Original Article
1151
Indicators for Cervical Length in Twin Pregnancies Frederik J. R. Hermans, MD1 Ewoud Schuit, PhD1,2,3 Sophie M. S. Liem, MD, PhD1 Arianne C. Lim, MD, PhD1 Johannes Duvekot, MD, PhD4 Liesbeth C. J. Scheepers, MD, PhD5 Mallory M. Woiski, MD, PhD6 Maureen M. Franssen, MD, PhD7 Martijn A. Oudijk, MD, PhD8 Kitty W. M. Bloemenkamp, MD, PhD9 Bas Nij Bijvanck, MD, PhD10 Dick J. Bekedam, MD, PhD11 Brent C. Opmeer, PhD12 Ben Willem J. Mol, MD, PhD13
Center, Amsterdam, The Netherlands 2 Julius Centre for Healthcare Research and Primary Care, University Medical Centre Utrecht, Utrecht, The Netherlands 3 Stanford Prevention Research Center, Stanford University, Stanford, California 4 Department of Obstetrics and Gynaecology, Erasmus Medical Centre, Rotterdam, The Netherlands 5 Department of Obstetrics and Gynaecology, Academic Hospital Maastricht, Maastricht, The Netherlands 6 Department of Obstetrics and Gynaecology, Radboud University Medical Centre, Nijmegen, The Netherlands 7 Department of Obstetrics and Gynaecology, University Medical Center Groningen, Groningen, The Netherlands 8 Department of Obstetrics and Gynaecology, University Medical Centre Utrecht, Utrecht, The Netherlands 9 Department of Obstetrics and Gynaecology, Leiden University Medical Centre, Leiden, The Netherlands 10 Department of Obstetrics and Gynaecology, Isala Clinics, Zwolle, The Netherlands 11 Department of Obstetrics and Gynaecology, Onze Lieve Vrouwe Gasthuis, Amsterdam, The Netherlands 12 Clinical Research Unit, Academic Medical Center, Amsterdam, The Netherlands 13 The Robinson Research Institute, School of Paediatrics and Reproductive Health, University of Adelaide, Australia
Address for correspondence Frederik Hermans, MD, Department of Obstetrics and Gynaecology, Academic Medical Center, Room H4-240, P.O. Box 22660 1100DD, Amsterdam, The Netherlands (e-mail: [email protected]).
Downloaded by: University of Pittsburgh. Copyrighted material.
1 Department of Obstetrics and Gynaecology, Academic Medical
Am J Perinatol 2015;32:1151–1157.
Abstract
Keywords
► ► ► ►
preterm birth cervical length indicators multiple pregnancies
received November 5, 2014 accepted after revision March 10, 2015 published online April 27, 2015
Objective Cervical length (CL) is associated with the risk of preterm birth (PTB) in multiple pregnancies. However, the position of CL within the pathophysiological pathway of PTB is unclear, and it is unknown which factors are predictive for CL. This study aims to investigate whether in twin pregnancies baseline maternal and obstetrical characteristics are potential indicators for CL, to improve insight in the pathophysiological pathway of PTB. Study Design Secondary analysis of data on twin pregnancies and CL measurement between 16 and 22 weeks. A set of 10 potential indicators, known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change
Copyright © 2015 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662.
DOI http://dx.doi.org/ 10.1055/s-0035-1549396. ISSN 0735-1631.
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
of CL were selected. We used multivariable linear regression with backward selection to identify independent indicators for CL. Results A total of 1,447 women with twin pregnancies were included. Mean CL was 43.7 ( 8.9) mm. In multivariable analysis, age (0.27 mm/y; 95% confidence interval [CI] 0.16 to 0.39), use of assisted reproductive technologies (ART) (1.42 mm, 95% CI 2.6 to 0.25), and having delivered at term in a previous pregnancy (1.32 mm, 95% CI 0.25 to 2.39) were significantly associated with CL. Conclusion This study shows that in twin pregnancies, age, use of ART and having delivered term in a previous pregnancy has an association with CL.
Preterm birth (PTB) is the main factor contributing to perinatal morbidity and mortality.1–3 Women with a multiple pregnancy are at an increased risk for spontaneous PTB before 37 weeks with or without preterm prelabor rupture of membranes compared with singleton pregnancies.1 In the Netherlands, 32% of multiple pregnancies, have a spontaneous PTB versus 4.3% of singleton pregnancies.1,4 The position of cervical length (CL) within the causal pathophysiological pathway toward PTB is unclear, and it is unknown which factors are in itself predictive for CL.5 Amongst many other risk indicators for PTB (e.g., ethnicity, social economic status, smoking, and previous PTB) CL measured at midtrimester ultrasound have been identified as a reliable predictor for spontaneous PTB in general, as well as for multiple pregnancies.6–9 In singleton pregnancies possible relationships between demographical and obstetrical characteristics, and CL, measured between 22 and 24 weeks, have been found.10–13 Since delivery starts with effacement and dilatation of the cervix, shortening of CL early in pregnancy is an important predictor of PTB. As such, in women with multiple pregnancies the cervix has been target of multiple intervention studies to prevent PTB (e.g., cerclage, pessary, or progesterone).14–17 Although none of the interventions were effective in the overall twin population, both pessary and progesterone had potential effects on both duration of pregnancy and adverse neonatal outcome for women with a short cervix. In view of this, it is important to understand the mechanism toward PTB and the associations between maternal and obstetrical characteristics and CL in twins. Therefore, our objective was to investigate whether women with twin pregnancies baseline maternal and obstetrical characteristics, known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change of CL, are potential indicators for midtrimester CL, to improve insight on the pathophysiological mechanism of PTB.
Methods Study Population We performed secondary analyses on data sets from two large multicenter randomized controlled trials evaluating interventions to prevent PTB in multiple gestations, the AMPHIAtrial and ProTWIN-trial,16,17 both conducted in the Netherlands within the Dutch Obstetric Research Consortium (available at: www.studies-obsgyn.nl). American Journal of Perinatology
Vol. 32
No. 12/2015
In short, in the AMPHIA-trial women were randomized for weekly intramuscular injections with 17-α hydroxyprogesterone caproate (17-OHPC) or placebo. In the ProTWIN-trial women were randomized between the treatment with a pessary or no treatment. In both trials women were recruited during the early midtrimester period and were offered transvaginal CL measurement during routine midtrimester ultrasound between 16 and 22 weeks of gestational age (GA). In the AMPHIA-trial CL measurement was not standardized, while in the ProTWIN-trial CL was measured in a sagittal view with a 5 MHz transducer in the anterior fornix. CL was the distance between the triangular area at the external ostium and the Vshape at the internal ostium. The Medical Ethics Committee of the Academic Medical Center had approved both trials (05/102 and 09/107). Detailed information on the design and the results of the studies can be found elsewhere.16,17
Outcome Midtrimester CL, the outcome of interest in this study, was measured through transvaginal ultrasound during the routine midtrimester screening between 16 and 22 weeks. CL is in clinical practice often dichotomized to categorize women as being at an increased risk of PTB or not. However, dichotomization of a continuous variable leads to the loss of information, so such practice is generally discouraged.18 Therefore, we choose to analyze CL in its original (continuous) form.
