A novel peer-support intervention using motivational interviewing for breastfeeding maintenance:a UK feasibility study

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/18/05. The contractual start date was in September 2014. The draft report began editorial review in February 2017 and was accepted for publication in August 2017. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Billie Hunter’s professorship is partly funded by the Royal College of Midwives.

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Paranjothy et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2017 Queen’s Printer and Controller of HMSO

Breastfeeding rates in the UK

The benefits of breastfeeding for the short- and longer-term health and well-being of babies and mothers in low- and high-income countries are well established. 1 If 45% of women in the UK exclusively breastfed for 4 months, it is estimated that at least £17M could be saved in treatment costs annually for acute illness in infants, in addition to incremental benefits over the lifetime of each annual cohort of first-time mothers. 2

The UK has exceptionally low breastfeeding rates. 1 The 2010 Infant Feeding Survey3 indicated that the majority of mothers in the UK (81%) initiate breastfeeding but that around two-thirds of these women stop breastfeeding before 6 weeks. For most women this is earlier than planned. 1 Only 1% of mothers in the UK currently exclusively breastfeed for 6 months,3 which is the World Health Organization (WHO)-recommended duration. 4 In 2010, eight in 10 mothers who stopped breastfeeding in the first 6 weeks did so before they had planned to, whereas, over the first 9 months, around three-quarters of mothers who stopped had intended to continue for longer. 3 Women’s experience of breastfeeding is not always straightforward, with around one-third reporting problems. In the first weeks, the most commonly given reasons for stopping breastfeeding are a perception of ‘insufficient milk’, a perception that the baby is rejecting the breast, ‘painful breasts/nipples’ and finding that breastfeeding takes too long or is tiring. 1

There are marked inequalities in breastfeeding rates. Geographically, low breastfeeding rates correlate with higher indices of social deprivation. 5 Younger mothers (aged < 20 years) of white British ethnicity and of lower socioeconomic status are less likely to start or continue breastfeeding beyond 6 weeks. 1 Midwives and health visitors provide professional support for breastfeeding and the majority of maternity units and health-visiting services are now working towards the United Nations International Children’s Emergency Fund (UNICEF)’s UK Baby Friendly Initiative (BFI) standards for support with infant feeding. 6 Nonetheless, mothers who give birth in the UK frequently feel underprepared for breastfeeding and many mothers who might benefit from breastfeeding support do not access professional help. 7

Medical conditions that prevent women from breastfeeding are rare;8 infant feeding decisions involve a complex interplay of psychological, social and cultural factors. 9 At the level of the individual mother, motivation to breastfeed and breastfeeding self-efficacy are associated with continuation of breastfeeding. 10–13 However, a broader set of social and societal drivers are at play, many of which can interact with individual mothers’ confidence, self-efficacy and determination to breastfeed. Extensive marketing of formula milk can denormalise breastfeeding;14 social attitudes and norms, including attitudes and norms of experiencing breastfeeding vicariously, can affect mothers’ confidence; and women’s work and employment conditions can make decisions to continue breastfeeding more difficult. 9,15

Women who are supported and encouraged to breastfeed by key members of their social network are more likely to start and continue breastfeeding for longer. 16 A review of the literature on differences between mothers who continue to breastfeed until 6 months and those who stop indicated that feeding intention and self-efficacy are inter-related with factors relating to social support. 17 Several UK studies have confirmed that attitudes, perceptions and experiences of immediate family members and friends have a strong influence on breastfeeding. 18–22 Mothers benefit from being part of a supportive community of other mothers who breastfeed,23 and women who have friends who have breastfed are more likely to breastfeed their own baby. 1 On the other hand, negative or mixed messages from partners, family, friends and health-care professionals can be confusing and can undermine decisions to breastfeed. 14,22,24 New approaches to support women who are at the highest risk of not continuing breastfeeding are urgently needed. There is a need for interventions that make appropriate use of theory and take into account the circumstances of women who are least likely to continue breastfeeding.

Breastfeeding peer-support interventions

In her concept analysis of ‘peer-support’ interventions, as applied to a wide range of health topics, Dennis25 notes that peer-support interventions seek to extend ‘natural embedded social networks and complement professional health services’. Dennis25 defines peer support as ‘the provision of emotional, appraisal and motivational assistance by a created social network member who possesses experiential knowledge of a specific behaviour or stressor, and has similar characteristics to the target population’. The importance of ‘similar characteristics’ relates to the principle of homophily,26 a key underpinning idea for peer-support interventions. This is the principle that (health education) messages will be more credible and support offered more acceptable because the peer delivering the message and offering the support is perceived by the recipient as being in some way similar to him- or herself – someone who is experiencing or who has experienced the problem being addressed and who shares the same frame of reference in terms of wider social and cultural norms and values.

Breastfeeding peer-support (BFPS) interventions that have been subject to experimental study have paid varying levels of weight to the principle of homophily. Peer supporters tended to be women who had experience of breastfeeding, in some cases from a similar sociodemographic and cultural background to the women whom they were supporting. This principle is itself linked to social learning theory,26 the idea that individuals compare themselves with others who occupy a social role to which they aspire and posits that people learn from one another through mechanisms of observation, imitation and modelling. Compared with health-care professionals, peer supporters may be more approachable, provide role models that mothers can relate to and have direct experience of the challenges of breastfeeding within a social context where it may not be the norm.

A second key underpinning principle for BFPS is the theory of social support, used to explain the ways in which social networks help individuals to manage stressful events. Four types of social support have been distinguished:27 emotional support, instrumental support, informational support and appraisal support. Informational support involves advice and suggestions that can be used to solve problems. Emotional support comes from sharing life experiences and providing empathy, love and caring built on relationships of trust. Instrumental support consists of providing tangible aid and services. Appraisal support facilitates self-evaluation through constructive feedback, for example through motivational interviewing (MI). 28 Intervention theorists distinguish between perceived support (i.e. the mother knows the help is there if she needs it and that in itself helps her to get through a stressful experience) and received support (direct interaction with the peer). The perception that help will be available when needed may be an attribute of BFPS interventions. 29

Breastfeeding peer-support interventions exhibit considerable design heterogeneity. 30 In some cases, BFPS is integrated within an existing health-care system, with peer supporters working alongside/reporting to health-care professionals. In other cases, the support is developed by a third-sector agency and delivered in community settings. Help is sometimes provided on a one-to-one basis (mother to peer) and is sometimes delivered in group settings. Modes of delivery include face to face, over the telephone, by text and online (e-mail, social media, forums). Some interventions are proactive (the peer contacts the mother directly), whereas others are reactive (the mother asks for help). Proactive interventions vary in intensity (the number of contacts) and in the timing of delivery (antenatal or postnatal; starting within 24 hours of the birth or several days afterwards).

Experimental evidence for the effectiveness of BFPS provides an unclear picture. A Cochrane review of ‘additional support’ from lay and professional supporters found that additional help can have an impact on breastfeeding rates. 31 The reviewers concluded that help is more likely to be effective in areas with high initiation rates, that face-to-face interventions are more likely to succeed and that reactive support is unlikely to be effective. The authors also recommended that support should be delivered in a predictable manner and be tailored to the needs of the population group targeted. A systematic review of 11 randomised controlled trials (RCTs) of antenatal BFPS found that targeted antenatal BFPS may lead to increases in breastfeeding initiation rates, whereas universal BFPS interventions tended not to be effective. 32

Another systematic review of RCTs looked at the impact of BFPS interventions on exclusive breastfeeding and breastfeeding continuation rates. 33 This review found these interventions to be effective in increasing breastfeeding maintenance in low- or middle-income countries, reducing the risk of not exclusively breastfeeding by up to 28%. However, the UK-based RCTs of BFPS interventions included in this review did not find increases in breastfeeding continuation rates,34–37 with no increase in breastfeeding maintenance at 8–10 weeks35,37 and 4 months. 36

Two quasi-experimental UK-based studies of BFPS have been published. The first, a community-based controlled trial of peer support delivered in a low-income area with very low breastfeeding rates (around 10%) resulted in no overall change in continuation rates at 6 weeks. 38 More recently, a time series analysis of a UK-based BFPS intervention delivered to a geographically defined population of adolescents found that, after adjusting for underlying trends, by the end of the study period, an additional 6.6 women in every 100 were initiating breastfeeding. 39

The UK-based interventions that have been studied were delivered as low-intensity interventions34,37,38 and several did not include an intended first contact with the mother in the days after the birth;34–36 in one UK-based study mothers were not contacted until their babies were around 3 months old. 36 In a UK context, this lack of early contact means that the intervention misses an important window, as breastfeeding discontinuation in the UK is highest during the early days and weeks after birth. 1 Four UK-based RCTs reported difficulty in achieving the intended number of contacts. 34,37–39

Motivational interviewing

Motivational interviewing is a counselling approach that is designed to build a client’s confidence and motivation for change, with a focus on eliciting the client’s reasons for behaviour change. MI is a well-defined approach that could be used within the context of BFPS to help provide a clear model for BFPS provision. Miller and Rollnick28 developed MI to be less confrontational, authoritarian and directing than other counselling styles that were available in the 1980s. MI addresses the therapist’s ‘righting reflex’ wherein therapists have the desire to ‘fix’ what seems to be wrong and tell the client how to change. MI has been called a person-centred collaborative conversation,40 as it is designed to ‘strengthen personal motivation for and commitment to a specific goal by eliciting and exploring the person’s own reasons for change within an atmosphere of acceptance and compassion’ (p. 29). 28 It was initially developed to help people with addictions, but has since been used in many areas such as medication adherence, weight loss and dental health care. 41,42

Motivational interviewing has been demonstrated to be effective across many areas of health. There have been at least 12 systematic reviews that have found statistically significant effects of MI in relation to health outcomes41–52 and three RCTs53–55 examining MI and breastfeeding support. A health educator,53 a research nurse54 and a practice nurse55 delivered the MI in the RCTs. Two of these RCTs53,54 found no significant effect of MI. The other RCT55 found statistically significant effects of MI on breastfeeding status at 2 and 4 months but not at 6 months post birth. In all three RCTs, it was difficult to disentangle the effects of MI from the other aspects of the intervention. Two of the interventions were MI-based interventions, in which only elements of MI were delivered. 53,55 None of the RCTs was based in the UK. It is possible that MI has a role in helping women to continue breastfeeding by increasing their intrinsic motivation to breastfeed, but there is limited evidence available on the effectiveness of MI in this context. The feasibility and acceptability of using a MI-based approach within a BFPS intervention has not been investigated.

The behaviour change wheel

Developing a complex intervention that utilises both peer support and MI-based techniques to support breastfeeding continuation requires integration of the MI and peer-support approaches, identification of potential mechanisms and consideration of how the intervention could be implemented. The behaviour change wheel (BCW) provides a unified and systematic framework for developing and characterising complex behaviour change interventions. 56 The BCW outlines a process for working out what needs to change (sources of behaviour), how to change it (intervention functions and behaviour change techniques) and how the intervention can be implemented (mode of delivery). 56 Within the BCW framework, the capability opportunity motivation – behaviour (COM-B) model helps to explain how interactions between people’s physical and psychological capability (C), social and physical opportunity (O) and automatic and reflective motivation (M) can influence behaviour (B). 57 The BCW uses the Behaviour Change Technique Taxonomy v1 (BCTTv1)58 to identify and classify the content of behaviour change interventions. The affordability, practicability, effectiveness and cost-effectiveness, acceptability, side effects and safety and equity (APEASE) criteria are used within the BCW process as a control point to consider the feasibility of an intervention. 56 The BCW process would therefore provide a unified framework for developing and characterising a MI-based BFPS intervention that could both guide its design and categorise the behaviour change techniques included in the intervention. This would provide a much greater understanding of the potential mechanisms of the intervention, which would facilitate future evaluation, refinement and replication of the intervention.

