Introduction

Preterm birth (PTB, birth at < 37 weeks’ gestation) remains a crucial issue in improving outcomes and quality of maternal, newborn and child health care. One in ten babies worldwide are born too soon and over a million die from complications associated to their prematurity [1]. Many of those who survive can face adverse lifelong health consequences which impact on families, societies, and health systems [2]. Despite efforts to reduce the prevalence, improve clinical management and decrease neonatal mortality and morbidity, PTBs continue to rise in most countries [3]. Although most PTBs are spontaneous and only a small proportion are provider-initiated due to maternal or fetal reasons, the cause is still unknown in up to half of the cases and involves multiple and overlapping factors (e.g. infections, chronic diseases, obstetric history such as previous PTB, smoking, psychosocial stress, domestic violence, social deprivation) [4]. Thus, achievement of public health strategies to prevent PTBs has been challenging and developing and evaluating specific maternity packages of care, that incorporate single effective interventions designed to address prematurity, remains a public health priority [5,6]. Models of midwifery continuity of care have been the only health system intervention shown to reduce PTB and increase perinatal survival, but no evidence exists for women with risk factors for PTB [7,8].

Complex interventions, such as continuity models and other maternity care packages, are composed of multiple and different interacting components, and non-linear causal pathways [9]. Thus, their impact is dependent on the degree to which an intervention is implemented as intended, in specific contexts and population groups. A clear understanding of the most appropriate strategies to deliver them and how they may work and be adapted in varying contexts is vital [10]. Randomised controlled trials (RCTs) are the most rigorous way to evaluate the clinical impact of interventions, regardless of their complexity, but are often criticised for providing little or no information about why and how an intervention worked (or not) and the context in which it was delivered [9]. Generalisability of findings may be limited, and interventions may be rejected if found to be ineffective or without a robust evaluation. Furthermore, knowing which components of an intervention are core, which components can be modified, and how they are delivered to produce an effect in a certain population is crucial for the resulting intervention to be adapted or scaled-up [11].

Multiple guides and tools are available to integrate implementation science concepts and methods to facilitate the identification and evaluation of factors likely to affect implementation efforts and evaluate implementation outcomes alongside the clinical impact [9,12,13]. Integrating mixed methods is crucial to understand why and how an intervention may work, how well it is delivered, for whom and in which circumstances. Effectiveness-implementation hybrid type 2 trials take a dual focus on assessing the effectiveness of an intervention as well as evaluating implementation efforts [14] and require a multidisciplinary approach [12]. These designs are increasingly recognised in the evaluation of health interventions and have the potential to speed up the translation of research findings into routine practice, ultimately enhancing public health impact [14], however their application in maternal and newborn health is very limited [15].

The Pilot study Of midwifery Practice in Preterm birth Including women’s Experiences (POPPIE) study is a hybrid type 2 pilot RCT of a model of midwifery continuity of care linked with specialist obstetric care for women at increased risk of PTB in London (United Kingdom), and was accompanied by a nested mixed-methods evaluation which was informed by the Medical Research Council (MRC) guideline for complex interventions [9]. The aim of POPPIE was to determine whether this care model was feasible and could improve a composite outcome of timely and appropriate interventions provided for the prevention and/or management of preterm labour and birth. These included: antibiotics for suspected/confirmed urinary tract infections, transvaginal scan assessments of the cervix, fetal fibronectin assessments, cerclage insertion, progesterone administration, corticosteroid administration, magnesium sulphate administration, admission for observation, in-utero transfer, and smoking cessation and domestic violence referrals.

It was hypothesized that initiation of these treatments would occur earlier and at a more appropriate time in the intervention group, as the POPPIE model would improve quality of care by providing a trusted safety net with midwives ensuring better care coordination, referral, and access to additional services [16] and by moderating the effects of women’s stress on her health through trust and confidence that relational continuity generates [17]. Together, this would result in more women receiving timely and safer care leading to more opportunities for early identification of complications and subsequent management, thus improving prevention, detection and management of preterm labour and birth. [13,18]. A logic model was created in collaboration with local stakeholders, healthcare providers (HCP) and service users (Fig 1) to present these assumptions, processes, and anticipated outcomes, and inform key areas for evaluation in this study.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1.

