Chapter 37: Puerperal Complications

Puerperal complications include many of those encountered during pregnancy, but there are some that are more common at this time. Typical of these is puerperal pelvic infection—a well-known killer of postpartum women. Other infections include mastitis and breast abscesses. Thromboembolism during the short 6-week puerperium is as frequent as during all 40 antepartum weeks.

Traditionally, the term puerperal infection describes any bacterial infection of the genital tract after delivery. These infections as well as preeclampsia and obstetrical hemorrhage form the lethal triad of maternal death causes before and during the 20th century. Fortunately, because of effective antimicrobials, maternal mortality from infection has become uncommon. Berg and associates (2010) reported results from the Pregnancy Mortality Surveillance System, which contained 4693 pregnancy-related maternal deaths in the United States from 1998 through 2005. Infection caused 10.7 percent of pregnancy-related deaths and was the fifth leading cause. In a similar analysis of the North Carolina population from 1991 through 1999, Berg and colleagues (2005) reported that 40 percent of infection-related maternal deaths were preventable.

Puerperal Fever

A number of factors can cause fever—a temperature of 38.0°C (100.4°F) or higher—in the puerperium. Most persistent fevers after childbirth are caused by genital tract infection. Using this conservative definition of fever, Filker and Monif (1979) reported that only about 20 percent of women febrile within the first 24 hours after giving birth vaginally were subsequently diagnosed with pelvic infection. This was in contrast to 70 percent of those undergoing cesarean delivery. It must be emphasized that spiking fevers of 39°C or higher that develop within the first 24 hours postpartum may be associated with virulent pelvic infection caused by group A streptococcus and is discussed on Uterine Infection.

Other causes of puerperal fever include breast engorgement, urinary infections, episiotomy and abdominal incisions, perineal lacerations, and respiratory complications after cesarean delivery (Maharaj, 2007). Approximately 15 percent of women who do not breast feed develop postpartum fever from breast engorgement. As discussed in Chapter 36 (Maternal Care During the Puerperium), the incidence of fever is lower in breast-feeding women. “Breast fever” rarely exceeds 39°C in the first few postpartum days and usually lasts < 24 hours. Urinary infections are uncommon postpartum because of the normal diuresis encountered then. That said, acute pyelonephritis has a variable clinical picture. The first sign of renal infection may be fever, followed later by costovertebral angle tenderness, nausea, and vomiting. Atelectasis following abdominal delivery is caused by hypoventilation and is best prevented by coughing and deep breathing on a fixed schedule following surgery. Fever associated with atelectasis is thought to follow infection by normal flora that proliferate distal to obstructing mucus plugs.

Uterine Infection

Postpartum uterine infection or puerperal sepsis has been called variously endometritis, endomyometritis, and endoparametritis. Because infection involves not only the decidua but also the myometrium and parametrial tissues, we prefer the inclusive term metritis with pelvic cellulitis.

Predisposing Factors

The route of delivery is the single most significant risk factor for the development of uterine infection (Burrows, 2004; Conroy, 2012; Koroukian, 2004). In the French Confidential Enquiry on Maternal Deaths, Deneux-Tharaux and coworkers (2006) cited a nearly 25-fold increased infection-related mortality rate with cesarean versus vaginal delivery. Rehospitalization rates for wound complications and endometritis were increased significantly in women undergoing a planned primary cesarean delivery compared with those having a planned vaginal birth (Declercq, 2007).

Vaginal Delivery

Compared with cesarean delivery, metritis following vaginal delivery is relatively infrequent. Women delivered vaginally at Parkland Hospital have a 1‑ to 2-percent incidence of metritis. Women at high risk for infection because of membrane rupture, prolonged labor, and multiple cervical examinations have a 5-to 6-percent frequency of metritis after vaginal delivery. If there is intrapartum chorioamnionitis, the risk of persistent uterine infection increases to 13 percent (Maberry, 1991). Finally, in one study, manual removal of the placenta, discussed in Chapter 41 (Causes of Obstetrical Hemorrhage), increased the puerperal metritis rate threefold (Baksu, 2005).

Cesarean Delivery

Single-dose perioperative antimicrobial prophylaxis is recommended for all women undergoing cesarean delivery (American College of Obstetricians and Gynecologists, 2011). Such single-dose antimicrobial prophylaxis has done more to decrease the incidence and severity of postcesarean delivery infections than any other intervention in the past 30 years. Such practices decrease the puerperal pelvic infection risk by 65 to 75 percent (Smaill, 2010).

The magnitude of the risk is exemplified from reports that predate antimicrobial prophylaxis. Cunningham and associates (1978) described an overall incidence of 50 percent in women undergoing cesarean delivery at Parkland Hospital. Important risk factors for infection following surgery included prolonged labor, membrane rupture, multiple cervical examinations, and internal fetal monitoring. Women with all these factors who were not given perioperative prophylaxis had a 90-percent serious postcesarean delivery pelvic infection rate (DePalma, 1982).

