Ultrasound Obstet Gynecol 2015; 46: 73–81 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.14665
Predictive value of cervical length in women with twin pregnancy presenting with threatened preterm labor N. MELAMED, L. HIERSCH, R. GABBAY-BENZIV, R. BARDIN, I. MEIZNER, A. WIZNITZER and Y. YOGEV Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva, Israel; Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
K E Y W O R D S: cervical length; prediction; preterm labor; twins
ABSTRACT Objective To assess the accuracy and determine the optimal threshold of sonographic cervical length (CL) for the prediction of preterm delivery (PTD) in women with twin pregnancies presenting with threatened preterm labor (PTL). Methods This was a retrospective study of women with twin pregnancies who presented with threatened PTL and underwent sonographic measurement of CL in a tertiary center. The accuracy of CL in predicting PTD in women with twin pregnancies was compared with that in a control group of women with singleton pregnancies. Results Overall, 218 women with a twin pregnancy and 1077 women with a singleton pregnancy, who presented with PTL, were included in the study. The performance of CL as a predictive test for PTD was similar in twins and singletons, as reflected by the similar correlation between CL and the examination-to-delivery interval (r, 0.30 vs 0.29; P = 0.9), the similar association of CL with risk of PTD, and the similar areas under the receiver–operating characteristics curves for differing delivery outcomes (range, 0.653–0.724 vs 0.620–0.682, respectively; P = 0.3). The optimal threshold of CL for any given target sensitivity or specificity was lower in twin than in singleton pregnancies. However, in order to achieve a negative predictive value of 95%, a higher threshold (28–30 mm) should be used in twin pregnancies. Using this twin-specific CL threshold, women with twins who present with PTL are more likely to have a positive CL test, and therefore to require subsequent interventions, than are women with singleton pregnancies with PTL (55% vs 4.2%, respectively). Conclusion In women with PTL, the performance of CL as a test for the prediction of PTD is similar in twin and
singleton pregnancies. However, the optimal threshold of CL for the prediction of PTD appears to be higher in twin pregnancies, mainly owing to the higher baseline risk for PTD in these pregnancies. Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
INTRODUCTION Prematurity remains the foremost cause of neonatal mortality and morbidity1,2 . Although the rate of preterm delivery (PTD) before 34 weeks’ gestation is approximately 1% among singleton pregnancies3 , it is reported to be as high as 13% in twin pregnancies4 , thus twin pregnancy is considered to be one of the strongest risk factors for PTD5,6 . One of the challenges in the management of women who present with preterm labor (PTL) is the distinction between true and false PTL, since fewer than 15% of those presenting with threatened PTL will actually deliver within 7 days of presentation or before 35 weeks’ gestation7,8 . One of the most widely investigated tools for the prediction of PTD in women with a singleton pregnancy and threatened PTL is sonographic measurement of cervical length (CL)7 – 11 . However, while there is a significant amount of evidence regarding the role of CL in asymptomatic women with twin pregnancies4,12 – 20 , there is only limited information regarding the performance of CL in women with twin pregnancies who present with threatened PTL21 – 25 . The aim of this study was to assess the accuracy of sonographic CL in the prediction of PTD in women with a twin pregnancy presenting with threatened PTL, using women with a singleton pregnancy as controls, and to determine the optimal threshold of CL for its prediction in twins.
Correspondence to: Dr N. Melamed, Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petah Tiqwa 49100, Israel (e-mail: [email protected]) Accepted: 8 September 2014
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
ORIGINAL PAPER
Melamed et al.
74
METHODS This was a retrospective cohort study of all women who presented with threatened PTL in the presence of intact membranes and underwent sonographic measurement of CL in a tertiary referral medical center between January 2007 and December 2012. The study group included women with twin pregnancies who presented with threatened PTL. The predictive accuracy of CL in the twin group was compared with that observed in a control group of women with a singleton pregnancy presenting with threatened PTL. Only women who presented at a gestational age of 24+0 to 33+6 weeks were included in either group. Exclusion criteria from both the study and control groups included women with high-order multiple gestations (triplets and above), monoamniotic twins, complicated monochorionic twins (twin–twin transfusion syndrome or twin anemia–polycythemia sequence), cervical cerclage, cervical dilatation > 3 cm at presentation, pregnancies complicated by placental abruption, chorioamnionitis, stillbirth or major fetal anomalies, and women who underwent indicated preterm delivery before 37+0 weeks’ gestation for any maternal or fetal indication or who did not deliver in our medical center. The study was approved by the local institutional review board. Women were identified using the comprehensive database of sonographic examinations in our ultrasound unit. This database contains data on all sonographic examinations performed in the ultrasound unit, which are entered in real time by the performer of the examination immediately after its completion. Initially, all women who underwent sonographic measurement of CL using transvaginal ultrasound were identified and the information regarding CL, presence of funneling, change in CL during the examination (either spontaneously by mild fundal pressure or in response to Valsalva maneuver) was extracted. The medical charts of these women were then reviewed thoroughly in order to identify only those for whom the indication for sonographic measurement of CL was threatened PTL. Of these women, the following information was extracted: demographic, medical and obstetric history, number of fetuses (singleton vs twins), complications during current pregnancy, validation of gestational age by first-trimester ultrasound, gestational age at presentation, frequency and regularity of uterine contractions, results of vaginal examination, interventions performed for threatened PTL, gestational age at delivery and onset of delivery (spontaneous vs induced). For women with repeat measurements of CL at different times during gestation, only the first measurement was included in the analysis. Threatened PTL was defined as the presence of at least three regular and painful uterine contractions within a 30-min period. Tocolysis with nifedipine, indomethacin or atosiban and betamethasone was administrated for fetal lung maturation, based on the decision of the treating physician. All sonographic examinations were performed by physicians specialized in ultrasonography,
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
or by experienced sonographers, using the following ultrasound systems: Voluson E8 and Voluson 730 Expert (GE Medical Systems, Kretz Ultrasound, Zipf, Austria) and ATL 5000 (Philips Healthcare, Best, The Netherlands). Measurement of CL was performed transvaginally after the patients had emptied their bladder and according to the standard technique26 . Briefly, the measurement was performed in the sagittal plane, visualizing the full length of the endocervical canal, from the internal cervical os to the external cervical os, while exerting as little pressure as possible with the transducer. At least three measurements were obtained and the shortest measurement was recorded. In addition, the presence of cervical funneling or a change in CL, either spontaneously or in response to Valsalva maneuver, was documented routinely.
Statistical analysis Data analysis was performed with SPSS v. 21.0 software (IBM Corp., Armonk, NY, USA). The accuracy of CL in predicting spontaneous PTD before 37, 34 and 32 weeks’ gestation and within 14 and 7 days of presentation in women with a twin pregnancy was calculated and compared with the predictive accuracy in singleton pregnancies. Student’s t-test and the Mann–Whitney U-test were used to compare continuous variables, with and without normal distribution, between the twin and singleton groups, respectively, while the chi-square and Fisher’s exact tests were used for categorical variables. The rate of PTD at a gestational age of less than 37, 34 and 32 weeks and within 14 days of presentation was calculated and plotted as a function of CL at presentation for women in the twin and singleton groups. Potential thresholds of CL, below which the risk of PTD appeared to increase, were identified initially through visual inspection of the graphs and were then confirmed by comparison of the slopes of the regression lines (describing the rate of PTD as a function of CL) above and below these thresholds, using the analysis of covariance (ANCOVA) test. Kaplan–Meier survival analysis was used to compare the proportion of pregnancies undelivered according to time from presentation between the twin and singleton groups, stratified by CL at presentation. The log-rank test was used to compare the survival distributions. Spearman’s correlation coefficient was used to assess the correlation between CL at presentation and examination-to-delivery interval. Correlation coefficients for the two groups were compared using Fisher’s Z-transformation. Multivariable logistic regression analysis was used to assess the association between CL as a continuous measure and the risk of PTD in the twin and singleton groups, while controlling for potential confounders. Receiver–operating characteristics (ROC) analysis was used to determine the area under the ROC curve (AUC) as an overall measure of the discriminative ability of CL. The AUCs for the twin and singleton
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
75
groups were compared using the method of Hanley and McNeil27 . ROC analysis was also used to calculate standardized outcome-specific thresholds, including: (1) the threshold associated with a sensitivity of 90%; (2) the threshold associated with a specificity of 90%; and (3) the threshold associated with a negative predictive value (NPV) of 95%. The following measures of predictive accuracy were calculated for each of the outcome variables: sensitivity, specificity, positive predictive value (PPV) and NPV; P < 0.05 was considered statistically significant.
