Demographics

In the United States, race is one of the most significant risk factors for preterm delivery. Black women have a prematurity rate of about 16% to 18%, in comparison with 7% to 9% for white women. Very low birth weight neonates (less than 1500 g) demonstrate the greatest risk of neonatal morbidity and death, and these neonates are disproportionately represented by African-American babies. Other factors have been implicated, including extremes of maternal age, less education, and lower socioeconomic status, all of which increase the risk of preterm delivery. However, even when these factors are controlled for, black women still have higher rates of preterm delivery.

In addition to race, various behavioral factors increase the risk of preterm birth. Nutritional status, either poor or excessive weight gain, seems to increase this risk. Smoking does as well, with 10% to 20% of all preterm births attributed to this habit, but tobacco use actually plays a more significant role in growth restriction than it does in preterm delivery. The increasing use of cocaine during pregnancy is another important behavioral factor. The pathophysiology of cocaine use is likely similar to that of smoking, primarily that of vasoconstriction leading to an increased rate of abruption (see Chapter 53).

Yet another behavioral factor related to preterm birth rates is the degree of physical activity and stress during pregnancy. Several studies have evaluated the effects of employment on preterm delivery, with disparate results ranging from an increase to an actual decrease in the risk of preterm birth in the working group. These variable results are likely related to the fact that physical activity levels probably impact the rate of preterm birth more than simple employment statistics. For example, activity in the standing position has been shown to increase uterine irritability, likely owing to uterine compression of pelvic vessels resulting in a decreased venous blood return to the heart, a phenomenon that may be temporarily relieved via contractive activity. Maternal stress also plays a role in the association between work activity and preterm birth.³³ It seems therefore reasonably clear that women who engage in hard, physical work for long hours under increased stress are at a greater risk of preterm birth than inactive women.

Obstetric History

Obstetric history can also be used as a predictor of preterm birth. A history of a prior preterm delivery is one of the most significant risk factors. The recurrence risk of preterm birth ranges from 17% to 40%. Both the number of prior preterm deliveries and the gestational age at which those deliveries occurred affect the risk of preterm birth in subsequent pregnancies. For example, whereas women with one prior preterm birth have a threefold increased risk for preterm delivery in comparison with women with no such history, a sixfold increased risk is seen in women with two previous preterm births.²⁷ Furthermore, the risk of recurrent preterm birth is inversely related to the gestational age at which a prior child was born.

Prior second-trimester abortions, whether single or multiple, also increase the risk of preterm delivery. However, the picture is less clear for women with a history of either spontaneous or induced first-trimester abortions. Studies evaluating one first-trimester abortion report no increased risk; however, data regarding multiple first-trimester abortions are inconsistent.

Cervical and Uterine Factors

Patients with congenital müllerian anomalies have an increased risk of preterm delivery. Approximately 3% to 16% of all preterm births are associated with a uterine malformation. The incidence of preterm labor varies greatly depending on the type of uterine anomaly. Unicornuate, didelphic, and bicornuate abnormalities have preterm labor rates ranging from 18% to 80%, whereas the rates for a septate uterus vary from 4% to 17%, depending on whether the division is complete or incomplete.

Uterine leiomyomata have also been associated with an increased risk of preterm delivery, primarily owing to an increased incidence of antepartum bleeding and PPROM. Of the various types of myomata, submucosal and subplacental myomata appear to be most strongly associated with preterm delivery.

Cervical incompetence is another important risk factor for preterm delivery. The classic clinical description of cervical incompetence involves a history of painless cervical dilation occurring between 12 and 20 weeks. A history of a second-trimester pregnancy loss has been the cornerstone of the diagnosis, but distinguishing between cervical incompetence and preterm labor can at times be difficult. Several techniques have been used to diagnose cervical incompetence in nonpregnant women, including passage of a No. 8 Hegar dilator with ease through the internal os, measurement of the pressure necessary to pull a Foley balloon catheter inflated with 1 mL of water through the internal os, and identification of an abnormal cervical canal and isthmic funnel angle by either hysterosalpingography or hysteroscopy. These methods evaluate cervical incompetence in the nonpregnant state, though, making the results at best suggestive of the diagnosis in pregnant women. Because of this inaccuracy as well as the potential trauma of such diagnostic techniques, these procedures are rarely performed.

