Introduction

The Diagnostic and Statistical Manual of Mental Disorders (DSM-V) (American Psychiatric Association 2013) described the symptoms for the most prevalent anxiety disorders: generalised anxiety disorder (GAD), panic disorder, agoraphobia, obsessive compulsive disorder, specific phobias and social anxiety disorder. Although specific anxiety disorders have specific symptoms, they share common symptoms which include excessive and intrusive worrying, feeling overwhelmed, angry or scared, irritability, fatigue, difficulty concentrating and sleeping, an elevated sensitivity to threat and a bias to interpret ambiguous information in a negative way (Craske et al. 2009; National Institute for Health and Care Excellence, NICE 2011; Highet et al. 2014; Staneva et al. 2015). In pregnancy, concerns over the wellbeing of the baby, the labour and birth or parenting may present as predominant features (Staneva et al. 2015; Vythilingum 2009). Pregnant women with anxiety have reported feeling a loss of control over their bodies and feeling confused by ambiguous information about pregnancy and labour (Highet et al. 2014; Keeton et al. 2008; Staneva et al. 2015). Women with a previous or existing mental illness, those who have poor partner or social support, women who are socially isolated, women from a low socio-economic background, those who are exposed to violence or abuse, women who are substance misusers, women with unplanned or unwanted pregnancies or those who have had a previous negative experience of pregnancy or birth are especially vulnerable to developing symptoms of anxiety in pregnancy (Biaggi et al. 2016; Staneva et al. 2015).

Reported prevalence of anxiety disorders in pregnancy varies from 10 to 16% (Goodman et al. 2014; National Institute for Health and Care Excellence (NICE) 2018; Rubertsson et al. 2014) and has been reported as 15–16% in UK and Canadian community samples (Heron et al. 2004; Fairbrother et al. 2016). Elevated and prolonged anxiety in pregnancy has been associated with pre-term birth, fetal growth restriction (Ding et al. 2014; Littleton et al. 2007; Rich-Edwards and Grizzard 2005) and childhood behavioural problems (Blair et al. 2011; Cardwell 2013; Davis and Sandman 2010; Glover 2014; Stein et al. 2014). Mild to moderate psychological distress can be debilitating and have a negative effect on women’s general functioning (Furber et al. 2009). It is associated with post-traumatic stress disorder (Czarnocka and Slade 2000; Iles et al. 2011) and postnatal depression (Heron et al. 2004; Coelho et al. 2011).

Rationale

Women with severe anxiety symptoms require assessment and management from specialist mental health services. In the management of women with mild to moderate mental health problems, the aim is to prevent an escalation of symptoms and improve a woman’s quality of life (NICE 2018). All women identified with mild to moderate mental health problems should have access to a range of support such as wellbeing advice, guided self help, motivational interviewing, cognitive behavioural therapy (CBT) and medication (Department of Health (DOH) 2012). However, services to support the emotional wellbeing of women need to be strengthened in order to provide suitable and timely support and treatment to help avoid illness (Maternal Mental Health Alliance 2013). The NICE guideline for perinatal mental health (NICE 2018) suggested that non-pharmacological interventions such as low intensity psychological interventions may benefit women with symptoms of mild to moderate anxiety. Social support, assisted self-help and CBT are proposed in The Healthy Child Programme (DOH 2009) as possible interventions to support pregnant women with anxiety. However, evidence of the effectiveness of such interventions has not yet been established.

Objectives

The review aimed to answer the following questions:

  • How acceptable for pregnant women are non-pharmacological interventions for reducing the symptoms of mild to moderate anxiety?

  • How beneficial do pregnant women consider non-pharmacological interventions to be in reducing the symptoms of mild to moderate anxiety in pregnancy?

Methods

Protocol and registration

A systematic review was conducted following the Centre for Reviews and Dissemination guidelines (CRD 2009). The narrative synthesis followed the guidelines by Popay et al. (2006). The review protocol was registered on the PROSPERO database at the CRD (Evans et al. 2015 CRD42015017841).

