Maternal and obstetrical factors associated with short cervical length at midtrimester in women with no history of preterm delivery

Abstract

Aim

To evaluate associations between maternal characteristics and a short cervix in patients without history of preterm delivery, and to determine if these characteristics can predict the presence of a short cervix.

Materials and methods

This is a retrospective cohort study that included 18,592 women with singleton pregnancies without history of previous preterm deliveries who underwent universal transvaginal cervical length (TVCL) screening between 18 + 0 and 23 + 6 weeks/days of gestation. A short cervix was defined as a cervical length (CL) ≤25 mm, ≤20 mm, and ≤15 mm. Associations between maternal age, weight, height, body mass index (BMI), previous term deliveries, and history of previous miscarriages, with a short cervix were evaluated using logistic regression models.

Results

The prevalence of a short cervix in our population was: CL ≤25 mm, 2.2% (n = 403); CL ≤20 mm, 1.2% (n = 224); and CL ≤15 mm, 0.9% (n = 161). Women with BMI >30 and/or previous abortions constituted 45.5% of the total population (8463/18,582). Significant associations with short cervix were observed for women with BMI ≥30, and for women with at least one previous abortion (p < .001). Parous women had a significantly lower association with a short cervix than nulliparous women (p < .001). Maternal age or height were not associated with a short cervix. Prediction of short cervix based on presence of any of the following: BMI ≥ 30 or previous abortions showed sensitivities of 55.8% (≤25 mm), 61.6% (≤20 mm), and 63.4% (≤15 mm) with similar specificity (50.1–54.6%) and likelihood ratio positive (1.2–1.5); and prediction based on BMI ≥ 30 and previous abortions showed sensitivities of 11.1% (≤25 mm), 14.7% (≤20 mm), and 16.7% (≤15 mm) with specificity 93%.

Conclusions

Among low risk women for spontaneous preterm delivery, those with a BMI ≥30 and/or previous miscarriages had a significantly increased risk for a short cervix at 18 + 0 and 23 + 6 weeks/days of gestation. Despite these significant associations, screening by maternal risk factors in a low risk population of pregnant women should not be an alternative to mid-trimester universal CL measurement.

Key message

In pregnant women evaluated at 18/0 and 23/6 weeks + days of gestation without history of preterm delivery, a 16.7% detection rate for short cervix ≤15 mm can be achieved by risk factors BMI ≥30, and at least one previous miscarriage. Nevertheless, screening for short cervix by risk factors among low risk women might not be an effective alternative to universal cervical length screening.

Keywords:

• Obesity
• obstetric history
• preterm delivery
• prevention
• risk factors
• screening

Introduction

Cervical length (CL) measured using transvaginal (TV) ultrasound is the most powerful parameter to identify pregnant women at risk of spontaneous preterm delivery (sPTD) [1–3]. Women with a CL ≤15 mm at 14–24 weeks have 50% risk of preterm delivery ≤32 weeks of gestation [4,5]. The importance of detecting women with a short cervix relies on the potential benefit of preventive strategies, such as vaginal progesterone, cerclage, and perhaps cervical pessary for reducing the incidence of sPTD [6–8]. Universal CL screening aims to obtain a transvaginal cervical length (TVCL) in all pregnant women during the anatomy scan between 18 and 24 weeks of gestation for identification of those with a short cervix and initiate preventive treatment [9–12]. However, arguments against universal TVCL screening include increased costs, prolonged scanning time, patient’s discomfort, and a low positive predictive value [13–17]. That said, cost-analysis studies favor universal CL screening [18–22]. Other authors also point out that differences in the prevalence of short cervix and in preterm delivery among populations may preclude universal CL screening [23–25]. An alternative approach is screening based on risk factors [26]. Although multiple gestations and history of preterm birth are widely recognized risk factors for a shortened cervix and spontaneous preterm birth [27], other maternal characteristics such as ethnicity and maternal weight have been also associated with CL and a short cervix [28,29]. The aim of this report was to evaluate associations between maternal characteristics and presence of a short cervix at 18–24 weeks in low risk pregnant women.

Materials and methods

Study design and participants

The present study was conducted in the Maternal-Fetal Medicine (MFM) ultrasound units of the University of Texas, McGovern Medical School, Department of Obstetrics and Gynecology between September 2017 and March 2020. This study was approved by the institutional review board from The University of Texas McGovern Medical School (HSC-MS-20-1050).