Intrauterine exposure to diethylstilbestrol (DES) is a significant factor associated with congenital causes of cervical incompetence and both upper and lower genital tract structural abnormalities. An estimated 1 to 1.5 million women were exposed in utero to DES between the late 1940s and 1971. These women have an increased risk of preterm delivery ranging from 15% to 28%, with an increased risk of spontaneous abortion of 20% to 40%. Women exposed to DES who have associated anomalies, such as T-shaped uterus, cervical incompetence, or vaginal structural anomalies, have a greater risk of preterm delivery than those who do not demonstrate these structural abnormalities. Because most women exposed to DES in utero are now older than 40, this risk factor is becoming less and less of an issue.

Procedure-related obstetric and gynecologic trauma can also cause cervical incompetence. First-trimester dilation and curettage appears to confer minimal risk for subsequent cervical incompetence. In contrast, cervical procedures performed for the diagnosis and treatment of cervical intraepithelial neoplasia, including cold knife conization, laser ablation, or loop electrosurgical excision, have been associated with cervical incompetence in subsequent pregnancies. The risk for preterm birth and PPROM appears to be directly related to the depth of tissue removed from the cervix in such procedures, with a greater than threefold increased risk of PPROM among women with the greatest depth of excision (i.e., 1.7 cm).¹⁶⁶

Multifetal Gestations

Multiple gestations carry one of the highest risks of preterm delivery. Approximately 50% of twin and nearly all of higher-multiple gestations deliver prior to 37 completed weeks, with an average length of gestation of 35 weeks for twins, 33 weeks for triplets, and 29 weeks for quadruplets. Owing to artificial reproductive techniques, the prevalence of multiple gestations has increased in the United States. Multiple gestations are discussed further in Chapter 22.

Bleeding

Vaginal bleeding is also associated with an elevated risk of preterm delivery. This is particularly true in the setting of placenta previa and placental abruption, but first or second trimester vaginal bleeding in the absence of these conditions is also associated with an increased risk of preterm delivery. The mechanism responsible for the development of preterm labor in the setting of vaginal bleeding appears to be related to thrombin deposition with subsequent production of prostaglandins and plasminogen activators that stimulate an array of degradative enzymes leading to destruction of the extracellular matrix.

Infection

Infections of the decidua, fetal membranes, and amniotic fluid have been associated with preterm delivery (Figure 20-3) (see Chapter 26).⁶⁰ For example, although intra-amniotic infection or chorioamnionitis complicates 1% to 5% of term pregnancies, this diagnosis occurs in nearly 25% of patients with preterm delivery.

There is a large body of evidence establishing a strong link between occult upper genital tract infection and spontaneous preterm delivery. For example, Watts and associates investigated a series of patients with preterm labor, demonstrating that positive amniotic fluid cultures were present in 19% of women with spontaneous preterm labor and intact membranes despite these women having no clinical evidence of intrauterine infection.¹⁹⁵ Interestingly, the likelihood of a positive amniotic fluid culture in this investigation was inversely proportional to the gestational age at delivery.¹⁹⁵ An association between histologic chorioamnionitis and preterm delivery has also been established. Organisms associated with histologic chorioamnionitis include Ureaplasma, Mycoplasma, Gardnerella, Bacteroides, and Mobiluncus species.¹⁹⁵ Again, an inverse relationship exists between colonization of the chorioamnion and the gestational age at delivery in women with spontaneous preterm labor.⁸² In fact, chorioamnion colonization is associated with up to 80% of very early spontaneous preterm births. In contrast, microbial colonization of the upper genital tract appears to play a much less important role in the initiation of parturition at or near term.⁸²

Numerous theories exist regarding the underlying pathogenesis of intra-amniotic infection: (1) ascending infection from the vagina and cervix, (2) transplacental passage through hematogenous dissemination, (3) retrograde seeding from the peritoneal cavity through the fallopian tubes, and (4) iatrogenic means as a result of intrauterine procedures such as amniocentesis and chorionic villus sampling. There is some evidence to support each of these theories, but ascending infection is the most well accepted theory based on data from several sources. For example, in cases of congenital pneumonia, inflammation of the chorioamniotic membrane is present, and the bacteria identified are similar to those found in the genital tract.

