UltrasoundAssessment of theCervixVINCENZO BERGHELLA, MD,* GEORGE BEGA, MD,*JORGE E. TOLOSA, MD,* andMICHELE BERGHELLA, MD†

*Division of Maternal–Fetal Medicine, Department of Obstetricsand Gynecology, Jefferson Medical College of Thomas JeffersonUniversity, Philadelphia, Pennsylvania; †Department of Obstetricsand Gynecology, Universitá degli studi “L’Aquila,” L’Aquila, Italy

IntroductionUltrasound of the cervix during pregnancyhas been the focus of much research in thelast few years, and significant advances havebeen made in understanding the proper roleof this procedure. In this chapter we will re-view the technique of transvaginal cervicalassessment and discuss its possible role inpredicting and preventing preterm birth(PTB).

Ultrasound of theCervix—TechniquesInitial attempts at evaluating the cervix usedtransabdominal ultrasound (TAU). Unfortu-nately, this technique was found to be imper-

fect, because of (1) fetal parts obscuringthe cervix, especially after 20 weeks, (2)the requirement of bladder filling, whichcan elongate the cervix and mask funneling,and (3) long distance from the probe to thecervix.

Translabial (also known as transperineal)ultrasound, first used in France in the early1980s, proved to be more useful. This tech-nique involves having the patient lie on tablewith the hips and knees flexed, while agloved transducer is positioned on the peri-neum in a sagittal orientation between thepatient’s labia majora. Elevation of the pa-tient’s hips with a cushion is sometimes usedto improve visualization. Compared withTAU, this technique is not impaired by ob-struction by fetal parts, and does not requirebladder filling, achieving close to 100% vi-sualization. Other advantages of this tech-nique are that the transducer is closer to thecervix, but does not enter the vagina (so nopressure can be exerted on the cervix), it

Correspondence: Vincenzo Berghella, MD, Division ofMaternal–Fetal Medicine, Department of Obstetricsand Gynecology, Jefferson Medical College of ThomasJefferson University, Philadelphia, Pennsylvania.E-mail:vincenzo.berghella@jefferson.edu

CLINICAL OBSTETRICS AND GYNECOLOGYVolume 46, Number 4, 947–962© 2003, Lippincott Williams & Wilkins, Inc.

CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 46 / NUMBER 4 / DECEMBER 2003

947

does not require an additional transducer,and it is well accepted by patients. The maindrawback of the transperineal approach isthat gas in the rectum can hamper visualiza-tion of the cervix, especially the external os.

The first studies of the human cervix us-ing transvaginal ultrasound (TVU) also dateback to the 1980s. The technique shares theadvantages of Translabial ultrasound, butthe probe is even closer to the cervix, and theproblem of obscuring bowel gas is elimi-nated. It has thus become the preferred, goldstandard method of evaluating the cervix inmost clinical settings. Current recommenda-tions for the performance of TVU of the cer-vix are as follows:

1. Have the patient empty her bladder;2. Prepare the clean probe covered by a con-

dom;3. Insert the probe (probe can be inserted by

patient for more comfort);4. Place the probe in the anterior fornix of the

vagina;5. Obtain a sagittal view of the cervix, with the

long axis view of echogenic endocervicalmucosa along the length of the canal;

6. Withdraw the probe until the image isblurred and reapply just enough pressure torestore the image (to avoid excessive pres-sure on the cervix which can elongate it);

7. Enlarge the image so that the cervix occu-pies at least 2/3 of the image, and externaland internal os are well seen;

8. Measure the cervical length from the inter-

nal to the external os along the endocervicalcanal (Fig. 1);

9. Obtain at least three measurements, and rec-ord the shortest best measurement in milli-meters;

10. Apply transfundal pressure for 15 seconds,and record any changes in cervical length orfunneling.

For best results, the internal os should beeither flat or at an isosceles angle with re-spect to the uterus, the whole length of thecervix should be visualized, a symmetricimage of the external os should be obtained,and the distance from the surface of the pos-terior lip to the cervical canal should beequal to the distance from the surface of theanterior lip to the cervical canal. Thereshould not be any increased echogenicity inthe cervix (a sign of excessive pressure).

While TVU of the cervix is usuallystraightforward, there is some anatomic ortechnical difficulty encountered in aboutone fourth of patients. For this reason, somehave recommended that a sonographer besupervised for his or her first 50 procedureswhile expertise is being acquired.

Cervical AssessmentDifferent cervical parameters have beenevaluated as predictors of PTB. Most com-monly, cervical length (CL), as measuredfrom the internal to the external os along theendocervical canal, is measured (Fig. 1). Ifthe cervical canal is curved (defined in the

FIGURE 1. Transvaginal ultrasonography of closed normal cervix (left)and of a short cervix with significant funneling (right)

948 Berghella et al

Maternal Fetal Medicine Network study as adeviation of the canal of >5 mm from astraight line connecting the external and in-ternal os), the CL can be either traced or thesum of two straight lines that essentially fol-low the curve can be used.1 It should benoted that a short CL is always straight, andthe presence of a curved cervix usually sig-nifies a CL >25 mm and is therefore a reas-suring finding. If the cervical canal is closed,CL is probably the only parameter that needsto be measured.

In about 25% to 33% of high-risk pa-tients,1,2 the internal os is open. In this case,the open portion of cervix (funnel length)and internal os diameter (funnel width) canbe measured (Fig. 2). Percent funneling isdefined as funnel length divided by total CL.Total CL is equivalent to the sum of funnellength and functional CL. Functional CL inthis case is the closed portion of the endocer-vical canal only. Functional CL is the sono-graphic CL used for calculations and predic-tions, and the term cervical length (CL), ifnot otherwise specified, refers to the func-tional CL.

If funneling is present, the shape can berecorded. Zilianti has described a continu-ous process of funneling, going from a nor-

mal T shape, to Y, then V, and finally a Ushape.3 It appears that U-shaped funneling ismore likely to be associated with PTB com-pared with a V-shaped funnel;4 however,these distinctions are somewhat subjective.

Uterine segment contractions may mimicthe appearance of funneling of the internalcervical os. In such cases, there is roundedmyometrium around the cervix and a normalcervix distal to the contraction. Carefulevaluation of the apparent funneling withendovaginal ultrasound over several min-utes should resolve any question of the mor-phology of the upper cervical canal. In someinstances, the depth of a true funnel may bedifficult to quantify, since the funneled por-tion may merge with the lower uterine seg-ment and the characteristic notch that de-picts the border between lower uterine seg-ment and the cervix may be flattened.Funneling has been reported to have higherinter-observer variability among examinersand different centers than CL.5

In spite of these difficulties, funneling hasbeen reported to have better or similar pre-dictive accuracy for PTB than CL.5,6 In onehigh-risk population, minimal funneling(<25%) noted between 14 and 22 weeks wasnot associated with a significant increase inPTB, while moderate (25% to 50%) and se-vere (>50%) funneling were associated witha 50% or more probability of PTB.2 The factthat <25% funneling is not associated withan increased risk of PTB is important, sincethis is a common finding that should notraise alarm or result in intervention.

