Prediction model for postpartum depression based on social psychological factors: Establishment and evaluation fulltext
10.16781/j.0258-879x.2017.04.0476
- Author:
Fei-Ya CAI
1
Author Information
1. Department of Psychiatry, University-Town Hospital Affiliated to Chongqing Medical University
- Publication Type:Journal Article
- Keywords:
Incidence;
Postpartum depression;
Predictive model;
Social factors psychological factors
- From:
Academic Journal of Second Military Medical University
2017;38(4):476-481
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the social psychological factors influencing postpartum depression (PPD), and to explore its predictive value for PPD, so as to establish and evaluate the prediction model of PPD. Methods We randomly selected 371 pregnant women (pregnancy≥28 weeks), who received antenatal examination and would be hospitalized for delivery in the First Hospital, Jinshan Hospital, and University-Town Hospital Affiliated to Chongqing Medical University from Sep. 2014 to Jun. 2016. The social demographic questionnaire (self-define), the Eysenck Personality Questionnaire (EPQ), the Symptom Checklist-90 (SCL-90), the Hamilton Anxiety Scale/Hamilton Depression Rating Scale (HAMA14/HAMD24), and the Beck Anxiety Inventory/Beck Depression Inventory (BAI/BDI) were used to assess the prenatal social psychological factors of pregnant women, and the Edinburgh postnatal depression scale (EPDS) was completed on the 42nd day after delivery. Then the evaluation of PPD was conducted to analyze and predict the social psychological factors influencing PPD. Multivariate stepwise logistic regression analysis was used to establish the prediction model, and receiver operating characteristic (ROC) curve was used to evaluate the predictive value of the model. Results Sixty of the 371 cases met the diagnostic criteria of PPD (PPD group) according to the assessment results of PPD, while 311 cases did not meet (N-PPD group); the incidence of PPD was 16.17% (60/371). Univariate analysis results showed that there were significant differences in having a fixed work or not, the degree of education, emotional stability, HAMA14 score, HAMD24 score, BAI score, BDI score, and each single factor score (except somatization) of the SCL-90 between the two groups (all P<0.05). The above indexes were selected for multivariate logistic regression analysis, and the following mathematical prediction model was established: Logit (PI)=0.042×total score of SCL-90+1.005×having a fixed work or not+2.498×relative company+0.108×BDI score-1.319×the degree of education-8.028, with the area under the ROC curve being 0.833 (P<0.001, 95%CI: 0.772-0.894). When PI=0.141 was selected as cut-off point, the Youden index of the model was maximum, the sensitivity was 0.900 and the specificity was 0.533. Conclusion The total score of SCL-90, having a fixed work or not, family company and BDI score are independent risk factors of PPD, while low degree of education is an independent protective factor. The prediction model for PPD based on social psychological factors has been successfully established, which has a great predictive value and is worthy of further study.
