Sepsis in Pregnancy

In most treatment trials, pregnant patients are explicitly barred from enrollment. Because severe sepsis and septic shock (aside from unsafe abortion) are not common in pregnancy, the epidemiology of sepsis in this population is not as well described as in a general medical-surgical population. The World Health Organization recently estimated 77,000 deaths worldwide per year from maternal sepsis, with 0.1% to 10% of all live births being complicated by some degree of maternal infection. Criteria for sepsis or severe sepsis have been met in 3 to 9 per 10,000 deliveries in Europe ; Callaghan, using data from the National Inpatient Sample, calculated 3 cases of sepsis per 10,000 delivery or postpartum hospitalizations in the United States, a figure that excludes antepartum hospitalizations not resulting in delivery.

The case-fatality rate for sepsis in the obstetric population is not known with any degree of certainty; however, the case-fatality rate for septic abortion specifically is as high as 20%. Calculations based on birth statistics and the National Inpatient Sample, although not provided in the original paper, would put the overall case-fatality rate for sepsis at delivery or postpartum at approximately 9%.

Sepsis may be obstetric or nonobstetric. Causes of obstetric sepsis include uterine infection (chorioamnionitis if undelivered, endomyometritis postpartum), septic abortion, and wound infection (cesarean or episiotomy wound); in addition, sepsis may follow invasive procedures such as amniocentesis, chorionic villus sampling, cervical cerclage, or percutaneous umbilical blood sampling. One of the few case series in the U.S. literature on septic shock in pregnancy reported half of the cases to have an obstetric cause whereas, of the 50% with nonobstetric causes, most were urinary in origin. However, more recent data from the U.K. Obstetric Surveillance System showed that 31% of severe sepsis cases in obstetric patients arose from the genital tract, another 20% were urinary in origin, and 26% had no identified source.

What Criteria Should Be Used to Diagnose Sepsis in a Pregnant or Postpartum Patient? Are These Different from the Generally Accepted Criteria?

Criteria for the diagnosis of sepsis, originally promulgated in 1992, were reconfirmed in an international multispecialty conference in 2001. Their applicability in pregnancy, labor, or the immediate puerperium, however, is limited. There is considerable overlap between the normal physiologic parameters of pregnancy and the criteria used to make a diagnosis of systemic inflammatory response syndrome (SIRS) or sepsis. This may contribute both to a delayed recognition of sepsis among experienced obstetric care providers and to an overdiagnosis of sepsis in obstetric patients by critical care professionals. A recent systematic review of normal physiologic variables during pregnancy, labor, and the puerperium demonstrated this dilemma in detail. In normal pregnancy, respiratory rates increase from the second to third trimester, with further increases in labor, and remain elevated postpartum. If the threshold for tachypnea is taken as a respiratory rate of 20 breaths per min, then the normal range in pregnancy includes this rate from the second trimester through the first few days postpartum. The normal range for maternal heart rate shows an overlap with the SIRS criterion for tachycardia (90 beats/min) in all stages of pregnancy. Normal white blood cell (WBC) count is also increased from the second trimester through the puerperium, again significantly overlapping the range identified as leukocytosis in the usual SIRS criteria. In fact, the mean WBC count in the first 2 days postpartum is 15 × 10 ⁹ /L, which would make it quite difficult to discriminate SIRS or sepsis by laboratory criteria in that period. Of the parameters analyzed, temperature alone is not affected by the fact of pregnancy (although the use of epidural analgesia in labor does increase the maternal temperature and could confound the diagnosis). In addition, both diastolic pressure and serum creatinine are known to decrease in pregnancy, which might call into question the threshold criteria commonly used for a diagnosis of severe sepsis.