Candidate Indicators for Cervical Length Candidate indicators for CL included maternal and obstetrical variables measured at study entry, which are known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change of CL (e.g., curettage or in vitro fertilization [IVF], a procedure in which cervix is manipulated).6,7 Maternal characteristics included maternal age, body mass index (BMI), Caucasian ethnicity, current smoking status, and education. BMI was defined as maternal weight in kilograms divided by maternal height in meters squared, and education was categorized as basic education (reference category), vocational education, and higher education. Obstetrical characteristics included the use of assisted reproductive technologies (ART, composite for intrauterine insemination, IVF or intracytoplasmic sperm injection [ICSI]) to conceive, a combination of parity and previous PTB, monochorionic pregnancy, composition of fetal sex, and termination of pregnancy (composite for history of
Downloaded by: University of Pittsburgh. Copyrighted material.
1152
Indicators for Cervical Length in Twin Pregnancies
Statistical Analyses Baseline characteristics were analyzed using descriptive statistics. Continuous variables were presented as mean with standard deviation (SD) or median and interquartile range (IQR), as appropriate. Categorical and dichotomous variables were presented as numbers and proportions of the total study sample (%). With conventional univariable linear regression models associations between candidate indicators and CL as continuous outcome were assessed. Restricted cubic spline analyses were used to assess linearity of the association of continuous indicators with the outcome.19 In case of nonlinearity indicators were either transformed using a log-transformation or collapsed into categorical variables. Since selection based on univariable statistics might result in unstable models, we choose not to perform any preselection and to include all candidate indicators in the multivariable analyses.19 In the subsequent multivariable linear regression analyses, independent indicators for CL were identified by backward stepwise selection using Akaike Information Criterion (AIC).20 A complete case analysis would yield biased results, especially when missing values are missing at random, because these values are missing selectively.21 This also applied to our study (►Appendix 1). Missing variables were assumed as missing at random and were imputed using multiple imputation techniques (n ¼ 10). The imputation model included maternal characteristics (BMI, ethnicity, education), obstetrical history (parity, previous PTB), and characteristics related to the current pregnancy (use of ART, chorionicity, fetal sex, preeclampsia, infection, corticosteroids, GA at delivery, and weight of the children), as well as the outcome CL.22 Since analyses of multiple imputed datasets may yield different results, indicator selection was performed separately in each imputation set. For inclusion in the final multivariable model, we used the majority method; that is, predictors were included if selected in at least 5 out of 10 imputed datasets.23 Pooled effect estimates and standard errors of these final indicators from the 10 data sets were obtained using Rubin rules to come to the final multivariable linear model.24 Statistical analyses were performed using R 3.0.1 (R Foundation for Statistical Computing; Vienna, Austria, available at: http://www.r-project.org).
Results Study Population The two trials included a total of 1,484 women carrying a multiple pregnancy. Overall, 35 women with a triplet pregnancy and 1 with a quadruplet pregnancy were excluded from analyses. One more woman was excluded because no
1153
baseline information was available because she was lost to follow-up. As a result, 1,447 women with a twin pregnancy were available for further analyses.
Missing Value Analysis A total of 85% (n ¼ 1,222) of all women had missing values for one or more of the indicators of interest or for the outcome. CL was missing in 265 out of 1,447 women (18.3%). The proportion of missing values ranged from 4 out of 1,447 (0.3%) for chorionicity to 490 out of 1,447 (34.0%) for education.
Baseline Characteristics Baseline characteristics are presented in ►Table 1. Women had a mean age of 32.9 years (SD 4.5 years), median BMI of Table 1 Characteristics of the study population Characteristics
Overall (N ¼ 1,447) Mean (SD) Median (IQR) n (%)
Maternal age (y)
32.9 ( 4.5) 2
Body mass index (kg/m )
23.4 (21.2–26.3)
Caucasian
1,299 (90.0%)
Smoking
178 (12%)
Education Basic
58 (4%)
Vocational
581 (40%)
Higher
808 (56%)
Parity Nulliparous
779 (54%)
Multiparous no PTB
570 (39%)
Multiparous with PTB Termination of pregnancy
98 (7%) a
474 (33%)
Assisted reproductive technology
533 (37.0%)
Monochorionic twin
294 (20.3%)
Composition of fetal sex Female–female
455 (31%)
Mixed
514 (36%)
Male–male
478 (33%)
Cervical length (mm)
43.65 ( 8.92)
Gestational age (weeks)
36.7 (34.7 to 37.7)
43.0 (38.0–49.0) Treatment allocation AMHPIA placebo
326 (22%)
AMPHIA progesterone
327 (23%)
ProTWIN usual care
400 (28%)
ProTWIN pessary
394 (27%)
Abbreviations: CL, cervical length; PTB, preterm birth. a Medical history of pregnancy termination, for example, induced abortion, curettage. American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
spontaneous or induced abortion or curettage). Parity and a history of PTB < 37 weeks were combined as women being multiparous with a previous term delivery, with a previous PTB < 37 weeks, or being nulliparous (reference). Fetal sex for twin pairs was categorized as two females (reference), two males, or one male and one female.
Hermans et al.
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
23.4 kg/m2 (IQR, 21.2–26.3) and mean CL of 43.7 mm (SD 8.9 mm) (IQR, 38.0–49.0 mm). There were 779 (54%) nulliparous women. Among the 570 (39%) multiparous women, 98 (7%) had a previous PTB < 37 weeks. Overall, 533 (37%) women conceived after the use of ART.
Associations between Indicators and Cervical Length In a univariable linear regression analysis; increasing maternal age (0.30 mm/y; 95% CI 0.19–0.41) and higher education (2.7 mm; 95% CI 0.0–5.5, as compared with basic education) were associated with an increase of CL (►Table 2 [univariable]). Women who smoked or stopped smoking during pregnancy had on average a shorter CL than nonsmokers (1.7 mm; 95% CI 3.4 to 0.00). Caucasian women had on average a longer cervix than non-Caucasian women (1.4 mm; 95% CI 0.3 to 3.2), but this association was not statistically significant (p ¼ 0.06). Multiparous women with a previous term delivery had on average a longer cervix compared with nulliparous women (2.1 mm; 95% CI 1.08–3.09). In women who conceived after ART CL appeared to be shorter on average compared with women who conceived spontaneously (1.0 mm; 95% CI 2.1 to 0.10, p ¼ 0.075). Multivariable linear regression resulted in a model containing three maternal indicators and three obstetrical indicators (►Table 2 [multivariable]). Identified indicators were
maternal age (0.27 mm/y; 95% CI 0.16 to 0.39), smoking (1.0 mm; 95% CI 2.8 to 0.74), and vocational (2.1 mm; 95% CI 1.1 to 5.3) and higher education (2.6 mm; 95% CI 0.3 to 5.4) as compared with basic education, ART (1.4 mm; 95% CI 2.6 to 0.25), monochorionicity (0.9 mm; 95% CI 2.3 to 0.45), being multiparous with a previous term delivery (1.3 mm; 95% CI 0.25 to 1.3), and being multiparous with previous PTB (0.3 mm; 95% CI 1.7 to 0.74) as compared with nulliparous women.
Discussion Main Findings This study found that an increase in maternal age and having a history of term delivery are positively associated with CL. Women who completed vocational or higher education seemed to have an increasingly longer cervix as well, however this association did not reach statistical significance. The use of ART (e.g., intrauterine insemination [IUI], IVF, or ICSI) was associated with a decrease in CL.