Other Risk Factors

It is generally accepted that pelvic infection is more frequent in women of lower socioeconomic status (Maharaj, 2007). Except in extreme cases usually not seen in this country, it is unlikely that anemia or poor nutrition predispose to infection. Bacterial colonization of the lower genital tract with certain microorganisms—for example, group B streptococcus, Chlamydia trachomatis, Mycoplasma hominis, Ureaplasma urealyticum, and Gardnerella vaginalis—has been associated with an increased postpartum infection risk (Andrews, 1995; Jacobsson, 2002; Watts, 1990). Other factors associated with an increased infection risk include general anesthesia, cesarean delivery for multifetal gestation, young maternal age and nulliparity, prolonged labor induction, obesity, and meconium-stained amnionic fluid (Acosta, 2012; Jazayeri, 2002; Kabiru, 2004; Leth, 2011; Siriwachirachai, 2010; Tsai, 2011).

Microbiology

Most female pelvic infections are caused by bacteria indigenous to the genital tract. Over the past 20 years, there have been reports of group A β-hemolytic streptococcus causing toxic shock-like syndrome and life-threatening infection (Aronoff, 2008; Castagnola, 2008; Nathan, 1994; Palep-Singh, 2007). Prematurely ruptured membranes are a prominent risk factor in these infections (Anteby, 1999). In reviews by Crum (2002) and Udagawa (1999) and their colleagues, women in whom group A streptococcal infection was manifested before, during, or within 12 hours of delivery had a maternal mortality rate of almost 90 percent and fetal mortality rate > 50 percent. In the past 10 years, skin and soft-tissue infections due to community-acquired methicillin-resistant Staphylococcus aureus—CA-MRSA—have become common (Chap. 64, Methicillin-Resistant Staphylococcus Aureus). Although this variant is not a frequent cause of puerperal metritis, it is commonly implicated in abdominal incisional infections (Anderson, 2007; Patel, 2007). Rotas and coworkers (2007) reported a woman with episiotomy cellulitis from CA-MRSA and hematogenously spread necrotizing pneumonia.

Common Pathogens

Bacteria commonly responsible for female genital tract infections are listed in Table 37-1. Generally, infections are polymicrobial, which enhances bacterial synergy. Other factors that promote virulence are hematomas and devitalized tissue. Although the cervix and vagina routinely harbor such bacteria, the uterine cavity is usually sterile before rupture of the amnionic sac. As the consequence of labor and delivery and associated manipulations, the amnionic fluid and uterus become contaminated with anaerobic and aerobic bacteria (Fig. 37-1). Intraamnionic cytokines and C-reactive protein are also markers of infection (Combs, 2013; Marchocki, 2013). In studies done before the use of antimicrobial prophylaxis, Gilstrap and Cunningham (1979) cultured amnionic fluid obtained at cesarean delivery in women in labor with membranes ruptured more than 6 hours. All had bacterial growth, and an average of 2.5 organisms was identified from each specimen. Anaerobic and aerobic organisms were identified in 63 percent, anaerobes alone in 30 percent, and aerobes alone in only 7 percent. Anaerobes included Peptostreptococcus and Peptococcus species in 45 percent, Bacteroides species in 9 percent, and Clostridium species in 3 percent. Aerobes included Enterococcus in 14 percent, group B streptococcus in 8 percent, and Escherichia coli in 9 percent of isolates. Sherman and coworkers (1999) later showed that bacterial isolates at cesarean delivery correlated with those taken from women with metritis at 3 days postpartum.

The role of other organisms in the etiology of these infections is unclear. Chlamydial infections have been implicated in late-onset, indolent metritis (Ismail, 1985). Observations of Chaim and colleagues (2003) suggest that when cervical colonization of U urealyticum is heavy, it may contribute to the development of metritis. Finally, Jacobsson and associates (2002) reported a threefold risk of puerperal infection in a group of Swedish women in whom bacterial vaginosis was identified in early pregnancy (Chap. 65, Vaginitis).

Bacterial Cultures

Routine pretreatment genital tract cultures are of little clinical use and add significant costs. Similarly, routine blood cultures seldom modify care. In two earlier studies done before perioperative prophylaxis was used, blood cultures were positive in 13 percent of women with postcesarean metritis at Parkland Hospital and 24 percent in those at Los Angeles County Hospital (Cunningham, 1978; DiZerega, 1979). In a later Finnish study, Kankuri and associates (2003) confirmed bacteremia in only 5 percent of almost 800 women with puerperal sepsis.

Pathogenesis and Clinical Course

Puerperal infection following vaginal delivery primarily involves the placental implantation site, decidua and adjacent myometrium, or cervicovaginal lacerations. The pathogenesis of uterine infection following cesarean delivery is that of an infected surgical incision (see Fig. 37-1). Bacteria that colonize the cervix and vagina gain access to amnionic fluid during labor. Postpartum, they invade devitalized uterine tissue. Parametrial cellulitis next follows with infection of the pelvic retroperitoneal fibroareolar connective tissue. With early treatment, infection is contained within the parametrial and paravaginal tissue, but it may extend deeply into the pelvis.