Eligibility criteria

Participants

Pregnant women of all parities across the three trimesters of pregnancy. Women less than 18 years of age and women who lacked capacity to provide informed consent were excluded from the study. In addition, pregnant women with complex social factors were not included (pregnant women who misuse alcohol and/or drugs; are recent migrants, asylum seekers or refugees; have difficulty reading or speaking English; experience domestic abuse) (NICE 2010). Women under the care of specialist mental health services or women with severe symptoms of anxiety were excluded. Studies used various measurement techniques to assess eligibility. Some studies used anxiety scales with dimensional cut-off scores for mild, moderate and severe anxiety to assess eligibility (Brunton et al. 2015). Studies using dimensional anxiety scales who included women with severe scores were excluded (Table 2).

Interventions

Non-pharmacological interventions were classified as (1) psychological, (2) mind-body, (3) educational and (4) supportive interventions.

Outcomes

The primary outcome was women’s views on the acceptability of and satisfaction with interventions.

Study design

Quantitative or qualitative studies which assessed women’s views on the acceptability of and satisfaction with an intervention.

Information sources

A systematic search of the following electronic databases was undertaken in January 2015 and updated in June 2018:

Medline (Medical Literature Analysis and Retrieval System Online), CINAHL (Cumulative Index to Nursing and Allied Health Literature), Maternity and Infant Care database from MIDIRS (Midwives Information and Resource Service), PsycINFO, The Cochrane Library, EMBASE (Excerpta Medica Database), CRD (Centre for Reviews and Dissemination), SSCI (Social Sciences Citation Index), ASSIA (Applied Social Sciences Index and Abstracts), HTA (Health Technology Assessment) Library, JBI (Joanna Briggs Institute) Evidence-Based Practice Database and AMED (The Allied and Complementary Medicine Database). Visually scanned reference lists from relevant primary studies and reviews identified two additional studies for inclusion.

Search

The search was limited to studies conducted in countries with similar maternity care to the UK and published in English since 1990. This period reflects the time that non-pharmacological interventions have been recommended to support women’s mental health during pregnancy (DOH 1999). Search terms included pregnancy, antenatal, anxiety, intervention, trial, review, women’s views, acceptability and satisfaction. A full search strategy is included in Appendix 1.

Study selection

Potentially eligible papers were retrieved for full text assessment which was conducted independently by two researchers. Any disagreements were resolved by a third researcher.

Data collection process

A pre-piloted data extraction form was completed independently by two researchers for each included study.

Quality assessment

The Critical Appraisal Skills Programme (CASP 2014) for assessing the methodological quality of qualitative studies and the Critical Appraisal Checklist for a Questionnaire Study (Boynton and Greenhalgh 2004) were used to assess the quality of studies included in the review.

Analysis strategy

Data analysis and synthesis followed the suggested frameworks for conducting a narrative synthesis (Popay et al. 2006). Qualitative and quantitative studies which addressed the research questions were used to explore similarities and/or differences in the common themes (Popay et al. 2006). Each study was first described with reference to the context as intended by the original research (Jensen and Allen 1996). Secondly, a table of key concepts was produced to explore the homogeneity of themes, noting any discordance. Themes emerged from the similarities and contradictions between the study findings (Walsh and Downe 2005). The next phase involved translating the study findings using concepts that could be applied to all or some of the studies.

CERQual assessment

The Confidence in the Evidence from Reviews of Qualitative Research (CERQual) approach was used to assess the extent to which the review findings from the qualitative studies represented the phenomenon of interest (Lewin et al. 2015; The Cochrane Collaboration 2011). The process required an individual assessment of the studies which contributed to a review finding. Assessment components included methodological limitations, relevance to the review questions, adequacy of data and coherence (whether the finding was well grounded in data with a convincing explanation). After assessing each of the four components, an assessment of the overall confidence in each review finding was made. Each review finding was assessed as having a high, moderate, low or very low confidence rating (Lewin et al. 2015).

Results

Study selection

The search identified 3522 potentially eligible papers which were assessed on the information provided in the abstract using the review eligibility criteria. Duplicate papers were removed. Potentially eligible papers (n = 3494) were retrieved for full text assessment. Excluded papers (n = 3643) (1) did not report interventions delivered in pregnancy, include women’s views or report non-pharmacological interventions; (2) included women with severe mental health concerns or complex social factors. The literature search and inclusion process is detailed in the PRISMA Flow diagram (Moher et al. 2009) (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram: women’s views of interventions

Full size image

Study characteristics

The 14 included studies, conducted in Australia, Canada, Germany, New Zealand, UK and USA, were reported from 2009 to 2015. Components of the interventions are detailed in Table 1. There were 800 women in the included studies. Sample sizes ranged from four women (Breustedt and Puckering 2013) to 298 women (Brugha et al. 2015). Overall, from the 800 participants, 204 women provided views about the interventions via questionnaires or qualitative interviews.