Strength and Limitations This study has several strengths. First, we used data of two randomized clinical trials that included a total of 1,447 asymptomatic twin pregnancies. By inclusion of
Table 2 Associations between risk indicators and cervical length as a continuous outcome Risk indicator
Univariable
Multivariable
Mean difference mm/unit (95% CI) 0.30 (0.19 to 0.41)
0.27 (0.16 to 0.39)a
BMI (kg/m )
0.03 (0.07 to 0.14)
–
Caucasian
1.4 (0.3 to 3.2)
–
0.99 (2.1 to 0.1)
1.4 (2.6 to 0.3)a
0.76 (2.1 to 0.57)
0.90 (2.3 to 0.45)
1.7 (3.4 to 0)
1.0 (2.8 to 0.74)
0.11 (1.2 to 0.97)
–
Female–female
Ref.
–
Mixed
0.25 (1.5 to 1)
–
Male–male
0.23 (1.5 to 1.1)
–
Nulliparous
Ref.
Ref.
Multiparous with term delivery
2.1 (1.1 to 3.1)
1.3 (0.25 to 2.4)a
Multiparous with preterm delivery
0.89 (1.1 to 2.9)
0.29 (1.7 to 2.3)
Ref.
Ref.
Age (y) 2
ART
b
Monochorionic Smoking Termination pregnancy
c
Composition of fetal sex
Parity
Education Basic Vocational
1.8 (1.3 to 5.0)
2.1 (1.1 to 5.3)
Higher
2.7 (0.05 to 5.5)
2.6 (0.27 to 5.4)
Abbreviations: ART, assisted reproductive technologies; BMI, body mass index; CL, cervical length; PTB, preterm birth. a p < 0.05. b Mode of conception. c Medical history of pregnancy termination, for example, induced abortion, curettage. American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
1154
asymptomatic twins, we believe these women are a representative sample of the Dutch women with a twin pregnancy. Additionally, the study population is in concordance with general characteristics of the pregnant Dutch population (e.g., for age, parity, and smoking).1,4 Second, the multivariable models include indicators that were collected around study entry that are easily available, and are not modifiable thus minimizing the risk of information bias. As such, the results found are reliable representations of the relationship between these indicators and the outcome. A potential limitation of the study could be the quality of the CL measurement. CL was not part of the inclusion criteria of the primary studies, and was therefore not measured according to a predetermined protocol in the AMPHIA-trial and was only measured once, which could have introduced possible inaccurate measurements.19,20 Consequently, CL data were missing in 18.3% of the included women. However, these missing values were imputed using multiple imputations, which has been shown to be a more valid method to deal with missing values than complete case analysis.25 A second limitation is that the outcome of our study—CL— is a soft endpoint, that is, a risk factor known to be associated with PTB, rather than a hard endpoint such as PTB itself. Consequently, the current study should be viewed as a first step in a broader range of studies looking into the pathway of PTB, that is, with CL as an intermediate factor between risk factors for CL and PTB. Therefore, current results can be used for generating hypotheses.
Interpretation Studies investigating second trimester CL in women with a singleton pregnancy, found that women who were AfroAmerican, had a low BMI < 18 kg/m2 or who had recurrent or spontaneous PTB had an increased risk for a short cervix or shorter CL.10–12 Similar studies in women with a twin pregnancy are currently not published. In our study sample, we found a similar effect of ethnicity in univariable analysis. However, since in our population, 90% of women are Caucasian, while the remaining 10% is a mixture of ethnicities, we were unable to reliably investigate this association. No relationship was found between (low) BMI and CL in this study, mainly because the prevalence of underweight women was low in our population (median BMI 23.4 kg/m2 [IQR 21.2– 26.3]). Other studies investigating the relationship between obstetrical history and CL found that CL in singleton pregnancies, women with isolated PTB and recurrent PTB is shorter compared with women with term births.11,26 This is supported by our results, which demonstrated that multiparous women with a history of term delivery had a longer CL compared with nulliparous women, while women with a history of preterm delivery had no longer CL compared with nulliparous women. This study found no decrease in CL in women with a previous PTB. This could be explained by the fact that PTB < 34 weeks was an exclusion criterion for one of the trials (AMPHIA), leading to a selection of women with a history of less severe PTBs, while the strongest effect of PTB on CL is expected in women with recurrent PTB < 34 weeks.
Hermans et al.
It is shown that there is a possible association between the use of ART and CL, with an average decrease in CL of 1.4 mm in case of ART. In IVF (or ICSI) and IUI, embryos or insemination is performed through the cervix. We hypothesize that this manipulation can cause trauma to the cervix that can lead to scarring or cervical insufficiency. This theory is supported by evidence that other procedures (e.g., induced abortion or curettage) in which the cervix is manipulated (cervical dilatation and evacuation) are associated with an increased risk for PTB.27 Additionally, another study found that surgical management of pregnancy loss was associated with an increased risk of PTB.28 However, the authors could not conclude whether this association was a causal. Higher education seems to have a positive association with CL in linear regression (2.6 mm). This positive association could be explained by considering the educational level as a surrogate marker for social economic status. Basic education was defined as women who had no higher completed education than primary school, which is often associated with a lower social economic class. In these women there could be a higher risk of unhealthy circumstances or behavior (e.g., infection of the genital tract) leading to a decrease in CL.29 A twin pregnancy in general is considered as an important risk factor for PTB and it has also been shown that the prevalence of short CL is higher in twin pregnancies as compared with singleton pregnancies.6,8,9 So it is likely to assume that having a twin pregnancy substantially shortens CL mediating the risk of PTB. However, the actual effect of type of gestation (singleton vs. twin) on CL is unknown. Second associations between other indicators and CL can be less pronounced in twin pregnancies than in singleton pregnancies because of the strong effect of a twin pregnancy on the risk of PTB. Therefore, ideally, these associations should be investigated in a cohort study in which type of gestation is incorporated as an indicator.
Conclusion This study demonstrated that several maternal and obstetric indicators known to be associated with PTB are also associated with CL. Therefore, investigation of possible pathways of cervical shortening should be encouraged to gain more insight in the pathophysiological pathway of PTB.
Conflict of Interest None.
Funding None.
Authors’ Contributions Conceived and designed the experiments: F. J. R. H., B. W. J. M., E. S., B. C. O. Analyzed the data: E. S., F. J. R. H. Wrote the first draft of the article: F. J. R. H. Contributed to the writing of the article: F. J. R. H., E. S., B. C. O., B. W. J. M. Contributed American Journal of Perinatology
Vol. 32
No. 12/2015
1155
Downloaded by: University of Pittsburgh. Copyrighted material.
Indicators for Cervical Length in Twin Pregnancies
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
and acquired data for the original trials, corrected article: S. M. S. L., A. C. L. HD, L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B. ICMJE criteria for authorship read and met: F. J. R. H., E. S., B. W. J. M., B. C. O., S. M. S. L., A. C. L , L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B. Agree with article results and conclusions: F. J. R. H, E. S., B. W. J. M., B. C. O., S. M. S. L., A. C. L HD, L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B.
