Estimating the individual singleton preterm birth risk: nomogram establishment and independent validation.

Abstract

To establish and independently validate nomograms for predicting singleton preterm birth (PTB) risk based on a large sample size comprising data from two independent datasets. This cohort study used data from 50 states and the District of Columbia in the National Vital Statistics System (NVSS) database between January 2016 and December 2020. Multivariate logistic regression analysis was used to confirm the independent risk factors for PTB. Statistically significant variables were incorporated into the logistic regression models to establish PTB prediction nomograms. The models were developed using the United States (US)-derived data and were independently validated using data from US Territories. A total of 16,294,529 mother-newborn pairs from the US were included in the training set, and 54,708 mother-newborn pairs from the US Territories were included in the validation set. In all, 4 nomograms were built: 1 to predict PTB probability, and another 3 to predict moderately and late PTB probability, very PTB probability, and extremely PTB probability, respectively. Hypertensive eclampsia and infertility treatment were found to be the top 2 contributors to PTB. We developed and validated nomograms to predict the individualized probability of PTB, which could be useful to physicians for improved early identification of PTB and in making individualized clinical decisions.