Longitudinal symptom network analysis of perinatal depression in 5002 women: A dynamic perspective from pregnancy to...

This study used network analysis to investigate temporal dynamics between specific depressive symptoms across pregnancy and postpartum periods.

A longitudinal study was conducted at Shenzhen Baoan Women's and Children's Hospital (China) from January 2020 to January 2024. The self-reported questionnaire used to evaluate perinatal depressive symptoms was the Edinburgh Postnatal Depression Scale (EPDS). We employed a multilevel mixed-effects model (MLM) to develop a predictive model and explored the associations and dynamics of perinatal depressive symptoms through network analysis. Network analysis was applied to identify central symptoms (via node strength) and directional relationships (via cross-lagged panel modeling).

A total of 5002 participants (mean [SD] age, 30.74 [3.59] years) were included. The proportion of individuals screening positive for depressive symptoms decreased from 18.03 % in the first trimester to 1.70 % in the postnatal period. The MLM showed that women with higher education, income, marital satisfaction, older age, and higher parity reported fewer depressive symptoms. Network analysis showed that specific symptoms-particularly anxiety or worries (EPDS4), fear or panic (EPDS5), and difficulty sleeping (EPDS7)-became more influential as pregnancy advanced. Unnecessary blame of themselves (EPDS3) and sad or miserable (EPDS8) predicted the temporal emergence of anxiety, indicating key symptom pathways.

Findings may lack generalizability due to a single-hospital sample in a highly urbanized region. Excluding non-completers introduces bias, potentially underreporting true screening positive rate and complexity.

This longitudinal study identified key risk factors and dynamic changes in perinatal depression, providing valuable insights for developing effective clinical intervention strategies.