A Study On Determinants And Forecasting Model Of The Five Top Birth Defects Of Zhengzhou City

Objectives:To explore the influencing factor of five top birth defects(neural tube defects,cleft lip, multi-finger(toe), congenital hydrocephalus and congenital heart disease) in Zhengzhou and build the discriminant analysis models of them.Methods:This was a hospital-based frequency matched case-control study. The source population were who gave birth to child in The Third Affiliated Hospital of Zheng Zhou University, The Fifth Affiliated Hospital of Zheng Zhou University,Armed Police Hospital, The Second People’s Hospital of Zhengzhou, The Sixth People’s Hospital of Zhengzhou and Erqi District Mazhai clinic from February 20,2013 to February 20, 2014. Total of 334 subjects with BD(cases) were selected, and then, according to the delivery time and the frequency of age constituted of the case group, 668 healthy subjects(controls) were selected. The ratio of cases to control was1:2. The data of case and control group were entryed by using Epidata 3.1. The statistical analysis software is SPSS 17.0. Logistic regression analysis applied the five birth defects to find the influence factors in univariate and multivariate analysis.Univariate Logistic regression, multivariate Logistic regression were used to analyze the influencing facttors of BD. The significant level was set as α=0.05. According to results of univariate logistic analysis, stepwise discriminant analysis was adopted to construct fisher liner discriminant analysis forecasting model.Results:1.The t test of two independent-samples showed that the height, pre-pregnancy weight, pre-pregnancy BMI and the number of birth defects crowd of the case group were similar with the control group, the difference was no statistically significant(P>0.05).2.Results from univariate analysis showed that the following factors are positively associated with the incidence of BD: radial, drugs, smoking, alcohol,passive smoking, exhaust gas and effluent presence near the residence, pregnancy eatmore eggs, beans and soy products, multivitamin supplementation, folic acid, vitamin E and iron, pregnancy diabete. Multivariate logistic regression analysis showed that factors associated with birth defects are: drugs(OR=2.038), alcohol(OR=14.336),passive smoking(OR=1.054), exhast gas(OR=4.338), reasonable eating eggs(OR=0.695), reasonable eating beans and soy products(OR=0.555), multivitamin supplementation(OR=0.364), folic acid supplementation(OR=0.456), VE supplementation(OR=0.066), Iron supplementation(OR=0.232), gestational diabetes(OR=1.787).3.The variable entered into forecasting model including radial(X1), drugs(X2),alcohol(X3), passive smoking(X4), exhaust gas(X5),effluent(X6), reasonable eating eggs(X7), reasonable eating beans and soy products(X8), folic acid supplementation(X9), VE supplementation(X10), gestational diabetes(X11). The function of forecasting model was: Z=0.809X1 + 0.484X2 + 1.671X3 + 1.395X4 + 0.905X5 +0.075X6-0.157X7-0.312X8-0.663X9-0.767X10+0.282X11. Use self-validating,jackknife-validating and ROC curve analysis verify the discriminant model. The accuracy of self-validating of forecasting model was 83.0%, the sensitivity was72.8%, the specificity was 88.2%; the accuracy of jackknife-validating was 81.9%,the sensitivity was73.4%, the specificity was 86.2%; area under ROC curve of forecasting model was 0.810.Conclusion:1. Lifestyle behaviors are important factors in the occurrence of birth defects.Including: pregnancy diet, nutrient supplement, surrounding air pollution and pregnancy complications.2. Discriminant analysis and prediction model based on this study can predict the occurrence of birth defects.