Parkland Hospital, 15 percent of liveborn neonates died. Perinatal morbidity—often severe—is also common in surviving newborns (Abdella, 1984). Studies by Matsuda and coworkers (2003, 2013) reported that 20 percent of survivors developed cerebral palsy. These observations are similar to ours from Parkland Hospital. Notably, 20 percent of liveborn neonates of women with an abruption had severe acidemia, defined by a cord arterial blood pH <7.0 or base deficit of ≥12 mmol/L. One review confirmed the associated risk for cerebral palsy (Downes, 2017). Even so, Ananth and coworkers (2017) attribute adverse neurodevelopmental outcomes to be largely attributable to preterm delivery.

Predisposing Factors

Demographic Factors

Several predisposing factors raise the risk for placental abruption, and some are listed in Table 41-4. Advancing maternal age is one, although data are conflicting regarding women of great parity (Okby, 2017; Pritchard, 1991). Race or ethnicity also appears to be important. In almost 366,000 deliveries at Parkland Hospital, abruption severe enough to kill the fetus was most common in black and white women—1 in 200, less so in Asian women—1 in 300, and least common in Latin-American women—1 in 350 (Pritchard, 1991). A familial association was found in an analysis of a Norwegian population-based registry (Rasmussen, 2009). If a woman had a severe abruption, the risk for her sister was doubled.

Pregnancy-Associated Hypertension

Some form of hypertension is the most frequent condition associated with placental abruption. This includes gestational hypertension, preeclampsia, chronic hypertension, or a combination thereof. In a report by Pritchard and colleagues (1991) that described 408 women with placental abruption and fetal demise, hypertension was apparent in half once hypovolemia was corrected. Half of these latter women—a fourth of all 408—had chronic hypertension. Looked at another way, one Maternal–Fetal Medicine Units (MFMU) Network study found that 1.5 percent of pregnant women with chronic hypertension suffered placental abruption (Sibai, 1998). As discussed in Chapter 50 (Adverse Pregnancy Effects), at Parkland Hospital, the frequency of placental abruption in treated chronically hypertensive women was almost 1 percent, which was threefold higher than the 0.3-percent baseline (Morgan, 2016).

Chronic hypertension with superimposed preeclampsia or with fetal-growth restriction confers an even greater risk (Ananth, 2007). Even so, the severity of hypertension does not necessarily correlate with abruption incidence (Morgan, 2016; Zetterstrom, 2005). The long-term effects of these associations are apparent from the significantly elevated cardiovascular mortality risk in women with prior abruption, with or without chronic hypertension (DeRoo, 2016; Pariente, 2013). Observations from the Magpie Trial Collaborative Group suggest that women with preeclampsia, with or without chronic hypertension, given magnesium sulfate may have a reduced risk for abruption (Altman, 2002).

Preterm Prematurely Ruptured Membranes

The abruption risk substantially rises when placental membranes rupture before term (American College of Obstetricians and Gynecologists, 2016a; Hackney, 2016). Major and colleagues (1995) reported that 5 percent of 756 women with ruptured membranes between 20 and 36 weeks’ gestation developed an abruption. It was 17 percent with previable prematurely ruptured membranes (Kibel, 2016). The risk for abruption with preterm rupture is further increased with comorbid infection (Ananth, 2004). In these cases, inflammation and infection as well as preterm delivery may be primary causes leading to abruption (Nath, 2007, 2008).

Somewhat related, Aviram and coworkers (2015) found an eightfold higher abruption risk in pregnancies ≥34 weeks if hydramnios was comorbid. Abrupt uterine decompression during membrane rupture may be an inciting factor.

Prior Abruption

Many of the predisposing factors are chronic, and in these cases, placental abruption has a high recurrence rate. Pritchard and associates (1970) identified a recurrence rate of 12 percent—and half of these caused another fetal death. Furuhashi and colleagues (2002) reported a 22-percent recurrence rate—half recurred at a gestational age 1 to 3 weeks earlier than the first abruption. In the Dutch study mentioned previously, Ruiter and coworkers (2015) cited a recurrence risk of 5.8 percent. Looked at a second way, Tikkanen and associates (2006) found that of 114 parous women who experienced an abruption, 9 percent had a prior abruption. A third perspective is provided by a population-based study of 767,000 pregnancies reported by Rasmussen and Irgens (2009). They found a 6.5-fold higher risk for recurrence of a “mild” abruption and 11.5-fold risk for a “severe” abruption. For women who had two severe abruptions, the risk for a third was increased 50-fold.

Management of a pregnancy subsequent to an abruption is difficult because another separation may suddenly occur, even remote from term. In many of these recurrences, fetal well-being is almost always reassuring beforehand. Thus, antepartum fetal testing is usually not predictive. Because term abruptions tend to be recurrent, Ruiter and coworkers (2015) recommend labor induction at 37 weeks. Our practice at Parkland Hospital is to induce labor at 38 weeks if other complications do not develop beforehand.

Other Associations

Cigarette smoking is linked to an elevated risk for abruption (Misra, 1999; Naeye, 1980). Results of a metaanalysis of 1.6 million pregnancies included a twofold risk for abruption in smokers (Ananth, 1999b). This risk was five- to eightfold if smokers had chronic hypertension, severe preeclampsia, or both. Similar findings are reported by others (Hogberg, 2007; Kaminsky, 2007). Antepartum Vitamin C and E were reported to be protective for abruption in smokers (Abramovici, 2015).

Cocaine abuse is linked with an alarming frequency of placental abruption (Addis, 2001; Cressman, 2014). Bingol and colleagues (1987) described 50 women who abused cocaine during pregnancy—eight had a stillbirth caused by placental abruption.

Uterine leiomyomas, especially if located near the mucosal surface behind the placental implantation site, can predispose to placental abruption. This was reviewed recently by Ezzedine and Norwitz (2016).

Isolated single umbilical artery is associated with a 3.4-fold increased risk for placental abruption (Gutvirtz, 2016). Twins resulting from infertility treatments also carry greater risk (Okby, 2017). Subclinical hypothyroidism or high levels of antithyroid antibodies have been associated with a two- to threefold higher risk for abruption (Abbassi-Ghanavati, 2010; Casey, 2014; Maraka, 2016).

Women affected by some of the thrombophilias have higher associated rates of thromboembolic disorders during pregnancy. However, the link with placental abruption is less clear (American College of Obstetricians and Gynecologists, 2017a,b). Lupus anticoagulant is associated with maternal floor infarction of the placenta but is less so with typical abruptions. No convincing evidence supports a role for thrombophilias and placental abruption.

Clinical Findings and Diagnosis

Most women with a placental abruption have sudden-onset abdominal pain, vaginal bleeding, and uterine tenderness. In a prospective study, Hurd and colleagues (1983) reported that 78 percent with placental abruption had vaginal bleeding, 66 percent had uterine tenderness or back pain, and 60 percent had a nonreassuring fetal status. Other findings included frequent contractions and persistent hypertonus. In a fifth of these women, preterm labor was diagnosed, and abruption was not suspected until fetal distress or death followed.

Importantly, the signs and symptoms of placental abruption can vary considerably. In some women, external bleeding can be profuse, yet placental separation may not be so extensive as to compromise the fetus. In others, there may be no external bleeding, but the placenta is sufficiently sheared off that the fetus is dead—a concealed abruption. In one unusual case, a multiparous woman cared for at Parkland Hospital presented with a nosebleed. She had no abdominal or uterine pain, tenderness, or vaginal bleeding. Her fetus was dead, however, and her blood did not clot. The plasma fibrinogen level was 25 mg/dL. Labor was induced, and a total abruption was confirmed at delivery.

Differential Diagnosis

With severe placental abruption, the diagnosis generally is obvious. From the previous discussion, it follows that less severe, more common forms of abruption cannot always be recognized with certainty. Thus, the diagnosis is one of exclusion. Unfortunately, no laboratory tests or other diagnostic methods accurately confirm lesser degrees of placental separation. Sonography has limited use because the placenta and fresh clots may have similar imaging characteristics. Glantz and Purnell (2002) reported only 24-percent sensitivity for sonography in 149 consecutive women with a suspected placental abruption. Importantly, negative findings with sonographic examination do not exclude placental abruption. Conversely, magnetic resonance (MR) imaging is highly sensitive for placental abruption and should be considered if the diagnostic information would change management (Masselli, 2011).

With abruption, some degree of intravascular coagulation is almost universal. Thus, elevated serum levels of d-dimers may be suggestive, but this has not been adequately tested. Preliminary data show that serum alpha-fetoprotein levels >280 μg/L have a positive-predictive value of 97 percent (Ngai, 2012).

Thus, in the woman with vaginal bleeding and a live fetus, it is often necessary to exclude placenta previa and other causes of bleeding by clinical and sonographic evaluation. It has long been taught—perhaps with some justification—that painful uterine bleeding signifies placental abruption, whereas painless uterine bleeding is indicative of placenta previa. The differential diagnosis is usually not this straightforward, and labor accompanying previa may cause pain suggestive of placental abruption. On the other hand, pain from abruption may mimic normal labor, or it may be painless, especially with a posterior placenta. At times, the cause of the vaginal bleeding remains obscure even after delivery.

Hypovolemic Shock

Placental abruption is one of several notable obstetrical entities that may be complicated by massive and sometimes torrential hemorrhage. Hypovolemic shock is caused by maternal blood loss. In an earlier report from Parkland Hospital, Pritchard and Brekken (1967) described 141 women with abruption so severe as to kill the fetus. Blood loss in these women often amounted to at least half of their pregnant blood volume. Importantly, massive blood loss and shock can develop with a concealed abruption. Prompt treatment of hypotension with crystalloid and blood infusion is essential, and resuscitation steps are described later (Hypovolemic Shock).

Consumptive Coagulopathy

Obstetrical events—mainly placental abruption and amnionic fluid embolism—led to the initial recognition of defibrination syndrome. This syndrome is currently referred to as consumptive coagulopathy or disseminated intravascular coagulation, which later is described more broadly in Obstetrical Coagulopathies. The major mechanism causing procoagulant consumption is intravascular activation of clotting. Abruption is the most common cause of clinically profound consumptive coagulopathy in obstetrics—and indeed, probably in all of medicine (Cunningham, 2015).

An important consequence of intravascular coagulation is the activation of plasminogen to plasmin, which lyses fibrin microemboli to maintain microcirculatory patency. With placental abruption severe enough to kill the fetus, there are always pathological levels of fibrinogen–fibrin degradation products and d-dimers in maternal serum (Erez, 2015). Their quantification is not clinically useful. In a third of women with an abruption severe enough to kill the fetus, the plasma fibrinogen level will be <150 mg/dL. These levels are dependent on the maternal preabruption fibrinogen level, and thus higher levels are “protective” (Cunningham, 2015; Wang, 2016). Clinically significant low levels may cause troublesome surgical bleeding. Levels of several other coagulation factors are also variably decreased. In addition, thrombocytopenia, sometimes profound, may accompany severe hypofibrinogenemia initially and becomes common after repeated blood transfusions.

Consumptive coagulopathy is more likely with a concealed abruption because intrauterine pressure is higher. This forces more thromboplastin into the large veins draining the implantation site. With a partial abruption and a live fetus, severe coagulation defects are less common. Our experience has been that if serious coagulopathy develops, it is usually evident by the time abruption symptoms appear.

Couvelaire Uterus

At the time of cesarean delivery, it is not uncommon to find widespread extravasation of blood into the uterine musculature and beneath the serosa (Fig. 41-18). It is named after Couvelaire, who in the early 1900s termed it uteroplacental apoplexy. These myometrial hemorrhages seldom cause uterine atony, and alone they are not an indication for hysterectomy. Effusions of blood are also seen beneath the tubal serosa, between the leaves of the broad ligaments, in the substance of the ovaries, and free in the peritoneal cavity.

End-Organ Injury

Acute kidney injury (AKI) is a general term describing renal dysfunction from many causes (Chap. 53, Acute Kidney Injury). Delayed or incomplete treatment of hypovolemia with severe placental abruption can be one. However, even with abruption complicated by severe disseminated intravascular coagulation, prompt and vigorous treatment of hemorrhage with blood and crystalloid solution usually prevents clinically significant renal dysfunction. The risk for renal injury with abruption is magnified when preeclampsia coexists (Alexander, 2015; Drakeley, 2002). Most cases of AKI are reversible and not so severe as to require dialysis. Generally, long-term outcomes are good (Arazi, 2015). That said, irreversible acute cortical necrosis encountered in pregnancy can be associated with abruption (Gopalakrishnan, 2015).

