Introduction

The perinatal period has been identified as a period of increased vulnerability to the onset or exacerbation of obsessive–compulsive disorder (OCD), particularly among women (Russell et al. 2013). Perinatal obsessions and/or compulsions often reflect an anxious preoccupation with the safety or wellbeing of the foetus or infant, or with one’s responsibility to prevent harm as a primary caregiver (Abramowitz et al. 2003; McGuinness et al. 2011). For example, individuals with perinatal OCD (pnOCD) often experience intrusive and distressing thoughts (‘obsessions’) about the foetus/infant being harmed in some way, or about being unable to care appropriately for their baby (Fairbrother & Abramowitz 2016). Certain obsessive–compulsive symptoms (OCS), including harming and sexual obsessions appear to be more common when OCD occurs in the postpartum period, compared with OCD in pregnancy or other life periods (Starcevic et al. 2020). Other common symptoms of OCD in the general population, including washing/cleaning compulsions, appear to be less frequent in the postpartum. Studies indicate that between 2.2% and 16.9% of women experience OCD in the postpartum (Fairbrother et al. 2016, 2021a, b; Fawcett et al. 2019; Osnes et al. 2019; Russell et al. 2013), with the higher prevalence rates reported in more recent studies most likely being attributable to broadening of the criteria for OCD in the Diagnostic and Statistical Manual of Mental Disorders: Fifth Edition (American Psychiatric Association, 2013; Fairbrother and Collardeau, 2021).

The increased prevalence, distinctive symptom presentation, and potentially unique impacts associated with OCD (McGuinness et al. 2011; Russell et al. 2013) in the perinatal period indicates the need for specific guidelines to inform best-practice in working with new/expecting parents with OCD. Comprehensive clinical guidelines for the assessment, treatment and management of OCD have been published (American Psychiatric Association, 2007a,b; National Institute for Health and Care Excellence [NICE], 2005), but lack specific guidance for OCD in the pre- and postnatal periods (Fairbrother & Abramowitz 2016; McGuinness et al. 2011; Russell et al. 2013). The American Psychiatric Association (APA) guidelines on OCD references treatment considerations during pregnancy and breastfeeding but focuses on pharmacotherapy (e.g., medication risks and side effects) rather than psychological considerations (American Psychiatric Association, 2007a, b).

Several broader guidelines exist for psychological considerations and/or treatment during the perinatal period, such as those from beyondblue (2011), the Centre of Perinatal Excellence (Austin and Highet 2017), and the National Institute for Health and Clinical Excellence (National Institute for Health and Care Excellence 2014Such guidelines are valuable for a general understanding of mental health care in the perinatal period (Austin and Highet 2017; beyondblue 2011; National Institute for Health and Care Excellence 2014); however, guidance on OCD is typically subsumed within anxiety disorder recommendations. Although OCD has been classified as an anxiety disorder (Stein et al. 2010), there is growing recognition that OCD has distinctive features, requires unique diagnostic (American Psychiatric Association, 2013) and treatment perspectives (McKay et al. 2015; Stein et al. 2010), and thus warrants specific guidance. The lack of development of pnOCD-specific clinical guidelines reflects a dearth of empirical research on the treatment and management of the disorder. Nonetheless, a small number of studies have provided evidence supporting pharmacological (Misri et al. 2004; Misri & Milis 2004; Sharma 2018) and psychological treatment for pnOCD (Challacombe et al. 2017; Challacombe & Salkovskis 2011; Christian & Storch 2009; Gershkovich 2003; Misri et al. 2004; Puryear & Treece 2017).

The current study aims to address this gap in the clinical literature on perinatal OCD, by systematically developing clinical practice recommendations for assessing, managing, and treating pnOCD, and supporting individuals with this disorder and their families. We used a Delphi survey methodology to collate the views of professionals with clinical and/or research expertise in, as well as consumers with lived experience of, pnOCD. We intend that the recommendations resulting from this study may be used to inform best-practice clinical care for individuals with pnOCD, and to increase health practitioners’ understanding of this disorder and its treatment.

Method

Research design

The current study used a Delphi survey methodology to identify best practice recommendations for the assessment and treatment of individuals with pnOCD. The Delphi technique is widely used in health research to systematically ascertain a reliable consensus from a group of experts to inform service planning and practice (Jorm 2015). It is done via a series of survey rounds in which each member of the expert panel individually rates the extent to which they consider recommendation statements to be important. The current study utilised a modified Delphi methodology as described by Keeney et al. (2010) and comprised two panels; consumers as experts in their lived experience with pnOCD (i.e., the Consumer panel), and clinicians (i.e., psychologists, psychiatrists, researchers) specialising in pnOCD (i.e., the Professional panel). Ratings were made over three successive survey rounds, in accordance with endorsement criteria detailed in Table 1. During the first survey round, panellists were given the opportunity to suggest statements to be added or amended for the next round. In each subsequent round, participants were provided with a summary of how both panels responded in the previous round to each statement being re-rated, as well as how they, individually, responded to the statements.