14 Schuit E, Stock S, Rode L, et al; Global Obstetrics Network
15
16
17
References 1 Schaaf JM, Mol BWJ, Abu-Hanna A, Ravelli ACJ. Trends in preterm
2
3 4
5 6 7 8
9
10
11
12
13
birth: singleton and multiple pregnancies in the Netherlands, 2000-2007. BJOG 2011;118(10):1196–1204 Botting BJ, Davies IM, Macfarlane AJ. Recent trends in the incidence of multiple births and associated mortality. Arch Dis Child 1987; 62(9):941–950 McCormick MC. The contribution of low birth weight to infant mortality and childhood morbidity. N Engl J Med 1985;312(2):82–90 EURO-PERISTAT ProjectEuropean Perinatal Health Report. The Health and Care of Pregnant Women and Babies in Europe in 2010. Available at: http://www.europeristat.com. Accessed April 10, 2014 Romero R, Espinoza J, Kusanovic JP, et al. The preterm parturition syndrome. BJOG 2006;113(Suppl 3):17–42 Berkowitz GS, Papiernik E. Epidemiology of preterm birth. Epidemiol Rev 1993;15(2):414–443 Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008;371(9606):75–84 Goldenberg RL, Iams JD, Miodovnik M, et al; National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. The preterm prediction study: risk factors in twin gestations. Am J Obstet Gynecol 1996;175(4 Pt 1):1047–1053 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(2):128.e1–128.e12 Palma-Dias RS, Fonseca MM, Stein NR, Schmidt AP, Magalhães JA. Relation of cervical length at 22-24 weeks of gestation to demographic characteristics and obstetric history. Braz J Med Biol Res 2004;37(5):737–744 Mercer BM, Macpherson CA, Goldenberg RL, et al; NICHD MaternalFetal Medicine Units Network. Are women with recurrent spontaneous preterm births different from those without such history? Am J Obstet Gynecol 2006;194(4):1176–1184, discussion 1184–1185 Heath VC, Southall TR, Souka AP, Novakov A, Nicolaides KH. Cervical length at 23 weeks of gestation: relation to demographic characteristics and previous obstetric history. Ultrasound Obstet Gynecol 1998;12(5):304–311 Erasmus I, Nicolaou E, van Gelderen CJ, Nicolaides KH. Cervical length at 23 weeks’ gestation—relation to demographic characteristics and previous obstetric history in South African women. S Afr Med J 2005;95(9):691–695
American Journal of Perinatology
Vol. 32
No. 12/2015
18 19
20 21
22
23
24
25
26
27
28
29
(GONet) collaboration. Effectiveness of progestogens to improve perinatal outcome in twin pregnancies: an individual participant data meta-analysis. BJOG 2015;122(1): 27–37 Rust O, Atlas R, Wells M, Rawlinson K. Cerclage in multiple gestation with midtrimester dilation of the internal os. Am J Obstet Gynecol 2001;185(6):S111 Lim AC, Schuit E, Bloemenkamp K, et al. 17α-hydroxyprogesterone caproate for the prevention of adverse neonatal outcome in multiple pregnancies: a randomized controlled trial. Obstet Gynecol 2011;118(3):513–520 Liem S, Schuit E, Hegeman M, et al. Cervical pessaries for prevention of preterm birth in women with a multiple pregnancy (ProTWIN): a multicentre, open-label randomised controlled trial. Lancet 2013;382(9901):1341–1349 Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006;332(7549):1080 Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15(4): 361–387 Atkinson AC. A note on the generalized information criterion for choice of a model. Biometrika 1980;67(2):413–418 Donders ART, van der Heijden GJMG, Stijnen T, Moons KGM. Review: a gentle introduction to imputation of missing values. J Clin Epidemiol 2006;59(10):1087–1091 Moons KGM, Donders RART, Stijnen T, Harrell FE Jr. Using the outcome for imputation of missing predictor values was preferred. J Clin Epidemiol 2006;59(10):1092–1101 Vergouwe Y, Royston P, Moons KGM, Altman DG. Development and validation of a prediction model with missing predictor data: a practical approach. J Clin Epidemiol 2010;63(2): 205–214 Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and some applications. Stat Med 1991;10(4): 585–598 van der Heijden GJMG, Donders ART, Stijnen T, Moons KGM. Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: a clinical example. J Clin Epidemiol 2006;59(10): 1102–1109 de Carvalho MHB, Bittar RE, Brizot Mde L, Bicudo C, Zugaib M. Prediction of preterm delivery in the second trimester. Obstet Gynecol 2005;105(3):532–536 Sedgh G, Singh S, Shah IH, Ahman E, Henshaw SK, Bankole A. Induced abortion: incidence and trends worldwide from 1995 to 2008. Lancet 2012;379(9816):625–632 McCarthy FP, Khashan AS, North RA, et al; SCOPE Consortium. Pregnancy loss managed by cervical dilatation and curettage increases the risk of spontaneous preterm birth. Hum Reprod 2013;28(12):3197–3206 Matser A, Luu N, Geskus R, et al. Higher Chlamydia trachomatis prevalence in ethnic minorities does not always reflect higher sexual risk behaviour. PLoS ONE 2013;8(6):e67287
Downloaded by: University of Pittsburgh. Copyrighted material.
1156
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
1157
Appendix 1 Missing value analysis p-Valuea
Missing (N ¼ 1,226)
Not missing (N ¼ 221)
Mean (SD) Median (IQR) n (%)
Mean (SD) Median (IQR) n (%)
Maternal age (y)
32.9 ( 4.5)
32.7 ( 4.6)
0.66
Height
170.3 ( 6.8)
170.3 ( 6.6)
0.90
Weight
68.0 (61.0–77.0)
70.0 (63.0–78.0)
0.11
Caucasian
1,044 (91%)
201 (91%)
0.65
Smoking
149 (13%)
25 (11%)
0.68
Basic
33 (4%)
4 (2%)
0.06
Vocational
284 (39%)
100 (45%)
Higher
419 (57%)
117 (53%)
Nulliparous
657 (54%)
122 (55%)
0.71
Previous PTB
81 (7%)
16 (7%)
0.85
Spontaneous abortion
339 (38%)
65 (29%)
0.73
Induced abortion
73 (6%)
13 (6%)
0.95
Curettages
79 (7%)
13 (6%)
0.82
Assisted reproductive technology
437 (36%)
92 (42%)
0.11
Monochorionic twin
257 (21%)
36 (16%)
0.12
GBS carrier
20 (7%)
19 (9%)
0.69
Preeclampsia
166 (14%)
30 (14%)
0.92 0.38
Fetal sex Female–female
390 (32%)
61 (28%)
Mixed
429 (35%)
81 (37%)
Male–male
395 (33%)
79 (36%)
43.9 ( 8.9)
42.4 ( 9.1)
0.03
Spontaneous
462 (38%)
96 (43%)
0.21
Iatrogenic
488 (41%)
76 (34%)
Cervical length Onset of labor
253 (21%)
49 (22%)
Gestational age
C-Section
36.4 (34.1–37.7)
36.9 (24.7–37.7)
0.21
Study participation (AMPHIA)
537 (44%)
116 (52%)
0.95
Randomization result (intervention)
610 (50%)
111 (50%)
0.46
Abbreviations: CL, cervical length; PTB, preterm birth. a Tested accordingly: t-test, chi-square, Mann–Whitney U test.
