Adapting Women's PLA Cycle for Nutrition

Discussion

The aim of this thesis is to provide insight into the adaptations required to theoretically adaption of the women’s group PLA cycle from multiple low-income contexts in the UK for using nutrition in infants of Bangladeshi origin as an example. In this chapter, I will present the main findings from study 1 and study 2. I will then contrast these findings in relation to the overall objectives of the project. Finally, I will represent the strengths and limitations of this research.

Summary of Main Findings

The findings of study 1 and study 2 are used as further evidence to support the adaptations that emerged from the narrative reviews. They also highlighted additional areas that could be included to the theoretical framework, particularly the micro-adaptations, where they suggested how to contextually adapt the intervention for the Bangladeshi population of Tower Hamlets. The following section will outline the main findings of each of the studies in more detail. The results of study 1 generated four themes relating to previous adaptation of the intervention across seven trial contexts. These themes allowed me to confirm my preconceived theories that the intervention had been adapted in response to the context, specifically that adaptation occurred systematically to preserve the fidelity of the core components of the intervention (Castro, Barrera Jr et al. 2004). Castro et al. (2004) suggested that, it was important to adapt the intervention for the local context (Castro, Barrera Jr et al. 2004). In the literature from the women’s group PLA cycle intervention, it was evident that, some adaptations had occurred to the number of community facilitators delivering the trials, to the frequency of the group sessions, to the length of the cycle period, and then to the micro-adaptations (language, creative work, content), but it was not obvious how these adaptations were tested, implemented or evaluated. The key informants from study 1 suggested that, adaptations were based on evidence but could also be made pragmatically by the implementation team and by the community facilitators, who were delivering the intervention. They confirmed that adaptations could be made to a set of components, but the participatory delivery of the intervention, peer-led facilitation, the PLA cycle, and the number of sessions that the peer facilitator is delivering in a rural context should be preserved. The main lessons that can be added to the theoretical adaptation framework for the intervention from study 1 are suggestions around the nature of the adaptation process, the core components and the micro-adaptations. The results of study 1 indicated that, adaptation can occur in two forms: evidence-based adaptation and pragmatic adaptation. The second reiterated the core components including the PLA cycle, the peer-led facilitation and the participatory tools. The number of sessions a facilitator should deliver within a rural context was excluded from the core components in this case, because it is not applicable as the intervention is being implemented in an urban context. This renders session frequency as an adaptable component in this instance. The final lesson from study 1 suggests that, we should include a set of micro-adaptations and that these micro-adaptations should include the design of the contents, delivery, logistics and materials. Logistics and delivery should also be adapted to emulate context, with specific adaptations made to the location of the groups, frequency of groups, length of the cycle and the size of the groups when working in an urban context. Within an urban context, there is also the need to pay attention to the elevated levels of heterogeneity and the potentially transient nature of the target population, and to adapt the materials, contents, delivery and logistics (micro-adaptations) for responding to these variations. Furthermore, the women’s group PLA cycle literature indicated that, local partners could potentially facilitate the adaptation process by assisting with the contextualising of the micro-adaptations, so that they can be tested, further adapted and implemented as appropriately and efficiently as possible (Azad, Barnett et al. 2010, Houweling, Azad et al. 2011, Tripathy, Nair et al. 2011, Nahar, Azad et al. 2012, Fottrell, Azad et al. 2013).