The mechanism of ascending infection appears to start with excessive overgrowth of certain organisms within the vagina and cervical canal. These microorganisms then gain access to the intrauterine cavity by infecting the decidua, the chorion, the amnion, and then finally the amniotic cavity itself. In this manner, pathogens are able to cross intact membranes. The fetus then becomes infected by aspirating or swallowing infected amniotic fluid or by direct contact with the organism within the fluid, leading to localized infections such as pneumonitis, otitis, or conjunctivitis. Seeding of these areas can lead, in turn, to generalized fetal sepsis. Alternatively, sepsis may result from maternal bacteremia, leading to placental infection with subsequent spread of organisms through the umbilical cord to the fetus. Supporting this theory is the high relative risk of premature birth in the setting of asymptomatic bacteriuria, a risk that is reduced with appropriate treatment of the infection.

Intra-amniotic infection can also lead to preterm labor through less direct methods, namely stimulation of prostaglandin. The amniotic fluid concentrations of prostaglandins PGF_2α and PGE₂, as well as their metabolites, are increased in patients with preterm labor, preterm premature rupture of membranes, and intra-amniotic infection. Synthesis of prostaglandins by the human amnion is likely mediated through bacterial endotoxins such as phospholipase A₂ and C, which are elevated in women with preterm labor and PPROM. Lipopolysaccharide is another well-known endotoxin that is created by gram-negative organisms. This endotoxin has been found in the amniotic fluid of women with chorioamnionitis, where it has been shown to stimulate prostaglandin synthesis by the amnion and decidua. However, it is unlikely that the quantities of endotoxin within the amniotic fluid of women with chorioamnionitis are sufficient enough to stimulate prostaglandin production. Instead, these endotoxins likely stimulate production of various cytokines by the host, which in turn stimulates prostaglandin production. However, the finding that 28% of patients with PPROM have positive amniotic fluid cultures in the absence of labor suggests that it takes more than the presence of bacteria to stimulate prostaglandin production and labor.

Various organisms have been implicated as causal agents for preterm labor and PPROM. For example, Neisseria gonorrhoeae infection is associated with an increased risk of preterm birth, with early delivery occurring in up to 25% of patients with positive cultures as compared to 12.5% of women with negative results.

Infection with Chlamydia trachomatis has also been associated with prematurity. Although no differences have been documented in the prematurity rate, premature rupture of membranes, both term and preterm, is more common in women with positive immunoglobulin M (IgM) titers. Furthermore, Andrews and colleagues reported that patients with a diagnosis of chlamydia during pregnancy have an odds ratio of 2 for spontaneous preterm birth.⁸ All together, evidence suggests that there is a correlation between C. trachomatis infection and prematurity, especially in the setting of PPROM.

Group B streptococcus (GBS) is another organism that has been associated with an increased risk of preterm delivery. Regardless of gestational age at delivery, GBS colonization plays a major role in neonatal morbidity and death, but this effect is particularly prominent in premature infants. A higher incidence of low birth weight occurs in colonized women, and the incidence of PPROM is higher as well. Studies demonstrate that asymptomatic bacteriuria with GBS is also a risk factor for preterm delivery and that eradication with antibiotics decreases that risk. Patients with urinary colonization also can have positive results on cervical and vaginal cultures, possibly indicating that the presence of GBS in the urine may be a marker of more severe forms of genital tract colonization. Unfortunately, treatment of GBS genital tract colonization has never been shown to decrease the risk of preterm delivery or PPROM.

In 1996, the Centers for Disease Control and Prevention (CDC), in conjunction with the American College of Obstetricians and Gynecologists and the American Academy of Pediatrics, set forth recommendations for obstetric providers to adopt either a culture-based or a risk-based approach for the prevention of GBS disease.²⁸ For the culture-based strategy, intrapartum antibiotic prophylaxis was offered to women identified as GBS carriers through prenatal screening cultures collected at 35 to 37 weeks’ gestation and to women who developed premature onset of labor or rupture of membranes before 37 weeks’ gestation. For the risk-based approach, cultures are not employed, and intrapartum antibiotic prophylaxis is provided to women who at the time of labor or membrane rupture develop one or more of the following risk conditions: (1) preterm labor at less than 37 weeks’ gestation, (2) PPROM at less than 37 weeks, (3) membranes ruptured for 18 hours or longer, or (4) maternal fever of 38° C or higher. In 2002, after a population-based comparison of the culture-based and risk-based strategies revealed a significantly lower incidence of early-onset neonatal GBS disease among women cared for using the culture-based approach, the CDC revised the recommendations to adopt the culture-based approach for the prevention of early-onset neonatal GBS infection.^172,173 These guideline changes have subsequently been supported by the American College of Obstetricians and Gynecologists. The guidelines were further modified by the CDC in 2010 and elaborated on by the Committee on Infectious Disease and Fetus and Newborn in 2011. In 2010, GBS disease remained the leading cause of early-onset neonatal sepsis. The CDC issued revised guidelines in 2010 based on evaluation of data generated after 2002. These revised and comprehensive guidelines, which have been endorsed by the AAP, reaffirm the major prevention strategy—universal antenatal GBS screening and intrapartum antibiotic prophylaxis for culture-positive and high-risk women—and include new recommendations for laboratory methods for identification of GBS colonization during pregnancy, algorithms for screening, and intrapartum prophylaxis for women with preterm labor and premature rupture of membranes, as well as clarification of the colony count threshold required for reporting GBS detected in urine of pregnant women, updated prophylaxis recommendations for women with a penicillin allergy, and a revised algorithm for the care of newborn infants.^31,191