Rarely, pituitary failure—Sheehan syndrome—follows severe intrapartum or early postpartum hemorrhage. Described in Chapter 58 (Acromegaly), the exact pathogenesis is not well understood, especially because endocrine abnormalities are infrequent even in women who suffer catastrophic hemorrhage (Matsuwaki, 2014; Robalo, 2012).

Management

Treatment of the woman with a placental abruption varies depending on her clinical condition, gestational age, and the amount of associated hemorrhage. With a living viable-aged fetus, and with vaginal delivery not imminent, emergency cesarean delivery is chosen by most. In some women, fetal compromise will be evident as shown in Figure 41-19. When evaluating fetal status, sonographic confirmation of fetal heart activity may be necessary because sometimes an electrode applied directly to a dead fetus will provide misleading information by recording the maternal heart rate. If the fetus has died or if it is not considered sufficiently mature to live outside the uterus, then vaginal delivery is preferable. In either case, prompt and intensive resuscitation with blood plus crystalloid is begun to replace blood lost from retroplacental and external hemorrhage. These measures are lifesaving for the mother and hopefully for her fetus. If the diagnosis of abruption is uncertain and the fetus is alive and without evidence of compromise, then close observation may be warranted provided that immediate intervention is available. Colón and coworkers (2016) performed a randomized trial and found no benefits to magnesium sulfate tocolysis given to women with a preterm “nonsevere” abruption at 24 to 34 weeks’ gestation.

Cesarean Delivery

The compromised fetus is usually best served by cesarean delivery, and the speed of response is an important factor in perinatal outcomes. Kayani and coworkers (2003) studied this relationship in 33 singleton pregnancies with a clinically overt placental abruption and fetal bradycardia. Of the 22 neurologically intact survivors, 15 were delivered within a 20-minute decision-to-delivery interval. However, eight of 11 infants who died or developed cerebral palsy were delivered with intervals >20 minutes.

A major hazard to cesarean delivery is imposed by clinically significant consumptive coagulopathy. Preparations include plans for blood and component replacement and assessment of coagulation—especially fibrinogen levels.

Vaginal Delivery

If the fetus has died, then vaginal delivery is usually preferred. As reviewed earlier, hemostasis at the placental implantation site depends primarily on myometrial contraction and not blood coagulability. Thus, after vaginal delivery, uterotonic agents and uterine massage are used to stimulate myometrial contractions. Uterine muscle fibers compress placental site vessels and prompt hemostasis even if coagulation is defective.

In some instances, vaginal delivery may not be preferable, even with a dead fetus. One example is brisk hemorrhage that cannot be successfully managed by vigorous blood replacement. Others are the myriad obstetrical complications that prohibit vaginal delivery in general. These are listed in Table 30-1.

In some women with extensive placental abruption, labor tends to be rapid because the uterus is usually persistently hypertonic. This can magnify fetal compromise. In some cases, baseline intraamnionic pressures reach 50 mm Hg or higher, and with contractions, pressures may attain levels exceeding 100 mm Hg. Overall, however, first-and second-stage labor do not appear to be shortened (Downes, 2016).

Early amniotomy has long been championed in the management of placental abruption. This ostensibly achieves better spiral artery compression to diminish implantation site bleeding and reduce thromboplastin infusion into the maternal vascular system. Although evidence supporting this theory is lacking, membrane rupture may hasten delivery. However, if the fetus is small, the intact sac may be more efficient in promoting cervical dilation. If rhythmic uterine contractions are not superimposed on baseline hypertonus, then oxytocin is given in standard doses. No data indicate that oxytocin augments thromboplastin escape into the maternal circulation to worsen coagulopathy (Clark, 1995; Pritchard, 1967). In light of hypertonus associated with abruption, misoprostol may be a less favored induction agent due to its association with uterine tachysystole.

In the past, some had set arbitrary time limits to permit vaginal delivery. Instead, experiences illustrate that maternal outcome depends on the diligence with which adequate fluid and blood replacement therapy are pursued rather than on the interval to delivery. Observations from Parkland Hospital described by Pritchard and Brekken (1967) are similar to those from the University of Virginia reported by Brame and associates (1968). Specifically, women with severe abruption who were transfused during 18 hours or more before delivery had similar outcomes to those in whom delivery was accomplished sooner.

Expectant Management with a Preterm Fetus

If possible, delaying delivery may benefit an immature fetus. Bond and colleagues (1989) expectantly managed 43 women with placental abruption before 35 weeks’ gestation, and 31 of them were given tocolytic therapy. The mean interval-to-delivery for all 43 was approximately 12 days. Cesarean delivery was performed in 75 percent, and there were no stillbirths. As discussed earlier, women with a very early abruption may develop chronic abruption-oligohydramnios sequence. In one report, Elliott and coworkers (1998) described four women with an abruption at a mean gestational age of 20 weeks who developed oligohydramnios and delivered at an average gestational age of 28 weeks. In a description of 256 women with an abruption at <28 weeks’ gestation, Sabourin and colleagues (2012) reported that a mean of 1.6 weeks was gained. Of the group, 65 percent were delivered <29 weeks, and half of all women underwent emergent cesarean delivery.

Unfortunately, even continuous fetal heart rate monitoring does not guarantee universally good outcomes. For example, a normal tracing may precede sudden further separation with instant fetal compromise. In some of these, if the separation is sufficient, the fetus will die before it can be delivered. Tocolysis is advocated by some for suspected abruption if the fetus does not display compromise. Some investigators have observed that tocolysis improved outcomes in a highly selected cohort of women with preterm pregnancies (Bond, 1989; Combs, 1992; Sholl, 1987). In another study, Towers and coworkers (1999) administered magnesium sulfate, terbutaline, or both to 95 of 131 women with abruption diagnosed before 36 weeks. The perinatal mortality rate was 5 percent in both groups with or without tocolysis. Similar results were reported from a randomized trial (Colón, 2016). We are of the opinion that suspected placental abruption contraindicates use of tocolytic agents.

The Latin previa means going before—and in this sense, the placenta goes before the fetus into the birth canal. In obstetrics, placenta previa describes a placenta that is implanted somewhere in the lower uterine segment, either over or very near the internal cervical os. Because these anatomical relationships cannot always be precisely defined, and because they frequently change across pregnancy, terminology can sometimes be confusing.

Placental Migration

Beginning with the use of sonography in obstetrics, the term placental migration was coined to describe the apparent movement of the placenta away from the internal os (King, 1973). Obviously, the placenta does not move per se, and the mechanism of apparent movement is not completely understood. To begin with, migration is clearly a misnomer, because decidual invasion anchors chorionic villi at the cervical os.

Explanations of placental migration are likely additive. First, apparent movement of the low-lying placenta relative to the internal os is related to the imprecision of two-dimensional sonography. Second, as pregnancy progresses, growth of the lower and upper uterine segments differs. With greater blood flow in the upper uterus, placental growth is more likely directed toward the fundus—trophotropism. Many of those placentas that “migrate” most likely never were circumferentially implanted with true villous invasion that reached the internal cervical os. Importantly, a low-lying placenta or placenta previa is less likely to “migrate” if there is a prior cesarean delivery scar.

The frequency of placental migration has been quantified. Sanderson and Milton (1991) studied 4300 women at midpregnancy and found that 12 percent had a low-lying placenta. Of placentas not covering the internal os, previa did not persist, and none subsequently had placental hemorrhage. Conversely, approximately 40 percent of placentas that covered the os at midpregnancy continued to do so until delivery. Thus, placentas that lie close to but not over the internal os up to the early third trimester are unlikely to persist as a previa by term (Heller, 2014; Parrott, 2015). However, other evidence from Bohrer and associates (2012) showed that a second-trimester low-lying placenta was associated with antepartum admission for hemorrhage and increased blood loss at delivery.

The likelihood that placenta previa persists after being identified sonographically at given epochs before 28 weeks’ gestation is shown in Figure 41-20. For twin pregnancies, similar findings are reported until 23 weeks, after which the previa persistence rate is much higher (Kohari, 2012). Stafford and coworkers (2010), but not Trudell and colleagues (2013), found that a previa and a third-trimester cervical length <30 mm elevated the risks for hemorrhage, uterine activity, and preterm birth. Friszer and associates (2013) showed that women admitted for bleeding had a greater chance of delivery in the subsequent 7 days when the cervical length was <25 mm, although Trudell (2013) again did not confirm this.

Classification

Terminology for placenta previa has evolved, and from a Fetal Imaging Workshop sponsored by the National Institutes of Health (NIH), the following classification was recommended:

• Placenta previa—the internal os is covered partially or completely by placenta (Figs. 41-21 and 41-22). In the past, these were further classified as either total or partial previa.

• Low-lying placenta—implantation in the lower uterine segment is such that the placental edge does not cover the internal os but lies within a 2-cm wide perimeter around the os. A previously used term, marginal previa, described a placenta that was at the edge of the internal os but did not overlie it (Reddy, 2014).

Clearly, the classification of some cases of previa will depend on cervical dilation at the time of assessment (Dashe, 2013; Reddy, 2014). For example, a low-lying placenta at 2-cm dilation may become a partial placenta previa at 4-cm dilation because the cervix has opened to expose the placental edge. Conversely, a placenta previa that appears to be total before cervical dilation may become partial at 4-cm dilation because the cervical opening now extends beyond the edge of the placenta. Digital palpation in an attempt to ascertain these changing relations between the placental edge and internal os as the cervix dilates usually causes severe hemorrhage!

With any degree of placenta previa, a certain amount of spontaneous placental separation is inevitable during lower uterine segment remodeling and cervical dilation. Although this frequently causes bleeding, and thus technically constitutes a placental abruption, this term is usually not applied in these instances.

Somewhat but not always related is vasa previa, in which fetal vessels course through membranes and present at the cervical os (Catanzarite, 2016). Vasa previa was recently reviewed by the Society for Maternal-Fetal Medicine (2015) and is discussed in Chapter 6 (Remnants and Cysts).

Incidence and Associated Factors

Demographic Factors

The incidence of placenta previa has risen during past 30 years. Reported incidences average 0.3 percent or 1 case per 300 to 400 deliveries. The frequency at Parkland Hospital from 1988 through 2003 for nearly 250,000 births was 2.6 per 1000. For the 2004 to 2015 epoch, it rose to 3.8 per 1000. Similar frequencies have been reported from Austria, Finland, and Israel (Kollmann, 2016; Räisänen, 2014; Rosenberg, 2011).

Several demographic factors may contribute to this higher risk for placenta previa. First, maternal age raises the frequency of placenta previa (Biro, 2012; Roberts, 2012). In the First- and Second-Trimester Evaluation of Risk (FASTER) trial, which included more than 36,000 women, the frequency of previa was 0.5 percent for women <35 years compared with 1.1 percent in those ≥35 years (Cleary-Goldman, 2005). At Parkland Hospital, this incidence differed from a low rate of approximately 0.65 per 1000 births for women ≤19 years to almost 10 per 1000 births for women older than 35 (see Fig. 41-16).

Multiparity also elevates the risk for previa (Räisänen, 2014). Obviously, the effects of advancing maternal age and parity are confounding. Still, Babinszki and colleagues (1999) reported that the 2.2-percent incidence in women with parity of five or greater was significantly higher than that of women with lower parity. The interpregnancy interval does not affect this rate (Fox, 2015).

Cigarette smoking increases the relative risk of placenta previa at least twofold (Usta, 2005). It has been postulated that carbon monoxide hypoxemia causes compensatory placental hypertrophy and more surface area. Smoking may also be related to decidual vasculopathy. Last, uterine leiomyomas are a risk factor for previa (Jenabi, 2017).

Clinical Factors

Several clinical characteristics also raise previa risks. Foremost, women with one or more prior cesarean deliveries are at greater risk for subsequent placental disorders that include placenta previa, abruption, or morbidly adherent placenta (Gibbins, 2018; Klar, 2014). The cumulative risks for placenta previa that accrue with the increasing number of cesarean deliveries are extraordinary. The risk rises even further if there was a prior prelabor cesarean delivery (Downes, 2015). In one MFMU Network study of 30,132 women undergoing cesarean delivery, the incidence was 1.3 percent for those with only one prior cesarean delivery, but it was 3.4 percent if there were six or more prior cesareans (Silver, 2006). In a retrospective cohort of nearly 400,000 women who were delivered of two consecutive singletons, those with a cesarean delivery for the first pregnancy had a 1.6-fold greater risk for previa in the second pregnancy (Gurol-Urganci, 2011). These same investigators reported a 1.5-fold higher risk from six similar population-based cohort studies. The likelihood of previa is increased more than eightfold in women with parity greater than four and who have more than four prior cesarean deliveries (Gesteland, 2004; Gilliam, 2002).

