Conclusions

Our aim was to provide high-quality evidence of the effectiveness of CPMHTs, in terms of increased treatment use and the impact of CPMHTs on women children and families. We also sought to examine the reliability and predictive validity of two observational measures assessing the quality of parent–infant interactions. We aimed to test whether brief forms of each tool might be identified with sufficient predictive validity, even over short observation periods, enhancing their suitability for routine clinical use.

The research spanned four distinct WPs, using a multi-method approach.

• Work package 1: What service variations are found in CPMHTs and their links to the wider system of universal services, primary care and generic secondary care?
• Work package 2: What is the most reliable, valid and clinically feasible observational measure for routinely assessing parent–infant interaction quality in PMH services?
• Work package 3: Which CPMHT components promote access to treatment and which components work, for whom, in what circumstances, how and why to reduce PMH problems?
• Work package 4: Are CPMHTs associated with higher levels of access to secondary care mental health services (generic and CPMHTs), and in women with pre-existing severe disorders, lower risk of relapse and improved birth outcomes compared with generic services?

Are CPMHTs effective and do they make a difference to maternal and neonate outcomes?

Extending previous research that has found a link in maternal mental health problems and negative maternal and neonate outcomes, in our linked data set analyses (WP4), we found that among the 9% of women who had a previous history of a specialist mental health contact those with more severe histories (had at least one hospitalisation) or those who had contact within the past year were at the highest risks of having negative obstetric, maternal and neonatal outcomes. These findings provide support for the argument for CPMHT, which seek to provide treatment for women with moderate to severe and complex mental health disorders; those at greatest risk of negative outcomes.

Analysing data from linked data sets, we found greater use of any specialist mental health service and lower use of acute care (psychiatric inpatient and crisis teams) in perinatal women in areas that had CPMHTs relative to areas without CPMHTs, suggesting that CPMHTs are effective in improving access to mental health care for perinatal women and their babies and likewise reducing acute care usage. However, this reduction in overall acute care was driven by lower use of crisis teams, with a higher proportion of women in CPMHT areas being admitted to hospital and having longer lengths of stay, on average, compared to areas without CPMHTs. As a result, our economic analysis found total secondary mental healthcare costs over the perinatal period to be significantly higher for areas with CPMHTs. This may be due to greater access to MBUs in CPMHT areas, with MBUs being associated with longer lengths of stay compared with traditional psychiatric wards. Notably, MBUs are for two individuals, the mother and the baby, and treat the mother and the relationship with the baby. Likewise, although there was a decrease in preterm births in areas with CPMHTs, there were higher rates of stillbirth, neonatal death and premature delivery.

Our findings provide support for investment in CPMHT. Without CPMHTs, treatment use rates among perinatal women are low, and previous economic modelling has demonstrated significant downstream costs associated with both maternal and negative child outcomes. Because the systematic and widespread implementation of CPMHT in England are still early, it is not yet possible to examine their longer-term clinical and economic impacts. So, although the upfront costs of CPMHTs are greater, they provide important opportunities to change longer-term outcomes, and to change broader clinical and economic impacts than it was possible to measure in the current study. However, our findings regarding neonate outcomes temper these results. Although they may be due to chance, it is essential to replicate these findings, investigate the ‘black box’ explaining mechanisms underlying this link, and improve co-ordinated work between mental health and health providers in order to ensure women with mental health needs receive health care tailored to these needs.

What service models are effective for mothers with PMH problems and their babies, and does this vary according to mental health needs? (i.e. who delivers these interventions, with what skill mix and competencies, in what settings?) Which interventions are attractive and acceptable to women, and clinically effective within/in conjunction with mainstream secondary mental health care, maternity and primary care?

