Women's health care has engendered innovations with enormous impact on the practice of medicine and on modern life in general--such innovations as hand-washing, the Papanicolaou smear, and oral contraceptives.
We have the technology to image the gravid cervix with precision, but can this really help to prevent preterm labor?
Women's health care has engendered innovations with enormous impact on the practice of medicine and on modern life in general--such innovations as hand-washing, the Papanicolaou smear, and oral contraceptives. In contrast to these shining contributions stand less lustrous examples of hastily adopted interventions with unproved benefit and potential risk. The most ignominious cases include the use of thalidomide and diethylstilbestrol during pregnancy.
Given the history of our specialty, practitioners should view any technology introduced into practice before the completion of rigorous testing with great caution. Let us critically examine the evidence for and against the use of cervical ultrasonography for the prediction of preterm labor.
Testing a Test
Any medical test should meet several conditions before it is accepted into practice. First, the test should be valid. Validity is commonly expressed in terms of sensitivity, specificity, and positive and negative predictive values. Second, the test must be acceptable in terms of risk, cost, and patient convenience. Third, it should represent an improvement over existing alternatives. Fourth, the test must have clinical utility; its result should aid the practitioner in treating or preventing illness. Even conceding that it is acceptable in terms of risk, cost, and convenience, cervical ultrasonography to prevent preterm labor fails to meet the other three of these four criteria.
In a large, prospective, multicenter, blinded study conducted by the National Institute of Child Health and Human Development (NICHD) Maternal Fetal Medicine Unit Network, more than 2500 healthy pregnant women under-went cervical ultrasonography at 24 and 28 weeks gestation.1 The study found an inverse relationship between cervical length and the risk of subsequent preterm birth. Other studies of symptomatic and asymptomatic patients have shown similar results.2-4
Thus, there is little doubt about the biologic reality: Shorter cervices are associated with shorter gestations. But how should knowledge of a patient's cervical length be used clinically, and is such information more useful when obtained ultrasonographically than by physical examination?
According to data from the NICHD study (Table), even when the ultrasonographic test uses a cutoff of 20 mm to define short cervical length, it fails to identify over two thirds of the deliveries that occur before 35 weeks gestation (sensitivity of 31%). Furthermore, only one of six patients with a positive test result actually delivers prematurely (positive predictive value of 17%). Thus, ultrasonographic measurement of cervical length cannot be considered a good test for predicting preterm birth, especially considering the treatment options available to the clinician.
Even conservative therapy (i.e., bed rest) entails enormous economic, social, medical, and emotional costs.5 More aggressive interventions (e.g., cerclage) to treat patients with these ultrasonographic findings would probably increase costs and morbidity. Furthermore, analysis of the data demonstrates that ultrasonography does not substantially improve on digital examination as reflected in either the Bishop Score or the 'Cervical Score' (Table).
A separate study of asymptomatic patients demonstrated slightly greater sensitivity and specificity of ultrasonography over digital assessment for prediction of delivery before 37 weeks.2 However, the small number of patients (n = 113) and the large overlap in confidence intervals render it impossible to determine whether one technique was superior to the other. Sensitivity was 47% (CI, 24% to 72%) for ultrasonography versus 36% (CI, 14% to 64%) for digital examination, and specificity was 84% (CI, 75% to 91%) versus 76% (CI, 65% to 85%), respectively (I calculated the confidence intervals). Whether these data are maintained at the more meaningful 35-week cutoff is also unclear.
Probe vs. Digit
The ultrasound probe is supposedly capable of assessing structures that the examining finger cannot specifically, the supraforniceal portion of the cervix and the internal os. Even if this assumption is true, it does not follow that more information necessarily equals better care. Furthermore, in contrast to the purported advantages of ultrasonography, digital examination provides superior information about consistency, station, and the shape of the lower uterine segment. The important question is whether ultrasonography is better than digital examination at predicting preterm birth--and the answer is no.
Studies comparing ultrasonographic cervical length with digital measures other than cervical length (e.g., dilatation, effacement, Bishop Score) are seriously flawed.1,3,4,6 Effacement of the cervix is an imprecise measure because of the variability among patients in the length of the uneffaced cervix and because cervical shortening may occur without appreciable effacement.