Table 1 Data extraction from the studies included in the review
Full size table

Participants

In four studies, women were recruited from a general pregnant population. In eight studies, pregnant women with a history of mood concerns or elevated anxiety or depression scores were recruited (Table 2). Two studies included women with social risk factors and pregnant women with a history of previous pregnancy loss. Many of the studies used one of the self-report measures listed in Table 3 for participant inclusion.

Table 2 Psychological characteristics of participants in the included studies
Full size table
Table 3 Anxiety self-report measures used in the included studies
Full size table

Women were mainly recruited into studies while attending antenatal appointments in hospital and community locations. Women either self-selected into studies or were referred by healthcare professionals (HCP).

RCT randomised controlled trial, IPT inter-personal therapy, CBA cognitive behavioural approach, CBT cognitive behavioural therapy, MBSR mindfulness-based stress reduction, MBCT mindfulness-based cognitive therapy, PND postnatal depression, NR not reported

Interventions

Five studies evaluated psychological interventions: Cognitive behavioural therapy (CBT) (Bittner et al. 2014; Mcgregor et al. 2013; Milgrom et al. 2015); cognitive behavioural approach (CBA) (Brugha et al. 2015); psycho-educational and inter-personal therapy (IPT) (Thomas et al. 2014); and psychological, practical techniques and social support to promote wellbeing (Breustedt and Puckering 2013).

Six studies evaluated mind body interventions: hypnotherapeutic techniques and stress management (Goodman et al. 2014), mindfulness-based cognitive therapy (MBCT) (Dunn et al. 2012), mindfulness (Woolhouse et al. 2014), mindful yoga and mindfulness-based stress reduction (Beddoe et al. 2009) and yoga (Davis et al. 2015).

One study evaluated a supportive intervention: home visits by nurses (Côté-Arsenault et al. 2014) and one study considered how perinatal psychosocial assessment may act as an intervention (Darwin et al. 2013). Many of the psychological and supportive interventions also included components of parent education, relaxation and/or social support.

Outcomes

The included studies reported women’s views and responses to questions about the level of satisfaction, perceived benefits, acceptability, and relevance of interventions.

Study type

Qualitative and quantitative studies were included. Four studies conducted interviews with pregnant women (Brugha et al. 2015; Cornsweet Barber et al. 2013; Darwin et al. 2013; Woolhouse et al. 2014). Three studies interviewed postnatal women about their participation during pregnancy (Breustedt and Puckering 2013; Côté-Arsenault et al. 2014; Dunn et al. 2012). Goodman et al. (2014) collected qualitative data during pregnancy from a post-intervention questionnaire.

There were six cross-sectional surveys, administered post-intervention during the second and/or third trimesters of pregnancy (Beddoe et al. 2009; Bittner et al. 2014; Davis et al. 2015; Mcgregor et al. 2013; Milgrom et al. 2015; Thomas et al. 2014). The quality assessment of the included studies is presented in Table 2.

Quality appraisal

A summary of the quality assessment of the included studies is presented in Table 1. Many surveys had limited reporting of the questionnaire design, validity and reliability, administration and analysis all included a questionnaire as part of a larger quantitative evaluation. Results were presented as numbers and percentages with individual questionnaire item scores and brief descriptive statements of agreement or disagreement from participants.

GRADE-CERQual assessment

The CERQual components were used to assess the overall confidence in the findings of the seven qualitative studies:

  • Two studies were assessed as having moderate methodological limitations (Cornsweet Barber et al. 2013; Dunn et al. 2012). Five studies were assessed as low for methodological limitations.

  • One study was assessed as having moderate coherence (findings well-grounded in the data) (Cornsweet Barber et al. 2013). Six studies were assessed as being highly coherent.

  • Two studies were assessed as being moderately relevant to the context of the review questions (Côté-Arsenault et al. 2014; Darwin et al. 2013). Five studies were assessed as being highly relevant.

