Methods
This is a prospective multicenter cohort study involving singleton pregnant women screened during a multicenter randomized controlled trial entitled “Pessary plus Progesterone for Preventing Preterm Birth” (P5 trial; Registration no. RBR-3t8prz, approved by the Brazilian National Review Board/CONEP - number 1.055.555) 11. The P5 trial was conducted by the University of Campinas (UNICAMP) and involved 17 centers in nine states of Brazil from July 2015 to March 2019. Women between 18 and 22(+6/7) gestational weeks were invited to participate in the P5 screening phase. A consent form was signed and TVU was performed to measure the CL.
The standard technique followed the P5 study protocol and the Fetal Medicine Foundation orientation for CL measurement. Briefly, with the woman in dorsal lithotomy position and empty bladder, a TVU probe was introduced inside the vagina until the anterior fornix avoiding pressure. A sagittal view of the cervix, including the edge, identified the internal and external ostium. Calipers were used to measure the linear distance (in mm) between the external and internal ostium. Funneling and Sludge were described. All data from the screening phase were included in the online database Gsdoctor. Every participating center stored their ultrasound images with the CL measurements to confirm that all centers were correctly applying the TVU technique.
All women with a CL ≤30mm who did not have exclusion criteria and who accepted to participate in the trial were randomized into two groups: 200 mg/day vaginal progesterone or 200 mg/day vaginal progesterone + cervical pessary. Randomized women have delivery information in the P5 database. Women with CL >30 mm had their childbirth and postnatal information collected from hospital medical registers and added to the P5 database.
The sample for this analysis considered all women with CL ≤30mm receiving only progesterone and a random selection of women with CL >30mm, keeping the populational distribution of cervical length. Women using cervical pessary were excluded since we did not have clear information of how it could influence the gestational age at birth and this treatment is not routine for preventing PTB. Considering that progesterone is an established evidence-based treatment for preventing PTB and women are encouraged to use it if they have a short CL identified in the mid-trimester, we included the P5 trial progesterone group in our cohort sample. The P5 trial total sample screened 13.7% women with CL≤30mm and 86.3% of CL>30mm. To maintain the same CL distribution, we projected the progesterone group to correspond to 13.7% of CL ≤30mm for our analysis. To complete our final sample and reach the complementary 86.3% of CL >30mm, we selected singleton women with CL >30 mm using a random model. We excluded women who had received a cervical pessary, multiple gestations and those with incomplete gestational outcome data. We kept very similar baseline characteristics percentages found in the total of singleton pregnant that participated in the P5 trial screening, maintaining homogeneity and avoiding any possible selection bias (Table S1). The primary outcome was PTB at <37 weeks’ gestation and secondary outcomes were sPTB at <37, <34, <32 and <28 weeks’ gestation.
Descriptive statistical analysis was performed for demographic characteristics, expressed as means and percentages. Logistic regression was used to estimate odds ratios for baseline characteristics, gestational age and CL at measurement. A multivariate logistic regression analysis was performed to estimate adjusted odds ratio for different gestational ages.
For our primary outcome, receiver operating characteristic (ROC) curve analysis was performed to identify the most effective cutoff point to predict a PTB (<37 weeks). Our secondary outcomes were ROC curve analysis to identify the most effective cutoff points to predict sPTB at different gestational ages (<37, <34, <32 and <28 weeks). Kaplan-Meyer survival curves were used to analyze time to delivery, considering CL intervals (≤10mm, 10-15mm, 15-20mm, 20-25mm, 25-30mm, 30-35mm, 35-40mm and >40mm). We calculated the number needed to screen (NNS) to detect one true positive sPTB<34 in women with CL ≤25mm. Considering a recent IPD-metanalysis that included RCTs involving women with CL ≤25mm treated with vaginal progesterone, the number needed to treat (NNT) with vaginal progesterone to prevent one sPTB <34 weeks is 1812. Therefore, we estimated the number of TVU necessary to identify 18 women with CL ≤25mm. P<0.05 was considered as statistically significant. All statistical analyses were performed using R version 3.6.2 software.
This study was funded by Bill & Melinda Gates Foundation [OPP1107597], the Brazilian Ministry of Health, and the Brazilian National Council for Scientific and Technological Development (CNPq) [401615/20138]. The funders had no role in the design, development of the study, analysis, interpretation of data, writing the manuscript and in the decision to submit the article for publication.