Digital examiners tend to underestimate true cervical length by an average of 10 to 14 mm.7,8 Although this reflects a relative inaccuracy in digital assessment, it does not demonstrate a lack of value. The fact that digital measurements consistently underestimate true cervical length does not mean they do not correlate as well as ultrasonographic determinations with the likelihood of preterm birth--as demonstrated by the NICHD study.1
Only one randomized study compares the impact of ultrasonography versus digital pelvic examination on the incidence of preterm birth.9 Fifty-seven women at high risk for preterm delivery by Creasy Score were randomized to weekly digital or ultrasonographic assessment from 20 to 37 weeks gestation. The major differences in outcome between the two groups were higher rates of diagnosis of preterm labor with ultrasonographic management (62% versus 25%), greater use of tocolytic drugs (55% versus 21%), and longer hospital stays (8.1 days versus 2.3 days). Neonatal outcomes in the two groups were similar.
Among the flaws of this study were the use of abdominal rather than vaginal ultrasonography, nonexclusion for cerclage (used in more than 25% of patients), nonstandardization of intervention in cases of threatened preterm delivery, use of home uterine monitoring, and the small number of subjects. Nonetheless, these results should discourage practitioners from utilizing ultrasonography rather than other assessments to predict preterm birth.
The evidence presented here demonstrates that ultrasonography is not very good for predicting preterm birth and that it is no better than physical examination. Other attempts have been made to develop scoring systems (based on history and physical assessment) to predict preterm delivery. One such system, the Creasy Score, has sensitivity, specificity, and predictive values similar to those for ultrasonographic data, but these methods have not been compared directly. Until clear evidence of the superiority of ultrasonography is produced, it should be reserved for use in properly designed studies. Meanwhile, efforts should be made to enhance the utility of the digital examination in these studies by replacing effacement with length as a descriptor of the unlabored cervix.
A medical test is only as good as the intervention it spawns. In our zeal to help patients, let us not rush to diagnose a disorder for which we have no consistently effective treatment. Serious doubts remain about the efficacy of bed rest, cerclage, and tocolysis for the prevention or management of preterm birth; all of these therapies have major theoretical and practical drawbacks. Even if we could devise a test that reliably predicts preterm birth, could we then safely prevent it from occurring? Ah, but there is a topic for a different debate.
Read the Opposing View:
1. Iams JD, Goldenberg RL, Meis PJ, et al. The length of the cervix and the risk of spontaneous premature delivery. N Engl J Med. 1996;334:567-572.
2. Andersen HF, Nugent CE, Wanty SD, Hayashi RH. Prediction of risk for preterm delivery by ultrasonographic measurement of cervical length. Am J Obstet Gynecol. 1990;163:859-867.
3. Crane J, Van den Hof M, Armson B, Liston R. Transvaginal ultrasound in the prediction of preterm delivery: singleton and twin gestations. Obstet Gynecol. 1997;90:357-363.
4. Gomez R, Galasso M, Romero R, et al. Ultrasonographic examination of the uterine cervix is better than cervical digital examination as a predictor of the likelihood of premature delivery in patients with preterm labor and intact membranes. Am J Obstet Gynecol. 1994;171:956Ã964.
5. Goldenberg R, Cliver S, Bronstein J, et al. Bed rest in pregnancy. Obstet Gynecol. 1994;84:131-136.
6. Iams JD, Paraskos J, Landon MB, et al. Cervical sonography in preterm labor. Obstet Gynecol. 1994;84:40-46.
7. Jackson GM, Ludmir J, Bader TJ. The accuracy of digital examination and ultrasound in the evaluation of cervical length. Obstet Gynecol. 1992;79:214-218.
8. Sonek JD, Iams J, Blumenfeld M, et al. Measurement of cervical length in pregnancy: comparison between vaginal ultrasonography and digital examination. Obstet Gynecol. 1990;76:172-175.
9. Lorenz RP, Comstock CH, Bottoms SF, Marx SR. Randomized prospective trial comparing ultrasonography and pelvic examination for preterm labor surveillance. Am J Obstet Gynecol. 1990;162: 1603-1610.
10. Newman RB. The Preterm Prediction Study: comparison of the Cervical Score and Bishop Score for the prediction of spontaneous preterm birth, abstract no. 293. San Diego, CA: Society for Gynecologic Investigation; 1997.
Originally published in The Female Patient -- April, 1998 Â© Copyright, 1998 Quadrant Publishing, All Rights Reserved Reprints are not allowed without the expressed written consent of Quadrant Publishing.