Suicidality in Postpartum Women With Unipolar and Bipolar Depression: A Secondary Analysis Comparing Self-Reported...

Objective: To investigate the alignment of self-harm ideation ratings with clinical assessments of suicidality in postpartum women diagnosed with unipolar and bipolar depression and the impact of trauma and psychiatric diagnosis on this alignment.

Methods: Data from the largest postpartum depression screening study (n=10,000) in the US were examined in this secondary analysis. Inclusion criteria were a positive depression screen (Edinburgh Postnatal Depression Scale [EPDS] ≥10), a psychiatric diagnosis (Structured Clinical Interview for DSM-IV), and a suicidality assessment derived from the Structured Interview Guide for the Hamilton Depression Rating Scale with Atypical Depression Supplement (SIGH-ADS). Trauma exposure, including both childhood and adult physical and sexual abuse, was measured using 4 yes/no questions from the Dissociative Disorders Interview Schedule. Associations between key variables were examined using independent samples t tests, analysis of variance, χ² tests, or Fisher exact tests. Nonparametric tests were used for skewed continuous data. To assess the consistency between the EPDS and SIGH-ADS scales, Cohen κ statistics were used, with weighted κ applied to severity ratings and simple κ for binary categorizations.

Results: Among 1,155 screen-positive postpartum women (68% White, 25.5% African American, 6.6% other; mean age 27.93 years), 21% endorsed self-harm ideation and 10.1% reported suicidality. Compared to those with unipolar depression, women with bipolar disorder had more than twice the odds of suicidality (odds ratio [OR] 2.77, 95% CI, 1.86 to 4.13, P<.001) and nearly 4 times the odds (OR 3.92, 95% CI, 1.18 to 13.00, P<.001) of not self-reporting self-harm ideation. Overall concordance between self-report (EPDS10) and clinical evaluation (SIGH-ADS11) was 78.6% (κ=0.28, 95% CI, 0.21 to 0.34, fair agreement) but varied significantly by diagnosis (P<.001), with lower concordance in the bipolar group (67.3%; κ=0.21) compared to the unipolar group (80.4%; κ=0.31). In the high-risk bipolar disorder group, concordance was no longer statistically significant, indicating poor alignment between self-report and clinical evaluation for these patients. Trauma was strongly associated with suicidality and a bipolar diagnosis.

Conclusion: The EPDS does not consistently detect suicidality in perinatal bipolar patients, with our study showing only slight and nonsignificant agreement with clinical assessment in this high-risk group. Given that risk can change quickly in postpartum bipolar patients, timely and frequent clinical assessments are needed to identify high-risk individuals. Tracking and integrating routine bipolar disorder screening and trauma assessments in perinatal care may enhance early identification of suicide risk and improve maternal mental health outcomes.