Nomogram models for predicting cerclage efficiency in women with different cerclage indications: A 10‐year retrospective observational study

BACKGROUND: Cervical cerclage is the most effective intervention for preventing preterm birth (PTB) caused by cervical insufficiency. However, its benefits remain controversial. This study aimed to develop and validate nomogram models to predict cerclage efficiency, defined as the prevention of PTB or extremely preterm birth (EPTB) and the prolongation of gestational latency, in singleton pregnancies with different cerclage indications. METHODS: A retrospective observational study was conducted using data from women who underwent cerclage at our institution, Wenzhou, China, between 2013 and 2023. The eligible participants were categorized into three groups based on the indications for cerclage: a history-indication group (n = 145), an ultrasound-indication group (n = 101), and an emergency group (n = 41). PTB (<37 weeks) was the primary outcome for the history- and ultrasound-indicated groups, while EPTB (<28 weeks) was the primary outcome for the emergency group, with gestational latency as a secondary outcome. Predictors were identified using univariate analysis, followed by binary logistic or Cox regression. Model performance was evaluated by discrimination, calibration, and decision curve analysis. Internal validation was performed using bootstrap resampling (500 repetitions). RESULTS: For history-indicated cerclage, the nomogram incorporated body mass index at cerclage, preoperative C-reactive protein, and prior cesarean deliveries to predict PTB. For ultrasound-indicated cerclage, PTB predictors were body mass index, miscarriage history, cervical dilation, and operation time. For emergency cerclage, the model for EPTB included vaginal bleeding, U-shaped cervical funneling, and postoperative combination tocolysis. A separate model for gestational latency in the emergency group incorporated U-shaped cervical funneling, cervical dilation, and postoperative ritodrine use. The models demonstrated good discrimination, with apparent areas under the curve (AUCs) of 0.77, 0.75, and 0.92 for predicting PTB/EPTB in the three groups, respectively. Calibration curves indicated favorable concordance between predicted and observed event probabilities. Bootstrap internal validation demonstrated acceptable internal consistency, with bias-corrected mean AUCs of 0.767, 0.728, and 0.914. CONCLUSION: These novel models preliminarily provide accessible and reliable predictions for PTB in women with history or ultrasound indicated cerclage, as well as for EPTB and prolonged gestational latency in women with emergency cerclage. The models have exhibited good predictive values and might assist clinicians in counseling and facilitating personalized management decisions.