Background
Mid-trimester transvaginal cervical length assessment can identify women who are at risk of preterm birth and afford opportunities for preterm birth prevention. However, the incidence of a short cervix is low, and some physicians have questioned whether a universal screening program among women without a previous preterm birth would be beneficial.
Objective
The purpose of this study was to examine whether the introduction of a universal transvaginal cervical length screening program is associated with a reduction in the preterm birth rate.
Study Design
This is a cohort study of women with singleton gestations and without any previous preterm births who underwent an obstetric sonogram at 18-24 weeks of gestation and who had their delivery at a single tertiary institution from January 2007 to January 2014. In July 2011, a program was implemented in which all pregnant women who had a sonogram at 18-24 weeks of gestation were to receive a transvaginal cervical length measurement. The preterm birth rates were compared before and after the implementation of the universal cervical length screening program. Multivariable analysis was used to identify whether the universal cervical length screening program was associated independently with the frequency of preterm birth. The Breslow-Day test for homogeneity was used to assess whether any interaction existed in the association based on parity.
Results
Of 64,207 eligible women, 46,598 underwent their mid-trimester sonogram before the universal cervical length screening program, and 17,609 underwent a sonogram after implementation of the program. Of the 17,590 women (99.9%) who agreed to cervical length measurement, 157 (0.89%) had a measurement of ≤25 mm. The introduction of the cervical length program was associated with a significant decrease in the frequency of preterm birth at <37 weeks of gestation (6.7% vs 6.0%; adjusted odds ratio, 0.82 [95% confidence interval, 0.76–0.88]), <34 weeks of gestation (1.9% vs 1.7%; adjusted odds ratio, 0.74 [95% confidence interval, 0.64–0.85]), and <32 weeks of gestation (1.1% vs 1.0%; adjusted odds ratio, 0.74 (95% confidence interval, 0.62–0.90]). This reduction in frequency of preterm birth primarily was due to a change in spontaneous (and not medically indicated) preterm births. The effect size for the reduction in preterm birth was similar in nulliparous and multiparous women with previous term births.
Conclusion
The introduction of a universal transvaginal cervical length screening program in women without a history of preterm birth is associated with a reduction in the frequency of preterm birth.
Preterm birth is a major cause of neonatal morbidity and death. It is estimated that 11-12% of neonates born each year in the United States are delivered preterm, which is defined as birth at <37 weeks of gestation. Most successful efforts to reduce the frequency of spontaneous preterm birth have focused on women with previous preterm births. Unfortunately, most preterm births occur in women who have no such history. Therefore, there remains an important need for the development of effective screening and treatment strategies for the prevention of preterm birth in women without a history of preterm birth.
One proposed strategy is the transvaginal measurement of cervical length at 18-24 weeks of gestation, with the recommendation for vaginal progesterone treatment if a short cervix is identified. Vaginal progesterone is associated with an approximate 40% reduction in the risk of preterm birth in women with a short cervix. However, the incidence of a short cervix in a low-risk population has been reported to be 1-2% ; therefore, the effectiveness of a program of routine cervical length screening in women without an increased risk for preterm birth is still controversial.
Therefore, our objective was to examine whether the introduction of a universal mid-trimester transvaginal cervical length (TVCL) screening program is associated with a reduction in the preterm birth rate in women without a history of preterm birth.
Material and Methods
This is a retrospective cohort study of nulliparous or multiparous women with no previous preterm birth who underwent an obstetric sonogram between 18 and 24 weeks of gestation and who had their delivery at Northwestern Memorial Hospital in Chicago, IL, from January 2007 to January 2014. In July 2011, our institution implemented a universal cervical length screening program in which all pregnant women who have an obstetric sonogram at 18-24 weeks of gestation were to receive a TVCL measurement. Women who had a mid-trimester obstetric sonogram before the implementation of the screening program (prescreen group) were compared with women who had a mid-trimester obstetric sonogram after the implementation of the screening program (screen group).
