Research On Risk Assessment And Preventive Control Strategies Of Adverse Pregnancy Outcomes Among Women In Late Pregnancy

After the implementation of the Two-child Policy in China,the proportion of pregnant women with advanced age has increased substantially.Accordingly,the incidence of pregnancy complications,birth defects and other adverse pregnancy outcomes also increased.The occurrence of adverse pregnancy outcomes not only brought serious damage to the health of pregnant women and newborns,but also imposed heavy burden on family economic conditions and health expenditure,and even had a serious adverse impact on the improvement of the quality of the entire social population and social harmony and stability.The organ function load of pregnant women in late pregnancy was approaching the highest value gradually,and they were facing various problems related to childbirth,so they were more likely to suffer from health damage.The health status of women during this period was closely related to pregnancy complications,delivery mode and birth outcome.Therefore,exploring the risk factors of adverse pregnancy outcomes and assessing the risk for adverse pregnancy outcomes in late pregnancy are of great significance to reduce the occurrence of adverse pregnancy outcomes and ensure maternal and child health.ObjectiveBased on the theory of risk management,following the research approach risk identification-risk measurement-risk assessment and prediction-risk management,this study used a variety of models and algorithms to comprehensively analyze the risk factors that may lead to adverse pregnancy outcomes from the aspects of physical status,psychological status,pregnancy and childbirth related factors,behavioral lifestyle and social interaction.Establishing risk assessment model for adverse pregnancy outcomes and evaluating the predictive effect of the model were also our aim.Then this study elaborated the possible risk diffusion caused by adverse pregnancy outcomes based on the risk diffusion theory.Finally,according to the results and problems we found in this study,we formulated practical and targeted prevention and intervention strategies to provide theoretical support for the prospective prevention and reduction of adverse pregnancy outcomes.MethodsThis study was conducted among pregnant women who underwent maternity examination in a hospital of Jilin Province from May to September,2017 by using convenient sampling method.A questionnaire included maternity,behavior,social interaction,sleep quality,stress and et al.We tracked the pregnancy outcomes of the respondents by consulting the electronic medical records.Epidata 3.0 was used to establish the database.SPSS 24.0 and R Stutio were used to analyze the data.Chi-square test,two-independent sample t test and one-way ANOVA were used to preliminarily explore the relations between variables and negative birth outcome,pregnancy complications and negative pregnancy outcome.Binary Logistic regression model,BP neural network model and random forest algorithm were used to analyze the factors may cause adverse pregnancy outcomes.Risk scoring algorithm was used to establish risk assessment model,and the risk score of adverse pregnancy outcomes was also calculated.The working characteristic curve of the subjects was drawn according to the risk score of the participants,and we divided participants into high-risk group and the low-risk group according to the highest score point of Youden index.Paired Chi-square test was used to compare the difference between the predicted risk group and the actual risk group.Results(1)Among the 966 women,518(53.6%)pregnant women had adverse pregnancy outcomes.The incidence of adverse birth outcomes was 8.4%,and the incidence of pregnancy complications was 51.9%.Premature birth was the main adverse birth outcome,while gestational diabetes mellitus was the main complication.(2)The results of binary Logistic regression showed that the parity,BMI index,weight gain,delivery mode,passive smoking,vitamin calcium supplementation,history of illness,marital relationship,social support,diastolic blood pressure and amniotic fluid index were independent risk factors for adverse pregnancy outcomes(P<0.05).In the BP neural network model,the variables influencing adverse pregnancy outcomes according to the importance were diastolic blood pressure,pre-pregnancy weight,social support,weight gain,delivery mode,abdominal circumference,history of illness,marital relationship,pregnant school,intake of vegetables and fruits,vitamin calcium supplement,anxiety and et al.Randomized forest model ranked the average reduction of Gini index from large to small,representing the importance of variables from high to low.Their ranks as followed,diastolic blood pressure,systolic blood pressure,social support,amniotic fluid index,BMI index,pre-pregnancy weight,anxiety,weight gain and et al.(3)According to the prediction effect,the accuracy,specificity and negative predictive value of random forest algorithm were the highest.The sensitivity of Logistic regression model was better than random forest algorithm,and its positive predictive value was equal to random forest algorithm.BP neural network model did not perform well in this test.(4)According to the risk score model established based on logistic regression results,the area under ROC curve was 0.771.Considering the risk score of the best critical point as the cut-off point,the high-risk group and the low-risk group were determined.Paired Chi-square test was used to analyze the differences between predictive condition and actual adverse pregnancy outcome.Mc Nemar test P value was 0.188,so we considered that the risk assessment model had better prediction effect.(5)Through the process of identifying risk sources,risk points,risk diffusion direction and forming risk diffusion mechanism,we elaborated the risk diffusion mechanism for adverse pregnancy outcomes.It was complex,including one-way direct diffusion,two-ways risk diffusion and even multi-ways risk diffusion.To reduce the formation of follow-up new risk points and the severity of risk diffusion,adopting preventive interventions at different stages of the path have better cost-effectiveness.Conclusion(1)Pregnant women with high BMI index,history of illness,low weight gain during pregnancy,poor marital relationship,low perceived social support,passive smoking,high diastolic blood pressure and low amniotic fluid index had a higher risk for adverse pregnancy outcomes.(2)Adverse birth outcome and pregnancy complications have common related factors,and they also have their own related factors respectively.(3)Logistic regression model had the highest sensitivity in predicting adverse pregnancy outcomes in this study.Different models had their own advantages and disadvantages.In the practice,we should combine the results of various models to analyze comprehensively.(4)The risk score model of adverse pregnancy outcomes based on logistic regression model could correctly predict the occurrence of adverse pregnancy outcomes among women in late pregnancy,but the ability of discrimination and prediction was still limited.It provided references for clinical risk assessment,but it needed further verification and improvement in application.(5)To reduce the incidence of adverse pregnancy outcomes and risk diffusion,enhancing prenatal care,closely monitoring weight variation,creating support network,giving psychological counseling,and establishing pregnancy risk prevention system,risk early warning and critical rescue system were of great significance.