For women who are culture positive for GBS, intrapartum chemoprophylaxis with intravenous penicillin G (5 million units initially and then 2.5 million units every 4 hours) is recommended until delivery. Intravenous ampicillin (2 g initially, and then 1 g every 4 hours) is an acceptable alternative to penicillin G. Because of emerging resistance of GBS to macrolides, guidelines have recently been modified for women who are allergic to penicillins.¹⁹¹ In penicillin-allergic women who are not at high risk for anaphylaxis, intravenous cefazolin (2 g initially, and then 1 g every 8 hours until delivery) is recommended. In penicillin-allergic women who are at high risk for anaphylaxis, clindamycin and erythromycin susceptibility testing of the GBS isolate is recommended. If the isolate is sensitive to both clindamycin and erythromycin, intravenous treatment with the appropriate agent is recommended (clindamycin 900 mg every 8 hours or erythromycin 500 mg every 6 hours).¹⁷² If the GBS isolate is resistant to either agent or susceptibility is unknown, treatment with intravenous vancomycin (1 g every 12 hours) is recommended. Van Dyke and co-workers evaluated the implementation of the 2002 guidelines using a multistate, retrospective cohort. ¹⁹⁰ They documented that the rate of screening for GBS before delivery increased from 48.1% in 1998-1999 to 85% in 2003-2004, whereas the percentage of infants exposed to intrapartum antibiotics increased from 26.8% to 31.7% in this same time period. Chemoprophylaxis was appropriately administered to 87% of the women who delivered at term with a positive GBS culture, but only 63.4% of women with an unknown colonization status who delivered preterm received antibiotic coverage. The overall incidence of early-onset group B streptococcal disease was 0.32 cases per 1000 live births, with a higher incidence in preterm infants than term infants (0.73 vs 0.26 cases per 1000 live births). Despite the increased incidence in the preterm infants, 74.4% of the cases of group B streptococcal disease occurred in term infants, of which 13.4% of cases were attributed to missed screening among mothers and 61.4% of cases were born to women who had tested negative for GBS before delivery. These results indicate that the recommendations for universal screening have been successfully implemented. Further improvement toward eradication of GBS colonization and disease may involve universal screening in conjunction with rapid diagnostic technologies or other novel approaches, including DNA techniques for the identification of GBS. Given the complications and potential limitations associated with maternal intrapartum prophylaxis, however, vaccines may be the most effective means of preventing neonatal GBS disease. Developing a universal vaccine has proved to be a daunting task, though, because of the variability of serotypes in diverse populations and geographic locations. Application of modern technologies, such as those involving proteomics and genomic sequencing, may hasten these efforts toward development of a universal vaccine against GBS.¹⁰⁵

Bacterial vaginosis (BV) is yet another infection associated with an increased risk for preterm labor and delivery. It is a common lower genital tract infection found in approximately 20% to 40% of African-American women and 10% to 15% of white women. This condition is a clinical syndrome characterized by a decrease in the normal vaginal lactobacilli-dominant microflora and a compensatory predominance of bacteria such as Gardnerella vaginalis, Prevotella, Bacteroides, Peptostreptococcus, Mobiluncus, Mycoplasma hominis, and Ureaplasma urealyticum. Characteristically, patients with symptomatic BV complain of a watery, homogeneous grayish discharge with a fishy amine odor.