https://doi.org/10.1371/journal.pone.0279695.g001

We report the evaluation of the implementation of the POPPIE model, describe the local context in which it was delivered, the mechanisms of impact by which the delivered model may produce change, and integrate results to determine whether the effect of POPPIE on the primary outcome in our setting could be explained. Specific objectives included:

• Implementation outcomes: to evaluate the perceived fit of POPPIE to address local needs (appropriateness) and intention to try its use (adoption); to determine whether POPPIE could be successfully carried out within this clinical setting (feasibility); to evaluate satisfaction among women, healthcare professionals and other key stakeholders with various aspects of the intervention (acceptability); to evaluate whether the intervention was implemented as outlined in the protocol and/or whether adaptations were made (fidelity); and to evaluate how well the intervention was integrated within the maternity system (penetration) and the extent to which it was maintained over time (sustainability)
• Context: to describe contextual factors which impact upon’ implementation efforts.
• Mechanisms: to explore if and how the intervention impacted on maternity care in this setting and identify possible reasons for this.
• Inter-relationships: to explore relationships between core implementation and process outcomes and the primary outcome.

Methods

Data collection

Overall, clinical and process outcome data were abstracted from medical records and electronic data systems, and implementation data (e.g. determinants of implementation, implementation outcomes and mechanisms) were collected from meeting records and key documents pre and mid implementation; and from postnatal surveys and qualitative interviews at 6–9, 14–18 months and post-implementation (22 months after implementation). Data collection for each outcome and determinants including measurement operationalisation are illustrated in Fig 2, and further detailed in S1 Table. The estimation of costs and resources are important outcomes of the implementation evaluation, but the economic evaluation for POPPIE will be reported separately.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2.

https://doi.org/10.1371/journal.pone.0279695.g002

Data from two business cases and records were collected from key documents (both public and provided by the implementation site e.g. local health profile, internal reports, protocols) and twelve executive and project board meetings (dates, duration, attendees’ roles, milestones and deliverables, timeframe, budget, risks) with directors, clinical managers, commissioners, public health specialists, academics and service users held over more than two years pre-implementation. Additional data were collected from more than fourteen monthly group meetings mid- and post-implementation with clinicians, researchers, implementers and users’ representatives (e.g. date, duration, attendees’ roles, project plan, workforce and funding arrangements, training needs, commitments, patient and public involvement, action points).

Postnatal surveys were completed by women at six-eight weeks after birth either online (through MedScinet) or by post (and posteriorly incorporated into MedScinet by research assistants). Up to three reminder phone calls and/or texts were sent every two weeks to non-responders. The survey, piloted with an advisory group (which included parents of preterm babies), was partially based on questionnaires from previous studies of models of maternity care and surveys conducted in the UK and included standardised scales to measure different aspects of women’s experiences of maternity care [27,28].

Semi-structured interviews were used as this style allows for a key list of questions to be addressed whilst providing flexibility for the interviewer to follow-up on points made by participants which are pertinent to their experiences [29]. Women (n = 30) were asked to share their experiences in their maternity care journey; and HCPs and local stakeholders (n = 23) were asked about their experiences in designing, implementing and/or providing the new care pathway. Interviews were conducted based on participants’ preferences (few over the phone and most face to face at home [for women] and hospital [for HCP/Stakeholders]) by three researchers with training and/or experience in qualitative methods not colleagues of participants and external to the hospital service provision. Reflective diaries were completed after each interview to encourage being continually reflexive [30]. Interviews lasted one hour on average and were digitally recorded and transcribed verbatim.

Results

Between May 2017 and September 2018, 334 women were recruited to the main pilot trial; 169 women were allocated to the POPPIE group and 165 to the standard group. Of the 149 women followed-up in the POPPIE group, 90 (60.4%) completed the postnatal survey; and out of the 154 women followed-up in the standard group, 78 (49.4%) completed the postnatal survey. The overall survey response rate in both groups was 55%. A total of 20 women in the POPPIE group, 29 women in the standard group and 28 HCPs and key stakeholders were invited for an interview; and 16, 14 and 23 accepted (acceptance rate of 80%, 48.3% and 82.1% respectively). The overall interview acceptance rate was 70.1%. An overview of the characteristics of the qualitative participants are shown in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of interview participants.

https://doi.org/10.1371/journal.pone.0279695.t001

Implementation outcomes

Appropriateness.