American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
Education
1151
Indicators for Cervical Length in Twin Pregnancies Frederik J. R. Hermans, MD1 Ewoud Schuit, PhD1,2,3 Sophie M. S. Liem, MD, PhD1 Arianne C. Lim, MD, PhD1 Johannes Duvekot, MD, PhD4 Liesbeth C. J. Scheepers, MD, PhD5 Mallory M. Woiski, MD, PhD6 Maureen M. Franssen, MD, PhD7 Martijn A. Oudijk, MD, PhD8 Kitty W. M. Bloemenkamp, MD, PhD9 Bas Nij Bijvanck, MD, PhD10 Dick J. Bekedam, MD, PhD11 Brent C. Opmeer, PhD12 Ben Willem J. Mol, MD, PhD13
Center, Amsterdam, The Netherlands 2 Julius Centre for Healthcare Research and Primary Care, University Medical Centre Utrecht, Utrecht, The Netherlands 3 Stanford Prevention Research Center, Stanford University, Stanford, California 4 Department of Obstetrics and Gynaecology, Erasmus Medical Centre, Rotterdam, The Netherlands 5 Department of Obstetrics and Gynaecology, Academic Hospital Maastricht, Maastricht, The Netherlands 6 Department of Obstetrics and Gynaecology, Radboud University Medical Centre, Nijmegen, The Netherlands 7 Department of Obstetrics and Gynaecology, University Medical Center Groningen, Groningen, The Netherlands 8 Department of Obstetrics and Gynaecology, University Medical Centre Utrecht, Utrecht, The Netherlands 9 Department of Obstetrics and Gynaecology, Leiden University Medical Centre, Leiden, The Netherlands 10 Department of Obstetrics and Gynaecology, Isala Clinics, Zwolle, The Netherlands 11 Department of Obstetrics and Gynaecology, Onze Lieve Vrouwe Gasthuis, Amsterdam, The Netherlands 12 Clinical Research Unit, Academic Medical Center, Amsterdam, The Netherlands 13 The Robinson Research Institute, School of Paediatrics and Reproductive Health, University of Adelaide, Australia
Address for correspondence Frederik Hermans, MD, Department of Obstetrics and Gynaecology, Academic Medical Center, Room H4-240, P.O. Box 22660 1100DD, Amsterdam, The Netherlands (e-mail: [email protected]).
Downloaded by: University of Pittsburgh. Copyrighted material.
1 Department of Obstetrics and Gynaecology, Academic Medical
Am J Perinatol 2015;32:1151–1157.
Abstract
Keywords
► ► ► ►
preterm birth cervical length indicators multiple pregnancies
received November 5, 2014 accepted after revision March 10, 2015 published online April 27, 2015
Objective Cervical length (CL) is associated with the risk of preterm birth (PTB) in multiple pregnancies. However, the position of CL within the pathophysiological pathway of PTB is unclear, and it is unknown which factors are predictive for CL. This study aims to investigate whether in twin pregnancies baseline maternal and obstetrical characteristics are potential indicators for CL, to improve insight in the pathophysiological pathway of PTB. Study Design Secondary analysis of data on twin pregnancies and CL measurement between 16 and 22 weeks. A set of 10 potential indicators, known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change
Copyright © 2015 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662.
DOI http://dx.doi.org/ 10.1055/s-0035-1549396. ISSN 0735-1631.
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
of CL were selected. We used multivariable linear regression with backward selection to identify independent indicators for CL. Results A total of 1,447 women with twin pregnancies were included. Mean CL was 43.7 ( 8.9) mm. In multivariable analysis, age (0.27 mm/y; 95% confidence interval [CI] 0.16 to 0.39), use of assisted reproductive technologies (ART) (1.42 mm, 95% CI 2.6 to 0.25), and having delivered at term in a previous pregnancy (1.32 mm, 95% CI 0.25 to 2.39) were significantly associated with CL. Conclusion This study shows that in twin pregnancies, age, use of ART and having delivered term in a previous pregnancy has an association with CL.
Preterm birth (PTB) is the main factor contributing to perinatal morbidity and mortality.1–3 Women with a multiple pregnancy are at an increased risk for spontaneous PTB before 37 weeks with or without preterm prelabor rupture of membranes compared with singleton pregnancies.1 In the Netherlands, 32% of multiple pregnancies, have a spontaneous PTB versus 4.3% of singleton pregnancies.1,4 The position of cervical length (CL) within the causal pathophysiological pathway toward PTB is unclear, and it is unknown which factors are in itself predictive for CL.5 Amongst many other risk indicators for PTB (e.g., ethnicity, social economic status, smoking, and previous PTB) CL measured at midtrimester ultrasound have been identified as a reliable predictor for spontaneous PTB in general, as well as for multiple pregnancies.6–9 In singleton pregnancies possible relationships between demographical and obstetrical characteristics, and CL, measured between 22 and 24 weeks, have been found.10–13 Since delivery starts with effacement and dilatation of the cervix, shortening of CL early in pregnancy is an important predictor of PTB. As such, in women with multiple pregnancies the cervix has been target of multiple intervention studies to prevent PTB (e.g., cerclage, pessary, or progesterone).14–17 Although none of the interventions were effective in the overall twin population, both pessary and progesterone had potential effects on both duration of pregnancy and adverse neonatal outcome for women with a short cervix. In view of this, it is important to understand the mechanism toward PTB and the associations between maternal and obstetrical characteristics and CL in twins. Therefore, our objective was to investigate whether women with twin pregnancies baseline maternal and obstetrical characteristics, known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change of CL, are potential indicators for midtrimester CL, to improve insight on the pathophysiological mechanism of PTB.
Methods Study Population We performed secondary analyses on data sets from two large multicenter randomized controlled trials evaluating interventions to prevent PTB in multiple gestations, the AMPHIAtrial and ProTWIN-trial,16,17 both conducted in the Netherlands within the Dutch Obstetric Research Consortium (available at: www.studies-obsgyn.nl). American Journal of Perinatology
Vol. 32
No. 12/2015
In short, in the AMPHIA-trial women were randomized for weekly intramuscular injections with 17-α hydroxyprogesterone caproate (17-OHPC) or placebo. In the ProTWIN-trial women were randomized between the treatment with a pessary or no treatment. In both trials women were recruited during the early midtrimester period and were offered transvaginal CL measurement during routine midtrimester ultrasound between 16 and 22 weeks of gestational age (GA). In the AMPHIA-trial CL measurement was not standardized, while in the ProTWIN-trial CL was measured in a sagittal view with a 5 MHz transducer in the anterior fornix. CL was the distance between the triangular area at the external ostium and the Vshape at the internal ostium. The Medical Ethics Committee of the Academic Medical Center had approved both trials (05/102 and 09/107). Detailed information on the design and the results of the studies can be found elsewhere.16,17
Outcome Midtrimester CL, the outcome of interest in this study, was measured through transvaginal ultrasound during the routine midtrimester screening between 16 and 22 weeks. CL is in clinical practice often dichotomized to categorize women as being at an increased risk of PTB or not. However, dichotomization of a continuous variable leads to the loss of information, so such practice is generally discouraged.18 Therefore, we choose to analyze CL in its original (continuous) form.