In our 1999 study, when we consideredeither a short CL or funneling as abnormal,compared with a CL <25 mm alone, the sen-sitivity for PTB was increased (from 61%–74%) without major changes in specificity,positive, and negative predictive values.7

Another dataset1 from the NICHD MFMUNetwork also showed that the presence offunneling when the CL is >25 mm is associ-ated with a high risk of later development ofa short CL and eventual PTB.

In less than 5% of TVU, CL may changedynamically during the course of a 5- to 10-

FIGURE 2. Schematic of TVU cervicalmeasurements

Ultrasound Assessment of the Cervix 949

minute examination, and in some cases fun-neling of the upper cervical canal may ap-pear and resolve. Similarly, the cervix mayshorten in response to transfundal pressure(TFP) in about 5% of cases. When changesoccur, the shortest CL should be recorded. Inmost of the cases in which the cervix short-ens spontaneously or in response to TFP, it isalready abnormal at baseline. There is con-flicting evidence concerning whether TFPincreases the screening potential of TVU. Inour study, only 1 of 9 patients with an ini-tially normal cervix who had abnormalitiesafter the application of TFP delivered pre-term.7 Adding response to TFP as a screen-ing criteria for PTB did not significantly in-crease predictive accuracy. In contrast,Owen1 showed that dynamic changes, eitherspontaneous or after TFP, significantly im-proved the predictive accuracy of TVU forPTB.

Many parameters other than CL and pres-ence or absence of a funnel have been stud-ied, including funnel width, funnel length,6

anterior and posterior cervical width, cervi-cal angle, cervical position (horizontal ver-sus vertical), lower uterine segment thick-ness, vascularity, visibility of chorion am-nion at internal os, cervical index (funnellength +1/functional length), etc., but noneof these has proven more reliable or predic-tive than CL.

Prediction of Preterm BirthThere have been numerous studies evaluat-ing the usefulness of TVU for predicting andpreventing PTB. Many different popula-tions have been screened, including asymp-tomatic women with singleton, twin, andtriplet pregnancies and symptomatic womenwith preterm labor or PPROM. Studies havebeen done on low-risk patients, high-riskpopulations, and patients with a cerclage inplace. While the technique in these studies issimilar, study population (in particular theincidence of prior PTB), gestational age atwhich TVU was done, frequency of TVU,cervical parameters studied, and outcomes

vary. In the following review of the avail-able studies, we emphasize those with thebest design (ie, those which describe propertechnique, blinding of managing physiciansto ultrasound results, and the inclusion oflarge numbers of patients). The cervical pa-rameter found in most studies to have thebest predictive accuracy, as determined byreceiver operating characteristic curves, wasa CL <25 mm. The most common primaryoutcome was spontaneous PTB at less than35 weeks’ gestation (PTB <35 w).

Cross-Sectional StudiesNumerous studies have reported normo-grams for CL in non-selected pregnantwomen with singleton gestations.5,8,9 Inlow-risk women, CL is a continuous vari-able, with a mean of 35 mm to 40 mm from14 to 30 weeks, with the lower 10th percen-tile being 25 mm and the upper 10th (90thpercentile) 50 mm.5,9 A progressive andnatural shortening of CL is noticed after 30weeks even in patients destined to deliver atterm.8 Some have reported a very slight pro-gressive linear reduction of CL even before30 weeks.10 There seems to be no differencein CL between nulliparous and multiparouswomen throughout pregnancy, while riskfactors for PTB such as African origin, ageless than 20 years, low ponderal index, andprior miscarriage or PTB, are associatedwith a shorter CL.10

Studies that have evaluated the useful-ness of TVU for predicting PTB in asymp-tomatic singleton pregnancies are shown inTable 1. In general, these studies found thatthe shorter the cervix, the higher the risk ofPTB. Using different cutoffs for CL rangingfrom 15 mm to 34 mm, the positive predic-tive values ranged from 6% to 44%. Thisrelatively low value is likely due at least inpart to the low incidence of PTB in thesestudies (0.8%–15%). In only 3 of these stud-ies were the clinicians caring for the patientblinded to the ultrasound results. It shouldbe noted that in one of the better-designed,larger studies,5 the sensitivity was only 37%

950 Berghella et al

and the positive predictive value only 18%.This means that 82% of these low-risk pa-tients who were found to have a short CL at24 weeks delivered at or after 35 weeks.

High-Risk SingletonsThere is intense interest in preventing recur-rence of PTB. It has been thought that manypreterm births are a result of a “weak” cer-vix, and that cervical cerclage may help pre-vent recurrence. However, three publishedrandomized studies on prophylactic cer-clage without the use of TVU have notshown that this procedure prevents PTB inhigh-risk singleton gestations, except for thesubgroup of women with three or more priorsecond trimester losses or PTBs.15 Re-searchers have hypothesized that a subgroupof patients in whom a cerclage may be ben-eficial could be identified with TVU of thecervix. Risk factors that have been used toidentify potential subjects for these studiesinclude one or more PTB between 14 and 32weeks’ gestation, Mullerian anomaly, twoor more voluntary terminations, diethylstil-bestrol (DES) exposure, cone biopsy, etc.Women with multiple gestations, previa,prophylactic cerclage, or a major fetalanomaly have been excluded. Predictive ac-curacy of the studies performed by us and

others on singleton high-risk pregnancieswith some of these risk factors are detailedin Table 2. In our study,7 each of the 168pregnancies analyzed received an average of2.3 TVUs between 14 and 24 weeks’ gesta-tion. Multivariate logistic regression analy-sis showed that TVU cervical change was asignificant independent predictor of sponta-neous PTB (OR 6.4, 95% CI 2.6–16.8),while demographic variables and risk fac-tors were not.