As above, BV has been associated with increased risk for preterm birth, particularly early preterm deliveries. Nearly 40% of early spontaneous preterm births, especially among African-American women, may be attributable to BV. Two randomized clinical trials demonstrated that treating BV in patients who are at high risk for preterm delivery resulted in substantial reductions in preterm birth rates.81,132 Importantly, though, treatment of BV appears to be effective only in this high-risk population. Data from studies involving treatment of low-risk women with asymptomatic BV failed to demonstrate a reduction in preterm delivery rates.^23,120

The majority of women with BV never manifest any signs or adverse outcomes related to the infection, and it is likely that BV is only a surrogate marker for a more important, presently unrecognized condition that may be the cause of the preterm birth. For example, one compelling theory is that BV is a marker for occult upper genital tract infection. According to this theory, premature births attributed to BV may instead be related to an associated upper genital tract infection, whereas treatment of BV in high-risk patients may be effective in preventing preterm delivery by inadvertently treating these unrecognized infections.

Many other maternal infections or colonizations have been reportedly associated with preterm birth. One of the difficult questions to address is whether these relationships are causal or associative. Gonorrhea, chlamydia, trichomonas, syphilis, and other genital pathogens are more frequently found in women who have a spontaneous preterm birth. For example, gonorrhea, chlamydia, and syphilis have all been associated with a twofold increased risk for preterm delivery, and trichomonas has been associated with a 1.3-fold increased risk. Furthermore, many other sexually transmitted infections (e.g., human immunodeficiency virus, hepatitis B, and genital herpes simplex virus) have been associated with an increased risk for spontaneous preterm birth in some, but not most, studies. Importantly, the women affected by these infections often have other risk factors for preterm birth (e.g., low socioeconomic status, malnutrition, smoking, substance abuse, bacterial vaginosis). These confounding variables make it difficult to establish causality.

Other Risk Factors

A variety of other factors have been associated with an increased risk for preterm labor (see Box 20-1), including the following:

Classic Predictors

Cervical change is the first of the classic clinical predictors of preterm birth. Mortensen and co-workers divided 1300 women into high- and low-risk groups on the basis of various risk factors, evaluating these patients for evidence of cervical change at 24, 28, and 32 weeks’ gestation.¹³⁴ Cervical change in the low-risk group was an extremely poor predictor of preterm delivery, with a positive predictive value of only 4%.¹³⁴ In the high-risk group, the predictive value was 25% to 30%, with a sensitivity of 65%, indicating improved, but far from perfect, predictability in this group.¹³⁴

Another potentially important clinical factor is the presence of uterine contractions. An association has been noted between the reported presence of contractions and preterm delivery. Despite this, monitoring of contractions (i.e., using a home tocometer) is not clinically useful in defining the population of women at a particularly high risk for spontaneous preterm birth.⁸⁹ Use of home uterine monitoring is therefore not recommended.

Beginning in the early 1980s, attempts were made to combine these various factors into a risk scoring system to determine which patients were at jeopardy for preterm delivery. Creasy and co-workers combined socioeconomic factors such as age, height, weight, previous medical history, smoking, work habits, and aspects of the current pregnancy into a risk scoring system.³⁵ A total of 10 points or more indicated a high risk of preterm delivery (Table 20-4). The initial study held promise, with a positive predictive value of 38%, but subsequent studies had much lower positive predictive values in the range of 18% to 22%. One of the limitations of the Creasy risk scoring system is the emphasis placed on previous preterm deliveries, a factor which by itself elevates a patient into a high-risk category.³⁵ Half of all preterm deliveries happen in primigravidas, a population whose obstetric histories lower the predictive value even further according to this system.³⁵ Overall, classic predictors have had limited success in predicting preterm delivery, and no well-performed study to date has demonstrated a reduction in preterm birth rates using cervical change assessment, contraction monitoring, or risk scoring systems.

Biochemical Predictors

The biochemical processes leading to the initiation of either term or preterm labor are complicated, such that these pathways have not been fully established in humans. Despite this limitation, important insights into the pathophysiology of spontaneous preterm labor have helped to identify various biochemical markers that may predict preterm delivery (see Figure 20-3).⁶²