Low risk pregnant women with singleton pregnancies between 18 + 0 and 23 + 6 weeks/days of gestation were analyzed. Pregnant women with a high risk for preterm delivery due to previous preterm delivery, multiple gestation, known short cervix, and those with a cerclage in place were not included in the analysis. Maternal characteristics such as age, pre-pregnancy weight, height, body mass index (BMI), and previous obstetric history (other than previous preterm deliveries) were documented in all cases. Associations between maternal age, weight, or height either <5th or >95th percentiles with a short cervix defined as ≤25 mm, or ≤20 mm, or ≤15 mm were estimated. BMI was categorized as: normal (18.5 to <25, underweight (<18.5), overweight (≥25 to <30), obese grade I (≥30 to <35), obese grade II (≥35 to 40), and obese grade III (≥40) [30], and their associations with a short cervix analyzed.

Ultrasound examinations

Ultrasound evaluations were performed by Registered Diagnostic Medical Sonographer (RDMS) and Cervical Length Education and Review (CLEAR) certified sonographers and reviewed by a Maternal Fetal Medicine board certified or board eligible specialist. The CLEAR program integrates an online course with an examination and image review for TVCL measurement [31]. Ultrasound examinations were done with GE Voluson E-8 and E10 systems using abdominal and TV ultrasound probes.

All patients had a transabdominal (TA) scan for detailed evaluation of fetal anatomy and biometry. Cervical length was measured following CLEAR guidelines for TV ultrasound as follows: empty bladder, identification of the mid-sagittal plane of the cervix, cervix image magnified occupying at least 2/3 of the ultrasound screen, clear view of the internal and of the external cervical os and of the endocervical canal, and similar thickness of the anterior and posterior cervical lips [31]. Calipers were placed from the internal to the external os following the endocervical canal. Three measurements were obtained and the lowest considered as representative of the patient was used for clinical care. A short cervix was defined as a CL ≤25.0 mm, ≤20.0 mm, and ≤15.0 mm.

Statistical analysis

Associations between CL with maternal age, weight, height, and BMI subgroups were evaluated using regression models. Magnitudes of association were calculated using logistic regression models adjusted for covariance with other maternal characteristics. Prediction was estimated with receiver-operating characteristic (ROC) curves and diagnostic performance analysis. Statistical analyses were performed using SPSS version 29 (SPSS Inc., Chicago, IL).

Results

Demographic characteristics of the study population are presented in Table 1. Women with BMI ≥25 accounted for 55% (10,225/18,592), with BMI ≥30 for 29.7% (5517/18,592), with BMI ≥35; for 16.2% (3021/18,592) and with BMI ≥40 for 7.1% (1329/1859) of the total study group; in total, 3804 (20.4%) women had previous abortions and 1831 (9.8%) had previous abortions and no term deliveries. The prevalence of CL ≤25 mm was 2.2% (n = 403), ≤20 mm was 1.2% (n = 224), and ≤15 mm 0.9% (n = 161).

Table 1. Demographic characteristics of the study population.

Maternal characteristic	Value
Age (years, mean, range)	30 (13–45)
Height (cm, mean, SD)	161.9 (7.3)
Weight (kg, mean, SD)	74.3 (20.5)
BMI (kg/M^2, median, range)	29 (14–60)
BMI <18.5 (n, %)	444 (2.2)
BMI 18.5 to <25 (n, %)	7472 (40.1)
BMI ≥25 to <30 (n, %)	4708 (25.3)
BMI 30–35 (n, %)	2496 (14.4)
BMI 35–40 (n, %)	1692 (9.1)
BMI >40 (n, %)	1329 (7.1)
Gestational age (weeks, median range)	20 (18–23)
CL (mm; mean, SD)	38.4 (7.2)
CL ≤25 mm (n, %)	403 (2.2)
CL ≤20 mm (n, %)	224 (1.2)
CL ≤15 mm (n, %)	161 (0.9)
Gravida (median, range)	2 (0–26)
Primigravida (n, %)	5592 (30.1)
Nulliparous (n, %)	6332 (34.1)
Previous abortions (n, %)	3804 (20.4)
Previous abortions/no term deliveries (n, %)	1831 (9.8%)

BMI: body mass index; CL: cervical length.

