Objective: This study aimed to develop a nomogram for the risk assessment of any type of birth defect in offspring using a large birth-defect database in Northwest China.
Methods: This study was based on a birth-defect survey, which included 29,204 eligible women who were pregnant between 2010 and 2013 in the Shaanxi province of Northwest China. The participants from central Shaanxi province were assigned to the training group, while the subjects from the south and north of Shaanxi province were assigned to the external validation group. The primary outcome was the occurrence of any type of birth defect in the offspring. A multivariate logistic regression model was used to establish a prediction nomogram, while the discrimination and calibration were evaluated by external validation.
Results: The multivariate analyses revealed that household registration, history of miscarriages, family history of birth defects, infection, taking medicine, pesticide exposure, folic acid supplementation, and single/twin pregnancy were significant factors in the occurrence of birth defects. The area under the receiver operating characteristic curve (AUC) in the prediction model was 0.682 (95% CI 0.653 to 0.710) in the training set. The validation set showed moderate discrimination, with an AUC of 0.651 (95% CI 0.614 to 0.689). Additionally, the prediction model had a good calibration (HL χ2 = 8.106, p= 0.323).
Conclusions: We developed a nomogram risk model for any type of birth defect in a Chinese population based on important modifying factors in pregnant women. This risk-prediction model could be a tool for clinicians to assess the risk of birth defects and promote health education.
Keywords: Chinese population; birth defects; nomogram; prediction model; pregnant women.