As Table 2 shows, some of the patientswith PTB had normal TVU during the studyperiod (false negatives). In our experience,most of these delivered at �32 weeks, a timeat which the consequences of PTB are not assevere. The positive predictive value of anabnormal test was higher if the change ap-peared between 14 and 18 weeks (PPV 70%)than if abnormalities were first noted be-tween 18 and 22 weeks (PPV 40%). A pa-tient with clinical risk factors for PTB whohad a normal TVU between 18 and 22 weekshad a risk of PTB of only 4%, which repre-sents an 80% reduction in risk. PTB within 4weeks of TVU occurred in 8 (13%) of preg-nancies with cervical changes and in none ofthe pregnancies without such change. It isimportant to note that, at least in high-riskpatients, PTB following the detection of ashort CL are most often preceded by

TABLE 1. Prediction of PTB by TVU in Asymptomatic Singleton Pregnancies StudiedCross-Sectionally

Author nPTB(%)

PTBdefined(wks)

GAstudied(wks)

CLcut-off(mm)

%abn Sens Spec PPV NPV RR

Andersen9 113 15 <37 7–30 34 25 47 84 35 90 3.4Tongsong11 730 13 <37 28–30 25 2 6 98 31 89 2.6Iams5 2915 4.3 <35 22–25 25 10 37 92 18 97 6.2*Iams5 " " " " 20 5 23 97 26 97 9.4*Heath12 1252 2.3 <35 23 15 2 38 " " " "Heath12 " " " " 20 3 58 93 11 99 1.1Taipale13 3694 0.8 <35 18–22 29 3 19 97 6 99 8.0Taipale13 " " " " 25 0.3 7 100 15 99 20.0Hibbard14 760 6.7 <35 16–22 27 5 29 97 44 94 7.5

Names in italics indicate studies in which clinicians were blinded to results of TVU. For studies listing two sets of numbers for predictiveaccuracy, different CL cut-offs were used. PTB% = incidence of preterm birth; GA = gestational age; CL = cervical length; % abu =percent abnormal sens = sensitivity; spec = specificity; PPV = positive predictive value; NPV = negative predictive value; RR = relativerisk compared to those with normal CL except * = compared to values above the 75th percentile.

Ultrasound Assessment of the Cervix 951

PPROM (48%–68%) instead of PTL.2,19 ACL <10 mm in these patients is particularlypredictive of PPROM.20

A recent well-designed blinded multi-center study of the Maternal Fetal MedicineUnits Network of the National Institute ofChild Health and Human Development onTVU in patients with a history of PTB <32weeks demonstrated that the best predictiveaccuracy was achieved with serial TVUs,and including the shortest cervix ever afterspontaneous or transfundal pressure-elicitedchanges. The sensitivity and positive predic-tive value reached 69% and 55%, respec-tively. This was the only blinded study per-formed in women at high-risk for PTB, andcontains therefore the most reliable andvalid predictive data in this population.1

No study has so far reported exclusivelyon other high-risk subgroups, such aswomen with Mullerian anomalies, exposureto diethylstilbestrol (DES), or cone or LEEPbiopsy.

TwinsDespite the significant contribution of PTBto perinatal morbidity and mortality in twinpregnancies, the prediction of PTB in twinsusing traditional clinical means remains lim-

ited. In a preterm prediction study in twingestation, Goldenberg found that a CL �25mm at 24 weeks’ gestation to be the best ofall the predictors of PTB that they eval-uated, including fetal fibronectin and bacte-rial vaginosis.21 (O.R. 3.2, 95% C.I. 1.3–7.9) Compared with singleton pregnancies,twin pregnancies that deliver at term havebeen shown to have a similar TVU CL at 14to 19 weeks, but have a progressively muchshorter cervix starting after 20 weeks’ ges-tation.22 Since cervical shortening occurs af-ter 20 weeks’ gestation even in twin preg-nancies destined to deliver at term,22 sono-graphic examination of the cervix before 20to 24 weeks may lead to better prediction ofPTB. A recent study found that the predic-tive value of sonographic CL determinationin twins between 24 and 34 weeks’ gestationwas low.23

Performance of ultrasound for predictingpreterm delivery in twins is outlined inTable 3. Again, predictive accuracy variesby study, possibly because of different CLcutoffs and gestational age at which womenwere screened. As with high-risk singletons,a short cervix is a good predictor of PTB.The shorter is the CL, and/or the more severethe funneling, the higher the risk of PTB.Importantly, in our study,25 only 4% of twin

TABLE 2. Prediction of PTB by TVU in Asymptomatic High-Risk Singleton Pregnancies

Author nPTB(%)

PTBDefined(wks)

GAStudied(wks)

CLcut-off(mm)

%2bn Sens Spec PPV NPV RR

Berghella7 96 18 <35 14–30 25 25 59 85 45 91 4.8Berghella7 168 18 <35 14–24 25 33 61 74 35 89 3.3Berghella7 " " " " 25# 38 74 70 37 92 4.8Andrews16 53 28 <35 16–20 25 15 33 100 100 79 4.8Cook17 120 20 <34 9–24 30+ 24 59 79 45 87 3.5Owen1 183 26 <35 16–19* 25 – 19 98 75 77 3.3Owen1 " " " 16–24� 25 – 69 80 55 88 4.5Guzman18 469 10 <34 15–24 25 15 76 68 20 96 2.3

Names in italics indicate studies in which clinicians were blinded to results of TVU.# Funneling of >25% also a criterion of abnormal.+ Funnel width >5 mm considered abnormal.

* Before dynamic changes.

� Serial TVUs and after dynamic and TFP changes.

Other abbreviations the same as on Table 1.

952 Berghella et al

pregnancies with a sonographic CL of >35mm between 18 and 26 weeks delivered be-fore 35 weeks. If this finding is confirmed byothers, it may allow obstetricians to avoidbedrest and other interventions commonlyused in twin pregnancies.

At least four studies have reported onTVU of the cervix and its prediction of PTBin triplet pregnancies. Our data show that aCL <20 mm at 16 to 24 weeks is associatedwith an increased risk for PTB <28 weeks(sensitivity 67%, specificity 78%, PPV 44%,NPV 90%; RR 4.4 [95% CI 1.0-20.0] ).

Women with CerclageTVU of the cervix has been evaluated inpatients with prophylactic, therapeutic, oremergent cerclage in place. Most studieshave shown that transvaginal cerclage isplaced in the middle part of the cervixin the majority of cases (Fig. 3).29,30,31 Eval-uation of pre- and post-cerclage TVUCL has shown that CL usually increasespost-cerclage, and that an increase in CLis associated with a higher rate of term de-livery.32,33

Several studies have evaluated the accu-racy of TVU for predicting PTB in patientswith cerclage.29–32 These studies all showthat TVU cervical parameters are predictiveof PTB. CL <25 mm and upper cervix (theclosed portion above the cerclage, see figure2) <10 mm are probably the two best predic-tive parameters. It is unclear what (if any)

intervention would prevent PTB once thescreening TVU of the cervix is found to beabnormal.

