Reduction in cervical length after vaginal progesterone in women with short cervix is significantly associated with preterm delivery at ≤ 34 weeks and < 37 weeks of gestation

ABSTRACT Objective To evaluate the association between changes in cervical length (CL) after vaginal progesterone treatment and preterm delivery (PTD). Methods This was a retrospective cohort study that included 197 singleton pregnancies without ( n = 178) and with ( n = 19) a history of PTD which were found to have a short cervix (≤ 25 mm) between 18 + 0 and 23 + 6 weeks' gestation with a follow‐up transvaginal CL measurement taken at least 1 week after vaginal progesterone treatment started. Receiver‐operating‐characteristics (ROC)‐curve analysis was performed and three CL shortening patterns were evaluated: (1) ≥ 10% reduction; (2) ≥ 20% reduction; and (3) ≥ 5 mm reduction relative to the first CL measurement. The predictive performance of each CL reduction cut‐off and its association with PTD ≤ 34 weeks and PTD < 37 weeks were evaluated. Results Overall, the rate of PTD ≤ 34 weeks was 16.8% (33/197) and that of PTD < 37 weeks was 36.5% (72/197). The area under the ROC curve of cervical shortening expressed in % for predicting PTD ≤ 34 weeks and PTD < 37 weeks was 0.703 and 0.608, respectively. Cervical shortening was observed in 60/197 (30.5%) patients, with 49/60 (81.7%) women showing ≥ 10% reduction, 32/60 (53.3%) ≥ 20% reduction and 27/60 (45.0%) ≥ 5 mm reduction in CL. Sensitivity and specificity for PTD ≤ 34 weeks were, respectively, 48.5% and 79.9% for ≥ 10% reduction; 36.4% and 87.8% for ≥ 20% reduction; and 27.3% and 89.0% for ≥ 5 mm reduction in CL. For PTD < 37 weeks, sensitivity and specificity were, respectively, 36.1% and 81.6% for ≥ 10% reduction; 27.8% and 90.4% for ≥ 20% reduction; and 20.8% and 90.4% for ≥ 5 mm reduction in CL. The highest positive likelihood ratios for PTD ≤ 34 and < 37 weeks were for ≥ 20% CL reduction (2.98 (95% CI, 1.62–5.49) and 2.89 (95% CI, 1.52–5.57), respectively). Despite significant differences in sensitivity among the different cut‐offs for cervical shortening, favoring the ≥ 10% reduction cut‐off, a reduction of ≥ 20% in CL showed the strongest association with PTD ≤ 34 weeks (odds ratio (OR), 4.11 (95% CI, 1.75–9.62)) and < 37 weeks (OR, 3.62 (95% CI, 1.65–7.96)), as compared with a less pronounced reduction in CL. Conclusions In women with a short cervix treated with vaginal progesterone, a reduction in CL on a subsequent ultrasound scan can predict PTD ≤ 34 and < 37 weeks. A ≥ 20% reduction in CL had the highest positive likelihood ratio and strongest association with PTD ≤ 34 and < 37 weeks compared with ≥ 10% or ≥ 5 mm reduction. © 2023 International Society of Ultrasound in Obstetrics and Gynecology.