Mam-Kind: a UK feasibility study

The National Institute for Health Research Health Technology Assessment programme called for primary research in the form of a feasibility study to answer the following research question: ‘Can peer support for breastfeeding in the UK contribute to the maintenance of breastfeeding, particularly in groups of the population who are less likely to breastfeed?’.

This study aimed to develop and test the feasibility and acceptability of delivering a BFPS programme using a MI approach. The target population was women living in areas with lower than average breastfeeding initiation rates and high levels of social deprivation, who have expressed an interest in breastfeeding. We identified the key feasibility questions to be addressed prior to a full trial (outlined below) and designed our study to address these.

Although BFPS is recommended as part of the strategy for increasing breastfeeding rates in the UK, current National Institute for Health and Care Excellence (NICE) guidance does not specify how this should be provided, resulting in a variety of models being used across the UK. We therefore planned a rapid evidence review to understand how BFPS is currently provided in the UK and identify any underpinning theoretical models, elements of best practice and facilitators of, and barriers to, implementation of the service within a NHS or community setting. These data would also allow us to contextualise usual care across the UK and inform the care pathway in the control group of a full trial.

We planned to carry out qualitative research in the form of focus groups and interviews with pregnant women, mothers, fathers, peer supporters and health-care professionals to inform the intervention content and design and develop a training module that was specific to the intervention being developed.

Four UK RCTs,34–37 both individual and cluster randomised trials, had already demonstrated that it was feasible and acceptable to randomise pregnant women in the antenatal period to receive either BFPS or usual care. Two RCTs that randomised individuals recruited 70%35 and 82%34 of their intended sample size and cited difficulties in recruitment resulting from resource constraints. We therefore planned to use the qualitative work in the developmental phase of our study to explore the challenges to recruitment, investigate specific barriers to participation and develop strategies for recruitment and consent. Our approach was based on the work by the Bristol Medical Research Council (MRC) ConDuCT (Collaboration and innovation in Difficult and Complex randomised controlled Trials In Invasive procedures) methods hub,59 which demonstrated the utility of qualitative work for developing and testing optimal strategies for recruitment during the feasibility stage of trials to improve recruitment.

The previous UK RCTs of BFPS34–37 also highlighted the uptake of, and adherence to, the intervention as possible reasons for non-effectiveness. We therefore included a process evaluation in the design of this feasibility study, to include an assessment of the extent to which the MI-based BFPS intervention that we developed can be delivered as intended to the target population. We planned to test the feasibility of the MI-based BFPS intervention in three community maternity sites that had high levels of social deprivation [most deprived quintile of the Welsh Index of Multiple Deprivation (WIMD)60 and below UK average rate of breastfeeding initiation (< 70%)]. We were specifically interested in establishing the feasibility and acceptability of:

• recruiting and retaining peer supporters from the community of women to whom MI-based BFPS is to be delivered

• delivering MI-based BFPS as specified, including an assessment of the extent to which peer supporters utilised MI techniques in their interactions with the mothers they were supporting

• the recruitment, data collection and follow-up strategies (for clinical outcome measures and resource usage and costs) and study materials.

A summary of the study aims and objectives are provided in Box 1. We intended to use the results from objectives 1–5 to inform the optimal strategy for recruitment, consent timing and approach and data collection in a full trial. Figure 1 provides a schematic representation of the different components of this study.

FIGURE 1.

Schematic representation of the Mam-Kind study.

Introduction

Peer support was recommended by NICE61 as part of the strategy to increase breastfeeding in the UK. Current NICE guidance on the commissioning of BFPS in England does not specify the theoretical basis, critical components or optimal delivery mode of BFPS. 62 This has resulted in a wide variety of models being used in current practice. 63,64 We wanted to explore the range of BFPS interventions that were currently available to identify lessons from current practice that could be used to inform intervention development. We carried out a survey of infant feeding co-ordinators (IFCs) in the four UK nations to map the current provision of BFPS and obtain an understanding of what peer-support services consist of and how these are provided. In addition to informing intervention development, these data would provide an understanding of ‘usual care’ to inform the planning of a future effectiveness trial, if warranted.

Method

Results

The following sections provide an overview of respondents’ characteristics, followed by the survey results in relation to six key themes: (1) recruitment, training and support for breastfeeding peer supporters, (2) peer-supporter roles, (3) descriptions of the management and implementation of breastfeeding groups, (4) the interaction between BFPS and health-care professionals, (5) the accessibility of BFPS and (6) resource and financial considerations.

Respondents

A total of 136 individual responses with usable data were received (response rate 19.5%), representing 58% of NHS trust/health board areas (Table 2). Within the 136 responses, there were multiple responses (total n = 34) from 21 NHS trust/health board areas. Seven of these were instances in which provision in England was split between the NHS trust and another provider, such as the local authority. We retained all individual responses in the analysis because the multiple responses from NHS trust/health board areas provided different information in response to open-text questions.

TABLE 2 - Sample and response rates

Nine participants from England did not indicate which NHS trust their response related to. Includes data from 16 non-NHS responses that were mapped to the relevant NHS trust.

Response	Location, n (%)				Total
	England	Scotland	Wales	Northern Ireland
Individual responses
Individual invitations	617	40	19	20	696
Individual responses	113	11	8	4	136
NHS trust/health board coverage
NHS trusts/health boards	151	14	7	5	177
Total response within NHS trust/health board areas	84 (56)a	9 (64)	7 (100)	2 (40)	102 (58)

Breastfeeding peer support was available in 80 (78%) NHS trust/health board areas and breastfeeding support groups were available in 92 (90%) NHS trust/health board areas of the UK for which we had survey data. These data should be interpreted with caution, as we attributed individual responses to NHS trusts and, for larger trusts, these responses may not be representative of the whole catchment area. Neither BFPS nor breastfeeding support groups were available in 10% (n = 10) of the areas for which we had survey data. About half of the survey respondents reported that there were breastfeeding groups in their NHS/health board area that were funded by local authority or non-NHS organisations (n = 62, 54%) and that there were breastfeeding counsellors working in the area who regularly received referrals from health-care professionals (n = 31, 46%).

In total, 62% of the survey respondents (n = 66) reported that there had been a review, evaluation or report of the breastfeeding support service in their NHS trust/health board area. We requested copies of evaluations and received six reports from two respondents and written descriptions of evaluations from two further respondents. These reports were aimed at evaluating service development, such as breast pump loans and attempting to standardise delivery of breastfeeding support within an area, as opposed to evaluating the effectiveness of BFPS.

We searched the internet for information on the availability of BFPS services in the 81 (46%) NHS trusts that did not provide data on breastfeeding support (either because there was no response to the survey or no data were provided in response to the question). In total, 58% (n = 47) of these non-responder trusts provided breastfeeding support groups, 15% (n = 12) provided a one-to-one peer-supporter service, 24% (n = 20) provided both groups and a one-to-one peer-supporter service and 3% (n = 2) provided neither.

Combining the results from the survey and online search, 40% (n = 71) of NHS trusts provided breastfeeding support groups, 7% (n = 13) provided a peer-supporter service, 49% (n = 86) provided both groups and a peer-supporter service and 3% (n = 5) provided neither. Data were unobtainable for two NHS trusts (1%).

Figure 2 illustrates the areas in which breastfeeding support (groups, peer support, both groups and peer support or neither groups nor peer support) was provided throughout the UK, with Figure 3 providing a detailed map of London. It should be noted that these data indicate which services were available in any given region, but do not provide information on service coverage within the region, accessibility or uptake.

FIGURE 2.

The provision of BFPS and breastfeeding support groups in the UK.

FIGURE 3.

The provision of BFPS and breastfeeding support groups in the London region.

The visual representation uses current health board boundaries in Scotland, Wales and Northern Ireland. We were unable to produce a map for England using current health board boundaries and, therefore, we used primary care trust boundaries as a proxy, as these were most closely matched. However, some NHS trusts in England covered multiple geographical areas and, in these cases, all areas of the map have been coloured. Service coverage may vary within these boundaries, as some participants reported variation within the areas they managed in open-text responses and it was not possible to tease out the exact coverage from the available data.

Peer supporters’ roles

Attending breastfeeding groups was the main activity that peer supporters were involved in, followed by working on the postnatal ward (Figure 4). In general, delivery seemed to be more focused on group support, with one-to-one forms of delivery being less common. Most peer supporters saw mothers in both the antenatal and postnatal periods (50% of survey responses, n = 68), but some saw mothers only in the postnatal period (29%, n = 39). The comprehensiveness of services was described through open-text responses, with some areas viewed as having a complete model of service delivery:

The Peer-support Service is a 7 days service 356 days of the year. Team of 10 members, total 7.5 whole time equivalent from 9–5 man a 24 telephone support line. The Service is integrated into [child health care], works alongside Health Visitors, School Nurses, and support staff. The service delivers Health Promotion sessions within Primary schools, they provide bedside support within the three feeder hospitals, provide support groups with Children’s Centre Groups. It is an excellent service provided by a dedicated team.

FIGURE 4.

What activities are peer supporters in your area engaged in?

In contrast, other areas indicated that they were not currently resourced to provide a comprehensive service:

I have one breastfeeding support worker who is employed by [the NHS], this isn’t enough for a birth rate of 2500. We are currently writing a business case for 10 × paid peer-support workers.

Breastfeeding support groups

It was most commonly reported that > 10 groups were running within individual areas (Figure 5). Respondents stated that NHS staff, children’s centre staff and trained peer supporters most commonly organised breastfeeding support groups. Alongside this, it was noted in open-text responses that coverage of breastfeeding support groups was not uniform throughout NHS trust areas and that this was not necessarily related to the numbers of births. Respondents also noted within the open-text boxes that group sessions took place in a broad range of settings, including community venues (café, garden centre café) and children’s centres, and alongside health visitor (weighing) clinics. Some groups ran as ‘baby cafes’, ‘first friends’ or generic ‘parenting support groups’ with a focus on breastfeeding, rather than explicitly as breastfeeding support groups. In response to questions about record-keeping in the groups, less than one-third of survey respondents reported that notes were kept on individual mothers who received support (n = 31, 26%), mothers who have problems (n = 34, 29%) and referrals for additional support (n = 32, 27%).

FIGURE 5.

If you have peer-support groups, how many groups are currently running?

Interaction between breastfeeding peer support and health-care professionals

The majority of participants (63%, n = 67) felt that peer support was well integrated with other NHS services (Figure 6). Alongside the closed question, which asked participants to describe how well they thought peer support was integrated with other health services, respondents were asked to explain their answer. The majority of open-text comments were consistent with respondents’ answers on the six-point Likert scale. Participants described a range of factors that were responsible for integration, including:

• guidance on peer-supporters’ roles, with clear responsibilities (n = 15)

• visibility to health-care professionals, such as working on postnatal wards or other shared working practices (n = 14)

• peer support being trusted and valued by local health-care professionals (n = 9), including as a result of an evaluation or feedback from parents (n = 3)

• a health-care professional acting as a ‘champion’ and communicating the value of peer supporters to other health-care professionals (n = 5).