Three-dimensional ultrasound makes itpossible to obtain an axial plane through thecervix at the level of the cerclage, demon-strating the entire stitch (Fig. 3). This view isnot obtainable with conventional 2DUS.Whether 3-dimensional imaging will im-prove clinical management in patients withor without a cerclage in place is unknown.

At What Gestational Age isCervical Assessment MostPredictive?Almost all patients, even those at the highestrisk, have a normal CL in the first and earlysecond trimesters. In a study of TVU of thecervix in high-risk women, only 5% had aCL <25 mm between 10 and 14 weeks.34

Sensitivity for the prediction of PTB is verylow in this time interval. An additionaldrawback of very early screening is that thelower uterine segment is difficult to distin-guish from the true cervix in the late first andearly second trimesters. The most commongestational age when short cervix or funnel-ing develops is 18 to 22 weeks.2,6 Therefore,if a screening program is to include only oneCL assessment, it should be done in this timeinterval. High-risk patients destined to de-liver preterm may have earlier cervicalchanges. The earlier the short CL is de-

TABLE 3. Prediction of PTB by TVU in Twin Pregnancies

Author nPTB(%)

PTBDefined(wks)

GAStudied(wks)

CLCut-off(mm)

(%)Abn Sens Spec PPV NPV RR

Goldenberg21 147 32 <35 22–24 25 18 30 88 54 74 3.2Wennerholm23 101 22 <35 28 33 N/A N/A N/A N/A N/A 2.1Imseis24 85 20 <34 24–26 <35 40 94 49 31 97 10.7Yang25 65 23 <35 18–26 25 9 27 96 67 81 4.6Yang25 <35 18–26 30 20 53 90 62 87 3.6Skentou26 464 19 <33 23 25 11 46 90 33 94 5.8Guzman27 131 17 <34 21–24 20 18 36 86 35 87 9.4Newman28 128 N/A <35 18–26 25 N/A 38 80 45 74 N/A

Abreviations same as table 1. N/A = not available.

Ultrasound Assessment of the Cervix 953

tected, the higher is the risk of PTB. A cervixof <25 mm had a positive predictive value of70% when detected between 14 and 18weeks, and of 40% when detected between18 and 22 weeks.2 Therefore, it may be thatpatients with the highest risk of PTB (eg, pa-tients with classic histories of cervical in-competence) may benefit from early (ie, 14–18 weeks) ultrasound examination to deter-mine their need for intervention.

The benefit of repeated TVU examina-tions and the ideal interval for repeatingTVU has not been clearly established. If ascreening program were employed in rela-tively low-risk women, one TVU of the cer-vix at around 18 to 22 weeks would probablybe most effective. It appears that one normalTVU CL between 14 and 18 weeks and an-other between 18 and 22 weeks is reassuringin most high-risk women.7 In women at veryhigh risk for PTB, such as those with priorsecond trimester loss or very early spontane-ous PTB, some have advocated TVU of thecervix every 2 weeks, at least between 14

and 24 weeks. The fact that TVU at 14 to 22weeks is at least as predictive of PTB asTVU after 22 weeks is important, since in-terventions to prevent PTB are most effec-tive when changes leading to PTB are de-tected early in the process.

Transvaginal Ultrasoundversus Digital Examination ofthe CervixCervical assessment by manual examinationis the traditional method for the prediction ofPTB. A study comparing the utility of digitalexamination and TVU for predicting PTBevaluated 96 high-risk patients every 2weeks—from 14 to 30 weeks’ gestation withexaminers blinded to the results of the alter-nate technique.2 The majority of patients(61%) were included because of a history ofone or more PTB and 18% had PTB (<35weeks) in the current pregnancy. The meansonographic CL of the PTB group was sig-nificantly shorter than that of the term deliv-

FIGURE 3. 3-dimensional multiplanar display of a cervix with a cerclagein place. In 2DUS, we normally see just two bright dots representing thesuture (B). In 3DUS in the axial plane, you can see the suture in its entirety(A). This view enables a complete assessment of the cerclage and its rela-tionship with the cervical canal. D is an enlargement of A.

954 Berghella et al

ery group while there was no difference inmanual cervical measurements between thetwo groups.

The relative lack of success of digital ex-amination in predicting PTB is probably dueto the fact that it is subjective (inter-observervariability of 52%),35 not accurate for evalu-ating the internal os (the whole upper halfof the cervix is not measurable by thismethod),36 and nonspecific (15% to 16% ofprimiparous women and 17% to 35% ofmultiparous women who are delivered atterm have cervices that are 1–2 cm dilatedby manual examination in the late secondtrimester).37 Studies have shown that sono-graphic CL measurements are, on the aver-age, 11 mm longer than manual estimations.74% of asymptomatic patients with funnel-ing have a closed and at least 2-cm-long cer-vix on manual examination.6 These datashow that TVU is clearly superior to manualexamination for evaluation of the cervix andprediction of PTB.

Why is Short CervixAssociated with PTB?The most obvious hypothesis is that a shortCL is caused by an intrinsic weakness of thecervix. This cervical weakness could possi-bly be due to traumatic or surgical damage, acongenital disorder, or a connective tissuedisease. This hypothesis, while appealing,has not been proven, and will undoubtedlybe addressed by future histologic and func-tional in-vivo studies. It is interesting to notethat almost no women, even the most high-risk, have a short CL in the first trimester.34

Another hypothesis is that a short CL canprovide easier access of potentially patho-logic vaginal organisms into the intrauterineenvironment, leading to prolonged subclini-cal chorioamnionitis and subsequent PTB.Women with a normal CL have mechanicaland immunologic protection against the as-cent of lower vaginal organisms. There is astrong association between a short CL onTVU and infection. High amniotic fluid IL-6, later development of chorioamnionitis,

and acute inflammatory lesions of the pla-centa have all been associated with a shortCL on TVU. In the preterm prediction studyconducted by the NICHD MFMU Network,patients with bacterial vaginosis and a shortCL have higher PTB than women with just ashort CL (22% vs. 16%) (personal commu-nication, Berghella). A short CL leading toPTB is often associated with PPROM in-stead of PTL, providing additional evidencefor the role of infection in these patients.19

Recent studies have shown that the ma-jority of asymptomatic women with CL <25mm before 24 weeks have some contrac-tions, more than controls with a normal cer-vix.38,39 It is unclear whether contractionscause the short CL, are a result of the shortcervix, or whether these two factors worksynergistically.

Interventions to PreventPreterm Birth Based onAbnormal TVUMany investigators believe the high nega-tive predictive value of TVU of the cervix isimportant, since patients with a normal CLcan be reassured and interventions avoided.However, a high negative predictive valuealone is not sufficient to justify a screeningprogram. For TVU to be judged cost-effective, there must be an effective inter-vention to prevent PTB if the test is positive.For asymptomatic patients with a short CL,the only intervention studied so far has beencervical cerclage.