Table 1 Statement endorsement criteria over 3 review rounds
Full size table

Participants

Recruitment of the Consumer panel occurred by contacting relevant consumer advocacy/support services in Australia, United Kingdom, and the United States of America. Consumer panellists (n = 18) were English-speaking females over 18 years of age, who reported living with, or having previously experienced, OCD during the perinatal period, diagnosed by a mental health professional. Consumers were deemed ineligible to participate in the study if, based on a pre-study measure of OCD severity, they reported ‘severe’ current symptoms on a measure of OCD severity (see ‘Measures’ section below). This was due to concerns their severity may impact insight and/or that reflecting on their care needs while experiencing severe symptoms could add to their distress. Potential participants for the Professionals panel (n = 20) were identified via relevant professional organisations (e.g., the International Marcé Society), perinatal mental health LISTSERVs, provider registers, authors of relevant pnOCD publications found via a literature review, and via the professional networks of the study researchers. Participants were invited to be a part of the Professionals panel if they had at least five years of relevant experience working with individuals with perinatal mental health disorders, including assessing and treating pnOCD, and/or had published peer-reviewed academic publications on pnOCD. A five-year minimum was set to ensure that professionals had adequate experience and expertise with pnOCD.

For homogeneous Delphi samples it is recommended that a minimum of 10–15 participants be used to ensure reliability of the survey results (Keeney et al. 2010). A total of 14 Consumer panellists and 15 Professional panellists completed the first survey round and, thus, were included in the final study sample. The retention of participants who responded to the first round ensured that the consensus reached in later rounds was not biased by participants dropping out of either panel (Keeney et al. 2010).

Materials

Consumer symptom severity screening. The 10-item Yale Brown Obsessive–Compulsive Scale – Self Report version (YBOCS-SR; Fineberg 2001) was used to screen potential members of the Consumer panel for OCD symptom severity prior to the survey rounds. Previous studies have found the YBOCS-SR to have strong internal consistency, test–retest reliability, and construct validity (Storch et al. 2011). A total summed score of < 24, corresponding with ‘mild’ or ‘moderate’ symptoms, was the cut-off for inclusion in the study.

Initial statement formation. Prospective guideline statements were created via review of the existing literature. Specific search terms included: perinatal obsessive–compulsive disorder, postpartum obsessive–compulsive disorder, perinatal OCD, postpartum OCD, postnatal OCD, pregnancy OCD, maternal OCD, antenatal obsessive–compulsive disorder, postnatal obsessions, postpartum obsessions, OCD best practice, and perinatal best practice. Databases searched included PsycArticles (Ovid), PsychINFO (Ovid), Psychiatry online, Science Direct, PubMed, ProQuest, Taylor & Francis, Google Scholar, and Cochrane Library. Existing OCD (American Psychiatric Association, 2007a, b; National Institute for Health and Care Excellence, 2005) and perinatal mental health guidelines (Austin and Highet 2017; beyondblue 2011; National Institute for Health and Care Excellence 2014) were also reviewed. Additionally, panel members were invited to submit information they thought should be included in the guidelines via open text submissions on Qualtrics Survey Software.

The proposed statements and suggestions from panellists were reviewed by a research team for relevance, clarity, and repetition of content. The research team comprised one perinatal psychiatrist (M.G.) and two senior clinical psychologists with clinical and research expertise in pnOCD (R.A. & C.R.), a doctoral student (M.M.), and two postgraduate masters students (C.L. & T.M.) conducting research on pnOCD. A total of 103 initial statements were generated and organised by topic. To support the generation of evidence-based guidance, a plain language summary of the current research literature on clinical approaches to pnOCD was created and provided to participants. The clinical literature summary was also reviewed by the research team for content and clarity.

Procedure

Ethics approval was obtained from the Curtin University Human Research Ethics Committee (No. HRE 2017-0087) before study commencement. The recommendation statements formed by the research team were presented to panellists for rating over three survey rounds. The statements, along with definitions of key terms and the clinical literature summary, were presented online via Qualtrics. During each round, panellists chronologically reviewed and rated each statement using a Likert scale; ratings of 1 (‘essential’) and 2 (‘important’) indicated endorsement of the item, and 3 (‘don’t know/depends’), 4 (‘unimportant’), and 5 (‘should not be included’) considered to be not endorsed. Panellists were instructed to respond based on any knowledge available to them, including clinical or lived-experience, research evidence, or any other experience related to perinatal OCD. They were also asked to consider the clinical literature summary when completing each round so that the recommendations would reflect evidence-based practice and practice-based evidence provided by experts (Jorm 2015).