RESULTS A total of 2680 sonographic examinations of CL were performed in 1950 women during the study period. Of those, 218 women with a twin pregnancy were found to be eligible for the study and were compared with 1077 eligible women with a singleton pregnancy (Figure 1). The demographic characteristics of women with twin pregnancies and those with singleton pregnancies were similar, except for a higher maternal age in the twin group (Table 1). There were no differences between
Sonographic measurements of CL in women presenting with preterm labor between 24 +0 and 33+ 6 weeks during study period (n = 2680 measurements)
Twin CL measurements (n = 685)
Excluded (n = 467)
Twin pregnancies (n = 218)
Exclusion criteria: High-order multiple gestation Repeated measurements TTTS or TAPS Monoamniotic twins Cervical cerclage Indicated preterm delivery < 37+ 0 weeks Major anomalies Placental abruption Amnionitis Stillbirth Dilatation > 3 cm Incomplete data or delivery in another center
Singleton CL measurements (n = 1995)
Excluded (n = 918)
Singleton pregnancies (n = 1077)
Figure 1 Flowchart summarizing the study groups. CL, cervical length; TAPS, twin anemia–polycythemia sequence; TTTS, twin–twin transfusion syndrome. Table 1 Demographic and pregnancy characteristics of women with twin or singleton pregnancy presenting with threatened preterm labor Characteristic Maternal age (years) Maternal age > 35 years Parity Nulliparous History of preterm delivery Chorionicity DC/DA twins MC/MA twins Unknown Gestational age at examination (weeks) Cervical length (mm) Intervention Progesterone Tocolysis Corticosteroids Delivery outcome Gestational age at delivery (weeks) Gestational age at delivery < 37 + 0 weeks Gestational age at delivery < 34 + 0 weeks Examination-to-delivery interval (days) Examination-to-delivery interval within 7 days Examination-to-delivery interval within 14 days
Twin (n = 218)
Singleton (n = 1077)
34.5 ± 5.2 84 (38.5) 2 (1–2) 108 (49.5) 22 (10.1)
33.1 ± 5.6 325 (30.2) 2 (1–3) 497 (46.1) 152 (14.1)
< 0.001 0.02 0.1 0.4 0.1
165 (75.7) 27 (12.4) 26 (11.9) 29.5 ± 2.8 24.9 ± 11.2
— — — 29.8 ± 2.8 30.9 ± 10.2
— — — 0.1 < 0.001
31 (14.2) 190 (87.2) 195 (89.4)
86 (8.0) 686 (63.7) 720 (66.9)
0.003 < 0.001 < 0.001
34.3 ± 2.5 180 (82.6) 63 (28.9) 36.3 ± 21.7 23 (10.6) 38 (17.4)
37.4 ± 2.7 274 (25.4) 77 (7.1) 56.3 ± 25.2 29 (2.7) 60 (5.6)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
P
Data are presented as mean ± SD, n (%) or median (interquartile range). DA, diamniotic; DC, dichorionic; MA, monoamniotic; MC, monochorionic. Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
76
the two groups with respect to parity, history of PTD and gestational age at presentation with threatened PTL (Table 1). Women in the twin group were characterized by a significantly shorter cervix at presentation and a higher rate of subsequent PTD, and were more likely to receive progesterone, tocolysis or corticosteroids for fetal lung maturation than were women in the singleton group (Table 1). Figure 2 presents the distribution curves of CL at presentation and gestational age at delivery according to the pregnancy group. Both curves were shifted to the left in twins, as compared to singletons (Figure 2). The rates of spontaneous PTD before 37, 34 and 32 weeks’ gestation and within 14 days from presentation, as a function of CL at presentation, are shown in Figure 3. The rate of PTD for any given CL at presentation was higher in twin than in singleton pregnancies (Figure 3). In addition, the threshold of CL at presentation, below which the risk of spontaneous PTD began to increase, was higher in twin than in singleton pregnancies (Figure 3): < 37 weeks, 40 mm vs 30 mm; < 34 weeks, 30 mm vs 20 mm; < 32 weeks, 30 mm vs 20 mm; within 14 days of assessment, 30 mm vs 20 mm, respectively (P < 0.001 using the ANCOVA test). Thus, the curve describing the risk of PTD according to CL at presentation in twin pregnancies seems to be shifted up and to the right compared with that of singleton pregnancies (Figure 3). In addition, the risk of PTD according to CL at presentation in twin pregnancies was assessed by means of survival analysis (Figure 4). For any given CL at presentation, the proportion of women remaining undelivered at a given interval from the time of presentation with PTL to delivery, was significantly lower in twins than in singletons (Figure 4). The performance of CL as a predictive test for PTD in twin pregnancies was compared with its performance in singletons using several measures. First, the correlation between CL and the examination-to-delivery interval in twin pregnancies was assessed and found to be significant and similar to that of singletons (r = 0.30 vs r = 0.29, respectively; P = 0.9) (Figure 5). Multivariable logistic regression analysis was used to assess the association between CL (as a continuous variable) and risk of PTD in the twin and singleton groups while controlling for potential confounders including maternal age, parity, history of PTD and gestational age at presentation (Table 2). For women with a twin pregnancy, CL was significantly associated with the risk of PTD, so the risk of each PTD outcome variable (i.e. delivery before 37, 34 and 32 weeks’ gestation and within 14 and 7 days from presentation) decreased consistently by a similar magnitude of 5–7% for each additional mm of CL at presentation (odds ratio, 0.93–0.95; Table 2). The degree of association between CL and the risk of each PTD outcome variable was similar to that observed in singleton pregnancies, as reflected by the overlapping CIs (Table 2). In addition, the ability of CL to discriminate between women who will or will not deliver preterm following presentation with PTL, as reflected in the AUC, was similar between twin and singleton pregnancies (range,
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Melamed et al. 0.653–0.724 vs 0.620–0.682, respectively; P = 0.3) (Table 3). Although the performance of CL as a predictive test for PTD appears to be similar in twin and singleton pregnancies, the optimal threshold of CL that should define the test as being positive and subsequently guide clinical decision-making (e.g. administration of steroids for fetal lung maturation) may differ between twins and singletons, considering the differences in the risk of PTD for any given CL observed between the two groups. Thus, in order to determine the optimal threshold of CL in twin pregnancies for the prediction of PTD, we analyzed the effect of the threshold used in the predictive accuracy of CL for PTD in twin and singleton pregnancies. The sensitivity and specificity of CL in the prediction of PTD before 34 weeks’ gestation, as a function of the CL threshold used to define the test as positive, is presented in Figure 6. For any given threshold of CL, the performance of CL in twin pregnancies was associated with a higher sensitivity but a lower specificity (or higher false-positive rate) compared with singletons (Figure 6). A similar pattern was observed for all other outcome variables of PTD (i.e. PTD before 37, 34 and 32 weeks and within 14 and 7 days from presentation, data not shown). This implies that, in order to achieve a certain sensitivity or specificity of CL for the prediction of PTD, a lower threshold of CL should be used for twin pregnancies than for singletons. This is demonstrated in Figure 6 for a hypothetical target of 80% sensitivity or 90% specificity. However, the predictive measures that are more relevant to obstetricians, from the perspective of clinical decision-making, are the PPV and NPV, which are affected not only by the sensitivity and specificity of CL but also by the rate of PTD in the study population. Thus, we subsequently analyzed the effect of the threshold of CL, used to define the test as positive, on the PPV and NPV of CL for predicting PTD (Figure 7). Overall, for any given threshold of CL, the NPV was lower and the PPV was higher in twin pregnancies than in singletons (Figure 7). Assuming that the aim of CL measurement in women presenting with PTL is to maximize the NPV for PTD (as is usually the case), the optimal threshold of CL in twins should be 28–30 mm, which is considerably higher than the threshold associated with a similar NPV in singletons (11.5–12 mm) (Figure 7 and Table 4). Thus, despite the fact that a lower threshold should be used in twin than in singleton pregnancies to achieve a target sensitivity and specificity, the threshold that should be used to achieve a certain NPV is higher for twins than for singletons. This is also illustrated in Table 4, which presents the thresholds that are associated with a sensitivity of 90%, specificity of 90% and NPV of 95% in twin and singleton pregnancies. The rate of a positive test result among the cases, reflecting the proportion of women who will require intervention following presentation with PTL (e.g. admission to hospital and administration of corticosteroids for fetal lung maturation), is also presented in Table 4. In the case of a threshold associated with an NPV of 95% being used,
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
(a)
77
(b)
25
20 20 15 Percent
Percent
15 10
10
5
5
0
0 0
5
10
15 20 25 30 Cervical length (mm)
35
40
45
24
26
28 30 32 34 36 38 40 Gestational age at delivery (weeks)
Figure 2 Distribution of cervical length at presentation (a) and gestational age at delivery (b) in women with singleton ( ) pregnancy presenting with threatened preterm labor. (
(b)
100 80 60 40 20 0
50 40 30 20 10 0
0
5
10
15
20
25
30
35
40
45
0
5
10
Cervical length (mm)
15
20
25
30
35
40
45
15 20 25 30 35 Cervical length (mm)
40
45
Cervical length (mm) (d)
40
Delivery within 14 days of presentation (%)
Delivery before 32 weeks (%)
(c)
) or twin
60 Delivery before 34 weeks (%)
Delivery before 37 weeks (%)
(a)
42
30
20
10
60 50 40 30 20 10 0
0 0
10 20 30 Cervical length (mm)
40
0
5
10
Figure 3 Rate of preterm delivery before 37 weeks (a), 34 weeks (b) and 32 weeks (c) and within 14 days of presentation (d), according to ) and singleton ( ) pregnancies with threatened preterm labor. cervical length at examination in twin (
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Melamed et al.
