An assets-based intervention before and after birth to improve breastfeeding initiation and continuation: the ABA feasibility RCT

Article history

The research reported in this issue of the journal was funded by the PHR programme as project number 15/53/04. The contractual start date was in November 2016. The final report began editorial review in November 2018 and was accepted for publication in May 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Kate Jolly reports grants from the National Institute for Health Research (NIHR), local authority funding for the intervention and part-funding by NIHR Collaborative Leadership for Applied Health Research and Care West Midlands during the conduct of the study. Alongside her Cardiff University role, Heather Trickey worked part time as a senior researcher for NCT (London, UK) during the period in which the research was conducted. NCT provides breastfeeding peer support services. NCT volunteers were not included in this study. Pat Hoddinott is a member of the Health Technology Assessment Commissioning Board (March 2014 to present). She is working on a funding application to take forward the FEeding Support Team (FEST) feasibility trial that she led and that is cited in this report. The FEST feasibility trial informed parts of the design of the Assets-based feeding help Before and After birth study. Alice Sitch is supported by the NIHR Birmingham Biomedical Research Centre, University Hospitals Birmingham NHS Foundation Trust and University of Birmingham, UK.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Clarke et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. 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.

2020 Queen’s Printer and Controller of HMSO

The benefits of breastfeeding for health and well-being

Breastfeeding (defined as the baby being put to the breast or receiving breast milk on at least one occasion) is associated with short- and long-term benefits to both the breastfed infant1–10 and the mother. 11 Internationally, the largest health gains are seen in low-income countries, as breastfeeding protects against infant mortality by reducing acute infections in infants. 1 However, considerable health gains from breastfeeding are also possible in high-income countries. Currently, only around 12% of babies in the UK are exclusively breastfed at 4 months. If this figure increased to 45% of women in the UK breastfeeding exclusively for 4 months, then at least £17M could be saved annually in NHS treatment costs for common acute illnesses in infants, with additional longer-term gains for mothers and children. 12,13

The benefits of breastfeeding are considerable. The evidence has been collated in systematic reviews1,9 and is consistent across cohort studies in a range of settings and from a randomised controlled trial (RCT) of a breastfeeding support intervention with long-term follow-up of the children. 14 For the infant and child, any breastfeeding is associated with reduced risk of gastrointestinal infection by 63% [95% confidence interval (CI) 50% to 94%],10 sudden infant death syndrome by 36% (95% CI 10% to 49%),15 otitis media by 33% (95% CI 28% to 38%),2 asthma when aged 5–18 years by 12% (95% CI 5% to 18%),3 being overweight in the future or obese by 26% (95% CI 22% to 30%),7 type 2 diabetes mellitus by 35% (95% CI 14% to 51%)7 and malocclusions by 68% (95% CI 60% to 75%). 5 Exclusive breastfeeding for > 4 months reduces the risk of hospital admission for lower respiratory tract infections in the first year by 72% [risk ratio (RR) 0.28, 95% CI 14 to 54]10 and for 3–4 months reduces the risk of eczema by 26% in children at < 2 years. 3 Exclusive breastfeeding for > 3 months is associated with a reduced risk of type 1 diabetes mellitus of up to 30%. 15 Meta-analyses show that being fed breast milk is associated with a 58% (95% CI 4% to 82%) reduced risk of necrotising enterocolitis in pre-term infants16 and with reduced mortality. It is also associated with improved performance in intelligence tests. 8 Mothers have a reduced risk of breast (26%, 95% CI 21% to 31%) and ovarian (37%, 95% CI 29% to 44%) cancers if they breastfeed for > 12 months11 and lower post-menopausal body mass index if they have ever breastfed. 17

Breastfeeding rates and duration in the UK

Breastfeeding duration in the UK is among the lowest worldwide, with routinely collected data and 5-yearly infant feeding surveys18–20 showing relatively small improvements over the past two decades, particularly for rates of exclusive breastfeeding. Although breastfeeding initiation increased from 76% in 2005 to 81% in 2010, exclusive breastfeeding at 6 weeks increased only from 21% to 23% during the same period. There are considerable health inequalities, despite government initiatives; breastfeeding initiation and duration rates are lowest in teenagers (58% initiated breastfeeding in 2010), women living in socioeconomically disadvantaged circumstances, women with lower educational outcomes and white women. 18 In 2010, 90% of UK mothers in managerial and professional occupations breastfed, compared with 74% of those in routine and manual occupations and 71% among those who had never worked. 18

The World Health Organization21 recommends exclusive breastfeeding for 6 months to optimise infant and maternal health, a recommendation endorsed by UK governments;22 however, < 1% of infants in the UK receive this. 18 Data from the Infant Feeding Survey 201018 show that the steepest decline in breastfeeding occurs soon after birth: 81% of women who give birth initiate breastfeeding but only 69% of babies are breastfed at 1 week, 66% at 2 weeks, 55% at 6 weeks and 34% at 6 months. Rates of exclusive breastfeeding are even lower: 46% at 1 week and 23% at 6 weeks. 18 More recent data collected from local authorities in England show that 44.4% of babies receive breast milk at 6–8 weeks, with the range being 19.3% to 75.6%. 23 Mothers express dissatisfaction with breastfeeding care,24,25 and 30% report feeding problems in the early weeks. 18 A 2017 survey by the National Federation of Women’s Institutes and the NCT (formerly known as the National Childbirth Trust) identified baby feeding as the greatest area of unmet need for support. 26 Women who reported that they did not receive support for breastfeeding difficulties in hospital or at home were more likely to discontinue breastfeeding within the first 2 weeks. 18

Effectiveness of peer support for breastfeeding initiation and continuation

In the UK, breastfeeding peer support has been widely recommended as a means of increasing breastfeeding initiation and continuation rates among women from disadvantaged communities. 27,28

Breastfeeding peer support has been defined as ‘support offered by women who have received appropriate training and either have themselves breastfed or have the same socio-economic background, ethnicity or locality as the women they are supporting’. 29 From a theoretical perspective, Dennis30 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’. In comparison with health-care professionals, peer supporters may be considered more approachable and operate as positive role models to whom women can relate because of their direct experience of the challenges of breastfeeding, and in contexts where breastfeeding may not be the social norm. 31

A systematic review32 of breastfeeding peer support interventions reported a significant increase in breastfeeding initiation in three trials that targeted this support at pregnant women who had decided to breastfeed (relative risk for not initiating breastfeeding 0.64, 95% CI 0.41 to 0.99), but no difference in the three trials that offered universal peer support to all pregnant women (relative risk for not initiating breastfeeding 0.96, 95% CI 0.76 to 1.22). Heterogeneity in the meta-analysis of targeted breastfeeding peer support was high, which might be because of differences in the settings and context where the peer support was offered and the intensity of the interventions. 32

A systematic review29 to assess the impact of breastfeeding peer support on breastfeeding continuation rates reported significant effects on any breastfeeding rates and exclusive breastfeeding rates at the last study follow-up (relative risk of not breastfeeding at last follow-up 0.85, 95% CI 0.77 to 0.94, and 0.82, 95% CI 0.76 to 0.88, respectively). Heterogeneity was high and was explored using subgroup analyses and meta-regression. Peer support interventions were found to have a significantly greater effect on any breastfeeding and exclusive breastfeeding in low- or middle-income countries than in high-income countries. However, in high-income countries, peer support reduced the risk of not breastfeeding by 7% (0.93, 95% CI 0.87 to 1.00). The risk of non-exclusive breastfeeding decreased significantly, by 10% (0.90, 95% CI 0.85 to 0.97). No significant effect on any breastfeeding or exclusive breastfeeding was observed in the three UK-based studies. Peer support had a greater effect on any breastfeeding rates when given at higher intensity (five or more planned contacts; p = 0.02).

A 2017 Cochrane review33 of support for breastfeeding mothers found strong evidence that providing extra professional, lay or peer support for women who wish to breastfeed increases the duration of exclusive breastfeeding (cessation of exclusive breastfeeding at 6 months, average RR 0.88, 95% CI 0.85 to 0.92) and of babies receiving breast milk alongside other liquids or solids (cessation of any breastfeeding at 6 months, average RR 0.91, 95% CI 0.88 to 0.95). The review found that the effects of lay support were broadly similar to those of professional support. 33 Lay support is broader than peer support and does not require that the supporter and mother share experience or characteristics. Nine trials of lay support compared with usual care reported a RR of stopping breastfeeding before the last study assessment up to 6 months of 0.85 (95% CI 0.77 to 0.93), but with considerable heterogeneity, and 13 trials reported a reduced risk of stopping exclusive breastfeeding before the last study assessment (RR 0.76, 95% CI 0.65 to 0.87). However, the generalisability of these findings to the UK context is uncertain. Nine UK trials since 2000 providing additional support using a range of models, including peer, lay and professional support, have failed to improve breastfeeding outcomes significantly. 34

Similar systematic review results to those for peer support29 were reported by Renfrew et al. 35 in relation to frequency of planned contact for lay support. Interventions with four to eight contacts had a larger effect size than combined interventions with fewer than four planned contacts in trials with a usual care control group.

There is evidence that, to be effective, peer support should be offered proactively. In Canada, peer supporters with 2.5 hours’ training proactively telephoned women (n = 256) using a woman-centred format;36 the relative risk for any breastfeeding at 4 weeks was 1.10 (95% CI 1.01 to 2.72).

Preliminary research suggests that early proactive telephone support might suit a UK context. 37 In a pilot trial (69 women),37 intensive early proactive telephone support (not peer support) for women who initiated breastfeeding, delivered by a postnatal ward feeding team with personal breastfeeding experience, increased any breastfeeding by 22% (RR 1.49, 95% CI 0.92 to 2.40) at 6–8 weeks compared with the opportunity to access reactive telephone support from the team. A Cochrane review38 of telephone support for women during pregnancy and up to 6 weeks after the birth showed that women who had received a telephone support intervention were more likely to be exclusively breastfeeding (RR 1.51, 95% CI 1.19 to 1.93) at 3–6 months postpartum than those in the comparator group, but this included only three trials, and no difference was observed in the four trials that reported breastfeeding at 4–8 weeks postpartum. 38

A UK study39 applying a theory of constraints model to investigate the barriers to effective lay feeding help recommended that (1) to gain wider acceptability, interventions should be mother-centred (rather than breastfeeding-centred), both enabling breastfeeding and giving help with formula milk, (2) there should be a greater focus on the early weeks after the birth, as establishing breastfeeding can be difficult and mothers frequently stop feeding before they had planned, and (3) support should be offered proactively to improve the take-up of breastfeeding.

A recent realist review40 of breastfeeding peer support interventions in high-income countries found that breastfeeding peer support appears to rely on a chain of mechanisms firing in sequence. The realist review found that intervention design should take account of needs as perceived by the target population; integration with health professional care can be critical, and so ensuring mutual respect and overcoming local barriers to integrated working practices, collaboration and feedback are important. Peers need to be accessible when mothers most need support; support around the time of the birth can help mothers who are unsure to firm up decisions to breastfeed. Peer support also needs to be proactive, as reactive support tends to be used by mothers who are motivated or confident, and is unlikely to be effective in improving rates overall. Mothers value friendly, competent and proactive peers, and these qualities may outweigh social similarity. Mothers who experience a warm and affirming relationship with their peer supporter often feel helped to overcome challenges and to meet their feeding goals. The review also found that peer supporters are motivated when they feel valued and are demotivated when their offers of help are rejected. As a result, peers tend to focus their energy on mothers who seek support and seem to be appreciative.