Importantly, women with a prior uterine incision and placenta previa have an elevated likelihood that cesarean hysterectomy will be necessary because of an associated morbidly adherent placenta (Wei, 2014). In one study, 6 percent of women with a primary cesarean delivery for previa required a hysterectomy. This rate was 25 percent for women with a previa undergoing repeat cesarean delivery (Frederiksen, 1999).

Maternal serum alpha-fetoprotein (MSAFP) levels, if abnormally elevated for otherwise unexplained reasons during prenatal screening, raise the risk for previa and a host of other abnormalities. Moreover, women with a previa and comorbid MSAFP level ≥2.0 multiples of the median (MoM) at 16 weeks’ gestation were at greater risk for late-pregnancy bleeding and preterm birth (Chap. 14, Maternal Serum AFP Elevation: Neural-Tube Defect Screening).

Last, assisted reproductive technology (ART) used for conception elevates previa risks. Some of this association may derive from overlapping effects. For example, older women comprise a significant portion of ART patients (Luke, 2017). In addition, multifetal gestation is a well-known risk of both in vitro fertilization and previa. However, even adjusting for these overlapping elements, ART is still associated with higher previa rates (Romundstad, 2006).

Clinical Features

Painless bleeding is the most characteristic event with placenta previa. Bleeding usually does not develop until near the end of the second trimester or later, but it can begin even before midpregnancy. And undoubtedly, some late abortions are caused by an abnormally located placenta. Bleeding from a previa usually begins without warning and without pain or contractions in a woman who has had an uneventful prenatal course. This so-called sentinel bleed is rarely so profuse as to prove fatal. Usually it ceases, only to recur. However, in perhaps 10 percent of women, particularly those with a placenta implanted near but not over the cervical os, there is no bleeding until labor onset. Bleeding at this time varies from slight to profuse, and it may clinically mimic placental abruption.

A specific sequence of events leads to bleeding in cases in which the placenta is located over the internal os. First, the uterine body remodels to form the lower uterine segment. With this, the internal os dilates, and some of the implanted placenta inevitably separates. Bleeding that ensues is augmented by the inherent inability of myometrial fibers in the lower uterine segment to contract and thereby constrict torn vessels. Similarly, bleeding from this lower segment implantation site also frequently continues after placental delivery. Last, there may be lacerations in the friable cervix and lower segment. These may be especially problematic following manual removal of a somewhat adhered placenta.

Morbidly adherent placentas are a frequent and serious complication associated with placenta previa. Described later (Morbidly Adherent Placenta), this abnormally firm placental attachment derives in part from poorly developed decidua that lines the lower uterine segment. Biswas and coworkers (1999) performed placental bed biopsies in 50 women with a previa and in 50 control women. Trophoblastic giant-cell infiltration of spiral arterioles—rather than endovascular trophoblast cells—was found in half of previa specimens. In contrast, only 20 percent of biopsies from normally implanted placentas had these changes. In another study of 514 cases of previa, abnormal placental attachment was identified in 7 percent (Frederiksen, 1999). As discussed, previa overlying a prior cesarean incision conveys a particularly high risk for morbidly adherent placenta.

Coagulation defects are rare complications of placenta previa, even when implantation site separation is extensive (Cunningham, 2015). Placental thromboplastin, which incites the intravascular coagulation seen with placental abruption, is presumed to readily escape through the cervical canal rather than be forced into the maternal circulation. The paucity of large myometrial veins in this area may also be protective.

Diagnosis

Whenever there is uterine bleeding after midpregnancy, placenta previa or abruption are always considered. In the Canadian Perinatal Network study discussed earlier (Uterine Atony), placenta previa accounted for 21 percent of women admitted from 22 to 28 weeks’ gestation with vaginal bleeding (Sabourin, 2012). Previa should not be excluded until sonographic evaluation has clearly proved its absence. If sonography is not readily available, diagnosis by clinical examination is done using the double set-up technique because it requires that a finger be passed through the cervix and the placenta palpated. A digital examination should not be performed unless delivery is planned. A cervical digital examination is done with the woman in an operating room and with preparations for immediate cesarean delivery. Even the gentlest examination can cause torrential hemorrhage. Fortunately, double set-up examination is rarely necessary because placental location can almost always be ascertained sonographically.

Quick and accurate localization can be accomplished using standard sonographic techniques (American Institute of Ultrasound in Medicine, 2013). This is usually done with transabdominal sonography. If the placenta clearly overlies the cervix or if it lies away from the lower uterine segment, the examination has excellent sensitivity and negative-predictive value (Olive, 2006; Quant, 2014). Obese women may have limitations of visualization of the lower uterine segment. Also, a full bladder may artificially elongate the cervix and compress the lower uterine segment to give the impression that the placenta overlies the cervix. If placental location remains in question, then transvaginal sonography is the most accurate method of assessment (Fig. 41-23). It is safe, even when there is bleeding.

Accuracy depends on the sonographic technique used. In a comprehensive study, the internal os was visualized in all cases with transvaginal sonography but in only 30 percent with transabdominal sonography (Farine, 1988). As discussed, according to the Fetal Imaging Workshop, if the placental edge is <2 cm from the internal os, but not covering it, the placenta is considered low lying (Reddy, 2014). In the absence of any other indication, sonography need not be frequently repeated simply to document placental position. At Parkland, women with a placenta previa identified at 18 to 22 weeks’ gestation with a prior cesarean delivery are evaluated again at 28 weeks and those without at 32 weeks. Restriction of activity is not necessary unless a previa persists beyond 28 weeks or if clinical findings such as bleeding or contractions develop before this time. At 32 weeks’ gestation, if the placental edge is still <2 cm from the os, then transvaginal sonography is repeated at 36 weeks.

Using MR imaging, several investigators have reported excellent results in visualizing placental abnormalities. That said, it is unlikely that this technique will replace sonography for routine evaluation anytime soon. However, MR imaging has proved useful for evaluation of morbidly adherent placenta (Clinical Presentation and Diagnosis).

Management

Women with a placenta previa are managed based on their individual clinical circumstances. Three prominent factors include fetal age and maturity, labor, and bleeding severity. In one study of 214 women with a previa, 43 percent had an emergency delivery, and half of these were preterm (Ruiter, 2015). But, if the fetus is immature and active bleeding subsides, close observation in an obstetrical unit is indicated. Data are sparse regarding tocolytic administration for uterine contractions. Although robust randomized trials are lacking, Bose and colleagues (2011) recommend that if tocolytics are given, they be limited to 48 hours of administration. We categorically recommend against their use in this setting.

After bleeding has ceased for approximately 2 days and the fetus is judged to be healthy, a woman can usually be discharged home with instructions for “pelvic rest.” Importantly, the woman and her family must fully appreciate the possibility of recurrent bleeding and be prepared for immediate transport back to the hospital. In other cases, prolonged hospitalization may be ideal.

The frequency of emergency delivery in women with placenta previa ranges from 25 to 40 percent (Gibbins, 2018; Kassir, 2017). But, in properly selected patients, long-term inpatient care does not appear to add benefits compared with outpatient management (Neilson, 2003). In one randomized study of 53 women who had a bleeding previa at 24 to 36 weeks’ gestation, maternal or fetal morbidity rates did not differ between management method (Wing, 1996). Of all study women, 60 percent had recurrent bleeding, and half eventually required expeditious cesarean delivery.

For women who are near term and who are not bleeding, plans are made for scheduled cesarean delivery. Timing balances fetal immaturity risks against antepartum hemorrhage. One NIH workshop suggested elective delivery at 36 to 37 completed weeks’ gestation (Spong, 2011). The Society for Maternal-Fetal Medicine (2017) recommends delivery between 34 and 37 weeks. At Parkland Hospital, we usually perform elective cesarean delivery at 38 weeks. With a suspected morbidly adherent placenta, delivery is recommended at 34 to 35 completed weeks by the NIH workshop (Management). Our practice is to schedule delivery at 36 completed weeks.

Delivery

Practically all women with placenta previa undergo cesarean delivery. Many surgeons recommend a vertical laparotomy incision to provide rapid entry in cases with torrential bleeding or operating space if hysterectomy is required. As discussed, cesarean delivery is emergently performed in more than half because of hemorrhage, for which about a fourth require blood transfusion (Boyle, 2009; Sabourin, 2012). Although a low transverse hysterotomy is usually possible, this may cause fetal bleeding if the placenta is implanted anteriorly and the placenta is incised. In such cases, fetal delivery should be expeditious (Silver, 2015a). A vertical uterine incision may be preferable in some instances. In either case, even when the incision extends through the placenta, maternal or fetal outcomes are rarely compromised.

Following placental removal, the placenta site may bleed uncontrollably due to poorly contracted smooth muscle, which is characteristic of the lower uterine segment. If hemostasis at the placental implantation site cannot be obtained by adequate uterotonic administration and pressure, it can be oversewn with 0-chromic sutures. Cho and associates (1991) described interrupted 0-chromic sutures at 1-cm intervals to form a circle around the bleeding portion of the lower segment to control hemorrhage. Others have reported success with compression sutures that traversed and compressed the anterior and posterior uterine wall (Kayem, 2011; Penotti, 2012).

Of other methods, Bakri or Foley balloon tamponade used alone or coupled with compression sutures has been described (Albayrak, 2011; Diemert, 2012; Kumru, 2013). Law and coworkers (2010) successfully used a hemostatic gel. Other surgical options are bilateral uterine or internal iliac artery ligation, illustrated later (Adjunctive Surgical Procedures). Finally, pelvic artery embolization has also gained acceptance.

Hysterectomy

If these more conservative methods fail and bleeding is brisk, hysterectomy is necessary. Placenta previa—especially with an abnormally adherent placenta—currently is the most frequent indication for peripartum hysterectomy at Parkland Hospital and other institutions (Jakobsson, 2015; Wong, 2011). When there is no associated accrete syndrome, the reported incidence of hysterectomy is 2 percent (Gibbins, 2018).

Thus, it is not possible to accurately estimate the effect on the hysterectomy rate from previa alone without considering the associated accrete syndromes. Again, for women whose placenta previa is implanted anteriorly at the site of a prior uterine incision, the likelihood of an associated morbidly adherent placenta and need for hysterectomy is increased. In a study of 318 peripartum hysterectomies performed in the United Kingdom, 40 percent were done for abnormal placentation (Knight, 2007). Similar results were reported for 211 hysterectomies from the Nordic Obstetric Surveillance Study (Jakobsson, 2015). At Parkland Hospital, 44 percent of cesarean hysterectomies were done for bleeding placenta previa or for a morbidly adherent placenta (Wortman, 2015). The technique for peripartum hysterectomy is described in Chapter 30 (Peripartum Hysterectomy).

Maternal and Perinatal Outcomes

Placenta previa and coexistent accrete syndromes both contribute substantively to maternal morbidity and mortality rates. The maternal mortality ratio is increased approximately threefold for women with a placenta previa (Gibbins, 2018; Oyelese, 2006). In another report of 5367 maternal deaths in the United States from 2006 to 2013, placenta previa alone accounted for nearly 3 percent of deaths from hemorrhage (Creanga, 2015, 2017).

The report from the Consortium on Safe Labor emphasizes the ongoing perinatal morbidity with placenta previa (Lai, 2012). Preterm delivery continues to be a major cause of perinatal death (Nørgaard, 2012). In deliveries with placenta previa in the United States in 1997, the neonatal mortality rate was threefold higher than that in unaffected pregnancies and stemmed primarily from preterm delivery (Salihu, 2003). Ananth and colleagues (2003) reported a comparably elevated risk of neonatal death even for fetuses who delivered at term. This is at least partially related to the fetal anomaly rate, which is two- to threefold higher in pregnancies with placenta previa (Crane, 1999).

The association of fetal-growth restriction with placenta previa is likely minimal after controlling for gestational age. In a population-based cohort of more than 500,000 singleton births, Ananth and associates (2001) found that most low-birthweight newborns associated with placenta previa resulted from preterm birth. Harper and coworkers (2010) reported similar findings from a cohort of nearly 58,000 women. In contrast, at least two studies reported a greater risk for fetal-growth restriction (Räisänen, 2014; Weiner, 2016).

Etiopathogenesis

The term morbidly adherent placenta describes aberrant placentation characterized by abnormally implanted, invasive, or adhered placenta. We also refer to these disorders collectively as accrete syndromes and use these terms interchangeably. Derivation of accrete comes from the Latin ac- + crescere—to adhere or become attached to (Benirschke, 2012).