• In our taxonomy work (WP1), we evidenced wide variations in CPMHT structures between 2019 and 2020. Notably, none of the CPMHTs met the staff mix and WTE for a comprehensive service as suggested by the Royal College of Psychiatry’s CR197 guidance.18 Using simplified models of CPMHT structures defined by staff mix alone (not WTE), we found that 85% met criteria for a ‘Basic’ service (psychiatrist, nurse, psychologist) and 64% met criteria for a modified ‘comprehensive’ service (Basic + nursery nurse and occupational therapist). Service provision among our programme theory domains was also variable. Greatest uniformity existed across reported collaboration with other statutory services, though even this varied considerably, with few examples of joint clinics held with maternity, for example. There were notable gaps in collaborative working with social care and voluntary sector organisations, but best practice existed in some places. Some services provided example of good collaboration between organisation through join clinical meetings, assessments and/or clinics run through community children centres or other public services (e.g. local libraries) In contrast, other components of CPMHT care were widely variable, with only 57% of services reporting any data on psychological interventions for maternal mental health and the parent–infant relationship, for example. Of those we received data from, only 41% of services between 3 and 5 types of psychological provision, and most of these were focused on maternal mental health. Few services using the core recommended outcome measures.31 Delivery of care was typically limited to clinics or in home, though some services delivered care in children’s centres. Pre-COVID, few services used remote delivery formats. Wait times likewise varied, with some services meeting wait times in < 2 weeks, but other services struggling to see clients within 11 weeks of referrals.
• In our realist evaluation (WP3), we used our taxonomy, plus the descriptive data we gathered in WP1 to purposively select 10 CPMHTs on their service component offers as highlighted from our programme theory on the effectiveness of treatment for PMH. We asked what CPMHTs were effective, for whom, and how?

As we had anticipated, the importance of co-ordinated, integrated care between CPMHTs and other relevant health providers emerged as critical for women. The presence of comprehensive CPMHTs, as compared to basic CPMHTs, facilitated integrated, co-ordinated care, leading to timely and appropriate CPMHT referrals and improved outcomes across different settings, pointing to the importance of mixed staff skill sets. This holistic approach played a pivotal role in ensuring the well-being of women, children and families, although results from WP4 suggest that additionally enhanced sharing of knowledge between health and mental health professionals is important to ensure women receive timely health interventions when needed.

Perinatal women with mental health problems expressed a strong preference for treatment that was family centred and delivered by health professionals with perinatal competence – an understanding of the complex needs they and their families faced. The manner in which treatment was delivered played a crucial role, with compassionate, open and consistent non-judgemental care leading to improved perceived relevance of treatments offered and overall higher treatment engagement. Many women commented on how CPMHTs provided them with transformative experiences of mental health treatment, interacting with women as individuals and mothers, rather than ‘labels’. A holistic approach, which typically combined pharmacological and relational/psychological treatments, proved beneficial. Particular strengths of the CPMHT were around optimising women’s medication by providing them with comprehensive specialist up-to-date information on risks and benefits of medication in the perinatal period, improving their sense of social belonging with other parents, building their social support and reducing conflict with family, helping them to build their relationships with their infant(s), and providing them with emotional management skills and supporting them to engage in valued activities. Women described how a flexible treatment approach, tailored to the distinct needs of their families and with the capacity for schedule and location adjustments (home, clinic, remote, etc.), played a pivotal role in sustaining their engagement with treatment.

For whom? Women with SMI described a notable appreciation for how the CPMHT effectively facilitated accurate medication optimisation, a crucial factor in improving their treatment outcomes which significantly contributed to their willingness and ability to adhere to or start their medication. Women with both acute and complex trauma experiences shared their concerns regarding the lack of psychological interventions targeting their trauma-related issues. They were also especially likely to report difficulties in bonding and often experienced heightened conflict in their relationships. Women with SMI and/or with histories of complex trauma consistently emphasised the value of receiving support from staff to connect with similar mothers, both within the CPMHT and in community motherhood groups.

What is the most reliable, valid and clinically feasible observational measure for routinely assessing parent–infant interaction quality in PMH services?

In WP2, we used film footage recorded at 6–8 months of age in a high-risk community sample and found evidence for the reliability and predictive validity of two observational tools suitable for routine use in perinatal clinical practice to identify parent–infant dyads who may benefit substantially from interventions designed to improve interaction quality. These measures comprise (1) a short form of the NICHD coding system, namely the NICHD-3, or (2) the PIIOS in its full form using the total score. No shorter form of PIIOS was identified with sufficient predictive validity. Optimal reliability (stability of ratings) and predictive (discriminative) validity to age 2 mental health outcomes was achieved by using 5 minutes observation for both NICHD-3 and PIIOS total score. In clinical practice, we therefore recommend filming for a 5-minute period, rather than shorter or longer periods which confer, respectively, disadvantage or no advantage in terms of predictive validity. Neither observational tool predicted longer-term mental health outcomes at school entry, but this is not surprising if one considers the many intervening events or circumstances in children’s lives that might influence outcomes between 6 and 8 months of age when parenting was assessed and school entry at age 4–5 years.