  • The adequacy of data was assessed as being highly adequate in six studies where the authors provided detailed accounts of women’s views and experiences and used the results to build theories and explanations (Popay et al. 1998). One study reported only a small number of examples of participant quotations to support the findings and was assessed as being moderately adequate (Cornsweet Barber et al. 2013).

Synthesis of the findings

Data analysis revealed five descriptive themes: (1) motives and barriers to participating in studies, (2) acceptability of interventions, (3) satisfaction with components of interventions, (4) overall satisfaction with interventions and (5) perceived benefit from participation. Table 4 outlines the reported positive views of intervention components and highlights components which were less beneficial or acceptable. The CERQual assessment of the confidence in the evidence contributing to the findings is presented in Table 5.

Table 4 Summary of the themes and data from the included studies
Full size table
Table 5 GRADE-CERQual assessment of the themes identified in the findings
Full size table

Motivation and barriers to participating in studies

Participants in studies of mindfulness interventions who had previous experience of anxiety and depression were motivated to participate (Dunn et al. 2012; Woolhouse et al. 2014). Women wanted to learn new ways to manage their symptoms; they considered that the intervention would help them achieve a positive experience of pregnancy. However, some women who were identified or referred for inclusion by a healthcare professional (HCP) had concerns about participation (Breustedt and Puckering 2013; Darwin et al. 2013). They were uncertain about the reason for their selection and were concerned that disclosing their symptoms may lead to unwanted interference from HCPs and social care services.

Acceptability of interventions

Studies with reported attrition rates below 25% included group yoga interventions (Beddoe et al. 2009; Davis et al. 2015) and interventions provided one-to-one (Brugha et al. 2015; Cornsweet Barber et al. 2013; Côté-Arsenault et al. 2014; Milgrom et al. 2015). Five out of seven of the studies with lower attrition rates did not include psychological assessment as part of the inclusion criteria. Rates of attrition greater than 45% were reported in studies of a group CBT intervention for women with elevated anxiety and depression scores (Bittner et al. 2014) and a psycho-social intervention for women with complex social factors (Breustedt and Puckering 2013).

Women assessed as vulnerable or at risk of developing anxiety and depression initially felt uncomfortable attending group sessions and feared judgement or disapproval from the group (Breustedt and Puckering 2013; Woolhouse et al. 2014). Creating a relaxed and non-judgemental atmosphere and visiting the women at home before the group began helped women to feel confident about attending and created a welcoming experience. Once the group was established, sharing time with other pregnant women was valued by most participants (Breustedt and Puckering 2013; Dunn et al. 2012; Woolhouse et al. 2014).

Satisfaction with components of interventions

Mcgregor et al. (2013) delivered a brief individual CBT intervention in 10-min sessions, but reported that some women would have liked more time and in-depth discussions about their emotional difficulties. Having time to discuss emotional issues with HCPs was highlighted as an important component by Darwin et al. (2013). Research interviews provided women with an opportunity to talk, which for some had been the first opportunity to discuss their feelings.

A number of participants in the study by Darwin et al. (2013) felt that completing psychological questionnaires resulted in them being confronted by the reality of their anxiety and depressive symptoms but they felt left without any further support. Brugha et al. (2015) reported that many women found completing the EPDS important and helpful. However, a few women found it difficult to discuss their emotions and felt apprehensive about the potential consequences resulting from elevated EPDS scores, such as the information being used by HCPs to raise child protection concerns.

Breustedt and Puckering (2013) discussed how the end of the group left some participants with a sense of loss and signalled a period of adjustment. This was addressed by the provision of follow-up postnatal groups and reunions. Some women in the study by Goodman et al. (2014) suggested that having partners included in at least one session would help support them with their new practices and would have welcomed on-going support to continue developing mindfulness techniques.

Some studies of mind-body interventions included homework exercises. Authors reported that participants had not completed some of the content (Cornsweet Barber et al. 2013) or at times, the homework had felt too much for the women to complete (Goodman et al. 2014). Certain exercises were reported as helpful to some women and unhelpful to others; however, women did not feel any specific exercises should be omitted. Women wanted an opportunity to learn a variety of techniques, having the choice to participate in exercises which they enjoyed or found useful (Goodman et al. 2014; Woolhouse et al. 2014).

