Background: Current studies have confirmed that fetal congenital heart diseases (CHDs) are caused by various factors. However, the quantitative risk of CHD is not clear given the combined effects of multiple factors. Objective: This cross-sectional study aimed to detect associated factors of fetal CHD using a Bayesian network in a large sample and quantitatively analyze relative risk ratios (RRs). Methods: Pregnant women who underwent fetal echocardiography (N = 16,086 including 3,312 with CHD fetuses) were analyzed. Twenty-six maternal and fetal factors were obtained. A Bayesian network is constructed based on all variables through structural learning and parameter learning methods to find the environmental factors that directly and indirectly associated with outcome, and the probability of fetal CHD in the two groups is predicted through a junction tree reasoning algorithm, so as to obtain RR for fetal CHD under different exposure factor combinations. Taking into account the effect of gestational week on the accuracy of model prediction, we conducted sensitivity analysis on gestational week groups. Results: The single-factor analysis showed that the RRs for the numbers of births, spontaneous abortions, and parental smoking were 1.50, 1.38, and 1.11 (P < 0.001), respectively. The risk gradually increased with the synergistic effect of ranging from one to more environmental factors above. The risk was higher among subjects with five synergistic factors, including the number of births, upper respiratory tract infection during early pregnancy, anemia, and mental stress as well as a history of spontaneous abortions or parental smoking, than in those with less than 5 factors (RR = 2.62 or 2.28, P < 0.001). This result was consistent across the participants grouped by GWs. Conclusion: We identified six factors that were directly associated with fetal CHD. A higher number of these factors led to a higher risk of CHD. These findings suggest that it is important to strengthen healthcare and prenatal counseling for women with these factors.
Read full abstract