In the USA, a pregnant individual will attend roughly 15 prenatal visits with a medical provider for the monitoring of an uncomplicated pregnancy. During these visits, an unlimited amount of demographic and clinical information can be collected and entered into the electronic health record (EHR). Much of the data is said to monitoring the pregnancy–comparable to weight gain, blood pressure, urinalysis; nonetheless, there may be information within the medical record that might be used to predict risk for perinatal depression. Several recent studies have sifted through this information gleaned from the electronic health record, using machine learning to generate algorithms that might be used to estimate risk for postpartum depression.
Estimation of Risk of PPD in First-Time Moms
In a retrospective cohort study analyzing data from the NIH Nulliparous Pregnancy Outcomes Study: Monitoring Moms-to-Be, Wakefield and colleagues examined the medical records of 10,038 first-time moms. They tested the performance of 4 different models for predicting risk:
- Model 1 utilized only readily obtainable sociodemographic data.
- Model 2 also included data on maternal mental health prior to pregnancy.
- Model 3 utilized recursive feature elimination to construct a parsimonious model.
- Model 4 further titrated the input data to simplify prepregnancy mental health variables.
The evaluation included 8,454 births; 338 (4%) received treatment for depression (as documented within the EHR) throughout the postpartum period. When it comes to predicting women who would later require treatment, model 3 performed one of the best, with an area under the receiver operating characteristics curve of 0.91 (±0.02). Which means this model would discover 91% of ladies who ultimately require treatment for depression.
The models identified nine variables that were probably the most robust predictors of postpartum depression treatment: maternal history of depression (highest), any current mental health condition, recent psychiatric medication use, BMI, income, age, history of tension, education level, and preparedness for pregnancy (lowest).
Estimation of Risk for PPD with Model Including EPDS Scores
In one other study, Amit and colleagues analyzed EHR data from 266,544 women in the UK who gave birth to their first child between 2000 and 2017. A subset of 5959 women also had Edinburgh Postnatal Depression Scores (EPDS) scores recorded within the EHR. The researchers extracted multiple socio-demographic and medical variables and constructed a machine learning model to predict the chance of PPD throughout the 12 months following childbirth.
On this cohort, the prevalence of PPD was 13.4%. PPD was defined based on the occurrence of considered one of the next documented within the EHR throughout the first 12 months postpartum: (1) diagnosis of depression; (2) latest treatment with antidepressant; or (3) non-pharmacological treatment for depression.
In a model using only data derived from the EHR, the world under the curve (AUC) of the prediction model ranged from 0.72 to 0.74. Interestingly, the model worked fairly well when only prepregnancy data was used to predict risk; the EHR-based prediction model administered before pregnancy identified not less than 70% of ladies who were later diagnosed with PPD.
When the model combined EHR-based data with EPDS scores, the world under the receiver operator characteristics curve (AUC) increased from 0.805 to 0.844, with a sensitivity of 0.76 at a specificity of 0.80. In other words, one of the best predictive model could discover not less than 80% of ladies who would later be diagnosed with PPD.
The aspects most strongly related to risk of PPD included history of antidepressant use, history of depression, variety of antidepressant prescriptions filled, younger age, BMI, smoking status, and deprivation index.
Can We Use the Electronic Medical Record to Predict Risk for PPD?
The reply is YES. Each studies indicate that machine learning could be used to construct a model using data collected from the EHR that could be used to predict risk for depression inside the first 12 months after childbirth. In these two studies, the predictive models were in a position to discover between 84% and 91% of ladies in danger for developing depression after delivery. Statisticians get pretty enthusiastic about screening tools when the AUC is larger than 0.8.
Now for the caveats. Each of those studies document the diagnosis or treatment of PPD using documentation within the electronic health record: either documentation of the diagnosis itself or treatment (antidepressant or non-pharmacologic treatment). While the prevalence of PPD within the Wakefield study carried out within the US was 4%, the Amit study from the UK reported that the prevalence of PPD was 13.4%.
Based on previous epidemiologic studies, the prevalence of PPD is often around 15%. There are not any studies indicating that the prevalence of PPD is lower within the US than within the UK; it ought to be noted that the US study is looking only on the prevalence of treatment for PPD, whereas the UK study is the prevalence of diagnosis and/or treatment. This discrepancy is consistent with previous studies indicating low rates of treatment amongst women with PPD where lower rates of treatment reflect underdiagnosis of perinatal mood and anxiety disorders, in addition to barriers to accessing treatment within the US.
The models described in these studies are probably identifying individuals in danger for more severe PPD and never women with less severe depressive symptoms. While it is important to discover women with probably the most severe symptoms with the intention to limit morbidity in each the mother and the kid, we could also be missing a chance to support other moms who’re also struggling throughout the postpartum period. Further studies are needed to check these predictive models in additional diverse populations, including multiparous moms, and using a broader definition of perinatal depression.
Ideally we would love to have the opportunity to discover women in danger for postpartum depression before it occurs. This could not only allow us to extend monitoring when needed and to treat early if PPD emerges, nevertheless it might also provide a chance to initiate preventative interventions. Currently our strongest predictors of risk include a history of depression prior to pregnancy and depressive symptoms while pregnant. These models construct on these robust risk aspects, and include other risk aspects (i.e., age, BMI) to enhance our ability to predict and quantify risk.
Ruta Nonacs, MD PhD
Wakefield C, Frasch MG. Predicting Patients Requiring Treatment for Depression within the Postpartum Period Using Common Electronic Medical Record Data Available Antepartum. AJPM Focus. 2023 Apr 27;2(3):100100.
Amit G, Girshovitz I, Marcus K, Zhang Y, Pathak J, Bar V, Akiva P. Estimation of postpartum depression risk from electronic health records using machine learning. BMC Pregnancy Childbirth. 2021 Sep 17;21(1):630.