Women were included in the cohort if they were at least 18 years old, had a singleton gestation, and had available sonogram and delivery records. Women were excluded if they had a previous preterm birth at <37 weeks of gestation. Women were also excluded if they experienced a spontaneous pregnancy loss at <20 weeks of gestation (because they may not have had the opportunity to receive routine cervical length screening and/or been offered intervention) or if they terminated the pregnancy at <24 weeks of gestation (given the inability to assess the natural progression of their pregnancies). A separate analysis was performed in which women who were found to have short cervices and subsequently terminated their pregnancies were included and counted as preterm births to ensure that any observed change in the frequency of preterm birth was not attributable solely to the exclusion of these pregnancies.
All potentially eligible women were identified through a query of the hospital’s electronic database, which includes information on all electronic medical records and sonographic reports. There were no significant changes to the electronic medical record database during the study period. Charts were abstracted for demographic and baseline clinical information and for obstetric outcomes. Demographic and baseline clinical data that were collected included maternal age, race/ethnicity, body mass index at delivery, smoking history, and any preexisting comorbidities (eg, chronic hypertension, diabetes mellitus) or a history of cervical excision procedures (ie, cold knife cone or loop electrosurgical excision procedure). Obstetric data that were abstracted included TVCL measurement (for women in the screen group), placement of physical examination–indicated cervical cerclage, initiation of vaginal progesterone therapy, gestational age at delivery, and indication for delivery (medically indicated vs spontaneous).
Cervical length images were obtained by staff sonographers, all of whom had been trained to obtain TVCL measurements according to the method of Iams et al ; the images were reviewed by attending perinatologists. A TVCL ≤25 mm was defined as a short cervix. Women with a TVCL between 20 and 25 mm were often asked to return for 1 follow-up TVCL at <24 weeks of gestation; in such cases, we incorporated the smaller measurement of the 2 for analyses. Vaginal progesterone was recommended routinely for women with a TVCL ≤20 mm. The finding of a short cervix on ultrasound examination typically prompted a digital cervical examination. A physical examination–indicated cerclage typically was recommended if there was cervical dilation on digital examination at <24 weeks gestational age.
Women were compared before (prescreen group) and after (screen group) implementation of our institution’s universal TVCL screening program in July 2011. The primary outcome was the frequency of preterm birth at <37 weeks of gestation. The secondary outcomes were the frequencies of preterm birth at <34 and <32 weeks of gestation and of spontaneous preterm birth at <37, <34, and <32 weeks of gestation. Spontaneous preterm birth was defined as a birth after preterm labor or preterm premature rupture of membranes. Bivariable comparisons were performed with the use of the Mann-Whitney U test or χ 2 analysis for continuous and categoric variables, respectively. Multivariable logistic regression analyses were performed to assess whether the use of a universal TVCL screening program was associated independently with the frequency of preterm birth. Variables that significantly differed by exposure ( P < .05) in the bivariable analyses were included in the multivariable logistic regression equations. The Breslow-Day test for homogeneity was used to assess whether there were differences in the odds of preterm birth after stratification by parity (nulliparous compared with multiparous women). A nonparametric test of trend was performed in the prescreen group to assess the frequency of preterm birth stratified by year before the institution of the screening program to test whether any secular trend in the preterm birth rate was occurring before the initiation of the program. All hypotheses tests were 2-tailed, and a probability value of <.05 was used to define statistical significance. All statistical analyses were performed with Stata software (version 13.1; StataCorp, College Station, TX). An a priori power calculation was performed with the use of an alpha of .05 and beta of .2. With an estimated risk for preterm birth of 6.6% in the prescreen group based on institutional data and the use of a 2:1 nonexposed:exposed ratio, 38,836 women in the prescreen group and 19,418 women in the screen group would be needed to detect at least a 10% reduction in preterm birth associated with the screening program. We projected that an observed period of 7 years would allow us to obtain the targeted sample size. Approval for this study was obtained from the Northwestern University Institutional Review Board with a waiver of informed consent.
