Objective:
To understand the current status of the incidence of childhood congenital heart disease (CHD) and its distribution characteristics in Xinjiang, to explore the influencing factors for the occurrence of CHD in Xinjiang, and to develop an effective prediction model for CHD as a preliminary screening tool based on comprehensive epidemiological data for identifying pregnant women at high risk for CHD in early pregnancy, which can help prenatal care providers to guide prenatal management and prevention, and provide an important theoretical basis for the prevention of congenital heart disease in children in the region.
Methods:
This study was a 1:1.8 matched case-control study in which children aged 1-14 years were selected for on-site congenital heart disease screening from 1 August 2023 to 21 December 2023 in a populous county in Xinjiang, and 196 cases of congenital heart disease diagnosed as the case group, and 360 cases of the same class of students judged to be healthy who underwent a health check-up during the same period of time, as the control group. The questionnaire "Questionnaire on the Prevalence Status of Congenital Heart Disease in Children and Related Risk Factors in Xinjiang Uygur Autonomous Region" was administered to the included study subjects. Statistical analyses were performed by using univariate and multifactorial logistic regression analyses to explore the correlation between congenital heart disease and the influencing factors by comparing the case group with the control. Further the predictors were screened by removing the covariate variables through variance inflation factor and four Machine Learning (ML) models were constructed for prediction, the four machine learning models were evaluated and thus the performance of the models were compared.The hypothesis was tested at level=0.05, and the difference was statistically significant by using a two sided test P<0.05.
Results:
congenital heart disease screening results and distribution: a total of 13,268 children were screened, of which 196 were screen positive, with a positive rate of 14.77 per thousand; 125 cases (63.78%) of all types of congenital heart disease had the highest percentage of oval foramen not closed.
2. General demographic characteristics of the mothers: there were differences between mothers in the CHD case group and the control group in terms of age, age at birth, BMI, ethnicity, occupation, place of residence, and educational level (P<0.05).
3. analysis of factors related to congenital heart disease: mothers in the CHD case group and control group differed statistically (P<0.05) in whether they had a full-term birth, mode of delivery, whether they were consanguineous or not, the number of pregnancies, the number of births, history of illnesses during pregnancy, folic acid intake during pregnancy, history of adverse environmental exposures during pregnancy, negative life events during pregnancy, history of exposure to tobacco, and history of exposure to tobacco; and whether they were multiparous, there was no difference (P<0.05); and whether they were multiparous or not, there was no difference (P<0.05). were not statistically different (P>0.05).
4. Single-factor logistic regression analysis of congenital heart disease: 11 risk factors and 7 protective factors were screened, of which age, age at delivery, education level of bachelor's degree or above, taking folic acid before and after cesarean section delivery in the first and second trimesters of pregnancy, and consuming alcohol 1-3 times a month were the independent protective factors of congenital heart disease (OR<1.0); occupation of farmer, parturition, consanguineous marriage, other family history of birth defects, pregnancy history of birth defects, and other family history of birth defects. other family history of birth defects, ≥2 pregnancies, ≥2 deliveries, illness during pregnancy, exposure to organic solvents during pregnancy, negative life events during pregnancy, and passive smoking were independent risk factors for congenital heart disease (OR>1.0).
5. Multifactorial logistic regression analysis of congenital heart disease: BMI (OR=0.66, 95% CI: 0.55-0.79), age at childbearing (OR=0.79, 95% CI: 0.74-0.85), education as a university college (OR=0.002, 95% CI: 0.001-0.09), and having taken folic acid all the way through pregnancy ( OR=0.02, 95% CI: 0.01 to 0.06) were negatively associated with the occurrence of congenital heart disease in children; consanguineous marriage (OR=7.23, 95% CI: 2.54-20.60), parturition (OR=3.56, 95% CI: 1.20-10.57), illness during pregnancy (OR=3.40, 95% CI: 1.29- 8.97), negative life events during pregnancy (OR=4.64, 95% CI: 1.90-11.31) were positively associated with the occurrence of congenital heart disease in children.
6. Dose-response relationship between BMI, age at childbearing and congenital heart disease: According to the dose-response relationship, BMI, age at childbearing and CHD showed a non-linear relationship (overall<0.01, nonlinear<0.01), and the risk levelled off when the BMI and age at childbearing were greater than 21.8kg/m2, and 26 years old, respectively.
7.Development and validation of machine learning models: construct GBM, Random Forest, SMoteRf and KNN model test set to validate the stability and generalisation ability of the 4 prediction models. The four models were evaluated by five dimensions: accuracy, precision, recall, F1 score, and AUC.The five indexes of the GBM model were higher than the other models, including the AUC of the training set (0.986, 0.978-0.9860), and the AUC of the test set (0.975, 0.957-0.975).
Conclusion:
The incidence of CHD in children in a county of Xinjiang is higher than the national average, age at birth, BMI, education level of university college, and adherence to folic acid during pregnancy are the protective factors for congenital heart disease in offspring, consanguineous marriage, parturition, history of illnesses during pregnancy, and negative life events during pregnancy are the risk factors for congenital heart disease in offspring, and the GBM model has demonstrated a better clinical predictive value and good efficiency and stability of classification models, which can effectively assist physicians in the diagnosis of congenital heart disease. Therefore, the target population can be regularly educated to popularise the knowledge about congenital heart disease and the risk factors. For pregnant women, especially to ensure that early pregnancy folic acid dosage, reasonable control of BMI, but also to maintain a healthy lifestyle and mood can reduce the occurrence of congenital heart disease to a certain extent.
谷歌
火狐