Fever is the most important criterion for the diagnosis of postpartum metritis. Intuitively, the degree of fever is believed proportional to the extent of infection and sepsis syndrome. Temperatures commonly are 38 to 39°C. Chills that accompany fever suggest bacteremia or endotoxemia. Women usually complain of abdominal pain, and parametrial tenderness is elicited on abdominal and bimanual examination. Leukocytosis may range from 15,000 to 30,000 cells/μL, but recall that cesarean delivery itself increases the leukocyte count (Hartmann, 2000). Although an offensive odor may develop, many women have foul-smelling lochia without evidence for infection, and vice versa. Some other infections, notably those caused by group A β-hemolytic streptococci, may be associated with scant, odorless lochia.

Treatment

If nonsevere metritis develops following vaginal delivery, then treatment with an oral antimicrobial agent is usually sufficient. For moderate to severe infections, however, intravenous therapy with a broad-spectrum antimicrobial regimen is indicated. Improvement follows in 48 to 72 hours in nearly 90 percent of women treated with one of several regimens. Persistent fever after this interval mandates a careful search for causes of refractory pelvic infection. These include a parametrial phlegmon—an area of intense cellulitis; an abdominal incisional or pelvic abscess or infected hematoma; and septic pelvic thrombophlebitis. In our experience, persistent fever is seldom due to antimicrobial-resistant bacteria or due to drug side effects. The woman may be discharged home after she has been afebrile for at least 24 hours, and further oral antimicrobial therapy is not needed (Dinsmoor, 1991; French, 2004).

Choice of Antimicrobials

Although therapy is empirical, initial treatment following cesarean delivery is directed against elements of the mixed flora shown in Table 37-1. For infections following vaginal delivery, as many as 90 percent of women respond to regimens such as ampicillin plus gentamicin. In contrast, anaerobic coverage is included for infections following cesarean delivery (Table 37-2).

In 1979, DiZerega and colleagues compared the effectiveness of clindamycin plus gentamicin with that of penicillin G plus gentamicin for treatment of pelvic infections following cesarean delivery. Women given the clindamycin-gentamicin regimen had a 95-percent response rate, and this regimen is still considered by most to be the standard by which others are measured (French, 2004). Because enterococcal cultures may be persistently positive despite this standard therapy, some add ampicillin to the clindamycin-gentamicin regimen, either initially or if there is no response by 48 to 72 hours (Brumfield, 2000).

Many authorities recommend that serum gentamicin levels be periodically monitored. At Parkland Hospital, we do not routinely do so if the woman has normal renal function. Once-daily dosing versus multiple-dosing with gentamicin provides adequate serum levels, and either method has similar cure rates (Livingston, 2003).

Because of potential nephrotoxicity and ototoxicity with gentamicin in the event of diminished glomerular filtration, some have recommended a combination of clindamycin and a second-generation cephalosporin to treat such women. Others recommend a combination of clindamycin and aztreonam, which is a monobactam compound with activity similar to the aminoglycosides.

The spectra of β-lactam antimicrobials include activity against many anaerobic pathogens. Some examples include cephalosporins such as cefoxitin, cefotetan, cefotaxime, and ceftriaxone, as well as extended-spectrum penicillins such as piperacillin, ticarcillin, and mezlocillin. β-Lactam antimicrobials are inherently safe and except for allergic reactions, are free of major toxicity. The β-lactamase inhibitors, clavulanic acid, sulbactam, and tazobactam, have been combined with ampicillin, amoxicillin, ticarcillin, and piperacillin to extend their spectra. Metronidazole has superior in vitro activity against most anaerobes. This agent given with ampicillin and an aminoglycoside provides coverage against most organisms encountered in serious pelvic infections. It is also used to treat Clostridium difficile colitis.

Imipenem and similar antimicrobials are in the carbapenem family. These offer broad-spectrum coverage against most organisms associated with metritis. Imipenem is used in combination with cilastatin, which inhibits its renal metabolism. Preliminary findings with ertapenem indicated suboptimal outcomes (Brown, 2012). It seems reasonable from both a medical and an economic standpoint to reserve these drugs for serious nonobstetrical infections.

Vancomycin is a glycopeptide antimicrobial active against gram-positive bacteria. It is used in lieu of β-lactam therapy for a patient with a type 1 allergic reaction and given for suspected infections due to Staphylococcus aureus and to treat C difficile colitis (Chap. 54, Upper Gastrointestinal Bleeding).

Perioperative Prophylaxis

As discussed, administration of antimicrobial prophylaxis at the time of cesarean delivery has remarkably reduced the postoperative pelvic and wound infection rates. Numerous studies have shown that prophylactic antimicrobials reduce the pelvic infection rate by 70 to 80 percent (Chelmow, 2001; Dinsmoor, 2009; Smaill, 2010; Witt, 2011). The observed benefit applies to both elective and nonelective cesarean delivery and also includes a reduction in abdominal incisional infection rates.

Single-dose prophylaxis with ampicillin or a first-generation cephalosporin is ideal, and both are as effective as broad-spectrum agents or a multiple-dose regimen (American College of Obstetricians and Gynecologists, 2011). Extended-spectrum prophylaxis with azithromycin added to standard single-dose prophylaxis has shown a further reduction in postcesarean metritis rates (Tita, 2008). These findings need to be verified. Women known to be colonized with methicillin-resistant Staphylococcus aureus—MRSA—are given vancomycin in addition to a cephalosporin (Chap. 64, Methicillin-Resistant Staphylococcus Aureus). Finally, it is controversial whether the infection rate is lowered more if the selected antimicrobial is given before the skin incision compared with after umbilical cord clamping (Baaqeel, 2012; Macones, 2012; Sun, 2013). The American College of Obstetricians and Gynecologists (2011) has concluded that the evidence favors predelivery administration. There may be additive salutary effects of preoperative vaginal cleansing with povidone-iodine rinse or application of metronidazole gel (Haas, 2013; Reid, 2011; Yildirim, 2012).