Candidate Indicators for Cervical Length Candidate indicators for CL included maternal and obstetrical variables measured at study entry, which are known to be associated with an increased risk of PTB and/or which have a plausible mechanism resulting in a change of CL (e.g., curettage or in vitro fertilization [IVF], a procedure in which cervix is manipulated).6,7 Maternal characteristics included maternal age, body mass index (BMI), Caucasian ethnicity, current smoking status, and education. BMI was defined as maternal weight in kilograms divided by maternal height in meters squared, and education was categorized as basic education (reference category), vocational education, and higher education. Obstetrical characteristics included the use of assisted reproductive technologies (ART, composite for intrauterine insemination, IVF or intracytoplasmic sperm injection [ICSI]) to conceive, a combination of parity and previous PTB, monochorionic pregnancy, composition of fetal sex, and termination of pregnancy (composite for history of
Downloaded by: University of Pittsburgh. Copyrighted material.
1152
Indicators for Cervical Length in Twin Pregnancies
Statistical Analyses Baseline characteristics were analyzed using descriptive statistics. Continuous variables were presented as mean with standard deviation (SD) or median and interquartile range (IQR), as appropriate. Categorical and dichotomous variables were presented as numbers and proportions of the total study sample (%). With conventional univariable linear regression models associations between candidate indicators and CL as continuous outcome were assessed. Restricted cubic spline analyses were used to assess linearity of the association of continuous indicators with the outcome.19 In case of nonlinearity indicators were either transformed using a log-transformation or collapsed into categorical variables. Since selection based on univariable statistics might result in unstable models, we choose not to perform any preselection and to include all candidate indicators in the multivariable analyses.19 In the subsequent multivariable linear regression analyses, independent indicators for CL were identified by backward stepwise selection using Akaike Information Criterion (AIC).20 A complete case analysis would yield biased results, especially when missing values are missing at random, because these values are missing selectively.21 This also applied to our study (►Appendix 1). Missing variables were assumed as missing at random and were imputed using multiple imputation techniques (n ¼ 10). The imputation model included maternal characteristics (BMI, ethnicity, education), obstetrical history (parity, previous PTB), and characteristics related to the current pregnancy (use of ART, chorionicity, fetal sex, preeclampsia, infection, corticosteroids, GA at delivery, and weight of the children), as well as the outcome CL.22 Since analyses of multiple imputed datasets may yield different results, indicator selection was performed separately in each imputation set. For inclusion in the final multivariable model, we used the majority method; that is, predictors were included if selected in at least 5 out of 10 imputed datasets.23 Pooled effect estimates and standard errors of these final indicators from the 10 data sets were obtained using Rubin rules to come to the final multivariable linear model.24 Statistical analyses were performed using R 3.0.1 (R Foundation for Statistical Computing; Vienna, Austria, available at: http://www.r-project.org).
Results Study Population The two trials included a total of 1,484 women carrying a multiple pregnancy. Overall, 35 women with a triplet pregnancy and 1 with a quadruplet pregnancy were excluded from analyses. One more woman was excluded because no
1153
baseline information was available because she was lost to follow-up. As a result, 1,447 women with a twin pregnancy were available for further analyses.
Missing Value Analysis A total of 85% (n ¼ 1,222) of all women had missing values for one or more of the indicators of interest or for the outcome. CL was missing in 265 out of 1,447 women (18.3%). The proportion of missing values ranged from 4 out of 1,447 (0.3%) for chorionicity to 490 out of 1,447 (34.0%) for education.
Baseline Characteristics Baseline characteristics are presented in ►Table 1. Women had a mean age of 32.9 years (SD 4.5 years), median BMI of Table 1 Characteristics of the study population Characteristics
Overall (N ¼ 1,447) Mean (SD) Median (IQR) n (%)
Maternal age (y)
32.9 ( 4.5) 2
Body mass index (kg/m )
23.4 (21.2–26.3)
Caucasian
1,299 (90.0%)
Smoking
178 (12%)
Education Basic
58 (4%)
Vocational
581 (40%)
Higher
808 (56%)
Parity Nulliparous
779 (54%)
Multiparous no PTB
570 (39%)
Multiparous with PTB Termination of pregnancy
98 (7%) a
474 (33%)
Assisted reproductive technology
533 (37.0%)
Monochorionic twin
294 (20.3%)
Composition of fetal sex Female–female
455 (31%)
Mixed
514 (36%)
Male–male
478 (33%)
Cervical length (mm)
43.65 ( 8.92)
Gestational age (weeks)
36.7 (34.7 to 37.7)
43.0 (38.0–49.0) Treatment allocation AMHPIA placebo
326 (22%)
AMPHIA progesterone
327 (23%)
ProTWIN usual care
400 (28%)
ProTWIN pessary
394 (27%)
Abbreviations: CL, cervical length; PTB, preterm birth. a Medical history of pregnancy termination, for example, induced abortion, curettage. American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
spontaneous or induced abortion or curettage). Parity and a history of PTB < 37 weeks were combined as women being multiparous with a previous term delivery, with a previous PTB < 37 weeks, or being nulliparous (reference). Fetal sex for twin pairs was categorized as two females (reference), two males, or one male and one female.
Hermans et al.
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
23.4 kg/m2 (IQR, 21.2–26.3) and mean CL of 43.7 mm (SD 8.9 mm) (IQR, 38.0–49.0 mm). There were 779 (54%) nulliparous women. Among the 570 (39%) multiparous women, 98 (7%) had a previous PTB < 37 weeks. Overall, 533 (37%) women conceived after the use of ART.
Associations between Indicators and Cervical Length In a univariable linear regression analysis; increasing maternal age (0.30 mm/y; 95% CI 0.19–0.41) and higher education (2.7 mm; 95% CI 0.0–5.5, as compared with basic education) were associated with an increase of CL (►Table 2 [univariable]). Women who smoked or stopped smoking during pregnancy had on average a shorter CL than nonsmokers (1.7 mm; 95% CI 3.4 to 0.00). Caucasian women had on average a longer cervix than non-Caucasian women (1.4 mm; 95% CI 0.3 to 3.2), but this association was not statistically significant (p ¼ 0.06). Multiparous women with a previous term delivery had on average a longer cervix compared with nulliparous women (2.1 mm; 95% CI 1.08–3.09). In women who conceived after ART CL appeared to be shorter on average compared with women who conceived spontaneously (1.0 mm; 95% CI 2.1 to 0.10, p ¼ 0.075). Multivariable linear regression resulted in a model containing three maternal indicators and three obstetrical indicators (►Table 2 [multivariable]). Identified indicators were
maternal age (0.27 mm/y; 95% CI 0.16 to 0.39), smoking (1.0 mm; 95% CI 2.8 to 0.74), and vocational (2.1 mm; 95% CI 1.1 to 5.3) and higher education (2.6 mm; 95% CI 0.3 to 5.4) as compared with basic education, ART (1.4 mm; 95% CI 2.6 to 0.25), monochorionicity (0.9 mm; 95% CI 2.3 to 0.45), being multiparous with a previous term delivery (1.3 mm; 95% CI 0.25 to 1.3), and being multiparous with previous PTB (0.3 mm; 95% CI 1.7 to 0.74) as compared with nulliparous women.
Discussion Main Findings This study found that an increase in maternal age and having a history of term delivery are positively associated with CL. Women who completed vocational or higher education seemed to have an increasingly longer cervix as well, however this association did not reach statistical significance. The use of ART (e.g., intrauterine insemination [IUI], IVF, or ICSI) was associated with a decrease in CL.
