Rate of Change in Cervical Length in Women With Vaginal Bleeding During Pregnancy

Nicholas Behrendt, MD, Ronald S. Gibbs, MD, Anne Lynch, Nancy A. West, PhD, and Jay D. Iams, MD
OBJECTIVE: To evaluate whether women with known risk factors for preterm birth will manifest different rates of cervical shortening preceding a spontaneous preterm birth. METHODS: We conducted a secondary analysis of data from the MaternalFetal Medicine Units Network Preterm Prediction Study. Known risk factors for preterm birth were recorded. Cervical lengths were measured between 22+0 weeks and 24+6 weeks, and again 4 weeks later. Cervical slope was defined as the change in cervical length between these visits divided by time (millimeters per week). Preterm birth was defined as preterm premature rupture of membranes or spontaneous preterm labor leading to delivery before 37 weeks of gestation. We analyzed the data for 2,584 women using logistic regression and tested for interaction between risk factors in the model to determine whether cervical shortening preceded preterm births in all variable groups.
MD, MSPH,

Jan Hart,

MSPH,

RESULTS: Cervical slope was not significantly associated with preterm birth (P5.9) in women with vaginal bleeding. Cervical slope was significantly associated with preterm birth in women without a history of vaginal bleeding (odds ratio 1.2, 95% confidence interval 1.11.4). CONCLUSIONS: Pregnancies without vaginal bleeding have a 20% increase in the risk of preterm birth for each additional millimeter per week increase in cervical slope. Pregnancies with vaginal bleeding are at risk for preterm birth but do not appear to undergo progressive cervical shortening. This suggests that women with vaginal bleeding undergo a different mechanism leading to preterm birth.
(Obstet Gynecol 2013;121:2604) DOI: http://10.1097/AOG.0b013e31827d8e1b

LEVEL OF EVIDENCE: II

From the Departments of Obstetrics and Gynecology, University of Colorado School of Medicine, Aurora, Colorado; and The Ohio State University Wexner Medical Center, Columbus, Ohio. Initial funding for the Preterm Prediction Study was provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The authors thank the Eunice Kennedy Shriver National Institute of Child Health and Human Development MaternalFetal Medicine Units Network, the MaternalFetal Medicine Units Preterm Prediction Study Protocol Subcommittee, and the Biostatistical Coordinating Center at George Washington University for making the database available for this analysis. Presented as a poster at the 32nd Society for MaternalFetal Medicine Annual Meeting, February 611, 2012, Dallas, Texas. The contents of this report represent the views of the authors and do not represent the views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development MaternalFetal Medicine Units Network. Corresponding author: Nicholas Behrendt, MD, University of Colorado School of Medicine, Department of Obstetrics and Gynecology, 12631 East 17th Avenue, Room 4403, B198-6, Aurora, CO 80045; e-mail: Nickbehrendt@gmail.com. Financial Disclosure Dr. Gibbs has been a consultant to Columbia Laboratories. The other authors did not report any potential conflicts of interest. 2013 by The American College of Obstetricians and Gynecologists. Published by Lippincott Williams & Wilkins. ISSN: 0029-7844/13

reterm birth remains the leading cause of infant mortality in the United States.1 Myriad etiologies and risk factors have been identified, including infection, inflammation, cervical length, uterine activity, pregnancy history including induced and spontaneous abortions, maternal behaviors such as tobacco use, and demographic characteristics.24 Preterm birth has been categorized as either indicated or spontaneous, with the latter including women who present for care with signs and symptoms of spontaneous preterm labor or preterm premature rupture of membranes. Some investigators consider spontaneous preterm labor and preterm premature rupture of membranes to be separate clinical entities arising from distinctly different pathways, whereas others assert that they are two clinical presentations of a common pathophysiologic process. We built on previous work investigating cervical length and preterm birth.5 In a previous analysis, we demonstrated that cervical shortening preceded and progressed at the same rate in women who later delivered preterm after spontaneous preterm labor and after preterm premature rupture of