78 (a) 100
(b) 100 P = 0.006 Proportion undelivered (%)
Proportion undelivered (%)
P = 0.001 80
60
40
20
80
60
40
20
0
0 0
20
40
60
80
100
0
Examination-to-delivery interval (days) (d)
100
P < 0.001
40
60
80
100
Proportion undelivered (%)
Proportion undelivered (%)
(c)
20
100
Examination-to-delivery interval (days)
80
60
40
20
P < 0.001
80
60
40
20
0
0 0
20
40
60
80
0
100
Examination-to-delivery interval (days)
20
40
60
80
100
Examination-to-delivery interval (days)
Figure 4 Kaplan–Meier survival plot showing difference in examination-to-delivery interval between twin ( ) and singleton ( pregnancies following presentation with threatened preterm labor according to cervical length at examination of: (a) < 10 mm; (b) 10–19 mm; (c) 20–29 mm; and (d) ≥ 30 mm. P-values refer to comparison of survival curves using log-rank test. (a)
(b)
r = 0.30, P < 0.001
Examination-to-delivery interval (days)
Examination-to-delivery interval (days)
)
r = 0.29, P < 0.001 120
120
100
80
60
40
100
80
60
40
20
20
0
120
0
10
20
30
40
50
60
Cervical length (mm)
0
0
10
20
30
40
50
60
Cervical length (mm)
Figure 5 Correlation between cervical length at presentation with preterm labor and time-to-delivery interval in twin (a) and singleton (b) pregnancies (r, 0.30 vs 0.29; P = 0.9).
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
79
Table 2 Multivariable logistic regression analysis to determine association of cervical length with risk of preterm delivery at different gestational ages in women with twin or singleton pregnancy presenting with threatened preterm labor Twin (n = 218) Timing of delivery
OR (95% CI)
< 37 weeks < 34 weeks < 32 weeks Within 14 days of examination Within 7 days of examination
0.95 (0.92–0.98) 0.94 (0.92–0.97) 0.93 (0.89–0.97) 0.93 (0.89–0.97) 0.94 (0.89–0.98)
Singleton (n = 1077) P 0.003 < 0.001 0.001 < 0.001 0.006
OR (95% CI) 0.94 (0.93–0.97) 0.94 (0.92–0.96) 0.94 (0.92–0.97) 0.96 (0.92–0.99) 0.94 (0.91–0.96)
P < 0.001 < 0.001 < 0.001 < 0.001 0.008
Maternal age, parity, history of preterm delivery and gestational age at presentation with threatened preterm labor were controlled for. OR, odds ratio. Table 3 Ability of cervical length to identify women, with twin or singleton pregnancy presenting with threatened preterm labor, who will deliver preterm, as reflected by area under receiver–operating characteristics curve (AUC) AUC (95% CI) Timing of delivery < 37 weeks < 34 weeks < 32 weeks Within 14 days of assessment Within 7 days of assessment
Twin
Singleton
P*
0.653 (0.568–0.739) 0.674 (0.589–0.758) 0.724 (0.612–0.835) 0.710 (0.624–0.796) 0.696 (0.595–0.797)
0.653 (0.611–0.694) 0.649 (0.575–0.723) 0.647 (0.538–0.756) 0.682 (0.603–0.761) 0.620 (0.496–0.744)
0.9 0.6 0.3 0.7 0.4
*Determined using method of Hanley and McNeil27 . 100
Sensitivity or specificity (%)
90 80 70 60 50 40 30 20 10 0
10
20
30
40
50
Threshold of cervical length (mm)
Figure 6 Effect of different cervical-length thresholds on sensitivity (solid lines) and specificity (dashed lines) for prediction of preterm delivery before 34 weeks’ gestation in twin (thick lines) and singleton (thin lines) pregnancies. A similar relationship was observed for prediction of preterm delivery before 37, 34 and 32 weeks and within 14 days and 7 days from presentation (data not shown). A target sensitivity of 80% and specificity of 90% are indicated.
the proportion of women with a positive test result would be 55–57% in twins vs 4.2% in singletons (Table 4).
DISCUSSION The aim of this study was to assess the accuracy of sonographic measurement of CL for the prediction of PTD in women with twin pregnancies presenting with PTL and to determine the optimal threshold of CL that should be used in these cases. The main findings of this
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
study are: (1) the risk of PTD for any given CL at the time of presentation with PTL is higher in twin pregnancies than in singletons; (2) the overall performance of CL as a predictive test for PTD in twin pregnancies is similar to that observed in singletons; and (3) the threshold of CL that is needed to achieve a target sensitivity or specificity is lower in twin pregnancies. However, if the aim of measuring CL in women with PTL is to maximize the NPV of the test, a higher threshold (in the range of 28–30 mm) should be used in twin than in singleton pregnancies. The association between CL and the risk of PTD in asymptomatic women with a twin pregnancy has been widely reported in the past4,12 – 20 . However, few studies have investigated the predictive role of CL in symptomatic women presenting with threatened PTL21 – 25 . In a recently published meta-analysis, Liem et al.28 included five studies with a total of 226 women with twin pregnancies presenting with PTL. The variation in definition of preterm delivery (i.e. before 37 weeks or before 34 weeks) and CL cut-off values (i.e. < 30 mm, < 25 mm, < 20 mm) used among studies limited the capacity to draw any conclusions on the effectiveness of CL for the prediction of PTD in symptomatic women with twin pregnancies. Moreover, only a single study25 examined the predictive ability of CL to identify those who are at risk for delivery within 7 days from presentation, which, from a clinical perspective, is probably the more important outcome variable since it affects the decision regarding the need for prompt interventions, such as administration of corticosteroids and tocolysis or transfer to a tertiary-care center. We found that the overall performance of CL as a predictive test for PTD in twins is similar to that observed in singleton pregnancies. This implies that, although PTD represents a final common pathway of different underlying
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Melamed et al.
80
90
90
80
80
70
70
PPV or NPV (%)
(b) 100
PPV or NPV (%)
(a) 100
60 50 40 30
60 50 40 30
20
20
10
10
0
10
20
30
40
50
0
10
20
Threshold of CL (mm)
90
90
80
80
70
70
PPV or NPV (%)
(d) 100
PPV or NPV (%)
(c) 100
60 50 40 30
60
40 30
10
10 20
50
50
20
10
40
60
20
0
30
Threshold of CL (mm)
30
40
50
0
10
20
Threshold of CL (mm)
30
40
50
Threshold of CL (mm)
Figure 7 Effect of cervical length (CL) threshold on positive (PPV) (solid lines) and negative predictive value (NPV) (dashed lines) of CL for prediction of preterm delivery before 34 weeks’ gestation (a), before 32 weeks (b) and within 14 days (c) and within 7 days (d) from presentation in twin (thick lines) and singleton (thin lines) pregnancies. Vertical lines represent threshold of CL that would maximize NPV of CL for preterm delivery in twin pregnancies. Table 4 Performance of cervical length (CL) for prediction of delivery within 14 or 7 days from presentation of threatened preterm labor according to twin or singleton pregnancy, calculated for standardized outcome-specific thresholds Twin
Threshold type Delivery within 14 days Sensitivity = 90% Specificity = 90% NPV = 95% Delivery within 7 days Sensitivity = 90% Specificity = 90% NPV = 95%
Singleton
CL cut-off (mm)
Sens. (%)
Spec. (%)
PPV (%)
NPV (%)
Rate of positive test (%)
CL cut-off (mm)
Sens. (%)
Spec. (%)
PPV (%)
NPV (%)
Rate of positive test (%)
29.5 10.5 30.0
92.1 21.1 86.8
41.7 89.4 48.9
25.0 29.6 26.4
96.2 84.3 94.6
64.2 12.4 57.3
40.0 18.0 12.0
86.4 37.3 13.6
20.3 90.1 96.4
6.2 18.6 18.6
96.1 95.9 94.8
80.1 11.5 4.2
27.5 10.5 28.0
91.3 21.7 82.6
46.7 88.7 48.2
16.8 18.5 15.8
97.8 90.6 95.9
57.3 12.4 55.0
41.5 17.0 11.5
90.0 30.0 3.3
14.0 90.5 95.8
3.1 8.7 2.3
97.9 97.7 97.1
86.1 10.1 4.2
NPV, negative predictive value; PPV, positive predictive value; Sens., sensitivity; Spec., specificity.
mechanisms29,30 and that the pathogenetic mechanisms involved in PTD in twin pregnancies have been suggested to be different in part from those of singletons31 , it appears that, as in singletons, the sonographic measurement of CL in women with twin pregnancies presenting with PTL can be used as a predictive marker for PTD.