FIGURE 6.

Do you think that BFPS provided in your area is well integrated with the breastfeeding support work that health-care professionals do?

In areas with less integration, the converse occurred: staff did not understand the peer-supporter role and did not feel that peer supporters were suitably qualified and, as such, there was a lack of trust, resulting in a lack of referrals from health-care professionals to peer supporters.

Accessibility of breastfeeding peer support

In response to the question about how accessible peer support was to mothers from poorer social backgrounds, 24% (n = 25) of respondents used the three negative options (not/moderately, not/somewhat, not accessible) (Figure 7). However, some responses did not corroborate with the additional open text provided, suggesting that some respondents may have unintentionally ticked the wrong box. An example of this was a response that indicated an accessible service on the Likert scale, but conversely stated that ‘the most vulnerable and poorer mothers may be “put off” from attending groups as the groups are seen to have women in a slightly higher social class than themselves’.

FIGURE 7.

Is the BFPS provided in your area accessible to breastfeeding mothers from poorer social backgrounds?

However, this could also mean that, although access is good overall, there are certain aspects of the service that could be improved. When asked to explain the rationale for their responses to the above question, participants suggested that services were working well when they:

• were held in areas of high deprivation, and worked with providers who aimed to support women in areas of high deprivation and used informal conversations in areas with low levels of breastfeeding (school gates, social media)

• were well accessed by mothers living in deprived areas (some responders had data to show this and, when there were no data, this was supported by the breastfeeding co-ordinators’ impressions)

• recruited peers from areas of deprivation and provided proactive contact.

Barriers to accessibility included:

• attracting women from deprived areas to breastfeeding support groups

• inadequate numbers of peer supporters

• being reactive as opposed to proactive

• not being able to provide home visits.

Discussion

Peer support for breastfeeding is recommended as part of a strategy to address low breastfeeding rates in the UK. 62 Within this context, our UK-wide survey of 136 UK-based IFCs found wide variation in service provision. As in previous studies,63,66,67 we found that there was wide availability of peer supporters across the country, both within and between NHS trust/health board areas. A key finding from our survey was that there was no standardised provision of BFPS in the UK. Services were regularly adapted in response to unstable financial circumstances, with services being reduced or increased in line with the funding available. None of the models that were implemented at the time of this survey had been robustly evaluated for effectiveness.

This survey has provided information about how peer support for breastfeeding is provided in the UK and identified some of the facilitators of, and barriers to, the delivery of peer-support services, which were further explored in our qualitative research and through discussions with the Stakeholder Advisory Group (see Chapter 3). A clear tension was reported in some areas between BFPS and health-care professionals, and this may be related to a lack of clear guidance regarding roles and responsibilities and a lack of visibility of BFPS to health-care professionals. We explored this further in our discussions with the Stakeholder Advisory Group and in qualitative work with health-care providers to ensure that the local health service context was considered in planning for intervention delivery in the feasibility sites (see Chapter 3). Accessibility to parents from poorer backgrounds was noted as a challenge. Services that considered themselves to be accessible to poorer mothers noted elements of good practice, including being held in areas of deprivation, working in partnership with organisations who support women from deprived areas and the use of informal conversations with peers.

A further issue was that, in many sites, the current level of support provided was not regularly documented or evaluated. Two-thirds of participants reported that registers of attendees were kept in their area, but details of any support provided or signposting to additional services were rarely kept. Documentation and record-keeping are important to enable evaluation in the context of a research study. This highlights the importance of engaging with peer supporters during the development phase of the study to explore the options for data collection to capture reliable information.

The most common theme found in the open-text responses was the challenge of running services with limited financial support, although this was not experienced equally by all services, with a minority of services reporting recent investment. Linked to this financial shortfall, some services reported the challenges of recruiting, training and providing ongoing supervision for peer supporters, with one-third of participants reporting no ongoing training and support in their area. Understanding this variation will enable the selection of sites in which a future RCT could be undertaken and may influence the extent to which sites are able to fully roll out an intervention.

The majority of survey respondents indicated that there had been a review or evaluation of BFPS in their NHS trust/health board area over the last 5 years; however, to our knowledge, these were service evaluations to inform service provision rather than formal assessments of efficacy or effectiveness.

This study is the first attempt to map and describe the provision of peer support for breastfeeding throughout all four nations of the UK. We received responses from around the UK and achieved a response rate that covered 58% of NHS trusts/health board areas. We searched the internet to obtain information about the availability of BFPS in the NHS trust/health board areas for which there were no survey data available to describe the current provision. We checked the consistency of the data provided in survey responses and those obtained from the internet and were satisfied that these two sources of information were compatible. The survey was open for a period of 3 weeks in December 2014, and it may be that we would have received further responses if the survey had been kept open for a longer period. We were also made aware that two participants were unable to access our online survey from their NHS computers. Although we provided support that enabled those participants to take part, it may be that other potential participants did not contact us and were thus unable to access the survey. The response rate to the survey was low. We did not receive a response from 42% of the NHS trust/health board areas, and, although we were able to obtain internet information on the types of BFPS available, we were unable to understand how these services were run and any potential facilitators or barriers in these areas. We mapped the data on BFPS availability to NHS trust and health board areas and, in doing so, we have had to assume that the responses for each health-care area were applicable across the board. This may not be the case for larger NHS trusts, in which the availability of services may vary greatly; however, our survey was not sensitive enough to capture availability at smaller geographical levels. Further information on the availability and accessibility of BFPS could potentially have been obtained through direct contact with NHS trust/local authorities and third-sector organisations, but we did not have the resources to do this within the scope of this study.

Introduction

The theoretical basis of BFPS and its active behaviour change components have not been well described or characterised, so there is currently limited understanding of what components are required for an effective BFPS intervention. Motivation, self-efficacy, affective attitudes, social norms and strong beliefs that breastfeeding is the normal and healthiest way to feed an infant are associated with continuation of breastfeeding. 10,11,18,68 Applying the theory of constraints thinking to breastfeeding problems, Trickey and Newburn7 found that support should be proactive and mother centred, consistent with the data from our survey of IFCs. It is therefore important to address the ‘why’ (motivation) and ‘how’ (confidence, skills, resources) of breastfeeding maintenance in this context.

Framework for intervention development

The BCW57 provided a framework for identifying the behaviours to be addressed, the required functions of the MI-based BFPS intervention and the relevant service requirements. 56 We utilised the COM-B theoretical model, which hypothesises that interactions between people’s physical and psychological capability (C), social and physical opportunity (O) and automatic and reflective motivation (M) can be used to understand what needs to change to achieve the desired behaviour change (B). 57

We used a flexible iterative process to develop this intervention, which is summarised in Figure 8. We used the survey findings reported in Chapter 2, a rapid literature review and qualitative research to identify the behaviours associated with breastfeeding to be addressed by the intervention and the underpinning theoretical models and design factors that influenced the delivery of the BFPS intervention. We aimed to explore parental, health-care professionals’ and peer-supporters’ views of what the features of a feasible, acceptable and effective BFPS intervention would be. We examined their perceived barriers to, and facilitators of, breastfeeding that a BFPS intervention would be able to address, what current peer support is like and what it should be like. We clarified the sources of behaviour to be addressed by the intervention through discussion with the Stakeholder Advisory Group, informed by the findings from the rapid evidence review and qualitative research, and categorised these according to the COM-B model. This enabled us to follow up on issues identified in the survey and literature review in more depth and explore issues directly relevant to the design of a MI-based BFPS intervention and define its function, content and mode of delivery.

FIGURE 8.

Development process.

Integrating the motivational interviewing approach with breastfeeding peer support

We proposed to develop a BFPS intervention that uses a MI approach. Self-determination theory (SDT) has been proposed as an explanatory theory for MI. 69,70 SDT is a theory about a person’s self-motivation to change behaviour. It shares a common principle with MI that ‘people have an innate organisational tendency toward growth, integration of the self and the resolution of psychological inconsistency’. 69 SDT seeks to explain what drives human behaviour and places motivation on a continuum of autonomy ranging from external regulation (no autonomy) to intrinsic regulation (full autonomy). By examining people’s different motivations for achieving goals, for example pleasure compared with obligation,71 SDT identifies three elements that are critical to support the process of changing motivation from external to intrinsic: competence support, autonomy support and relatedness. MI provides support for each of the psychological needs that are identified by SDT. 72

Motivational interviewing involves four key processes: engaging [establishing a ‘mutually trusting and respectful helping relationship to collaborate toward agreed-upon goals’ (p. 3)73], evoking (eliciting the client’s own motivation for a particular change), focusing (clarifying a particular goal or direction for change)73 and planning (developing a specific change plan that the client is willing to implement). 28 These processes are sequential and recursive and are used throughout the MI session. 28 A MI practitioner aims to clarify and resolve a client’s ambivalence about choosing a particular behaviour and does so in conversation with the client in a spirit of acceptance and collaboration. The process aims to help make complex behaviour changes feel clear and more achievable by allowing the client to decide on the changes in their life that they feel that they can make. Skilful MI practitioners use a range of skills and techniques to guide clients through this process. Central to these is a technical skill of attending differentially to client language that is indicative of change (change talk), while consciously not reinforcing client sustain talk, which is language in favour of the status quo. As MI is a conversation style, these techniques are not specific to a type of behaviour change; rather, they are used in response to where the client is in the MI process. In the case of breastfeeding, these techniques can be used to support initiation and continuation (Box 2).

Self-determination theory is one of the theories that was considered during the development of the BCW framework56,74 and, therefore, it is likely that the BCW framework will be applicable in the context of MI-based BFPS. However, it has been suggested that the BCW may not capture some of the therapist skills that form a key element of the MI approach. 75 Therefore, in addition to using the BCW as a guide in developing and characterising the MI-based BFPS intervention, we assessed its applicability in the context of a MI-based approach and investigated whether or not there were any important potential mechanisms for change included in the intervention that were not incorporated in the current BCW framework and BCTTv1 taxonomy.

Methods

Production of the intervention specification

The results of the rapid review and qualitative work were mapped onto the COM-B model and the BCW to identify the relevant sources of behaviour to be addressed and the relevant behaviour change techniques and implementation requirements using the BCTTv1. 58 We then produced the specification of the MI-based BFPS intervention and corresponding logic model for discussion and endorsement by the Stakeholder Advisory Group.

Results

Rapid evidence review

The rapid literature review aimed to (1) describe heterogeneity in intervention theory and design among one-to-one BFPS interventions delivered in developed country settings that have been subject to experimental study and (2) identify opportunities and modifiable weak points associated with the design of BFPS interventions delivered across different contexts that could be addressed in the design and implementation of the MI-based BFPS intervention that we were developing.

The rapid review identified 15 BFPS interventions that were subject to experimental study, which were reported in 16 papers published between the start of January 2000 and the end of January 2016. 29,34,35,37–39,80–89 Nine interventions were delivered in the USA, all associated with the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC). 80–87,89 Five papers related to UK-based interventions34,35,37–39,88 and one intervention was delivered in Canada. 29

Studies were included if they pertained to a model of BFPS that included planned one-to-one contact between a mother and a peer supporter, reported breastfeeding rates (initiation, continuation or exclusivity) as an outcome measure and if they had been delivered in a developed country setting. Studies were excluded if the support was primarily intended to be group based. A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram90 describing the different stages of study identification is presented in Figure 9.