260

VOL. 121, NO. 2, PART 1, FEBRUARY 2013

OBSTETRICS & GYNECOLOGY

membranes.6 That study demonstrated that a decrease in cervical length is a common but not universal antecedent to spontaneous preterm labor and preterm premature rupture of membranes, and suggested that change in cervical length over time may distinguish different physiologic mechanisms leading to preterm birth. In this current analysis from the same dataset, we sought to relate the presence or absence of cervical shortening to clinical risk factors for preterm birth in an effort to identify pathways that did or did not progress through cervical shortening. Specifically, we sought to evaluate whether cervical shortening is associated with preterm delivery independently of known risk factors for preterm delivery (eg, maternal African American race, parity, smoking history, body mass index (BMI, calculated as weight (kg)/[height (m)]2), infection, vaginal bleeding, chronic disease, and pregnancy history including spontaneous miscarriages and previous induced abortions) and to evaluate whether the effect of cervical shortening is modified by known risk factors for preterm delivery. By determining which groups had significantly increased cervical slopes, we hypothesized that we could provide insight as to the pathophysiologic process leading to preterm birth.

MATERIALS AND METHODS

We performed a secondary analysis of data collected in the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal Fetal Medicine Units Network Preterm Prediction Study. These data were collected between 1992 and 1994. This de-identified dataset is available through the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The current analysis was conducted at the University of Colorado Medical Campus and was considered exempt by the Colorado Multiple Institutional Review Board at the University of Colorado Denver Anschutz Medical Campus. Data were collected from 3,073 women at 10 different medical centers. Four clinical visits occurred over the course of the pregnancy starting at 2224 weeks of gestation. Cervical length measurements were obtained by certified ultrasonographers following a predetermined protocol5 at visits 1 (between 22 0/7 and 23 6/7 weeks of gestation) and 3 (4 weeks later, between 26 0/7 and 28 6/7 weeks of gestation). Cervical length measurements were not obtained at visits 2 and 4. Data regarding medical history, current pregnancy information, and delivery outcomes were collected for each patient. Women with pregnancies complicated by pla-

centa previa, cervical insufficiency, or fetal anomalies were not enrolled. As in our previous study, this analysis was limited to women who completed cervical ultrasound examinations at both visits 1 and 3, and for whom delivery data were available.5 For this study, preterm birth is defined as delivery before 37 0/7 weeks of gestation and is further categorized as occurring after spontaneous preterm labor or after preterm premature rupture of membranes. Women with delivery because of a medical indication were excluded from this analysis. Our previous analysis demonstrated that cervical length measurements from this group were not different at visits 1 and 3 from measurements obtained in women who delivered at term. Thus, we excluded data collected from women whose preterm birth was accompanied by a clinical diagnosis of abruption, but we retained data from women who reported vaginal bleeding but had no diagnosis of abruption. Women with placenta previa were not enrolled in the Preterm Prediction Study. Statistical analysis was performed using SAS 9.2. For each woman, the change in cervical length over time, otherwise referred to as cervical slope, was calculated by dividing the difference in cervical length at visits 1 and 3 by the number of weeks between visits 1 and 3, therefore giving a value of millimeters per week. The cervical slopes for each patient in the defined categories of spontaneous preterm labor, preterm premature rupture of membranes, and term deliveries were calculated, and then the mean cervical slope for each category was determined. The following 12 variables were analyzed: age, race, parity, vaginal bleeding, smoking, vaginal infection, chronic disease, BMI, previous spontaneous miscarriage, previous spontaneous delivery at 1319 weeks, previous induced abortion, and cervical slope. Vaginal bleeding was defined as patientreported bleeding occurring in the first or second trimester. Variables also were created for the following: smoking history (defined as any patient reporting smoking at any point during the current pregnancy); infection during pregnancy (defined as a positive test for bacterial vaginosis, chlamydia, or trichomonas); and chronic medical illness during pregnancy (including diabetes, endocrine disorders, hypertension, lung disease, and renal disease). We conducted x2 and t tests to assess characteristics of the sample by preterm delivery status. A logistic regression model was used to investigate the association of cervical slope with preterm birth, independent of maternal age, race, parity, smoking, vaginal infection, chronic disease, BMI, and vaginal bleeding. An interaction term in each model was included