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
When considering the predictive value of CL for PTD, the predictive measures that are more relevant to obstetricians from the perspective of clinical decision-making are the PPV and NPV for delivery within 7 or 14 days from the time of examination. Our results indicate that, in order to achieve an NPV of 95% for delivery within 7 or
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy 14 days from examination, the optimal CL threshold in twin pregnancies should be 28–30 mm. This is in contrast to singleton pregnancies, in which the optimal threshold that should be used for that purpose is 11.5–12 mm. We speculate that the reason for this difference is the a-priori increased risk for PTD, which is higher in twins than in singletons, so a higher threshold is needed to achieve reassurance. Our study has several limitations. Apart from its retrospective nature, it is limited by the lack of information regarding certain potential confounders, such as maternal body mass index and the use of assisted reproductive technology. Nevertheless, the study has several points of strength. We evaluated a relatively large cohort of women with strict exclusion criteria, managed in a single center. Moreover, all CL sonographic assessments were carried out by highly-trained personnel according to a well-defined standardized protocol. Finally, we performed a thorough analysis evaluating different outcome variables according to both gestational age at delivery and interval from examination to delivery. In conclusion, the overall performance of CL as a predictive test in women with twin pregnancies who present with PTL is similar to that observed in singleton pregnancies with PTL. However, in order to decrease the risk of a false-negative result to the level achieved in singleton pregnancies, a higher threshold of CL, below which the result is to be considered positive, should be used in twin pregnancies. This implies that women with twin pregnancies who present with PTL are much more likely to have a positive CL test and thus would require interventions more frequently, such as admission to hospital and administration of corticosteroids (>50% of cases). We believe that this information will be helpful to obstetricians for the purpose of decision-making as well as for consultation purposes for women with twin pregnancies presenting with threatened PTL.
REFERENCES 1. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008; 371: 75–84. 2. Robertson PA, Sniderman SH, Laros RK Jr, Cowan R, Heilbron D, Goldenberg RL, Iams JD, Creasy RK. Neonatal morbidity according to gestational age and birth weight from five tertiary care centers in the United States, 1983 through 1986. Am J Obstet Gynecol 1992; 166: 1629–1641; discussion 1641–1645. 3. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das A, Thom E, McNellis D, Copper RL, Johnson F, Roberts JM. The length of the cervix and the risk of spontaneous premature delivery. National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996; 334: 567–572. 4. Conde-Agudelo A, Romero R, Hassan SS, Yeo L. Transvaginal sonographic cervical length for the prediction of spontaneous preterm birth in twin pregnancies: a systematic review and metaanalysis. Am J Obstet Gynecol 2010; 203: 128.e1–12. 5. Makrydimas G, Sotiriadis A. Prediction of preterm birth in twins. Best Pract Res Clin Obstet Gynaecol 2014; 28: 265–272. 6. Hee L. Likelihood ratios for the prediction of preterm delivery with biomarkers. Acta Obstet Gynecol Scand 2011; 90: 1189–1199. 7. Tsoi E, Fuchs IB, Rane S, Geerts L, Nicolaides KH. Sonographic measurement
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
81
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29. 30.
31.
of cervical length in threatened preterm labor in singleton pregnancies with intact membranes. Ultrasound Obstet Gynecol 2005; 25: 353–356. ´ ´ Palacio M, Sanin-Blair J, Sanchez M, Crispi F, Gomez O, Carreras E, Coll O, ´ E. The use of a variable cut-off value of cervical length in Cararach V, Gratacos women admitted for preterm labor before and after 32 weeks. Ultrasound Obstet Gynecol 2007; 29: 421–426. Wulff CB, Ekelund CK, Hedegaard M, Tabor A. Can a 15-mm cervical length cutoff discriminate between low and high risk of preterm delivery in women with threatened preterm labor? Fetal Diagn Ther 2011; 29: 216–223. Gomez R, Romero R, Nien JK, Chaiworapongsa T, Medina L, Kim YM, Yoon BH, Carstens M, Espinoza J, Iams JD, Gonzalez R. A short cervix in women with preterm labor and intact membranes: a risk factor for microbial invasion of the amniotic cavity. Am J Obstet Gynecol 2005; 192: 678–689. Sotiriadis A, Papatheodorou S, Kavvadias A, Makrydimas G. Transvaginal cervical length measurement for prediction of preterm birth in women with threatened preterm labor: a meta-analysis. Ultrasound Obstet Gynecol 2010; 35: 54–64. Soriano D, Weisz B, Seidman DS, Chetrit A, Schiff E, Lipitz S, Achiron R. The role of sonographic assessment of cervical length in the prediction of preterm birth in primigravidae with twin gestation conceived after infertility treatment. Acta Obstet Gynecol Scand 2002; 81: 39–43. Imseis HM, Albert TA, Iams JD. Identifying twin gestations at low risk for preterm birth with a transvaginal ultrasonographic cervical measurement at 24 to 26 weeks’ gestation. Am J Obstet Gynecol 1997; 177: 1149–1155. To MS, Fonseca EB, Molina FS, Cacho AM, Nicolaides KH. Maternal characteristics and cervical length in the prediction of spontaneous early preterm delivery in twins. Am J Obstet Gynecol 2006; 194: 1360–1365. Sperling L, Kiil C, Larsen LU, Qvist I, Bach D, Wøjdemann K, Bladh A, Nikkila¨ A, Jørgensen C, Skajaa K, Bang J, Tabor A. How to identify twins at low risk of spontaneous preterm delivery. Ultrasound Obstet Gynecol 2005; 26: 138–144. Klein K, Gregor H, Hirtenlehner-Ferber K, Stammler-Safar M, Witt A, Hanslik A, Husslein P, Krampl E. Prediction of spontaneous preterm delivery in twin pregnancies by cervical length at mid-gestation. Twin Res Hum Genet 2008; 11: 552–557. Fox NS, Rebarber A, Klauser CK, Peress D, Gutierrez CV, Saltzman DH. Prediction of spontaneous preterm birth in asymptomatic twin pregnancies using the change in cervical length over time. Am J Obstet Gynecol 2010; 202: 155.e1–4. Fox NS, Rebarber A, Roman AS, Klauser CK, Saltzman DH. Association between second-trimester cervical length and spontaneous preterm birth in twin pregnancies. J Ultrasound Med 2010; 29: 1733–1739. Yang JH, Kuhlman K, Daly S, Berghella V. Prediction of preterm birth by second trimester cervical sonography in twin pregnancies. Ultrasound Obstet Gynecol 2000; 15: 288–291. Guzman ER, Walters C, O’reilly-Green C, Kinzler WL, Waldron R, Nigam J, Vintzileos AM. Use of cervical ultrasonography in prediction of spontaneous preterm birth in twin gestations. Am J Obstet Gynecol 2000; 183: 1103–1107. Crane JM, Van den Hof M, Armson BA, Liston R. Transvaginal ultrasound in the prediction of preterm delivery: singleton and twin gestations. Obstet Gynecol 1997; 90: 357–363. Gonzalez N, Bige V, Kandoussi S, Graesslin O, Quereux C, Gabriel R. [Ultrasonographic measurement of cervical length in twin pregnancies with preterm labor: comparison with singleton pregnancies]. Gynecol Obstet Fertil 2004; 32: 122–127. Vendittelli F, Volumenie J. Transvaginal ultrasonography examination of the uterine cervix in hospitalised women undergoing preterm labour. Eur J Obstet Gynecol Reprod Biol 2000; 90: 3–11. Yoshizato T, Inoue Y, Fukami T, Sanui A, Miyamato S, Kawarabayashi T. Longitudinal changes in canal length at 16–35 weeks in normal twin pregnancies and twin pregnancies with preterm labor and delivery. J Obstet Gynaecol Res 2010; 36: 733–738. Fuchs I, Tsoi E, Henrich W, Dudenhausen JW, Nicolaides KH. Sonographic measurement of cervical length in twin pregnancies in threatened preterm labor. Ultrasound Obstet Gynecol 2004; 23: 42–45. Mella MT, Berghella V. Prediction of preterm birth: cervical sonography. Semin Perinatol 2009; 33: 317–324. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: 839–843. Liem SM, van de Mheen L, Bekedam DJ, van Pampus MG, Opmeer BC, Lim AC, Mol BW. Cervical length measurement for the prediction of preterm birth in symptomatic women with a twin pregnancy: a systematic review and meta-analysis. Obstet Gynecol Int 2013; 2013: 125897. Romero R, Yeo L, Chaemsaithong P, Chaiworapongsa T, Hassan SS. The preterm parturition syndrome. BJOG 2006; 113: 17–42. Gotsch F, Gotsch F, Romero R, Erez O, Vaisbuch E, Kusanovic JP, Mazaki-Tovi S, Kim SK, Hassan S, Yeo L. The preterm parturition syndrome and its implications for understanding the biology, risk assessment, diagnosis, treatment and prevention of preterm birth. J Matern Fetal Neonatal Med 2009; 22: 5–23. Gardner MO, Goldenberg RL, Cliver SP, Tucker JM, Nelson KG, Copper RL. The origin and outcome of preterm twin pregnancies. Obstet Gynecol 1995; 85: 553–557.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Predictive value of cervical length in women with twin pregnancy presenting with threatened preterm labor N. MELAMED, L. HIERSCH, R. GABBAY-BENZIV, R. BARDIN, I. MEIZNER, A. WIZNITZER and Y. YOGEV Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva, Israel; Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
K E Y W O R D S: cervical length; prediction; preterm labor; twins