FIGURE 9.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram. BF, breastfeeding. Adapted from Moher et al. 90 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Adapted from Moher et al. 90 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Study quality

Eleven studies were RCTs,29,34,35,37,80,81,84,86–89 three were area-based controlled studies38,82,83 and two were natural experiments. 39,85 Applying Cochrane risk of bias criteria,91 we found that only three studies had a low risk of bias. 34,81,86 Six of the evaluations were at a risk of selection bias,29,38,39,82,83,85 attrition may have affected six evaluations35,37,80,83,87,88 and one evaluation89 was at a risk of detection bias. Implementation issues affected 11 of the 15 evaluations. 34,35,37–39,80,84,86–89 Among the five UK studies, there were difficulties in achieving the intended number of contacts34,38,39,88 and in ensuring intervention fidelity. 35,37,88^. Of the nine studies of US-based cases, four reported significant implementation problems. 80,86,87,89

Heterogeneity between studies

Within the relatively narrow inclusion criteria, the rapid review found considerable heterogeneity in the type and scale of the breastfeeding rate ‘problems’ that were addressed, the background social norms and the wider health-care delivery context, as well as the underpinning intervention theory and specific intervention components.

With regard to the breastfeeding rate problem being addressed in the studies, nine interventions aimed to increase initiation rates33,35,38,39,80,83,85,87,88 and 12 sought to improve continuation rates,29,34,35,37–39,80,81,84,85,87,89 measured at varying time points. Two interventions were intended to improve the rates of exclusive breastfeeding. 86,89 The scale of the problem of low rates of breastfeeding varied. One intervention was implemented against a backdrop of a breastfeeding continuation rate of around 10% at 6 weeks,38 whereas two interventions were delivered in the context of background initiation rates of 90% in a low-income population of Latina – predominantly Puerto Rican – women. 86,87 The majority of included studies focused on populations that were socially deprived. 38,39,80–87,89 Many interventions were interlinked with a wider agenda to reduce health inequalities. Only three studies were not specifically located/targeted to address the needs of mothers experiencing social disadvantage. 29,34,35

Overwhelmingly, study papers did not include a description of the theoretical models underpinning the interventions that they described. The theory of social support27 seemed to be implied across many studies; however, different configurations of intervention components suggested differences of emphasis on informational, emotional, instrumental or appraisal support among the included interventions. Two studies explicitly referred to social influence or role modelling as a concept underpinning intervention design. 29,89

The emphasis on a similarity between the mother and the peer (adherence to the principle of homophily) varied. Most used peers selected on a locality basis, suggesting an intention to recruit from a similar social background. Six interventions attempted to individually match mothers to peers by ethnicity or language. 80,82,84,86,87,89 Two interventions targeted to adolescents used adolescent peers. 39,84 One intervention, targeted to recent Spanish-speaking immigrants, did match peers by first language. 87

Interventions differed considerably in accordance with indicators of peer professionalisation,25 and were not equally well embedded in existing health services. At one extreme, peers were trained to university diploma level34 whereas, at the other, peers received only 2 hours of orientation. 29 In some cases, peers were employed or were managed by health-care professionals. 37,39,80,83,85–88 Several settings had previous experience of BFPS. 81,85–87 In most settings, the intervention was funded for the lifetime of the evaluation only.

Comparison across interventions indicated considerable heterogeneity in the timing, frequency and intensity of contacts and the setting for contacts. Two interventions did not include an antenatal contact29,84 whereas one was entirely reactive in the postnatal period. 34 Six interventions intended more than five contacts with each mother. 35,80,81,84–86 Three interventions involved contact in hospital prior to discharge. 80,81,86

Design opportunities and modifiable weak points

Within these included studies, peer supporters were described as role models, affirming and normalising experiences and empowering mothers to identify solutions that work for them. The BFPS interventions across these studies aimed to address issues related to the mothers’ own capacity and resources (lack of knowledge, unhelpful beliefs, attitudes, low breastfeeding self-efficacy) and issues related to health-care professional support or capacity. All studies applied the principle of homophily (the tendency for people to identify with others who are similar to themselves)92 such that peer supporters were mothers who had breastfed, but there was variation between studies in the extent to which peers had similar social or cultural backgrounds to the mothers who they supported. The content of the interventions included in these studies is summarised in Appendix 6 (see Table 22).

Thematic cross-case analysis of extracted data resulted in the identification of five areas for consideration in intervention development. These were:

1. achieving cultural acceptance

2. successfully integrating with existing health-care services

3. ensuring that peer qualities enhance uptake and acceptance

4. ensuring that the peer is practically and emotionally accessible to the mother

5. ensuring that the peer–mother relationship promotes change in line with intervention goals.

We summarised the facilitators of, and barriers to, the delivery of BFPS within the context of each theme and identified delivery implications for the MI-based BFPS intervention. These lessons for design were endorsed by the Stakeholder Advisory Group in January 2015 (Table 4).

TABLE 4 - Barriers to, and facilitators of, delivering one-to-one BFPS interventions identified through the rapid review

Characteristic of BFPS	Facilitators	Barriers	Implications for the Mam-Kind BFPS intervention design
Sociocultural norms	Breastfeeding-centric interventions can improve initiation rates in formula-normal cultures if the intervention is intensive around the time of the birth Breastfeeding-centric interventions can improve continuation rates in formula-normal cultures if the intervention provides intensive support and is sustained Under conditions of scarcity, a change in outcomes may be more likely if the BFPS intervention is targeted to a subpopulation inclined to value the offer of help	The intervention will not ‘take’ if mothers and others perceive the gulf between the intervention goal and pre-existing priorities to be very wide. For example, breastfeeding-centric interventions to improve initiation rates in a formula-normal culture will be perceived as irrelevant to needs by a high proportion of the target population Community-based aspects of the intervention may be challenged by contexts in which the target population is transient or has many competing pressures and demands	Intervention needs to be targeted at women most inclined to breastfeed. In the context of this study it was agreed that this would be women who were considering breastfeeding Breastfeeding peer supporters have an important role in acting as advocates for breastfeeding mothers within their own social groups and helping to change the culture around breastfeeding, as well as providing social support directly to mothers
Qualities of the peer	It may be more important to ensure that the peer is perceived as friendly, competent and prepared to be proactive than ensuring that she is well matched to the target population or has specialised breastfeeding knowledge. However, if a target population has specific social, cultural or other needs that relate directly to their feeding experience or ability to access support, then using peers with experiential knowledge of the defining characteristic may be helpful	Volunteer status can cause mothers to feel that they are imposing themselves on the peer, undermining receipt of support. If the target population has complex social needs and multiple competing pressures, then selecting and retaining peers who closely resemble this population will be challenging	The peer supporters need not be socially matched but should have experience of breastfeeding and credibility, an effective listening style, the appropriate social skills to be able to develop personal relationships and be sensitive to the mother’s agenda in relation to breastfeeding
Accessibility of the peer	If the infant feeding context is one in which there is a rapid decline in breastfeeding rates soon after the birth, then early contact is essential to success. In-hospital support for early feeds can help mothers who were unsure to firm up a decision to breastfeed. The early weeks are an important adjustment period during which breastfeeding is becoming established. Incorporating maternal control into the schedule of contacts may help mothers to feel that the support is tailored to their own needs and increase/decrease the period of support-giving	When a mother is already strongly inclined to formula feed, a proactive antenatal visit will be unlikely to be effective in changing her decision. In any context, mothers have strong social and emotional barriers to seeking help from a peer supporter. In general, only mothers who are strongly motivated to overcome breastfeeding challenges and/or who are unusually confident to seek help take up reactive support	Intervention needs to be targeted at women who are considering breastfeeding Intervention needs to be proactive and peer-supporter led in the early postnatal period Clear guidance needs to be in place for health-care professionals to know how to notify the peer supporters of births promptly, to facilitate the provision of support as soon as possible after birth Peer supporters need to be provided with access to postnatal wards Incorporation of a mother-led schedule of contacts to be incorporated into intervention guidelines Guidelines need to be developed to ensure that peer supporters have clear professional boundaries
Peer–mother relationship	Antenatal support is used to provide health education, inform about the intervention and to start a relationship. When mothers are considering breastfeeding but are feeling unsure, or when they feel unusual in making a decision to breastfeed, an antenatal contact may help affirm the decision. Antenatal support may be effective in conveying a specific health education message (e.g. myth busting or addressing a common cultural practice). Mothers who experience a warm and affirming relationship with the peer supporter often feel supported to overcome challenges and meet their feeding goals The promise of postnatal support at this stage may help the mother to feel that her decision will be supported	Extrinsic motivation arising from the direct presence of the peer may not be translated into intrinsic motivation once the peer is absent	Antenatal contact needs to be included as part of the intervention Underpinning intervention theory should encompass the development of warm and affirming relationships between the peer and the mother Integration of peer supporters into the health-care team requires several lengthy stages and checks including occupational health and criminal records check
Successfully integrating with existing health-care services	Well-specified role boundaries and referral pathways, positive prior experience of working alongside peer supporters and the presence of a health-care professional champion can enhance intervention acceptance and help peers to feel comfortable in their role	Ambivalent attitudes to breastfeeding among health-care professionals and incongruent policies may lead to mothers receiving countervailing messages that undermine the BFPS intervention	Guidelines need to be developed to ensure that peer supporters have clear professional boundaries Preliminary work to ensure that the goals of health-care staff are aligned with the goals of the intervention will be needed Integration of peer supporters into the health-care team requires several lengthy stages and checks, including occupational health and criminal records check

Discussion

We used a systematic, user-informed approach, guided by the BCW, to develop a novel MI-based peer-support intervention to support breastfeeding maintenance. We pooled data from a range of sources (published and unpublished literature, qualitative research findings and discussions with experts in our Stakeholder Advisory Group) to understand the behaviours associated with breastfeeding that can be targeted by this intervention. We found that psychological capability, social opportunity and reflective motivation were the key sources of behaviour that the intervention should address to support women to continue breastfeeding. We used the BCW framework to link these behaviours with the relevant functions of the intervention and service requirements for implementation. To our knowledge, the BCW has not previously been used in this context.

The BCW is an extensive but not exhaustive model. 56 The BCW was developed in the context of professional-led interventions, primarily in the areas of diet and smoking. 57 Components of the BCW have since been used in a variety of behaviour change contexts, including improving the uptake of health checks for children,93 medication management for multimorbidity94 and understanding prescribing errors by junior doctors. 95 Using the BCW as a guide in this study enabled us to characterise the sources of behaviour targeted by the intervention and its functions, service implications and mode of delivery.