Associations

There was a mild but significant positive correlation between maternal weight (r = 0.06; p < .001; Figure 1) and BMI (r = 0.07; p < .001; Figure 2) with CL. As weight and BMI increased, the mean CL was slightly longer, but there was also a wider distribution of data and, therefore, high maternal weight and high BMI were significantly associated with a short cervix. Women with weight >95th percentile or BMI >30 had odds ratios (ORs, 95%CI) of 1.7 (1.4–2.2) and 1.4 (1.2–1.8) for CL ≤25 mm, respectively; ORs 2.3 (1.7–3.1) and 2.0 (1.4–2.7) for CL ≤20 mm, respectively, and ORs 2.6 (1.8–3.6) and 2.0 (1.4–2.9) for CL ≤ 15 mm, respectively. The highest association was observed for women with BMI >40 (CL ≤ 25 mm, OR 2.0 (1.4–2.8); CL ≤20 mm, OR 3.0 (2.0–4.7); and CL ≤15 mm, OR 3.7 (2.3–5.9)) (Table 2). No associations were observed between maternal age or maternal height with a short cervix (Figure 3).

Figure 1. Plot of maternal weight and cervical length; there is a mild but significant association. Regression model: y = 0.0209x + 36.863; R = 0.60, R² = 0.0036; p < .001.

Figure 2. Plot of maternal body mass index (BMI) and cervical length; there is a mild but significant association. Regression model: y = 0.0697x + 36.447; R = 0.071, R² = 0.0051; p < .001.

Figure 3. Plot of maternal height and cervical length; no association was documented. Regression model: y = 0.014x + 40.763; R = –0.14; R² = 0.0002; p < .05.

Table 2. Associations between maternal characteristics and short cervix.

	CL ≤25 mm	CL ≤20 mm	CL ≤15 mm
BMI >40 vs. BMI 18 to <25	2.0 (1.4–2.8); <0.001	3.0 (2.0–4.7): <0.001	3.7 (2.3–5.9); <0.001
BMI >35 vs. BMI 18 to <25	1.6 (1.2–2.1); <0.001	2.3 (1.6–3.3); <0.001	2.5 (1.6–3.7); <0.001
BMI >30 vs. BMI 18 to <25	1.4 (1.1–1.8; 0.004	2.0 (1.4–2.7); <0.001	2.0 (1.4–2.9); <0.001
BMI 25 to <30 vs. BMI 18 to <25	1.1 (0.8–1.4); 0.6	1.2 (0.8–1.8); 0.3	1.1 (0.7–1.7); 0.7
BMI <18 vs. BMI 18–25	0.7 (0.3–1.7); 0.4	0.5 (0.1–2.1); 0.4	0.7 (0.2–3.0); 0.6
Weight >95th	1.7 (1.4–2.2); <0.001	2.3 (1.7–3.1); 0.001	2.6 (1.8–3.6); <0.001
Weight <5th	0.6 (0.2–1.8); 0.3	0.4 (0.1–2.7); 0.3	Na;
Height >95th	1.3 (0.9–1.9); 0.2	1.1 (0.6–2.0); 0.7	0.5 (0.2–1.3); 0.13
Height <5th	0.9 (0.6–1.6); 0.9	1.1 (0.6–2.1); 0.7	1.1 (0.5–2.3); 0.7
Mat age ≤17 years	0.9 (0.2–4.1); 0.9	0.9 (0.1–6.5); 0.9	na
Mat age ≥38 years	1.3 (0.9–1.8); 0.2	1.3 (0.9–2.0); 0.2	1.4 (0.9–2.3); 0.1
Parous vs. no parous	0.7 (0.5–0.8); <0.001	0.5 (0.4–0.7); <0.001	0.4 (0.3–0.6);
Multiparous vs. no parous	0.7 (0.5–0.9); 0.03	0.6 (0.4–1.03); 0.06	0.6 (0.3–1.01); 0.06
Abortions vs. no abortions	1.5 (1.2–1.9); <0.001	1.8 (1.4–2.4); <0.001	2.0 (1.4–2.8); <0.001
One previous abortion vs. no abortions	1.5 (1.1–1.9); 0.005	1.7 (1.2–2.3); 0.004	1.8 (1.2–2.7); 0.003
One or two previous abortions vs. no abortions	1.3 (1.1–1.5); 0.002	1.5 (1.3–1.9); 0.001	1.6 (1.3–2.1); <0.001
≥3 abortions vs. no abortions	1.2 (1.1–1.3); 0.002	1.1 (0.9–1.3); 0.1	1.2 (1.01–1.4); 0.04
Stillbirths vs. no stillbirths	0.9 (0.3–3.0); 0.9	1.2 (0.3–4.8); 0.8	1.7 (0.4–0.7); 0.4
Terminations vs. no terminations	1.2 (0.8–1.9); 0.4	1.7 (1.01–2.9); 0.04	1.9 (1.1–3.5); 0.03
BMI and previous abortions	2.0 (1.4–3.0) <0.001	3.3 (2.1–5.3) <0.001	3.8 (2.3–6.5) <0.001