Non-Randomized StudiesFive non-randomized studied have studiedthe benefit of therapeutic cerclage whenshort CL is identified (Table 4). Three em-ployed screening in a cross-section of lowand high-risk patients,14,40,41 and twoscreened only high-risk women.7,42 Becausecerclage was performed at the obstetricians’discretion in these studies, selection biasmay have been important. Two14,41 of the 3cross sectional and one7 of the two high-risk

Ultrasound Assessment of the Cervix 955

studies reported no benefit from cerclage.It is impossible to do a metaanalysis of thesestudies given the different population char-acteristics, CL cutoffs, and outcomes stud-ied.

A recent analysis of our non-randomizeddata at Thomas Jefferson University hasshown that in 43 women without obstetricalrisk factors but a CL of <25 mm, placementof cerclage did not decrease the incidence ofPTB (RR 1.1, 95% CI 0.5–2.7). In contrast,in 73 women with prior delivery at less than35 weeks and with a CL <25 mm, placementof a cerclage reduced the incidence of recur-rent PTB (<35 weeks) by 40%, from 63% to39% (RR 0.6, 95% CI 0.4–1.0). In 57women with a prior second trimester lossand CL <25 mm, the incidence of PTB <35weeks was reduced from 65% without cer-clage to 41% with cerclage (RR 0.6, 95% CI0.4–1.0).

Randomized TrialsRandomized trials of cerclage are extremelydifficult to perform, particularly when this

procedure is believed by many patients andtheir doctors to be beneficial. Randomizedtrials on therapeutic cerclage to prevent PTBin women with a short CL have had differentresults, depending on the type of patientsstudied. Two trials which have focused onan unselected population showed no preven-tion of PTB in women who were random-ized to receive cerclage. These included astudy by Rust44 of 113 women with eitherCL <25 mm or funneling >25%, and an un-published study by Prof Nicolaides’s groupin London, which randomized women witha CL <15 mm at 23 weeks (personal com-munication Nicolaides). The conclusionfrom these 2 studies, including now (Ruststudy is still in progress) over 500 women, isthat therapeutic cerclage in unselectedwomen with short CL on TVU does not pre-vent PTB. In contrast, a small trial (n = 35)by Althuisius et al, which included only pa-tients at high risk for PTB, many of whomwere suspected of having cervical incompe-tence,43 showed a decrease in PTB and neo-natal morbidity and mortality with cerclage.

TABLE 4. Therapeutic Cerclage for TVU Short Cervix

Author Group n

HistoryPTB(%)

GAstudied(wks)

CLcut-off(mm)

PTB < 35+

wks (%)PTB < 32wks (%)

Non-Randomized, Cross-SectionalHeath40 Cerclage� 22 18 23 �15 2 (9) 1 (5)

Control 21 10 23 �15 11 (52)* 11 (52)*Hibbard14 Cerclage 15 <10 14–24 <25 8 (53) 6 (40)

Control 14 <10 14–24 <25 9 (64) 9 (64)Hassan41 Cerclage 25 32 14–24 <15 17 (68) 14 (56)

Control 45 16 14–24 <15 24 (53) 19 (42)Non-Randomized, High RiskGuzman42 Cerclage 28 N/A 16–24 <20 6 (21) 2 (7)

Control 17 N/A 16–24 <20 11 (65)* 4 (25)Berghella7 Cerclage 39 69 14–24 <25 18 (46) 10 (26)

Control 24 58 14–24 <25 5 (21) 4 (17)Berghella7 Cerclage 22 65 18–24 <25 6 (27) 3 (14)

Control 22 55 18–24 <25 5 (23) 4 (18)RandomizedAlthuisius43 Cerclage 19 74 16–27 <25 0 (0)

Control 16 75 16–27 <25 7 (44)*Rust44 Cerclage 55 54 16–24 <25 19 (35)

Control 58 36 16–24 <25 21 (36)

+ PTB < 34 weeks in the two randomized studies. �Used Shirodkar cerclage. All others used McDonald. Abbreviations same as Table I.*Indicates p < .05.

956 Berghella et al

Interestingly, when Rust45 analyzed a sub-group of his study population that had a priorPTB (n = 87), again he did not find a benefitfrom cerclage.

Many design details are available for theRust and Althuisius studies. Similarities in-clude race distribution, CL <25 mm requiredfor inclusion, same mean CL in study pa-tients (20 mm), hospital admission prior tocerclage placement, use of the McDonaldtechnique, bedrest and followup TVUs forall patients, some rescue cerclages (8 in Rustand 2 in Althuisius), and cerclage removal at36 weeks. The main difference was that Rustincluded all women detected to have a shortCL, while Althuisius studied only those witha probable history of cervical incompetence(in fact, their patients were first randomizedto prophylactic cerclage or not before beingfollowed with TVU). Thus, in Rust and Al-thuisius studies, the incidence of prior PTBwas 47% versus 74%, and of prior secondtrimester loss 20% versus 46%, respec-tively. Also, 16% of the patients in the Ruststudy had twins and 9% of patients had uni-dentifiable historic risk factors, while Al-thuisius excluded such patients. A possibleconfounder was that Rust gave indometha-cin to all patients, while Althuisius used itonly in the cerclage arm. This raises the pos-sibility that indomethacin may have contrib-uted significantly to the prevention of PTBin the Althuisius study.

Avoiding Cerclage if TVU isNormalOne randomized study46 and 3 non-ran-domized studies47,49 have evaluated wheth-er women with suspected incompetent cer-vix can be safely followed with TVU, withplacement of a cerclage only in those withcervical change compared to prophylacticcerclage. At least 60% of these high-riskwomen maintain a normal CL until after 24weeks and deliver at term, and can be sparedany intervention. Only about 40% develop ashort CL and are at true risk of PTB, and canbe offered intervention. While this latter

management with serial TVU of the cervixappears to be a safe alternative to traditionalprophylactic cerclage, larger randomizedtrials are needed to prove that this manage-ment approach is appropriate.

Other Uses of UltrasoundCervical Assessment

EVALUATION OF PATIENTS WITHSUSPECTED PRETERM LABORTVU of the cervix has been studied exten-sively as a predictor of PTB in patients withsymptoms of PTL (Table 5) While inclusioncriteria in these studies were all slightly dif-ferent, all showed a statistically significantpredictive accuracy of TVU for PTB. Unfor-tunately, no prospective intervention ran-domized trials have been performed basedon this association. Two non-randomizedtrials have been reported. Zalar57 reported adecrease in incidence of birth weigh <2,500grams when TVU of the cervix was used totriage patients to bed rest and tocolysis,compared with historic controls. Also com-pared with historic controls, Rageth58

showed that using TVU in symptomatic pa-tients for management could decrease the in-cidence of hospitalization and costs, but didnot decrease PTB. Randomized trials shouldbe performed in this area before clinical usecan be recommended.