Motivational interviewing is currently included in the BCTTv1 under ‘emotional social support’. 58 However, this does not reflect the complexity of MI, which has both technical and interpersonal/relational dimensions. 96 There were several techniques used within MI that could be mapped on to behaviour change techniques included in other taxonomies. 75 One of the challenges that we encountered was that the terminology used in the BCTTv1 was not always consistent with the MI approach. The types of intervention that the BCW and BCTTv1 have been built on are in the main part directive and are often focused on doing things to a person to elicit behaviour change. The underlying ethos of MI is quite different, in that it values autonomy and draws on a person’s own strengths and assets. One of the guiding principles of MI is to resist the ‘righting reflex’, wherein health-care workers have a powerful desire to ‘set things right’, which can have paradoxical effects of increasing resistance to a given behaviour. 28

In working through the process of characterising the intervention, it became apparent that the BCTTv1 was not well suited to describing how the intervention content would be delivered in terms of style of communication, the quality of the interpersonal relationship between peer supporters and mothers and the underlying ethos of the intervention approach. Much of MI focuses on relational and interpersonal techniques, known as ‘the MI spirit’, which can strongly influence its effectiveness. 28 Building a collaborative relationship is a key feature of MI. 28 MI interventions that had a strong emphasis on improving fidelity to intervention content (i.e. using manuals) were less effective,97 whereas those focusing on adherence to relational–interpersonal style techniques had larger effects. 98 We addressed this by developing a set of ‘guiding principles’ to clarify the underlying ethos and interpersonal style that guides the way in which the technical behaviour change content of the intervention is delivered. These ‘non-specific’ factors apply to many types of intervention and can be challenging to measure and assess, but they are important and need to be given sufficient attention in future iterations of the BCTTv1 and other behaviour change technique taxonomies.

Introduction

We aimed to assess (1) whether or not the intervention could be delivered as intended, with proactive Mam-Kind buddy-led support during the early postnatal period; and (2) whether or not the intervention and research processes were acceptable to pregnant women who lived in areas with high levels of social deprivation, Mam-Kind buddies and health-care professionals. We used quantitative data to assess adherence to the Mam-Kind intervention by examining the number, timing and duration of one-to-one contacts with Mam-Kind buddies. We used qualitative data from interviews with participants, Mam-Kind buddies and health-care professionals involved with the delivery of the Mam-Kind intervention in the study sites to assess the feasibility and acceptability of the method of recruiting pregnant women, and the study materials to understand what worked well and any areas for improvement. We also explored fidelity to the intervention by examining the extent to which Mam-Kind buddies met the objectives of the intervention as intended and the range of MI techniques used in their interactions with the mothers who they supported.

We considered the possible inclusion of a control group and randomisation process, but concluded that this would be premature at this stage. Previous studies had already demonstrated that it was feasible to conduct RCTs of BFPS, but highlighted problems with poor uptake and adherence to the intervention. We therefore considered that it was most important to establish the feasibility of providing a MI-based BFPS intervention as intended to the target population and designed the study to address this.

Ethics considerations

Ethics approval for the feasibility study was granted by the NHS Health Research Authority, Wales REC 3 Panel, in June 2015 (reference 15/WA/0149). All participants provided written informed consent. Health-care professionals provided audio-recorded verbal consent for their interviews following a standardised script.

Study design

This was a non-randomised multisite before-and-after study with a process evaluation.

Setting

The study was carried out within community maternity services in three areas with high levels of social deprivation and low breastfeeding initiation rates. Each site was in the 20% of most deprived communities based on the English Indices of Deprivation99/WIMD60 and had lower rates of breastfeeding than the UK average (< 70%) and a higher-than-average proportion of teenage pregnancies (> 41.9 conceptions per 1000 women aged < 18 years). One of the sites (site 3) had an existing paid peer-support service. This allowed us to test the feasibility of implementing the intervention within an existing service, which required a shift in the way of working to deliver Mam-Kind as specified in the context of a research study. The other two sites had a voluntary peer-support service that provided group support in community settings. In sites 1 and 2, we estimated that there would be approximately 528 women with an expected date of delivery during the recruitment period. Based on experience in other studies, we assumed that approximately half of these women (n = 264) would be considering breastfeeding and that 55% of these women (n = 145, based on breastfeeding initiation rates) would be interested in the study, with at least half (n = 73) consenting into the study. In site 3, we estimated that there would be approximately 264 women with an expected date of delivery during the recruitment period. By applying the same assumptions that were used in the other two sites, we estimated that approximately 34 women would consent to take part in the study.

Inclusion and exclusion criteria

All English-speaking pregnant women (at least 28 weeks’ gestation) who were considering breastfeeding were eligible for inclusion in this study. We chose to recruit and consent women who were considering breastfeeding in the antenatal period because the primary outcome that we wanted to affect was maintenance of breastfeeding. This was consistent with the finding from the rapid literature review, which suggested that a change in outcome was more likely if the intervention was targeted to a subpopulation that was inclined to value the offer of help. We anticipated that recruitment from 28 weeks onwards would provide sufficient time for women to have an antenatal contact with their Mam-Kind buddy in accordance with the intervention specification. Women with multiple pregnancies (twins, triplets, etc.) were eligible for inclusion in the study as long as they met the other inclusion criteria. This would have allowed us to explore the feasibility of providing MI-based BFPS to this group of women. We excluded women who were unable to provide written informed consent, who were unable to use conversational English, who did not plan to breastfeed, who had a clinical reason that precluded breastfeeding (e.g. baby with a major congenital anomaly) or who had a planned admission to a neonatal unit following birth.

Recruitment of Mam-Kind buddies

We recruited seven Mam-Kind buddies to work in the two sites that did not have a pre-existing service. In the site with an existing paid peer-support service, two peer supporters took on the Mam-Kind buddy role. An additional Mam-Kind buddy was recruited and trained to provide cover in sites 1 and 2 if needed; however, we retained all Mam-Kind buddies throughout the study and no cover was required. The post of breastfeeding peer supporter was advertised through the Cardiff University job vacancies web page. Local breastfeeding groups and peer supporters were approached via social media to reach potential applicants. All candidates (n = 15) who applied for the post were invited to attend the first 3 days of training, which included information about the study and an introduction to MI. When possible, video conferencing was used to deliver training for those who were unable to attend. The 3 days of training were unpaid with the exception of travel expenses and childcare costs. Following this training event, candidates were shortlisted and asked to attend for an interview to complete the selection process. Following recruitment, Mam-Kind buddies were given further study-specific training over 4 consecutive days to integrate breastfeeding and MI skills. One further training day was provided in final preparation for the role. Training was provided by Cardiff University and, although childcare costs and travel expenses were covered for the Mam-Kind buddies, attending further training was not feasible for the Mam-Kind buddies in site 3 because of the geographical distance and family commitments. All nine Mam-Kind buddies were provided with an intervention handbook containing information about the intervention and training materials and a resource pack with information about MI (see Report Supplementary Material 1). All eight Mam-Kind buddies who delivered the intervention were asked to take part in a process evaluation interview to understand their experiences of the study.

Recruitment of supervising midwives

We recruited one practising midwife in each site to provide supervision to the Mam-Kind buddies delivering the intervention at their site. We requested expressions of interest for the role, in collaboration with heads of midwifery. All three midwives attended the 1-day study-specific training, which included an introduction to MI.

Site set-up

Outcome measures

We used a range of measures to assess the feasibility of delivering the Mam-Kind intervention, as shown in Box 4.

Data collection

Baseline data items (see Report Supplementary Material 1) were collected at the time of obtaining consent, including sociodemographic variables, infant feeding intentions and maternal health and well-being. The study manager conducted a brief follow-up telephone interview at 10 days (see Report Supplementary Material 1) that asked about skin-to-skin contact post delivery, breastfeeding initiation, breastfeeding self-efficacy, support received, sources of influence and feeding method. A further, more detailed follow-up interview was conducted at 8–10 weeks after birth (see Report Supplementary Materials 1and 2) to ascertain the duration of exclusive or partial breastfeeding, breastfeeding attitudes, use of health-care professionals or groups for breastfeeding support and maternal and child health and well-being. A 10-day minimum data set questionnaire was completed at 8–10 weeks for participants who could not be contacted at 10 days. Well-being was measured using the Edinburgh Postnatal Depression Scale. 100 Breastfeeding status has been linked to mental health status and so we wanted to capture this as a potential confounder. Although we were not testing confounders in the feasibility study, we were testing the acceptability of collecting these types of data. Mam-Kind buddies completed a diary to provide data about their contacts with the women they were supporting. Mam-Kind buddies were also asked to audio-record all of their face-to-face sessions with participants who had consented to this. Midwives completed screening logs to record the number of women who did not want to participate in the study. Supervising midwives completed contact logs of their contacts with the Mam-Kind buddies. A summary of data items collected is provided in Table 7.

The process evaluation interviews were conducted by experienced qualitative researchers, facilitated by a topic guide (Box 5). The interviews were conducted by telephone, audio-recorded and transcribed verbatim by a professional transcription company. The duration of the interviews ranged from 16 to 41 minutes for health-care professionals, from 15 to 70 minutes for mothers who received the intervention, from 36 to 75 minutes for Mam-Kind buddies and from 29 to 35 minutes for supervising midwives.

For the telephone interviews with mothers across the three sites, we used a purposive sampling process based on four factors: study site, Mam-Kind buddy delivering the intervention, success at breastfeeding at 10 days and level of engagement with the intervention. We approached a total of 43 mothers and 29 (67%) agreed to take part in the interviews. We originally aimed for the sample to consist of at least 50% of young mothers (aged ≤ 21 years). However, as the feasibility study had recruited only four mothers who were aged ≤ 21 years, we interviewed all eight mothers who were aged ≤ 25 years. We further interviewed mothers aged 26–30 years (n = 12), 31–35 years (n = 8) and ≥ 36 years (n = 1).

Fidelity to the content of the intervention sessions and MI was assessed using audio-recordings of face-to-face sessions between Mam-Kind buddies and mothers. Sessions were recorded on a Dictaphone (Olympus digital voice recorder VN-731PC, Olympus, Southend-on-Sea, UK) with the participants’ permission and transcribed verbatim for analysis. We initially selected two sessions per Mam-Kind buddy to be coded: the first antenatal session that the Mam-Kind buddy had audio-recorded and a postnatal session that occurred 2 months after the first session. An additional five audio-recordings were also included in the sample to ensure full representation of all key time points in the intervention (antenatally, 48 hours, 2–13 days and 2–6 weeks), resulting in an overall sample size of 21.

We used a validated measure of MI, the Motivational Interviewing Treatment Integrity (MITI 4.1) tool,104 to measure the extent to which the peer supporters used MI. Twenty-minute segments are usually randomly selected for MITI 4.1 assessment; however, we adapted this method of measurement for pragmatic reasons, as follows. In our sessions, Mam-Kind buddies shifted focus across a number of different topic areas, which meant that there was not necessarily a continuous 20-minute section in which they talked about ‘feeding baby’. Therefore, we first conducted a content analysis of the audio-recordings, which helped us to identify conversations focused on ‘feeding baby’, and then applied the MITI 4.1 tool to a 20-minute collection of these segments per audio-recording.

Quantitative data analysis

Summary measures [frequencies/percentages and means/standard deviations (SDs)] were tabulated for all quantitative data at each time point (baseline, 10 days and 8 weeks), including missing data for each measure.

Process evaluation data analysis

An initial coding framework for the interview data was developed, based on the analysis of three interviews with participants. The themes were further updated and refined in an iterative manner throughout the analysis.

The content of the audio-recordings was assessed using deductive content thematic analysis. The coding framework was based on the content guide of the intervention (see Table 6), to outline the content of the conversation against the specified intervention objectives at each stage of the intervention (antenatally – six objectives; 48 hours – five objectives; postnatally – five objectives). The coding framework was then applied to all of the participant interview and session data by two researchers (LC and LM). The team discussed any new analytical themes that emerged; these were added to the framework and previous transcripts were recoded accordingly until all of the data had been coded. The transcripts and the coding framework were uploaded to NVivo 10, where both researchers coded a subset of the data to test reliability.