Results
Of the 64,207 eligible women during the period of study, 46,598 (72.6%) underwent a mid-trimester obstetric sonogram before the introduction of the universal TVCL program (none of whom received a routine transvaginal sonogram), and 17,609 (27.4%) underwent an obstetric sonogram after implementation of the program. The 34,856 women (54.3%) in the overall cohort were nulliparous. Women in the screen group had a higher median body mass index and were more likely to be of non-Hispanic white race, have a history of a cervical excision procedure, and have chronic hypertension. Conversely, they were less likely to be using cigarettes or have pregestational diabetes mellitus ( Table 1 ).
| Characteristic | Prescreen group (n = 46,598) | Screen group (n = 17,609) | P value |
|---|---|---|---|
| Age, y a | 32 (29-35) | 32 (29-35) | .412 |
| Nulliparous, n (%) | 25,287 (54.3) | 9,569 (54.3) | .864 |
| Race/ethnicity, n (%) | < .001 | ||
| Non-Hispanic white | 29,333 (62.9) | 11,166 (63.4) | |
| Non-Hispanic black | 4,454 (9.6) | 1,690 (9.6) | |
| Hispanic | 10,190 (21.9) | 3,305 (18.8) | |
| Asian | 2,376 (5.1) | 1,036 (5.9) | |
| Other/unknown | 245 (0.5) | 412 (2.3) | |
| Body mass index at delivery (n = 64,140), kg/m 2 a | 28.7 (26.0-32.3) | 29.1 (26.4-32.7) | < .001 |
| Previous cervical excision procedure, n (%) | 2,647 (5.7) | 1,281 (7.3) | < .001 |
| Cigarette smoker (n = 62778), n (%) | 3,632 (7.9) | 599 (3.4) | < .001 |
| Chronic hypertension, n (%) | 681 (1.5) | 396 (2.3) | < .001 |
| Pregestational diabetes mellitus, n (%) | 344 (0.7) | 93 (0.5) | .004 |
In the prescreen group, 57 women underwent placement of a physical examination–indicated cerclage because of the incidental finding of cervical dilation on physical examination, but none of the women received vaginal progesterone therapy. In the screen group, routine TVCL measurement was performed in 17,590 women (ie, 99.9% of eligible women accepted transvaginal sonography) and occurred at a median gestational age of 20.7 weeks of gestation (interquartile range, 20.1-21.4). One hundred fifty seven women (0.89%) had a TVCL measurement of ≤25 mm, and 83 women (0.47%) had a TVCL measurement of ≤20 mm. Forty-three women received vaginal progesterone therapy (40 had TVCL ≤20 mm and 3 had TVCL 20-25 mm); 6 women initially received vaginal progesterone and then ultimately required placement of physical examination–indicated cerclage (all had TVCL ≤20 mm), and 31 women underwent placement of physical examination–indicated cerclage without progesterone supplementation (25 had TVCL ≤20 mm and 6 had TVCL 20-25 mm). An additional 20 women who had a TVCL >25 mm at the time of screening later underwent placement of physical examination–indicated cerclage because of subsequent finding of cervical dilation. There were no women who received a cervical pessary. Twelve women with a TVCL ≤20 mm did not receive any interventions, either because of a lack of physician recommendation or patient declination of the intervention. Of the women who were found to have a TVCL ≤25 mm, 50 of 157 women (31.8%) delivered at <37 weeks of gestation. Of the women who were found to have a TVCL ≤20 mm, 37 of 83 women (44.6%) delivered at <37 weeks of gestation.
The introduction of the TVCL program was associated with a significant decrease in the frequencies of preterm birth at <37, <34, and <32 weeks of gestation ( Table 2 ). A nonparametric test of trend by year in the prescreen group did not reveal any preexisting trends in the preterm birth frequency before the initiation of the TVCL screen program ( P = .10). Furthermore, when the preterm birth outcome was restricted to spontaneous causes, the TVCL screening program was also associated with a significant decrease in the frequencies of spontaneous preterm birth at <37, <34, and <32 weeks of gestation. There were no statistically significant differences between the groups in the frequencies of medically indicated preterm births at <37, <34, and <32 weeks of gestation. The differences in total and spontaneous preterm birth rates persisted after adjustment for potential confounders in multivariable regression models ( Table 3 ). The number needed to screen (NNS) to prevent 1 case of preterm birth at <37 weeks of gestation was 143 (95% confidence interval, 120–172); the NNS to prevent 1 case of preterm birth at <34 weeks of gestation was 500 (95% confidence interval, 362–721), and the NNS to prevent 1 case of preterm birth at <32 weeks of gestation was 1000 (95% confidence interval, 619–1649).