In the accrete syndromes, abnormal placental adherence to the myometrium stems in part from partial or total absence of the decidua basalis and imperfect development of the fibrinoid or Nitabuch layer, described in Chapter 5 (Decidual Histology). If the decidual spongy layer is lacking either partially or totally, then the physiological line of cleavage is absent, and some or all cotyledons are densely anchored. Microscopically, placental villi attach to smooth muscle fibers rather than to decidual cells. This decidual deficiency then prevents normal placental separation after delivery. The surface area of the implantation site involved and the depth of trophoblastic tissue ingrowth are variable between women, but all affected placentas can potentially cause significant hemorrhage.

Substantiated data now suggest that accrete syndromes are not solely caused by this anatomical layer deficiency (Duzyj, 2017; Tantbirojn, 2008). Indeed, the cytotrophoblasts may control decidual invasion through factors such as angiogenesis (Duzyj, 2015; Goh, 2016; Wehrum, 2011). Also, accrete syndrome tissue specimens show “hyperinvasiveness” (Pri-Paz, 2012). Myometrial fibers attached to the basal plate in an antecedent pregnancy are predictive markers for a subsequent placenta accreta (Linn, 2015; Miller, 2016). This implies an antecedent “constitutional endometrial defect” in most cases. The greater risk conveyed by previous surgical uterine trauma may be partially explained by an enhanced vulnerability to trophoblast invasion (Garmi, 2012; Gill, 2015; Jauniaux, 2017).

This association with prior trauma is reinforced by the close relationship between cesarean-scar pregnancy (CSP) and later development of placenta accreta in the same pregnancy. Indeed, accruing evidence suggests that CSP and accrete syndromes lie on a spectrum and that CSP is a precursor, as both share the same histopathology (Happe, 2018; Timor-Tritsch, 2014). CSP frequency has been reported to approximate 1 in 2000 pregnancies (Berhie, 2015; Rotas, 2006). Described in Chapter 19 (Cesarean Scar Pregnancy), early rupture and hemorrhage are not uncommon with CSP, and women often elect pregnancy-terminating interventions to avoid these (Michaels, 2015; Timor-Tritsch, 2015).

Classification

Variants of the morbidly adherent placenta are classified by the depth of trophoblastic growth (Figs. 41-24 and 41-25). Placenta accreta indicates that villi are attached to the myometrium. With placenta increta, villi actually invade the myometrium, and placenta percreta defines villi that penetrate through the myometrium and to or through the serosa (Bailit, 2015; Silver, 2015a). In clinical practice, these three variants are encountered in an approximate ratio of 80:15:5, respectively (Wong, 2008). In all three varieties, abnormal adherence may involve all lobules—total placenta accreta. If all or part of a single lobule is abnormally attached, it is described as a focal placenta accreta. Histological diagnosis cannot be made from the placenta alone, and myometrial samples are necessary for confirmation (Benirschke, 2012).

Incidence

The frequency of accrete syndromes was 1 in 20,000 births almost 100 years ago (McKeogh, 1951). As late as 1971, Hellman and Pritchard in the 14th edition of Williams Obstetrics described accreta to be the subject of case reports. Since then, the incidence has grown remarkably in direct relationship to the rising cesarean delivery rate. For example, incidence was 1 in 2500 births in the 1980s, but it was 1 per 731 births in the report from the MFMU Network comprising 115,502 women (Bailit, 2015). And a Canadian study of more than 570,000 births found an incidence of 1 in 700 deliveries (Mehrabadi, 2015). In the Nationwide Inpatient Sample, the prevalence of accreta was 3.7 per 1000 births—1 per 270 (Mogos, 2016).

This rising frequency has made accrete syndromes one of the most formidable problems in obstetrics. In one review of 5367 pregnancy-related maternal deaths in the United States from 2006 to 2013, 13 percent were due to hemorrhage caused by accrete syndromes (Creanga, 2015, 2017). In addition, they are a leading cause of hemorrhage and emergency peripartum hysterectomy (Awan, 2011; Eller, 2011; Rossi, 2010). The American College of Obstetricians and Gynecologists (2017c) and the Society for Maternal-Fetal Medicine (2010) have taken the lead to address and optimize management.

In subsequent pregnancies following placenta accreta, recurrence risks are high. Women in whom hysterectomy is avoided have an estimated 20-percent incidence of recurrence (Cunningham, 2016; Roeca, 2017). In addition, some evidence shows that these women have greater risks for previa, uterine rupture, and hysterectomy (Eshkoli, 2013).

Risk Factors

These are similar in many aspects to those for placenta previa (Classification). That said, the two most important risk factors are an associated previa, a prior cesarean delivery, and more likely a combination of the two (Klar, 2014). A classical hysterotomy incision has a higher risk for a subsequent accrete placenta (Gyamfi-Bannerman, 2012). In fact, almost half of women with a prior cesarean delivery had myometrial fibers seen microscopically adhered to the placenta (Hardardottir, 1996; Miller, 2016). An associated previa confers an even higher risk. This is shown in Figure 41-26, and the astonishing increase in frequency of associated accrete syndromes is apparent with a concomitant previa.

Dysfunctional decidual formation also may follow any other type of myometrial trauma such as curettage or endometrial ablation (Benirschke, 2012; Gill, 2015). Even without a prior hysterotomy, coexisting placenta previa is additive to frequency, and in one study, 10 percent of such women with a previa had an associated accrete syndrome. A shorter cervical length with placenta accrete syndromes did not confer a greater risk for preterm delivery (Rac, 2017).

Another risk marker became apparent with widespread use of MSAFP and human chorionic gonadotropin (hCG) screening for neural-tube defects and aneuploidies. In one study of more than 9300 women screened at 14 to 22 weeks’ gestation, the risk for accrete syndromes was eightfold higher with MSAFP levels >2.5 MoM, and it was increased fourfold with maternal serum free β-hCG levels >2.5 MoM (Hung, 1999).

Clinical Presentation and Diagnosis

In cases of first- and second-trimester accrete syndromes, there is usually hemorrhage that is the consequence of coexisting placenta previa. Such bleeding will typically prompt evaluation and management. In some women who do not have an associated previa, accreta may not be identified until third-stage labor when an adhered placenta is encountered. Unfortunately, imaging modalities are less than perfect to identify all of these placentas early.

Ideally, sonography is used for antepartum identification of abnormal placental ingrowth (Chantraine, 2013; Jauniaux, 2016; Reddy, 2014; Tam Tam, 2012). Happe and colleagues (2018) found that first-trimester measurement of the smallest myometrial thickness can be used to predict the necessity for peripartum hysterectomy with an accrete syndrome. Other findings include loss of the normal hypoechoic retroplacental zone between the placenta and uterus, placental vascular lacunae, and placental bulging into the posterior bladder wall (Fig. 41-27). Using these criteria, Warshak and associates (2006) calculated the following values: sensitivity of 77 percent; specificity of 96 percent; positive-predictive value of 98 percent. Similar values are cited by the American College of Obstetricians and Gynecologists (2017c) and others (Chalubinski, 2013; Elhawary, 2013; Maher, 2013).

Despite these findings, some investigators report less spectacular results with sonography (Jauniaux, 2016; Primo, 2014). Bowman and colleagues (2014) described the sensitivity of sonography to be 54 percent; specificity, 88 percent; positive-predictive value, 82 percent; negative-predictive value, 65 percent; and accuracy, 65 percent. Location affects sonographic accuracy. In one study, the detection rate was 90 percent for anterior placenta accreta compared with 50 percent for posterior wall ones (Pilloni, 2016). Nageotte (2014) concluded that identification of the morbidly adherent placenta with sonography should be interpreted along with clinical and operative findings.

Better results have been reported by some using three-dimensional (3-D) sonography and power Doppler (Collins, 2015; Doyle, 2015). We too have found that the addition of Doppler color flow mapping is highly predictive of myometrial invasion (Fig. 41-28). This is suspected if the distance between the uterine serosa–bladder wall interface and the retroplacental vessels measures <1 mm and if there are large intraplacental lacunae (Rac, 2015a; Twickler, 2000). Similarly, Cali and associates (2013) reported that hypervascularity of the uterine serosa–bladder wall interface had the highest positive- and negative-predictive values for placenta percreta.

MR imaging can be added to outline anatomy and to identify invasion of adjacent structures, including possible ureteral involvement (Chalubinski, 2013; Reddy, 2014). Although gadolinium is usually not added during pregnancy, this contrast may enhance images (Millischer, 2017). Lax and coworkers (2007) described three MR imaging findings that suggest accreta: uterine bulging, heterogeneous signal intensity within the placenta indicative of lacunae, and dark intraplacental bands on T2-weighted imaging. Some recommend use of MR imaging if sonography results are inconclusive or there is a posterior previa (American College of Obstetricians and Gynecologists, 2017c; Silver, 2015a).

Management

Preoperative assessment ideally begins once a possible accrete syndrome is recognized antenatally (Fitzpatrick, 2014; Sentilhes, 2013). A major decision concerns the timing of and the ideal facility for delivery. Considerations include appropriate surgical, anesthesia, intensive care, and blood banking capabilities. An obstetrical surgeon or gynecological oncologist and surgical, urological, and interventional radiological consultants should be available (Brennan, 2015; Shamshirsaz, 2015). The American College of Obstetricians and Gynecologists (2017c) and the Society for Maternal-Fetal Medicine (2010) recommend planned delivery in a tertiary-care facility. In some of these, specially designed teams have been assembled and are on call (Al-Khan, 2014; Erfani, 2017a; Smulian, 2017; Walker, 2013).

Silver and colleagues (2015b) have provided criteria for accreta centers of excellence. Shown in Table 41-5 are some criteria to consider transfer to a higher level-of-care facility. Women who refuse blood or its derivatives pose especially difficult management dilemmas (Barth, 2011). If possible, delivery is best scheduled for peak availability of all resources and team members. Even so, a third of cases require unscheduled delivery, and contingency plans should be ready (Pettit, 2017).

Timing of Delivery

Timing balances fetal immaturity risks against serious adverse maternal consequences of emergency cesarean delivery (Stephenson, 2016). The American College of Obstetricians and Gynecologists (2017c) recommends individualization of delivery timing. It cites a decision-analysis study that justifies elective delivery without fetal lung maturity testing after 34 completed weeks (Robinson, 2010). The Society for Maternal-Fetal Medicine (2017) recommends delivery between 34 and 37 weeks. Two recent surveys found that most practitioners do not deliver these women until 36 weeks or later (Esakoff, 2012; Wright, 2013). At Parkland Hospital, we generally schedule these procedures after 36 completed weeks but are prepared also to manage them in nonelective situations (Rac, 2015b). Perlman and colleagues (2017) recommend individualization based on specific risk criteria.

In some cases, placenta accrete syndrome is not recognized until laparotomy. If there are inadequate resources to surgically manage the percreta, and if the woman is stable and not bleeding, then the fetus is not delivered, the abdominal incision is closed, and she is transferred to a tertiary-care facility.

Preoperative Prophylactic Catheterization

In cases that may involve one or both ureters, catheterization may aid in dissection or identification and repair of injury. Some, but not all, advocate preoperative ureteral catheterization (Eller, 2011; Society for Maternal-Fetal Medicine, 2010; Tam Tam, 2012).

Balloon-tipped intraarterial catheters to mitigate blood loss and thereby enhance surgical visibility have also gained supporters. Catheters are advanced preoperatively into the internal iliac arteries, and then after delivery, they are inflated to occlude pelvic blood flow (Ballas, 2012; Desai, 2012). Alternatively, the catheters can be used to deliver occluding emboli to bleeding arterial sites. Others have concluded that these procedures offer borderline efficacy and have serious risks (Salim, 2015; Sentilhes, 2009). Complications have included thromboses of the common and left iliac arteries (Bishop, 2011). At this time, the American College of Obstetricians and Gynecologists (2017c) concludes that a firm recommendation cannot be made for or against intraarterial catheter use. Similarly, there are no obvious benefits to internal artery ligation (Eller, 2011; Po, 2012).

Cesarean Delivery and Hysterectomy

Before commencing with delivery, the risk of hysterectomy to prevent exsanguination should be estimated. Some of these abnormal placentations, especially if partial, may be amenable to placental delivery with hemostatic suture placement. Confirmation of a percreta or increta almost always mandates hysterectomy. Because the scope of invasion may not be apparent before delivery of the fetus, we complete many dissection steps early. This also minimizes blood loss during potentially tedious dissection after hysterotomy. Thus, we usually attempt to create a wide bladder flap before making the hysterotomy incision (Cunningham, 2017b). The round ligaments are divided, and the lateral edges of the peritoneal reflection are dissected downward. If possible, these incisions are extended to encircle the entire placental implantation site that visibly occupies the prevesical space and posterior bladder wall. Following this, a classical hysterotomy or transverse fundal incision is made to avoid the placenta (Kotsuji, 2013).