We mapped session content against the objectives in the intervention content guide to assess whether or not (1) objectives had been met and (2) the content of the sessions was appropriate to the stage of the intervention.

Motivational interviewing fidelity was assessed using the MITI 4.1 tool. The MITI rating tool consists of a number of count and score variables. ‘Global’ ratings are divided into technical (cultivating change talk and softening sustain talk) and relational (partnership and empathy). These are scored on a scale from 1 to 5. Behaviour count scores are provided for giving information, persuade, persuade with permission, questions, simple reflections and complex reflections, affirm, seek collaboration, emphasising autonomy and confront.

A number of composite scores are calculable from these variables, including the technical rating (which encompasses scores on cultivating change talk and softening sustain talk), relational rating (which encompasses scores on partnership and empathy), percentage complex reflection, reflection-to-question ratio, total MI adherent and total MI non-adherent. There is no overall score to indicate ‘basic competence’ or ‘proficiency’ incorporating all of the measures. MITI guidance indicates that, to reach basic competence in each of the measures, a practitioner must achieve an average global technical score of 3, an average global relational score of 4 and a reflection-to-question ratio of 1 : 1 and that at least 40% of all reflections should be complex reflections. To reach proficiency, a practitioner must achieve 4, 5, 2 : 1 and 50% for these items, respectively. The Mam-Kind buddies were not expected to reach these thresholds, but we used this assessment to obtain an understanding of the extent to which peer supporters were able to use MI techniques in their interactions with mothers.

Feasibility study results

Feasibility and acceptability of the method of recruiting pregnant women

Feasibility of recruitment

Feasibility of participant recruitment was demonstrated by the numbers recruited in each site, as shown in Figure 10 (for recruitment by site see Appendices 10–12). Of the 292 women who were assessed and met the eligibility criteria for the study, 39% (n = 115) expressed an interest in taking part. Of these 115 women, 61% (n = 70) completed the consent process and were recruited into the study. The majority of women (94%, n = 108) who expressed an interest in taking part in the study were contacted by the study team to carry out the recruitment process; however, of those contacted, 35% (n = 38) declined to participate. The percentage who declined to participate varied between sites (site 1: 34%, 11/32; site 2: 46%, 27/59; site 3: 0%, 0/17). The reasons given for declining were that it was a very busy time for them and they did not have time to take part in a study or they felt that it was not right for them. In site 3, the high rate of non-participation in the study may be explained by the availability of an existing one-to-one peer-support service that women could access without having to consent to a research study. There were four withdrawals across the three sites because of participants feeling that they had sufficient breastfeeding support systems in place.

FIGURE 10.

Recruitment flow diagram (all sites). EDD, estimated due date.

Demographic characteristics of participants

The mean age of participants was 29 years (Table 9). Three participants (4%) had no qualifications, nine participants (13%) had General Certificate of Secondary Education (GCSE) qualifications as their highest qualifications and 23 participants (33%) had a degree or higher qualification (Table 10). The majority of participants were employed in full-time (n = 46, 66%) or part-time (n = 7, 10%) work, with 14 participants (20%) unemployed. The majority of participants (43%) were in either managerial or professional occupations or intermediate occupations (23%).

TABLE 9 - Age distribution by site

Max., maximum; min., minimum.

Site	Age (years)
	Mean (SD)	Min., max.
1 (n = 21)	28.3 (5.09)	19.2, 41.1
2 (n = 32)	29.6 (4.14)	20.5, 36.6
3 (n = 17)	27.9 (4.13)	20.9, 35.6
Total (n = 70)	28.8 (4.44)	19.2, 41.1

TABLE 10 - Participant demographics

A level, Advanced level; AS level, Advanced Subsidiary level; BA, Bachelor of Arts; BSc, Bachelor of Science; NS-SEC, National Statistics Socio-economic Classification; S level, Subsidiary level.

Characteristic	Participants (N = 70), n (%)
Marital status
Married/civil partnership	27 (38.6)
Living together	27 (38.6)
Have partner but not living together	6 (8.6)
Single	9 (12.9)
Widowed/divorced/separated	0 (0.0)
Missing	1 (1.4)
Highest educational achievement
GCSE grades A–C	9 (12.9)
A or AS or S levels	7 (10.0)
Certificate or diploma in higher education	14 (20.0)
First degree (e.g. BA or BSc)	20 (28.6)
Higher degree	3 (4.3)
Other	13 (18.6)
None of these qualifications	3 (4.3)
Missing	1 (1.4)
Employment status
Full-time work	46 (65.7)
Part-time work	7 (10.0)
Unemployed/seeking work	3 (4.3)
Unemployed/not seeking work	11 (15.7)
Full-time education/training	2 (2.9)
Part-time education/training	0 (0.0)
Rather not say	0 (0.0)
Missing	1 (1.4)
NS-SEC – socioeconomic code
Managerial and professional occupations	30 (42.9)
Intermediate occupations	16 (22.9)
Small employers and own-account workers	3 (4.3)
Lower supervisory and technical occupations	8 (11.4)
Semiroutine and routine occupations	8 (11.4)
Not applicable (never worked)	2 (2.9)
Missing	3 (4.3)
Ethnicity
White British	66 (94.3)
Any other white background (Romanian)	1 (1.4)
White and black Caribbean	1 (1.4)
White and Asian	1 (1.4)
Missing	1 (1.4)
First baby?
Yes	43 (61.4)
No	26 (37.1)
Missing	1 (1.4)

The majority of participants were married (n = 27, 39%) or living with a partner (n = 27, 39%). In total, 9% (n = 6) of participants reported that they had a partner that they were not living with and 13% (n = 9) were single. The majority of participants (n = 66, 94%) were white British. All participants were expecting one baby and for 61% (n = 43) it was their first baby. Among the 26 participants who had other children, about one-third had previously breastfed. Nearly all participants (n = 65, 93%) planned to exclusively breastfeed their child or to combine breast and formula milk (n = 2, 3%). The majority of women (n = 26, 37%) reported that the mothers they knew had mostly formula fed, with only 6% (n = 4) reporting that they knew mothers who had breastfed (Table 11). Half of participants reported that, as babies, they had been entirely formula fed (n = 35).

TABLE 11 - Breastfeeding experiences

Breastfeeding experiences	Participants (N = 70), n (%)
How participants were fed as a baby
Breastfed entirely	19 (27.1)
Formula entirely	35 (50.0)
Mixed feeding	8 (11.4)
Do not know	0 (0.0)
Missing	8 (11.4)
Know any mothers with young babies?
Yes	55 (78.6)
No	14 (20)
Missing	1 (1.4)
How other mothers you know have fed
Mostly formula only	26 (37.1)
Mostly breastfed	4 (5.7)
Mostly mixed	8 (11.4)
Mixture of above	16 (22.9)
Do not know	1 (1.4)
Not applicable	14 (20.0)
Missing	1 (1.4)
Plan to feed baby
Formula (not eligible)	0 (0.0)
Breastfeed	65 (92.9)
Mixed	2 (2.9)
Undecided	1 (1.4)
Missing	2 (2.9)

Acceptability of recruitment methods to participants and health-care professionals (data from qualitative interviews)

The study training provided to midwives and maternity care assistants about introducing the study to potential participants was received positively, although fewer than expected attended because it took place during the summer holiday period. Midwives expressed that the training was clear and that they had a good understanding of their role within the study. The following insights were obtained through the qualitative interviews that were carried out during the process evaluation.

Community midwives felt that it was appropriate and acceptable to introduce the study to potential participants at 28 weeks’ gestation, as it coincided with the stage in their pregnancy when participants were starting to consider how they would feed their baby. However, midwives reported that they occasionally found that they had little time to explain the study to women. Participants felt reassured that they would get extra support with breastfeeding, and this was expressed as the primary reason for participating in the study:

And she said oh she said there’s people that are going to, you know are doing a study and they’re going to be able to support you through breastfeeding and I read all about it and I was like oh yes, you know it made me feel more at ease knowing that I’m going to get support.

Participant 103

Participants and midwives expressed that meeting with Mam-Kind buddies would be beneficial for study promotion and recruitment in a future trial. Midwives were requested to complete a screening log for all women who were approached about the study, but acknowledged that sometimes they forgot because of other time pressures:

And we have so many things to, to discuss and talk about and obviously we’ve only got 20-minute appointment, we’d obviously ask all the information but then sometimes it may just be that, because we’re so busy, we’ve forgotten to put it on the screen, the log at that point.

Midwife 2001

The process for referring potential participants who expressed an interest in the study for a consent visit needed refinement, and additional support was provided from the research and development office at the largest site (site 2). Written consent was obtained by a member of the research team, usually in the participant’s home, and this was acceptable to participants. The research team monitored recruitment and, when there was a noticeable plateau, a site visit was initiated or the local study team was contacted to remind midwives about the study and find out if there were any issues that needed to be resolved.

Overall, midwives expressed satisfaction with the training that they received. They felt that their role was clearly explained and expressed that they had the opportunity to discuss the study further. Midwives liked the study resource pack that they were given, consisting of BFI information leaflets for participants, screening logs, copies of the study invitation letter and procedures for introducing the study to women. There was only a short interval between the health-care professionals’ training and the commencement of study recruitment and some health-care professionals stated that they would have liked a few weeks to consider the study information before introducing the study to women.

Discussion

In this section we summarise the key findings from the feasibility study and discuss the implications for a future trial.

Introduction

Given the importance of health economics in informing decision-making, the incorporation of a health economics work package would be fundamental to the success of a future RCT evaluating clinical effectiveness and cost-effectiveness. Within the context of public health interventions and programmes, attention should be paid to ensure that the aims and methods are congruent with UK decision-making bodies such as NICE, which published a generic manual for guideline development, including in public health, in 2014,111 with specific reference to consideration of incorporating economic evidence within the context of public health interventions. Key questions, such as which perspective should be adopted in the analysis and which measures of health and non-health benefits should be considered, are essential questions to address in the early stages of planning.

This part of the study (study objective 5) evaluated the feasibility of undertaking an economic evaluation of the effectiveness of the Mam-Kind intervention. Specifically, it was designed to assess whether or not accurate and comprehensive data could be collected on health and relevant resource use in this population and consider the key cost drivers associated with the Mam-Kind intervention and which relevant perspectives should be incorporated into a future trial. Early-stage identification of the resources needed to deliver the Mam-Kind intervention and the associated costs was undertaken as part of the feasibility testing. Alongside the data provided in Chapter 4, we considered the most appropriate measures to inform the collection of salient health and non-health benefits.

Objectives

The objectives of the economic evaluation were to:

• assess the feasibility of collecting resource usage and costs associated with the implementation of the Mam-Kind intervention

• provide an early-stage assessment of the costs associated with implementing the Mam-Kind intervention

• identify the relevant perspectives to adopt in a future trial and related resource use categories for health economic analysis and consider outcome measures (and challenges associated), taking into account current thinking and practice in the design of economic evaluations of public health interventions

• provide commentary on a suitable economic evaluation framework for a future RCT.