The results of study 2 generated evidence to support the literature which further encourages careful consideration of the heterogeneity in minority ethnic group populations and the influence of disparate environments and potentially the effect of acculturation on socio-cultural practices (Asad and Kay 2015, Sussman, Baezconde-Garbanati et al. 2018). It also supported the literature that challenged whether the Bangladeshi population would be receptive to a community-led intervention that was targeted solely at Bangladeshis, based on the UK census ethnicity data, because they considered themselves bicultural (British and Bangladeshi) (Berry 2005). This has implications for the theoretical micro-adaptations, specifically around the group composition, linguistic accessibility, delivery, contents and materials adaptation, because it indicates that, the population may have varying degrees of acculturation and those facets will need to be adapted accordingly. The main lessons that can be extracted from study 2 could form the theoretical micro-adaptations of the intervention. The results indicated heterogeneity within the Bangladeshi population of Tower Hamlets and the barriers and limitations that could occur in the urban inner-city context. They suggested that the population are influenced both by their Bangladeshi heritage and their lived environment. This could have critical implications for their socio-cultural preferences in relation to social mixing, food, language, family structure, employment situations and religious beliefs. It also suggested that, we should consider how long the participants have been living in the UK, as some may have migrated to the UK recently, some have spent several years in the UK and some were born in the UK. Each facet could have a different preference regarding the aforementioned socio-cultural preferences. It was also suggested that, we should recognise the inherent differences between LIC rural contexts and HIC urban context, particularly regarding the availability of space to hold the groups, transport to and from the groups, the influence that the structural barriers have on the population sense of community, and the ability of the groups to facilitate behavioural changes in the presence of free and good quality healthcare. The data also demonstrated that, male participants may prefer a medical professional-led intervention, whereas female participants may opt for a peer-led intervention. Moreover, it was suggested that, a mosque-based intervention may not be appropriate because not all mosques in Tower Hamlets accommodate women. Finally, the participants indicated that, they felt uneasy in expressing their negative experiences in front of the strangers, particularly male strangers. This further suggests that, there should be more relationship-building activities at the start of the cycle and that groups should be smaller and more intimate. Furthermore, the contents and materials should include both negative and positive experiences and messages; so that the participants feel comfortable expressing their views in a group environment once they have been reassured that their challenges are experienced by other families in their context.

Interpretation of findings in relation to objectives

The following section will represent the main findings from study 1 and study 2 in relation to each of the objectives and outline some suggestions for the NEON study to explore when they are testing the theoretical adaptations that have emerged from the results of this research.

Determining how the women’s group PLA cycle was previously adapted by interviewing key informants who have experience applying the theoretical model in a resource-limited setting

The data from the key informant interviews has attempted to demonstrate that, the intervention was adapted for each trial context; however, theoretical adaptation was not always recorded or evaluated. The MIRA Makwanpur trial was the original operational model from which all subsequent trials were adapted. Micro-adaptations had to be made to materials, contents, logistics and delivery in this trial to make the model relevant to the context. Each trial team emulated these adaptations for their context. This supported the evidence on adaptation from the literature around the frequency of groups (More, Bapat et al. 2008), number of groups led by each group facilitator (Fottrell, Azad et al. 2013), using a local artist to draw culturally and contextually appropriate materials (Manandhar, Osrin et al. 2004, Rosato, Laverack et al. 2008, Rath, Nair et al. 2010) and making adaptations to the contents (Tripathy, Nair et al. 2011, Prost, Colbourn et al. 2013). What the literature did not describe was the details of this adaptation and processes and how they were informed. Therefore, the topic guide in study 1 (5.10.4) was designed to explore previous adaptations with a particular focus on identifying the context, the process of adaptations, and the cultural and contextual adaptations. The results of study 1, theme 3 the process of adaptation described two ways that adaptations were implemented in the previous women’s group PLA cycle trials (section 9.1). This decision by the key informants not to pilot certain adaptations was neither intentional nor unintentional; instead, the key informants viewed these pragmatic adaptations as “obvious” areas of adaptation that would occur along the life course of the intervention development process. They suggested that, it is “common sense” to adapt certain areas of the women’s group PLA cycle to echo the context, for example adapting to the local language, and this did not require new evidentiary support. The key informants also indicated that the pragmatic adaptations were made by advisory and evaluation team from UCL or the implementation partners who had previously implemented RCTs in the trial context (Bangladesh (Azad, Barnett et al. 2010), India (More, Bapat et al. 2008, Tripathy, Nair et al. 2016) and Nepal (Manandhar, Osrin et al. 2004) (Bolam, Manandhar et al. 1998). The team were already familiar with the context prior to adaptation and were able to make educated decisions on the delivery, language and logistics for the intervention. Therefore, their prior knowledge of the intervention development and implementation process allowed them to made evidence-based decisions on adaptations without needing to conduct formal research. Most trials, however, it has been started off by applying formal research to conduct a needs assessment, but then continued to adapt the intervention through development and implementation, which suggests that, theoretical adaptation must be a continuous process.