VOL. 121, NO. 2, PART 1, FEBRUARY 2013

Behrendt et al

Cervical Shortening, Bleeding, and Preterm Birth

261

and tested to evaluate heterogeneity of associations between cervical slope and age, race, parity, vaginal bleeding, smoking, vaginal infection, chronic disease, and BMI. P#.05 was considered significant.

RESULTS
There were 2,584 women who had paired cervical length measurements performed, and all variables and delivery outcomes were available for these women. Of these, 141 women delivered after spontaneous preterm labor, 107 women delivered with preterm premature rupture of membranes, and 2,336 delivered at term. Table 1 shows the characteristics of the sample as well as delivery outcomes. When analyzing the chosen risk factors and cervical slope, we observed a significant interaction between vaginal bleeding and cervical slope (P5.01). Therefore, we separated the patients into two groups: patients who experienced vaginal bleeding in pregnancy (n5662) and patients who experienced no vaginal bleeding in pregnancy (n51,922). In patients without vaginal bleeding, increased cervical slope was significantly associated with preterm birth with an adjusted odds ratio of 1.2 (confidence interval 1.11.4). Cervical slope was not associated with preterm birth among patients who experienced vaginal bleeding (P5.9). Table 2 shows associations for other selected risk factors for preterm birth stratified by patients with and without bleeding in pregnancy.

We further divided preterm birth into the categories of spontaneous preterm labor and preterm premature rupture of membranes to evaluate whether cervical slope may have differing effects on these outcomes. We found similar relationships within each of these groups (Table 3).

DISCUSSION
In this large cohort study of pregnancy outcomes, increased cervical slope was significantly associated with preterm birth, but only among women without a history of vaginal bleeding; cervical slope was not associated with preterm birth in women with a history of vaginal bleeding. This was shown by an odds ratio of 1.2, or a 20% increase in the risk of preterm birth for each additional millimeter per week increase in cervical slope. The inclusion of standard risk factors for preterm birth in the models did not substantially change these results. These results may have an important clinical implication regarding evaluating preterm birth. Women with vaginal bleeding may experience preterm delivery through a different pathway manifested by their differences in cervical slope. These women may need to be analyzed separately for research and clinical purposes. Analyzing preterm births as spontaneous preterm labor compared with preterm premature rupture of membranes still showed this relationship, once again suggesting a common pathway between spontaneous preterm labor and preterm premature rupture of membranes.

Table 1. Characteristics of the Cohort

Characteristics
Maternal age Maternal race African American White Multiparity Smoking Maternal BMI category at first visit (kg/m2) Less than 18.5 18.524.0 25.029.9 30 or more Vaginal infection Vaginal bleeding Chronic disease Previous spontaneous miscarriage Previous delivery at 1319 wk Previous induced abortion

Term (n52,336)
23.765.3 1,443 893 989 522 27 776 692 834 158 582 290 496 64 485 (62) (38) (42) (22) (1) (33) (30) (36) (7) (25) (12) (21) (3) (21)

Spontaneous Preterm Labor (n5141)

24.465.6 102 39 46 28 2 75 37 26 12 42 17 33 3 32 (72) (28) (33) (20) (1) (54) (26) (19) (9) (30) (12) (2) (2) (23)

Preterm Premature Rupture of Membranes (n5107)

23.365.7 78 29 37 25 5 43 30 29 10 38 16 25 3 28 (73) (27) (35) (23) (5) (40) (28) (27) (9) (36) (15) (23) (3) (26)

.004 .03 .76 ,.001

.45 .02 .73 .91 .73 .36

BMI, body mass index. Data are mean6standard deviation or n (%) unless otherwise specified.