Other Methods of Prophylaxis

Several studies have evaluated the value of prenatal cervicovaginal cultures. These are obtained in the hope of identifying pathogens that might be eradicated to decrease incidences of preterm labor, chorioamnionitis, and puerperal infections. Unfortunately, treatment of asymptomatic vaginal infections has not been shown to prevent these complications. Carey and coworkers (2000) reported no beneficial effects for women treated for asymptomatic bacterial vaginosis. Klebanoff and colleagues (2001) reported a similar postpartum infection rate in women treated for second-trimester asymptomatic Trichomonas vaginalis infection compared with that of placebo-treated women.

Technical maneuvers done to alter the postpartum infection rate have been studied with cesarean delivery. For example, allowing the placenta to separate spontaneously compared with removing it manually lowers the infection risk. However, changing gloves by the surgical team after placental delivery does not (Atkinson, 1996). Exteriorizing the uterus to close the hysterotomy may decrease febrile morbidity (Jacobs-Jokhan, 2004). Postdelivery mechanical lower segment and cervical dilatation have not been shown to be effective (Liabsuetrakul, 2011). No differences were found in postoperative infection rates when single- versus two-layer uterine closure was compared (Hauth, 1992). Similarly, infection rates are not appreciatively affected by closure versus nonclosure of the peritoneum (Bamigboye, 2003; Tulandi, 2003). Importantly, although closure of subcutaneous tissue in obese women does not lower the wound infection rate, it does decrease the wound separation incidence (Chelmow, 2004; Magann, 2002; Naumann, 1995). Similarly, skin closure with staples versus suture has a higher incidence of noninfectious skin separation (Mackeen, 2012; Tuuli, 2011).

Complications of Uterine and Pelvic Infections

In more than 90 percent of women, metritis responds to treatment within 48 to 72 hours. In some of the remainder, any of several complications may arise. These include wound infections, complex pelvic infections such as phlegmons or abscesses, and septic pelvic thrombophlebitis (Jaiyeoba, 2012). As with other aspects of puerperal infections, the incidence and severity of these complications are remarkably decreased by perioperative antimicrobial prophylaxis.

Abdominal Incisional Infections

Wound infection is a common cause of persistent fever in women treated for metritis. Other wound infection risk factors include obesity, diabetes, corticosteroid therapy, immunosuppression, anemia, hypertension, and inadequate hemostasis with hematoma formation. If prophylactic antimicrobials are given as described above, the incidence of abdominal wound infection following cesarean delivery ranges from 2 to 10 percent depending on risk factors (Andrews, 2003; Chaim, 2000).

Incisional abscesses that develop following cesarean delivery usually cause persistent fever or fever beginning about the fourth day. In many cases, antimicrobials had been given to treat pelvic infection, yet fever persisted. There is wound erythema and drainage. Although organisms that cause wound infections are generally the same as those isolated from amnionic fluid at cesarean delivery, hospital-acquired pathogens may also be causative (Emmons, 1988; Owen, 1994). Treatment includes antimicrobials, surgical drainage, and debridement of devitalized tissue. The fascia is carefully inspected to document integrity.

Local wound care is typically completed twice daily. Before each dressing change, procedural analgesia is tailored to wound size and location, and oral, intramuscular, or intravenous dosage routes are suitable. Topical lidocaine may also be added. Necrotic tissue is removed, and the wound is repacked with moist gauze. At 4 to 6 days, healthy granulation tissue is typically present, and secondary en bloc closure of the open layers can usually be accomplished (Wechter, 2005). With this closure, a polypropylene or nylon suture of appropriate gauge enters 3 cm from one wound edge. It crosses the wound to incorporate the full wound thickness and emerges 3 cm from the other wound edge. These are placed in series to close the opening. In most cases, sutures may be removed on postprocedural day 10. Wound vacuum device use is gaining popularity. However, its efficacy remains unproven in randomized trials.

Wound Dehiscence

Wound disruption or dehiscence refers to separation of the fascial layer. This is a serious complication and requires secondary closure of the incision in the operating room. McNeeley and associates (1998) reported a fascial dehiscence rate of approximately 1 per 300 operations in almost 9000 women undergoing cesarean delivery. Most disruptions manifested on about the fifth postoperative day and were accompanied by a serosanguineous discharge. Two thirds of 27 fascial dehiscences identified in this study were associated with concurrent fascial infection and tissue necrosis.