Most local stakeholders recognised PTB as a leading cause of child mortality in the area. A pre-implementation audit conducted by public health specialists confirmed a high number of early births and identified local complex needs. Consequently, the majority of HCP and stakeholders perceived the introduction of a continuity of care model as a fundamental change with possible benefits on local maternity services and preterm births, and decision to provide funding for a midwife leader post to lead organisational change was made. The most illustrative quotes on appropriateness and the rest of implementation outcomes are presented in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Selected qualitative quotes to illustrate implementation outcomes.

https://doi.org/10.1371/journal.pone.0279695.t002

Adoption.

There was a clear intention by most local stakeholders and HCPs to try to implement the model. However, some stakeholders did acknowledge numerous and tedious steps required for adopting the intervention in clinical practice due to lack of other continuity models and maternity service reconfiguration. The midwife leader was the only post externally funded and, initially, there were particular concerns on creating a new team with no additional staff (but a relocation of an already limited workforce or hospital funding for core staff recruitment) and lack of awareness of how a continuity team functions within the maternity care system of the hospital.

Feasibility.

A total 553 women were screened for trial participation: 334 women met all inclusion criteria and were randomised (169 POPPIE group; 165 standard group), 123 did not meet the inclusion criteria and 96 declined to participate [23]. Eleven (6.5%) women in POPPIE and eight (4.8%) in standard group either moved out of area antenatally or postnatally or experienced in-utero transfers in threatened preterm labour to a tertiary hospital. Data for analysis was collected from 168 women (one woman withdrew, with consent to use all data withdrawn) in the POPPIE group, and 163 women in the standard group (two women lost to follow up) [23].

Fidelity.

The named or partner midwife provided nearly 75% of antenatal and postnatal visits for women in the POPPIE group and was present at birth for nearly 57% of those women. A POPPIE midwife provided more than 85% of antenatal and postnatal visits and was present at birth for more than 80% of women in the POPPIE group [23]. Nearly 90% of POPPIE midwives submitted monthly audits to monitor own clinical activity and quality of continuity. Qualitative interview data on fidelity found that the majority of women in the POPPIE group met at least two to three midwives (and more than a third met most team members), they had a known midwife at birth, and a team member was available most times during pregnancy, birth, and the postnatal period. Qualitative interview data from HCPs showed that certain aspects of the intervention were adapted and tailored to meet certain needs in certain circumstances such as the organization of team’s on-calls from two to one per week during busy or under-staff periods, or the recommendation of further training on bereavement care.

Acceptability.

Nearly 97% of women who completed the postnatal survey in the intervention group reported they would prefer a POPPIE midwife to be the main person for their maternity care. All women interviewed described their satisfaction with various aspects of the continuity model such as having direct access to their midwife and the team (mobile number to call and text anytime; flexibility in timing and location of visits); building a relationship over time (personalised advice, involvement of the family) and receiving coordinated care (consistency of information, practical support). Most midwives delivering the intervention highlighted positive elements of working in the POPPIE team including autonomy over own diary; job satisfaction through providing continuity and being part of research; a supportive team, and team culture, and access to training and development. Some reported flexible working and weekly on calls enhanced their work-life balance, whilst others reported exactly the same features disrupted their work-life balance, particularly during periods of sickness or annual leave, when the flexibility that the team depended on causes midwives to ‘stand in’ for each other on days off and to volunteer for ‘informal’ or ‘silent’ on calls. Overall, two midwives left the team but new team members were promptly recruited to replace them.

Penetration.

When assessing the integration of the POPPIE model within the maternity services, many HCPs working on the new team encountered issues at the ‘boundaries’ between themselves and established services (especially labour ward and community teams) and potentially due to lack of knowledge about how the team worked (e.g. philosophy of care, visits, inductions of labour, on-calls). A couple of women also perceived a disconnection between their midwives and core hospital staff; they described care as being disjointed and observed unpleasant behaviours by core hospital staff towards POPPIE midwives.