Exploring why it was adapted by examining the experiences of key informants who have applied the women’s group PLA cycle in a resource-limited context

Arguably, the model was implemented in response to contextual demands and pre-ordained research agendas, set by international donors or governments and this was supported by the results of study 1. For example, the key informants from the Nepali trial suggested that, interventions that use didactic methods to provide health information had failed to reduce neonatal and perinatal mortality; therefore a new method of delivery was required (Bolam, Manandhar et al. 1998). This spans across both the external and internal factors, because the government and international agendas at the time of the MIRA Makwanpur trial focused on a reduction in neonatal morality, but the research team had to reassess how they could deliver the intervention in a way that would have an impact on neonatal mortality in their context (Manandhar, Osrin et al. 2004). The theme of context that influences adaptation demonstrates that there are many factors, which can influence the theoretical adaptation and that external factors may address what you do while internal factors address how you do it. This has implications for the adaptation process, because it illustrates that, the research agenda should address NHS and local priorities; by addressing both of these areas, there will be good support from NHS providers and population buy-in.

Identifying the modifiable components that could be adapted to emulate the context

The modifiable components were identified in Study 1, theme 5 the micro-adaptations and on-going adaptations are the areas of the women’s group PLA cycle framework, which encompass contextual adaptation. These adaptable components were grouped into a set of micro-adaptations, the set of adaptable components that allows the model to be adjusted specifically for the context. Only some of the key informants reported population heterogeneity in their context. This finding is curious because, although most of the clusters had a largely homogenous population, there was a lot of heterogeneity between the clusters in terms of the local languages spoken, education levels, socioeconomic status and ethnic groups. The levels of heterogeneity were not fully explored in the literature, however, differences in ethnic groups were described by Manandhar et al. (2004) (Manandhar, Osrin et al. 2004) and More et al. (2008) (More, Bapat et al. 2008) in terms of ethnicity, literacy levels, linguistic and culture. This further demonstrates that, the heterogeneity of the population had less influence on the design of the contents, materials and delivery because a lot of the health messages being disseminated focused on generic practices around health service delivery, hygiene, pregnancy, home deliveries and antenatal care (Prost, Colbourn et al. 2013). These care practices offer the same advice for all the participants in the cluster area; however, infant feeding practices in a non-indigenous population in an urban setting are more complex. This adaptation requires an understanding of how the context has influenced behaviour and practices.

Alternative methods to design tools and materials based on available resources

In study 2, the FGD participants were asked, where they accessed their health information. The participants said, they used the Internet and social media channels for this purpose. The male participants used more formal websites like NICE guidelines and NHS online, whereas the female participants would consult their peers on Facebook or WhatsApp as a first option, and then they are asked about their GDP as a second option. All of the participants agreed that, they would benefit from having the information from the groups available online in the event that they missed a session, or they wanted to share the information with friends and family who were in Tower Hamlets, in other areas of the UK or located overseas. They said this information could be available in a PDF or on a website, but they felt that, a PowerPoint presentation would not be suitable. They did not think that, many people would have access to PowerPoint in their community and that they did not have the digital literacy to use it effectively. They also suggested that, mobile phone penetration, particularly availability of smart phones, was not high in this population. Women in particular would often use a family phone or their husband’s phone and it was often monitored by another family member. The purpose of this research is to explore how the women’s group PLA cycle could be adapted from multiple LICs to the UK NHS and theoretically adapted for the Bangladeshi population of Tower Hamlets. The previous section has indicated some further suggestions for the theoretical adaptation framework that I will now represent it in the next section.

Theoretical adaptation framework for the reverse innovation of the Women’s Group PLA Cycle

This research aims to add to the existing knowledge of adaptation literature by presenting a theoretical framework for the reverse innovation of the women’s group PLA cycle based on the narrative reviews of the literature, the experiences of professionals who have previously adapted the model for their context and based on insights from the target population for the UK context. By listening to the reflections of the key informants in study 1, I was able to understand the process of adaptation and how each trial team used local sources and existing data to inform these adaptations. Drawing on the literature, I was able to identify the core components of the interventions, where some theoretical adaptations have occurred, and how they have been implemented (15-18, 20, 21, 51, 121). I will now present a four-step sequential process which illustrates factors that need to be considered, when adapting the intervention with so, it can undergo reverse innovation into the UK context.