ABSTRACT Objective To assess the accuracy and determine the optimal threshold of sonographic cervical length (CL) for the prediction of preterm delivery (PTD) in women with twin pregnancies presenting with threatened preterm labor (PTL). Methods This was a retrospective study of women with twin pregnancies who presented with threatened PTL and underwent sonographic measurement of CL in a tertiary center. The accuracy of CL in predicting PTD in women with twin pregnancies was compared with that in a control group of women with singleton pregnancies. Results Overall, 218 women with a twin pregnancy and 1077 women with a singleton pregnancy, who presented with PTL, were included in the study. The performance of CL as a predictive test for PTD was similar in twins and singletons, as reflected by the similar correlation between CL and the examination-to-delivery interval (r, 0.30 vs 0.29; P = 0.9), the similar association of CL with risk of PTD, and the similar areas under the receiver–operating characteristics curves for differing delivery outcomes (range, 0.653–0.724 vs 0.620–0.682, respectively; P = 0.3). The optimal threshold of CL for any given target sensitivity or specificity was lower in twin than in singleton pregnancies. However, in order to achieve a negative predictive value of 95%, a higher threshold (28–30 mm) should be used in twin pregnancies. Using this twin-specific CL threshold, women with twins who present with PTL are more likely to have a positive CL test, and therefore to require subsequent interventions, than are women with singleton pregnancies with PTL (55% vs 4.2%, respectively). Conclusion In women with PTL, the performance of CL as a test for the prediction of PTD is similar in twin and
singleton pregnancies. However, the optimal threshold of CL for the prediction of PTD appears to be higher in twin pregnancies, mainly owing to the higher baseline risk for PTD in these pregnancies. Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
INTRODUCTION Prematurity remains the foremost cause of neonatal mortality and morbidity1,2 . Although the rate of preterm delivery (PTD) before 34 weeks’ gestation is approximately 1% among singleton pregnancies3 , it is reported to be as high as 13% in twin pregnancies4 , thus twin pregnancy is considered to be one of the strongest risk factors for PTD5,6 . One of the challenges in the management of women who present with preterm labor (PTL) is the distinction between true and false PTL, since fewer than 15% of those presenting with threatened PTL will actually deliver within 7 days of presentation or before 35 weeks’ gestation7,8 . One of the most widely investigated tools for the prediction of PTD in women with a singleton pregnancy and threatened PTL is sonographic measurement of cervical length (CL)7 – 11 . However, while there is a significant amount of evidence regarding the role of CL in asymptomatic women with twin pregnancies4,12 – 20 , there is only limited information regarding the performance of CL in women with twin pregnancies who present with threatened PTL21 – 25 . The aim of this study was to assess the accuracy of sonographic CL in the prediction of PTD in women with a twin pregnancy presenting with threatened PTL, using women with a singleton pregnancy as controls, and to determine the optimal threshold of CL for its prediction in twins.
Correspondence to: Dr N. Melamed, Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petah Tiqwa 49100, Israel (e-mail: [email protected]) Accepted: 8 September 2014
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
ORIGINAL PAPER
Melamed et al.
74
METHODS This was a retrospective cohort study of all women who presented with threatened PTL in the presence of intact membranes and underwent sonographic measurement of CL in a tertiary referral medical center between January 2007 and December 2012. The study group included women with twin pregnancies who presented with threatened PTL. The predictive accuracy of CL in the twin group was compared with that observed in a control group of women with a singleton pregnancy presenting with threatened PTL. Only women who presented at a gestational age of 24+0 to 33+6 weeks were included in either group. Exclusion criteria from both the study and control groups included women with high-order multiple gestations (triplets and above), monoamniotic twins, complicated monochorionic twins (twin–twin transfusion syndrome or twin anemia–polycythemia sequence), cervical cerclage, cervical dilatation > 3 cm at presentation, pregnancies complicated by placental abruption, chorioamnionitis, stillbirth or major fetal anomalies, and women who underwent indicated preterm delivery before 37+0 weeks’ gestation for any maternal or fetal indication or who did not deliver in our medical center. The study was approved by the local institutional review board. Women were identified using the comprehensive database of sonographic examinations in our ultrasound unit. This database contains data on all sonographic examinations performed in the ultrasound unit, which are entered in real time by the performer of the examination immediately after its completion. Initially, all women who underwent sonographic measurement of CL using transvaginal ultrasound were identified and the information regarding CL, presence of funneling, change in CL during the examination (either spontaneously by mild fundal pressure or in response to Valsalva maneuver) was extracted. The medical charts of these women were then reviewed thoroughly in order to identify only those for whom the indication for sonographic measurement of CL was threatened PTL. Of these women, the following information was extracted: demographic, medical and obstetric history, number of fetuses (singleton vs twins), complications during current pregnancy, validation of gestational age by first-trimester ultrasound, gestational age at presentation, frequency and regularity of uterine contractions, results of vaginal examination, interventions performed for threatened PTL, gestational age at delivery and onset of delivery (spontaneous vs induced). For women with repeat measurements of CL at different times during gestation, only the first measurement was included in the analysis. Threatened PTL was defined as the presence of at least three regular and painful uterine contractions within a 30-min period. Tocolysis with nifedipine, indomethacin or atosiban and betamethasone was administrated for fetal lung maturation, based on the decision of the treating physician. All sonographic examinations were performed by physicians specialized in ultrasonography,
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
or by experienced sonographers, using the following ultrasound systems: Voluson E8 and Voluson 730 Expert (GE Medical Systems, Kretz Ultrasound, Zipf, Austria) and ATL 5000 (Philips Healthcare, Best, The Netherlands). Measurement of CL was performed transvaginally after the patients had emptied their bladder and according to the standard technique26 . Briefly, the measurement was performed in the sagittal plane, visualizing the full length of the endocervical canal, from the internal cervical os to the external cervical os, while exerting as little pressure as possible with the transducer. At least three measurements were obtained and the shortest measurement was recorded. In addition, the presence of cervical funneling or a change in CL, either spontaneously or in response to Valsalva maneuver, was documented routinely.
Statistical analysis Data analysis was performed with SPSS v. 21.0 software (IBM Corp., Armonk, NY, USA). The accuracy of CL in predicting spontaneous PTD before 37, 34 and 32 weeks’ gestation and within 14 and 7 days of presentation in women with a twin pregnancy was calculated and compared with the predictive accuracy in singleton pregnancies. Student’s t-test and the Mann–Whitney U-test were used to compare continuous variables, with and without normal distribution, between the twin and singleton groups, respectively, while the chi-square and Fisher’s exact tests were used for categorical variables. The rate of PTD at a gestational age of less than 37, 34 and 32 weeks and within 14 days of presentation was calculated and plotted as a function of CL at presentation for women in the twin and singleton groups. Potential thresholds of CL, below which the risk of PTD appeared to increase, were identified initially through visual inspection of the graphs and were then confirmed by comparison of the slopes of the regression lines (describing the rate of PTD as a function of CL) above and below these thresholds, using the analysis of covariance (ANCOVA) test. Kaplan–Meier survival analysis was used to compare the proportion of pregnancies undelivered according to time from presentation between the twin and singleton groups, stratified by CL at presentation. The log-rank test was used to compare the survival distributions. Spearman’s correlation coefficient was used to assess the correlation between CL at presentation and examination-to-delivery interval. Correlation coefficients for the two groups were compared using Fisher’s Z-transformation. Multivariable logistic regression analysis was used to assess the association between CL as a continuous measure and the risk of PTD in the twin and singleton groups, while controlling for potential confounders. Receiver–operating characteristics (ROC) analysis was used to determine the area under the ROC curve (AUC) as an overall measure of the discriminative ability of CL. The AUCs for the twin and singleton
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
75
groups were compared using the method of Hanley and McNeil27 . ROC analysis was also used to calculate standardized outcome-specific thresholds, including: (1) the threshold associated with a sensitivity of 90%; (2) the threshold associated with a specificity of 90%; and (3) the threshold associated with a negative predictive value (NPV) of 95%. The following measures of predictive accuracy were calculated for each of the outcome variables: sensitivity, specificity, positive predictive value (PPV) and NPV; P < 0.05 was considered statistically significant.