Methods

The evaluation was divided into the following consecutive stages, following the overall study process. We reviewed the health economic literature derived from the literature search and, when data allowed, aimed to provide a narrative synthesis of the current evidence base on the use of MI-based BFPS. Based on the results of the review (when applicable) and discussion with the research team on the findings from the qualitative component, we identified the initial perspective and key resource categories to adopt in the adaption or, if required, construction of a bespoke or adapted resource use measure, drawing on a search of the DIRUM (Database of Instruments for Resource Use Measurement) website112 for candidate measures.

Intervention costs

A summary of the intervention costs associated with the Mam-Kind intervention is reported in Table 20. The overall cost of delivering the Mam-Kind intervention, based on the current feasibility study, was £33,595.

The Mam-Kind intervention costs were based on delivery as in the feasibility study, and important factors should be considered in determining the reliability and validity of these costs, particularly with regard to how the Mam-Kind intervention would be delivered in the context of a RCT or in usual practice. The training costs were based on senior experts providing the MI-based training, with follow-up from members of the research team. An important consideration in developing this intervention for use in any future trial is to ensure that the skill mix of the training team is carefully considered, for example intervention costs could be reduced by ‘training’ others to deliver this important component of the Mam-Kind intervention.

With respect to the midwifery supervision, there was variation in practice across the three sites, with different modes of supervision provided (e.g. telephone vs. face to face; individual vs. group supervision). Although the cost of the supervision time is largely determined by the time spent, efficiencies could be achieved by optimising the model of supervision. Travel costs were not recorded and, thus, training and the provision of information to ensure accurate data collection for this component of the intervention should be undertaken for any future trial. Acknowledgement is also made that we used local financial records rather than published unit costs120 to calculate salaries and may have underestimated the on-costs associated with these salaries.

The opportunity cost of employing the peer supporters warrants further consideration, as it represents the most costly driver in delivering the Mam-Kind intervention. One important factor to consider is the workload of the peer supporters, with variation seen across the peer supporters, ranging from a workload of six participants to a workload of 11 participants. The base-case analysis showed that, for the 70 participants recruited to the study, the intervention cost per participant was £480. This did not take into account participants who withdrew from the Mam-Kind intervention. Although further in-depth examination was not carried out, the costs were re-examined based on a scenario in which the peer supporters recruited in accordance with the maximum number of participants allowed in the study protocol (12 per peer supporter; total of 96 participants). In this case, the costs would be reduced to £350, reflecting the need for any future trial or roll-out of the Mam-Kind intervention to fully consider the capacity of peer supporters to ensure the most effective and efficient delivery of the intervention. However, these costs associated with the Mam-Kind intervention must be considered early-stage estimations. Further consideration of how the feasibility study reflects the broader clinical sites that would be included in a future trial will also play a part in assessing the validity of these costs.

Health-care utilisation costs

The costs associated with NHS health-care resource usage are available from the authors, and a summary of the costs is shown in Table 21.

Important caveats need to be placed around the estimations provided in Table 20. In total, 45 out of 70 questionnaires were available for analysis and, thus, these costs may not be reflective of all participants within the feasibility study. Even if a full response had been achieved, given the small sample size and the spread across the three sites, it is doubtful that these costs would be representative.

As expected, resource utilisation was skewed, with six hospital admissions (all related to the baby) being the biggest cost driver. This was weighted by one admission resulting from a urinary tract infection, resulting in a hospital stay of 10 days (NHS reference cost £9137). In addition, with the reliance on self-reported data, these costs could not be verified against routine records. Self-report also required participants to make a judgement on whether or not this was the result of a problem with feeding. In a subsequent trial, attention should be paid to the wording of questions, with a focus on whether this should be ‘specific to feeding issues’ (as used in this feasibility study) or should capture all health-care costs within the post-partum and longer period in a future economic analysis.

Nonetheless, this illustration of costs associated with the intervention indicates that a number of important cost drivers should be captured in a future study. Although not formally costed, the costs to the individual and the family (costs of over-the-counter medication, equipment, travel, carer time and paid help) will also be important determinants in establishing the wider costs associated with the Mam-Kind intervention.

Discussion

There is little health economic evidence to date on the use of peer-support interventions in breastfeeding. We identified the resources and associated costs involved in delivering the Mam-Kind intervention and considered the most appropriate measures to inform the collection of selected health and non-health benefits. We found that the questionnaire that was used captured salient information about the wider costs associated with the Mam-Kind intervention. With the focus on the impact of the Mam-Kind intervention on NHS maternity services, a NHS/Personal Social Services perspective should be adopted in a future trial. However, the feasibility study confirmed that capturing the wider costs to the family should also be considered. Feedback from the Mam-Kind research team suggested that it is feasible to use a telephone-administered RUQ. Based on the conduct of the feasibility study and recruitment of 70 participants, an intervention cost of £480 per participant was estimated. The number of participants recruited is an important factor in determining the intervention cost; if recruitment had been at the optimum capacity level for each peer supporter (12 participants), the intervention would have an estimated cost of £350 per participant. Appropriate consideration should be paid to the method of training peer supporters in the use of the Mam-Kind intervention. The cost estimation for the peer supporters included all of the development work that was conducted. If this intervention was to be rolled out, then a different training model could be used. This warrants further consideration, as efficiencies in delivery of the training and economies resulting from a longer duration of intervention delivery will have an impact on the cost of the intervention. An initial examination of health-care costs and costs to the individual and the family indicates that the costs of hospital admissions and outpatient/day-case and primary care costs make up the majority of the costs; prescribed medication for the baby and the purchase of equipment to support feeding are other important costs to consider, including potential cost offsets, such as reduced purchase of formula milk, over-the-counter medications and feeding equipment.

Although not formally explored, the impact of participants lost to follow-up is an important determinant in establishing intervention costs in a future trial; similarly, the importance of ensuring optimal capacity of peer supporters and variance in workload should also be considered in a future trial of clinical effectiveness and cost-effectiveness. Although brief consideration of the use of the EQ-5D-5L within the feasibility study suggests that it is feasible to collect data from mothers, careful consideration is needed to establish the most appropriate choice of outcomes to be included (and measure) to provide as comprehensive a picture as possible of the economic impact of the Mam-Kind intervention in a future trial. We propose that consideration should be given to generating a cost per quality-adjusted life-year (QALY)124 as a secondary objective in both any planned within-trial analysis and any subsequent model-based analysis to estimate the longer-term cost-effectiveness of the Mam-Kind intervention. A future framework for health economics analysis should consider the short- and longer-term outcomes of the Mam-Kind intervention, drawing on similar work that goes beyond trial modelling to examine the cost-effectiveness of breast feeding over the longer term. We suggest that the use of the NICE reference case,125 with specific consideration of the economic evidence to inform public health interventions, provides a useful framework for setting out the key requirements for such an analysis.

Summary of key findings

Our first objective was to carry out a rapid evidence review to describe the range of BFPS interventions in use in the UK. Our survey of UK IFCs showed that the provision of BFPS across the UK was not standardised. Services were regularly adapted in response to changing financial contexts, and none of the models in use were reported to have an underpinning theoretical model or to include results from robust evaluations of clinical effectiveness or cost-effectiveness. Although there was wide availability of peer supporters across the country, the majority were engaged in providing group support, with little provision of one-to-one support in the early postnatal period. The findings from this survey and our rapid literature review highlighted a number of potential facilitators of, and barriers to, implementing BFPS. Specifically, these included the need for clarity about the peer-supporter role and responsibilities and professional boundaries, the need for integration with existing local health-care services and close working with health-care professionals and a strategy for ensuring that peer supporters are informed of births and have access to postnatal wards.

We further clarified the content of the intervention and the requirements for implementation (our second objective) using focus groups with pregnant women, mothers, fathers and peer supporters and qualitative interviews with health-care professionals. The focus groups highlighted that peer supporters had an important role to play in normalising breastfeeding and supporting mothers, particularly in areas where breastfeeding was not the norm. In training for the peer-supporter role, the need to provide consistent advice (and the need for integration into local health services) was a prominent theme for both peer supporters and health-care professionals. A Stakeholder Advisory Group (n = 23) was also convened to advise on all aspects of study design, including intervention development. The qualitative work was finalised and used to inform the development of a more detailed specification of the intervention, which was presented to and endorsed by the Stakeholder Advisory Group.

We used the BCW framework to understand the behaviours associated with breastfeeding that can be targeted by this intervention. We determined that the intervention should address psychological capability, social opportunity and reflective motivation to enable women to continue breastfeeding. We used the BCW framework to link these behaviours with the relevant functions of the intervention and service requirements for implementation. To our knowledge, the BCW has not previously been used in this context. By the end of the process we produced the specification and corresponding logic model for the Mam-Kind intervention, fulfilling our third objective.

Our fourth objective was to assess the feasibility of delivering the Mam-Kind intervention to women living in areas with high levels of social deprivation. We were specifically interested in the recruitment and retention of peer supporters (Mam-Kind buddies), the recruitment of women to the study, the uptake of the intervention and intervention fidelity. We had no difficulties in recruiting Mam-Kind buddies who met our specification and all of the Mam-Kind buddies who successfully completed the initial job check continued in their role for the full duration of the study. In total, 39% of eligible women (n = 115) approached expressed an interest in taking part in the study, of whom 61% (n = 70) completed the consent process and were recruited. Although broadly consistent with our assumptions, the proportion of eligible women expressing an interest in the study was lower than we expected, which is possibly explained by a number of factors. In one site, only a proportion of midwives working in the area received training about the study. In another site with a pre-existing peer-support service, women were reluctant to take part in the Mam-Kind study when they could access peer support without having to consent to the research process. The recruitment rate, although reasonable, could be improved by a number of strategies discussed later in this section. Further to the recruitment and retention of participants, we were also interested in the recruitment and retention of supervising midwives. We managed to recruit and retain all three supervising midwives, who reported that they found the role to be acceptable and that it was feasible to deliver the supervision required by the Mam-Kind buddies. It was noted that the midwife supervision role should be kept distinct from other roles associated with the study, such as being a local champion, and recruitment in a full RCT.

Uptake of the intervention was good and the majority of participants had an antenatal contact with their Mam-Kind buddy, demonstrating that it was feasible to deliver this aspect of the intervention. In total, 73% of participants received a contact within 48 hours and this did not vary according to whether or not there was antenatal contact with the Mam-Kind buddy. The qualitative interviews carried out during the process evaluation provided some useful insights to inform future strategies to improve the birth notification process. Fidelity testing showed that the Mam-Kind buddies delivered the intervention content with high fidelity and, encouragingly, some Mam-Kind buddies developed MI skills to basic competency level. We did not make reaching competency on the MITI 4.1 scale a requirement for delivering the intervention, as this is the level expected by a trained and experienced MI counsellor. Instead, we used the MITI 4.1 tool to assess the extent to which the Mam-Kind buddies were using MI techniques in their contacts with the women who they were supporting. The process evaluation highlighted that there was a tension between information-giving and providing support in a MI-consistent manner. We have used these insights to refine the scope and objectives of the Mam-Kind sessions and refined the training module to increase the emphasis on providing support.

Strengths and limitations

The main strength of this study was the use of a systematic approach, consistent with MRC guidance for the development of complex interventions,126 to intervention development. We were able to validate the key findings from the rapid evidence review in the qualitative work that was carried out to support intervention development, and through discussion with the Stakeholder Advisory Group. In doing so, we have developed a detailed specification and logic model for the Mam-Kind intervention.