Sustainability.

Despite initial challenges in penetration, long term support for the continuity model remained high, amongst women, HCPs and stakeholders. Clinical managers recognised the importance of the support received by the POPPIE midwives when needed, the improved experiences of care among women looking after them and lack of complains from the team. A follow up interview with one of the stakeholders reported that the continuity model was sustained post-implementation and adapted to provide care to a mixed risk caseload of women, and felt current maternal policy on continuity of care facilitated the scale up of four continuity models targeting other local populations (e.g. vulnerable or socially disadvantaged women).

Context of implementation

Intervention characteristics.

Data from the initial business case and project meetings proved the intervention initially entered into the Trust through external sources (university and local public health department). However, the POPPIE model was then designed in partnership with researchers, clinical managerial teams, maternity commissioners, public health specialists and service users, after discussing at various meetings the strong evidence, quality, and cost-effectiveness behind continuity models. To avoid centrally driven and controlled leadership, researchers built a coalition between stakeholders through information sharing and decision making, conducted local consensus discussions to develop and agree on the characteristics of the POPPIE model (e.g. team size, targeted population), and also involved executive boards in the intervention development. A patient and public involvement group was actively involved in discussions of the intervention and outcome measures important to women at risk of PTB and their partners through biannual meetings, social media groups, and service user events. A relative flexibility of the POPPIE model was agreed but guarantee of core components (e.g. relationship continuity, autonomy over own work) was important. Thus, adaptability was promoted by allowing certain aspects of the intervention to be adapted and tailored (e.g. organisation of team’s on-calls, rota and workload; skills mixed; further training). A six-month piloting phase, led by the team leader, enabled researchers and clinicians to build experience and expertise in the complex reconfiguration of local maternity services before commenting the trial (see Table 3 for quotes to illustrate context of implementation).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Selected quotes to illustrate determinants of context.

https://doi.org/10.1371/journal.pone.0279695.t003

Outer setting.

The review of local documents revealed that approximately 30% of the local population are from Black and minority ethnic groups and levels of deprivation were higher than the national average. Health profile reports from Public Health England at pre-trial time found that there were more than 6000 births in the hospital and approximately 8.2% of women delivered preterm babies, which is slightly above the UK rate of 7.8% [40]. Important financial implications related to local PTBs and associated child mortality were acknowledged by most stakeholders. In addition, findings from the pre-implementation audit identified a care and quality gap and encouraged maternity commissioners to fund clinical innovation, developing a financial formula for the midwife leader post to implement the new care pathway. National maternity policy set up just before the pilot trial commenced and recommending scale up of high-quality continuity of care [41] was a strong facilitator for the implementation and sustainability of POPPIE, as well as the NHS Long Term Plan endeavouring continuity of care for most women [42], and the WHO antenatal and intrapartum guidelines for positive pregnancy experiences highlighting the benefits of continuity in settings with well-trained midwives [43,44]. National policy might have affected clinical practice in standard care group, particularly during the antenatal period when more than half of the women were seen by the same two midwives. The implementation of POPPIE was also influenced by a robust network between maternity services and other organisations (e.g. maternity networks, commissioning and strategic clinical groups, university). There was presence of mimetic pressure from competing organizations who already implemented similar care models and events were organized for management and leadership tips from external managers who had successfully reconfigured services at other hospitals.

Inner setting.

Local documents reported that the hospital had quality improvement processes in place (e.g. sharing on maternity key performance indicators, staff survey results) and its vision and mission statement included three key principles underpinning the care delivered at every stage of their maternity pathway: informed choice, accessibility and continuity of care [45]. Although there were no existing continuity models in the hospital, there was an organisational commitment and a shared vision among hospital managers, commissioners, and service-users of the importance of implementing this care pathway. Thus, the organisation mandated a change by highlighting in the maternity services specification report their formal commitment to deliver high quality continuity of care for women. A piloting phase of 6–8 months, led by the midwife team leader, enabled stakeholders and care providers to build experience and expertise in reconfiguring local services before commenting the trial (e.g. in terms of workforce, pathway for referrals). To actively support adoption and raise awareness of sustainability of similar models, multiple implementation strategies were used e.g. sharing of experiences with other hospital managers; using advisory boards and workshops to monitor progress and review risks; and sharing educational materials (e.g. standard operating guidelines continuity of care toolkit, preterm birth prevention). Similarly, to address some of the initial penetration challenges and enhance positive relationships with managers and colleagues working in other services, other strategies were used e.g. attendance to clinical, community and multidisciplinary meetings; sharing of relevant team materials (e.g. operating guidelines); and support of core maternity services when possible.

