Abstract
Background: We used prepregnant and gestational characteristics as predictors to develop and validated a nomogram predicting the risk of preterm birth (PTB) in assisted reproductive technology (ART) treated women. Methods: The National Vital Statistics System (NVSS) was queried for singleton ART-treated pregnant women from 2015 to 2019. Multivariable cox regression was used to develop the early (< 32 weeks) or late (< 37 weeks) PTB risk model using both statistical significance and clinical importance criteria for variable selection. The predictive accuracy was assessed, and bootstrapping was used for validation. A nomogram was constructed for the presentation of the final model. Results: ART-treated women who were over 45 years old, black, obese, had a history of cesarean section and PTB, restarting ART within 3 months, prepregnant diabetes, chronic hypertension, gestational diabetes, gestational hypertension, and eclampsia, had the highest risk for late and early-stage PTB. The nomogram with these variables accurately predicted PTB in ART women with a singleton pregnancy. (Brier score:0.121, calibration slope: 0.99, c-index: 0.684). Conclusion: We created a nomogram predicting the risk of early or late PTB in ART women with a singleton pregnancy, which could identify potentially at-risk women who seeking ART treatment and inform appropriate preterm care.