Two-stage nomogram models in mid-gestation for predicting the risk of spontaneous preterm birth in twin pregnancy.

Abstract

This study aimed to develop two-stage nomogram models to predict individual risk of preterm birth at < 34weeks of gestation in twin pregnancies by incorporating clinical characteristics at mid-gestation. We used a case-control study design of women with twin pregnancies followed up in a tertiary medical centre from January 2018 to March 2019. Maternal demographic characteristics and transvaginal cervical length data were extracted. The nomogram models were constructed with independent variables determined by multivariate logistic regression analyses. The risk score was calculated based on the nomogram models. In total, 65 twin preterm birth cases (< 34weeks) and 244 controls met the inclusion criteria. Based on univariate and multivariate logistic regression analyses, we built two-stage nomogram prediction models with satisfactory discrimination and calibration when applied to the validation sets (first-stage [22-24weeks] prediction model, C-index: 0.805 and 0.870, respectively; second-stage [26-28weeks] prediction model, C-index: 0.847 and 0.908, respectively). Restricted cubic splines graphically showed the risk of preterm birth among individuals with increased risk scores. Moreover, the decision curve analysis indicated that both prediction models show positive clinical benefit. We developed and validated two-stage nomogram models at mid-gestation to predict the individual probability of preterm birth at < 34weeks in twin pregnancy.