Second-trimester dyadic coping and risk of third-trimester depressive symptoms: development and internal validation of a prediction model.

BACKGROUND: Depressive symptoms in late pregnancy are common and may adversely affect maternal well-being and perinatal outcomes. Current antenatal screening mainly relies on symptom scales administered at a single time point, which may be insufficient for identifying women who are likely to develop depressive symptoms later in pregnancy. Dyadic coping reflects how partners communicate, support each other, and respond jointly to stress, but its value in risk prediction for third-trimester depressive symptoms remains unclear. METHODS: This prospective observational study recruited pregnant women receiving routine antenatal care at a tertiary hospital in Shanghai, China. Recruitment lasted from January 2025 to August 2025. Baseline predictors were collected at 21-24 weeks of gestation and included demographic characteristics, pregnancy-related clinical data, laboratory indicators, and dyadic coping. Third-trimester depressive symptoms were assessed at 28 to 36 + 6 weeks of gestation using the 24-item Hamilton Depression Rating Scale. Multivariable logistic regression was used to develop a basic clinical-demographic model and a dyadic coping-enhanced model. Model performance was evaluated using discrimination, calibration, Brier score, decision curve analysis, and bootstrap internal validation. RESULTS: A total of 370 complete cases were included, of whom 86 developed third-trimester depressive symptoms. During baseline screening, 11 women were excluded solely for meeting the exclusion criterion of baseline HAMD-24 total score ≥8. Women with depressive symptoms had significantly lower dyadic coping scores than those without symptoms. In the basic model, younger age, higher mean platelet volume, gestational hypertension, intrauterine growth and amniotic fluid abnormalities, and immune and autoimmune diseases were associated with third-trimester depressive symptoms. After adding dyadic coping, the area under the curve increased from 0.739 to 0.826, and the Brier score decreased from 0.152 to 0.127. The enhanced model also showed significant improvement in integrated discrimination improvement and net reclassification improvement. After bootstrap correction, the area under the curve was 0.719 for the basic model and 0.810 for the dyadic coping-enhanced model. Decision curve analysis showed that the enhanced model provided greater net benefit across most clinically relevant threshold probabilities. CONCLUSION: Dyadic coping assessed in the second trimester provided incremental predictive value for third-trimester depressive symptoms beyond routinely available clinical information. Incorporating dyadic coping into antenatal risk assessment may provide preliminary risk screening reference for our single-center cohort to identify women requiring enhanced psychological and family support. However, multi-center external validation across diverse obstetric populations is required before this model can be routinely implemented in general clinical settings.