Depressive symptom profiles across pregnant, postpartum, and nonperipartum women: measurement invariance and differential item functioning analyses of 2005-2021 NHANES data.

BACKGROUND

The DSM-5-TR classifies peripartum depression as symptomatically indistinguishable from major depressive disorder (MDD). However, evidence suggests that depressive symptom profiles may differ among pregnant, postpartum, and nonperipartum (NPP) women. This study used 2005-2021 National Health and Nutrition Examination Survey data to evaluate Patient Health Questionnaire-9 (PHQ-9) depressive symptoms and assessed measurement equivalence across these groups.

METHODS

Measurement invariance (MI) and differential item functioning (DIF) analyses included 632 pregnant, 385 postpartum, and 6,503 NPP women (aged 15-49 years). After confirming the PHQ-9 factor structure, we evaluated MI using multigroup confirmatory factor analysis (MGCFA) and assessed DIF using ordinal logistic regression and item response theory. In MGCFA, model changes supported invariance (measurement equivalence between groups) if ΔCFI ≥ -0.010 together with ΔRMSEA ≤ 0.015 and ΔSRMR ≤ 0.030 for metric invariance and ΔCFI ≥ -0.010 together with ΔRMSEA ≤ 0.015 and ΔSRMR ≤ 0.015 for scalar and strict invariance. DIF, defined as differential item responses between groups, was statistically significant at p < 0.01 (likelihood-ratio χ2 test) and clinically significant if the pseudo-R2 change was ≥ 0.035.

RESULTS

A two-factor PHQ-9 model was used. The cognitive-affective factor included anhedonia, depressed mood, guilt/worthlessness, concentration difficulty, psychomotor changes, and suicidal ideation. The somatic factor included sleep disturbance, fatigue, and appetite changes. MGCFA supported metric (ΔCFI = -0.0005, ΔRMSEA = -0.0001, ΔSRMR = 0.0027), scalar (ΔCFI = 0.0004 and ΔSRMR = -0.0024), and strict invariance (ΔCFI and ΔSRMR < 0.0001). Statistically significant DIF emerged between pregnant and NPP women (guilt/worthlessness, psychomotor changes, fatigue, and appetite changes), postpartum and NPP women (sleep disturbance and fatigue), and pregnant and postpartum women (sleep disturbance). However, all statistically significant DIF effects were modest and of minimal clinical relevance (total DIF pseudo-R2 difference ≤ 0.012).

CONCLUSION

The PHQ-9 assesses depressive symptoms equivalently in pregnant, postpartum, and NPP women, allowing meaningful comparisons across groups. At equivalent depression severity, differences in PHQ-9 item responses across these groups are modest and of minimal clinical relevance. The absence of symptom bias does not support a distinction between peripartum depression and DSM-5-TR MDD.