RESULTS A total of 2680 sonographic examinations of CL were performed in 1950 women during the study period. Of those, 218 women with a twin pregnancy were found to be eligible for the study and were compared with 1077 eligible women with a singleton pregnancy (Figure 1). The demographic characteristics of women with twin pregnancies and those with singleton pregnancies were similar, except for a higher maternal age in the twin group (Table 1). There were no differences between
Sonographic measurements of CL in women presenting with preterm labor between 24 +0 and 33+ 6 weeks during study period (n = 2680 measurements)
Twin CL measurements (n = 685)
Excluded (n = 467)
Twin pregnancies (n = 218)
Exclusion criteria: High-order multiple gestation Repeated measurements TTTS or TAPS Monoamniotic twins Cervical cerclage Indicated preterm delivery < 37+ 0 weeks Major anomalies Placental abruption Amnionitis Stillbirth Dilatation > 3 cm Incomplete data or delivery in another center
Singleton CL measurements (n = 1995)
Excluded (n = 918)
Singleton pregnancies (n = 1077)
Figure 1 Flowchart summarizing the study groups. CL, cervical length; TAPS, twin anemia–polycythemia sequence; TTTS, twin–twin transfusion syndrome. Table 1 Demographic and pregnancy characteristics of women with twin or singleton pregnancy presenting with threatened preterm labor Characteristic Maternal age (years) Maternal age > 35 years Parity Nulliparous History of preterm delivery Chorionicity DC/DA twins MC/MA twins Unknown Gestational age at examination (weeks) Cervical length (mm) Intervention Progesterone Tocolysis Corticosteroids Delivery outcome Gestational age at delivery (weeks) Gestational age at delivery < 37 + 0 weeks Gestational age at delivery < 34 + 0 weeks Examination-to-delivery interval (days) Examination-to-delivery interval within 7 days Examination-to-delivery interval within 14 days
Twin (n = 218)
Singleton (n = 1077)
34.5 ± 5.2 84 (38.5) 2 (1–2) 108 (49.5) 22 (10.1)
33.1 ± 5.6 325 (30.2) 2 (1–3) 497 (46.1) 152 (14.1)
< 0.001 0.02 0.1 0.4 0.1
165 (75.7) 27 (12.4) 26 (11.9) 29.5 ± 2.8 24.9 ± 11.2
— — — 29.8 ± 2.8 30.9 ± 10.2
— — — 0.1 < 0.001
31 (14.2) 190 (87.2) 195 (89.4)
86 (8.0) 686 (63.7) 720 (66.9)
0.003 < 0.001 < 0.001
34.3 ± 2.5 180 (82.6) 63 (28.9) 36.3 ± 21.7 23 (10.6) 38 (17.4)
37.4 ± 2.7 274 (25.4) 77 (7.1) 56.3 ± 25.2 29 (2.7) 60 (5.6)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
P
Data are presented as mean ± SD, n (%) or median (interquartile range). DA, diamniotic; DC, dichorionic; MA, monoamniotic; MC, monochorionic. Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
76
the two groups with respect to parity, history of PTD and gestational age at presentation with threatened PTL (Table 1). Women in the twin group were characterized by a significantly shorter cervix at presentation and a higher rate of subsequent PTD, and were more likely to receive progesterone, tocolysis or corticosteroids for fetal lung maturation than were women in the singleton group (Table 1). Figure 2 presents the distribution curves of CL at presentation and gestational age at delivery according to the pregnancy group. Both curves were shifted to the left in twins, as compared to singletons (Figure 2). The rates of spontaneous PTD before 37, 34 and 32 weeks’ gestation and within 14 days from presentation, as a function of CL at presentation, are shown in Figure 3. The rate of PTD for any given CL at presentation was higher in twin than in singleton pregnancies (Figure 3). In addition, the threshold of CL at presentation, below which the risk of spontaneous PTD began to increase, was higher in twin than in singleton pregnancies (Figure 3): < 37 weeks, 40 mm vs 30 mm; < 34 weeks, 30 mm vs 20 mm; < 32 weeks, 30 mm vs 20 mm; within 14 days of assessment, 30 mm vs 20 mm, respectively (P < 0.001 using the ANCOVA test). Thus, the curve describing the risk of PTD according to CL at presentation in twin pregnancies seems to be shifted up and to the right compared with that of singleton pregnancies (Figure 3). In addition, the risk of PTD according to CL at presentation in twin pregnancies was assessed by means of survival analysis (Figure 4). For any given CL at presentation, the proportion of women remaining undelivered at a given interval from the time of presentation with PTL to delivery, was significantly lower in twins than in singletons (Figure 4). The performance of CL as a predictive test for PTD in twin pregnancies was compared with its performance in singletons using several measures. First, the correlation between CL and the examination-to-delivery interval in twin pregnancies was assessed and found to be significant and similar to that of singletons (r = 0.30 vs r = 0.29, respectively; P = 0.9) (Figure 5). Multivariable logistic regression analysis was used to assess the association between CL (as a continuous variable) and risk of PTD in the twin and singleton groups while controlling for potential confounders including maternal age, parity, history of PTD and gestational age at presentation (Table 2). For women with a twin pregnancy, CL was significantly associated with the risk of PTD, so the risk of each PTD outcome variable (i.e. delivery before 37, 34 and 32 weeks’ gestation and within 14 and 7 days from presentation) decreased consistently by a similar magnitude of 5–7% for each additional mm of CL at presentation (odds ratio, 0.93–0.95; Table 2). The degree of association between CL and the risk of each PTD outcome variable was similar to that observed in singleton pregnancies, as reflected by the overlapping CIs (Table 2). In addition, the ability of CL to discriminate between women who will or will not deliver preterm following presentation with PTL, as reflected in the AUC, was similar between twin and singleton pregnancies (range,
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Melamed et al. 0.653–0.724 vs 0.620–0.682, respectively; P = 0.3) (Table 3). Although the performance of CL as a predictive test for PTD appears to be similar in twin and singleton pregnancies, the optimal threshold of CL that should define the test as being positive and subsequently guide clinical decision-making (e.g. administration of steroids for fetal lung maturation) may differ between twins and singletons, considering the differences in the risk of PTD for any given CL observed between the two groups. Thus, in order to determine the optimal threshold of CL in twin pregnancies for the prediction of PTD, we analyzed the effect of the threshold used in the predictive accuracy of CL for PTD in twin and singleton pregnancies. The sensitivity and specificity of CL in the prediction of PTD before 34 weeks’ gestation, as a function of the CL threshold used to define the test as positive, is presented in Figure 6. For any given threshold of CL, the performance of CL in twin pregnancies was associated with a higher sensitivity but a lower specificity (or higher false-positive rate) compared with singletons (Figure 6). A similar pattern was observed for all other outcome variables of PTD (i.e. PTD before 37, 34 and 32 weeks and within 14 and 7 days from presentation, data not shown). This implies that, in order to achieve a certain sensitivity or specificity of CL for the prediction of PTD, a lower threshold of CL should be used for twin pregnancies than for singletons. This is demonstrated in Figure 6 for a hypothetical target of 80% sensitivity or 90% specificity. However, the predictive measures that are more relevant to obstetricians, from the perspective of clinical decision-making, are the PPV and NPV, which are affected not only by the sensitivity and specificity of CL but also by the rate of PTD in the study population. Thus, we subsequently analyzed the effect of the threshold of CL, used to define the test as positive, on the PPV and NPV of CL for predicting PTD (Figure 7). Overall, for any given threshold of CL, the NPV was lower and the PPV was higher in twin pregnancies than in singletons (Figure 7). Assuming that the aim of CL measurement in women presenting with PTL is to maximize the NPV for PTD (as is usually the case), the optimal threshold of CL in twins should be 28–30 mm, which is considerably higher than the threshold associated with a similar NPV in singletons (11.5–12 mm) (Figure 7 and Table 4). Thus, despite the fact that a lower threshold should be used in twin than in singleton pregnancies to achieve a target sensitivity and specificity, the threshold that should be used to achieve a certain NPV is higher for twins than for singletons. This is also illustrated in Table 4, which presents the thresholds that are associated with a sensitivity of 90%, specificity of 90% and NPV of 95% in twin and singleton pregnancies. The rate of a positive test result among the cases, reflecting the proportion of women who will require intervention following presentation with PTL (e.g. admission to hospital and administration of corticosteroids for fetal lung maturation), is also presented in Table 4. In the case of a threshold associated with an NPV of 95% being used,
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
(a)
77
(b)
25
20 20 15 Percent
Percent
15 10
10
5
5
0
0 0
5
10
15 20 25 30 Cervical length (mm)
35
40
45
24
26
28 30 32 34 36 38 40 Gestational age at delivery (weeks)
Figure 2 Distribution of cervical length at presentation (a) and gestational age at delivery (b) in women with singleton ( ) pregnancy presenting with threatened preterm labor. (
(b)
100 80 60 40 20 0
50 40 30 20 10 0
0
5
10
15
20
25
30
35
40
45
0
5
10
Cervical length (mm)
15
20
25
30
35
40
45
15 20 25 30 35 Cervical length (mm)
40
45
Cervical length (mm) (d)
40
Delivery within 14 days of presentation (%)
Delivery before 32 weeks (%)
(c)
) or twin
60 Delivery before 34 weeks (%)
Delivery before 37 weeks (%)
(a)
42
30
20
10
60 50 40 30 20 10 0
0 0
10 20 30 Cervical length (mm)
40
0
5
10
Figure 3 Rate of preterm delivery before 37 weeks (a), 34 weeks (b) and 32 weeks (c) and within 14 days of presentation (d), according to ) and singleton ( ) pregnancies with threatened preterm labor. cervical length at examination in twin (
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Melamed et al.
