Postpartum psychosis (PP) occurs in 1–2 per 1000 women after delivery [1]. However, considering the large number of women who experience childbirth each year, a substantial number of women suffer from this disorder. PP typically manifests in the early days or weeks following delivery with the onset of psychotic symptoms including delusional thinking, paranoia, and hallucinations; symptoms most often occur in the context of an episode of mania or depression [2]. Although uncommon, the severity of PP cannot be overstated; it is often a psychiatric emergency and requires acute attention from medical professionals with knowledge of appropriate treatment options. Women experiencing PP are at significant risk for maternal suicide, and infanticide can be a tragic complication of untreated illness [3]. Aside from mortality risk, PP is associated with increased risk of child displacement, social services involvement, and insecure infant attachment compared to healthy control groups without postpartum psychiatric illness [4]. Because PP is the index episode of a psychotic or major affective disorder for many women, it is important to identify those women who are at greatest risk for this illness [5]. Therefore, describing clinical predictors and patterns of symptom presentation is of significant clinical and public health importance.

The classification of PP is the subject of much debate, and the disorder is not recognized as a distinct diagnosis in the most recent Diagnostic and Statistical Manual (DSM) of Mental Disorders, DSM-5 [6]. While the current version of the DSM-5 allows for the diagnosis of major depressive disorder or a psychotic disorder with postpartum onset, some argue that such classification is incomplete and insufficient [7]. It has been proposed that PP be included in the DSM as a unique diagnosis based on its distinctive features (e.g., temporal onset following childbirth, cognitive disorganization or delirium in some affected patients) and that failure to appreciate such features may hinder diagnosis and treatment [7]. Furthermore, there is substantial evidence that PP may lie on the spectrum of bipolar disorders [5]. Although the presentation of PP often includes mania and/or depression, there is additional debate regarding whether PP should be classified as an episode of bipolar disorder or be considered a discrete illness [8]. For many patients without preexisting psychiatric disorders, PP is in fact the index episode of bipolar disorder, and over time a bipolar diagnosis is apparent as these patients go on to experience subsequent episodes of mania/hypomania and/or depression. A prospective study found that among women who experienced postpartum onset mania or psychosis, 32% subsequently experienced a recurrent manic, hypomanic and/or depressive episode following an index episode of PP [9]. However, for a substantial proportion of patients (43.5%), the disorder appeared to be more circumscribed, and patients did not manifest recurrent manic or psychotic illness outside of the postpartum period. This finding as well as other relevant literature suggests that while PP has significant overlap with bipolar disorder, the postpartum onset descriptor appended to a bipolar disorder diagnosis may not be sufficiently inclusive in describing PP [9, 10]. Since the long-term outcomes and risk of recurrence are still incompletely delineated, further understanding of the course of PP following the onset of an episode is of particular importance. This insight would contribute to clinical decisions regarding which patients need long-term maintenance therapy with mood stabilizers and which may be candidates to safely discontinue medications following a certain period subsequent to symptom resolution of the PP episode.

The relatively low prevalence of PP, inconsistent classification, and history of stigmatization associated with the illness makes its systematic study particularly challenging but nonetheless important. To recruit a large sample of women with histories of PP, some investigators have relied primarily on inpatient populations or populations with pre-existing diagnoses (such as bipolar disorder) for recruitment [8]. The lack of consistency in PP diagnostic criteria and naming conventions, as well as the absence of an International Classification of Diseases (ICD) code used for categorization and billing purposes, makes the conduct of system-wide analyses of patient records particularly challenging. Such heterogeneity of definitions also contributes to the challenge of collaborative research on PP. For example, some investigators include patients with isolated mania (even in the absence of psychotic symptoms) under the diagnostic umbrella of PP [8], and the definition of the postpartum window varies significantly across studies (from within 4 weeks to within 6 months) [11, 12].

In response to this identified need to more adequately describe and classify PP, research has sought to further elucidate the phenomenology, symptomatology, and clinical status of those who suffer from PP. Common clinical features identified include affective symptoms (mania and/or depression), visual or auditory hallucinations, disorganized thoughts, ideas of reference, and delusional ideas [2]. A prospective cohort study performed in the Netherlands assessed the symptoms and phenotypes of 130 inpatient participants with PP and classified them into mutually exclusive, homogeneous subgroups (i.e., depressive, manic, and atypical) [13]. Given the observed familial clustering of PP/mania, previous research has also examined genetic contributions to PP; Jones et al. [14] conducted a genome-wide linkage study in families with women with bipolar disorder or schizoaffective disorder who experienced PP and observed linkage signals on specific chromosomes, suggesting genetic risk factors. Although there have been few treatment studies, the rate of response to lithium in the available data further suggest a bipolar diathesis [15]. Risk factors for PP include a personal or family history of PP, a history of bipolar disorder or psychotic illness prior to delivery, a family history of bipolar disorder, and cessation of anti-manic medication during pregnancy [12]. Other associated factors include fluctuations in reproductive hormones [16], loss of sleep [17], environmental stressors, autoimmune dysregulation [18], and thyroid dysregulation [19–22].

Despite these known risk factors, there are still significant gaps in knowledge and challenges in identifying patients at risk for this illness. Given the ongoing debate about classification of PP, further research is needed to better understand the etiology and phenomenology of this illness. This report describes the methods of The Massachusetts General Hospital Postpartum Psychosis Project (MGHP3), which aims to address the above-mentioned gaps in knowledge by enrolling a large and geographically diverse sample of women with histories of PP. This report describes the design, rationale, and assessments used to study the sample enrolled to-date in this ongoing study.