After fetal delivery, the extent of placental invasion is assessed without attempts at manual placental removal. In a report from the United Kingdom, attempts for partial or total placental removal prior to hysterectomy were associated with twice as much blood loss (Fitzpatrick, 2014). Generally speaking, with obvious percreta or increta, hysterectomy is usually the best course, and the placenta is left in situ (Eller, 2011). With more extensive placental ingrowth, there may be little or no bleeding until manual placental removal is attempted. Unless there is spontaneous separation with bleeding that mandates emergency hysterectomy, the operation begins after full assessment is made. With bleeding, successful treatment depends on immediate blood replacement therapy and other measures that can include uterine or internal iliac artery ligation, balloon occlusion, or embolization.

The group at Baylor College of Medicine has described a modified radical hysterectomy for surgical management of the morbidly adherent placenta (Shamshirsaz, 2015). For a description of this technique, refer to Cunningham and Gilstrap’s Operative Obstetrics (Yeomans, 2017). At Parkland Hospital, we have had cases in which a traditional radical hysterectomy was necessary to excise all abnormally implanted placenta.

Conservative Management

Occasionally, it may be possible to trim the umbilical cord, repair the hysterotomy incision, leave the placenta in situ, and not pursue hysterectomy. This option may be used for women in whom abnormal placentation was not suspected before cesarean delivery and in whom uterine closure stops bleeding. After this, she can be transferred to a higher-level facility for definitive management. Another consideration is the woman with a strong desire for fertility and who has received extensive counseling.

Conservative management was reviewed by Perez-Delboy (2014) and Fox (2015) and their colleagues. In some of these cases, the placenta spontaneously resorbed between 1 and 12 months with a mean of 6 months. Numerous complications can occur and include sepsis, disseminated intravascular coagulation, pulmonary embolism, and arteriovenous malformation (Fox, 2015; Judy, 2015; Roach, 2015).

In some of these women, a subsequent hysterectomy—either planned or prompted by bleeding or infection—is performed days to weeks postpartum when blood loss might be lessened (Al-Khan, 2014; Sentilhes, 2009). In one study, only 21 percent of such women ultimately required hysterectomy (Bretelle, 2007). In other reports, however, up to 60 percent eventually required emergency hysterectomy (Clausen, 2013; Pather, 2014). Evidence that treatment with methotrexate aids resorption is lacking. Last, for women in whom the placenta is left in situ, serial serum β-hCG measurements are not informative, and serial sonographic or MR imaging is recommended (Timmermans, 2007; Worley, 2008).

At this time, we agree with the American College of Obstetricians and Gynecologists (2017c) that leaving the placenta in situ is seldom indicated. Exceptions are for temporization to permit transfer to a higher level of care.

Pregnancy Outcomes

In sum, these syndromes can have disastrous outcomes for both mother and fetus. Although the depth of placental invasion does not correspond with perinatal outcome, it is of paramount maternal significance (Seet, 2012). Shown in Table 41-6 are outcomes from reports of women from tertiary-care hospitals and in whom the diagnosis of morbidly adherent placenta was made preoperatively. Despite these advantages, a litany of complications included hemorrhage, urinary tract injury, intensive care unit admission, and secondary surgical procedures. Some of these reports chronicle outcomes in a second cohort of women in whom care was not given at a tertiary-care facility or in whom the diagnosis of percreta was not made until delivery, or both. In these cohorts, morbidity was higher, and there was one maternal death.

The terms consumptive coagulopathy, defibrination syndrome, or disseminated intravascular coagulation (DIC) are often used interchangeably, but there is an important distinction in these terms. An event related to actual consumption of procoagulants within the intravascular tree results in a consumptive coagulopathy. In contrast, massive loss of procoagulants from hemorrhage results in a dilutional coagulopathy. Semantics aside, the clinicopathological coagulation disturbances with consumptive coagulopathy culminate in a systemic intravascular activation that completely disrupts natural hemostasis. As a result, an ineffective balance of natural anticoagulant mechanisms leads to widespread fibrin deposition that can cause multiorgan failure (Levi, 2013).

Disseminated Intravascular Coagulation in Pregnancy

Because of the many definitions and variable severity, citing an accurate incidence for consumptive coagulopathy in pregnant women is problematic, but it ranges from 0.03 to 0.35 percent (Erez, 2014; Rattray, 2012). For example, some degree of significant coagulopathy is found in virtually all cases of placental abruption and amnionic fluid embolism. Other instances in which frequently occurring but less recognized degrees of coagulation activation can be found include sepsis, thrombotic microangiopathies, acute kidney injury, acute fatty liver, severe preeclampsia, and hemolysis, elevated liver enzyme levels, low platelet count (HELLP) syndrome (Cunningham, 2015). The overall contribution of each of these obstetrical disorders also varies depending on the population studied (Erez, 2015).

When consumptive coagulopathy is severe, the likelihood of maternal and perinatal morbidity and mortality is increased. In one study of 49 cases, antecedent causes included those listed above, and 59 percent received blood transfusions, 18 percent underwent hysterectomy, 6 percent were dialyzed, and three mothers died (Rattray, 2012). The perinatal mortality rate was 30 percent. Callaghan and associates (2012) reviewed data from the Nationwide Inpatient Sample and found a rising prevalence of DIC from 1998 to 2009. And, from 2010 to 2011, DIC was the second most common severe maternal morbidity indicator (Creanga, 2014). Notably, DIC was associated with nearly a fourth of maternal deaths during this study period. Despite these statistics, consumptive coagulopathy as the sole cause of maternal death is relatively uncommon and accounts for only 0.2 percent of pregnancy-related deaths in the United States (Creanga, 2015).

Pregnancy-Induced Coagulation Changes

During normal pregnancy, extensive changes in coagulation and fibrinolysis develop to create a procoagulant state. Some of these include appreciable increases in the plasma concentrations of factors I (fibrinogen), VII, VIII, IX, and X. A partial list of these normal values is found in the Appendix (Serum and Blood Constituents). At the same time, plasminogen levels rise considerably, but levels of plasminogen activator inhibitor-1 and 2 (PAI-1 and PAI-2) also grow. Thus, plasmin activity usually declines until after delivery (Hale, 2012; Hui, 2012). The mean platelet count drops by 10 percent during pregnancy, and platelet activation is enhanced (Kenny, 2015).

The net results of these changes include greater levels of fibrinopeptide A, β-thromboglobulin, platelet factor 4, and fibrinogen-fibrin degradation products, which includes d-dimers. Along with lower concentrations of anticoagulant protein S, hypercoagulability, and decreased fibrinolysis, there is augmented—yet compensated—intravascular coagulation that may function to maintain the uteroplacental interface.

Activation of Normal Coagulation

Instead of the “waterfall” sequential activation of clotting, a current theory proposes that tissue factor—an integral membrane glycoprotein—serves as the principal initiator of coagulation (Levi, 2010b). Coagulation then moves forward but incorporates a feedback loop. To begin, tissue factor forms complexes with factor VII/VIIa to activate factors IX and X. Tissue factor is found in highly vascularized organs such as the brain, lungs, and placenta; in amnionic fluid; and in certain other cell types (Kuczyński, 2002; Østerud, 2006; Uszyński, 2001).

Tissue factor-factor VIIa complexes ultimately generate activated factor X (Xa) to initiate clotting. Subsequently, the previously labeled “intrinsic” pathway amplifies this process. Specifically, the initial thrombin produced directly activates factor XI by providing a feedback amplification loop. This primary role of tissue factor–factor VIIa complex in coagulation and consequent amplification loop of thrombin is depicted in Figure 41-29 (Rapaport, 1995). The end result of this amplified coagulation process is fibrin formation. This is then counterbalanced by the fibrinolytic system, in which plasminogen is activated. As shown in Figure 41-29, even this process is tied initially to tissue factor. The final result is production of fibrinogen/fibrin degradation products, which include d-dimers.

Activation of Pathological Coagulation

The initiation of DIC begins with the release of tissue factor by pathological entities. Tissue factor is released by subendothelial tissue and stimulated monocytes, which in turn provoke release of cytokines from the endothelium. With generalized endothelial activation, diffuse activation of coagulation follows. This pathological cycle of coagulation and fibrinolysis becomes clinically important when coagulation factors and platelets are sufficiently depleted to create consumptive coagulopathy.

Several obstetrical syndromes can trigger consumptive coagulopathy. The best known and most common is placental abruption with its significant release of thromboplastin. Another is embolization of amnionic fluid and debris into the maternal circulation. This causes activation of factor X by abundant mucin found in fetal squames. Other causes include endotoxins from gram-negative bacteria and exotoxins from gram-positive bacteria.

Diagnosis

Bioassay is an excellent method to detect or suspect clinically significant coagulopathy. Excessive bleeding at sites of modest trauma characterizes defective hemostasis. Examples include persistent bleeding from venipuncture sites, nicks from shaving the perineum or abdomen, trauma from bladder catheterization, and spontaneous bleeding from the gums, nose, or gastrointestinal tract. Purpura or petechiae at pressure sites such as sphygmomanometer cuffs or tourniquets suggest significant thrombocytopenia. Any surgical procedure provides the ultimate bioassay and elicits generalized oozing from abdominal wall layers, the retroperitoneal space, the episiotomy, or incisions and dissections for cesarean delivery or hysterectomy.

Of laboratory tests, fibrinogen, fibrin, and degradation product levels can be informative. In late pregnancy, plasma fibrinogen levels typically have risen to 300 to 600 mg/dL. Even with severe consumptive coagulopathy, levels may sometimes be sufficiently high to protect against clinically significant hypofibrinogenemia. For example, defibrination caused by a placental abruption might lower an initial fibrinogen level of 600 mg/dL to 250 mg/dL. Although this would indicate massive fibrinogen consumption, levels are still adequate to promote clinical coagulation—usually about 150 mg/dL. If serious hypofibrinogenemia—less than 50 mg/dL—is present, the clot formed from whole blood in a glass tube may initially be soft but not necessarily remarkably reduced in volume. Then, over the next half hour or so, as platelet-induced clot retraction develops, the clot becomes quite small. When many of the erythrocytes are extruded, the volume of liquid in the tube clearly exceeds that of clot.

As depicted in Figure 41-29, fibrinolysis cleaves fibrin and fibrinogen into various fibrin degradation products that are detected by several sensitive assays. There are many fragment types, and monoclonal antibodies in assay kits usually measure d-dimers specific for that assay. These values are always abnormally high with clinically significant consumptive coagulopathy. At least in obstetrical disorders, quantification has not been correlated with outcomes. Examples of the magnitude of fibrin split product elevations in various obstetrical coagulopathies is shown in Figure 41-30.

Thrombocytopenia is likely if petechiae are abundant or if clotted blood fails to retract within an hour or so. Confirmation is provided by a low platelet count. If severe preeclampsia syndrome is comorbid, there may also be qualitative platelet dysfunction (Chap. 40, Maternal Thrombocytopenia).

Prothrombin time (PT) and partial thromboplastin time (PTT) are standard coagulation tests. Prolongation may stem from very low fibrinogen concentrations, from appreciably reduced levels of the procoagulants needed to generate thrombin, or from large amounts of circulating fibrinogen-fibrin degradation products.

Thromboelastometry and thromboelastography are point-of-care tests used as adjuncts to conventional laboratory studies (Abdul-Kadir, 2014). Their current role may serve to guide blood product replacement, discussed later (Viscoelastic Assays).

Using many of these tests, several organizations have attempted to establish a more uniform definition of DIC. One is the International Society on Thrombosis and Haemostasis (ISTH) scoring system. The score is used only after a condition known to cause intravascular coagulation is identified and is calculated using a combination of laboratory tests. Composite ISTH-DIC scores <5 suggest nonovert DIC, whereas scores ≥5 are compatible with overt DIC. Other than one report of acute fatty liver of pregnancy, this scoring system has not been applied widely in obstetrics (Nelson, 2014).

General Management

To halt ongoing defibrination, prompt identification and removal of the inciting source of the coagulopathy is a priority. With surgical incisions or extensive lacerations accompanied by severe hemorrhage, rapid replacement of procoagulants is usually indicated. Vigorous restoration and maintenance of the circulation to treat hypovolemia cannot be overemphasized. Adequate perfusion restores hepatic and endothelial synthesis of procoagulants and permits prompt removal of activated coagulation factors, fibrin, and fibrin degradation products by the reticuloendothelial system.