In Round 1, panellists were able to write-open ended responses regarding any additional statement suggestions, or amendments to existing statements, they felt necessary. Any additional recommendation statements generated by participants were reviewed by the research team and included for rating by the panels in Round 2. Consensus criteria for whether each statement is considered endorsed, retained for re-rating, and not endorsed is available in Table 1. Those recommendations that were endorsed or rejected in previous rounds were not included in subsequent rounds. Participants were given two weeks to complete each round, with up to three reminder emails sent to participants yet to complete the round, and they were able to return to the survey form to complete the round at a later time if they were unable to do so in one sitting. Participants who completed at least 50% of Round 1 were invited to respond to Round 2 and 3. They were asked to consider the following information provided to them about the statements from the previous round retained for re-rating: how they previously responded to each item, and the level of endorsement of the item by each panel and overall. Statements considered ‘endorsed’ after the three rounds were included and those ‘not endorsed’ were excluded from the final recommendation set. This list of final recommendations grouped by topic area can be found in Table 2.

Table 2 List of endorsed statements for final recommendation, by topic
Full size table

Data analysis

One-hundred and three statements across eight topic areas were rated in Round 1. Comments from panellists in Round 1 contributed a further 18 statements that were included in Round 2. Figure 1 indicates the number of items which were included, retained, and excluded for re-rating during each round.

Fig. 1
figure 1

Number of items included, re-rated and excluded at each round of the study

Full size image

Results

Endorsed items

From a total of 121 statements, 102 were endorsed; 69 from Round 1, 23 from Round 2, and 10 from Round 3. Six of the 18 participant-generated recommendations received endorsement. A list of the endorsed statements is provided in Table 2. Of note, eleven statements (10.8% of endorsed statements) were unanimously endorsed by all members of both panels as ‘essential’ or ‘important’ to include (see Table 2).

Excluded items

Nineteen statements were excluded; six from Round 1, seven from Round 2, and six from Round 3. These excluded statements are available in the supplementary materials. Certain statements were definitively rejected by both the Professional and Consumer Panels. Items with the lowest endorsement rate across both panels included the statement that “The assessing clinician should also provide the pnOCD treatment” and “Benzodiazepines and psychoeducation should be considered in some pnOCD cases as an initial treatment approach, instead of SSRI’s or other medications”. Some statements that were suggested by panellists in Round 1were also rejected strongly by both panels in the following round (e.g., “Other psychotherapies should be offered (e.g., Jungian-Feminist Therapy) as options for treatment”). Finally, some statements were excluded due to disagreement between panels.

Differences between consumer and professional panels. Post hoc analysis was conducted to explore any differences between Professional and Consumer Panels’ endorsement of recommendations throughout the three stages of the study. Overall, there was substantial agreement Consumers were less likely to endorse items in the first round compared to the Professional Panel, although this difference appeared to diminish in subsequent rounds (see Table 3). The results within the Consumer Panel were also more varied in the first round than within the Professional Panel whereby consumers had a greater number of items to be re-rated. Again, these differences appeared to decrease in subsequent rounds. between the panels, with over 75% agreement in Round 1, and over 65% agreement in Round 2. Statements were deemed to have considerable disagreement if the difference in endorsement across panels was ≥ 30% (Rosenthal 1996). Only five items met the criteria for considerable disagreement, as shown in Table 4. Consumers were more likely to rate those statements with large differences as important or essential than the professional panel.

Table 3 Summary of items endorsed, re-rated, or rejected by consumers, professionals and by both panels combined for rounds one, two, and three
Full size table
Table 4 Statements with large differences (≥ 30%) in endorsement between panels
Full size table

Discussion

The aim of the current study was to address the gap in literature regarding the clinical management of pnOCD; more specifically, to develop recommendations for the assessment, treatment, and management of pnOCD, and other considerations to best support individuals with this disorder. To our knowledge, this is the first study to systematically collate expert views, and generate best-practice recommendations for pnOCD, using a consensus-based methodology.

Results indicated that participants agreed with most (96/103) of the initial recommendation statements which were created based on the initial literature review and panelist suggestions. Of particular significance is the high degree of agreement between the Professional and Consumer Panels as to which statements should be endorsed, which may confer confidence as to the veracity of the final recommendation set. A strength of the Delphi method is that participants can provide input about amendments to statements, or additional statements which they think should be included (Keeney et al. 2010). This process resulted in the endorsement of an additional six practice recommendations generated by professionals or consumers, therefore adding the comprehensiveness of the final recommendation set and highlighting the importance of including consumers in the creation of such recommendations.