We engaged a stakeholder expert advisory group during the early stages of the study, which enabled us to adopt an interactive and iterative process for intervention development. This allowed us to develop an intervention that was informed by the views and perspectives of intended service users, peer supporters and health-care service providers and to ensure that the design of the intervention included practical strategies for recruitment, consent and data collection. We identified the key feasibility questions to be addressed prior to a full trial, and designed our study to address these.

However, there were some limitations. The response rate to our web-based survey was low. To mitigate this, we used publicly available data from organisational websites to supplement our survey findings. Although this provided data about the availability of peer support, it did not provide detail about how services were run or how training and supervision for peer supporters were provided. We mapped the data on BFPS availability to NHS trust and health board areas and, in doing so, we have had to assume that the responses for each health-care area were applicable across the board. This may not be the case for larger NHS trusts in which the availability of services may vary greatly within the trust; however, our survey was not sensitive enough to capture availability at smaller geographical levels. We did not have the resources within the scope of the project to directly contact local authority or voluntary groups to obtain further information about available services.

We included fathers in our intervention development work and used a pre-existing fathers’ parenting group to ensure that they felt comfortable talking to us about this emotive topic. Although this had the advantage of securing open talk about infant feeding, it attracted a smaller number of fathers than we had hoped to engage with.

Delivery of the Mam-Kind intervention was hindered in some cases by significant delays in the process of obtaining research passports for Mam-Kind buddies, which resulted in their inability to access postnatal wards and provide support within 48 hours, as specified by the intervention. A further limitation was that, although we managed to recruit in areas with high levels of social deprivation and low breastfeeding initiation rates, our participants had, on average, high levels of education and were mostly employed. The small number of mothers recruited who were aged < 20 years limited our ability to assess the acceptability of the Mam-Kind intervention to this group. The sample that we recruited may not be representative of the whole population in the study sites. A further limitation was the increase in loss to follow-up at 8 weeks, which has implications for the RCT; these are described in Strategies for the recruitment and retention of peer supporters and participants. These areas need further consideration before undertaking a future RCT. We have discussed these issues with our Stakeholder Advisory Group and used their input to develop strategies for addressing them in a full trial.

This study has demonstrated the levels of study uptake, levels of intervention usage, follow-up rates and levels of data completion. It has also provided an indication of how some of these processes may either be streamlined or improved (e.g. by providing additional training and redesigning some of the participant materials).

The incorporation of an early-stage assessment of the requirements and parameters for an economic analysis in a future trial has allowed a provisional assessment of this, with evidence from the feasibility study providing a useful basis for the development of a suitable framework. Because of resource constraints, in-depth examination was not intended and the provisional assessments (e.g. early assessment of the intervention costs associated with the Mam-Kind intervention) must be interpreted with appropriate caution. A key learning point was the need for health economics to be embedded into the feasibility study to ensure that important insights and lessons can be used to develop the framework for an economic analysis within a future trial. The considerations given to economic evaluation in this feasibility study provide a good basis for further development of the design of an economic analysis in a full trial.

Patient and public involvement in the study

In this section we describe how we involved the public in the design and conduct of the research. The aim of patient and public involvement (PPI) in this study was to improve the design, conduct and dissemination of the research. We achieved this using a range of methods to fulfil the following objectives:

• inform the design of the study so that it addresses relevant questions and is acceptable and feasible from the perspective of service users and members of the public

• contribute to oversight of the conduct of the research

• provide a lay perspective on the analysis and interpretation of the data

• facilitate public engagement and dissemination activities

We included a PPI representative as a co-applicant on the project and as a member of the Study Management Group for the duration of the project, to advise on study conduct and the materials to be used with service users and the public (Box 6). Our representative was a member of the Involving People (Wales) Network who had completed the network formal training days and was provided with ongoing support from the Involving People Network. We ensured that she was briefed about the project and was given opportunities to debrief with a member of the research team. She also attended the stakeholder events (see Box 6) and assisted with communication with other mothers and peer supporters through her own contacts with peer-support groups and online communities for breastfeeding mothers/peer supporters.

We collaborated with the NCT, the UK’s largest charity for new and expectant parents, to secure additional PPI input. The Research Operations Manager, a collaborator on this study, was our link with the NCT. The NCT assisted with publicising the study and disseminating findings via web-based information, its information for volunteers and Maternity Services Liaison Committee representatives and through articles in NCT’s range of publications for parents, practitioners and commissioners.

We formed an online closed Facebook group to gather additional feedback from mothers and peer supporters. There were 30 members of this closed group. We asked them to provide feedback on the name of the study and to review patient-facing materials, such as the patient information sheet and the leaflets. The group also assisted with disseminating information about the peer-support job opportunities associated with the feasibility study.

Our Stakeholder Advisory Group included representation from two service users, three peer supporters and breastfeeding groups. The group worked across all of the study objectives, advising on the development of the intervention, the study materials and the management of any potential issues that arose during the course of the study and the interpretation of the findings, plans for dissemination and future plans for the research (including the design of a full trial). The group met on four occasions over the course of the project.

Our PPI representatives on the Study Management Group and the Stakeholder Advisory Group were mothers who had had a child (or children) in the last 5 years and who had experience of breastfeeding. Mothers were welcome to bring young babies along to meetings or childcare costs could be provided for older children if required. We provided payment at the Involving People Network-recommended rate (equivalent to £150 per day) to our PPI representatives for attending meetings and stakeholder events.

The PPI input on this project was invaluable in developing study procedures that were feasible and acceptable, given the sensitive nature of the topic at hand and that we were working with women during a busy and potentially vulnerable period in their lives. This meant that all study procedures and service user-facing materials were designed with the needs of mothers in mind. This mother-centred ethos was carried through all aspects of the study, ensuring that the information was relevant and easy to read, the burden of the study assessments was kept to a minimum and women were engaged at all stages of the research. It has also been invaluable in planning and designing the next steps in this research, particularly in terms of ensuring that planned future work is relevant and acceptable and that barriers to participation are addressed at the early design stage.

Implications for a future randomised trial

Overall, we consider that the intervention can be delivered with fidelity and is acceptable to women, and that the core study procedures appear to be feasible to deliver. Further, we still consider that there is a pressing clinical need to better support women to breastfeed, as only 1% of mothers in the UK currently exclusively breastfeed for 6 months, which is the WHO-recommended duration. 129 Currently, there remains no good UK-based evidence of effective interventions to support breastfeeding, as indicated by the findings of our rapid review. In this section, we describe the implications for a future RCT to test the clinical effectiveness and cost-effectiveness of the Mam-Kind intervention for breastfeeding. We have focused on recommendations for the design of a future trial, strategies for the recruitment and retention of participants, measurement of primary and secondary outcomes and a framework for economic evaluation.

Conclusion

We have established that there is no standardised provision of BFPS in the UK and that the models that are currently in use are not explicitly linked to theory nor have they been robustly evaluated for effectiveness. We have systematically developed a BFPS intervention that uses a MI approach, using the BCW and informed by service users, peer supporters and health-care professionals. Previous UK RCTs have not shown that BFPS interventions are effective at increasing breastfeeding maintenance, possibly because of their use of low-intensity interventions that relied on mothers to seek support. The Mam-Kind intervention, by contrast, is characterised by proactive Mam-Kind buddy (peer-supporter)-led support in the first 2 weeks post-partum. We have tested and established the feasibility of delivering the Mam-Kind intervention, with high uptake of the intervention and fidelity to intervention content. We have addressed the findings from the feasibility study by applying modifications and refinements to the intervention specification and have developed study materials that can be adapted for use in a full trial. We have established data collection methods for outcome data, including the feasibility of collecting salient information about the wider costs of the Mam-Kind intervention. This feasibility study was focused on developing and testing the delivery of the Mam-Kind intervention and the feasibility of research processes, such as data collection. A full RCT is now required to test the effectiveness of the Mam-Kind intervention at increasing the duration of breastfeeding, with a built-in pilot phase to ensure the recruitment and retention of participants.

Contributions of authors

Shantini Paranjothy (Mansel Talbot Professor of Preventive Medicine, Population Health) was the Chief Investigator. She led the design of the study, interpretation of the data and drafting and approval of the final study report and provided oversight for all aspects of the study.

Lauren Copeland (Research Associate, Behaviour Change and MI) co-managed the study, providing a significant contribution to the final report, and led on several chapters. She also contributed to the design of the intervention and to the evaluation of MI fidelity and was involved in the recruitment of participants and data collection.

Laura Merrett (Research Associate, Midwifery) co-managed the study and contributed to the recruitment of participants and data collection. She also conducted, analysed and wrote up the qualitative research as part of the process evaluation and provided input into the final report.

Aimee Grant (Research Associate, Qualitative Research) oversaw the qualitative research. She conducted data collection for the peer-supporter interviews, provided an analysis of the survey data and wrote up the relevant part of the report.

Rhiannon Phillips (Research Fellow, Health Psychology) applied the BCW to the qualitative data analysis to develop the intervention. She contributed to the intervention development chapter of the report.

Nina Gobat (Research Fellow, MI and Behaviour/Motivation) developed and delivered the MI training and supervision to peer supporters, including integration of MI in the intervention. She also contributed to the process evaluation of MI fidelity and commented on the final report.

Julia Sanders (Consultant Midwife and Reader in Midwifery) was a study co-investigator providing senior clinical input to the study and was involved in the design of the study, intervention development and delivery and report writing.

Deborah Fitzsimmons (Professor of Health Outcomes Research and Academic Director, Health Economics) supervised the health economics aspects of the study, wrote the health economics plan and drafted the health economics chapter.

Billie Hunter (Royal College of Midwives Professor of Midwifery, Qualitative Research and Midwifery Practice) contributed to the conception and design of the study and the interpretation of the data and commented on and revised the final study report.

Sian Regan (PPI Representative, Involving People Network) contributed to the Stakeholder Advisory Group, provided feedback on the study documents and provided reflections in the PPI section of the report.

Rebecca Playle (Senior Lecturer, Medical Statistics) was responsible for the statistical design and analysis of the study.

Amy Brown (Associate Professor in Child Public Health, Infant Feeding Research) contributed to the design of the study and commented on the final report.

Sally Tedstone (Infant Feeding Specialist) contributed as a midwifery advisor to the development of the intervention, site set-up and breastfeeding peer-supporter training.

Heather Trickey (Research Associate, Social Science/Public Health, Infant Feeding Policy and Intervention) contributed to the rapid evidence review and to the qualitative research for the intervention development.

Mike Robling (Director of Population Health Trials, Psychology) was a co-investigator and funding co-applicant. He contributed to the design of the study, interpretation of the data and drafting and approval of the final study report.

Publications

Grant A, McEwan K, Tedstone S, Greene G, Copeland L, Hunter B, et al. Availability of breastfeeding peer support in the UK: a cross-sectional study. Matern Child Nutr 2017;e12476.

Trickey H, Thomson G, Grant A, Sanders J, Mann M, Murphy M, Paranjothy S. A realist review of one-to-one breastfeeding peer support experiments conducted in developed country settings. Matern Child Nutr 2017; in press.

Data sharing statement

We will make data available to the scientific community with as few restrictions as feasible, while retaining exclusive use until the publication of major outputs.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.

Search methods: breastfeeding peer support

Ovid MEDLINE(R) (1946 to January week 4 2016)

Health economic review search strategy

Ovid MEDLINE(R) (1946 to November week 1 2014)