A Sepsis in Obstetrics Score was recently proposed as an indicator of maternal morbidity ( Table 78-1 ). On the basis of the Rapid Emergency Medicine Score, itself a derivative of the APACHE (Acute Physiology and Chronic Health Evaluation) score, and the SIRS/sepsis criteria of the Surviving Sepsis Campaign (SSC), which were modified for pregnancy-specific parameters such as blood pressure, heart rate, and WBC count, the authors hoped to refine a tool for predicting admission to the ICU within 48 hours after presentation to the emergency department in a group of pregnant and postpartum women in whom sepsis was suspected ( Table 78-1 ). A cutoff score of 6 or more was found to have 89% sensitivity and 99% specificity for the outcome of interest, but the ICU admission rate of approximately 1% in the study population meant that the positive predictive value of the scoring system was only approximately 17%. This test performance was nevertheless better in predicting outcomes for obstetric patients with suspected sepsis than that of standard, nonmodified scoring systems such as the SIRS score alone or the Modified Early Warning System (MEWS).

Can the Surviving Sepsis Campaign Guidelines Be Applied in Cases of Sepsis in Obstetric Patients?

The SSC is a multiorganizational effort to improve outcomes in sepsis and septic shock that is based on best available evidence. It recommends several goals, which are summarized in the following list, with commentary relating specifically to obstetric patients. There is no evidence base for these guidelines in a pregnant or postpartum patient, but there is no evidence against them either.

• 1.
  

  Initial resuscitation for patients with sepsis- induced hypotension.

  
  

  During the first 6 hours, target central venous pressure (CVP) 8 to 12 mm Hg, mean arterial pressure (MAP) greater than 65 mm Hg, urine output greater than 0.5 mL/kg/min, and mixed venous oxygen saturation 65%. Normalize lactate in patients with elevated lactate levels.

  
  

  MAP is commonly lower in pregnancy, which is a volume-loaded vasodilated state. MAP as low as 60 mm Hg may still be normal, and normal CVP is commonly lower than 8 mm Hg. Oncotic pressure is also lower in normal pregnancy; therefore, fluid loading with isotonic solution may predispose more easily to pulmonary edema. No pregnancy-specific guidelines or cutoffs have been proposed for these parameters.

  
  
• 2.
  

  Blood cultures before antibiotic therapy, if this does not significantly delay starting antimicrobial therapy.

  
  

  There is no reason this would not apply in pregnant/postpartum patients. One study in Finland reported on this specific policy for obstetric patients; 2% (of >40,000) were cultured for fever and had broad-spectrum antibiotics immediately administered. Bacteremia was confirmed in 5% of cases; only 1 of the 798 patients cultured had septic shock, for an incidence of 0.1%.

  
  
• 3.
  

  Imaging studies performed promptly to ascertain the source of infection.

  
  

  Pregnant women can be imaged despite the fact of pregnancy, although there are some issues related to ionizing radiation. The American College of Obstetricians and Gynecologists recommends limiting the total radiation dose during pregnancy to 5 rad (5 cGy) because no fetal effects are known this low. Substitute nonionizing modality if feasible (e.g., ultrasound, magnetic resonance imaging). If ionizing radiation is to be used, then shield the abdomen if possible. If ionizing radiation is required and the abdomen/pelvis is to be included in the field, then modify the technique to minimize the dose delivered to the fetus, and use dosimetry to tally the fetal dose.

  
  
• 4.
  

  Initiation of broad-spectrum antibiotic therapy within 1 hour of diagnosis.

  
  

  There is no reason this would not be feasible; however, the hemodynamic picture that characterizes normal pregnancy may result in overdiagnosing sepsis. The central hemodynamics of normal pregnancy include increased cardiac output, increased heart rate, decreased systemic vascular resistance, and a somewhat lower blood pressure. Most but not all antibiotics can be used in pregnancy, although dose adjustments may be needed because of changes in pharmacokinetics (e.g., expanded plasma volume, increased glomerular filtration rate, increased protein binding). Broad-spectrum coverage is reasonable in obstetrics patients. In a recent Finnish study of peripartum sepsis, more than 40 organisms were cultured, including aerobic gram-positive and gram-negative as well as anaerobic bacteria, and extended-spectrum beta-lactamase–producing microbes should be considered in high-risk patients who do not respond to initial antibiotics.