Unique and salient features of this theoretical adaptation framework

The following section will outline the theoretical adaptation framework for the women’s group PLA cycle, which should assist with the reverse innovation of the intervention from multiple LICs to the UK context using nutrition in infants of Bangladeshi origin as an exemplar. Although there are other theoretical adaptation frameworks available within the literature (6, 181, 195-197), the framework for the adaptation of the women’s group PLA cycle has multiple unique and salient features that should be distinguished. Specifically, throughout the theoretical adaptation process, this framework aims to:

Utilising existing knowledge within the context in the form of stakeholders from government, members of the target population, representatives from the health sector, and local NGOs

Highlighting the necessity of utilising routinely available national and international health statistics (census data and/or existing health data), qualitative data collection (interviews and FGDs) and ethnography to understand the extent of the issue in each context

Recognising that some theoretical adaptations can be made efficiently through pragmatic adaptation (based on existing knowledge of the context) and some require piloting (through pretesting methodologies)

Suggesting the adaptation that should be an iterative process and occur throughout the intervention life cycle

Indicating the domains that can be adapted to emulate local context (materials, contents, delivery and logistics) and suggests how these can be adapted

The following section will outline each step of the theoretical adaptation framework and how it was informed by the results of the narrative literature review, study 1 and study 2:

Step 1: Setting up your intervention

Step 1 of the theoretical framework outlines areas that need to be considered, when setting up the intervention. This includes the core components, methods of group facilitation, systems strengthening, site selection, and engaging several local partners, the community and local services (section. These areas were all considered across all the seven women’s group PLA cycle cluster RCTs and should be considered when translating the intervention across incongruent geographical contexts. Each will be explained in the following section.

The core components

Core components maintain the fidelity of the intervention (7, 197). This PhD research demonstrates that, the women’s group PLA cycle has three core components. The examples, listed below, demonstrate how the women’s group PLA cycle was initially designed and the following section shows how these design features informed the core components of the intervention:

The theoretical underpinnings of the PLA cycle embody the philosophical principles of Brazilian Paolo Freire’s Critical Conscientiousness, which asserts that, through reflection and action, an individual can develop an awareness of their social reality (91).

The Warmi Project, which inspired the women’s group PLA cycle format, was led by American Lisa Howard-Grabman and implemented by a local team in Bolivia (84).

Robert Chambers’ PRA toolkit, a creative and iterative means to assess the issues in one’s surroundings (100, 198): and these tools offered creative methods of conducting research that could be led by the community (100).

The key informants considered these components as essential influences in the design of the women’s group PLA cycle. Based on this, I assumed that, these were the core components of the women’s group PLA cycle intervention:

The PLA cycle – inspired by the Warmi Project (minus the post-implementation evaluation stage) (84)

The participatory toolkit – inspired by Robert Chambers (100)

Community-led facilitation – inspired by Robert Chambers (100) and utilised in the Warmi Project (84)

Encouraging change by a critical examination of lived environment (Paolo Freire) (91)

Each of these elements is recorded in theme 1: setting up your intervention in the framework. Other factors mentioned by the key informants that were considered essential parts of the implementation process, but not essential parts of women’s group PLA cycle core components were:

Engaging with a local NGO, the community and local services

Framing the topic to create buy-in from government, health services and the target population

The remaining components of the intervention can be adapted to emulate the context. The context specific adaptation will be addressed in 4. Micro-Adaptations.

Method of group facilitation

The key informants described using local respected women to facilitate their women’s groups through the PLA cycle. Some reported that, although a literate facilitator is more practical, they are not always available, and that further training and assistance could help to illiterate facilitator and manage each group successfully. The group facilitator should be fully trained and remunerated for their time. It is important to deliver rigorous training, observe their first few sessions and conduct refresher training every six months to optimise their ability to deliver the intervention in each group. Sub-optimal training can lead to the group facilitators offering their own advice or interpreting the facilitator manual incorrectly, therefore compromising delivery of the intervention.

