Predicting spontaneous preterm birth

The literature on predicting spontaneous preterm delivery deals with two areas. The first is predicting preterm delivery in asymptomatic patients. This is the risk you tell a woman in mid pregnancy when she asks you if she will deliver early. The second concerns symptomatic patients. These are women who report contractions between 24 and 36 weeks of gestation. To separate the two in the literature is difficult because both sets of data involve the same terminology and tests. The reader must therefore be aware of the populations being studied. The predictive values for any test are directly related to the incidence of the disease in the population studied. Therefore, a test that is predictive in a higher risk group may be of little use in the general population.

The initial attempts to predict preterm delivery in asymptomatic patients involved the use of risk factors in patients. These produced the Creasy risk scoring system.¹ This system was able to predict two thirds of those who delivered prematurely in a middle class population. However, when the scoring system was used in a higher risk indigent population the sensitivity was less impressive.²

The next area of study concerned the use of transvaginal ultrasonography to assess the length of the cervix. Iams et al evaluated the cervical length of 2915 women at 22 to 24 weeks' gestation and noted that as the cervix shortened, the risk of preterm delivery increased.³ Unfortunately, there was no cervical length below which all women delivered prematurely and no cervical length above which none of the women delivered early. The best you can do is to note an increased risk for early delivery with a shortened cervix.

The most recent attempt to predict increased risk for preterm delivery in asymptomatic patients is the use of the fetal fibronectin test. The presence of this substance in vaginal secretions is correlated with an increased risk of spontaneous preterm delivery.⁴ Honest et al (p 301) evaluated 28 studies of asymptomatic women and calculated that the likelihood ratio for spontaneous preterm delivery before 34 weeks' gestation with a positive fetal fibronectin test was 4.01 (95% confidence interval 2.93 to 5.49).⁵ Although this is statistically significant, it currently has little clinical value. The argument against screening asymptomatic patients is best made by Goldenberg et al: “Less clear is whether fibronectin screening of the general population, or even high risk groups, will result in an improved outcome for the woman or her fetus. Because there is no intervention to date that has demonstrated clear benefit in preventing preterm birth, the knowledge of which women are at increased risk for this outcome may have little practical benefit. In fact, performing interventions that are of no value in women thought to be at high risk may make outcomes worse, not better, and will certainly increase costs.”⁴ This argument also applies to the other modalities mentioned here.

The evaluation of symptomatic patients for preterm delivery used to involve simply the digital examination of the cervix and monitoring of contractions. This method has been shown to have a poor sensitivity and specificity for predicting preterm delivery.⁶,⁷ The difficulty with digital examination of the cervix is the lack of consistency between examinations and examiners. The most accurate and reproducible method of cervical evaluation is transvaginal ultrasonography.⁸,⁹ Iams et al, as well as others, have noted that if the cervical length is more than 30 mm the patient is most likely not in preterm labour.⁹ If the cervical length is less than 20 mm, the patient is more likely to be in preterm labour.

The next test used in predicting preterm birth is the fetal fibronectin test. Honest et al did a meta-analysis of 40 previous studies in symptomatic women and noted that the likelihood ratio for preterm delivery in 7–10 days when the test had positive results was 5.42 (4.36 to 6.74).⁵ Despite the increased likelihood of preterm delivery most of the women remained pregnant 10 days after the test. Of more interest, the negative likelihood ratio of preterm delivery was 0.25 (0.20 to 0.31) with negative results on the test. This is consistent with a study by Peaceman et al.¹⁰ In the group with a negative result 99.5% were still pregnant seven days later and 99.2% were still pregnant 14 days later.

Thus we have no reliable method of confirming preterm labour. We do have two tests that seem to rule out labour. If a patient presents to the labour room and has a cervical length more than 30 mm or a negative fetal fibronectin she is most likely not in preterm labour. From a clinical standpoint we can use these tests to avoid unnecessary treatment and perhaps to give corticosteroids in a more judicious manner. From a research standpoint we can better select our patient populations when evaluating modalities to decrease the rate of preterm delivery.

Papers p 301