Perhaps one of the most important biochemical markers identified to date is fetal fibronectin.¹⁵⁶ This glycoprotein is found within the extracellular matrix that surrounds the extravillous trophoblast at the uteroplacental junction. Clinically, it serves as a prototypic example of a marker of choriodecidual disruption. Fetal fibronectin is usually absent from cervicovaginal secretions starting from the 20th week of gestation until near term. Detection of elevated cervicovaginal levels of fetal fibronectin has therefore been strongly associated with an increased risk for preterm delivery in high-risk patients.^57,133,150 Sensitivities in the 80% to 90% range with positive predictive values of 30% to 60% have been reported. For example, Goldenberg and associates demonstrated that, in asymptomatic women at 24 weeks’ gestation, elevated cervicovaginal fetal fibronectin levels (greater than 50 ng/mL) were strongly associated with subsequent spontaneous preterm delivery, with ORs of 59.2 (95% CI, 35.9-97.8) for delivery before 28 weeks’ gestation, 39.9 (95% CI, 25.6-62.1) for delivery less than 30 weeks’ gestation, and 21.2 (95% CI, 14.3-62.1) for delivery less than 32 weeks’ gestation.⁵⁷ Additionally, fetal fibronectin levels have been correlated with other risk factors for preterm birth, including cervical shortening, bacterial vaginosis, elevated IL-6 levels, and peripartum infection.^55,57 Further supportive data comes from Tanir and associates, who prospectively confirmed the clinical value of cervicovaginal fetal fibronectin, showing that a positive fetal fibronectin test was associated with an increased likelihood of preterm delivery in women with signs and symptoms of preterm labor.¹⁸⁰ Quantitative fetal fibronectin testing may be even more useful. For example, Abbott and colleagues evaluated women with preterm labor symptoms, demonstrating an increasing positive predictive value for preterm birth of 19%, 32%, 61%, and 75% with increasing fetal fibronectin thresholds of 10 ng/mL, 50 ng/mL, 200 ng/mL, and 500 ng/mL respectively.¹ These data clearly suggest that fetal fibronectin is one of the most potent markers for spontaneous preterm delivery identified to date, a fact related to the theory that infection of the upper genital tract with disruption of the choriodecidual interface is a feature common to many cases of preterm birth. Clinically, most experts rely primarily on the negative predictive value of fetal fibronectin, using negative results to justify management of women in an ambulatory fashion, thereby limiting inpatient hospitalization to only women at the highest risk for preterm birth.

Estriol is another potential biochemical marker that may be of use in predicting preterm delivery.¹⁵⁶ Estriol is a unique hormone of pregnancy that is produced almost entirely by the trophoblast using precursors derived from the fetal adrenal gland and liver. Levels of this hormone rise throughout pregnancy, with a characteristic exponential increase 2 to 4 weeks before the spontaneous onset of labor at term. Interestingly, patients undergoing induction of labor at term fail to demonstrate this increase in estriol, indicating that it plays a role in the onset of spontaneous labor. This finding has led to the theory that salivary estriol levels may be used to identify patients at risk for preterm delivery.¹²¹ Studies have subsequently shown that detection of an early estriol surge may identify patients at risk for preterm labor and delivery, but estriol appears to be a better marker for late, rather than early, spontaneous preterm birth. As with many of the other markers, no reduction in the preterm birth rate has been demonstrated with the use of these assays, but they may play a role in limiting hospitalization of women who are ultimately unlikely to deliver prematurely.

A variety of other serologic, salivary, cervical, and amniotic fluid markers have been evaluated as predictors for preterm delivery, including alpha-fetoprotein, human chorionic gonadotropin, human placental lactogen, corticotropin-releasing factor, C-reactive protein, alkaline phosphatase, ferritin, placental isoferritin, progesterone, estradiol, matrix metalloproteinase, IL-6, TNFα, and granulocyte colony-stimulating factor. Each of these markers has been shown to have a modest correlation with spontaneous preterm delivery.

Corticotropin-releasing hormone (CRH) is a placental peptide produced during the second and third trimesters. This peptide appears to play a role in the initiation of parturition, with elevated levels noted weeks before the onset of preterm labor.¹⁵⁹ Tropper and co-workers demonstrated that CRH levels were significantly higher in maternal serum and umbilical cord serum in patients who delivered preterm than in gestational age-matched controls who delivered at term.¹⁸⁵ Although at least two studies later refuted this finding, observing that CRH levels were not predictive of preterm birth or PPROM, they showed that CRH binding proteins decreased near term, suggesting that it is the bioavailability of CRH that may specifically be related to the onset of labor.^19,109 Using an arbitrary cutoff of 1.9 multiples of the median to predict preterm birth, Leung and colleagues showed that elevated CRH had a sensitivity of 72.9%, a specificity of 78.4%, a positive predictive value of 3.6%, and a negative predictive value of 99.6%.¹⁰⁹

A variety of degradative enzymes has been studied with regard to their ability to predict preterm birth. Serum collagenases are potential markers. IL-1 stimulates cervical, decidual, and other cells to produce various collagenases, which then act to break down the collagen matrix of the cervix. Serum levels of these collagenases remain relatively constant until the onset of labor, when a marked increase occurs. This increase appears to be exaggerated in women who deliver prematurely, with an up to eightfold greater elevation in preterm births.