These findings from the realist review are in line with a meta-synthesis41 of women’s perceptions and experiences of breastfeeding support that recommended person-centred approaches and qualitative studies24,42 of women’s experiences of infant feeding support that found that structured approaches to support-giving are unpopular24 but flexible support is acceptable. 42 How breastfeeding interventions are delivered and the intervention-context fit are important determinants of outcomes. 43 The timing of support in the very early postnatal period may be an important feature of effective breastfeeding support. 40,44,45 Continuity of targeted peer support by having an antenatal visit and postnatal support from the same local supporter is associated with psychosocial benefits for mothers, health professionals and peer supporters. 42

Existing provision of breastfeeding support in the UK

In hospital, midwives deliver breastfeeding support, with breastfeeding counsellors and hospital peer supporters also available in some areas. However, length of stay following a singleton vaginal delivery in the UK is one of the shortest internationally (1.5 days). 46 Many women, including first-time mothers, go home 6 hours after giving birth; 19.8% of women in 2016/17 were discharged on the same day as the birth. 47 This gives insufficient time to establish breastfeeding. Reduced hospital stay following birth provides a suboptimal context to support establishing breastfeeding for many new mothers. Care is transferred from midwives to health visitors between 10 and 30 days postnatally. Much community breastfeeding support is provided by lay workers in children’s centres and by peer supporters. Breastfeeding peer support is offered by a range of organisations, including voluntary and charitable organisations, local authorities and the NHS. Peers may be paid or voluntary, and training and supervision are offered by a range of providers. The UNICEF UK (United Nations Children’s Fund, London, UK) Baby Friendly Initiative stage 1 accreditation48 requires a meaningful discussion about infant feeding in the antenatal period and identifies that this might be delivered by a peer supporter. Additionally, for accreditation, local maternity services are required to have mechanisms in place to enable mothers to access support for breastfeeding with basic problem-solving via their local maternity service or other local routes, for example breastfeeding support groups or peer support, and to ensure that mothers know about these services.

The characteristics of peer support provided for pregnant and breastfeeding women across the UK are not routinely collected. A survey in 2014 of all known infant feeding co-ordinators in the UK had a 19.5% response rate, and covered of 58% of NHS trust/health board areas. 49 This study identified wide availability of breastfeeding support across the UK, with peer support available in 78% of areas and breastfeeding support groups available in 90% of areas. However, these may not be representative of all areas and the support may be provided only in selected localities within trust/health board areas. The survey identified a lack of standardisation of the provision of breastfeeding peer support across the UK and the challenging context of limited financial support. Services were reduced and increased in line with funding availability. 49 The most common providers were third-sector organisations, such as the NCT and the Breastfeeding Network, and most peer supporters were volunteers. 49 In the 2010 Infant Feeding Survey,18 69% of women reported being given the details of a voluntary organisation or community group that helped new mothers to breastfeed, and 64% were aware of the National Breastfeeding Helpline. A report of breastfeeding support in London50 found that the proportion of new mothers receiving breastfeeding support from a peer supporter varied from < 5% to 52% in London boroughs.

Research into the role of UK fathers in supporting breastfeeding reported that they wanted to be able to support their partners, but they were often excluded from antenatal breastfeeding education or were considered unimportant in postnatal support. 51 Many fathers feel ignored throughout the whole journey of pregnancy and postnatal care. 52 Men want more information about how they can practically support their partner,51 and a survey of women with young children suggested that engaging the fathers in breastfeeding education was as a way of increasing breastfeeding support. 53

Apart from the father of the child, many other ‘significant others’ influence a woman’s decisions about breastfeeding,54 and the composition of a woman’s social network changes over time. 55 Women’s support needs can be mapped in an infant feeding genogram55 that records the feeding history of family and friends and the strength of relationships with these social network members. This genogram can be used as a tool to support discussion around breastfeeding and to identify support needs.

Information needs of and risks for mothers who feed their babies formula milk

Previous UK studies to promote breastfeeding have focused solely on breastfeeding, excluding any discussion of formula feeding with mothers. The evidence shows that, for infant-feeding interventions to be acceptable to women, it is important to address issues related to mixed feeding and formula feeding. 40,56 This need is now explicitly recognised by several UK key providers of peer support; for example, NCT (the UK’s largest charity for expectant and new parents) policy is to provide support to all women with their infant feeding decisions, regardless of how they are feeding their babies. 57 The 2010 Infant Feeding Survey showed that 54% of babies had received formula milk by the age of 1 week, 88% had received it by 6 months and 95% had received it by 9 months. 18 Furthermore, the survey highlighted that half of mothers who prepared powdered infant formula did not follow all three key NHS recommendations (making only one feed at a time, making feeds within 30 minutes of the water boiling and adding the water to the bottle before the powder), which are intended to reduce the risk of infection and overconcentration of feeds. Other studies have also highlighted a high frequency of errors in formula feed preparation. 58,59 Current guidance for mothers is available on the NHS website60 and includes a 13-point set of instructions for making up a bottle of formula. The evidence indicates that an intervention to increase breastfeeding rates that fails to address mothers’ needs in relation to formula feeding (particularly in a culture where mixed feeding is common) risks alienating potential beneficiaries, limiting intervention reach and retention, and decreasing the likelihood of achieving breastfeeding-related outcomes. 40,61 Improving the preparation of formula feeds will incur additional infant health benefits from reduced gastrointestinal infections. 59 Moreover, by focusing on the mothers’ needs, there may be less guilt associated with feeding decisions. 62

Assets-based approaches in public health

The use of peer support and encouragement to access community support for breastfeeding and social opportunities for new mothers is an exemplar of an assets-based approach to public health. An assets-based approach focuses on the positive capabilities of individuals and communities, rather than solely on their needs, deficits and problems. This approach is linked to the theory of salutogenesis (health origin),63,64 which conceptualises health as a continuum and focuses on what helps individuals retain positive health and well-being rather than on factors that cause disease. 63–65 It also has parallels with economic theories of capability and well-being, from a broad physical, psychological, social and community perspective. 66

Assets-based approaches are about recognising and making the most of people’s strengths to change the balance between meeting the needs of people and communities and nurturing their strengths and resources. 67 This is accompanied by a corresponding shift in focus from the determinants of ill health to the determinants of health and well-being. Although assets can include material resources,68,69 in public health more typically, the primary focus is on valuing individual and collective psychosocial attributes. These include confidence, optimism, self-esteem, knowledge and skills, as well as features of social capital such as social networks and reciprocity. 70–73

Longitudinal qualitative research with families living in disadvantaged areas suggests that family well-being rather than potential future health benefits is the outcome that matters most and that drives decisions to stop breastfeeding. 56 In the context of breastfeeding, assets may include intrinsic personal resources such as willingness to ask for and accept help, self-efficacy in relation to infant feeding,74 and motivation and drive to maintain feeding. 74–77 These assets also include extrinsic resources such as availability of social support from partner,78–80 family and friends, wider social networks of new mothers and women who have breastfed, and community assets such as children’s centres, mother-and-baby groups, breastfeeding groups or baby cafes. Local breastfeeding peer supporters are also community assets for breastfeeding. Hopkins and Rippon’s73 theory of change approach to assets-based working focuses on recognising and mobilising assets. An assets-based approach, by focusing on a woman’s priorities, is woman-centred.

Rationale for the ABA study

In 2015, the National Institute for Health Research Public Health Research programme called for studies to determine the effectiveness of community-based interventions that promote the uptake and maintenance of breastfeeding. Our study aimed to assess the feasibility of delivering a new Assets-based feeding help Before and After birth (ABA) infant feeding helper (IFH) intervention within a RCT. The ABA intervention was built on systematic review evidence,29,32,33,41 behaviour change theory,81 extensive qualitative research41,42,56,82 and learning from the FEST (FEeding Support Team) pilot trial about woman-centred feeding support after birth. 37,83

The study took place in geographical areas of socioeconomic disadvantage, as the largest potential public health gain is obtained from improving health outcomes for disadvantaged infants. 84

The ABA intervention used an assets-based approach, drawing on the community, social network, family and personal assets of each woman. This enabled the extent of support to be tailored to the assets a woman has available for infant feeding. This assets-based approach was enhanced with behavioural change theory. In addition, we developed a new feeding helper approach that is woman-centred, aims to establish a strong supportive relationship with continuity of care from pregnancy until after birth, respects a woman’s choices, is non-judgemental and discusses both breastfeeding and formula feeding issues, should a mother wish to. 56,62,82,83 This is because trials of breastfeeding peer support in the UK have had unexpected null results contrary to the worldwide systematic review evidence. 29 One hypothesis is that women who engage with breastfeeding-centred intervention research are those who are highly motivated to breastfeed. In taking a broader feeding approach, we were compliant with current UNICEF guidance,85 and at the same time aimed not to alienate women who were considering mixed or formula feeding43,56,62 by using the term ‘infant feeding’ in ABA information materials.

Peer support is a behaviour change technique found to be effective in increasing breastfeeding initiation and continuation. 29,33,35,86 Peer support is recommended by the National Institute for Health and Care Excellence (NICE);87 many programmes are in existence in the NHS and are suggested by UNICEF as a potential mechanism for achieving effective onward community support. 85

This assets-based feeding intervention is a new approach to peer support that seeks to overcome some of the pitfalls identified in previous studies, while building in methods to enable women to identify and activate assets within their family and friendship networks and in the wider community.

Aim and objectives

Setting

The study was undertaken in two distinct geographical areas in England (site A and site B). Both areas had existing programmes of peer support, but these were provided reactively through, for example, midwife referral or self-referral. In site A paid peer supporters employed by a social enterprise organisation delivered the programme, whereas in site B peer supporters were volunteers managed by a national charity. The sites were selected from five that were initially identified as interested in participating in the study. Sites were chosen (1) to reflect the diversity of existing peer support services, but those with no proactive peer support offered antenatally, (2) as they had relatively high levels of socioeconomic disadvantage and low rates of breastfeeding initiation and continuation, and (3) because they were reasonably local to the investigators to enable oversight of the study.

Study design

We undertook a feasibility individually randomised controlled trial (1 : 1) in two UK sites with a mixed-methods process evaluation.

Study management

The ABA study was overseen by a Trial Steering Committee (TSC) made up of two subject experts, a statistician and a public representative. A study management group, comprising the principal investigator, the trial co-ordinator and the 10 co-investigators, met regularly to guide study conduct.

Ethics approval and study registration

Ethics approval was obtained on 28 November 2016 from South West – Cornwall and Plymouth Research Ethics Committee (16/SW/0336). The study was registered with the International Standard Randomised Controlled Trial Register Number ISRCTN14760978. During the course of the study, some minor revisions were made to the protocol (see Appendix 1). The final protocol was published as a journal article. 88

Participant identification

We aimed to recruit 100 women to the study (at least 50 from each site), with half randomly allocated either to the intervention group or to the usual care group. We hoped to recruit sufficient numbers of teenagers, women of low socioeconomic status and women with limited social network breastfeeding exposure to allow us to investigate their experiences of the intervention. The two study sites were selected to reflect our target population.

Community midwives in the study areas were asked to hand out a summary participant information leaflet (PIL) to women who were pregnant with their first child at their 25-week antenatal appointment. At their 28-week antenatal appointment, women were approached in clinic by a researcher. The researcher provided the woman with further information about the study, including a full PIL (see Appendix 2), and gave her an opportunity to ask any questions. The woman was then asked if she would like to take part in the study. Women were given the option of signing up to the study there and then, or having time to think about it and/or discussing with others before contacting the researcher to arrange a time and place to sign up. Women were able to enrol in the study only up until 32 weeks’ gestation; this was to allow sufficient time for intervention participants to meet with their IFH before the birth. Researchers completed screening logs to record the number of women who were approached. At recruitment, women were told that if they completed and returned follow-up questionnaires at both 8 weeks and 6 months, they would receive a £25 shopping voucher at the end of the study to thank them for their time.

