Buzzapp Master 
LOS ANGELES — A machine-learning model may be able to predict the risk of postpartum depression among new mothers, according to a retrospective cohort study.
Using electronic health record (EHR) data routinely collected before discharge after delivery, the model had good discrimination in the external validation data, with an area under the receiver operating characteristic curve of 0.721 (95% CI 0.709-0.736), reported Mark Clapp, MD, MPH, of Massachusetts General Hospital and Harvard Medical School in Boston, at the American Psychiatric Association annual meeting.
At a specificity of 90%, the positive predictive value was 28.8%, and the negative predictive value was 92.2%. The Brier calibration score was 0.087 (95% CI 0.083-0.091).
“By identifying groups of individuals at greater risk, implementing these models may allow the application of interventions aimed at prevention or more intensive screening that may otherwise be unnecessary or infeasible in the entire clinical population,” the authors wrote in the American Journal of Psychiatry, where the study was published.
“This tool could help identify patients within a practice at the highest risk and facilitate individualized postpartum care planning for the prevention of, screening for, and management of postpartum depression at the start of the postpartum period and potentially the onset of symptoms,” they concluded.
In the 6 months following delivery, about 2,700 women were diagnosed with postpartum depression, about a 9% rate, Clapp said.
“We are close to moving this model into clinical practice,” Clapp told MedPage Today at a press briefing. “It is easy to make models, but it is a lot harder getting them into clinical practice at bedside. We are now in the process of doing that conversion.”
Postpartum depression is common, affecting about 15% of recently pregnant women, and is a major contributor to morbidity and mortality following pregnancy. It is linked to an increased risk for suicide and self-harm and is estimated to be responsible for 10% or more of all pregnancy-related deaths, the authors explained.
“The next steps of this work involve translating this model into clinical practice and studying how it can be used effectively and appropriately by patients and clinicians to reduce the incidence, severity, and subsequent consequences of postpartum depression,” they wrote.
Commenting on the study, Misty Richards, MD, of the University of California Los Angeles, said, “What we try to do in our preventive clinic is to catch people with postpartum depression before it becomes a forest fire. Yet, oftentimes we miss it. If we could have predictive tools like this program, especially with people who have no history of depression, this would be incredibly important.”
“That is the big takeaway here: Those folks without a history of depression do not tend to be diagnosed. So this machine-learning system is very important clinically, so that we can reduce that,” she added.
To develop the machine-learning model, Clapp and colleagues included all live births that occurred from 2017 to 2022 at two academic medical centers, six community-based hospitals, and their affiliated outpatient clinics, which all shared an EHR system. The model only included women who had received prenatal care at the facilities, and the study excluded patients who had a history of psychiatric illnesses.
Predictors used in the model included sociodemographic factors, medical history, and prenatal depression screening information.
Postpartum depression was determined by hospital codes indicating a mood disorder, a prescription for an antidepressant medication, or a positive screen on the Edinburgh Postnatal Depression Scale (EPDS).
The median age at delivery was 33 years, 70% were white, 13% were Asian, 7% were Black, and 11% were Hispanic. The median prenatal EPDS was 3, and the median length of hospital stay was about 3 days.
Disclosures
The study was supported by grants from the National Institute of Mental Health, the National Institute of Child Health and Human Development, and the Simons Foundation.
Clapp reported serving on the scientific advisory board of and holding equity in Delfina Care; receiving research support from grants to his institution from the Agency for Healthcare Research and Quality; and receiving a stipend for editorial services from the American College of Obstetricians and Gynecologists.
Richards disclosed no relevant relationships with industry.
Primary Source
American Journal of Psychiatry
Source Reference: Clapp MA, et al “Stratifying risk for postpartum depression at time of hospital discharge” Am J Psychiatry 2025; DOI: 10.1176/appi.ajp.20240381.
GIPHY App Key not set. Please check settings