Aside from these fundamental steps, few other agents have proven soundly effective. Although seemingly counterintuitive, unfractionated heparin had been recommended but has now been abandoned. Other examples include use of antifibrinolytic agents—either tranexamic acid or epsilon-aminocaproic acid (Amicar) (American College of Obstetricians and Gynecologists, 2017d; Pacheco, 2017). Currently, use of these two agents is not recommended because the fibrinolytic system is necessary for dissolution of widespread fibrin thromboses caused by generalized intravascular coagulation (Hunt, 2014). Discussed later (Packed Red Blood Cells), recombinant factor VIIa (rFVIIa) has been used to help control severe obstetrical hemorrhage from other causes. However, current clinical evidence is insufficient to make firm recommendations on its administration for obstetrical coagulopathies.

Specific Comorbid Conditions

Placental abruption is the most common cause of severe consumptive coagulopathy in obstetrics and is discussed more fully in Placental Abruption. Typical quantified levels of fibrin-split products with abruption are shown in Figure 41-30. With preeclampsia, eclampsia, and HELLP syndrome, endothelial activation is a hallmark and is discussed in Chapter 40 (Pathogenesis). In general, the clinical severity of preeclampsia is directly correlated with thrombocytopenia and fibrinogen-fibrin degradation products (Kenny, 2015; Levi, 2010b). As shown in Figure 41-30, intravascular coagulation is seldom severe enough to be clinically worrisome (Pritchard, 1976).

Fetal Death and Delayed Delivery

Consumptive coagulopathy associated with prolonged retention of a dead fetus is unusual today because fetal death can be easily confirmed and there are highly effective methods for labor induction. With singleton pregnancies, if the dead fetus is undelivered, most women enter spontaneous labor within 2 weeks. Gross disruption of maternal coagulation rarely develops before 4 weeks (Pritchard, 1959, 1973). After 1 month, however, almost a fourth will develop consumptive coagulopathy.

Obvious coagulation derangement occasionally develops in a multifetal pregnancy in which one fetus dies while the other survives (Chescheir, 1988; Landy, 1989). This situation is uncommon, and in one study of 22 such pregnancies, none developed a coagulopathy (Petersen, 1999). Most cases are seen in monochorionic twins with shared circulations, which are described in Chapter 45 (Monochorionic Twins and Vascular Anastomoses).

Amnionic Fluid Embolism

The classic triad of abrupt hemodynamic and respiratory compromise along with DIC underpins its diagnosis (Clark, 2016). Most reports describe a frequency of 1 in 40,000 to 1 in 50,000 (Clark, 2014; Knight, 2010; Kramer, 2012). The case-fatality rate in all of these studies ranges from 11 to 43 percent. From another perspective, amnionic fluid embolism was the cause of 5 to 15 percent of all pregnancy-related deaths in the United States and Canada (Berg, 2003, 2010; Creanga, 2015; Kramer, 2012).

Predisposing conditions are rapid labor, meconium-stained fluid, and tears into uterine and other large pelvic veins that permit an exchange of fluids between the maternal and fetal compartment (Society for Maternal-Fetal Medicine, 2016). Other commonly cited risks include older maternal age; postterm pregnancy; labor induction or augmentation; eclampsia; cesarean, forceps, or vacuum delivery; placental abruption or previa; and hydramnios (Knight, 2010, 2012; Kramer, 2012). The association of uterine hypertonus appears to be the effect rather than the cause because uterine blood flow ceases when intrauterine pressures exceed 35 to 40 mm Hg. Thus, a hypertonic contraction would be the least likely circumstance for amnionic fluid and other debris to enter uterine veins (Clark, 1985). For this reason, hypertonus from oxytocin is not implicated.

Diagnosis

Proposed criteria for diagnosis of amnionic fluid embolism are shown in Table 41-7. The classic example is dramatic, and a woman in the late stages of labor or immediately postpartum begins gasping for air. Seizures or cardiorespiratory arrest rapidly follows accompanied by massive hemorrhage from consumptive coagulopathy. Clinical manifestations are variable. For example, we and others have managed several women in whom otherwise uncomplicated vaginal or cesarean delivery was followed by severe acute consumptive coagulopathy without overt cardiorespiratory difficulties. In those women, consumptive coagulopathy appears to be the forme fruste of amnionic fluid embolism (Kramer, 2012; Porter, 1996).

Because of this clinical variability, other sources of acute cardiac or respiratory failure should be considered. These include myocardial infarction, pulmonary or air embolism, high spinal blockade, eclampsia, and anaphylactic shock. In some cases, the temporal relationship of events aids diagnosis. Unfortunately, no specific diagnostic laboratory test confirms or refutes the diagnosis of amnionic fluid embolism, and it remains a clinical diagnosis. Importantly, women suffering from excessive blood loss and resulting coagulopathy may be misdiagnosed with amnionic fluid embolism, when the true culprit is unrecognized or underappreciated hemorrhage (Clark, 2016). In either event, a woman with cardiopulmonary compromise should receive immediate resuscitation (Society for Maternal-Fetal Medicine, 2016).

Pathophysiology

The mechanism of injury from amnionic fluid embolism has evolved. Early theories proposed that amnionic fluid and debris entered maternal circulation and obstructed pulmonary artery flow, which led to hypoxia, right heart failure, and death. However, during normal delivery, amnionic fluid commonly enters the maternal circulation through venous channels at the placental implantation site or from small lacerations. Accordingly, squames, fetal cells, and trophoblasts can often be identified in maternal peripheral blood at delivery (Clark, 1986; Lee, 1986). And, infused amnionic fluid is generally innocuous, even in large amounts (Adamsons, 1971; Stolte, 1967).

Current explanations describe disruption of the maternal-fetal interface, which allows material from the fetal compartment to enter maternal circulation. This leads to abnormal activation of proinflammatory mediator systems, similar to the systemic inflammatory response syndrome (SIRS), and causes initial, transient pulmonary vasoconstriction and hypertension. Acute right ventricular failure is then followed by hemodynamic collapse from right ventricular infarction coupled with interventricular septum displacement to the left and ultimately decreased left-sided cardiac output. This right and now left ventricular dysfunction is followed by cardiogenic pulmonary edema and systemic hypotension. Concurrently in this process, acute respiratory failure with severe hypoxemia from shunting develops. Notably, the resulting multiorgan dysfunction is an interrelated process, with both the cardiac and pulmonary systems affecting each other.

Women who survive beyond these first phases invariably have the third component of the classic triad—a consumptive coagulopathy. Similar to the coagulation process described earlier, the material from the fetal compartment containing tissue factor activates factor VII. This leads to the development of DIC (see Fig. 41-29).

In those who succumb, postmortem histopathological findings may be obvious (Fig. 41-31). However, detection of such material may require special stains, and even then, debris may not be seen. In one study, fetal elements were detected in 75 percent of autopsies and in 50 percent of specimens prepared from concentrated buffy coat aspirates taken antemortem from a pulmonary artery catheter (Clark, 1995).

Management

The initial period of systemic and pulmonary hypertension with amnionic fluid embolism is transient. Thus, immediate high-quality cardiopulmonary resuscitation and advanced cardiac life support must be initiated without delay (Society for Maternal-Fetal Medicine, 2016). These are discussed in detail in Chapter 47 (Cardiopulmonary Resuscitation).

If resuscitation is successful, hemodynamic instability is common in survivors. Both fever and hyperoxia will worsen ischemia–reperfusion injury to the brain, and thus both are avoided. A suitable goal for temperature is 36°C and for mean arterial pressure is 65 mm Hg (Society for Maternal-Fetal Medicine, 2016). Additional supportive care measures such as intubation are usually necessary. During the phase of right ventricular failure, inotropic agents such as dobutamine may improve right heart output, and later systemic hypotension should be treated with vasopressors such as norepinephrine. Excess fluid administration is discouraged due to risks of worsening dilation of an already engorged right ventricle, which may cause right-sided myocardial infarction and displacement of the interventricular septum.

Beginning either immediately after cardiopulmonary collapse or during the ensuing phases of injury, a coagulopathy develops in most cases from activation of factor VII and X. This may be exacerbated by ongoing hemorrhage. A common source of obstetrical bleeding is uterine atony. Therefore, immediate evaluation of coagulation parameters is prudent with concurrent clinical management of bleeding.

Clinical Outcomes

Most reports describe dismal outcomes with amnionic fluid embolism. This is likely influenced by underdiagnosis and reporting biases that favor the most severe cases with the highest mortality rates. Several reports are illustrative. From a California database of 1.1 million deliveries, the mortality rate with amnionic fluid embolism was 60 percent (Gilbert, 1999). In a report of 34 mothers from China, 90 percent died (Weiwen, 2000). Death can be amazingly rapid, and 12 of the 34 died within 30 minutes. The mortality rate was somewhat better in the largest study from Canada. Of 120 women with an amnionic fluid embolism, only a fourth died. Survivors commonly have profound neurological impairment. Clark (1995) observed that only 8 percent of women who lived despite cardiac arrest survived neurologically intact. Overall, prognosis appears to be more associated with disease severity and the attendant cardiac arrest than with any specific treatment modality (Clark, 2014).

As perhaps expected, perinatal outcomes are also poor and are inversely related to the maternal cardiac arrest-to-delivery interval. Even so, neonatal survival rate is 70 percent, but unfortunately, up to half of survivors suffer residual neurological impairment. In the Canadian study, 28 percent of infants were considered to be asphyxiated at birth (Kramer, 2012).

Sepsis Syndrome

Various infections that are accompanied by endo- or exotoxin release can lead to sepsis syndrome. Although a feature of this syndrome includes activation of coagulation, seldom does sepsis alone cause massive procoagulant consumption. Escherichia coli bacteremia is frequently seen with antepartum pyelonephritis and puerperal infections, however, accompanying consumptive coagulopathy is usually not severe. Some notable exceptions are septicemia associated with puerperal infection or septic abortion caused by exotoxins released from infecting organisms such as group A Streptococcus pyogenes, Staphylococcus aureus, or Clostridium perfringens, C sordellii, or C novyi (Herrera, 2016). Treatment of sepsis syndrome and septic shock is discussed in Chapter 47 (Sepsis Syndrome).

Purpura Fulminans

This severe—often lethal—form of consumptive coagulopathy is caused by microthrombi in small blood vessels leading to skin necrosis and sometimes vasculitis. Debridement of large areas of skin over the extremities and buttocks frequently requires treatment in a burn unit. Purpura fulminans usually complicates sepsis in women with heterozygous protein C deficiencies and low protein C serum levels (Levi, 2010b). Note that homozygous protein C or S deficiency results in fatal neonatal purpura fulminans (Chap. 52, Protein S Deficiency).

Abortion

Septic abortion—especially associated with the organisms just discussed—can incite coagulation and worsen hemorrhage, especially with midtrimester abortions. Indeed, sepsis syndrome accompanied by intravascular coagulation accounts for 25 percent of abortion-related deaths (Saraiya, 1999). In the past, especially with illegal abortions, infections with C perfringens were a frequent cause of intense intravascular hemolysis at Parkland Hospital (Pritchard, 1971). More recently, however, septic abortions from infection with C sordellii have emerged as important causes (Chap. 18, Inevitable Abortion).

Second-trimester induced abortions can stimulate intravascular coagulation even in the absence of sepsis. Ben-Ami and associates (2012) described a 1.6-percent incidence in 1249 late second-trimester pregnancies terminated by dilation and evacuation. Two thirds were done for fetal demise, which may have been contributory to coagulopathy. Another source of intense coagulation is from instillation of hypertonic solutions to effect midtrimester abortions. These are not commonly used currently for pregnancy terminations. The mechanism is thought to initiate coagulation by thromboplastin release into maternal circulation from the placenta, fetus, and decidua by the necrobiotic effect of hypertonic solutions (Burkman, 1977).

Recognition of obstetrical hemorrhage severity is crucial to its management. Visual estimation of blood loss, especially when excessive, is notoriously inaccurate, and true blood loss is often two to three times the clinical estimate. Consider also that in obstetrics, part and sometimes even all of the lost blood may be concealed. Estimation is further complicated in that peripartum hemorrhage also includes the pregnancy-induced augmented blood volume. After pregnancy hypervolemia is lost at delivery, blood loss can be estimated by calculating 500 mL loss for each 3 volume percent drop in hematocrit. The hematocrit nadir depends on the speed of resuscitation with intravenous crystalloids. With acute blood loss, the real-time hematocrit is at its maximum whenever measured in the delivery, operating, or recovery room.

A prudent rule is that any time blood loss is considered more than average, then the hematocrit is determined and plans are made for close observation for potential physiological deterioration. Urine output measured hourly is one of the most important “vital signs.” Unless diuretic agents are given—and these are seldom indicated with active bleeding—accurately measured urine flow reflects renal perfusion, which in turn reflects perfusion of other vital organs. Urine flow of at least 30 mL, and preferably ≥50 mL per hour, should be maintained.