While PIIOS had broader predictive validity to age 2 internalising (emotional) and externalising (behavioural) mental health outcomes, it was developed for use with infants between 2 and 8 months which limits application in services with a broader age range of infants. The NICHD-3 which focuses on parental sensitivity, intrusiveness and positive regard (warmth) may have greater clinical utility due to its comparative brevity to train (~2 days) and code (~15 minutes), its predictive validity to both attachment security and externalising problems and application from 3 to 24 months of age which matches the broader needs of specialist PMH services. Services who have already invested in staff training on these observational tools can use them with confidence for the infants within scope developmentally at the time of assessment.

Patient and public involvement

We had strong lived experience and stakeholder representation on the grant in the form of the team, the project oversight group (two lived experience members) and our PAG, which was comprised of 12 women individuals with lived experience of PMH problems and 3 male partners. We had PPI input embedded in ESMI-II from the design and writing of the grant to supporting the development of the interview guide in WP3 to feeding back on the interpretation of findings and dissemination of the findings, including input to the content of our policy briefs and participation in our discussion panels on our stakeholder days.

The PAG met routinely four times a year, and members were available via e-mail in-between meetings. We provided costings for a service user researcher, Clare Dolman, to manage and run the PPI group (2 hours/week). Clare presented updates from the PAG at each project management meeting.

Both during the grant planning phase and in early meetings on the grant, the PAG prioritised the following areas:

• Focus on which aspects of CPMHTs work well and help women the most, highlighting improvement in symptoms and quality of life were of primary importance.
• Examine whether care plans were collaborative and were implemented.
• Examine support for fathers/wider family and assessment of their needs.
• Examine support around medication choices including preconception advice, liaison with GPs and suggested resources for women not needing specialist care.
• Examine long-term support for women following acute illnesses (e.g. post admission to MBU) and integrated recovery support.
• Examine integration of services that facilitate consistent messages and education/training of health professionals in the broader care pathway.

We also sought advice from the PAG when recruitment from individuals from diverse ethnic backgrounds was slow. Based on their advice, we updated our recruitment strategy to include social media and third-sector organisations.

Equality, diversity and inclusion

Our team, although primarily White English and female, was comprised of three individuals from ‘white, other’ backgrounds, two from East/South Asian backgrounds, and one from a Black African background. Four of these individuals were first-generation immigrants to England. At the analysis and dissemination stage, we sought to enhance input from diverse voices and brought in additional PPI from diverse backgrounds to input into our understanding of our findings. The strong interaction between our project team and our PAG has helped to ensure the meaningfulness of the findings to women, families, services and policy-makers, for example with ‘tweets’ from our stakeholder day reaching over 4500 individuals and reports from clinicians that they have been drawing on ESMI-II results in their service designs and business plans.

Our data, however, were ultimately reliant on the representativeness of the data sets from which we drew, or as in our realist evaluation, on the individuals who were in CPMHTs. Where possible, in WP3 we oversampled participants from diverse and underserved populations. Over 20.8% of our sample was from diverse background, compared to 16.7% in services, with the data from WP4 suggesting that CPMHTs are failing to adequately engage diverse populations. In WP2, we used longitudinal data from the Medical Research Council funded WCHADS. No other cohort in the UK has filmed footage of early mother–infant interactions and infant attachment and mental health outcomes in the pre-school period. The community sample was stratified for psychosocial risk by design and was also drawn from a geographic location spanning affluent to deprived neighbourhoods. The sample represented the local population well, with higher levels of socioeconomic deprivation than the national population of England (~40% in bottom quintile of UK neighbourhoods). However, only 3% families were non-White British, which although representative of the local population, means that findings reported here may not be generalisable to more ethnically diverse populations.

Impact and learning

We had excellent recruitment to the study, despite starting during the first COVID lockdown. Key factors contributing to success included the swift responsiveness of the NIHR and streamlined NHS ethics, allowing a seamless shift to fully remote recruitment and interviews in WP3.

Our focus on recruiting women who either were concluding their time with the CPMHT or had received care in the past year enabled us to compare women’s experiences before, during and after the pandemic-driven shift to remote treatment provision. This evaluation revealed both positive and negative unintended consequences resulting from COVID’s impact on service delivery, which we highlighted in rapid response briefs to the House of Lords and NHS England.

On the negative side, rapid service expansion coupled with the sudden shift to remote treatment led to significant changes in how the NHS LTP was implemented, at times slowing progress in some areas. On the positive side, the forced adoption of remote treatment delivery, although not universally appropriate, effectively overcame transportation and time barriers for some women and infants, leading to more efficient outreach.