As of July 1, 2022, 311 subjects have been enrolled in MGHP3. Data collection began in 2018 and currently has a goal of enrolling 400 patients into the index cohort. Primary recruitment sources include web presence, social media, digital advertising partnerships, the Postpartum Psychosis Project website (MGHP3.org), and Massachusetts General Hospital (MGH) Psychiatry Department physician referral. The study has recruited from 44 states, Washington D.C., and 7 different countries to-date, and the distribution of participants captures the experience of patients across various healthcare systems, from urban specialized medical centers to rural emergency care settings.

Through continued enrollment of the study sample and the pairing of genomic and clinical data collection, we aim to assess the degree to which PP risk reflects higher polygenic loading for schizophrenia or bipolar disorder versus a novel genetic signature. Because the Psychiatric Genomics Consortium [38] polygenic risk analyses currently include primarily subjects of European ancestry, we will limit genome-wide polygenic risk analyses to that population; the Psychiatric Genomics Consortium does not yet contain summary statistics for non-white groups. MGHP3 will examine a genetic underpinning for this serious disorder and address critical questions about clinical presentation, attendant morbidity, and potential mortality of PP. Furthermore, the neuroimaging arm of MGHP3 seeks to identify whether or not women who experience psychotic symptoms within 6 months of delivery show abnormalities in brain function that are similar to those previously observed in psychotic disorders, mood disorders, or both. For this arm of the study, women without history of psychiatric illness, but who have delivered a child within the past 10 years are eligible to be enrolled as a comparison group.

A significant strength of this study is its large sample size, given the paucity of systematically-collected clinical data for PP, the low prevalence of the condition, and the aim for international diversity of the sample. Previous studies of PP phenotype had a cohort size of up to 130 [13] and included relatively homogenous national samples. The effort to recruit an international sample as well as participants from a wide range of healthcare settings and access increases the generalizability of our findings. A further strength of this study is that women who have experienced psychotic symptoms are recruited both with and without bipolar disorder diagnoses. For example, some previous research on postpartum psychosis has only included women with bipolar disorder [11, 14]. Inclusion of any individual who experienced a psychotic episode within 6 months of delivery, our data provides insight into a broader range of psychosis diagnoses, including Brief Psychotic Disorder and MDD with psychotic features, and we will explore the ability to identify phenomenological differences by factors such as time of onset.

There are important limitations to this study. Although the MINI-PDS is a gold-standard for structured clinical interviews examining psychotic and other psychiatric disorders, the reliability of collected data is limited by the time elapsed since the episode and the impact of psychosis on participant recall. In order to capture a broader sample, women are eligible to participate in the study at any point within the ten years following an episode of postpartum psychosis. Given this wide window, there is significant variation in the time elapsed between onset of episode and the study interview among the sample. For example, women whose episode took place many years ago may have flawed recall of their psychiatric symptoms and medication trials. Notably, poor insight and impaired or incomplete memory of psychotic episodes are common. Therefore, it is not unusual for study participants to report difficulty remembering the exact time of onset and presentation of psychotic symptoms; on such occasions, corroborating family or medical information is sought. The challenge inherent in symptom recall from the distant past is somewhat mitigated by the real temporal landmark (proximity to childbirth), and memorable nature associated with a postpartum psychotic episode. In addition, many women are hospitalized as part of treatment, history of which further consolidates the timeline of events. Further research on the presentation of PP should include assessment more proximate to the episode, medical records as a standard data source, and prospective study of women at risk for PP who are planning a pregnancy. Another limitation of the study may derive from our decision not to exclude women with histories of intermittent postpartum marijuana use (as long as such use was not associated with emerging psychotic symptoms) while excluding women with other frank substance use disorder. This qualification derived from our observation of the frequency of marijuana use among postpartum women and an effort not to limit the generalizability of our findings by excluding such potential subjects.

Despite these limitations, by working with invested stakeholders and patient advocates, investing in an online presence, and sharing the nature of this research on social media, an innovative recruitment strategy and subsequent engagement of a large and geographically diverse sample has been possible. Rather than recruit a group of subjects who derive from an exclusively inpatient setting, we aim to capture a broad range of experiences and clinical presentations, including women who never received inpatient hospitalization or women who live in areas with rural healthcare systems. In addition to geographic diversity, we also seek to recruit a representative population across racial, ethnic, and socioeconomic groups.

We will continue to engage with maternal mental health advocates and women with lived experience of PP to further identify participants that represent this broad range of experiences. MGHP3 gathers essential data on the clinical presentation of PP as well as genetic contributions, and the ongoing rigorous phenotyping and genetic sampling will contribute to our understanding of PP’s phenomenology. Modeled after existing psychiatric genetics initiatives [38, 39], MGHP3 is the first U.S.-based initiative in reproductive neuroscience to explore the phenomenology, timing of PP episode onset, symptom expression, and underlying psychiatric comorbidities with rigorous clinical evaluation in a large study sample. While previous genetic studies of psychiatric disorders have revealed inconsistent conclusions in terms of diagnostic classification in part due to study population heterogeneity, the broad range of MGHP3 recruitment platforms allows accession of a geographically and diagnostically diverse population of participants, permitting delineation of phenotypes in understudied populations using established, validated tools. Clarification of the heterogeneity of PP at the level of clinical symptom expression to genomic characteristics to neural circuitry can lead to a richer understanding of the disease which may have clinical implications with respect to specific pharmacologic treatment received, duration of treatment, and long-term prognosis.