Necrotizing Fasciitis

This uncommon, severe wound infection is associated with high mortality rates. In obstetrics, necrotizing fasciitis may involve abdominal incisions, or it may complicate episiotomy or other perineal lacerations. As the name implies, there is significant tissue necrosis. Of the risk factors for fasciitis summarized by Owen and Andrews (1994), three of these—diabetes, obesity, and hypertension—are relatively common in pregnant women. Like pelvic infections, these wound complications usually are polymicrobial and are caused by organisms that make up the normal vaginal flora. In some cases, however, infection is caused by a single virulent bacterial species such as group A β-hemolytic streptococcus. Occasionally, necrotizing infections are caused by rarely encountered pathogens (Swartz, 2004).

Goepfert and coworkers (1997) reviewed their experiences with necrotizing fasciitis at the University of Alabama Birmingham Hospital. Nine cases complicated more than 5000 cesarean deliveries—frequency of 1.8 per 1000. In two women, the infection was fatal. In a report from Brigham and Women’s and Massachusetts General Hospitals, Schorge and colleagues (1998) described five women with fasciitis following cesarean delivery. None of these women had predisposing risk factors, and none died.

Infection may involve skin, superficial and deep subcutaneous tissues, and any of the abdominopelvic fascial layers (Fig. 37-2). In some cases, muscle is also involved—myofasciitis. Although some virulent infections, for example, from group A β-hemolytic streptococci, develop early postpartum, most of these necrotizing infections do not cause symptoms until 3 to 5 days after delivery. Clinical findings vary, and it is frequently difficult to differentiate more innocuous superficial wound infections from an ominous deep fascial one. A high index of suspicion, with surgical exploration if the diagnosis is uncertain, may be lifesaving. We aggressively pursue early exploration. Certainly, if myofasciitis progresses, the woman may become ill from septicemia. Profound hemoconcentration from capillary leakage with circulatory failure commonly occurs, and death may soon follow as described in Chapter 47 (Clinical Manifestations).

Early diagnosis, surgical debridement, antimicrobials, and intensive care are paramount to successfully treat necrotizing soft-tissue infections (Gallup, 2004; Urschel, 1999). Surgery includes extensive debridement of all infected tissue, leaving wide margins of healthy bleeding tissue. This may include extensive abdominal or vulvar debridement with unroofing and excision of abdominal, thigh, or buttock fascia. Death is virtually universal without surgical treatment, and rates approach 50 percent even if extensive debridement is performed. With extensive resection, synthetic mesh may ultimately be required later to close the fascial incision (Gallup, 2004; McNeeley, 1998).

Adnexal Abscesses and Peritonitis

An ovarian abscess rarely develops in the puerperium. These are presumably caused by bacterial invasion through a rent in the ovarian capsule (Wetchler, 1985). The abscess is usually unilateral, and women typically present 1 to 2 weeks after delivery. Rupture is common, and peritonitis may be severe.

Peritonitis is infrequent following cesarean delivery. It is almost invariably preceded by metritis. It most often is caused by uterine incisional necrosis and dehiscence, but it may be due to a ruptured adnexal abscess or an inadvertent bowel injury at cesarean delivery. Peritonitis is rarely encountered after vaginal delivery, and many such cases are due to virulent strains of group A β-hemocyte streptococci or similar organisms.

Importantly in postpartum women, abdominal rigidity may not be prominent with puerperal peritonitis because of abdominal wall laxity from pregnancy. Pain may be severe, but frequently, the first symptoms of peritonitis are those of adynamic ileus. Marked bowel distention may develop, and these findings are unusual after uncomplicated cesarean delivery. If the infection begins in an intact uterus and extends into the peritoneum, antimicrobial treatment alone usually suffices. Conversely, peritonitis caused by uterine incisional necrosis as discussed subsequently, or from bowel perforation, must be treated promptly with surgical intervention.

Parametrial Phlegmon

For some women in whom metritis develops following cesarean delivery, parametrial cellulitis is intensive and forms an area of induration—a phlegmon—within the leaves of the broad ligament (Fig. 37-3). These infections should be considered when fever persists longer than 72 hours despite intravenous antimicrobial therapy (Brown, 1999; DePalma, 1982).

Phlegmons are usually unilateral, and they frequently are limited to the parametrium at the base of the broad ligament. If the inflammatory reaction is more intense, cellulitis extends along natural lines of cleavage. The most common form of extension is laterally along the broad ligament, with a tendency to extend to the pelvic sidewall. Occasionally, posterior extension may involve the rectovaginal septum, producing a firm mass posterior to the cervix. In most women with a phlegmon, clinical improvement follows continued treatment with a broad-spectrum antimicrobial regimen. Typically, fever resolves in 5 to 7 days, but in some cases, it persists longer. Absorption of the induration may require several days to weeks.

In some women, severe cellulitis of the uterine incision may lead to necrosis and separation (Treszezamsky, 2011). Extrusion of purulent material as shown in Figure 37-4 leads to intraabdominal abscess formation and peritonitis as described above. Surgery is reserved for women in whom uterine incisional necrosis is suspected because of ileus and peritonitis. For most, hysterectomy and surgical debridement are needed and are predictably difficult because the cervix and lower uterine segment are involved with an intense inflammatory process that extends to the pelvic sidewall. The adnexa are seldom involved, and one or both ovaries can usually be conserved. There is often appreciable blood loss, and transfusion may be necessary.