Overall satisfaction with interventions

Women who participated in psychological or mind-body interventions reported an overall satisfaction and described interventions as enjoyable, valuable and beneficial. Group interventions received positive comments, women were able to discuss their thoughts and experiences which they had found difficult to discuss with professionals or their family (Breustedt and Puckering 2013). Groups provided a supportive environment where they could make friends, knowing that others had similar thoughts and experiences helped women develop an acceptance of their feelings and feel less isolated (Breustedt and Puckering 2013; Dunn et al. 2012; Goodman et al. 2014).

Perceived benefit from participation

Some women felt they had derived benefit from learning practical breathing techniques and developing an ability to reflect on their thoughts and emotions (Cornsweet Barber et al. 2013; Woolhouse et al. 2014). Women said that exercises such as the body scan (being aware of different areas of the body) had helped them to sleep better.

Some participants in the studies of mindfulness and CBT interventions reported a greater understanding of the causes of stress and anxiety in their lives and greater self-awareness of their thought patterns. This helped them respond in a more positive way to situations and feelings, before negative thought patterns could escalate (Beddoe et al. 2009; Goodman et al. 2014; McGregor et al. 2013; Woolhouse et al. 2014). For some women, learning to recognise their feelings helped them to accept their anxious thoughts (Goodman et al. 2014). Rather than becoming annoyed or frustrated, they had learned to be kinder to themselves and felt more confident and positive about the future (Beddoe et al. 2009; Breustedt and Puckering 2013; Côté-Arsenault et al. 2014).

Discussion

The review was conducted to evaluate women’s views on the acceptability of and satisfaction with non-pharmacological interventions to reduce the symptoms of anxiety in pregnancy. Fourteen studies from six countries were included which accessed women’s views through qualitative interviews or questionnaires.

The review followed a narrative synthesis framework (Popay et al. 2006) and used the CERQual approach to assess the confidence in the findings of the review. Themes assessed as having a high confidence were seen in at least six of the included studies, all of which were assessed as being at least moderately well conducted.

Quality of included studies

Only two survey studies used validated questionnaires to access participant feedback. Such feedback can be used to improve intervention design, recruitment of and study retention in clinical trials. However, validated surveys and benchmarks need to be developed to assess the experience of participation in clinical trials (Planner 2015). In many of the studies, data were collected from all or a sub-section of participants who had successfully completed interventions which was a potential source of selection bias. Five of the 14 studies collected data from all or a sub-set of participants in the postnatal period which may introduce positive or negative recall bias.

Recruitment and data collection methods were only described in three studies. Four of the seven studies which used qualitative interviews to access women’s views provided detailed descriptions of the analytic method. All of the qualitative studies presented participant quotations to support the findings. Lewin et al. (2009) described how qualitative components are included in RCTs of complex interventions to explore participants’ experiences; however, the quality of qualitative components can be variable and often lacks justification. Recent reviews of interventions focused on psychological health and wellbeing in pregnancy have highlighted the need to improve the reporting of study methods, recruitment strategies and study quality (Fontein-Kuipers et al. 2014; Marc et al. 2011; Morrell et al. 2016; Ryan 2013).

Participants

Studies which included women from general antenatal populations aimed to help women develop coping strategies to prevent the development of symptoms of anxiety/depression, whereas, studies which recruited women with elevated scores or risk factors for anxiety and/or depression aimed to reduce or improve existing anxiety symptoms.

Milgrom et al. (2015) reported that 54% of the initial study population declined to complete symptom checklists; however, other studies which conducted psychological eligibility assessment did not report the rates of consent (Bittner et al. 2014; Goodman et al. 2014; McGregor et al. 2013). Reporting the rate for agreeing or declining eligibility assessment would help researchers to consider the design of effective recruitment strategies (Williams et al. 2007). Recruitment could be maximised through discussion and providing information early in the recruitment process, addressing women’s concerns about psychological screening and fear of stigma (Brintnall-Karabelas et al. 2012; NICE 2018). Women’s apprehensions about joining group interventions may be eased by conducting welcome visits, prior to group commencement, in order that women feel more confident to participate (Breustedt and Puckering 2013).