Strength and Limitations This study has several strengths. First, we used data of two randomized clinical trials that included a total of 1,447 asymptomatic twin pregnancies. By inclusion of
Table 2 Associations between risk indicators and cervical length as a continuous outcome Risk indicator
Univariable
Multivariable
Mean difference mm/unit (95% CI) 0.30 (0.19 to 0.41)
0.27 (0.16 to 0.39)a
BMI (kg/m )
0.03 (0.07 to 0.14)
–
Caucasian
1.4 (0.3 to 3.2)
–
0.99 (2.1 to 0.1)
1.4 (2.6 to 0.3)a
0.76 (2.1 to 0.57)
0.90 (2.3 to 0.45)
1.7 (3.4 to 0)
1.0 (2.8 to 0.74)
0.11 (1.2 to 0.97)
–
Female–female
Ref.
–
Mixed
0.25 (1.5 to 1)
–
Male–male
0.23 (1.5 to 1.1)
–
Nulliparous
Ref.
Ref.
Multiparous with term delivery
2.1 (1.1 to 3.1)
1.3 (0.25 to 2.4)a
Multiparous with preterm delivery
0.89 (1.1 to 2.9)
0.29 (1.7 to 2.3)
Ref.
Ref.
Age (y) 2
ART
b
Monochorionic Smoking Termination pregnancy
c
Composition of fetal sex
Parity
Education Basic Vocational
1.8 (1.3 to 5.0)
2.1 (1.1 to 5.3)
Higher
2.7 (0.05 to 5.5)
2.6 (0.27 to 5.4)
Abbreviations: ART, assisted reproductive technologies; BMI, body mass index; CL, cervical length; PTB, preterm birth. a p < 0.05. b Mode of conception. c Medical history of pregnancy termination, for example, induced abortion, curettage. American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
1154
asymptomatic twins, we believe these women are a representative sample of the Dutch women with a twin pregnancy. Additionally, the study population is in concordance with general characteristics of the pregnant Dutch population (e.g., for age, parity, and smoking).1,4 Second, the multivariable models include indicators that were collected around study entry that are easily available, and are not modifiable thus minimizing the risk of information bias. As such, the results found are reliable representations of the relationship between these indicators and the outcome. A potential limitation of the study could be the quality of the CL measurement. CL was not part of the inclusion criteria of the primary studies, and was therefore not measured according to a predetermined protocol in the AMPHIA-trial and was only measured once, which could have introduced possible inaccurate measurements.19,20 Consequently, CL data were missing in 18.3% of the included women. However, these missing values were imputed using multiple imputations, which has been shown to be a more valid method to deal with missing values than complete case analysis.25 A second limitation is that the outcome of our study—CL— is a soft endpoint, that is, a risk factor known to be associated with PTB, rather than a hard endpoint such as PTB itself. Consequently, the current study should be viewed as a first step in a broader range of studies looking into the pathway of PTB, that is, with CL as an intermediate factor between risk factors for CL and PTB. Therefore, current results can be used for generating hypotheses.
Interpretation Studies investigating second trimester CL in women with a singleton pregnancy, found that women who were AfroAmerican, had a low BMI < 18 kg/m2 or who had recurrent or spontaneous PTB had an increased risk for a short cervix or shorter CL.10–12 Similar studies in women with a twin pregnancy are currently not published. In our study sample, we found a similar effect of ethnicity in univariable analysis. However, since in our population, 90% of women are Caucasian, while the remaining 10% is a mixture of ethnicities, we were unable to reliably investigate this association. No relationship was found between (low) BMI and CL in this study, mainly because the prevalence of underweight women was low in our population (median BMI 23.4 kg/m2 [IQR 21.2– 26.3]). Other studies investigating the relationship between obstetrical history and CL found that CL in singleton pregnancies, women with isolated PTB and recurrent PTB is shorter compared with women with term births.11,26 This is supported by our results, which demonstrated that multiparous women with a history of term delivery had a longer CL compared with nulliparous women, while women with a history of preterm delivery had no longer CL compared with nulliparous women. This study found no decrease in CL in women with a previous PTB. This could be explained by the fact that PTB < 34 weeks was an exclusion criterion for one of the trials (AMPHIA), leading to a selection of women with a history of less severe PTBs, while the strongest effect of PTB on CL is expected in women with recurrent PTB < 34 weeks.
Hermans et al.
It is shown that there is a possible association between the use of ART and CL, with an average decrease in CL of 1.4 mm in case of ART. In IVF (or ICSI) and IUI, embryos or insemination is performed through the cervix. We hypothesize that this manipulation can cause trauma to the cervix that can lead to scarring or cervical insufficiency. This theory is supported by evidence that other procedures (e.g., induced abortion or curettage) in which the cervix is manipulated (cervical dilatation and evacuation) are associated with an increased risk for PTB.27 Additionally, another study found that surgical management of pregnancy loss was associated with an increased risk of PTB.28 However, the authors could not conclude whether this association was a causal. Higher education seems to have a positive association with CL in linear regression (2.6 mm). This positive association could be explained by considering the educational level as a surrogate marker for social economic status. Basic education was defined as women who had no higher completed education than primary school, which is often associated with a lower social economic class. In these women there could be a higher risk of unhealthy circumstances or behavior (e.g., infection of the genital tract) leading to a decrease in CL.29 A twin pregnancy in general is considered as an important risk factor for PTB and it has also been shown that the prevalence of short CL is higher in twin pregnancies as compared with singleton pregnancies.6,8,9 So it is likely to assume that having a twin pregnancy substantially shortens CL mediating the risk of PTB. However, the actual effect of type of gestation (singleton vs. twin) on CL is unknown. Second associations between other indicators and CL can be less pronounced in twin pregnancies than in singleton pregnancies because of the strong effect of a twin pregnancy on the risk of PTB. Therefore, ideally, these associations should be investigated in a cohort study in which type of gestation is incorporated as an indicator.
Conclusion This study demonstrated that several maternal and obstetric indicators known to be associated with PTB are also associated with CL. Therefore, investigation of possible pathways of cervical shortening should be encouraged to gain more insight in the pathophysiological pathway of PTB.
Conflict of Interest None.
Funding None.
Authors’ Contributions Conceived and designed the experiments: F. J. R. H., B. W. J. M., E. S., B. C. O. Analyzed the data: E. S., F. J. R. H. Wrote the first draft of the article: F. J. R. H. Contributed to the writing of the article: F. J. R. H., E. S., B. C. O., B. W. J. M. Contributed American Journal of Perinatology
Vol. 32
No. 12/2015
1155
Downloaded by: University of Pittsburgh. Copyrighted material.
Indicators for Cervical Length in Twin Pregnancies
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
and acquired data for the original trials, corrected article: S. M. S. L., A. C. L. HD, L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B. ICMJE criteria for authorship read and met: F. J. R. H., E. S., B. W. J. M., B. C. O., S. M. S. L., A. C. L , L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B. Agree with article results and conclusions: F. J. R. H, E. S., B. W. J. M., B. C. O., S. M. S. L., A. C. L HD, L. C. J. S., M. A. O., K. W. M. B., M. M. W., M. M. F., B. N. B., D. J. B.
