Abstract
BACKGROUND: The identification and assessment of environmental risks are crucial for the primary prevention of congenital heart disease (CHD). We were aimed to establish a nomogram model for CHD in the offspring of pregnant women and validate it using a large CHD database in Northwest China. METHODS: A survey was conducted among 29,204 women with infants born between 2010 and 2013 in Shaanxi province, Northwest China. Participants were randomly assigned to the training set and to the validation set at a ratio of 7:3. The importance of predictive variables was assessed using random forest. A multivariate logistic regression model was used to construct the nomogram for the prediction of CHD. RESULTS: Multivariate analyses revealed that the gravidity, preterm birth history, family history of birth defects, infection, taking medicine, tobacco exposure, pesticide exposure and singleton/twin pregnancy were significant predictive risk factors for CHD in the offspring of pregnant women. The area under the receiver operating characteristic curve for the prediction model was 0.716 (95% CI: 0.671, 0.760) in the training set and 0.714 (95% CI: 0.630, 0.798) in the validation set, indicating moderate discrimination. The prediction model exhibited good calibration (Hosmer-Lemeshow χ(2) = 1.529, P = 0.910). CONCLUSIONS: We developed and validated a predictive nomogram for CHD in offspring of Chinese pregnant women, facilitating the early prenatal assessment of the risk of CHD and aiding in health education.