Interviews with HCPs and analysis of local documents found a lack of research culture in the maternity services and the need for innovation was recognised. Thus, the organisation mandated a change by highlighting in the maternity services specification report their formal commitment to deliver high quality continuity of care for women and started organisational changes (e.g. layout change of several rooms to provide a small office space for the team, new equipment). The implementation of POPPIE meant the hospital became actively engaged in maternal health research and started to participate in clinical trials in collaboration with other sites. At the same time the trial was being set up, a preterm surveillance clinic was also being established as the main obstetric care pathway for women identified at risk of preterm birth and this could have improved clinical management and quality of care for the study population.

Characteristics of individuals involved.

In stakeholder group discussions, continuity of care was not supported by all HCP and some midwives did not want to work in the context of continuity models with the belief that they would experience burnout (e.g. with on-call system), they would perceive difficulties in autonomous decision making (e.g. particularly if only worked in routine models) or they would be routinely pulled to cover conventional service needs (e.g. labour ward, community). Recruiting and retaining midwives in Trusts is often difficult and recruiting midwives to implement these care models can be a challenge [35]. Two job advertisements were required and promotion of a learning culture, development and enhancement of research and clinical skills though the continuity model was emphasised. To avoid potential inter-professional tensions, both stakeholders and the team proposed it would be beneficial to introduce medical and midwifery teams to work together on the implementation project, with the aim of promoting positive relationships amongst and between teams.

Process of implementation.

Planning, developing, and piloting the care model in advance enabled local stakeholders to work together to build capacity and expertise and provided time to reflect and support each other’s learning before commencing the trial, thus promoting adoption of the POPPIE model. Weekly and monthly research and clinical meetings were scheduled to find practical solutions to possible implementation challenges for women, midwives, and healthcare service. For quality and monitoring purposes, daily/queries logs with agreed action plans were in place, and monthly audit tools were designed to monitor fidelity of the intervention. Throughout the project a stakeholder engagement strategy was utilised a combination of social marketing, training and education, and other similar activities and involvement of team leaders, champions, project coordinators and services users.

Mechanisms of action

Midwife-woman relationship.

Women in the POPPIE group reported how important it was for them to have a trusting relationship with a midwife who provided individualized and respectful care for them and their families; and how reassuring it was to have telephone access 24/7 to a team, with whom they were familiar, for personal advice and support. They felt at the centre of their care, involved in discussions and informed choices, with no need to repeat own story every time which made them feel safer and calmer, and appreciated the advocacy through their midwives e.g. during birth choices or interactions with other health professionals. The importance of relationships as a pathway for safe and quality of care was also recognized by midwives and stakeholders (quotes to illustrate mechanisms in Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Selected quotes to illustrate mechanisms of action.

https://doi.org/10.1371/journal.pone.0279695.t004

Processes and clinical pathways.

Most women accessed maternity care relatively early (mean gestation at booking 10.2 weeks, SD 2.7). However, timing of booking to care is unlikely to be affected by the intervention as nearly all women were recruited after the first antenatal visit with a community midwife or scan appointment (mean gestation at randomisation 15.1 weeks, SD 3.5). Women felt supported with access to midwives at any time and more informal and flexible visits (e.g. more time, home visits, re-arrangements) and reported they could openly discuss more sensitive or personal circumstances (e.g. vaginal infections, mental illness). The majority of these women had medical and/or obstetric risk factors and reported experiences of management of complications (e.g. urine infections, induction of labour, pre-eclampsia); and highlighted how timely interventions and referrals to other professionals were done. Details of process outcomes (e.g. total number of visits, total inpatient nights, total number of referrals) are provided in S4 Table. Women in the POPPIE group were more likely to be referred to the safeguarding team–a multi-agency team that protects women and their unborn child’s health, wellbeing and human rights to avoid harm, abuse and neglect (main reasons being housing, social care, and mental health) although safeguarding referrals and overall antenatal referrals were not related with the primary outcome (RR 0.62, 0.36 to 1.06; and RR 1.25, 0.78 to 1.98 respectively). They were also more likely to have more postnatal visits, and women highlighted during the interviews the practical and emotional support received during the postnatal period (especially with breastfeeding). Some women and midwives reported feelings of frustration due to poor organization, communication and cooperative relationships between HCPs working in different wards or departments (e.g. hospital wards, community, neonatal units) particularly at time of labour and after birth.