PREDICTING LATENCY IN PPROMThree studies have examined the utility ofTVU of the cervix in patients with pretermpremature rupture of membranes (PPROM).Carlan59 demonstrated in a randomized trialthe safety of performing TVU in this group.In patients studied between 24 and 34weeks,he found that the latency was 2 days shorterif the CL was �30 mm (9 days vs. 11 days).Rizzo60 studied 92 women with PPROM be-tween 24 and 32 weeks, and showed that aCL �20 mm was associated with a latencyof 2 days (range 0–14) versus 6 days (range0–36) if the CL was >20 mm. Most recently,

Ultrasound Assessment of the Cervix 957

Gire61 reported on 101 singleton pregnan-cies with PPROM at <34 weeks. A CL <20mm was associated with a latency of 2.5days versus 10 days if the CL was �20 mm.

PREDICTING THE SUCCESS OFLABOR INDUCTIONSeveral studies have evaluated the predic-tive accuracy of TVU of the cervix in termgravidas for length of induction and inci-dence of success (vaginal delivery). ShortCL (<30 mm,62 <28 mm,63 <26 mm,64 orwedging65) were associated with a short du-ration of labor and a higher incidence ofvaginal delivery compared with longer cer-vix. Two studies instead did not find thatTVU of the cervix added significantly to theprediction obtained by dilatation of the cer-vix on manual examination,66,67 but 3 didfind that TVU CL was a better predictor thanany Bishop score parameter.62–64 TVU ofthe cervix was of limited success in predict-ing the onset of labor in term patients.68

ConclusionsPTB remains the leading cause of neonatalmorbidity and mortality. TVU for cervicalassessment is one of the best if not the best ofavailable techniques for predicting PTB. A

cervical length of <25 mm between 16 and24 weeks has been shown to be the most re-liable threshold for an increased risk of PTB.However, the shorter the cervix, the higheris the risk of PTB, and the earlier in gesta-tional age the shortening occurs, the higheris the risk. The role of TVU of the cervix hasbeen studied in a wide variety of settings andwith different patient populations. Screen-ing frequency should depend on severity ofobstetrical history, with serial TVUs of thecervix having a better predictive accuracythan one, especially in high-risk popu-lations. The use of TVU of the cervixhas been found to be safe and acceptable topatients.

Since there is still limited evidence show-ing any benefit of cerclage placement in re-sponse to cervical changes, we believe large,properly conducted, randomized controlledtrials are necessary before this interventioncan become part of standard clinical prac-tice. A screening test such as TVU of the cer-vix should be used in clinical practice onlywhen treatment of the condition is finallyproven to be effective, not just based exclu-sively on its high negative predictive value.There is accumulating evidence that thistechnique can be used to safely avoid cer-clage placement unless or until cervical

TABLE 5. Prediction of PTB by TVU in Patients with Symptoms of Preterm Labor (SingletonPregnancies)

Author nPTB(%)

PTBDefined(wks)

GAStudied(wks)

CLcut-off(mm)

%2bn Sens Spec PPV NPV RR

Murakawa50 32 34 N/A 25–35 <30 53 100 71 65 100 *<25 31 64 86 70 82 3.9

Iams51** 60 40 < 36 24–34 <30 73 100 44 55 100 *Gomez52 59 37 < 37 20–35 <18 41 73 78 67 83 3.9" " " " " >0.52� 31 76 94 89 86 6.9Timor-Trisch53 70 27 < 37 20–35 wedging 54 100 75 59 100 *Rizzo54 108 44 < 37 26–30 <20 N/A 68 79 71 76 3.0Rozenberg55 76 26 < 37 24–39 <26 39 75 73 50 89 4.6Crane56 136 27 < 37 23–33 <30 50 81 65 46 90 2.3

Names in italics indicate clinicians were blinded to results of TVU. Abbreviations same as Table 1.

* = infinity, since one of the boxes was zero.

** = 12 twin pregnancies included.

� = cervical index.

958 Berghella et al

change occurs, but this management alsoneeds to be confirmed by appropriate pro-spective trials. There are still limited dataconcerning whether TVU of the cervix isclinically useful in caring for patients beingevaluated for preterm labor or who experi-ence PPROM, or for prediction of success oflabor induction. Future research on all theseclinical applications of TVU of the cervixhas the potential to significantly improve thehealth outcomes of pregnant women andtheir babies.

References1. Owen J, Yost N, Berghella V, et al for the

National Institute of Child Health and Hu-man Development, Maternal-Fetal Medi-cine Units Network. JAMA. 2001;286:1340–1348.

2. Berghella V, Tolosa JE, Kuhlman KA, et al.Cervical ultrasonography compared tomanual examination as a predictor of pre-term delivery. Am J Obstet Gynecol. 1997;177(4):723–730.

3. Zilianti M, Azuaga A, Calderon F, et al.Transperineal sonography in second trimes-ter to term pregnancy and early labor. J Ul-trasound Med. 1991;10:481–485.

4. Berghella V. The natural history of cervicalfunneling in high-risk women. Obstet Gyne-col. 2003;101:120–121s.

5. Iams JD, Goldenberg RL, Meis PJ, et al. Thelength of the cervix and the risk of sponta-neous premature delivery. N Eng J Med.1996;334(9):567–572.

6. Berghella V, Kuhlman K, Weiner S, et al.Cervical funneling: sonographic criteriapredictive of preterm delivery. UltrasoundObstet Gynecol. 1997;10:161–166.

7. Berghella V, Daly SF, Tolosa JE, et al. Pre-diction of preterm delivery with transvagi-nal ultrasonography of the cervix in patientswith high-risk pregnancies: does cerclageprevent prematurity. Am J Obstet Gynecol.1999;181:809–815.

8. Andersen HF. Transvaginal and transab-dominal ultrasonography of the uterine cer-vix during pregnancy. J Clin Ultrasound.1991;19:77–81.

9. Andersen HF, Nugent CE, Wanty SD, et al.Prediction of risk for preterm delivery by ul-

trasonographic measurement of cervicallength. Am J Obstet Gynecol. 1990;163:859–867.

10. Heath VCF, Southall TR, Souka AP, et al.Cervical length at 23 weeks of gestation: re-lation to demographic characteristics andprevious obstetric history. Ultrasound Ob-stet Gynecol. 1998;12:304–311.