BMI: body mass index; CL: cervical length.

Bold values are statistically significant.

Women with previous abortions had a significant association with a short cervix (CL ≤25 mm, OR 1.5 (1.2–1.9), CL ≤20 mm, OR 1.8 (1.4–2.4), and CL ≤15 mm, OR 2.0 (1.4–2.8)). The combination of maternal BMI >30 and history of previous abortions showed an even greater association with a short cervix: CL ≤25 mm, OR 2.0 (1.4–3.0); CL ≤20 mm, OR 3.3 (2.1–5.3); and CL ≤15 mm, OR 3.8 (2.3–6.5). Of note, parous women had a significantly lower association with a short cervix than nulliparous women (CL ≤25 mm, OR 0.7 (0.5–0.8); CL ≤20 mm, OR 0.5 (0.4–0.7); and CL ≤15 mm, OR 0.4 (0.3–0.6)).

Prediction

The prediction performance of maternal characteristics is presented in Table 3. In total, 45.6% (8463/18,592) of all women had a BMI ≥30 and/or history of previous abortion. Sensitivities and positive likelihood ratios for short cervix were: for CL ≤25 mm: 55.5% and 1.2 (1.1–1.3); for CL ≤20 mm, 61.6% and 1.35 (1.2–1.5); and for CL ≤15 mm 63.4% and 1.4 (1.2–1.6), respectively. By reducing half the number of TVCL scans based on maternal risk factors, the detection of short cervix is also reduced by 50–40%. In women with both, BMI ≥30 and previous abortions (7.04% [1309/18,592]), a detection of: 11.1% for CL ≤25 mm, 14.7% for CL ≤20 mm, and 16.7% for CL ≤15 mm with specificity of 93% can be achieved by performing TVCL in less than 8% of all pregnant women.

Table 3. Prediction of short cervix by maternal BMI and history of previous abortions.

	BMI ≥30 and/or previous abortions			BMI ≥30 and previous abortions
	CL ≤25	CL ≤20	CL ≤15	CL ≤25	CL ≤20	CL ≤15
Sens	55.8	61.6	63.4	11.1	14.7	16.7
Spec	54.5	54.5	54.6	93	93	93
LR +	1.2 (1.1–1.3)	1.35 (1.2–1.5)	1.4 (1.2–1.6)	1.6 (1.2–2.1)	2.1 (1.5–2.9)	2.4 (1.7–3.4)
LR –	0.8 (0.7–0.9)	0.7 (0.6–0.8)	0.7 (0.6–0.8)	0.95 (0.92–0.99)	0.92 (0.87–0.97)	0.87 (0.7–1.0
PPV	2.6	1.6 (1.5–1.8)	1.2 (1.1–1.4)	3.4	2.5	2.1
NPV	98	99.2	99.4	97.9	98.9	99.2

BMI: body mass index; CL: cervical length; Sens: sensitivity; Spec: specificity; LR+: likelihood ratio positive; LR –: likelihood ratio negative; PPV: positive predictive value; NPV: negative predictive value.

Discussion

The principal findings of this study are as follows: women with BMI ≥30, weight >95th percentile, or with history of previous abortions have significant associations with a short cervix. Multiparous women had a significantly lower association with short cervix than nulliparous women. Screening by presence of any of these risk factors will detect approximately 55% of women with a short cervix (≤25 mm) with 40% false positive rate. Despite the significance of these associations, screening for short cervix by risk factors among low risk women might not be an effective alternative to universal CL screening.