Recruitment ran from 28 February 2017 until 23 May 2017 in site A and from 21 April 2017 until 31 August 2017 in site B. Follow-up took place between 24 April 2017 and 12 March 2018 in site A and between 21 June 2017 and 23 May 2018 in site B. Recruitment ended when at least 50 participants had been recruited from each site.

Inclusion and exclusion criteria

Women were eligible to take part in the study if they were:

• pregnant with their first child (excluding previous stillbirth)

• aged ≥ 16 years.

Women were excluded if they had had a previous live birth or were aged < 16 years.

We decided to include only first-time mothers as they would have had no experience of infant feeding and might have been more likely to be influenced by the intervention. This is because mothers often repeat the infant feeding method they used for their first baby with their second baby. 89

Consent-taking process

The researcher checked the woman’s eligibility to participate in the study before asking her to complete, sign and date three copies of the consent form (see Appendix 3). The consent form was also signed and dated by the researcher. One copy of the consent form was given to the participant, one was kept by the research team and the third was stapled into the woman’s maternity notes. Details of the discussion about informed consent were recorded in the woman’s maternity notes (including date, name of study, discussion summary, PIL and consent form version numbers). Participants’ contact details were recorded and a baseline questionnaire was completed at the time of recruitment. Women were given a fridge magnet with the ABA study team contact number and were asked to notify the team as soon as their baby was born.

Randomisation

Women were randomised to intervention or usual care in a 1 : 1 ratio. The randomisation process differed by site.

In site A, a randomisation list was developed by the clinical trials unit, minimised by age group (< 25 years and ≥ 25 years). The list was stored securely and was not available to the researcher enrolling participants. After the participant had signed the consent form, the researcher telephoned the randomisation service who checked the participant’s eligibility and assigned her to the intervention or usual care. The researcher informed the participant there and then of her allocation. If the telephone randomisation service was unavailable, the researcher contacted the woman later, by letter, to inform her of her allocation.

In site B, a different method of randomisation was required to be able to match the number of IFHs available in the different areas of the study site (in site A, all IFHs were available to cover the whole study site). Therefore, block randomisation was used to randomise multiple women from each area of the site. Each block of women was randomised simultaneously by a researcher who was not undertaking recruitment; the recruiting researcher was then informed of allocation and notified women in writing.

Intervention design

Intervention design was informed by the Medical Research Council Complex Interventions and RE-AIM (Reach, Efficacy, Adoption, Implementation, Maintenance) frameworks. 90,91 We also used information from systematic reviews, surveys and qualitative studies and discussions with our patient and public involvement (PPI) group to ascertain barriers to both breastfeeding initiation and breastfeeding continuation. We used the behaviour change wheel framework [in which behaviour is analysed in context with respect to capability, opportunity, motivation and behaviour (COM-B)] alongside the theoretical domains framework to identify a number of behaviour change functions and behaviour change techniques (BCTs) from the behaviour change taxonomy. 81,92 Following this, we considered possible BCTs using the APEASE (Affordability, Practicality, Effectiveness and cost-effectiveness, Acceptability, Side-effects/safety, Equity) criteria. Components of the intervention were identified that were simple, low cost, practical and acceptable. A review of multicomponent incentive interventions to support breastfeeding had mapped BCTs and discovered social support to be predominant. 93 Social support is fundamental to peer support. 30

The chosen BCTs, with definitions and prespecified examples based on the ABA intervention, are shown in Table 1. Core and non-core BCTs for the antenatal part of the ABA intervention were agreed by the research team (core BCTs were social support and restructuring the social environment; non-core BCTs were emotional social support and instruction of how to perform behaviour). Some of these BCTs overlapped with the social support and use of social networks within the assets-based approach.

Table 2 provides details of the rationale for including the intervention components drawing on behaviour change theory and assets-based approaches.

The ABA intervention consisted of proactive peer support underpinned by behaviour change theory and an assets-based approach. The intervention delivered person-centred care41 and used best evidence in terms of setting and frequency, duration and manner of support provision from the ABA IFH. The ABA intervention aimed to be inclusive of all feeding methods (i.e. breastfeeding, formula or mixed feeding) and to provide support to all women.

Before the intervention commenced, researchers developed an ‘assets leaflet’ at each study site that was designed by a graphic designer. This leaflet (developed with the assistance of local contacts and using internet searches) was specific to the study areas and included information on local community ‘assets’ (e.g. antenatal or postnatal groups, breastfeeding drop-in centres, details of local breastfeeding counsellors and baby groups), as well as details of national helplines and internet resources. The leaflet, entitled ‘What’s available locally for you and your baby?’, had input from two PPI groups (mothers of young babies attending children’s centre groups) that provided constructive feedback on making the leaflet more user-friendly. Quotations from previous qualitative work were included in the leaflet (e.g. concerning the usefulness of breastfeeding drop-in centres), as well as tips on what to do when feeling uncomfortable about going along to a new group. Contact details for the ABA study were included on the leaflet, along with a space for the IFH to put their name and contact details. All details were checked prior to the start of the intervention to make sure they were up to date. For an anonymised example of the leaflet, see Appendix 4.

The two PPI groups were also asked for their opinions on a library of text messages produced by the research team, intended to be used by IFHs to engage with and support women in the intervention group. These texts aimed to take a woman-centred approach, to be infant feeding rather than breastfeeding centred and to draw on BCTs and an assets-based approach. The PPI groups were given cards with the various text messages and were asked to put them into one of three piles (yes, no or maybe). A group discussion was then facilitated by a researcher, and feedback on and suggestions about the various messages were noted. The PPI feedback was then used to finalise the library of suggested text messages available for use by the IFHs (see Appendix 5).

The ABA intervention outline is shown in Table 3. The intervention started at around 30 weeks’ gestation and could continue up until 5 months postnatally. At around 30 weeks’ gestation, the IFHs contacted women by telephone to arrange a face-to-face meeting, either at home (site A only) or at a suitable location, such as a café or children’s centre. The IFHs in site B (volunteer peer supporters) were not able to offer home visits because of local policies, unlike the paid peer supporters working in site A. Women were welcome to include partners or family members in this and subsequent meetings. The purpose of this face-to-face meeting was to talk about infant feeding and investigate the woman’s personal, family and social network assets for infant feeding. An approach of ‘narrative storytelling’ was used to produce a simple family tree diagram (‘genogram’) of experiences with infant feeding,55 incorporating the woman’s social network, to allow her to reflect on future feeding relationships and sources of support. 54 [See Appendix 6 for an example genogram from the training session (real names are not used).] At the antenatal visit, IFHs gave the woman the assets leaflets and explained the range of support available for infant feeding. In addition, contact details were swapped and a ‘Let us know when you’ve had your baby’ fridge magnet was given to the woman to encourage her to include the IFH on the list of people she would notify of the birth of the baby.

Following the face-to-face antenatal visit, the IFHs were asked to call and/or text the women fortnightly during the pregnancy. The aim was to encourage a strong rapport between the IFHs and the women that would facilitate successful immediate engagement after birth. In addition, IFHs were encouraged to facilitate a visit (antenatally) by the woman to a local breastfeeding group. The aim of this was so that the woman would know how and where to access support for infant feeding once her baby was born.

Postnatally, support from IFHs in both sites was by telephone calls or text messages every day until the baby was 2 weeks old. Additionally, at site A, IFHs were asked to arrange a postnatal home visit/Skype call as soon as possible after the mother and baby were discharged home.

From 2 to 8 weeks, frequency of contact was reduced based on the preferences of the mother. Text messages were sent to those still breastfeeding (or mixed feeding) at 3, 4 and 5 months. Women were able to ask for telephone calls or text messages to stop at any point. At site A, IFHs were unable to support women from 8 weeks postnatally because of their working practices. Therefore, in site A, a researcher sent out the 3-, 4- and 5-month texts and provided signposting support for women if required.

The intervention timeline is presented in Figure 1.

FIGURE 1.

The ABA intervention timeline.

Recruitment of ABA infant-feeding helpers

At site A, a paid peer support service providing reactive postnatal support (and no antenatal support) already existed. Although the service was available across the entire local authority area, peer supporters had traditionally worked within certain inner-city areas. Within the local authority, community midwives were split into three teams, each serving a distinct geographical area. For the ABA study, site A was selected as it was the area with the lowest rate of breastfeeding and highest rate of teenage pregnancy within the local authority. This area had traditionally not been served by the peer support service. The peer support service manager was willing to support the ABA study following discussions with the research team.

In site A, researchers attended part of the peer supporters’ regular team meeting on two occasions prior to the ABA training day. The purpose of these meetings was to introduce the ABA study, to encourage the peer supporters to take part in the study and to provide background information that would facilitate the smooth running of the ABA training day. Researchers also attended a team meeting after the training day to answer any questions arising from the session.

At site B, a volunteer peer support service was already in existence, which was overseen by a national charity. Again, this service provided reactive postnatal support, based around breastfeeding support groups, and no antenatal support. The geographical areas chosen within site B were those with the lowest breastfeeding rates within the local authority and with active peer supporters. Researchers met the charity manager and the peer supporter co-ordinator to discuss details of the study and recruitment of peer supporters.

Within the two sites, existing peer supporters were asked if they would like to support the ABA study. At site A, six of the seven existing peer supporters agreed to participate. At site B, seven of 11 peer supporters volunteered to be involved.

Training for ABA infant-feeding helpers

Infant feeding helpers were provided with 6 hours of training plus a study folder. The folder included details of all aspects covered in the training day. The intervention training was delivered face to face on 1 day in site A and over 2 half-days in site B. Heather Trickey led on the development of the training materials and led the training days, with input from Dr Kirsty Darwent (Programme Director, Family Therapy Training Network Ltd).

The aims of the training were (1) to promote competence and confidence in delivering the ABA intervention and (2) to facilitate understanding of the ABA study (to enhance fidelity to the intervention delivery). The training was designed to enable IFHs to learn about how to deliver the ABA intervention and to practise the skills required to deliver the intervention effectively. The training was interactive and involved simulations and role-play of contact with women as well as group-based learning activities.

Although the training did not explicitly present the BCTs to IFHs, it specifically focused on the delivery of the prespecified core BCTs (social support and restructuring the social environment) via the genogram and assets leaflet.

The training included the following.

1. Study information

   Kate Jolly gave an overview of the study. She explained that the aim of the ABA study was to compare two ways of delivering feeding help to first-time mothers in areas where breastfeeding rates were low. Half of the mothers recruited to the study would have the usual feeding support from midwives, health visitors and any voluntary agencies or peer support that they choose to access; the other half would, in addition, receive the new ABA intervention.

2. Overview of the intervention

   The ‘assets-based’ approach was outlined as an approach that encourages women to draw on support and help from their family, social and community networks. It was explained that the intervention was more intensive than usual peer support, and that is was peer supporter initiated rather than mother initiated. Every woman in the intervention would be offered antenatal contact, and continuity of care would be given wherever possible by women having the same IFH throughout the intervention. Trainees were informed that the intervention would end when the baby was 5 months old (if still being breastfed), or at a time when formula feeding had been established. Key principles of the ABA intervention were presented: the importance of a woman-centred approach and building a strong rapport, the use of open questions and active listening, seeing the woman (not the IFH) as the solution and viewing relationships as assets. The intervention timeline (see Figure 1) was presented to the IFHs, with opportunities given for discussion and clarification of any uncertainties.