11. Tongsong T, Kamprapanth P, Srisomboon J,et al. Single transvaginal sonographic mea-surement of cervical length early in the thirdtrimester as a predictor of preterm delivery.Obstet Gynecol. 1995;86:184–187.

12. Heath VCF, Southall TR, Souka AP, et al.Cervical length at 23 weeks of gestation:prediction of spontaneous preterm delivery.Ultrasound Obstet Gynecol. 1998;12:312–317.

13. Taipale P, Hiilesmaa V. Sonographic mea-surement of uterine cervix at 18-22 weeks’gestation and the risk of preterm delivery.Obstet Gynecol. 1998;92:902–907.

14. Hibbard JU, Tart M, Moawad AT. Cervicallength at 16–22 weeks’ gestation and risk ofpreterm delivery. Obstet Gynecol. 2000;96:972–978.

15. MRC/RCOG Working Party on CervicalCerclage. Final report of the Medical Re-search Council/Royal College of Obstetri-cians and Gynaecologists multicentre ran-domized trial of cervical cerclage. Br J Ob-stet Gynecol. 1993;100:516–523.

16. Andrews WW, Cooper R, Hauth JC, Golden-berg RL, Neely C, Dubard M. Second-trimester cervical ultrasound: associationswith increased risk of recurrent early spon-taneous delivery. Obstet Gynecol. 2000;95:222-226.17.

17. Cook C-M, Ellwood DA. The cervix as apredictor of preterm delivery in ‘at-risk’women. Ultrasound Obstet Gynecol. 2000;15:109–113.

18. Guzman ER, Mellon C, Vintzileos AM, etal. Longitudinal assessment of endocervicalcanal length between 15 and 24 weeks’ ges-tation in women at risk for pregnancy loss orpreterm birth. Obstet Gynecol. 1998;92(1):31–37.

19. Odibo AO, Berghella V, Reddy U, et al.Does transvaginal ultrasound of the cervixpredict preterm premature rupture of mem-branes in a high-risk population? Ultra-sound Obstet Gynecol. 2001;18:223–227.

Ultrasound Assessment of the Cervix 959

20. Odibo AO, Talucci M, Berghella V. Predic-tion of preterm premature rupture of mem-branes by transvaginal ultrasound featuresand risk factors in a high-risk population.Ultrasound Obstet Gynecol. 2002;20:245–251.

21. Goldenberg RL, Iams J, Miodovnik M, et al.The preterm prediction study: Risk factorsin twin gestation. Am J Obstet Gynecol.1996;175:1047–1053.

22. Kushnir O, Izquierdo LA, Smith JF, et al.Transvaginal sonographic measurement ofcervical length: Evaluation of twin pregnan-cies. J Rep Med. 1995;40(5):380–382.

23. Wennerholm UB, Holm B, Mattsby-BaltzerI, et al. Fetal fibronectin, endotoxin, bacte-rial vaginosis and cervical length as predic-tors of preterm birth and neonatal morbidityin twin pregnancies. Br J Obstet Gynecol.1997;104:1398–1404.

24. Imseis HM, Albert TA, Iams JD. Identifyingtwin gestations at low risk for preterm birthwith a transvaginal ultrasonographic cervi-cal measurement at 24 to 26 weeks’ gesta-tion. Am J Obstet Gynecol. 1997;177:1149–1155.

25. Yang JH, Kuhlman K, Daly S, et al. Predic-tion of preterm birth by second trimester cer-vical sonography in twin pregnancies. Ultra-sound Obstet Gynecol. 2000;15:288–291.

26. Skentou C, Souka AP, Lioa AW, et al. Pre-diction of preterm delivery in twins by cer-vical assessment at 23 weeks. UltrasoundObstet Gynecol. 2001;17:7–10.

27. Guzman ER, Walters C, O’Reilly-Green C,et al. Use of cervical ultrasonography in pre-diction of spontaneous preterm birth in twingestations. Am J Obstet Gynecol. 2000;183:1103–1107.

28. Newman RB, Krombach RS, Myers MC.CNM, McGee DL. Effect of cerclage on ob-stetrical outcome in twin gestations with ashortened cervical length. Am J Obstet Gy-necol. 2002;186:634–640.

29. Quinn. Vaginal ultrasound and cervical cer-clage; a prospective study. Ultrasound Ob-stet Gynecol. 1992;2:410–416.

30. Andersen HF, Karimi A, Sakala EP, et al.Prediction of cervical cerclage outcome byendovaginal ultrasonography. Am J ObstetGynecol. 1994;171:1102–1106.

31. Berghella V, Davis G, Wapner RJ. Trans-vaginal ultrasound of the cervix in pregnan-

cies with prophylactic cerclage. Am J ObstetGynecol. 1999;180:s173.

32. Guzman ER, Houlihan C, Vintzileos A, etal. The significance of transvaginal ultraso-nographic evaluation of the cervix inwomen treated with emergency cerclage.Am J Obstet Gynecol. 1996;175:471–476.

33. Althuisius SM, Dekker GA, van Geijn HP,et al. The effect of therapeutic McDonaldcerclage on cervical length as assessed bytransvaginal ultrasonography. Am J ObstetGynecol. 1999;180:366–370.

34. Berghella V, Talucci M, Desai A. Doestransvaginal sonographic measurement ofcervical length before 14 weeks predict pre-term delivery in high-risk pregnancies? Ultra-sound Obstet Gynecol. 2003;21:140–144.

35. Phelps JY, Higby K, Smyth MH, et al. Ac-curacy and intraobserver variability ofsimulated cervical dilatation measurements.Am J Obstet Gynecol. 1995;173:942–945.

36. Michaels WH, Montgomery C, Karo J, et al.Ultrasound differentiation of the competentfrom the incompetent cervix: Prevention ofpreterm delivery. Am J Obstet Gynecol.1986;154:537–546.

37. Floyd WS. Cervical dilatation in the mid-trimester of pregnancy. Obstet Gynecol.1961;18:380–381.

38. Berghella V, for the NICHD MFMU Net-work. Frequency of uterine contractions inasymptomatic pregnant women with orwithout a short cervix on transvaginal ultra-sound. Am J Obstet Gynecol. 2003;187(6):S127.

39. Lewis D, Pelham J, Sawhney H, et al. Mostasymptomatic pregnant women with a shortcervix on ultrasound are having uterine con-tractions. Am J Obstet Gynecol. 2001;185:S144.

40. Heath VCF, Souka AP, Erasmus I, GibbDMF, Nicholaides KH. Cervical length at23 weeks of gestations: the value of Shirod-kar suture for the short cervix. UltrasoundObstet Gynecol. 1998;12:318–322.