In addition to the collagenases, the metalloproteinases and their inhibitors have received increasing interest in regard to their role in predicting preterm birth. In particular, elevated metalloproteinase-2 (MMP-2) and MMP-8 levels have been associated with preterm labor, especially PPROM, likely owing to their role in membrane weakening through degradation of the chorioamnion basement membrane.12,51 On the other hand, MMP-9 levels do not seem to be useful in predicting which patients are at risk for preterm labor because they increase with, but not before, labor.¹⁸⁶

Granulocyte elastase is another granulocyte degradative enzyme that may play a role in parturition. Granulocytes are stimulated by IL-8 (an inflammatory marker that is elevated in amniotic fluid in labor) and then degraded by granulocyte elastase. The activity of granulocyte elastase has been shown to be increased in the cervix in both term and preterm labor, suggesting that it may be involved in cervical ripening and degradation of fetal membranes.

Given the association of occult upper genital tract infection with early spontaneous preterm birth, a variety of serum, amniotic fluid, and cervicovaginal inflammatory markers has also been evaluated as potential markers for the prediction of spontaneous preterm delivery. Both serum and cervical IL-6 levels are significantly elevated at 24 weeks’ gestation in women with subsequent spontaneous preterm birth at less than 35 weeks’ gestation.55,139 Rizzo and colleagues further characterized the inflammatory milieu of the cervix and amniotic fluid, showing that levels of several cytokines (IL-1, IL-6, TNFα) were elevated in a series of patients with preterm labor and intact membranes, a finding that was significantly associated with the presence of intra-amniotic infection.¹⁶⁰ In that study, cervical IL-6 was the most potent marker for infection, with an RR of 7.7 (95% CI, 3.5-17.8) in the presence of elevated levels (greater than 410 pg/mL) of this protein.¹⁶⁰ Serum granulocyte colony-stimulating hormone, ferritin, and lactoferrin are examples of other inflammatory markers that may be elevated in asymptomatic women who subsequently deliver prematurely.

To safely implement the use of these biochemical markers clinically, tests with improved sensitivity and specificity are needed. Of particular interest is the potential for a test combining a number of these assays. For example, Goldenberg and co-workers demonstrated that the use of a serum multiple marker test may enhance the predictive value of the presently available serologic markers for spontaneous preterm birth.⁵⁹ Again, further studies are needed to address these issues before they are widely implemented.

Ultrasound Predictors

Regardless of the gestational age at delivery, cervical changes begin to occur approximately 3 to 4 weeks beforehand. Detection of these changes previously involved digital examination only, a test that is problematic because of such possible factors as the introduction of infection, interobserver differences, and an inability to evaluate the internal cervical os when the external os is closed. In fact, detection of these changes digitally may be possible only late in the process, limiting the clinician’s ability to initiate potential treatments. Ultrasonography therefore has several potential benefits, allowing for a more objective approach to examination of the cervix with visualization of changes earlier in the process of parturition.¹⁴⁶ Cervical changes visible via ultrasonography include cervical length, dilation of the internal cervical os, dynamic changes that occur in the cervix with time, the presence of intra-amniotic debris, and cervical funneling or wedging.

Several studies have compared digital examination with ultrasonographic assessment of the cervix. For example, Sonek and colleagues assessed patients at risk of preterm labor, reporting that digital examination tended to underestimate cervical length by about 1 to 1.5 cm.¹⁷⁶ Furthermore, in 45% of these patients, ultrasound detected changes that were not yet identified using digital examination.¹⁷⁶ Subsequent reports have confirmed that ultrasound changes are more predictive of preterm delivery than digital examination.⁶⁴ Alterations in cervical length and internal os dilation tend to occur approximately 10 weeks before delivery, whereas the more dynamic changes occur only 4 weeks before delivery.

Transvaginal ultrasound is far superior to transabdominal ultrasound. Transabdominal ultrasound is technically more difficult because the distance between the transducer and the cervix is relatively long, particularly in obese patients. Transabdominal assessment of cervical length and internal os dilation may also be affected by bladder filling and emptying. Finally, fetal parts can cause acoustic shadowing of the cervix when looking abdominally. Transperineal ultrasonography is also effective in the assessment of cervical length, revealing findings that correlate well to those obtained via digital cervical examination and/or transvaginal imaging.