3. Antenatal contact

   In this part of the training, expectations for the antenatal visit were detailed. The trainers simulated the visit to facilitate good practice. Role-play techniques were then used to allow the IFHs to try out the approach. This session included:

   • Introducing themselves as an ABA IFH and explaining the purpose of the antenatal visit.

   • Learning how to explain the ABA timeline so the woman knows what to expect.

   • Learning how to open a conversation on infant feeding, and being led by the woman. The importance of good listening skills and using open questions was emphasised.

   • Discussing support from family and friends, including completion of a simple genogram.

   • Discussing support available in the community, and introducing the ‘assets leaflet’.

   • Offering to accompany the woman on an antenatal visit to a local breastfeeding support group and encouraging the use of the group after the baby is born.

   • Swapping telephone numbers, encouraging the women to let the IFH know when they have given birth, and discussing plans for keeping in touch.

  The role plays were also used as an opportunity for IFHs to practise person-centred listening skills, with IFHs working in groups of three to give feedback to each other on the use of verbal and non-verbal active listening techniques.

1. Supporting mothers who use formula milk

   The importance of being inclusive of all feeding types was stressed. A group discussion on supporting mothers who formula feed was facilitated, with the aid of a ‘myths and truths about formula feeding’ exercise. Key information about different kinds of formula milk and preparation of feeds, and up-to-date advice on formula feeding in response to the baby’s cues was delivered to IFHs during the session and supplemented with a key messages information leaflet and links to further information.

2. Postnatal contact

   In this session, scenarios and group work were used to facilitate understanding. Groups worked together to decide how they would support women in the different scenarios. The ‘assets-based approach’ was underlined in the support provided, for example by encouraging women to use their personal and community-level assets for infant feeding.

Comparator group

Women allocated to the comparator (or ‘usual care’) group received usual care for infant feeding available in the study areas; this included routine support from midwives and health visitors. We describe the support for infant feeding that was available and accessed by women, which included local services, such as breastfeeding support groups and peer support, and national breastfeeding helplines. This usual care was available only reactively to women (i.e. the woman had to ask for support or the midwife asked for support on behalf of the woman).

In site A, peer supporters who did not volunteer to participate in the study were available to cover any requests for support received from the usual care women (as per usual care). In site B, breastfeeding support was available at any of the breastfeeding groups in the area.

Outcome assessment

All women were asked to notify the ABA study team about their baby’s birth by text message, e-mail or telephone call. To ensure that we found out about as many births as soon as possible, researchers from the ABA study team in site A also maintained daily telephone contact with the community midwives to find out if any of the women had given birth. On notification of an intervention participant giving birth, researchers contacted the IFHs to let them know. In site B, we relied on women notifying the research team or their IFH directly. Details of those who did not use either method in site B were obtained from midwives.

Feasibility outcomes

The feasibility of intervention delivery and the research methods were determined by:

• Reach of recruitment of women to reflect required sociodemographic profile.

• Ability to recruit, train and engage current peer supporters to the new ABA IFH role.

• Ability to deliver planned number of contacts at a time and location convenient to participants.

• Acceptability to women.

• Fidelity of delivery and whether or not woman-centred care was provided.

• Unintended consequences of the intervention.

• The feasibility of a future definitive trial assessed by recruitment rates, willingness to be randomised, follow-up rates at 3 days, 8 weeks and 6 months and level of completion of assessments by text94 (see Criteria for progression to main trial).

• Potential cases of intervention contamination in the usual care group; at 8 weeks’ follow-up, all women were asked if they had used national breastfeeding helplines or any breastfeeding support, whether or not there was a home visit or one-to-one meeting at a children’s centre and number of contacts by the IFHs. They were asked in interviews whether or not they had met other women taking part in the study and whether or not they had discussed the study.

Assessment of feasibility outcomes

A number of methods were used to assess feasibility outcomes. Table 4 summarises these.

Qualitative research

Semistructured interviews with women were carried out in the woman’s own home or in another convenient location. Sampling was purposive, aiming for a diverse range of experiences, and included teenagers, unemployed women (as indicated on the baseline questionnaire), women in socioeconomically disadvantaged areas, women with disparate feeding methods, women with different levels of contact with the IFH and women in the usual care group where intervention contamination was suspected (based on responses to the 8-week questionnaire).

With the exception of the first four, interviews took place after return of the 8-week questionnaire. After the first four interviews (all at site A), we decided to wait until after the 8-week questionnaire had been completed to avoid any possible interference with the primary outcome of a future definitive trial (any breastfeeding at 8 weeks). We aimed to interview around 15 women at each site (10 intervention, 5 usual care). Women were able to have a person of their choice present for the interview (research team experience has been that this can boost participation among socioeconomically disadvantaged groups). Interviews with women were conducted by Joanne L Clarke (site A) or Debbie Johnson (site B), both experienced qualitative researchers.

Interviews with women allocated to the intervention group explored the acceptability of the ABA intervention, as well as investigating the interaction of the intervention with other sources of support that are available, particularly with respect to existing community assets (e.g. breastfeeding support groups and baby groups). Interviews with women in the usual care group looked at their experiences of ‘usual care’ for infant feeding, and investigated possible cases of contamination. In addition, all women were asked about their experiences of being part of the ABA study, including the acceptability of the recruitment and randomisation process and follow-up methods.

We conducted FGs or interviews with all IFHs (n = 13), with the IFH manager at site B and with health-care providers (midwives and other providers of infant feeding support) working in the study areas (n = 17). FGs and interviews with IFHs investigated intervention acceptability, satisfaction with the ABA training, experiences of delivering the intervention and any barriers to or facilitators of intervention implementation. We also explored any additional training or supervision requirements.

Focus groups and interviews with health-care providers investigated how the ABA intervention fitted with existing support and whether or not the intervention had in any way changed ‘usual care’, as well as issues concerning referral or delivery. Possible cases of contamination were investigated with both IFHs and health-care providers.

Focus groups took place in a convenient location, and interviews with those unable to attend the FGs were conducted over the telephone. At the FGs with the IFHs, there was a lead facilitator (GT) who had no prior interactions with the IFHs, and at least one other member of the project team (JLC, DJ or JI) to record notes, interpersonal issues and ask follow-up questions, as appropriate. The FGs with health-care providers were conducted by Joanne L Clarke at site A and by Debbie Johnson and Jenny Ingram at site B. At the start of the FGs, all participants were asked to be mindful of confidentiality issues, whereby they should refrain from providing personal information about individual cases, and not to share what was discussed outside the FG. During the FGs, the lead facilitator encouraged all individuals to share their views, such as through seeking confirmatory or disconfirming views, and questions were directed to different individuals. At the end of the FGs, a summary of all key issues discussed was provided, with participants invited to offer any final comments.

Semistructured interview schedules were developed (see Appendix 8) based on research literature, team discussions, our logic model, PPI input and the ‘stages of breastfeeding peer support intervention design model’ constructed from a realist review of peer support intervention studies. 40

All interviews and FGs were audio-recorded. An external transcription company was employed to transcribe the recordings verbatim, including pauses and laughter, and anonymise them (by removing names of people and places). Transcriptions were checked for accuracy by researchers and reflective notes were made after every interview.

Qualitative analysis

For the qualitative analysis, we undertook thematic analysis using Braun and Clarke’s96 thematic approach supported by NVivo version 11 (QSR International, Warrington, UK). First, in line with the adopted approach, three researchers (JLC, DJ and GT) listened to the recordings and read and reread the transcripts of four participant interviews (one usual care and one intervention from each study site) before independently performing line-by-line inductive coding. Codes were discussed and developed into an initial and tentative coding framework of themes and subthemes. The remaining transcripts from all participant groups were then coded by Joanne L Clarke and Debbie Johnson using the coding framework that was iteratively refined (i.e. new codes added and/or refined), as appropriate. A number of external checks were undertaken to ensure that all data were represented within the coding framework. This involved the coding framework and NVivo files being reviewed by Gill Thomson, followed by discussions with Joanne L Clarke and Debbie Johnson, and amendments made as appropriate. The final coding framework was agreed by all researchers. In this report, we describe the qualitative results relevant to the feasibility outcomes. The full qualitative findings will be published in Health Expectations.

For each of the participant interviews, BCTs delivered by IFHs were coded as standalone themes. Coding of BCTs was based on reports of the behaviour of the IFH, regardless of the participant’s response. BCTs delivered by people other than the IFHs (e.g. midwives) were not coded for the purpose of this analysis.

Outcome measures for a future trial

The primary outcome for a future trial was any breastfeeding at 8 weeks.

Secondary outcomes for a future trial were:

• breastfeeding initiation (at 2–3 days, as defined by the UK Infant Feeding Survey 2010,18 even if on one occasion only; includes giving expressed breastmilk)

• exclusive breastfeeding at 6–8 weeks (exclusive breastfeeding defined in accordance with the WHO definition of infants who received only breastmilk during the previous 24 hours)97

• any/exclusive breastfeeding at 6 months

• duration of any and exclusive breastfeeding, if ceased breastfeeding

• maternal well-being [Warwick–Edinburgh Mental Well-being Scale (WEMWBS)]98

• maternal satisfaction with feeding experience and support provided at 8 weeks and 6 months (using a single-item question used in a previous trial37 and coproduced with PPI).

Outcomes for a future economic evaluation included in the feasibility trial were:

• self-reported use of health and feeding support services

• overall feeding support activity during the intervention

• use of child care.

Assessment of outcomes

Table 5 presents a summary of the data items collected. At baseline (around 28 weeks’ gestation), women were asked to complete a baseline questionnaire. This included questions on demographic characteristics, feeding intentions, how they were fed as a baby, whether or not they knew anyone who had breastfed a baby, maternal well-being (WEMWBS) and use of health services. A researcher was present during questionnaire completion to answer any queries or clarify any points on the questionnaire.

At 2–3 days postnatally, women were sent a text message by the study team asking them how they had fed their baby since the birth. They were asked to text back a response: 1 for formula milk, 2 for breastmilk or 3 for formula milk and breastmilk (see Appendix 9).

At 8 weeks and at 6 months postnatally, women were sent a brief questionnaire in the post (with a prepaid return envelope). These included questions on the delivery of their baby, the length of hospital stay, feeding methods, feeding support received and satisfaction with feeding support, maternal well-being (WEMWBS) and social support. 99 Women who did not return their questionnaire within 2 weeks received a text-message reminder followed by a telephone call giving them the option of completing the questionnaire over the telephone. If women were reluctant to complete a questionnaire, attempts were made to secure the primary outcome (feeding status at 8 weeks) over the telephone.

See Putz et al. 100 for copies of the questionnaires.

For women who did not return their 8-week questionnaire, and who could not be contacted by telephone to secure data on the primary outcome of a definitive trial (feeding status at 8 weeks), local health visiting teams were contacted to request this information.

Assessment of adverse events

Information on possible adverse events was collected at 8 weeks using an open question asking about any difficulties experienced in feeding their baby and any hospital admissions related to infant feeding for mother or baby. The research team contacted the woman for more information, as required. The chief investigator reviewed adverse events to define their severity and causality. Only serious adverse events that could be related to the intervention were to be reported to the Research Ethics Committee.

Sample size

The selected sample size (n = 100) allowed us to estimate feasibility outcomes with reasonable precision, allowing us to estimate recruitment, follow-up and questionnaire completion rates to within ± 15% with 95% confidence. This was based on a worst-case (in terms of precision) estimate of 50% for each outcome, the targets being 75% for recruitment, 75% for follow-up and 70% for questionnaire completion.