41. Hassan SS, Romero R, Maymon E, et al.Does cervical cerclage prevent preterm de-livery in patients with a short cervix? Am JObstet Gynecol. 2001;184:1325–1331.

42. Guzman ER, Benito CW, Yeo L, VintzileosAM, Walters C, Meirowitz N. UMDNJ-Robert Wood Johnson Medical School/St.Peter’s Medical Center, New Brunswick,

960 Berghella et al

NJ. Bed rest versus cervical cerclage in thetreatment of cervical incompetence mani-fested by ultrasound around the time of fetalviability. Am J Obstet Gynecol. 1999;180(1):S78.

43. Althuisius SM, Dekker GA, Hummel P, etal. Final results of the cervical incompetenceprevention randomized cerclage trial(CIPRACT): Therapeutic cerclage with bedrest versus bed rest alone. Am J Obstet Gy-necol. 2001;185:1106–1112.

44. Rust OA, Atlas RO, Reed J, et al. Revisitingthe short cervix detected by transvaginal ul-trasound in the second trimester: Why cer-clage therapy may not help. Am J Obstet Gy-necol. 2001;185:1098–1105.

45. Rust O, Atlas R, Fischl S, et al. Does cer-clage therapy improve perinatal outcome inpatients with a history of previous pretermbirth and cervical changes on 2nd trimestertransvaginal ultrasound? Am J Obstet Gyne-col. 2002;187:S58.

46. Althuisius SM, Dekker GA, van Geijn HP,et al. Cervical incompetence prevention ran-domized cerclage trial (CIPRACT): Studydesign and preliminary results. Am J ObstetGynecol. 2000;183:823–829.

47. To MS, Palaniappan V, Skentou C, et al.Elective cerclage vs. ultrasound-indicatedcerclage in high-risk pregnancies. Ultra-sound Obstet Gynecol. 2002;19:475–477.

48. Berghella V, Haas S, Chervoneva I, et al. Pa-tients with prior second-trimester loss: Pro-phylactic cerclage or serial transvaginalsonograms? Am J Obstet Gynecol. 2002;187:747–751.

49. Kelly S, Pollock M, Maas B, et al. Earlytransvaginal ultrasonography versus earlycerclage in women with an unclear historyof incompetent cervix. Am J Obstet Gyne-col. 2001;184:1097–1099.

50. Murakawa H, Utumi T, Hasegawa I, et al.Evaluation of treatened preterm delivery bytransvaginal ultrasonographic measurementof cervical length. Obstet Gynecol. 1993;82:829–832.

51. Iams JD, Paraskos J, Landon MB, TeterisJN, Johnson FF. Cervical sonography inpreterm labor. Obstet Gynecol. 1994;84:40–46.

52. Gomez R, Galasso M. Romero R, et al. Ul-trasonographic examination of the uterinecervix is better than cervical digital exami-

nation as a predictor of the likelyhood ofpremature delivery in patients with pretermlabor and intact membranes. Am J ObstetGynecol. 1994;171:956–964.

53. Timor-Trisch IE, Boozarjomehri F, Masa-kowski Y, et al. Can a ‘snapshot’ sagittalview of the cervix by transvaginal ultraso-nography predict active preterm labor? Am JObstet Gynecol. 1996;174:990–995.

54. Rizzo G, Capponi A, Arduini D, et al. Thevalue of fetal fibronectin in cervical andvaginal secretions and of ultrasonographicexamination of the uterine cervix in predict-ing premature delivery for patients with pre-term labor and intact membranes. Am J Ob-stet Gynecol. 1996;175:1146–1151.

55. Rozenberg P, Goffinet F, Malagrida L, et al.Sage-Femme, Nisand I. Evaluating the riskof preterm delivery: a comparison of fetalfibronectin and transvaginal ultrasono-graphic measurement of cervical length. AmJ Obstet Gynecol. 1997;176:196–199.

56. Crane JMG, Van Der Hof M, Armson BA,Liston R. Transvaginal ultrasound in theprediction of preterm delivery: singletonand twin gestations. Obstet Gynecol. 1997;90:357–363.

57. Zalar RW. Transvaginal ultrasound and pre-term prelabor: a nonrandomized interven-tion study. Obstet Gynecol. 1996;88:20–23.

58. Rageth JC, Kernen B, Saurenmann E, et al.Premature contractions: possible influenceof sonographic measurement of cervicallength on clinical management. UltrasoundObstet Gynecol. 1997;9:183–187.

59. Carlan SJ, Richmond LB, O’Brien WF.Randomized trial of endovaginal ultrasoundin preterm premature rupture of membranes.Obstet Gynecol. 1997;89:458–461.

60. Rizzo G, Capponi A. Angeline E, et al. Thevalue of transvaginal ultrasonographic ex-amination of the uterine cervix in predictingpreterm delivery in patients with pretermpremature rupture of membranes. Ultra-sound Obstet Gynecol. 1998;11:23–29.

61. Gire C, Faggianelli P, Nicaise C, et al. Ul-trasonographic evaluation of cervical lengthin pregnancies complicated by preterm pre-mature rupture of membranes. UltrasoundObstet Gynecol. 2002;19:565–569.

62. Ware V, Raynor BD. Transvaginal ultraso-nographic cervical measurement as a pre-

Ultrasound Assessment of the Cervix 961

dictor of successful labor induction. Am JObstet Gynecol. 2000;182:1030–1032.

63. Pandis GK, Papageorghiou AT, RamanathanVG, et al. Preinduction sonographic measure-ment of cervical length in the prediction ofsuccessful induction of labor. Ultrasound Ob-stet Gynecol. 2001;18:623–628.

64. Gabriel R, Darnaud T, Chalot F, et al. Trans-vaginal sonography of the uterine cervixprior to labor induction. Ultrasound ObstetGynecol. 2002;19:254–257.

65. Boozarjomehri F, Timor-Tritsch E, ChaoCR, et al. Transvaginal ultrasonographicevaluation of the cervix before labor: pres-ence of cervical wedging is associated with

shorter duration of induced labor. Am J Ob-stet Gynecol. 1994;171:1081–1087.

66. Chandra S, Crane JMG, Hutchens D, et al.Transvaginal ultrasound and digital exami-nation in predicting successful labor induc-tion. Obstet Gynecol. 2001;98:2–6.

67. Watson WJ, Stevens D, Wleter S, et al. Fac-tors predicting successful labor induction.Obstet Gynecol. 1996;88:990–992.

68. Rozenberg P, Goffinet F, Hessabi M. Sage-Femme. Comparison of the Bishop score, ul-trasonographically measured cervical length,and fetal fibronectin assay in predicting timeuntil delivery and type of delivery at term. AmJ Obstet Gynecol. 2000;182:108–113.

962 Berghella et al