Imaging Studies

Persistent puerperal infections can be evaluated using computed tomography (CT) or magnetic resonance (MR) imaging. Brown and associates (1991) used CT imaging in 74 women in whom pelvic infection was refractory to 5 days of antimicrobial therapy. They found at least one abnormal radiological finding in 75 percent of these women, and in most, these were nonsurgical. Maldjian and coworkers (1999) used MR imaging in 50 women with persistent fever and found a bladder flap hematoma in two thirds. In three women, parametrial edema was seen, and another two had a pelvic hematoma. In most cases, imaging can be used to dissuade surgical exploration.

Uterine incisional dehiscence such as shown in Figure 37-4 is sometimes suspected based on CT scanning images. Another example is shown in Figure 37-5. These findings must be interpreted within the clinical context because apparent uterine incisional defects thought to represent edema can be seen even after uncomplicated cesarean delivery (Twickler, 1991).

Occasionally, a parametrial phlegmon suppurates, forming a fluctuant broad ligament mass that may point above the inguinal ligament. These abscesses may dissect anteriorly as shown in Figure 37-5 and be amenable to CT-directed needle drainage. Occasionally they dissect posteriorly to the rectovaginal septum, where surgical drainage is easily effected by colpotomy incision. A psoas abscess is rare, and despite antimicrobial therapy, percutaneous drainage may be required (Shahabi, 2002; Swanson, 2008).

Septic Pelvic Thrombophlebitis

This was a frequent complication in the preantibotic era. Collins and colleagues (1951) described suppurative thrombophlebitis in 70 women cared for from 1937 through 1946 at Charity Hospital in New Orleans. Septic embolization was common and caused a third of maternal deaths during that period. With the advent of antimicrobial therapy, the mortality rate and need for surgical therapy for these infections diminished.

Septic phlebitis arises as an extension along venous routes and may cause thrombosis as shown in Figure 37-6. Lymphangitis often coexists. The ovarian veins may then become involved because they drain the upper uterus and therefore the placental implantation site. The experiences of Witlin and Sibai (1995) and Brown and coworkers (1999) suggest that puerperal septic thrombophlebitis is likely to involve one or both ovarian venous plexuses (Fig. 37-7). In a fourth of women, the clot extends into the inferior vena cava and occasionally to the renal vein.

The incidence of septic phlebitis has varied in several reports. In a 5-year survey of 45,000 women who were delivered at Parkland Hospital, Brown and associates (1999) found an incidence of septic pelvic thrombophlebitis of 1 per 9000 following vaginal delivery and 1 per 800 with cesarean delivery. The overall incidence of 1 per 3000 deliveries was similar to the 1 per 2000 reported by Dunnihoo and colleagues (1991). In a cohort of 16,650 women undergoing primary cesarean delivery, Rouse and coworkers (2004) reported an incidence of 1 per 400 cesarean deliveries. Incidences approximated 1 per 175 surgeries if there was antecedent chorioamnionitis but only 1 per 500 if there was no intrapartum infection.

Women with septic thrombophlebitis usually have symptomatic improvement with antimicrobial treatment, however, they continue to have fever. Although there occasionally is pain in one or both lower quadrants, patients are usually asymptomatic except for chills. Diagnosis can be confirmed by either pelvic CT or MR imaging (Klima, 2008; Sheffield, 2001). Using either, Brown and colleagues (1999) found that 20 percent of 69 women with metritis who had fever despite ≥ 5 days of appropriate antimicrobial therapy had septic pelvic thrombophlebitis.

It has been disproven that intravenous heparin causes fever to dissipate with septic phlebitis (Brown, 1986; Witlin, 1995). And although Garcia and coworkers (2006) and Klima and Snyder (2008) advocate heparin therapy, we do not recommend anticoagulation. In a randomized study of 14 women by Brown and associates (1999), the addition of heparin to antimicrobial therapy for septic pelvic thrombophlebitis did not hasten recovery or improve outcome. Certainly, there is no evidence for long-term anticoagulation as given for non-infection-related venous thromboembolism.

Perineal Infections

Episiotomy infections are not common because the operation is performed much less frequently now than in the past (American College of Obstetricians and Gynecologists, 2013). Reasons for this are discussed further in Chapter 27 (Episiotomy). In an older study, Owen and Hauth (1990) described only 10 episiotomy infections in 20,000 women delivered at the University of Alabama at Birmingham. With infection, dehiscence is a concern. Ramin and colleagues (1992) reported an episiotomy dehiscence rate of 0.5 percent at Parkland Hospital—80 percent of these were infected. In a more recent study, Uygur and associates (2004) reported a 1-percent dehiscence rate and attributed two thirds to infection. There are no data to suggest that dehiscence is related to faulty repair.

Infection of a fourth-degree laceration can be more serious. Goldaber and coworkers (1993) described 390 women with a fourth-degree laceration, of whom 5.4 percent had morbidity. In these women, 2.8 percent had infection and dehiscence, 1.8 percent had only dehiscence, and 0.8 percent only infection. Stock and associates (2013) reported wound complications in 7.3 percent of women with anal sphincter injuries. They further observed that intrapartum antimicrobials were protective against infection. Although life-threatening septic shock is rare, it may still occur as a result of an infected episiotomy (Soltesz, 1999). Occasionally also, necrotizing fasciitis develops as discussed on Abdominal Incisional Infections.