To enable us to calculate the necessary sample size for a future definitive trial, we determined the percentages of participants initiating breastfeeding and breastfeeding at 6–8 weeks for both groups; 95% CIs were given for the estimates acquired.

Statistical analysis

The study statistician was blinded to allocation group. We calculated recruitment and follow-up rates (with 95% binomial exact CIs) as a measure of trial feasibility.

To determine intervention implementation and contamination levels in the usual care group, we report the frequency and method of IFH and peer support contacts for both the intervention and the usual care groups.

Although the feasibility trial was not powered to detect a difference between the groups, we calculated the percentages of women breastfeeding and exclusively breastfeeding at 6–8 weeks by allocation, with 95% CIs presented. We also determined the dropout rate and the completeness of data, which will inform the sample size calculation and feasible outcome selection for a future definitive trial.

Women’s characteristics were described by allocation group, with simple summaries presented for each outcome measure. The primary analysis was by modified intention to treat, which included all randomly assigned patients for whom data on the primary end point were available.

The variability between IFHs in the primary outcome of a future trial was assessed by calculating the intracluster correlation coefficient (ICC) using a null linear model with a random effect for IFH. These data will inform the sample size calculation for a future definitive trial.

Feasibility of data collection for a future economic evaluation

This feasibility study explored whether or not it would be possible to collect the data required for a future economic evaluation. In a future definitive trial, we would want to be able to estimate the health service costs associated with the intervention (e.g. the costs of training the IFHs, telephone calls, text messaging service, one-to-one meetings with women, staff time to respond to women’s requests and any payments to IFHs).

In a future economic evaluation, we would calculate the extra cost for each additional case of breastfeeding in the intervention group compared with the usual care group. We may also consider how appropriate it is to link the increase in the uptake of breastfeeding to longer-term health benefits using a model-based economic evaluation.

Criteria for progression to a main trial

For the Phase III trial to be considered, the following criteria (predetermined by the TSC) needed to be met:

• The process evaluation suggests that the intervention is acceptable to a majority of mothers, their partners, ABA infant feeding team members and local services.

• At least 75 women need to be recruited in 5 months.

• The study must be able to recruit women of low socioeconomic status, teenagers and ethnic minorities.

• Intervention needs to be implemented with fidelity in 75% of mothers (this will be defined as contacts made in both the antenatal period and the postnatal period).

• 75% of participants should receive the assets-based antenatal face-to-face contact.

• > 70% follow-up should be achieved at 8 weeks and 6 months, with additional missing data able to be obtained from routine sources.

The criteria agreed by the ABA study TSC are presented in Table 6.

Feasibility of recruitment

In total, 135 women expecting their first child were approached at an antenatal clinic. Of these, 103 women (76.3%) agreed to participate in the study. Recruitment ran from 28 February 2017 until 23 May 2017 in site A and from 21 April 2017 until 31 August 2017 in site B.

Figure 2 shows the recruitment flow diagram for the ABA study. (See Appendix 10 for the recruitment flow diagrams for sites A and B.) The total number of births to primiparous women during the 3–4 month period was not routinely available. The researchers attended only some of the antenatal clinics during the period, so the 135 women who were approached were a sample of all of the eligible women.

FIGURE 2.

Flow diagram of participant flow through the ABA study. Reproduced from Clarke et al. 101 © 2019 The Authors. Maternal & Child Nutrition published by John Wiley & Sons, Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Reproduced from Clarke et al. 101 © 2019 The Authors. Maternal & Child Nutrition published by John Wiley & Sons, Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Reach of recruitment of women to reflect the required sociodemographic profile

The demographic and delivery characteristics of women who took part are presented in Table 7. (See Appendix 11 for additional details and characteristics by site.) The mean age of participants was 28.5 years. At both sites, the proportion of teenagers recruited was higher than the overall proportion for the area. Overall, nine of the study participants (8.7%) were teenagers, indicating successful reach of recruitment and acceptability of the recruitment process to this age group. The majority of women (n = 88, 86.3%) described their ethnicity as white British, a proportion that is reflected in local census data for the two sites. Most of the women were in paid employment (n = 90, 88.2%), and two (2.0%) were in full-time education or training. Nine participants (8.8%) were unemployed. Forty-six participants (45.1%) were educated to degree level or higher, 33 (32.4%) were educated to Advanced/Advanced Subsidiary (A/AS) level or equivalent, 22 (21.6%) were educated to General Certificate of Secondary Education (GCSE) or equivalent, and one participant (1%) had no formal qualifications. The majority of women were either married or in a civil partnership (n = 48, 47.5%) or living with their partner (n = 38, 37.6%). Fifteen participants (14.9%) were single. Overall, 38 participants (37.3%) resided in areas of the most deprived two Index of Multiple Deprivation quintiles, although this differed by site [33 participants (63.5%) in site A and five participants (10%) in site B]. The mean baseline score for mental well-being (WEMWBS) was 54.1 (the maximum possible score was 70).

The majority of women planned to feed their baby either breastmilk only (n = 35, 34.7%) or mainly breastmilk (n = 30, 29.7%). Twenty-two participants (21.8%) intended to feed their baby half breastmilk and half formula milk. Nine participants (8.9%) intended to feed their baby only formula milk, and a further five (5.0%) intended to feed mainly formula milk. More participants had been breastfed entirely as a baby (n = 36, 35.3%) than formula fed entirely (n = 29, 28.4%). Most of the women (n = 93, 91.2%) knew someone who had breastfed her baby.

Mean gestational age at birth was 39.5 weeks. Thirty-seven babies (44.1%) were delivered by normal vaginal birth, and 23 (27.4%) were delivered by forceps, ventouse or vacuum delivery. There were 18 births (21.4%) by emergency caesarean section and six (7.1%) by planned caesarean section. The proportion of women reporting staying in hospital for more than 48 hours after the birth was high (41.7%).

Baseline imbalances

A visual inspection of the baseline participant demographic and delivery characteristics (see Table 7) revealed some imbalances. Those in the intervention group were more likely to be unemployed, less likely to be educated to degree level and more likely to be single. In addition, there were more premature deliveries and more admissions to the neonatal unit in the intervention group than in the usual care group.

Birth notifications

The median age of babies at the time the study team found out about the birth was 3 days (interquartile range 0–30 days) (Table 8). Overall, 50.5% of births were reported to the study team when the baby was ≤ 3 days old. The source of birth notification to the study team was the community midwife in the majority of cases (57.4%), followed by the participant (35.6%) and the IFH (6.9%). However, the proportions for this differed by site. At site A, the primary source of birth notifications was the community midwife (78.4%), followed by the participant (17.1%) and the IFH (3.9%). At site B, 54.0% of birth notifications were from the participant, 36.0% were from the community midwives and 10.0% were from the IFHs.

Those women who notified the study or the IFH about the birth were more likely to be breastfeeding at 8 weeks and less likely to have had a long hospital stay (Table 9).

Feasibility of postnatal text at 3 days

The research team sent postnatal texts (asking women how they were feeding their baby) to 81.6% of women before their baby was 10 days old (Table 10). Postnatal text responses were received from 68% of participants overall.

Follow-up rates

Follow-up data were obtained from 68% of women at 3 days postnatally. The 8-week questionnaire was completed by 85.4% of participants (including three women who provided primary outcome data only over the telephone). Primary outcome data were available from 95.1% of participants (local health visiting teams provided 6- to 8-week feeding data for 10 participants who had not returned the 8-week questionnaire). At 6-month follow-up, 80.6% of participants returned a completed questionnaire. Follow-up rates were higher at all three time points at site B than at site A (Table 11).

See Appendix 12 for tables showing data completeness.

Table 12 shows follow-up rates by site and group. Follow-up rates were consistently higher in the usual care group than in the intervention group.

Study withdrawals

In site A, there were three withdrawals from the study. One participant withdrew immediately after recruitment (no reason given; she requested that her data be destroyed), and another withdrew between the 8-week and 6-month follow-up points (she no longer wanted to be part of the study). One participant was withdrawn by the study team after midwives reported that she had experienced a stillbirth. There were no study withdrawals at site B.

Characteristics of participants who were followed up compared with characteristics of those who were lost to follow-up or withdrew

Demographic characteristics of responders (i.e. those who responded to the postnatal text and the 8-week questionnaire) were compared with those of non-responders (Table 13). At both time points, when compared with responders, non-responders were younger, more likely to be of white British ethnicity, less likely to be in paid work, less likely to be educated to degree level or higher, less likely to be married or in a civil partnership or living with their partner, less likely to be intending to breastfeed at baseline and less likely to be breastfeeding at 8 weeks.

Qualitative study participants

Thirty of the study participants were purposively sampled to take part in a qualitative interview, 15 from each site. Twenty-one interviewees were from the intervention group and nine were from the usual care group. The sample included three teenagers (site A) and two from the 35–39 years age group (site A), with the remaining spread between 20–34 years (25–29 years, n = 10; 30–34 years, n = 11). The majority were white British (n = 26, 86.6%) and in paid work (n = 25, 83.3%); all those who were unemployed (n = 4) or students (n = 1) came from site A. Over half the sample were still breastfeeding at 8 weeks (n = 17, 56.6%); the majority of these (11 out of 17, 64.7%) were from site B. One woman at site A was interviewed together with her partner.

All IFHs took part in either a FG (n = 9) or a one-to-one interview (n = 4). At site B, the IFH manager also participated in the FG. Seventeen maternity service providers (including midwives and infant-feeding staff) also participated in either a FG (n = 14) or an interview (n = 3).

Quotations in this report are labelled by type of interviewee [P (woman participant), IFH or maternity services], by number (for women participants and IFHs), by intervention or usual care (for women participants), by site (site A or site B) and by whether the data were collected during an interview or a FG (for IFHs and health-care providers). For example, ‘P1 – intervention, site A’ would be woman participant number 1 from the intervention group in site A.

Women’s and maternity services providers’ views on recruitment and randomisation process

Feasibility and process outcomes for the intervention

Ability to deliver planned number of contacts at a time and location convenient for participants

Figure 3 shows the flow of participants through the intervention, and Table 14 reports IFHs’ ability to deliver the planned number of contacts.

FIGURE 3.

Participant engagement flow diagrams: (a) site A and (b) site B.

TABLE 14 - Ability to deliver planned number of contacts

Two ‘visits’ completed over the telephone at the participants’ request.

One participant not contacted because of stillbirth (withdrawn from study by research team).

IFH log unavailable for one participant; data not included for stillbirth.

Contact	Site A	Site B	Overall
Antenatal contact attempted, n/N (%)	25/25 (100)	25/25 (100)	50/50 (100)
Antenatal visit completed, n/N (%)	17/25a (68)	22/25 (88)	39/50 (78)
Postnatal contact attempted, n/N (%)	24/25b (96)	22/25 (88)	46/50 (92)
Postnatal support provided, n/N (%)	18/25 (72)	22/25 (88)	40/50 (80)
Contact attempted by IFH within 48 hours of birth, n/N (%)	6/25 (24)	18/25 (72)	24/50 (48)
Number of days contact made/attempted by IFH in 2 weeks postnatally, mean (N)	2.5 (23)c	8.6 (25)	5.7 (48)
Number of days two-way contact established in 2 weeks postnatally, mean (N)	1.2 (23)c	6.9 (25)	4.1 (48)

At site A, 17 of the 25 women randomised to the intervention group received an antenatal visit. Of these visits, 14 took place at the woman’s home, two took place in a coffee shop and two were conducted over the telephone at the woman’s request.