  
  
• 5.
  

  Reassessment of antibiotic therapy with clinical and microbiologic data to narrow antibiotic coverage when appropriate.

  
  

  There are no data specific to pregnancy. When narrowing coverage, some consideration should be given to whether transplacental coverage is appropriate; some drugs do not cross the placenta well and may result in inadequate fetal treatment (e.g., erythromycin or azithromycin in the treatment of syphilis).

  
  
• 6.
  

  Seven to ten days of antibiotic therapy.

  
  

  There is no evidence base specific to pregnancy, and no reason to recommend alteration in this goal.

  
  
• 7.
  

  Source control.

  
  

  There are no data specific to pregnancy. A significant proportion of cases of sepsis in pregnant/postpartum women localize to the uterus and would therefore require the uterus be emptied. This would generally equate to delivery. Fetuses less than 23 weeks gestational age are unlikely to survive outside of the uterus; at 23 and 24 weeks, survival rates of 26% and 55%, respectively, have been reported, at least at the highest-level neonatal intensive care facilities, although only 10% of those infants survive without major morbidity. There are no data on antibiotics without delivery for women diagnosed with clinical sepsis attributed to intra-amniotic infection. Women with a diagnosis of subclinical intra-amniotic infection who were treated with antibiotics alone, in the hope of delaying delivery to a more favorable gestational age, have been observed to have a prolongation of pregnancy by days to weeks, with the only maternal morbidity recorded being a 3% rate of postpartum endometritis, but with an infant death rate of 33% and major infant morbidity greater than 75%. It should be emphasized that patients with subclinical chorioamnionitis, who typically present only with preterm labor or membrane rupture, are unlikely to come to the ICU; if these patients cannot reasonably be managed without delivery, then there is no argument for managing clinical chorioamnionitis without it. There appears to be no place for deferring source control in pregnancy.

  
  
• 8.
  

  Crystalloid as the fluid of choice for resuscitation.

  
  

  There is no evidence to recommend one versus the other in pregnant patients with sepsis. Trials of crystalloid versus colloid have been performed to assess preloading before elective cesarean delivery with the patient under regional anesthesia, but extrapolation to sepsis would be inappropriate. Because the gradient between colloid oncotic pressure and pulmonary artery occlusion pressure is lower in pregnancy, there may be more risk of pulmonary edema than in the nonobstetric patient.

  
  
• 9.
  

  Norepinephrine as the first-choice vasopressor, to target initial MAP greater than 65 mm Hg.

  
  

  No data exist to recommend a lower limit of MAP in pregnancy, but MAP is normally lower in pregnancy than in healthy nonpregnant controls. Thus, a MAP of 65 mm Hg or greater may be too stringent. Although the MAP difference is 4 to 5 mm Hg lower in pregnancy, one cannot extrapolate to a recommendation to target 60 mm Hg instead. The uteroplacental circulation does not autoregulate, and compromised placental perfusion may be apparent by examination of the electronic fetal heart rate tracing: the tracing may allow individualization of the target MAP for the mother. Although norepinephrine, similar to epinephrine, vasopressin, and dopamine, has been used clinically in obstetric crises such as shock, there are limited data on safety or efficacy of any of these drugs in human pregnancy. During pregnancy, the response to vasoconstrictors, both endogenous and exogenous, is blunted; therefore, the usual therapeutic doses may not result in the expected effect. No recommendation can be made. This decision must be individualized.

Space limits a discussion of the supportive therapies reviewed in SSC. Such therapies as transfusion of red cells and platelets, sedation, venous thrombosis prophylaxis, and stress ulcer prophylaxis are not proscribed during pregnancy. To avoid hyperinsulinemia in the fetus, one would generally target an upper maternal blood glucose level of 140 mg/dL or less rather than the SSC recommendation of 180 mg/dL or less.