Both before and after COVID-19, we found that CPMHTs performed optimally when they were able to deliver on the LTP aims, but there was wide variability in the extent to which services were able to incorporate these and this was often linked to CPMHTs receiving promised levels of investment at the local level.

During WP4, we encountered substantial challenges when attempting to obtain linked data from NHS Digital. From the initial data request to its actual receipt took over 24 months. These delays were partly exacerbated by the prioritisation of COVID-related linked data requests.

The quality of the data we eventually received posed some challenges that necessitated an intensive 6-month effort by our team to transform the data into a usable format. This meant that even with the original 3-year grant time frame we were left with limited time for data analysis, leading to requests for no-cost extensions and additional funding for the WP4 statistical team. These delays and extensions created intense work periods with tight deadlines, which would have been avoidable with timely data receipt.

Despite these challenges, the data ultimately proved valuable, offering critical insights for policy-makers on CPMHT access and maternal/neonate outcomes. However, we could not analyse CPMHT impact on clinical outcomes across specific PMH issues due to significant missing data on diagnosis, clinical severity and functioning at the start and end of care. This issue, which is driven by both poor recording of data in the NHS and by problems between data systems and uploading data to MHSDS, is prevalent across mental health services, and critically hampers the examination of clinical effectiveness.

While efforts are ongoing in adult mental health to define core outcome measures and incentivise regular recording of these measures, without making them mandatory, regular completion is unlikely, affecting quality-improvement and investment in all mental health services in the NHS.

We held two stakeholder events, one remote in 2021 with 69 attendees from 35 NHS and third-sector organisations and in April 2023, we presented ESMI results at a London hybrid stakeholder event. Attendees included 120 PPI, service users, CPMHT service leads, clinicians, commissioners, perinatal researchers, and policy-makers from National Health Service England and DHSC, representing over 60 organisations. The event involved presentations from the research team on the various WPs followed by panel discussions involving PPI, NHS policy-makers and clinicians. Feedback was excellent, ‘I wanted to say [the ESMI-II event] … was the most interesting conference-style event I have been to. The panel discussions really brought it all to life … [especially] the focus on barriers to access. The work itself is so impressive and needed – I am looking forward to seeing more of the results!’ [DHSC government policy lead]. Visual briefings for teams and policy-makers, plus social media distribution, have been prepared, and we are consulting with National Health Service England’s perinatal policy team on ESMI-II results.

Implications for decision-makers

We found that 9–10% of perinatal women had histories of complex and severe mental health problems – a significant proportion of birthing individuals. Our results demonstrate that CPMHTs are effective at improving access to needed and effective treatments, but pregnancy/neonate outcomes are still an area of concern. Staff training on comorbid health/mental health conditions is needed in both CPMHTs and medical settings. Collaborative working between mental health and health settings to adequately address issues of multimorbidity is critical.

Qualitative results point to the importance of perinatal competencies in staff. CPMHT staff need perinatal specific training and supervision in parental mental health, couples approaches and, critically, parent–infant interventions.

Staff’s ‘soft skills’ – being genuine, warm, non-judgemental, reliable, and providing outreach to women underpinned women’s initial and ongoing engagement with treatment. These skills were especially important to women with more complex problems and those from under-represented groups. Staff ‘soft skills’ are often unmeasured and undervalued in mental health services relative to ‘knowledge’ and direct intervention skills, despite being critical to women’s outcomes and the success of the service. Our findings point to the need for a culture change in services where ‘soft skills’ are valued as equally important to high-quality care. All parts of systems across care pathways should signal and support this to be the norm, through:

• training and education
• staffing/workloads (prioritise providing staff with time to engage in ‘soft skills’ and outreach, as this is often the first to go when services are busy)
• measurement of ‘soft skills’ and outreach
• performance management
• staff attitudes (both mental health and health providers).

Care pathways and treatment provision in pregnancy, for women who have had recent contact with secondary mental health care and for women with SMI, are strong and well-developed. There was less pathway/treatment development and greater variability in women’s experiences where problems developed postnatally, women had not had previous or recent contact with secondary mental health care, and where women struggled with PTSD, depression, anxiety, OCD and EUPD. We found CPMHTs with well-developed comprehensive teams with full investment from commissioners had stronger strategic leadership and greater clarity on treatment pathways within their teams, providing access to higher-quality evidence-based treatment for women.