In a study by Smith and colleagues, low-risk patients were observed serially via transvaginal ultrasonography, showing that the average cervical length of 37 mm remained stable between 10 and 30 weeks of gestation and then began to decrease slightly after week 32.¹⁷⁵ In general, most additional studies reported lengths of more than 30 mm as normal, although this estimate depends upon the population studies and the gestational age at which the length is measured. For example, Anderson and co-workers examined low-risk patients using transvaginal ultrasound and digital examination prior to 30 weeks’ gestation, showing that a cervical length of less than 39 mm (50th percentile) had a sensitivity of 76% and a specificity of 59% in predicting preterm labor.⁶ Other studies used cutoff points for cervical shortening ranging from 17.6 to 30 mm, demonstrating sensitivities of 72% to 100% and positive predictive values of 55% to 76%.

Iams and colleagues, in a large multicenter trial, provided the clearest insights into the relationship between cervical length and spontaneous preterm delivery.⁸⁸ In this prospective study of 2915 women with a singleton pregnancy, transvaginal sonographic determination of cervical length was obtained at 24 weeks’ and again at 28 weeks’ gestation. A cervical length less than the 10th percentile (26 cm) at 24 weeks was significantly associated with an increased risk for spontaneous preterm birth at less than 35 weeks’ gestation (RR, 6.19; 95% CI, 3.84-9.97). An inverse relationship between cervical length and the rate of preterm delivery was noted in this study. The investigators concluded that the risk of spontaneous preterm birth is increased in women who are found to have a short cervix by transvaginal ultrasonography during pregnancy.

Andrews and associates further evaluated the use of cervical ultrasonography to determine whether early cervical changes may predict spontaneous preterm delivery.⁹ In this study, cervical length measurements were obtained prior to 20 weeks’ gestation in women with a history of a previous spontaneously preterm birth. In this cohort of patients, the presence of either a short cervical length (<22 mm consistent with <10th percentile for the study population) or cervical funneling was significantly associated with an increased risk for spontaneous preterm delivery prior to 35 weeks’ gestation. Owen and colleagues furthered these insights, evaluating the use of cervical length assessment in a cohort of high-risk women screened between 16 and 18 weeks’ gestation.¹⁴⁵ In this study, a cervical length of 25 mm or less was significantly associated with increased risk for spontaneous preterm delivery prior to 35 weeks (RR, 3.4; 95% CI, 2.1-5.0). It should also be noted that, in contrast to singleton pregnancies, cervical lengths differ significantly in higher-order gestations, probably reflecting a greater risk for subsequent preterm delivery.^149,153

Dilation of the internal cervical os may also be assessed via transvaginal ultrasonography. Increased dilation appears to correlate with an increased risk for preterm birth. For example, dilation of greater than 5 to 7 mm has been shown to carry a sensitivity of 70% and a positive predictive value of 33.3% for preterm delivery.⁶⁴

The dynamic nature of the cervix may also be evaluated. Parulekar and associates studied a cohort of patients with a reported history of cervical incompetence. In nearly 30% of these patients, dynamic findings were noted in which the internal os changed from 0-mm dilation to 42-mm dilation with no alteration of cervical length.¹⁴⁹ These changes occurred over a period of 1 to 3 minutes. Interestingly, 50% of these patients with dynamic cervical changes delivered at less than 37 weeks’ gestation. These dynamic findings may be influenced by such factors as the patient’s respiration, Valsalva maneuver, fundal pressure, and amniotic fluid levels.

Ultrasound assessment of the cervix represents a relatively easy way to identify patients who may be at higher risk for spontaneous preterm delivery, but the positive predictive value of this test is relatively low, and universal screening of all pregnant women remains controversial.⁹ Cervical length measurements have therefore been combined with biochemical parameters in an attempt to improve the specificity and sensitivity of these individual screening tests. For example, Defranco and colleagues performed a systematic review of nine studies that included a combination of fetal fibronectin testing and cervical length measurement to evaluate women with preterm labor symptoms.⁴² They concluded that this combined screening approach appears to be useful in identifying women at risk for preterm birth and guiding acute management in such situations.⁴² Further research is needed to determine if these combined tests will prove to be effective.