All three teenagers at site A allocated to the intervention group received an antenatal visit. Of the eight women who did not receive an antenatal visit, four gave birth prematurely before contact could be established, three could not be contacted by the IFHs and did not respond to messages, and one arranged to receive a home visit and then refused entry to the IFH. One woman who did not receive an antenatal visit at site A intended to feed her baby formula milk only, and the others wanted to breastfeed at least partially. Eighteen women at site A received some postnatal support from their IFH (counted as at least one two-way communication via text, telephone or face-to-face contact). Five women could not be contacted postnatally, and one woman was not contacted because of safety concerns. One woman experienced a stillbirth and was withdrawn from the study by the research team.

Beyond 2 weeks postnatally, 14 women at site A received some support from their IFH. Six women received text support at 3–5 months.

At site B, 22 of the 25 women randomised to the intervention group received an antenatal visit. Of these visits, 10 took place in a children’s centre, nine were in a coffee shop and three were at the woman’s home. Of the three women who did not receive a visit, one woman intended to feed her baby formula milk only and the other two wanted to breastfeed at least partially. Two women withdrew from the intervention (but not from the study) without meeting their IFH antenatally. Despite having telephone conversations with her IFH antenatally, the third woman did not meet antenatally with her IFH, and had one postnatal conversation in which she said that she had opted to formula feed her baby. The IFH continued to offer ongoing support postnatally with no response, so contact was discontinued. This participant is counted as receiving the offer of some postnatal support. One woman met with her IFH antenatally, but sent a text after her baby was born to say she had initiated formula feeding and no IFH help was required.

Twenty-two women at site B received some postnatal support in the first 2 weeks. IFHs contacted 17 women on the day of the birth or the following day. IFHs continued to contact 15 women beyond 2 weeks postnatally. At least 10 of these women were attending breastfeeding groups run by their IFHs. Many of these postnatal contacts took the form of text conversations of varying length; in some up to 20 texts were exchanged. Most of these conversations were initiated by the IFHs; there were very few occasions when women made unsolicited contact.

Views and experiences of organising antenatal contacts

The IFHs in site A were used to visiting women in their own homes, so they generally tried to arrange the antenatal meeting at this location. However, as they were working in an unfamiliar area this created travel difficulties, with comments such as it being ‘too far’ and ‘a nightmare’ to get to, and ‘you spend half your time travelling’. The geographical distance was also reported by one woman and her partner to have had an impact on how comfortable they felt about asking for visits postnatally, leading to conflicting feelings about the ABA intervention:

It’s meant to be like a service that you can call out and that you can have support, and if you think oh no I won’t call her because she’s going to need an hour on the bus then that’s really not fair on her, it’s not that bad and we’ll wait . . . it didn’t directly put us off, but you could see that it would.

Partner of P2 – intervention, site A

In site B, travel was not an issue as the IFHs were all seeing women from their local area. These IFHs did not normally visit women in their homes and were encouraged to hold the ABA antenatal meeting at their local children’s centres whenever possible. This had the advantage of familiarising women with the location of the breastfeeding group(s) and other services offered in the children’s centres. Several women commented on the convenience of this location:

Yeah, she dropped me a text and then we met up . . . at the [place] children’s centre. But we met up in another room down there and had a coffee and a chat and that was really good.

P20 – intervention, site B

Occasionally, antenatal meetings took place in coffee shops as these offered a mutually convenient location (site A) or time, for example at the weekend (site B). Being able to sit down and have a drink together helped create a relaxed atmosphere in a familiar and comfortable environment:

All of my meetings actually on the weekend, so we did have to meet in a café which that wasn’t a children’s centre or anywhere . . . I had four ladies altogether and I think three of them were still at work when we did the meeting, they hadn’t yet gone on maternity leave, so actually that suited them fine to do it on the weekend.

IFH8 – site B, interview

The organisational differences between IFHs at both sites meant that they had different approaches to arranging antenatal meetings with women. Site B IFHs worked locally to their own homes and were able to arrange visits flexibly to suit both themselves and the women (e.g. evenings or weekends when child care was available at home). Site A IFHs were restricted to providing support during working hours. Although this could create difficulties in co-ordinating meetings for mothers who were working full time, one woman was willing to take annual leave to meet her IFH:

She called me to say ‘I’m going to be part of the study, can I arrange to meet you?’, I was still working at the time so I had to arrange the day off.

P1 – site A, intervention

Making these arrangements was achieved through discussion between IFHs and women, with agreement on a time and place to suit both parties. However, the meeting often took place later than at the intended time of 32 weeks of pregnancy. This was sometimes because women were still at work, or had not been recruited until this stage of pregnancy:

We didn’t get the details in time for the 32 weeks, so I think a lot of ours were after that weren’t they from the [place] group? We didn’t know if that mattered or not in terms of the research timeline.

IFH11 – site B, FG

Despite numerous attempts, a few women were uncontactable during the antenatal period, causing frustration for IFHs:

It was sad that the women didn’t actually respond back, so it was very difficult to get hold of them, especially with the making the antenatal appointments and getting to know them as a person before they actually have the baby, that was quite difficult, they didn’t really engage.

IFH4 – site A, FG

Infant-feeding helpers’ and women’s experiences of maintaining contact throughout the intervention

Maintaining contact, and women’s willingness to engage in ongoing communication, was unproblematic for some IFHs who negotiated the frequency and mode (i.e. text, telephone) of contact with each woman. This could involve reducing contact as a result of knowing, for example, ‘she is doing really well’ (at breastfeeding) or providing additional help as it was felt to be ‘the best thing to do’:

My first lady, the one that went into hospital . . . we did text for quite a long time not necessarily about baby stuff but about her being ill and that, probably longer than the 2 or 3 weeks just because I thought we were getting on quite well . . . I probably messaged her more than I was supposed to but I just felt like that was the best thing to do really, I didn’t want to just abandon her when she was mid treatment, so I followed it through.

IFH8 – site B, interview

The IFHs also encouraged women to contact them and seek out help as needed:

I made it clear that they could text me whenever they wanted and I would get back to them as soon as I could.

IFH7 – site B, FG

Positive comments were made about having the schedule of planned contacts as, for example, ‘you could see what you had to do’. The smaller caseloads at site B meant that providing the agreed number of contacts was manageable, although one IFH reported difficulties associated with managing a work–life balance:

When you’ve got your own children it’s trying to fit it all in, and I think there might have been a few times where I missed by a few days.

IFH12 – site B, interview

The IFHs in site A found fitting ABA postnatal contacts around their busy working schedules more difficult. Home visits, which they would normally offer, were often not possible, and calls and texts might have to be done at weekends instead:

Yeah, it was time, so if you couldn’t manage to always get the ABA mums in the week that you had that weekend time to just focus on them.

IFH4 – site A, FG

Some IFHs in both sites expressed concerns that the frequency of proactive contacts could be construed as ‘hassling’, particularly when there was a lack of response, and they could be unsure how to proceed:

Yeah, a couple of times I did feel I was [hassling] . . . do I keep texting her or am I bothering her?

IFH8 – site B, FG

One IFH reported that although one mother had been engaged and responsive during the antenatal period, this had tapered off to ‘one-word answers’ in the postnatal period, leaving the helper with the feeling that the woman was ‘getting annoyed with me’. There were also challenges when women replied saying that they would recontact the IFH at a convenient time and then no further contact was received. A lack of response could result in the IFH reducing the number of contacts to give the mother ‘a bit of space’, suggestive of a sensitive, woman-based approach.

On the other hand, mothers seemed grateful for a proactive approach to contact, finding it reassuring that help was there if they needed it. Failure to respond to messages on their part was often as a result of the demands of caring for their new baby and not needing to ask for advice rather than because they did not want to be contacted. One woman articulated that receiving texts gave her ‘permission’ to continue seeking advice for longer than if she had had to instigate the contact:

If they hadn’t offered their help I’m not sure how good I would have been about asking for help . . . I suppose I kept feeling like I should be beyond the stage of needing their help . . . but with them asking how I was it gave me permission.

P4 – intervention, site A

The IFHs used various methods (telephone, text, e-mail) to maintain contact with women, with texts proving women’s most preferred and the most effective method of contact. This was mainly because it was ‘easy’ and women could respond in their own time and ‘have time to process it’:

I preferred that. I didn’t really have much energy to form proper sentences at that point . . . so texting was much better.

P27 – intervention, site B

The value of having met their IFH antenatally was also apparent in enabling women to feel confident about the postnatal contacts:

Once I had met her and I can put a face to the name, just gives you that reassurance again really that there’s somebody there . . . so then didn’t have a problem . . . if I need to text her then I would.

P22 – intervention, site B

One IFH recommended that further guidance should be given about frequency of contact and how to manage when women do not reply.

Fidelity of delivery and whether or not woman-centred care was provided

Fidelity of delivery: analysis of behaviour change techniques discussed in qualitative interviews

All qualitative interviews with intervention participants (n = 21) were checked for evidence of BCTs. A summary of the BCTs reported in qualitative interviews with women is presented in Table 15. The core BCTs were those most frequently used, with 20 out of 21 participants demonstrating evidence of ‘restructuring the social environment’ (BCT 12.2); many of these were related to the use of the genogram in the antenatal meeting:

They did the . . . diagram of support to show you and who you’ve got, so your parents, people that are around you, friends, and then the groups you can go to, and the stuff like that.

P2 – intervention, site A

It helped me think about the pathways in my head a little bit better in terms of ‘Oh yeah I forgot about that person’. Just by going through the process of drawing it . . . that process of seeing it laid out, my network of support as it were.

P20 – intervention, site B

Some talked about the IFH inviting them to breastfeeding groups:

Yeah, different groups, yeah. She said if I wanted to she would meet me at them and to come with me, yeah she went through all the different groups and stuff.

P21 – intervention, site B

Evidence of core BCT social support (unspecified) (BCT 3.1) was also widespread, with 18 participants demonstrating that they felt reassured by knowing where to go for appropriate advice and support:

There was always the opportunity for somebody to come and visit me if I wanted them to, or a phone or a text whenever I needed or wanted.

P6 – intervention, site A

I had someone to ask what this cluster feeding was and had the reassurance that you don’t run out of milk.

P19 – intervention, site B

Social support was also demonstrated by women reporting positive feedback and encouragement from their IFH:

I would say fed him X amount of times and it’s tiring, I was so tired or something like that, and [she] would be going you’re doing a great job, and keep going, just support in just telling me I’m doing really well and keep going.

P22 – intervention, site B

The other non-core social support-related BCTs were also well represented, with 12 women demonstrating social support (practical) (BCT 3.2) and six having examples of social support (emotional) (BCT 3.3). Most examples of practical social support (BCT 3.2) concerned receiving appropriate advice from IFHs during telephone calls or in text messages, which women found particularly helpful for specific problems such as mastitis or the frequency of breastfeeding. Examples of emotional social support demonstrated the value of the IFHs, with women saying that they found them ‘reassuring’, ‘kind’ and ‘supportive’:

They helped me a lot, they give me a lot of positive energy really. Whenever she come she makes me really happy, she makes me really good.

P7 – intervention, site A

Seven participants had examples of verbal persuasion about capability (BCT 15.1) and eight had examples of problem-solving (BCT 1.2).

Acceptability of the ABA intervention to women, infant-feeding helpers and maternity services

Potential cases of intervention contamination in the usual care group

Overall, there was no clear evidence of contamination in the study. Although there were some indications of women sharing resources and learning with others (either those in the usual care group or other expectant/postnatal women), this is inevitable and to be expected within a community of mothers. However, as the key components of the ABA intervention were an antenatal contact and targeted, intensive one-to-one tailored support in the early postnatal period, these were non-transferable elements:

When I was pregnant as well I mentioned it [ABA study] to a few people, and . . . they just want to know what I’m doing sort of thing, yeah, and offer some support, and a lot of people said that’s really well needed after pregnancy . . . and I thought it was a really good idea actually.

