Predicting postpartum depression risk with use of predelivery psychological assessments and quality of recovery tools: A prospective observational study
Ines Debbiche, Feyce M. Peralta, Lynn M. Yee, Kaitlyn Neumann, Sarah M. Cocoma, Robert J. McCarthyAbstract
Introduction
Postpartum depression (PPD) is a common and serious complication of childbirth that affects maternal functioning and disrupts mother–infant bonding. Validated psychological and Obstetric Quality of Recovery 10 (ObsQoR‐10) assessments have been used to screen for risk of PPD, but not in combination. The aim of this study was to determine if predelivery validated measures of anxiety and depression in combination with ObsQoR‐10 improved early screening of PPD.
Methods
A prospective, observational study was conducted between June 2023 and October 2024. Participants admitted for a trial of labor completed the Patient Health Questionnaire 4 (PHQ‐4), Generalized Anxiety Disorder 7 (GAD‐7), and the Pain Catastrophizing Scale (PCS‐13). Prior to discharge, participants completed the ObsQoR‐10 questionnaire. The PHQ‐4 and ObsQoR‐10 were repeated at 1 week postpartum. The Edinburgh Postnatal Depression Scale (EPDS) was completed at 6 weeks postpartum. An EPDS score ≥ 13 was the primary outcome and was modeled using 18 patient, obstetric and patient self‐reported factors collected shortly before and after delivery. A parsimonious model determined by minimization of the Akaike information criteria was compared to the full model. Net benefits of the models were evaluated and a schema for early risk identification on the day of discharge and 1 week postpartum was developed using recursive partitioning.
Results
A total of 603 participants were analyzed with 46 (7.6%) reporting a 6‐week EPDS ≥13. The model AUC was not different between the full (0.787) and parsimonious model (0.754), difference 0.033 (95% confidence interval [CI], −0.014 to 0.081; p = 0.167); however, model complexity was significantly reduced. Variable importance of factors from the full and parsimonious models identified the ObsQoR‐10 and the GAD‐7 among the most important variables. The variables in the parsimonious model included the GAD‐7, ObsQoR‐10, and duration of hospital stay. Similar net benefit of the full and parsimonious models for detection of an EPDS ≥ 13 was observed above threshold probabilities of 5% and superior to the treat all or none conditions. Recursive partition using the parsimonious model identified 16 of 36 (44%) women at hospital discharge as high risk for an EPDS ≥ 13 in terminal branches at a threshold risk > 15%. An additional 15 of 67 (22%) were identified using the ObsQoR‐10 and PHQ‐4 obtained at 1 week postpartum.
Discussion
Predelivery anxiety and depression assessments combined with quality of obstetric recovery measures, such as the ObsQoR‐10 score, at discharge and at 1 week postpartum can improve early identification of high‐risk patients for an EPDS ≥ 13 at 6 weeks postpartum. Implementation into clinical practice of these findings could potentially lead to earlier PPD screening and treatment of this vulnerable population.