Pathogenesis and Clinical Course

Episiotomy dehiscence is most commonly associated with infection as discussed above. Other factors include coagulation disorders, smoking, and human papillomavirus infection (Ramin, 1994). Local pain and dysuria, with or without urinary retention, are common symptoms. Ramin and colleagues (1992), evaluating a series of 34 women with episiotomy dehiscence, reported that the most common findings were pain in 65 percent, purulent discharge in 65 percent, and fever in 44 percent. In extreme cases, the entire vulva may become edematous, ulcerated, and covered with exudate.

Vaginal lacerations may become infected directly or by extension from the perineum. The mucosa becomes red and swollen and may then become necrotic and slough. Parametrial extension may result in lymphangitis.

Cervical lacerations are common but seldom are noticeably infected, which may manifest as metritis. Deep lacerations which extend directly into the tissue at the base of the broad ligament may become infected and cause lymphangitis, parametritis, and bacteremia.

Treatment

Infected episiotomies are managed like other infected surgical wounds. Drainage is established, and in many cases, sutures are removed and the infected wound debrided. In women with obvious cellulitis but no purulence, close observation and broad-spectrum antimicrobial therapy alone may be appropriate. With dehiscence, local wound care is continued along with intravenous antimicrobials. Hauth and colleagues (1986) were the first to advocate early repair after infection subsided, and other studies have confirmed the efficacy of this approach. Hankins and coworkers (1990) described early repair of episiotomy dehiscence in 31 women with an average duration of 6 days from dehiscence to episiotomy repair. All but two had a successful repair. Each of the two women with failures developed a pinpoint rectovaginal fistula that was treated successfully with a small rectal flap. With episiotomy dehiscence due to infection, Ramin and coworkers (1992) reported successful early repair in 32 of 34 women (94 percent), and Uygur and colleagues (2004) noted a similarly high percentage. Rarely, intestinal diversion may be required to allow healing (Rose, 2005).

Early Repair of Infected Episiotomy

Before performing early repair, diligent preparation is essential as outlined in Table 37-3. The surgical wound must be properly cleaned and cleared of infection. Once the surface of the episiotomy wound is free of infection and exudate and covered by pink granulation tissue, secondary repair can be accomplished. The tissue must be adequately mobilized, with special attention to identify and mobilize the anal sphincter muscle. Secondary closure of the episiotomy is accomplished in layers, as described for primary episiotomy closure (Chap. 27, Repair of Episiotomy or Perineal Laceration). Postoperative care includes local wound care, low-residue diet, stool softeners, and nothing per vagina or rectum until healed.

Toxic Shock Syndrome

This acute febrile illness with severe multisystem derangement has a case-fatality rate of 10 to 15 percent. There is usually fever, headache, mental confusion, diffuse macular erythematous rash, subcutaneous edema, nausea, vomiting, watery diarrhea, and marked hemoconcentration. Renal failure followed by hepatic failure, disseminated intravascular coagulation, and circulatory collapse may follow in rapid sequence. During recovery, the rash-covered areas undergo desquamation. For some time, Staphylococcus aureus was recovered from almost all afflicted persons. Specifically, a staphylococcal exotoxin, termed toxic shock syndrome toxin-1 (TSST-1) was found to cause the clinical manifestations by provoking profound endothelial injury. A very small amount of TSST-1 has been shown to activate 5 to 30 percent of T cells to create a “cytokine storm” as described by Que (2005) and Heying (2007) and their coworkers.

During the 1990s, sporadic reports of virulent group A β-hemolytic streptococcal infection began to appear. Heavy colonization or infection is complicated in some cases by streptococcal toxic shock syndrome, which is produced when pyrogenic exotoxin is elaborated. Serotypes M1 and M3 are particularly virulent (Beres, 2004; Okumura, 2004). Finally almost identical findings of toxic shock were reported by Robbie and associates (2000) in women with Clostridium sordellii colonization.

Thus, in some cases of toxic shock syndrome, infection is not apparent and colonization of a mucosal surface is the presumed source. Ten to 20 percent of pregnant women have vaginal colonization with S aureus. Thus, it is not surprising that the disease develops in postpartum women when there is luxuriant bacterial growth vaginally (Chen, 2006; Guerinot, 1982).

Delayed diagnosis and treatment may be associated with maternal mortality (Schummer, 2002). Crum and colleagues (2002) described a neonatal death following antenatal toxic shock syndrome. Principal therapy is supportive, while allowing reversal of capillary endothelial injury (Chap. 47, Clinical Manifestations). Antimicrobial therapy that includes staphylococcal and streptococcal coverage is given. With evidence of pelvic infection, antimicrobial therapy must also include agents used for polymicrobial infections. Women with these infections may require extensive wound debridement and possibly hysterectomy. Because the toxin is so potent, the mortality rate is correspondingly high (Hotchkiss, 2003).