P6 – intervention, site A

A few mothers (and, on one occasion, the woman’s partner) referred to how they had passed on information (e.g. about the breastfeeding groups), shared the assets leaflet and discussed the value of the genogram, but with no clear evidence of take-up. Others referred to how discussions with friends who were either pregnant or new mothers highlighted the value of and need for this supplementary support, rather than sharing specific details that would facilitate breastfeeding:

I know I suggested it [breastfeeding group(s)] to quite a few of the girls in the NCT group because I knew about it and they didn’t know about it, especially one that was having quite a few troubles.

P16 – intervention, site B

On discharge from hospital, women at site A received information on local breastfeeding groups in a leaflet included in the discharge packs. Women at site B were signposted to the local council website to find out about local breastfeeding groups, and as a result some of the IFHs at site B considered that mothers ‘may well have come anyway’. However, this was not supported by women at site B, many of whom said that they would not have known about the group had it not been for their IFH telling them about it.

Some women referred to ‘mentioning’ or discussing their participation in the ABA study with health professionals who either were aware or were not aware of their involvement: ‘I’ve told them about it and they were very interested, but they didn’t seem to know about it before’. During interviews, health professionals gave mixed responses, either being unaware that women were involved in the study or being vague about women’s involvement.

Outcomes for a definitive trial

Outcome data for any breastfeeding at 8 weeks and 6 months are presented in Table 16. For data by site, see Appendix 13.

Achievement of progression criteria

Our achievement of the progression criteria approved by the TSC is shown in Table 22. The criteria have all been met, although, with the small sample size, the lower boundary of the 95% CIs for the proportion of teenagers, the fidelity of intervention delivery and the receipt of assets-based face-to-face contact do exclude the percentage for progression for these criteria.

Summary of findings of the feasibility study

The main aim of the ABA study was to assess the feasibility of delivering the ABA intervention within a definitive RCT. To achieve this aim, we assessed the feasibility of recruiting peer supporters into the new ABA IFH role, the fidelity of the intervention and training needs of the peer supporters, the acceptability of the intervention to women and peer supporters, women’s engagement with the ABA service, recruitment and retention into the study and the collection of outcome measures. In addition, we explored two different models of ABA delivery: by employed, paid peer supporters and by volunteer peer supporters.

We found that it was feasible to recruit peer supporters, to train them to take on the ABA role and for them to deliver the intervention with adequate fidelity. We were able to recruit women to the study in the planned time frame and the women reported the ABA support to be acceptable. The study demonstrates that the processes for a future definitive RCT were feasible. These findings are discussed in more detail below.

Interpretation of findings of the feasibility study

Potential improvements for future intervention delivery

In terms of future intervention delivery, we have identified the need to use a locality with stable commissioning of an IFH service (volunteer or paid) and with good management support for the IFHs to facilitate the change in mode of working (antenatal contact, one-to-one support for volunteers and supporting women who might formula feed). The IFHs need to be embedded in their communities so that their travel is reduced and they have an understanding and experience of the local ‘assets’.

A big challenge at both sites was in receiving notification of the birth. Ideally, the midwife who delivered the baby would notify the IFHs and/or the research team; however, it was clear from our discussions with the community midwives that this was very unlikely to happen. To try to ensure that the IFHs were told about the birth, a focus of the antenatal contacts was the IFH developing a relationship with the pregnant woman so that she would notify the IFH about the birth, along with family and friends. We had fridge magnets made to remind the women to call/text the IFH and encouraged the IFHs and women to swap telephone numbers to ensure that barriers to letting the IFH know were few. Other trials of peer support in the UK have struggled to achieve early contact after birth. 104,105

Comparison of findings with those from other research

The ABA feasibility study was not powered to determine the difference in breastfeeding between the study groups, so conclusions about the effectiveness of the intervention cannot be drawn between ABA and other trials of breastfeeding peer support in the UK. Unlike previous RCTs of breastfeeding peer support in the UK,104–107 we offered a more intensive intervention, which included antenatal contact but had a particular focus on the key period of the 2 weeks after birth, when rates of breastfeeding fall dramatically. Other studies that aimed to support breastfeeding included daily calls in the first 2 weeks37,108 and showed promise in increasing breastfeeding rates.

A recent uncontrolled feasibility trial of breastfeeding peer support, the Mam-Kind study,31 reported that delivery of that intervention, which included motivational interviewing by paid peer supporters, was feasible and acceptable. A particular strength of the Mam-Kind report was its detailed process evaluation, which enables us to make comparisons with the ABA study’s intervention design and findings. In Mam-Kind, the invitation to the study was made by the community midwife but, unlike in ABA, the midwife passed on the woman’s details to the research team for recruitment at a later date. In ABA, we were more successful with recruitment, and recruited a higher proportion of women approached (76% vs. 24% in Mam-Kind). We recruited to target, unlike Mam-Kind, which recruited only 78% of planned participants, and we obtained a broader reach, with higher proportions of teenagers, women with lower educational attainment and women from minority ethnic groups. Our inclusion criteria were wider, with women pregnant for the first time being eligible for ABA regardless of feeding intention, whereas the Mam-Kind study recruited only women who planned to breastfeed and included women expecting their first and subsequent babies. Mam-Kind’s requirement for women to be intending to breastfeed may have excluded teenagers and women in more socioeconomically disadvantaged circumstances.

The Mam-Kind study31 had a midwife who supervised the peer supporter teams and worked to encourage the hospital midwives to notify the peer supporters about the birth. As a consequence, Mam-Kind contacted 73% of women within 48 hours of birth. In ABA, contact was attempted with 48% of women within 48 hours of birth, but with wide variation between the sites (24% and 72%, respectively). Mam-Kind recruited and employed peer supporters through the university, in contrast to our study, in which we worked within existing peer support services. All the Mam-Kind peer supporters were paid, and they experienced similar challenges to those in site A, where an existing, paid peer support service took on the trial intervention participants in addition to its usual workload.

Our early follow-up rates were similar to that of Mam-Kind. 31 We had data on feeding method within 14 days from 68% of women and Mam-Kind achieved 63% follow-up at 10 days using telephone follow-up. At 8 weeks, ABA obtained 85% follow-up and data on feeding status on an additional 10% of women from health visitor records. Rates were lower in Mam-Kind, suggesting that offering a shopping voucher at follow-up and using paper-based questionnaires may have been successful strategies.

Both ABA and Mam-Kind31 achieved reasonable intervention fidelity, while identifying aspects for improvement and additional training. Mam-Kind used a motivational interviewing approach as a key component,31 whereas in ABA we focused on an assets-based approach with behaviour change theory. Mam-Kind reported that the peer supporters found it challenging to move from information giving to a collaborative approach, something also identified by our IFHs who worked in the paid peer support service (site A).

The Mam-Kind intervention31 ceased at 14 days, with the peer supporters asked to facilitate the transition to a breastfeeding group or another community support group. Some women reported that the exit felt rather abrupt, which validates the ABA approach of withdrawing support much more gradually and maintaining contact up to 5 months postnatally for women still breastfeeding to encourage maintenance of breastfeeding.

Strengths and limitations

This feasibility study used robust methods, including a detailed process evaluation and a usual care group, unlike the Mam-Kind study. 31 The usual care comparator enabled us to characterise the breastfeeding support received in usual care and showed a low uptake of peer support. The randomised feasibility design enabled us to assess the feasibility of recruiting to and delivering the ABA study, as well as investigate the intervention’s acceptability to women, IFHs and maternity services. Furthermore, the use of two study sites that had existing peer support services enabled the comparison of processes between paid and volunteer services and an exploration of how the ABA intervention could fit into an existing service. The recruitment of trained peer supporters enabled us to reduce the duration of training required to become an IFH. Although we had sufficient interest from peer supporters in taking on the IFH role, we did not have enough interest to enable us to select peer supporters for the role based on woman-centredness or listening skills. To ensure transparency and replicability, we have reported the intervention using all the items in the TIDieR checklist. 109 The organisational disruptions and threats to the peer support service at site A, although challenging, provided an important insight into the risks associated with changing public health budgets and the short-term commissioning of services. We have used statistically robust methods; estimates of the primary outcome in the intervention and control groups have been presented with 95% binomial exact CIs to demonstrate the uncertainty of results. The ICC is also estimated with a 95% CI to indicate the variability between IFHs, which will occur only in the intervention group.

Limitations of the study include the small number of recordings of antenatal contacts at site A. The IFHs were reluctant to be recorded and also cited concerns about making recordings in noisy public places and with women living in particularly challenging social circumstances. This means that the fidelity assessment of the antenatal intervention can really be applied only to site B.

Another limitation was that our qualitative interviews were with women who had provided follow-up at 8 weeks; we did not interview women who were lost to follow-up. The sociodemographic characteristics of women who were not followed up are those associated with a lower likelihood of breastfeeding and it is likely that the responses of the women who were interviewed will have had a positive bias.

The participant recruitment process was resource intensive as it required a researcher to sit in community midwives’ antenatal clinics awaiting eligible women; nevertheless, this approach is likely to have contributed considerably to the successful recruitment. To recruit the required 103 participants, researchers spent approximately 130 hours in clinics. Despite this time commitment, we were successful in recruiting women living in areas of socioeconomic disadvantage, pregnant teenagers and women intending to formula feed. Feedback from women in the interviews indicated that the face-to-face explanation of the study was helpful in encouraging them to participate. If the ABA intervention were rolled out to usual care, we would envisage community midwives signposting women to the service during a routine appointment.

Encouraging timely notification of birth was problematic, with only half of women notifying the study team within 3 days of the birth. This resulted in delays in the collection of postnatal feeding status data and, in the case of the intervention participants, delays in the offer of postnatal support. We encouraged women to let us know of the birth by giving them a ‘let us know when you’ve had your baby’ fridge magnet recruitment and sending a reminder text at 38 weeks’ gestation. Intervention women were also encouraged to let their IFH know of their baby’s arrival as soon as possible after birth (with processes in place for the IFH to then let the study team know). In cases in which the participant did not inform the study of the birth, we relied on collecting this information from community midwifery teams. However, as community midwives receive information about only those women who have been discharged from hospital, this led to delays in receiving information about women who had a longer hospital stay. Women who notified the study or their IFH of the birth were more likely to be breastfeeding at 8 weeks than those women for whom we relied on community midwife contact, indicating some possible early disengagement of mothers who were formula feeding. From qualitative interviews and PPI discussions, women indicated that more text reminders before and around the due date would be acceptable. Going forward, processes for increasing the rate of early birth notification would need to be put in place, as would protocols for notifications of preterm births or perinatal death.

From the process data, it is clear that the ABA intervention was less successful in accommodating women who delivered prematurely, with only three in seven of these women receiving any support from the IFHs either antenatally or postnatally. Going forward, methods to include women who deliver prematurely must be considered. In the case of women who miss the antenatal assets-based component because of preterm birth (or other reasons), the process needs to ensure that the assets-based component is offered in the postnatal period. This needs to be addressed in the training of the IFHs.

Although we received follow-up questionnaires from 85.5% and 80.6% of participants at 8 weeks and 6 months, respectively, there were differences between those who were followed up and those who were lost to follow-up. In particular, non-responders were less likely to intend to breastfeed at baseline, and less likely to be breastfeeding at 8 weeks (from health visitor data), indicating possible further disengagement of formula-feeding mothers, who may have perceived the intervention to be about breastfeeding support, despite intentions for it to apply to all women.