System resources.

Lack of sufficient and/or trained staff working in a stretched, overworked and underfunded NHS was acknowledged by few women and may have impacted on delays on receiving support and elective and emergency interventions (e.g. inductions, cesarean sections).

Discussion

This study reports the results of a mixed methods evaluation alongside a hybrid type 2 pilot RCT. The introduction of a model of midwifery continuity of care linked with specialist obstetric care for women at increased risk of PTB (POPPIE) was found to be appropriate to address local needs in a UK inner-city maternity service despite challenges in early adoption (e.g. lack of similar models, service reconfiguration). Collaborative working and availability of additional funding certainly facilitated adoption efforts. We demonstrated it is feasible to set up and maintain appropriate levels of fidelity of the POPPIE model with overall good acceptability among women and HCPs in both our quantitative and qualitative analysis. The model was eventually embedded within established maternity services and, after the trial, it was sustained and adapted to provide care to a mixed risk caseload of women with further scale up of other continuity models. Process outcomes were also similar in both models of care. Overall, we found no association between specific implementation and process measures and the primary outcome.

Implementation fidelity, thought to be critical to successful translation of evidence-based interventions into practice [46], was prospectively identified as a key measure used to address and change implementation concerns during the trial to maximise the likelihood of potential scale-up of similar continuity models in other contexts. We retained core components of the intervention in terms of relational continuity (e.g. visits with named/partner/team midwife, midwife at birth) and demonstrated that it was possible to adapt certain components of the model (e.g. on-calls, training) whilst maintaining a very high proportion of the intervention fidelity and acceptability among women. Although support for the POPPIE model also remained high throughout implementation among midwives and clinical managers, work-life balance was sometimes hard to maintain with “silent” on calls for labour care and cover for each other on days off, particularly during busy periods, and this could have potentially influenced implementation fidelity. These findings are aligned with inconsistent evidence on the impact of flexible working in continuity models among midwives, some studies suggest it increases wellbeing and satisfaction [47,48] whereas others suggest it increases burnout [49,50]. It is possible that these models may work for some midwives but not others (e.g. with caring responsibilities) [51]. Although the extent of support midwives have (e.g. with childcare, specially overnight or during unsocial hours) and their individual ability to ‘relax’ or ‘switch off’ during down time is likely to play an important role.

Despite multiple strategies adopted, the embedment of the model within existing services required longer than expected, and collaboration and good communication among HCPs were crucial to address issues in relation to how the POPPIE team worked. Although this was not very surprising; until recently, there was very little guideline on implementation and scale-up of continuity models, and the introduction of a first continuity model in our context, involved a complex, large-scale transformation of the organisation of maternity services requiring whole system support and close alignment between top and frontline leadership [13]. Eventually, maternal policy on continuity of care strongly facilitated the penetration and sustainability of the POPPIE model with the scale up of additional continuity teams planned based on other local women’s needs.

This is one of the few studies in maternity care (and the first of a midwifery continuity of care model) that rigorously uses implementation frameworks and mixed methods research to examine multiple process measures and analyse them alongside the primary outcome [15]. We demonstrated that evaluation of implementation and process outcomes alongside a pilot trial is feasible and valuable in understanding quality and quantity of diverse aspects of the implementation and exploring potential mechanisms of impact in our specific setting. We provide invaluable learning on implementation strategies and scale-up of continuity models, crucial for NHS England plans to expand midwifery continuity of care pathways for the majority of women by 2025 [42].