Hypovolemic Shock

Shock from hemorrhage evolves through several stages. Early in the course of massive bleeding, mean arterial pressure, stroke volume, cardiac output, central venous pressure, and pulmonary capillary wedge pressure decline. Increases in arteriovenous oxygen content difference reflect a relative rise in tissue oxygen extraction, although overall oxygen consumption falls.

Blood flow to capillary beds in various organs is controlled by arterioles. These are resistance vessels that are partially controlled by the central nervous system. However, approximately 70 percent of total blood volume is contained in venules, which are passive resistance vessels controlled by humoral factors. Catecholamine release during hemorrhage prompts greater venular tone, which provides an autotransfusion from this capacitance reservoir (Barber, 1999). This is accompanied by compensatory rises in heart rate, systemic and pulmonary vascular resistance, and myocardial contractility. In addition, cardiac output and blood volume are redistributed from the effect of selective, centrally mediated arteriolar constriction or relaxation—autoregulation. Thus, although perfusion to the kidneys, splanchnic beds, muscles, skin, and uterus is diminished, relatively more blood flow is diverted to the heart, brain, and adrenal glands.

When the blood volume deficit exceeds approximately 25 percent, compensatory mechanisms usually are inadequate to maintain cardiac output and blood pressure. Importantly, additional small losses of blood will now cause rapid clinical deterioration. Following an initial augmented total oxygen extraction by maternal tissue, maldistribution of blood flow results in local tissue hypoxia and metabolic acidosis. This creates a vicious cycle of vasoconstriction, organ ischemia, and cellular death.

Another important clinical effect of hemorrhage is activation of lymphocytes and monocytes, which in turn causes endothelial cell activation and platelet aggregation. These promote release of vasoactive mediators that occlude small vessels and further impair microcirculatory perfusion. Other common obstetrical syndromes—preeclampsia and sepsis—also lead to loss of capillary endothelial integrity, additional loss of intravascular volume into the extracellular space, and platelet aggregation. These then can incite DIC.

The pathophysiological events just described create important but often overlooked extracellular fluid and electrolyte shifts involved in both the genesis and successful treatment of hypovolemic shock. These include changes in the cellular transport of various ions such as sodium and water into skeletal muscle as well as potassium loss. Replacement of extracellular fluid and intravascular volume are both necessary. Survival is enhanced in acute hemorrhagic shock if blood plus crystalloid solution is given compared with blood transfusions alone.

Fluid Resuscitation

Whenever excessive blood loss is suspected in a pregnant woman, steps are simultaneously taken to identify the bleeding source and to begin resuscitation. If she is undelivered, restoration of blood volume is beneficial to mother and fetus, and it also prepares for emergent delivery. If she is postpartum, it is essential to immediately identify uterine atony, retained placental fragments, or genital tract lacerations. At least one and preferably more large-bore intravenous infusion systems are established promptly with rapid administration of crystalloid solutions, while blood is made available. An operating room is readied, and a surgical and anesthesia team are assembled immediately. Specific management of hemorrhage is further dependent on its etiology.

It cannot be overemphasized that treatment of serious hemorrhage demands prompt and adequate refilling of the intravascular compartment with crystalloid solutions. These rapidly equilibrate into the extravascular space, and only 20 percent of crystalloid remains intravascularly in critically ill patients after 1 hour (Zuckerbraun, 2010). Because of this, initial fluid is infused in a volume two to three times the estimated blood loss.

Resuscitation of hypovolemic shock with colloid versus crystalloid solutions has been debated. In a Cochrane review of resuscitation of nonpregnant critically ill patients, Perel and coworkers (2013) found equivalent benefits but concluded that colloid solutions were more expensive. Similar results were found in the Saline versus Albumin Fluid Evaluation (SAFE) randomized trial of almost 7000 nonpregnant patients (Finfer, 2004). We concur with Zuckerbraun and colleagues (2010) that acute volume resuscitation is preferably done with crystalloid and blood.

Blood Replacement

The hematocrit level or hemoglobin concentration that mandates blood transfusion is controversial. Cardiac output does not substantively drop until the hemoglobin concentration falls to approximately 7 g/dL or hematocrit of 20 volume percent. At this level, several organizations recommend consideration for red cell transfusions (Carson, 2017). Also, Military Combat Trauma Units in Iraq used a target hematocrit of 21 volume percent (Barbieri, 2007). In general, with ongoing obstetrical hemorrhage, we recommend rapid blood infusion when the hematocrit is <25 volume percent. This decision is dependent on whether the fetus has been delivered; surgery is imminent or ongoing operative blood loss is expected; or acute hypoxia, vascular collapse, or other factors are present.

Scant clinical data elucidate these issues. In a study from the Canadian Critical Care Trials Group, nonpregnant patients were randomly assigned to restrictive red cell transfusions to maintain hemoglobin concentration >7 g/dL or to liberal transfusions to maintain the hemoglobin level at 10 to 12 g/dL. The 30-day mortality rate was similar—19 versus 23 percent in the restrictive versus liberal groups, respectively (Hébert, 1999). Transfusion therapy in nonpregnant patients with septic shock had similar mortality rates when 7 g/dL was compared with 9 g/dL as targets for transfusions (Holst, 2014). The number of units transfused in a given woman to reach a target hematocrit depends on her body mass and on expectations of additional blood loss.

Blood Component Products

Contents and effects of transfusion of various blood components are shown in Table 41-8. Compatible whole blood is ideal for treatment of hypovolemia from catastrophic hemorrhage. It has a shelf life of 40 days, and 70 percent of the transfused red cells function for at least 24 hours following transfusion. One unit raises the hematocrit by 3 to 4 volume percent. Important for obstetrical hemorrhage, whole blood replaces many coagulation factors in obstetrics—especially fibrinogen—and its plasma treats hypovolemia. A collateral derivative is that women with severe hemorrhage are resuscitated with fewer blood donor exposures than with packed red cells and components (Shaz, 2009).

Evidence supports the preferable use of whole blood for massive hemorrhage, including our experiences at Parkland Hospital (Alexander, 2009; Hernandez, 2012). Of more than 66,000 deliveries, women with obstetrical hemorrhage treated with whole blood had significantly lower incidences of renal failure, acute respiratory distress syndrome, pulmonary edema, hypofibrinogenemia, intensive care unit admissions, and maternal death compared with those given packed red cells and component therapy. Freshly donated whole blood has also been used successfully for life-threatening massive hemorrhage at combat support hospitals (Murdock, 2014; Stubbs, 2016).

In most institutions today, however, whole blood is rarely available. Thus, most women with obstetrical hemorrhage and ongoing massive blood loss are given packed red cells and crystalloid. In these instances, no data support a 1:1 plasma: red cell transfusion ratio. As subsequently discussed, many institutions use massive transfusion protocols designed to anticipate all facets of massive obstetrical hemorrhage. These “recipes” commonly contain a combination of red cells, plasma, cryoprecipitate, and platelets (Cunningham, 2015; Pacheco, 2011; Shields, 2011).

Several studies have assessed plasma:red cell ratio with massive transfusion protocols used in civilian trauma units and military combat hospitals (Borgman, 2007; Gonzalez, 2007; Hardin, 2014; Johansson, 2007). Patients undergoing massive transfusion—defined as 10 or more units of blood—had much higher survival rates as the ratio of plasma to red cell units neared 1:1.4, that is, one unit of plasma given for each 1.4 units of packed red cells. By way of contrast, the highest mortality group had a ratio of 1:8. Most of these studies found that component replacement is rarely necessary with acute replacement of 5 to 10 units of packed red cells.

From the foregoing, when red cell replacement exceeds five units or so, evaluation of platelet count, clotting studies, and plasma fibrinogen concentration is reasonable. In the woman with obstetrical hemorrhage, the platelet count should be maintained >50,000/μL by the infusion of platelet concentrates. A fibrinogen level <150 mg/dL or a sufficiently prolonged PT or PTT in a woman with surgical bleeding is an indication for replacement. Fresh-frozen plasma is administered in doses of 10 to 15 mL/kg, or alternatively, cryoprecipitate is infused (see Table 41-8).

Dilutional Coagulopathy

A major drawback of treatment for massive hemorrhage with crystalloid solutions and packed red blood cells is depletion of platelets and clotting factors. This can lead to a dilutional coagulopathy that is clinically indistinguishable from DIC (Hossain, 2013).

Thrombocytopenia is the most frequent coagulation defect found with blood loss and multiple transfusions (Counts, 1979). In addition, packed red cells have only very small amounts of soluble clotting factors, and stored whole blood is deficient in platelets and in factors V, VIII, and XI. As discussed, massive replacement with red cells only and without factor replacement can also cause hypofibrinogenemia and prolongation of the PT and PTT. Because many causes of obstetrical hemorrhage also cause consumptive coagulopathy, the distinction between dilutional and consumptive coagulopathy can be confusing. Fortunately, treatment for both is similar.

Type and Screen versus Crossmatch

A blood type and antibody screen should be performed for any woman at significant risk for hemorrhage. Screening involves mixing maternal serum with standard reagent red cells that carry antigens to which most of the common clinically significant antibodies react. Crossmatching involves the use of actual donor erythrocytes rather than the standardized red cells. This process is efficient, and only 0.03 to 0.07 percent of patients identified as having no antibodies are subsequently found to have antibodies (Boral, 1979). Importantly, administration of screened blood rarely results in adverse clinical sequelae.

Packed Red Blood Cells

One unit of packed erythrocytes is derived from one unit of whole blood to have a hematocrit of 55 to 80 volume percent. One unit will increase the hematocrit by 3 to 4 volume percent.

Platelets

With surgical delivery or with lacerations, platelet transfusions are considered with ongoing obstetrical hemorrhage when the platelet count falls below 50,000/μL (Kenny, 2015). In the nonsurgical patient, bleeding is rarely encountered if the platelet count is 10,000/μL or higher (Murphy, 2010). The preferable source of platelets is one “bag” obtained by single-donor apheresis. This contains the equivalent of six units from six individual donors. Depending on maternal size, each single-donor apheresis six-unit bag raises the platelet count by approximately 20,000/μL (Schlicter, 2010). If these bags are not available, then individual-donor platelet units are used, and six to eight such units are generally transfused one at a time.

Importantly, the donor plasma in platelet units must be compatible with recipient erythrocytes. Further, because some red blood cells are invariably transfused along with the platelets, only units from D-negative donors should be given to D-negative recipients. If it is necessary to give these, however, adverse sequelae are unlikely (Lin, 2002).

Fresh-Frozen Plasma

This component is prepared by separating plasma from whole blood and then freezing it. Approximately 30 minutes are required for frozen plasma to thaw. It is a source of all stable and labile clotting factors, including fibrinogen. Thus, it is often used for treatment of women with consumptive or dilutional coagulopathy. Plasma is not appropriate for use as a volume expander in the absence of specific clotting factor deficiencies. It should be considered in a bleeding woman with a fibrinogen level <150 mg/dL or with an abnormal PT or PTT.

An alternative to frozen plasma is liquid plasma (LQP). This never-frozen plasma is stored at 1 to 6°C for up to 26 days, and in vitro, it appears to be superior to thawed plasma (Matijevic, 2013).

Cryoprecipitate and Fibrinogen Concentrate

Each unit of cryoprecipitate is prepared from one unit of fresh-frozen plasma. Each 10- to 15-mL unit contains at least 200 mg of fibrinogen along with factor VIII:C, factor VIII:von Willebrand factor, factor XIII, and fibronectin (American Association of Blood Banks, 2014). It is usually given as a “pool” or “bag” using an aliquot of fibrinogen concentrate taken from 8 to 120 donors. Cryoprecipitate is an ideal source of fibrinogen when levels are dangerously low and there is oozing from surgical incisions. Another alternative is virus-inactivated fibrinogen concentrate. Each gram of this raises the plasma fibrinogen level approximately 40 mg/dL (Ahmed, 2012; Kikuchi, 2013).

Recombinant Activated Factor VII

This synthetic vitamin K-dependent protein is available as NovoSeven. It binds to exposed tissue factor at the site of injury to generate thrombin that activates platelets and the coagulation cascade. Since its introduction, rFVIIa has been used to help control hemorrhage from surgery, trauma, and obstetrical causes (Goodnough, 2016; Murakami, 2015). Most Level I trauma centers include it in their massive transfusion protocols, and it is included in the one used at Parkland Hospital. Importantly, rFVIIa will not be effective if the plasma fibrinogen level is <50 mg/dL or the platelet count is <30,000/μL.