In the future, the use of electronic questionnaires may help to boost response rates. Reliance on postal questionnaires was problematic, with a small number of women reporting that they had returned questionnaires that never arrived at the study offices. In addition, waiting for questionnaires to arrive back in the post led to delays in sending out reminders to women. On the whole, PPI input and findings from the qualitative interviews indicate that electronic questionnaires should be offered going forward.

Communication difficulties between IFHs and women sometimes caused issues. In one example, explored in a qualitative interview, a woman texted her IFH for breastfeeding advice, but phone records showed that the text message did not get through. By the time contact was re-established, the woman had moved on to formula feeding. In future, IFHs need to be aware of these possible issues; backup telephone calls should be considered when a woman does not respond to text messages. To mitigate technology failure, the FEST study37 used a combination of text and telephone.

In some cases, IFHs failed to establish any contact with women despite numerous attempts. More in-depth explanation of the intervention at randomisation may assist in increasing contact in future. It could be that, despite the woman-centred approach to recruitment and explanation that the intervention is inclusive of all feeding types, women were put off by the initial approach from IFHs. In a future trial, the content of initial text or telephone messages should be more fully considered to be the crucial first contact with women.

We have presented data by site to capture the differences in feasibility and in outcomes. Given that sites differed in more than just type of feeding helper (voluntary or paid), caution must be taken not to attribute differences to this factor alone.

Economic evaluation methods

Any future definitive trial of the ABA intervention would include a cost-effectiveness work package to assess the costs and the relative cost-effectiveness of the ABA intervention compared with the status quo from an NHS/Personal Social Services perspective. The ABA feasibility study included an objective to measure the features of the ABA infant feeding team provision and service utilisation to inform the design of a future cost-effectiveness study and specifically determine the feasibility of data collection.

We collected self-reported use of health and feeding support services using the 8-week questionnaire given to women. Although this had high rates of completion, the questionnaire would require further refinements to ensure accurate data capture. For example, we asked women about contact with peer supporters and attendance at breastfeeding support groups; in some cases, women commented on the questionnaire that they had met a breastfeeding peer supporter at a feeding support group and that there was overlap.

Overall IFH activity for each woman supported was obtained from logs kept by the IFHs and from their mobile phone records. A considerable amount of activity went into trying to contact women, particularly at site A, and this would need to be recorded and differentiated from research costs in a definitive trial.

The research proposal’s peer reviewers requested data collection on the use of child care and the uptake of benefits. It was clearly acceptable to women to ask about their use of benefits and we had good completion of the questions, although we are unable to confirm the accuracy of the responses. The interpretation of these data, however, is challenging. In particular, the relevance of the uptake of benefits and demand for child care and how that affects infant feeding decisions was not clear. We would not recommend collecting these data in a future trial.

Another challenge for a future trial that we identified is how to cost a volunteer IFH/peer supporter. Conventionally, health economic analyses cost volunteers at the national minimum wage, as there is an opportunity cost of their time. However, many peer supporters volunteer for a relatively short time prior to returning to work or having another baby, often citing that they want to give something back after receiving support themselves. 110 Additionally, the peer support role enables them to take their baby with them, which may not be possible in many other volunteer roles. A future study should explore with peer supporters whether or not they would volunteer in a different capacity if the breastfeeding peer supporter role was not available. A future cost-effectiveness analysis should undertake sensitivity analyses of a range of costs of the peer supporters’ time.

Training may have to be provided on an annual basis to ensure that, with high turnover of peer supporters, sufficient numbers are trained to deliver the intervention and this will be an important component of the overall costs.

In the Mam-Kind study31 the peer supporters were paid, as in one of our sites, and the overall cost of supporting a woman was estimated at £480, but this would have been as low as £350 had the study recruited women at the planned rate.

From the qualitative interviews, we did not identify additional costs to family or social networks in supporting a mother in her breastfeeding; therefore, the need to undertake a societal perspective in a future economic evaluation is not supported.

Patient and public involvement in the study

In this section, we describe the PPI in the ABA study, using the GRIPP2 short form111 as a guide to content.

Recommendations for future research

We consider that the intervention was feasible to deliver with acceptable fidelity and it was acceptable to the women in the intervention group, to the IFHs who delivered it and to maternity services. This was an individually randomised trial in two sites, and we showed that it was feasible to recruit women from more disadvantaged socioeconomic areas, teenagers and women who planned to formula feed. These women were willing to be randomised and the follow-up rates were acceptable.

On the basis of the findings from this feasibility study, the ABA study’s independent TSC agreed that the criteria for progression to a definitive trial have been met. Our main recommendation for future research is that there is a need for a future definitive trial to evaluate the effectiveness and cost-effectiveness of the ABA intervention. Below we discuss the factors that need to be modified from the feasibility trial and other issues for consideration.

Conclusions

This feasibility RCT has demonstrated that the ABA intervention is essentially acceptable to women and peer supporters. We have tested the feasibility of delivering the ABA intervention and have developed a training package for peer supporters. We have identified modifications that would enhance the training for intervention delivery and some of the trial processes. We were able to demonstrate the feasibility of recruiting women to, and retaining women in, the ABA study, as well as of collecting breastfeeding and health-related outcomes.

A full trial should be considered to test the effectiveness of the ABA intervention in increasing breastfeeding initiation and maintenance.

Contributions of authors

Dr Joanne L Clarke (https://orcid.org/0000-0003-2563-5451) (Research Fellow, Public Health) was study co-ordinator for the duration of the ABA study. She undertook the recruitment and data collection at site A, led on PPI, undertook the statistical analysis, contributed substantially to the qualitative analysis and process evaluation, and co-ordinated the writing of the final report, leading on the writing of Chapters 2 and 3.

Dr Jenny Ingram (https://orcid.org/0000-0003-2366-008X) (Senior Research Fellow, Child Health) oversaw the conduct of the study at site B and contributed to the qualitative analysis. She was a co-investigator and contributed to the study design and interpretation of findings.

Mrs Debbie Johnson (https://orcid.org/0000-0001-7390-6514) (Research Associate, Child Health) undertook the recruitment and data collection at site A. She contributed substantially to the qualitative analysis and process evaluation.

Dr Gill Thomson (https://orcid.org/0000-0003-3392-8182) [Reader (Associate Professor) in Perinatal Health] oversaw the qualitative research, conducted the data collection for the IFH interviews, and contributed substantially to the qualitative analysis and process evaluation. She was a co-investigator and contributed to the study design and interpretation of findings.

Ms Heather Trickey (https://orcid.org/0000-0002-3707-456X) (Research Fellow, Public Health) led on the development and delivery of the ABA training for IFHs. She also contributed substantially to the process evaluation, leading on the development and analysis of the antenatal fidelity checking. She was a co-investigator and contributed to the study design and interpretation of findings.

Dr Stephan U Dombrowski (https://orcid.org/0000-0001-9832-2777) (Assistant Professor, Psychology) contributed to the qualitative analysis and process evaluation, in particular the use of BCTs. He was a co-investigator and contributed to the study design and interpretation of findings.

Ms Alice Sitch (https://orcid.org/0000-0001-7727-4497) (Statistician, Medical Statistics) was a co-investigator and contributed to the study design and interpretation of findings. She oversaw the statistical analysis.

Professor Fiona Dykes (https://orcid.org/0000-0002-2728-7967) (Professor of Maternal and Infant Health) was a co-investigator and contributed to the study design and interpretation of findings.

Dr Max G Feltham (https://orcid.org/0000-0002-3891-5068) (Team Lead – Women’s Health, Birmingham Clinical Trials Unit) oversaw the running of the study from a trials unit perspective, including the randomisation process and setting up the database.

Professor Christine MacArthur (https://orcid.org/0000-0003-0434-2158) (Professor of Maternal and Child Epidemiology) was a co-investigator and contributed to the study design and interpretation of findings.

Professor Tracy Roberts (https://orcid.org/0000-0002-0624-0537) (Professor in Health Economics) oversaw the economic analysis. She was a co-investigator and contributed to the study design and interpretation of findings.

Professor Pat Hoddinott (https://orcid.org/0000-0002-4372-9681) (Professor in Primary Care) was a co-investigator. She had the original idea for ABA and for combining an assets-based approach with some of the learning gained from the FEST pilot trial, which she led. She contributed to the study design and interpretation of findings.

Professor Kate Jolly (https://orcid.org/0000-0002-6224-2115) (Professor of Public Health) led the study as principal investigator, leading on the design and conduct of the study throughout. She oversaw the writing of the final report, leading on the writing of Chapters 1 and 4.

Publications

Jolly K, Ingram J, Clarke J, Johnson D, Trickey H, Thomson G, et al. Protocol for a feasibility trial for improving breast feeding initiation and continuation: assets-based infant feeding help before and after birth (ABA). BMJ Open 2018;8:e019142.

Clarke JL, Ingram J, Johnson D, Trickey H, Dombrowski SU, Sitch A, et al. The ABA intervention for improving breastfeeding initiation and continuation: feasibility study results. Matern Child Nutr 2020;16:e12907.

Ingram J, Thomson G, Johnson D, Clarke JL, Trickey H, Hoddinott P, et al. Women's and peer supporters' experiences of an assets-based peer support intervention for increasing breastfeeding initiation and continuation: a qualitative study [published online ahead of print March 21 2020]. Health Expect 2020.

Data-sharing statement

All data are available on request from the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

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 PHR programme or the Department of Health and Social Care. 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 PHR programme or the Department of Health and Social Care.

Amendment 1, 4 April 2017

1. Several minor errors were identified on the 8-week and 6-month questionnaires. These included:

   • duplicate requests to tick and cross response boxes

   • incorrect numbering in 8-week questionnaire

   • incorrect signposting to next relevant question.

These were amended in the questionnaires.

1. We started the study using paper randomisation, rather than the web-based programme specified in the protocol. As the recruitment period was quite short, we proposed to continue using telephone randomisation at site A. Practical challenges led to a reconsideration of the randomisation method at site B. As volunteers were to deliver the intervention, they had constraints on how many women they could support at any one time. We amended the randomisation process to blocks for each locality at site B. When the number of women required to complete a block was recruited at a site, the women would be randomised at the same time and notified of their allocation status by telephone and/or letter. Changes were made in the protocol to reflect the changes to the randomisation process.

2. We clarified the procedure for following up participants if they did not return questionnaires at 8 weeks and 6 months. This was not clearly specified in the original protocol. The amended protocol reads:

   ‘Follow-up questionnaires will either be posted to participants to be completed at home, or participants will be called to complete the questionnaire over the telephone (according to preference indicated at recruitment).

   In the case of follow up questionnaires not being returned by participants, the following methods will be employed:

   • Sending another copy of the questionnaire in the post

   • A telephone call/text to the participant from the research team to encourage completion, or offer phone completion

   • Collection of the primary outcome only by telephone or text’

3. At the request of one of the organisations delivering the intervention, a sentence was inserted into the consent form:

   ‘The Sponsor, subject to agreement from the [ORGANISATION], may appoint a third party to access my identifiable data.’

4. In the study eligibility form, which was completed by the research fellow at recruitment, we added a question about whether the participant is aged < 25 years/≥ 25 years and a question about whether the participant is aged at least 16 years. This was to make the telephone randomisation process easier.

All amendments to the protocol were agreed on 12 June 2017 by South West – Cornwall and Plymouth Research Ethics Committee (16/SW/0336).

Site A

Site B

Data completeness: participant characteristics

Data completeness: estimates from feasibility study

Data completeness: data for economic evaluation

Use of child care at 8 weeks

Use of child care at 6 months

Receipt of benefits at baseline

Receipt of benefits at 8 weeks