Most items that were endorsed by less than 70% of consumers fell within the ‘Differential Diagnosis’ and ‘Case Care Considerations’ categories. Post hoc analysis identified that these items were not rated as ‘unimportant’ by the consumers, but as ‘don’t know/depends’. This likely shows the Consumer Panel’s ability to acknowledge the limits of understanding in these areas and, therefore, a preference not to make judgment. A theme of statements which were rejected by either Panel were those relating to child safeguarding. The Professional Panel’s rejection of such statements may be due to child safeguarding being a challenging decision for health practitioners to make, requiring careful consideration of risks of both direct and indirect infant harm, often in the context of comorbid maternal mental health diagnoses (i.e., Poinso et al. 2002). It is likely that Consumers rejected recommendations regarding child safeguarding given the potential risk of pnOCD symptoms being misidentified as indicating actual risk of harm to the infant, which may prevent parents from disclosing pnOCD symptoms to health practitioners from the outset (Megnin-Viggars et al. 2015). Regardless, these differences in professional and consumer perspectives highlight the potential consequences of the mischaracterisation of pnOCD, and the need for targeted guidance to increase clinicians’ understanding of how to assess and manage pnOCD.

This study has made a unique and much needed contribution to the pnOCD literature. Existing OCD and perinatal mental health best-practice guidelines do not detail specific considerations relevant to perinatal OCD which are important to explore (American Psychiatric Association, 2007a,b; Austin and Highet 2017; beyondblue 2011; National Institute for Health and Care Excellence, 2005, 2014). For example, the importance of assessing whether any intrusive thoughts of harming the baby are experienced as ego-syntonic or ego-dystonic, which is central to distinguishing between infant harm-related thoughts in OCD or psychosis (Fairbrother & Abramowitz 2016). This is particularly important given that parents’ may be reluctant to disclose intrusive thoughts for fear they will be misunderstood by healthcare providers (Challacombe & Wroe 2013), specific treatment indications for OCD (e.g., exposure and response prevention; Stein et al. 2010), and the potentially aggravating effects of misdiagnosis or mistreatment of pnOCD (Challacombe & Wroe 2013; Gershkovich 2003; Mulcahy et al. 2020; Storch 2015; Veale et al. 2009).

A strength of this study is that the endorsed statements provide guidance that spans the need for prevention via psychoeducation through to complex differential diagnostic and treatment considerations. By having participants rate, rather than rank, whether an item is important for inclusion in the guidelines, we have identified a comprehensive list that does not prioritise any one aspect of assessment, treatment or management over another. Each item may therefore be considered important both in isolation and in conjunction with the other items listed. For example, the provision of psychoeducation “…that there is no evidence that parents experiencing harm-related intrusions, no matter how horrific in content, will act on these thoughts” is important corrective information for someone experiencing pnOCD intrusions. However, it is equally as important to appropriately assess the client and follow the clear differential diagnosis statements to ensure the person is presenting with pnOCD and not another mental health disorder warranting a different treatment and management approach.

Certain limitations of the Delphi methodology used in this study must be acknowledged. First, five members of the Professional Panel and four members of the Consumer panel did not complete Round 1 of the study. A possible reason for this is that Round 1 included 103 items which may have been too time-demanding for some panellists to complete (Keeney et al. 2010; van Zolingen & Klaassen 2003). This may have particularly been the case given that this research was conducted during the unprecedented time of the initial COVID-19 global pandemic outbreak. However, there were still enough participants in each round as part of a homogeneous panel for results to be deemed representative of the larger population and therefore generalisable (Keeney et al. 2010). The relative similarity in the size of both the final Professional and Expert Panels also means that the views of both groups were equally weighted in reaching the study findings. Second, this study was focused on mothers with OCD, although fathers are also known to experience OCD in the perinatal period (Abramowitz et al. 2001). Thus, the guidelines developed in the present study may not apply to fathers with pnOCD, and there is a critical need for further research in this area more broadly. Finally, the specificity of recommendations was somewhat impacted by the global recruitment of experts, and the recruitment of both consumer and professional expert panels. For example, the lack of availability of mother-baby units in some locations led to the recommendation that “Health professionals be aware of the availability” of such services, rather than specifying that individuals should be referred to a mother-baby unit. Furthermore, we did not consider that consumers would likely had sufficient training to make a valid judgement on the psychometric properties of screening or assessment tools to recommend any particular one over another.

Given the paucity of pnOCD literature, this set of consensus findings may be a useful basis for identifying gaps in the pnOCD literature. For example, we now know that both consumers and experts agree with the need for routine screening via pnOCD specific screening items, but there is a need to empirically evaluate the existing screening tools to determine a preferred tool. In sum, the current study represents a novel contribution to the emerging clinical literature on pnOCD and offers an important step forward in efforts to develop and disseminate best-practice for pnOCD.