Abstract
INTRODUCTION: Postpartum depression (PPD) is a prevalent mental health issue that poses significant challenges to maternal wellbeing and infant development. We aimed to determine the prevalence of PPD and to investigate its associated determinants and predictors in in Shenyang, China. METHODS: This cross-sectional study, conducted between January and December 2021, included 1,065 postpartum women in Shenyang, China, at 6 weeks postpartum. PPD was screened using the Edinburgh Postnatal Depression Scale (EPDS, score ≥ 9). Key risk factors were identified through machine learning techniques, including LASSO regression and the Boruta algorithm, and their associations were evaluated using logistic regression. Non-linear relationships for continuous variables were examined using restricted cubic splines and threshold effect analysis. Feature importance was ranked via a random forest model based on the change in ROC-AUC after predictor removal. Statistical significance was defined as p < 0.05. RESULTS: A total of 1,065 postpartum mothers were included in this study, of whom 23.57% were identified as having postpartum depressive symptoms. Significant risk factors included prenatal anxiety (OR = 7.16, 95% CI: 4.67-11.11), poor sleep quality (OR = 5.30, 95% CI: 3.45-8.20), poor marital relationships (OR = 4.41, 95% CI: 2.47-7.99), poor in-law relationships (OR = 4.89, 95% CI: 3.11-7.74), unplanned pregnancy (OR = 2.92, 95% CI: 2.01-4.27), and lack of prenatal education (OR = 1.7, 95% CI: 1.15-2.52). A non-linear relationship was observed with gestational weight gain: gains <11 kg were associated with reduced risk (OR = 0.91, 95% CI: 0.83-0.99), whereas gains >11 kg increased risk (OR = 1.06, 95% CI: 1.02-1.10). Other factors included smoking history (OR = 1.8, 95% CI: 1.08-2.98) and expected fetal sex (OR = 3.39, 95% CI: 2.02-5.70). CONCLUSION: PPD affects a significant proportion of postpartum women in Shenyang, China. Routine screening and targeted interventions are essential to identify and support high-risk individuals with PPD.