78 (a) 100
(b) 100 P = 0.006 Proportion undelivered (%)
Proportion undelivered (%)
P = 0.001 80
60
40
20
80
60
40
20
0
0 0
20
40
60
80
100
0
Examination-to-delivery interval (days) (d)
100
P < 0.001
40
60
80
100
Proportion undelivered (%)
Proportion undelivered (%)
(c)
20
100
Examination-to-delivery interval (days)
80
60
40
20
P < 0.001
80
60
40
20
0
0 0
20
40
60
80
0
100
Examination-to-delivery interval (days)
20
40
60
80
100
Examination-to-delivery interval (days)
Figure 4 Kaplan–Meier survival plot showing difference in examination-to-delivery interval between twin ( ) and singleton ( pregnancies following presentation with threatened preterm labor according to cervical length at examination of: (a) < 10 mm; (b) 10–19 mm; (c) 20–29 mm; and (d) ≥ 30 mm. P-values refer to comparison of survival curves using log-rank test. (a)
(b)
r = 0.30, P < 0.001
Examination-to-delivery interval (days)
Examination-to-delivery interval (days)
)
r = 0.29, P < 0.001 120
120
100
80
60
40
100
80
60
40
20
20
0
120
0
10
20
30
40
50
60
Cervical length (mm)
0
0
10
20
30
40
50
60
Cervical length (mm)
Figure 5 Correlation between cervical length at presentation with preterm labor and time-to-delivery interval in twin (a) and singleton (b) pregnancies (r, 0.30 vs 0.29; P = 0.9).
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy
79
Table 2 Multivariable logistic regression analysis to determine association of cervical length with risk of preterm delivery at different gestational ages in women with twin or singleton pregnancy presenting with threatened preterm labor Twin (n = 218) Timing of delivery
OR (95% CI)
< 37 weeks < 34 weeks < 32 weeks Within 14 days of examination Within 7 days of examination
0.95 (0.92–0.98) 0.94 (0.92–0.97) 0.93 (0.89–0.97) 0.93 (0.89–0.97) 0.94 (0.89–0.98)
Singleton (n = 1077) P 0.003 < 0.001 0.001 < 0.001 0.006
OR (95% CI) 0.94 (0.93–0.97) 0.94 (0.92–0.96) 0.94 (0.92–0.97) 0.96 (0.92–0.99) 0.94 (0.91–0.96)
P < 0.001 < 0.001 < 0.001 < 0.001 0.008
Maternal age, parity, history of preterm delivery and gestational age at presentation with threatened preterm labor were controlled for. OR, odds ratio. Table 3 Ability of cervical length to identify women, with twin or singleton pregnancy presenting with threatened preterm labor, who will deliver preterm, as reflected by area under receiver–operating characteristics curve (AUC) AUC (95% CI) Timing of delivery < 37 weeks < 34 weeks < 32 weeks Within 14 days of assessment Within 7 days of assessment
Twin
Singleton
P*
0.653 (0.568–0.739) 0.674 (0.589–0.758) 0.724 (0.612–0.835) 0.710 (0.624–0.796) 0.696 (0.595–0.797)
0.653 (0.611–0.694) 0.649 (0.575–0.723) 0.647 (0.538–0.756) 0.682 (0.603–0.761) 0.620 (0.496–0.744)
0.9 0.6 0.3 0.7 0.4
*Determined using method of Hanley and McNeil27 . 100
Sensitivity or specificity (%)
90 80 70 60 50 40 30 20 10 0
10
20
30
40
50
Threshold of cervical length (mm)
Figure 6 Effect of different cervical-length thresholds on sensitivity (solid lines) and specificity (dashed lines) for prediction of preterm delivery before 34 weeks’ gestation in twin (thick lines) and singleton (thin lines) pregnancies. A similar relationship was observed for prediction of preterm delivery before 37, 34 and 32 weeks and within 14 days and 7 days from presentation (data not shown). A target sensitivity of 80% and specificity of 90% are indicated.
the proportion of women with a positive test result would be 55–57% in twins vs 4.2% in singletons (Table 4).
DISCUSSION The aim of this study was to assess the accuracy of sonographic measurement of CL for the prediction of PTD in women with twin pregnancies presenting with PTL and to determine the optimal threshold of CL that should be used in these cases. The main findings of this
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
study are: (1) the risk of PTD for any given CL at the time of presentation with PTL is higher in twin pregnancies than in singletons; (2) the overall performance of CL as a predictive test for PTD in twin pregnancies is similar to that observed in singletons; and (3) the threshold of CL that is needed to achieve a target sensitivity or specificity is lower in twin pregnancies. However, if the aim of measuring CL in women with PTL is to maximize the NPV of the test, a higher threshold (in the range of 28–30 mm) should be used in twin than in singleton pregnancies. The association between CL and the risk of PTD in asymptomatic women with a twin pregnancy has been widely reported in the past4,12 – 20 . However, few studies have investigated the predictive role of CL in symptomatic women presenting with threatened PTL21 – 25 . In a recently published meta-analysis, Liem et al.28 included five studies with a total of 226 women with twin pregnancies presenting with PTL. The variation in definition of preterm delivery (i.e. before 37 weeks or before 34 weeks) and CL cut-off values (i.e. < 30 mm, < 25 mm, < 20 mm) used among studies limited the capacity to draw any conclusions on the effectiveness of CL for the prediction of PTD in symptomatic women with twin pregnancies. Moreover, only a single study25 examined the predictive ability of CL to identify those who are at risk for delivery within 7 days from presentation, which, from a clinical perspective, is probably the more important outcome variable since it affects the decision regarding the need for prompt interventions, such as administration of corticosteroids and tocolysis or transfer to a tertiary-care center. We found that the overall performance of CL as a predictive test for PTD in twins is similar to that observed in singleton pregnancies. This implies that, although PTD represents a final common pathway of different underlying
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Melamed et al.
80
90
90
80
80
70
70
PPV or NPV (%)
(b) 100
PPV or NPV (%)
(a) 100
60 50 40 30
60 50 40 30
20
20
10
10
0
10
20
30
40
50
0
10
20
Threshold of CL (mm)
90
90
80
80
70
70
PPV or NPV (%)
(d) 100
PPV or NPV (%)
(c) 100
60 50 40 30
60
40 30
10
10 20
50
50
20
10
40
60
20
0
30
Threshold of CL (mm)
30
40
50
0
10
20
Threshold of CL (mm)
30
40
50
Threshold of CL (mm)
Figure 7 Effect of cervical length (CL) threshold on positive (PPV) (solid lines) and negative predictive value (NPV) (dashed lines) of CL for prediction of preterm delivery before 34 weeks’ gestation (a), before 32 weeks (b) and within 14 days (c) and within 7 days (d) from presentation in twin (thick lines) and singleton (thin lines) pregnancies. Vertical lines represent threshold of CL that would maximize NPV of CL for preterm delivery in twin pregnancies. Table 4 Performance of cervical length (CL) for prediction of delivery within 14 or 7 days from presentation of threatened preterm labor according to twin or singleton pregnancy, calculated for standardized outcome-specific thresholds Twin
Threshold type Delivery within 14 days Sensitivity = 90% Specificity = 90% NPV = 95% Delivery within 7 days Sensitivity = 90% Specificity = 90% NPV = 95%
Singleton
CL cut-off (mm)
Sens. (%)
Spec. (%)
PPV (%)
NPV (%)
Rate of positive test (%)
CL cut-off (mm)
Sens. (%)
Spec. (%)
PPV (%)
NPV (%)
Rate of positive test (%)
29.5 10.5 30.0
92.1 21.1 86.8
41.7 89.4 48.9
25.0 29.6 26.4
96.2 84.3 94.6
64.2 12.4 57.3
40.0 18.0 12.0
86.4 37.3 13.6
20.3 90.1 96.4
6.2 18.6 18.6
96.1 95.9 94.8
80.1 11.5 4.2
27.5 10.5 28.0
91.3 21.7 82.6
46.7 88.7 48.2
16.8 18.5 15.8
97.8 90.6 95.9
57.3 12.4 55.0
41.5 17.0 11.5
90.0 30.0 3.3
14.0 90.5 95.8
3.1 8.7 2.3
97.9 97.7 97.1
86.1 10.1 4.2
NPV, negative predictive value; PPV, positive predictive value; Sens., sensitivity; Spec., specificity.
mechanisms29,30 and that the pathogenetic mechanisms involved in PTD in twin pregnancies have been suggested to be different in part from those of singletons31 , it appears that, as in singletons, the sonographic measurement of CL in women with twin pregnancies presenting with PTL can be used as a predictive marker for PTD.
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
When considering the predictive value of CL for PTD, the predictive measures that are more relevant to obstetricians from the perspective of clinical decision-making are the PPV and NPV for delivery within 7 or 14 days from the time of examination. Our results indicate that, in order to achieve an NPV of 95% for delivery within 7 or
Ultrasound Obstet Gynecol 2015; 46: 73–81.