262

Behrendt et al

Cervical Shortening, Bleeding, and Preterm Birth

OBSTETRICS & GYNECOLOGY

Table 2. Unadjusted and Adjusted Odds Ratios of Cervical Slope and Other Select Risk Factors for Preterm Birth* Among Women With and Without a History of Vaginal Bleeding in the Current Pregnancy
With a History of Vaginal Bleeding (n5662) Risk Factor Unadjusted OR (95% CI) P Adjusted OR (95% CI) P Without a History of Vaginal Bleeding (n51,922) Unadjusted OR (95% CI) P Adjusted OR (95% CI) P

Cervical slope 1.0 (0.871.2) .9 1.0 (0.861.2) .9 1.3 (1.21.4) ,.001 1.2 (1.11.4) ,.001 Maternal age (y) .99 (0.961.0) .4 .99 (0.961.0) .7 1.0 (0.991.1) .17 1.0 (0.991.1) .08 African American 1.6 (1.02.7) .05 2.0 (1.13.4) .02 1.7 (1.22.4) .005 1.6 (1.12.4) .01 .95 (0.812.7) .2 1.6 (0.861.2) .13 .93 (0.621.3) ,.001 .93 (0.910.96) ,.001 BMI (kg/m2) Multiparity 1.8 (1.13.1) .02 1.6 (0.892.8) .12 1.3 (0.941.8) .11 1.4 (1.02.1) .05 Chronic disease .87 (0.521.5) .6 1.2 (0.682.0) .6 1.5 (0.812.7) .2 1.6 (0.861.2) .13 Smoking .91 (0.621.3) .6 1.1 (0.691.6) .8 1.0 (0.591.8) .9 .98 (0.531.8) .9 Vaginal infection 1.7 (1.02.9) .05 1.6 (0.932.8) .09 .71 (0.271.8) .5 .64 (0.241.7) .4 Previous spontaneous miscarriage 0.94 (0.541.6) .8 0.96 (0.541.7) .9 1.2 (0.841.8) .3 1.3 (0.902.0) .2 Previous delivery at 1319 wk 1.3 (0.374.5) .7 0.97 (0.273.6) .9 0.66 (0.202.1) .5 0.70 (0.212.3) .6 Previous induced abortion 1.4 (0.802.3) .3 1.2 (0.712.1) .5 1.1 (0.781.7) .5 1.1 (0.741.6) .6
OR, odds ratio; CI, confidence interval; BMI body mass index. * Spontaneous preterm birth and preterm premature rupture of the membranes. Adjusted for cervical slope, maternal age, African American race, multiparity, body mass index, smoking, vaginal infection, chronic disease, previous spontaneous miscarriage, previous spontaneous delivery at 1319 weeks, and any previous induced abortion.

Ramaeker and Simhan previously showed that cervical length combined with vaginal bleeding is a better predictor for preterm birth than either alone.7 Although their findings might appear to be inconsistent with ours, there are some important methodologic differences. They used a single cervical length measurement recorded at either visit 1 or visit 3 and a history of bleeding to predict subsequent preterm birth, whereas our analysis required cervical length measurements at both visits. Therefore, we were able to study change in cervical length over time in women with vaginal bleeding to investigate whether bleeding may cause or be caused by cervical change. Thus, our results suggest that their findings of improved prediction arise from the combination of two risk factors that are not causally related. Because their study did not analyze cervical shortening over time, this may help explain the difference between this report and the previous analysis if patients with vaginal bleeding started at a shorter cervical length. However, patients in our

study had similar mean starting cervical lengths within each group, making this a less likely conclusion (Fig. 1). For their study, it is not possible to prove whether vaginal bleeding caused the shortened cervical lengths, or vice versa, or if they were independently associated. Our results suggest that this interaction is not attributable to an increased rate of cervical shortening. Strengths of this analysis include a large sample of prospectively collected data and cervical lengths that were measured at defined intervals by certified ultrasonographers. Weaknesses include the limited sample sizes in various subgroups and lack of quantification of vaginal bleeding. Although this study does not definitely prove the different mechanisms leading to preterm birth, we can make a useful observation about the rate of cervical shortening and its interaction with vaginal bleeding. Although others have premature cervical shortening related to genital tract colonization and

