| Part I.Epidemic characteristics and incidence prediction of postpartum depressionObjective:The goals of this study were to:1)determine the epidemiological characteristics of postpartum depression,2)build ARIMA model and predict incidence of postpartum depression,3)explore the associated factors of postpartum depression and their interaction from the aspects of demographic characteristics and gestational development.Methods:1)A cross-sectional study was conducted at Baoan Maternal and Child Health Hospital and cluster sampling was adopted.A total of 8065 puerperal women who took a postpartum depression screening at 6 weeks after delivery in department of postpartum healthcare and meet the inclusion criteria were invited to participate in this study.We collected the information on maternal demographic characteristics,pregnancy related factors and other influencing factors via questionnaires and medical records.2)The Chinese version of the Edinburgh Postpartum Depression Scale(EPDS)was used to screen postpartum depression.3)Time series analysis was used to analyze the monthly occurrence rates of postpartum depression from January 2016 to December 2018,then the ARIMA prediction model was built.The monthly occurrence rate of postpartum depression in 2019 was predicted by the ARIMA model.The predictive effect was evaluated.The monthly incidence of postpartum depression in2020-2021 was then predicted.4)A binary logistic regression analysis was conducted to assess the risk factors for postpartum depression,and a likelihood ratio test in the logistic regression model was used to test multiplicative interaction among different influencing factors.Results:1)In total,8065 puerperal women were enrolled in this study,the median age of the participants was 29 years,the interquartile range was 27-32.The incidence of postpartum depression was 11.92%.The incidence of postpartum depression from2016 to 2019 was 12.42%,11.82%,11.11%,10.83%,respectively.Besides,the incidence of postpartum depression vary from person to person.Primipara,having a family history of mental illness,personality,living arrangement,depression during pregnancy,anxiety during pregnancy,stressful life events,low birth weight infant,and had feeding methods,attending pregnancy school,was associated with risk of postpartum depression with statistical significance.2)ARIMA(2,1,2)was built according to the parameter and model test of goodness of fit and Ljung-Box Q test,R~2=0.521,BIC=-0.117,Ljung-Box Q=17.99,P=0.207.The actual values of postpartum depression monthly occurrence rate in 2019 were in 95%CI.The monthly incidence of postpartum depression in 2020-2021 was between 8.21%and 10.66%.3)The results of multivariate logistic regression analysis showed that anxiety during pregnancy(OR=3.58;95%CI,3.04-4.22;P<0.001),depression during pregnancy(OR=4.02;95%CI,3.26-4.95;P<0.001),stressful life events(OR=1.48;95%CI,1.10-1.40;P=0.011),mixed feeding(OR=1.24;95%CI,1.10-1.40;P<0.001)and living with parents-in-law(OR=1.26;95%CI,1.08-1.46;P=0.003)were associated with a high prevalence of postpartum depression.4)The results of interaction analysis showed that there were interactions between anxiety during pregnancy and puerperal occupation,preterm birth,and negative life events(P interaction was 0.002,0.032,0.007,respectively).There was an interaction between mixed feeding with puerperal age and puerperal occupation(P interaction was 0.033 and 0.019,respectively).There was an interaction between living with parents-in-laws and puerperal personality(P interaction=0.004).Conclusions:1)Postpartum depression is still a common disease that harms maternal physical and mental health.The incidence rate was 11.92%.ARIMA(2,1,2)model can simulate and predicate the affection of the occurrence trend of postpartum depression.The model could conduct dynamic analysis and short-term prediction.2)Postpartum depression is influenced by many factors,including depression during pregnancy,anxiety during pregnancy,stressful life events,mixed feeding,living with parents-in-laws.Besides,these factors are affected by other factors.Therefore,the development of postpartum depression intervention strategies should focus on depression and anxiety during pregnancy,the relationship with family members,especially the relationship with parents-in-laws,and encourage breastfeeding.Part II.Analysis of influencing factors of postpartum depression based on BP neural network modelObjective:1)A combination of epidemiology,social psychology and clinical medicine to further enrich and verify the influencing factors of postpartum depression from the perspective of individual,family,hospital and community.2)To build BP neural network model and obtain the sensitivity of each variable,and to further evaluate the influence of each factor on postpartum depression.Methods:1)A case-control study was conducted at Baoan Maternal and Child Health Hospital.According to the inclusion and exclusion criteria,120 cases were included in the depression group and 240 controls were included in the control group.2)The Chinese version of the Edinburgh Postpartum Depression Scale(EPDS)was used to screen postpartum depression.Self-made questionnaire,Eysenck Personality Questionnaire(EPQ)and Social support rating scale(SSRS)were used to collect information about general features,characteristics of pregnant and puerperal period,family relationship,family support,social support,and hospital support.3)The influencing factors of postpartum depression were analyzed by univariate and multivariate Logistic regression analysis methods.4)BP neural network model was built to obtain the sensitivity of each variable,and then to evaluate the influence of each factor on postpartum depression.Results:1)The results of multivariate logistic regression analysis showed that history of adverse pregnancy outcomes(OR=5.77;95%CI,1.44-23.15;P=0.013),multipara(OR=0.42;95%CI,0.19-0.94;P=0.036),dysmenorrhea(OR=2.51;95%CI,1.21-5.20;P=0.013),depression during pregnancy(OR=6.87;95%CI,1.87-25.27;P=0.004),bad relationship with parents-in-law(OR=2.62;95%CI,1.24-5.52;P=0.011),having a low birth weight newborn(OR=19.19,95%CI,2.98-123.69;P=0.002)and psychology guidance of community doctor(OR=0.17;95%CI,0.08-0.40;P<0.001)were the influencing factors of postpartum depression.Multipara,psychology guidance of community doctor was associated with the reduced risks of postpartum depression.2)The training time of BP neural network model was 0.27s,the cross-entropy error of training samples was 91.783,and the correct prediction percentage was 87.4%.The cross-entropy error value of the test sample was 40.493,the correct prediction percentage was 82.2%,and the AUC was 0.898.When the importance of independent variables in the model was standardized,the top five factors affecting the occurrence of postpartum depression were depression during pregnancy,history of adverse pregnancy outcomes,having a low birth weight newborn,the relationship with spouse,and the relationship with parents,respectively.Conclusions:Our study explored the influencing factors of postpartum depression from the perspectives of individual,family,hospital,and community.The BP neural network model constructed in this study could predict postpartum depression well and be used to analyze the influencing factors of postpartum depression. |
PDF Full Text Request