Cervical length in twin pregnancy 14 days from examination, the optimal CL threshold in twin pregnancies should be 28–30 mm. This is in contrast to singleton pregnancies, in which the optimal threshold that should be used for that purpose is 11.5–12 mm. We speculate that the reason for this difference is the a-priori increased risk for PTD, which is higher in twins than in singletons, so a higher threshold is needed to achieve reassurance. Our study has several limitations. Apart from its retrospective nature, it is limited by the lack of information regarding certain potential confounders, such as maternal body mass index and the use of assisted reproductive technology. Nevertheless, the study has several points of strength. We evaluated a relatively large cohort of women with strict exclusion criteria, managed in a single center. Moreover, all CL sonographic assessments were carried out by highly-trained personnel according to a well-defined standardized protocol. Finally, we performed a thorough analysis evaluating different outcome variables according to both gestational age at delivery and interval from examination to delivery. In conclusion, the overall performance of CL as a predictive test in women with twin pregnancies who present with PTL is similar to that observed in singleton pregnancies with PTL. However, in order to decrease the risk of a false-negative result to the level achieved in singleton pregnancies, a higher threshold of CL, below which the result is to be considered positive, should be used in twin pregnancies. This implies that women with twin pregnancies who present with PTL are much more likely to have a positive CL test and thus would require interventions more frequently, such as admission to hospital and administration of corticosteroids (>50% of cases). We believe that this information will be helpful to obstetricians for the purpose of decision-making as well as for consultation purposes for women with twin pregnancies presenting with threatened PTL.
REFERENCES 1. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008; 371: 75–84. 2. Robertson PA, Sniderman SH, Laros RK Jr, Cowan R, Heilbron D, Goldenberg RL, Iams JD, Creasy RK. Neonatal morbidity according to gestational age and birth weight from five tertiary care centers in the United States, 1983 through 1986. Am J Obstet Gynecol 1992; 166: 1629–1641; discussion 1641–1645. 3. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das A, Thom E, McNellis D, Copper RL, Johnson F, Roberts JM. The length of the cervix and the risk of spontaneous premature delivery. National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996; 334: 567–572. 4. Conde-Agudelo A, Romero R, Hassan SS, Yeo L. Transvaginal sonographic cervical length for the prediction of spontaneous preterm birth in twin pregnancies: a systematic review and metaanalysis. Am J Obstet Gynecol 2010; 203: 128.e1–12. 5. Makrydimas G, Sotiriadis A. Prediction of preterm birth in twins. Best Pract Res Clin Obstet Gynaecol 2014; 28: 265–272. 6. Hee L. Likelihood ratios for the prediction of preterm delivery with biomarkers. Acta Obstet Gynecol Scand 2011; 90: 1189–1199. 7. Tsoi E, Fuchs IB, Rane S, Geerts L, Nicolaides KH. Sonographic measurement
Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd.
81
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29. 30.
31.
of cervical length in threatened preterm labor in singleton pregnancies with intact membranes. Ultrasound Obstet Gynecol 2005; 25: 353–356. ´ ´ Palacio M, Sanin-Blair J, Sanchez M, Crispi F, Gomez O, Carreras E, Coll O, ´ E. The use of a variable cut-off value of cervical length in Cararach V, Gratacos women admitted for preterm labor before and after 32 weeks. Ultrasound Obstet Gynecol 2007; 29: 421–426. Wulff CB, Ekelund CK, Hedegaard M, Tabor A. Can a 15-mm cervical length cutoff discriminate between low and high risk of preterm delivery in women with threatened preterm labor? Fetal Diagn Ther 2011; 29: 216–223. Gomez R, Romero R, Nien JK, Chaiworapongsa T, Medina L, Kim YM, Yoon BH, Carstens M, Espinoza J, Iams JD, Gonzalez R. A short cervix in women with preterm labor and intact membranes: a risk factor for microbial invasion of the amniotic cavity. Am J Obstet Gynecol 2005; 192: 678–689. Sotiriadis A, Papatheodorou S, Kavvadias A, Makrydimas G. Transvaginal cervical length measurement for prediction of preterm birth in women with threatened preterm labor: a meta-analysis. Ultrasound Obstet Gynecol 2010; 35: 54–64. Soriano D, Weisz B, Seidman DS, Chetrit A, Schiff E, Lipitz S, Achiron R. The role of sonographic assessment of cervical length in the prediction of preterm birth in primigravidae with twin gestation conceived after infertility treatment. Acta Obstet Gynecol Scand 2002; 81: 39–43. Imseis HM, Albert TA, Iams JD. Identifying twin gestations at low risk for preterm birth with a transvaginal ultrasonographic cervical measurement at 24 to 26 weeks’ gestation. Am J Obstet Gynecol 1997; 177: 1149–1155. To MS, Fonseca EB, Molina FS, Cacho AM, Nicolaides KH. Maternal characteristics and cervical length in the prediction of spontaneous early preterm delivery in twins. Am J Obstet Gynecol 2006; 194: 1360–1365. Sperling L, Kiil C, Larsen LU, Qvist I, Bach D, Wøjdemann K, Bladh A, Nikkila¨ A, Jørgensen C, Skajaa K, Bang J, Tabor A. How to identify twins at low risk of spontaneous preterm delivery. Ultrasound Obstet Gynecol 2005; 26: 138–144. Klein K, Gregor H, Hirtenlehner-Ferber K, Stammler-Safar M, Witt A, Hanslik A, Husslein P, Krampl E. Prediction of spontaneous preterm delivery in twin pregnancies by cervical length at mid-gestation. Twin Res Hum Genet 2008; 11: 552–557. Fox NS, Rebarber A, Klauser CK, Peress D, Gutierrez CV, Saltzman DH. Prediction of spontaneous preterm birth in asymptomatic twin pregnancies using the change in cervical length over time. Am J Obstet Gynecol 2010; 202: 155.e1–4. Fox NS, Rebarber A, Roman AS, Klauser CK, Saltzman DH. Association between second-trimester cervical length and spontaneous preterm birth in twin pregnancies. J Ultrasound Med 2010; 29: 1733–1739. Yang JH, Kuhlman K, Daly S, Berghella V. Prediction of preterm birth by second trimester cervical sonography in twin pregnancies. Ultrasound Obstet Gynecol 2000; 15: 288–291. Guzman ER, Walters C, O’reilly-Green C, Kinzler WL, Waldron R, Nigam J, Vintzileos AM. Use of cervical ultrasonography in prediction of spontaneous preterm birth in twin gestations. Am J Obstet Gynecol 2000; 183: 1103–1107. Crane JM, Van den Hof M, Armson BA, Liston R. Transvaginal ultrasound in the prediction of preterm delivery: singleton and twin gestations. Obstet Gynecol 1997; 90: 357–363. Gonzalez N, Bige V, Kandoussi S, Graesslin O, Quereux C, Gabriel R. [Ultrasonographic measurement of cervical length in twin pregnancies with preterm labor: comparison with singleton pregnancies]. Gynecol Obstet Fertil 2004; 32: 122–127. Vendittelli F, Volumenie J. Transvaginal ultrasonography examination of the uterine cervix in hospitalised women undergoing preterm labour. Eur J Obstet Gynecol Reprod Biol 2000; 90: 3–11. Yoshizato T, Inoue Y, Fukami T, Sanui A, Miyamato S, Kawarabayashi T. Longitudinal changes in canal length at 16–35 weeks in normal twin pregnancies and twin pregnancies with preterm labor and delivery. J Obstet Gynaecol Res 2010; 36: 733–738. Fuchs I, Tsoi E, Henrich W, Dudenhausen JW, Nicolaides KH. Sonographic measurement of cervical length in twin pregnancies in threatened preterm labor. Ultrasound Obstet Gynecol 2004; 23: 42–45. Mella MT, Berghella V. Prediction of preterm birth: cervical sonography. Semin Perinatol 2009; 33: 317–324. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: 839–843. Liem SM, van de Mheen L, Bekedam DJ, van Pampus MG, Opmeer BC, Lim AC, Mol BW. Cervical length measurement for the prediction of preterm birth in symptomatic women with a twin pregnancy: a systematic review and meta-analysis. Obstet Gynecol Int 2013; 2013: 125897. Romero R, Yeo L, Chaemsaithong P, Chaiworapongsa T, Hassan SS. The preterm parturition syndrome. BJOG 2006; 113: 17–42. Gotsch F, Gotsch F, Romero R, Erez O, Vaisbuch E, Kusanovic JP, Mazaki-Tovi S, Kim SK, Hassan S, Yeo L. The preterm parturition syndrome and its implications for understanding the biology, risk assessment, diagnosis, treatment and prevention of preterm birth. J Matern Fetal Neonatal Med 2009; 22: 5–23. Gardner MO, Goldenberg RL, Cliver SP, Tucker JM, Nelson KG, Copper RL. The origin and outcome of preterm twin pregnancies. Obstet Gynecol 1995; 85: 553–557.
Ultrasound Obstet Gynecol 2015; 46: 73–81.