14 Schuit E, Stock S, Rode L, et al; Global Obstetrics Network
15
16
17
References 1 Schaaf JM, Mol BWJ, Abu-Hanna A, Ravelli ACJ. Trends in preterm
2
3 4
5 6 7 8
9
10
11
12
13
birth: singleton and multiple pregnancies in the Netherlands, 2000-2007. BJOG 2011;118(10):1196–1204 Botting BJ, Davies IM, Macfarlane AJ. Recent trends in the incidence of multiple births and associated mortality. Arch Dis Child 1987; 62(9):941–950 McCormick MC. The contribution of low birth weight to infant mortality and childhood morbidity. N Engl J Med 1985;312(2):82–90 EURO-PERISTAT ProjectEuropean Perinatal Health Report. The Health and Care of Pregnant Women and Babies in Europe in 2010. Available at: http://www.europeristat.com. Accessed April 10, 2014 Romero R, Espinoza J, Kusanovic JP, et al. The preterm parturition syndrome. BJOG 2006;113(Suppl 3):17–42 Berkowitz GS, Papiernik E. Epidemiology of preterm birth. Epidemiol Rev 1993;15(2):414–443 Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008;371(9606):75–84 Goldenberg RL, Iams JD, Miodovnik M, et al; National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. The preterm prediction study: risk factors in twin gestations. Am J Obstet Gynecol 1996;175(4 Pt 1):1047–1053 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(2):128.e1–128.e12 Palma-Dias RS, Fonseca MM, Stein NR, Schmidt AP, Magalhães JA. Relation of cervical length at 22-24 weeks of gestation to demographic characteristics and obstetric history. Braz J Med Biol Res 2004;37(5):737–744 Mercer BM, Macpherson CA, Goldenberg RL, et al; NICHD MaternalFetal Medicine Units Network. Are women with recurrent spontaneous preterm births different from those without such history? Am J Obstet Gynecol 2006;194(4):1176–1184, discussion 1184–1185 Heath VC, Southall TR, Souka AP, Novakov A, Nicolaides KH. Cervical length at 23 weeks of gestation: relation to demographic characteristics and previous obstetric history. Ultrasound Obstet Gynecol 1998;12(5):304–311 Erasmus I, Nicolaou E, van Gelderen CJ, Nicolaides KH. Cervical length at 23 weeks’ gestation—relation to demographic characteristics and previous obstetric history in South African women. S Afr Med J 2005;95(9):691–695
American Journal of Perinatology
Vol. 32
No. 12/2015
18 19
20 21
22
23
24
25
26
27
28
29
(GONet) collaboration. Effectiveness of progestogens to improve perinatal outcome in twin pregnancies: an individual participant data meta-analysis. BJOG 2015;122(1): 27–37 Rust O, Atlas R, Wells M, Rawlinson K. Cerclage in multiple gestation with midtrimester dilation of the internal os. Am J Obstet Gynecol 2001;185(6):S111 Lim AC, Schuit E, Bloemenkamp K, et al. 17α-hydroxyprogesterone caproate for the prevention of adverse neonatal outcome in multiple pregnancies: a randomized controlled trial. Obstet Gynecol 2011;118(3):513–520 Liem S, Schuit E, Hegeman M, et al. Cervical pessaries for prevention of preterm birth in women with a multiple pregnancy (ProTWIN): a multicentre, open-label randomised controlled trial. Lancet 2013;382(9901):1341–1349 Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006;332(7549):1080 Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15(4): 361–387 Atkinson AC. A note on the generalized information criterion for choice of a model. Biometrika 1980;67(2):413–418 Donders ART, van der Heijden GJMG, Stijnen T, Moons KGM. Review: a gentle introduction to imputation of missing values. J Clin Epidemiol 2006;59(10):1087–1091 Moons KGM, Donders RART, Stijnen T, Harrell FE Jr. Using the outcome for imputation of missing predictor values was preferred. J Clin Epidemiol 2006;59(10):1092–1101 Vergouwe Y, Royston P, Moons KGM, Altman DG. Development and validation of a prediction model with missing predictor data: a practical approach. J Clin Epidemiol 2010;63(2): 205–214 Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and some applications. Stat Med 1991;10(4): 585–598 van der Heijden GJMG, Donders ART, Stijnen T, Moons KGM. Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: a clinical example. J Clin Epidemiol 2006;59(10): 1102–1109 de Carvalho MHB, Bittar RE, Brizot Mde L, Bicudo C, Zugaib M. Prediction of preterm delivery in the second trimester. Obstet Gynecol 2005;105(3):532–536 Sedgh G, Singh S, Shah IH, Ahman E, Henshaw SK, Bankole A. Induced abortion: incidence and trends worldwide from 1995 to 2008. Lancet 2012;379(9816):625–632 McCarthy FP, Khashan AS, North RA, et al; SCOPE Consortium. Pregnancy loss managed by cervical dilatation and curettage increases the risk of spontaneous preterm birth. Hum Reprod 2013;28(12):3197–3206 Matser A, Luu N, Geskus R, et al. Higher Chlamydia trachomatis prevalence in ethnic minorities does not always reflect higher sexual risk behaviour. PLoS ONE 2013;8(6):e67287
Downloaded by: University of Pittsburgh. Copyrighted material.
1156
Indicators for Cervical Length in Twin Pregnancies
Hermans et al.
1157
Appendix 1 Missing value analysis p-Valuea
Missing (N ¼ 1,226)
Not missing (N ¼ 221)
Mean (SD) Median (IQR) n (%)
Mean (SD) Median (IQR) n (%)
Maternal age (y)
32.9 ( 4.5)
32.7 ( 4.6)
0.66
Height
170.3 ( 6.8)
170.3 ( 6.6)
0.90
Weight
68.0 (61.0–77.0)
70.0 (63.0–78.0)
0.11
Caucasian
1,044 (91%)
201 (91%)
0.65
Smoking
149 (13%)
25 (11%)
0.68
Basic
33 (4%)
4 (2%)
0.06
Vocational
284 (39%)
100 (45%)
Higher
419 (57%)
117 (53%)
Nulliparous
657 (54%)
122 (55%)
0.71
Previous PTB
81 (7%)
16 (7%)
0.85
Spontaneous abortion
339 (38%)
65 (29%)
0.73
Induced abortion
73 (6%)
13 (6%)
0.95
Curettages
79 (7%)
13 (6%)
0.82
Assisted reproductive technology
437 (36%)
92 (42%)
0.11
Monochorionic twin
257 (21%)
36 (16%)
0.12
GBS carrier
20 (7%)
19 (9%)
0.69
Preeclampsia
166 (14%)
30 (14%)
0.92 0.38
Fetal sex Female–female
390 (32%)
61 (28%)
Mixed
429 (35%)
81 (37%)
Male–male
395 (33%)
79 (36%)
43.9 ( 8.9)
42.4 ( 9.1)
0.03
Spontaneous
462 (38%)
96 (43%)
0.21
Iatrogenic
488 (41%)
76 (34%)
Cervical length Onset of labor
253 (21%)
49 (22%)
Gestational age
C-Section
36.4 (34.1–37.7)
36.9 (24.7–37.7)
0.21
Study participation (AMPHIA)
537 (44%)
116 (52%)
0.95
Randomization result (intervention)
610 (50%)
111 (50%)
0.46
Abbreviations: CL, cervical length; PTB, preterm birth. a Tested accordingly: t-test, chi-square, Mann–Whitney U test.
American Journal of Perinatology
Vol. 32
No. 12/2015
Downloaded by: University of Pittsburgh. Copyrighted material.
Education