Only one study was focused on women with elevated symptoms of anxiety, with seven studies selecting women with symptoms or risk factors for anxiety alongside other psychosocial symptoms or risk factors. Although a multidimensional approach has been reported as an important factor to promote psychological wellbeing in pregnancy (Jomeen 2004), interventions targeting one condition may not be effective for the other co-morbid condition (Garber and Weersing 2010). Interventions that focus on improving symptoms of anxiety and depression need to define the underpinning theory of change before testing the mechanism by which an improvement in symptoms is likely to occur for each condition.

Interventions

Only three studies reported details of the facilitator training to deliver interventions. In most studies, women were not asked to provide their views on the acceptability or relevance of intervention facilitators. Such information could be helpful for researchers to consider the type, skill requirement and appropriate expertise of intervention facilitators, making efficient use of the available resources.

Developing an awareness of the causes of anxiety and the ability to reflect on thoughts and emotions was reported as beneficial by women across all categories of interventions. Darwin et al. (2013) highlighted that some women felt distressed when confronted by their emotions and suggested that self-reflection needed to be followed with further support and discussion. Facilitating time for women to discuss their feelings and experiences was highlighted as an important component across the included studies. Discussions with HCPs were reported as helpful for women with symptoms of or risk factors for mental illness (Brugha et al. 2015; Côté-Arsenault et al. 2014; Darwin et al. 2013; McGregor et al. 2013). In group interventions, women who felt isolated found comfort when they discovered other women had similar thoughts and experiences (Breustedt and Puckering 2013; Dunn et al. 2012; Goodman et al. 2014; Woolhouse et al. 2014). Most studies of psychological and social support interventions included multiple components: psychological therapy, discussion sessions, parent education and/or social support. An investigation into the acceptability and satisfaction of specific components was only reported in the qualitative studies, possibly because these studies had greater scope to report in-depth qualitative findings.

The location of interventions and level of commitment were important factors for women (Beddoe et al. 2009; McGregor et al. 2013). Work commitments and other responsibilities may restrict women’s ability to regularly attend sessions and complete additional homework. Most interventions were held during the daytime in hospital clinics, although some were also offered in community centres and during the evening which may have made it easier for women to attend.

Strengths of the review

To our knowledge, this is the only review of women’s views on the acceptability of and satisfaction with interventions to reduce the symptoms of anxiety in pregnancy. A comprehensive search strategy increased the likelihood that all potentially relevant studies were included. The review was strengthened by using good quality, independent and appropriate assessment methods. The use of the CERQual tool helped assess the certainty of the findings. A narrative synthesis approach (Popay et al. 2006) involved a textual and thematic exploration of the data, identifying common themes, contradictions and highlighting where the evidence was absent (Lucas et al. 2007). This helped to develop recommendations for the design and reporting of future research (Craig et al. 2008).

Limitations of the review

Studies not published in English were not included in the review. Most of the included studies had small sample sizes (n = 4–30); many were feasibility studies or additional components to larger trials. Due to the limited reporting of the study methods in many of the included studies, a full quality assessment was not possible although methodological limitations were assessed and informed the overall CERQual findings. There was considerable heterogeneity between the intervention designs, participants and time frames in the included studies. Participation and experiences of interventions may differ for particular groups of women. The narrative synthesis explored and compared the different approaches to inform discussion and consideration of future intervention designs (Lucas et al. 2007).

Conclusion

The review findings are limited due to the small number of included studies, many with small sample sizes and limited reporting of methods. Women’s views on the acceptability of and satisfaction with a range of interventions were overwhelmingly positive. The review has highlighted the importance of creating a welcoming non-judgemental context for group interventions. Most women valued individual or group discussions about their symptoms of anxiety. Discussions helped women to feel supported and develop supportive networks.

Responding to women’s views and experiences will help to inform the design of interventions which are acceptable to women and to develop an understanding of how and why intervention components may contribute to outcomes. Many qualitative studies accessed the views of women who had successfully completed interventions which introduced the potential for selection bias. Future studies need to access and report the views of women who did not participate or complete interventions to identify where further improvements are required. Researchers need to consider the acceptability of eligibility screening and identify ways to effectively communicate the purpose of screening to potential participants.

Study reports should include the methodological approach, recruitment strategy, intervention provider details and data analysis procedures. The use of validated evaluation questionnaires, following quality frameworks and reporting process evaluations will help researchers compare intervention studies and assess whether interventions may produce similar or different effects in other settings.