Development and validation of a prediction model on spontaneous preterm birth in twin pregnancy: a retrospective
Xiaofeng Yang1,2, Qimei Zhong1,2, Li Li1,2
1Department of Obstetrics and Gynecology, Chongqing Health Center for Women and Children, No.120 Longshan Road, Yubei District, Chongqing, 401147, China.
Background:
This study was conducted to develop and validate an individualized prediction model for spontaneous preterm birth (sPTB) in twin pregnancies.
Methods:
This a retrospective cohort study included 3845 patients who gave birth at the Chongqing Maternal and Child Health Hospital from January 2017 to December 2022. Both univariable and multivariable logistic regression analyses were performed to find factors associated with sPTB. The associations were estimated using the odds ratio (OR) and the 95% confidence interval (CI). Model performance was estimated using sensitivity, specificity, accuracy, area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA).
Results:
A total of 1313 and 564 cases were included in the training and testing sets, respectively. In the training set, univariate and multivariate logistic regression analysis indicated that age ≥ 35 years (OR, 2.28; 95% CI 1.67-3.13), pre-pregnancy underweight (OR, 2.36; 95% CI 1.60-3.47), pre-pregnancy overweight (OR, 1.67; 95% CI 1.09-2.56), and obesity (OR, 10.45; 95% CI, 3.91-27.87), nulliparity (OR, 0.58; 95% CI 0.41-0.82), pre-pregnancy diabetes (OR, 5.81; 95% CI 3.24-10.39), pre-pregnancy hypertension (OR, 2.79; 95% CI 1.44-5.41), and cervical incompetence (OR, 5.12; 95% CI 3.08-8.48) are independent risk factors for sPTB in twin pregnancies. The AUC of the training and validation set was 0.71 (95% CI 0.68-0.74) and 0.68 (95% CI 0.64-0.73), respectively. And then we integrated those risk factors to construct the nomogram.
Conclusions:
The nomogram developed for predicting the risk of sPTB in pregnant women with twins demonstrated good performance. The prediction nomogram serves as a practical tool by including all necessary predictors that are readily accessible to practitioners.