One major concern with rFVIIa use is arterial—and to a lesser degree venous—thrombosis. In a review of 35 randomized trials with nearly 4500 subjects, arterial thromboembolism developed in 55 percent (Levi, 2010a). A second concern is that it was found to be only marginally effective (Pacheco, 2011).

Tranexamic Acid

This antifibrinolytic drug has been used for traumatic and obstetrical hemorrhage. Tranexamic acid inhibits clot lysis to help forestall bleeding by preventing plasmin from degrading fibrin. Its use has been associated with a higher incidence of renal cortical necrosis (Frimat, 2016). The evidence supporting its use as an adjunct in obstetrical hemorrhage is limited, and its routine use for prophylaxis is not recommended American College of Obstetricians and Gynecologists, 2017d; Pacheco, 2017).

Massive Transfusion Protocols

These function to speed blood product delivery to the bedside or operating room, which permits product infusion early in the resuscitation process. The rationale is to prevent adverse effects of aggressive resuscitation solely with crystalloid and packed red blood cells. That said, it is not necessary to activate massive transfusions until at least four to five units of red cells have been given within 2 hours or so. Once activated, red cells, plasma, platelets, and fibrinogen are given by protocol in amounts shown in Table 41-9. Some protocols include rFVIIa and others include tranexamic acid.

As expected, studies attesting to the superiority for survival with massive transfusion protocols are limited. Most reports describe nonpregnant trauma victims, but some observational studies address obstetrical hemorrhage (Green, 2016; Pacheco, 2016). More data with use of these protocols is needed.

Viscoelastic Assays

Thromboelastography (TEG) and rotational thromboelastometry (ROTEM) are point-of-care tests that assess coagulation in whole blood during massive transfusions. These tests work by analyzing both clot formation and breakdown in a whole blood sample from a given patient. Testing produces a profile of coagulation dynamics, and displayed values indicate the speed and quality of clot formation (Fig. 41-32). These assays provide information regarding time to clot formation, clot strength, and fibrinolysis. Currently, they guide blood product replacement in trauma, liver transplant, and cardiac surgery patients. Studies of TEG and ROTEM techniques in pregnant women have confirmed the hypercoagulable state of pregnancy and provide reference ranges for use in this population (Butwick, 2015; de Lange, 2014; Solomon, 2012).

Although these point-of-care tests appear promising, they also have several limitations. For example, they cannot be used to detect disorders of primary hemostasis (Solomon, 2012). Additionally, these tests cannot diagnose coagulopathies stemming from platelet dysfunction or antiplatelet drugs. A major drawback is the risk of misinterpretation when tests are used by inadequately trained personnel. Further study is necessary before these tests are widely applied for treatment of obstetrical hemorrhage.

Topical Hemostatic Agents

Several agents can be used to control persistent surgical oozing. These were recently reviewed by Miller and colleagues (2015). Other than for cesarean hysterectomy, these are seldom used in obstetrical hemorrhage.

Cell Salvage and Autologous Transfusion

Preoperative patient phlebotomy and autologous blood storage for transfusion has been disappointing. Exceptions are women with a rare blood type or with unusual antibodies. Most have concluded that autologous transfusions are not cost effective (Etchason, 1995; Pacheco, 2011, 2013).

Intraoperative blood salvage with reinfusion is considered to be a safe intervention in obstetrical patients. As discussed in Chapter 30 (Patient Preparation), this practice may be helpful for women declining transfusion. Prior concern centered on amnionic fluid contamination and embolism (Dhariwal, 2014; Goucher, 2015; Pacheco, 2011). A recent randomized trial involving 3028 women compared routine cell salvage use against routine care, in which salvage was employed only for bleeding indications. The rate of nonautologous donor blood transfusion was reduced in the cell salvage group—2.5 versus 3.5 percent, but this was not a significant difference (Khan, 2017). Similar to prior reports, no cases of amnionic fluid embolism were reported.

Transfusion Complications

Of serious known risks, transfusion of an incompatible blood component may result in acute hemolysis. If severe, this can cause DIC, acute kidney injury, and death. Preventable errors responsible for most of such reactions frequently include mislabeling of a specimen or incorrectly transfusing a patient not slated for those products. The rate of such errors in the United States is estimated to be 1 in 14,000 units, but these events are likely underreported (Lerner, 2010). A transfusion reaction is characterized by fever, hypotension, tachycardia, dyspnea, chest or back pain, flushing, severe anxiety, and hemoglobinuria. Immediate supportive measures include stopping the transfusion, treating hypotension and hyperkalemia, provoking diuresis, and alkalinizing the urine.

Transfusion-related acute lung injury (TRALI) is the most common cause of transfusion-related mortality. The syndrome is characterized by severe dyspnea, hypoxia, and noncardiogenic pulmonary edema that develop within 6 hours of transfusion (Peters, 2015). TRALI is estimated to complicate at least 1 in 12,000 transfusions (Carson, 2017). Although the pathogenesis is incompletely understood, injury to the pulmonary capillaries may arise from anti-human leukocyte antigen (HLA) and neutrophil (HNA) antibodies in donor plasma (Lerner, 2010). A delayed form of TRALI has been reported to begin 6 to 72 hours following transfusion (Marik, 2008). Management is supportive and may include mechanical ventilation (Chap. 47, Clinical Course).

Bacterial infection from transfusion of a contaminated blood component is unusual because organism growth is discouraged by refrigeration. The most often implicated contaminants of red cells include Yersinia, Pseudomonas, Serratia, Acinetobacter, and Escherichia species. The more important risk is from bacterial contamination of platelets, which are stored at room temperature. Current estimates are that 1 in 1000 to 2000 platelet units are contaminated. Death from transfusion-related sepsis is 1 per 17,000 for single-donor platelets and 1 per 61,000 for apheresis-donor packs (Lerner, 2010).

Viral infection risks from transfusion have been curtailed. The risk of HIV or hepatitis C virus infection in screened blood is estimated to be 1 case per 1 to 2 million units transfused (Carson, 2017; Stramer, 2004). The risk for HIV-2 infection is less. Other viral infections include hepatitis B transmission, which is estimated to be <1 per 100,000 transfused units (Jackson, 2003). Because of its high prevalence, cytomegalovirus-infected leukocytes are often transfused. Thus, precautions are taken for immunosuppressed recipients, keeping in mind that this includes the fetus.

Also, risks for transmitting West Nile virus, human T-lymphotropic virus type I, parvovirus B19, and toxoplasmosis are slight (American Association of Blood Banks, 2013; Foroutan-Rad, 2016). Finally, Zika virus has emerged as another relevant transfusion-transmitted infection (Motta, 2016). The Food and Drug Administration (2016) revised recommendations for collection of all whole blood components to include testing for Zika virus. This practice has been affirmed by the Centers for Disease Control and Prevention (2016).

Adjunctive Surgical Procedures

Several invasive procedures can help arrest postpartum hemorrhage. A report from the Agency for Healthcare Research and Quality concluded that most studies addressing these methods are of poor quality (Likis, 2015). In one study of 6660 women with postpartum hemorrhage, 4.4 percent underwent an invasive procedure, and 1.1 percent had a hysterectomy (Kayem, 2016). The failure rate of conservative measures was 15 percent in surgical and embolization procedures.

Uterine Artery Ligation

The technique for unilateral or bilateral uterine artery ligation is used primarily for lacerations at the lateral part of a hysterotomy incision (Fig. 41-33). In our experiences, this procedure is less helpful for hemorrhage from uterine atony.

Uterine Compression Sutures

This surgical technique uses a no. 2 chromic suture to compress the anterior and posterior uterine walls together (B-Lynch, 1997). Because they give the appearance of suspenders, they are also called braces (Fig. 41-34). Several modifications of the B-Lynch technique have been described (Cho, 2000; Hayman, 2002; Matsubara, 2013; Nelson, 2007). Indications vary for its application, and this will affect the success rate. For example, B-Lynch (2005) cited 948 cases with only seven failures. Conversely, Kayem and associates (2011) described 211 women who had an overall failure rate of 25 percent, which did not differ between B-Lynch sutures and their modifications. In another series, the failure rate was 20 percent (Kaya, 2016). From their review, Sathe and coworkers (2016) reached similar conclusions.

Some unique complications can rarely follow compression sutures (Matsubara, 2013). Most involve variations of uterine ischemic necrosis with peritonitis (Gottlieb, 2008; Joshi, 2004; Ochoa, 2002; Treloar, 2006). In one case, total uterine necrosis followed B-Lynch sutures that were placed in combination with bilateral ligation of uterine, uteroovarian, and round ligament arteries (Friederich, 2007). In most cases, subsequent pregnancies are uneventful if compression sutures are used (An, 2013). A few women, however, with B-Lynch or Cho sutures developed uterine wall defects (Akoury, 2008). Another long-term complication is uterine cavity synechiae (Alouini, 2011; Ibrahim, 2013; Poujade, 2011).

Internal Iliac Artery Ligation

For years, ligation of one or both internal iliac arteries has been used to reduce pelvic hemorrhage. Drawbacks are that the procedure may be technically difficult and is only successful half of the time (American College of Obstetricians and Gynecologists, 2017d). It is not particularly helpful for abating hemorrhage with postpartum atony (Clark, 1985).

For ligation, adequate exposure is obtained by opening the peritoneum over the common iliac artery and dissecting down to the bifurcation of the external and internal iliac arteries (Fig. 41-35). Branches distal to the external iliac arteries are palpated to verify pulsations at or below the inguinal area. Ligation of the internal iliac artery 5 cm distal to the common iliac bifurcation will usually avoid the posterior division branches (Bleich, 2007). The areolar sheath of the artery is incised longitudinally, and a right-angle clamp is carefully passed just beneath the artery from lateral to medial. Care must be taken not to perforate contiguous large veins, especially the internal iliac vein. Suture—usually nonabsorbable—is passed under the artery with a clamp, and the vessel is then securely ligated.

The most important mechanism of action with internal iliac artery ligation is an 85-percent reduction in pulse pressure in those arteries distal to the ligation (Burchell, 1968). This converts an arterial pressure system into one with pressures approaching those in the venous circulation. This creates vessels more amenable to hemostasis via pressure and clot formation.

Even bilateral internal iliac artery ligation does not appear to interfere with subsequent reproduction. Nizard and colleagues (2003) reported follow-up in 17 women who had bilateral artery ligation. From a total of 21 pregnancies, 13 were normal, three ended with miscarriage, three were terminated, and two were ectopic.

Angiographic Embolization

This modality is now used for many causes of intractable hemorrhage when surgical access is difficult. In more than 500 women reported, embolization was 90-percent effective (Grönvall, 2014; Lee, 2012; Poujade, 2012; Zhang, 2015). After his review, Rouse (2013) concluded that embolization can be used to arrest refractory postpartum hemorrhage. Other reports have been less enthusiastic. Fertility is not impaired, and many subsequent successful pregnancies have been reported (Chauleur, 2008; Fiori, 2009; Kolomeyevskaya, 2009). An important caveat for these procedures is that women with hemodynamic instability related to active bleeding should not be removed from the operating room.

Complications of embolization are relatively uncommon but can be severe. Case reports detail instances of iatrogenic iliac artery rupture, uterine ischemic necrosis, and uterine infection (Grönvall, 2014; Katakam, 2009; Nakash, 2012). Finally, Al-Thunyan and coworkers (2012) described a woman with massive buttock necrosis and paraplegia following bilateral internal iliac artery embolization.

In a few instances, massive blood loss and difficult surgical dissection is anticipated. The use of balloon-tipped catheters preoperatively inserted into the iliac or uterine arteries was described earlier in management of placenta accrete syndromes (Management).

Pelvic Packing

For significant bleeding refractory to suture or topical hemostats, pelvic packing with gauze and termination of the operation may be considered. Rolls of gauze are packed to provide constant local pressure. This may serve as a temporizing step prior to interventional embolization. In other cases, packing alone may be left for 24 to 48 hours. If the patient is stable and bleeding appears to have stopped, packing is removed.

The umbrella or parachute pack uses a similar concept (Logothetopulos, 1926). Although seldom used today, it can be lifesaving if all other measures have failed, especially in low-resource areas (Dildy, 2006; Howard, 2002). The pack is constructed of a sturdy sterile plastic bag that is filled with gauze rolls that are unwound and knotted together. Sufficient rolls are used to provide enough volume to fill the pelvis. The pack is introduced transabdominally with the stalk exiting the vagina. Mild traction is applied by tying the stalk to a 1-liter fluid bag, which is hung over the foot of the bed. The umbrella pack is removed vaginally after 24 hours.