Parenchymal infection of the mammary glands is a rare antepartum complication but is estimated to develop in up to a third of breast-feeding women (Barbosa-Cesnik, 2003). Excluding breast engorgement, in our experiences, as well as that of Lee and associates (2010), the incidence of mastitis is much lower and probably approximates 1 percent. There is no evidence that any of several prophylactic measures prevent breast infection (Crepinsek, 2012). Risk factors include difficulties in nursing and mothers with outside employment (Branch-Elliman, 2012). Symptoms of suppurative mastitis seldom appear before the end of the first week postpartum and as a rule, not until the third or fourth week. Infection almost invariably is unilateral, and marked engorgement usually precedes inflammation. Symptoms include chills or actual rigors, which are soon followed by fever and tachycardia. There is severe pain, and the breast(s) becomes hard and red (Fig. 37-8). Approximately 10 percent of women with mastitis develop an abscess. Detection of fluctuation may be difficult, and sonography is usually diagnostic as also shown in Figure 37-8.

Etiology

Staphylococcus aureus, especially its methicillin-resistant strain, is the most commonly isolated organism. Matheson and coworkers (1988) reported it in 40 percent of women with mastitis. Other commonly isolated organisms are coagulase-negative staphylococci and viridans streptococci. The immediate source of organisms that cause mastitis is almost always the infant’s nose and throat. Bacteria enter the breast through the nipple at the site of a fissure or small abrasion. The infecting organism can usually be cultured from milk. Toxic shock syndrome from mastitis caused by S aureus has been reported (Demey, 1989; Fujiwara, 2001).

At times, suppurative mastitis reaches epidemic levels among nursing mothers. Such outbreaks most often coincide with the appearance of a new strain of antibiotic-resistant staphylococcus. A contemporaneous example is community-acquired methicillin-resistant S aureus (CA-MRSA), which has rapidly become the most commonly isolated staphylococcal species in some areas (Klevens, 2007; Pallin, 2008). At Parkland Hospital from 2000 to 2004, Laibl and associates (2005) reported that a fourth of CA-MRSA isolates were from pregnant or postpartum women with puerperal mastitis. Hospital-acquired MRSA may cause mastitis when the infant becomes colonized after contact with nursery personnel who are colonized (Centers for Disease Control and Prevention, 2006). Stafford and colleagues (2008) found a higher incidence of subsequent abscess in those with CA-MRSA-associated mastitis. In one study of puerperal breast abscesses, 63 percent were due to methicillin-resistant strains (Berens, 2010).

Management

Provided that appropriate therapy for mastitis is started before suppuration begins, the infection usually resolves within 48 hours. As discussed, abscess formation is more common with S aureus infection (Matheson, 1988). Most recommend that milk be expressed from the affected breast onto a swab and cultured before therapy is begun. Bacterial identification and antimicrobial sensitivities provide information mandatory for a successful program of nosocomial infection surveillance (Lee, 2010).

The initial antimicrobial choice is influenced by the current experience with staphylococcal infections at the institution. Although most are community-acquired organisms, as discussed above, these frequently include CA-MRSA. Dicloxacillin, 500 mg orally four times daily, may be started empirically. Erythromycin is given to women who are penicillin sensitive. If the infection is caused by resistant, penicillinase-producing staphylococci or if resistant organisms are suspected while awaiting the culture results, then vancomycin or another anti-MRSA antimicrobial should be given. Even though clinical response may be prompt, treatment should be continued for 10 to 14 days.

Marshall and coworkers (1975) demonstrated the importance of continued breast feeding. They reported that of 65 women with mastitis, the only three who developed abscesses were among the 15 women who quit breast feeding. Vigorous milk expression may be sufficient treatment alone (Thomsen, 1984). Sometimes the infant will not nurse on the inflamed breast. This probably is not related to any changes in the milk taste but is secondary to engorgement and edema, which can make the areola harder to grip. Pumping can alleviate this. When nursing bilaterally, it is best to begin suckling on the uninvolved breast. This allows let-down to commence before moving to the tender breast.

In resource-poor countries, breast feeding in women infected with the human immunodeficiency virus (HIV) is not contraindicated. In the setting of mastitis or breast abscess, it is recommended to stop feeding from the infected breast. This is because HIV RNA levels increase in affected breast milk. These levels return to baseline after symptoms resolve (Semrau, 2013).

Breast Abscess

In a population-based study of nearly 1.5 million Swedish women, the incidence of breast abscess was 0.1 percent (Kvist, 2005). An abscess should be suspected when defervescence does not follow within 48 to 72 hours of mastitis treatment or when a mass is palpable. Again, sonographic imaging is valuable (see Fig. 37-8). Breast abscesses can be quite large, and in one case report, 2 liters of pus was drained (Martic, 2012). Traditional therapy is surgical drainage, which usually requires general anesthesia. The incision should be made corresponding to Langer skin lines for a cosmetic result (Stehman, 1990). In early cases, a single incision over the most dependent portion of fluctuation is usually sufficient. Multiple abscesses, however, require several incisions and disruption of loculations. The resulting cavity is loosely packed with gauze, which should be replaced at the end of 24 hours by a smaller pack.

A more recently used technique that is less invasive is sonographically guided needle aspiration using local analgesia. This has an 80- to 90-percent success rate (O’Hara, 1996; Schwarz, 2001). In a randomized trial, Naeem and colleagues (2012) compared surgical drainage and aspiration. They found aspiration resulted in quicker healing at 8 weeks, 77 versus 93 percent, respectively.