Table 3. Adjusted Odds Ratios of Cervical Slope for Preterm Birth Resulting From Spontaneous Preterm Labor or Preterm Premature Rupture of Membranes Among Women With and Without a History of Vaginal Bleeding
History of Vaginal Bleeding (n5662) No History of Vaginal Bleeding (n51,922) Adjusted OR* (95% CI)
Spontaneous preterm labor Preterm premature rupture of membranes 1.0 (0.841.2) 0.99 (0.811.2)

P
.9 .9

Adjusted OR* (95% CI)

1.3 (1.11.4) 1.2 (1.11.4)

P
,.001 .001

OR, odds ratio; CI, confidence interval. * Adjusted for maternal age, African American race, multiparity, body mass index, smoking, vaginal infections, chronic disease, previous spontaneous miscarriage, previous spontaneous delivery at 1319 weeks, and any previous induced abortion.

VOL. 121, NO. 2, PART 1, FEBRUARY 2013

Behrendt et al

Cervical Shortening, Bleeding, and Preterm Birth

263

Term (no bleeding) Preterm (no bleeding) 37 36 Cervical length (mm) 35 34 33 32 31 30 29 28 27 32.3 (8.1) 35.8 (7.9) 35.7 (8.1) 32.9 (9.3)

Term (bleeding) Preterm (bleeding)

selecting candidates for progesterone prophylaxis based on obstetric history of preterm birth. REFERENCES

34.2 (8.2) 33.9 (8.3)

1. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2007 period linked birth/infant death data set. Natl Vital Stat Rep 2011;59:130. 2. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008;371:7584. 3. Romero R, Espinoza J, Kusanovic JP, Gotsch F, Hassan S, Erez O, et al. The preterm parturition syndrome. BJOG 2006; 113(suppl 3):1742. 4. Swingle HM, Colaizy TT, Zimmerman MB, Morriss FH. Abortion and the risk of subsequent preterm birth: a systematic review with meta-analyses. J Reprod Med 2009;54:95108. 5. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad AH, Das A, et al. The length of the cervix and the risk of spontaneous premature delivery. N Engl J Med 1996;334:567572. 6. Iams JD, Cebrik D, Lynch CD, Behrendt N, Das A. The rate of cervical change and the phenotype of spontaneous preterm birth. Am J Obstet Gynecol 2011;205:130.e1-6. 7. Ramaeker DM, Simhan HN. Sonographic cervical length, vaginal bleeding, and the risk of preterm birth. Am J Obstet Gynecol 2011;206:1.e1-4. 8. Goldenberg RL, Gravett MG, Iams JD, Papageorghiou AT, Waller SA, Kramer M, et al. The preterm birth syndrome: issues to consider in creating a classification system. Am J Obstet Gynecol 2012;206:113118.

29.9 (8.2)

28.3 (10.4) 22.024.6 26.028.6 Gestational age (weeks)

Fig. 1. Mean starting cervical lengths in patients with term and preterm labor as well as no bleeding and bleeding during the current pregnancy.
Behrendt. Cervical Shortening, Bleeding, and Preterm Birth. Obstet Gynecol 2013.

infection, accelerated cervical shortening does not appear to cause or be caused by vaginal bleeding. This suggests different pathways that may be distinguished by cervical length measurement8 and that may require different intervention strategies, eg, in

264

Behrendt et al

Cervical Shortening, Bleeding, and Preterm Birth

OBSTETRICS & GYNECOLOGY