Although high quality implementation has previously been associated with better outcomes on individual level in health promotion programmes, it may not be the case for organisational-level implementation measured in a multidisciplinary healthcare context [52]. Actually, aligned with larger studies of maternal and primary health interventions [31,53] we did not find any correlation between core implementation outcomes and the primary outcome. However, this must be interpreted with caution; this was a trial designed for feasibility and with limited power to detect significant improvements in clinical or processes outcomes. Therefore, it is possible there was limited power to detect a significant relationship between the primary outcome and implementation measures. Understanding variability in implementation of complex interventions is important and future research is needed to determine how useful existing or new ranking approaches are for measuring implementation strength.

Nearly a third of women in the pilot study were from ethnic minority groups, more than two thirds lived in areas of social deprivation and more than a quarter had at least one pre-existing medical condition and multiple obstetric risk factors for PTB [22]. Hypothesised mechanisms as to how the model might work (improved care coordination and referral with engagement and access) do appear to have limited influence where there are pathological physiological mechanisms influencing PTB. It is possible that relational continuity would be more likely to improve outcomes in other populations suffering social determinants of PTB such as socially disadvantaged communities (e.g women who find services hard to access). Some mechanisms of impact identified during the qualitative evaluation (e.g. access, safety, trust, advocacy) may be limited to a part of the study population group due to limited socio-demographic and social risk diversity among interview participants [54]. Although it is known that people from Black and minority ethnic and socially disadvantaged groups are less likely to engage in research follow ups [55]. Our findings are similar to a mixed-methods comparison of women’s experiences that found that POPPIE improved experiences of trust, safety and quality of care compared to standard care, but further research should focus on women with more complex needs who have multiple social risk factors [54].

It is also possible that community-based models of care that focus on local people and communities and enhance place-based level of health and social care services can have a different impact on clinical and processes outcomes and inequalities when compared to hospital-based care models like POPPIE [56,57]. Recent pioneering Australian research with Indigenous women has shown an effect on preterm birth of a collaborative model of midwifery continuity of care integrated with family services and a community-based hub [58]. Aligned with current maternal policy on midwifery continuity of care models to improve safety and quality of care for women living in diverse and socially disadvantage areas, further investigation is required to evaluate the role of community-based models among women with complex social factors and vulnerability and to understand their experiences of care, potential mechanisms, and the long term impact. To do so, supporting and increasing community participatory research of diverse and socially disadvantaged groups is a key initial step [59].

Conclusion

To our knowledge, POPPIE is the first hybrid type 2 pilot trial of a model of midwifery continuity of care for women at risk of preterm birth London (UK) that has evaluated core implementation and processes outcomes, using a mixed-methods approach, and integrating these with the primary outcome in order to understand the findings in our context. We have demonstrated that, overall, the POPPIE model of care was appropriate, feasible and implemented with high fidelity; was acceptable among women and HCPs; and was embedded, adapted, and sustained in maternity services. As expected for a pilot study focused on feasibility of a larger trial and with limited statistical power, there were no differences in the primary outcome and there was no association between implementation outcomes and the primary outcome. Larger appropriately powered trials considering primary outcomes such as preterm birth are feasible and needed, including in community settings, to evaluate the impact and implementation of continuity models in other populations such as women with social disadvantage and vulnerability.

Supporting information

Acknowledgments

We thank the POPPIE Pilot co-investigators and collaborators including Debra Bick (University of Warwick); Annette L Briley (Flinders University); Andy Healey (King’s College London); Haode Wang (York University); Lia Brigante, Sergio A Silverio, Claire Singh, Rachel Tribe (King’s College London); Helen Knower, Jackie Moulla, Mahishee Mehta, Chloe Saad and the POPPIE midwifery team (Lewisham and Greenwich NHS Trust). We also thank the independent trial steering committee, Nigel Simpson (chair, University of Leeds and Leeds Teaching Hospital NHS Trust), Julia Sanders (Cardiff University), Soo Downe (University of Central Lancashire), and David Edwards (King’s College London, Evelina London Children’s Hospital), and the lay advisor contributing to patient and public involvement and engagement, Mary Newburn. We thank Lia Brigante and Chloe Saad for collecting some of the qualitative and survey data, and all women, healthcare providers and stakeholders for participating in the study.