Our results in the context of what is already known

Previous preterm delivery is a well-known risk factor for a short cervix and subsequent preterm delivery; however, other components of the obstetric history may be also related to CL [32]. Our results show that women with history of previous abortions have a higher association with short cervix. This was previously showed by Boelig et al. [33] who reported a significant association with a short cervix (OR, 2.99; 95%CI 1.4–6.4) in women having one or more previous dilation and evacuation for either incomplete spontaneous miscarriage, or an induced abortion. Similar to our results, the authors reported a sensitivity of 52% and specificity of 73% among these women for prediction of a short cervix. Erasmus et al. [34] reported that women with history of previous miscarriages had shorter CLs (–5.3 mm averaged difference) and a higher risk (relative risk (RR) 4.4) for a short cervix, than nulliparous women. Whereas it is difficult to differentiate between spontaneous abortion and termination, the association with short cervix seems to be significant even when both events are included.

Increased maternal BMI has been associated with longer CLs [35,36]; however, this association is mainly related to maternal weight, as maternal height seem not to have an effect on CL [32,37]. We were able to corroborate a mild but significant positive association between maternal weight and CL, but also a significant association between increased maternal weight and short cervix. Whereas these results may appear contradictory, they can be explained by a wide dispersion of CL data in obese patients; whereas the overall mean CL increases, there are also more obese women with short cervix.

Other maternal characteristics such as ethnicity have been also related to CL [38]. Our population is mixed in demographic with a high prevalence of Latin-Hispanics; however, the individual definition of ethnicity varies according to parental origin, place of residence, and to the individual concept each patient has on their own racial origin. Unfortunately, we do not keep track of ethnicity in our ultrasound units.

Screening by risk factors instead of universal cervical length measurement

Screening by risk factors for identification of women with a short cervix can identify 40–60% of women with a short cervix [39]; however, these risk factors are present in about 35% of all pregnant women. This strategy will reduce significantly the number of TSV as compared to universal CL screening, but missing more than 50% of women with a short cervix [40–42]. Specificity and positive predictive values may improve by using the combination of two or more risk factors, but the detection rate will be lower [42]. The conclusion of these studies is that universal CL screening is a better and cost-effective strategy to detect women with a short cervix [43] unless costs for TV ultrasound are extremely high, if the prevalence of a short cervix is very low, if reduction of preterm delivery using progesterone is less than 20%, or if screening by risk factors can detect more than 80% of women with a short cervix [39].

Clinical implications

Arguments against universal CL may need to be reevaluated. Transvaginal ultrasound should be available for any ultrasound unit performing obstetric evaluations. Transvaginal scan takes about 5–10 extra minutes than the TA ultrasound, and by experienced sonographers should not exceed the allotted time for a complete anomaly scan. The direct costs related to TV scan are for ultrasound covers and disinfection, which should not dramatically increase the overall costs of the procedure. Patient’s refusal for TV scan is very low if benefits are well explained [12]. Therefore, universal CL screening is the most appropriate approach for identification of women with a short cervix. Nevertheless, in limited clinical facilities where universal CL screening cannot be implemented, based on our results in low risk pregnant women population, those with BMI ≥30 and/or history of previous abortions have a higher risk of a short cervix and should be strongly considered as candidates for TVCL measurement.

Strengths and limitations

The strengths of this study include a large number of patients, the use of different CL threshold to define a short cervix, and the evaluation of maternal obstetric characteristics other than previous preterm deliveries. All operators involved in the acquisition of the data were experienced sonographers with CLEAR certification who underwent a period of training before performing the TV ultrasound. As limitations, we did not include the prevalence of preterm delivery, as the outcome of our analysis was short cervix.

Conclusions

In low risk pregnant women, those with BMI ≥30 or weight >95th percentile, and or history of previous abortions have a higher association with a short cervix at 18 + 0 and 23 + 6 weeks/days of gestation. Despite this significant association, screening by risk factors in a low risk population of pregnant women may not be recommended as an alternative for universal CL screening.

Author contributions

All authors contributed to the conception or design of the work and participated in the review, drafting, and final approval of the manuscript.

Ethical approval

Data evaluation complies with the guidelines for human studies, and data collection was conducted ethically in accordance with the World Medical Association and the Declaration of Helsinki, under Institutional Review Board approval (HSC-MS-14-0632).

Acknowledgements

The authors thank all RDMS sonographers who contributed in the implementation and establishment of the universal transvaginal cervical length-screening program in the University of Texas McGovern Medical School Department of Obstetrics and Gynecology.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

No funding sources were required for the preparation of this manuscript.