Parent–infant treatment within teams was largely underdeveloped in most CPMHTs at the time of our evaluation. Provision was provided primarily by nursery nurses in most services, and evidence-based parent–infant interventions were rarely available. When parent–infant support was available, women and families reported notably improved experiences of their time with the CPMHT. Since our evaluation, more teams have expanded to meet the LTP flexible aim for expansion to 24 months and have invested in training some staff in parent–infant interventions, though there is still considerable variability among services. For CPMHTs to achieve the economic and health promises of preventing negative child outcomes, our findings suggest it is important to realise the full implementation of parent–infant provision in services. A critical component to ensuring appropriate recognition of parent–infant problems and treatment is having feasible outcomes tools. In WP2, we demonstrated the clinical validity and feasibility of two observational tools of the parent–infant relationship. Though the PIIOS showed good predictive ability, it currently can only be used on a narrow infant age cohort. The NICHD-3 measure is brief, easier to train and administer in busy clinical settings and is valid for use in infants up to age 2, making it more widely appropriate for specialist PMH teams and parent–infant services. Its extensive use as a research tool internationally makes it highly appropriate for future evaluations of the effectiveness and cost-effectiveness of interventions and mental health services.

Women and families both reported that women’s treatment experiences were improved when, with the woman’s consent, couples and families were involved in their treatment. Though staff described they sometimes felt unsure about how to engage families, our results highlighted a number of mechanisms through which to do this, consistent with recommendations from NHS England’s partners and family members’ good practice guide.

Recommendations for policy-makers, commissioners and service leads:

• Perinatal parents and their infants are at high risk and are less likely to access mental health services than at other periods of life. Perinatal-specific services increase access to timely, evidence-based treatment but require investment to deliver on their aims.
• CPMHT and health staff require training and close collaborative working with perinatal women with multimorbidities to reduce negative childbirth, obstetric and neonate outcomes.
• To ensure perinatal competences in CPMHT staff, ongoing perinatal-specific training and supervision is required.
• Paired measurement of parental and child-infant outcomes is critical for demonstrating the effectiveness and cost-effectiveness of services both in the short and longer term. Our study showed the NICHD-3 is a feasible and valid observational tool for assessing the parent–infant relationship.
• Family involvement in treatment improved treatment engagement and perceived outcomes.
• Diverse ethnic groups are less likely to access mental health treatment in statutory settings. Lack of trust and a history of discrimination are critical factors to overcome and require dedicated outreach from services into communities to (re)build trust.

Research recommendations

1. We found increases in neonate risk in areas with CPMHTs. We do not know the biological, social and healthcare determinants of these effects.
2. Mental health outcomes and diagnostic data are largely unavailable in existing mental health data sets. Mandatory reporting of this (and other important contextual factors such as domestic and sexual abuse) would significantly improve the quality of data. CPMHTs have also had limited time to fully embed themselves into practice, making it difficult to retrospectively assess their impact on women and their babies. Prospective studies that recruit women from case identification and follow them through the perinatal period and with longer-term follow-up assessing mental health, health and child outcomes are needed to reliably assess the effectiveness and cost-effectiveness of CPMHTs.
3. There is a growing research base about barriers and facilitators to mental health care and treatment preferences in underserved and diverse parents. Research could now focus on the development of feasible and effective interventions tailored for underserved and diverse perinatal populations and how best to implement them within community specialist PMH teams.
4. Two specific groups require further study. From this project, we do not know what happens to women who have PMH problems but do not reach the threshold of being seen by CPMHTs. Further, complex trauma and loss presentations are common in PMH, though often are not identified because of concerns about stigma associated with this presentation. CPMHTs struggled with decision-making about what treatments to provide within their relatively short treatment windows (pregnancy to 24 months postnatal). Research is needed on women who have moderate to severe but non-complex mental health problems (and would be eligible to be seen in primary care mental health) and examining the effects of models of care incorporating briefer perinatally focused interventions administered alongside co-ordinated care in community mental health teams on parental emotion regulation, interpersonal conflict and parenting outcomes will give services critical information about how to provide best care to perinatal parents with complex trauma and loss.
5. We have established that both NICHD-3 and PIIOS are reliable and valid tools to identify parent–infant dyads who require parent–infant focused interventions in order to improve infant mental health outcomes. Future research needs to evaluate the measures’ sensitivity to change following intervention.