Systems strengthening

The women’s group PLA cycle contributed not only in strengthening health system but also wider “system” strengthening including provision of healthcare, availability of health facilities (particularly in rural contexts), food accessibility/ availability (rural India) and availability of information of not only neonatal health outcomes but hygiene, nutrition, and sexual and reproductive health. Previously, the intervention successfully reduced neonatal and perinatal morality in multiple LICs (15, 18, 20, 102, 104, 111, 121) through a two-pronged approach that offered participatory learning rather than didactic health information delivery (19). It also addressed problems at the in the community, from the demand and in healthcare facilities, at the supply-side (92). Nevertheless, five out of the seven previous trials conducted an antenatal, delivery and new-born care service audit and then provided health service strengthening in intervention and control areas (15, 16, 19-21).

Site Selection

The key informants all suggested that, they used routinely available quantitative data to help them identifying a prominent health problem within each context. They also suggested that, collecting information on prevalence and incidence of the chosen problem could help with the monitoring and evaluation of the trial. For acute problems such as neonatal mortality this is easier than collecting data on chronic illness reduction or general behaviour change. They used routine channels to collect information and they selected a site. Site selection was influenced by the local NGO’s experience conducting research in the specific locality. Previous trials reported health systems strengthening as part of their intervention (15, 16, 18, 20, 21, 94, 102). Prost et al. (2013) reported that, service strengthening activities were conducted in all seven trials but differed across each context (2). The key informants from study 1 further implied that, these activities did take place in both the intervention and the control clusters but did not report a standard set of health strengthening activities. This is an important finding because it demonstrates that service strengthening should occur within the limitations of the project and also in response to the context. Interestingly More et al. (2008) reported that, community mobilisation through women’s groups was successful in areas where there were high neonatal mortality rates, and that areas with low rates of neonatal mortality should focus on health service strengthening (18).

Engaging a with implementation partners through a local NGO, the community and local services

In previous women’s group PLA cycle trials, the UCL team worked in collaboration with multiple local partners to facilitate the adaptation and implementation process. These included NGOs, local government, ministries of health and influential figures within the target population, community health workers, health professionals and applied health professionals. The purpose of the local partner was to advise on adaptation and implementation and to assist with the community entry process and formative research stage. The key informants described the value of local partners’ existing knowledge of the context, ability to facilitate the community entry process and their ability to navigate local and national policy. Using local partners was a strategy also used by Wingwood et al. (2008) in the ADAPT ITT-model for the inter-contextual adaptation of effective HIV interventions (195). Regarding implementation, most of the key informants described engaging with a local NGO who assisted with the formative research, and the adaptation, implementation and management of the trial. UCL were partners, but mainly managed the monitoring and evaluation. The local partners led the data collection, management of groups, training of the ground facilitators and the initial exploratory phase. They were instrumental in the implementation of the intervention, where they took a lead role in coordinating human resources and quality control. The NEON project has engaged several local partners; some had a working understanding of Tower Hamlets and were able to recruit community facilitators, others were familiar with the intervention and were able to offer advice on the intervention design. It may be that, the UK NHS adapted the model that does not require the same level of input from the local partner because the Bangladeshi population of Tower Hamlets is potentially more accessible than the rural villages in six of the cluster-RCTs. The Bangladeshi population of Tower Hamlets can also access information about the NEON project and their partner NGO on the Internet, and this offers credibility. Additionally, UCL is recognised as a credible institution within the UK and was considered respectable by the community facilitators. This raises the question of whether or not a local partner is needed to advice on the adaptation and implementation of the women’s group PLA cycle intervention in the UK context or whether it would be better to use them for the community entry process alone.

Step 2: Context that influences adaptation

Step 2 describes different factors that influence the adaptation process. These included internal and external factors that may change a research team’s aims and objectives. These are important to consider when designing the intervention from both the (1) stakeholder and (2) target population perspective. The aims and objectives should align with national and international health agendas to encourage support from stakeholders at these levels. The target population should identify the exemplar as an issue within their context to increase engagement with the intervention, encourage ownership, and improve sustainability of the intervention within the trial context. Apart from the physical context where the women’s group PLA cycle is being implemented, I wanted to consider the research context and how it has evolved over the last twenty years in terms of donor requirements and compliance, regulatory bodies and ethics committees. The first women’s group PLA cycle trial was conducted before the publication of the MRC’s guidelines for developing and evaluating complex interventions (Craig, Dieppe et al. 2008), and a few of the key informants mentioned that before these guidelines were published projects could be funded without formative research. Furthermore, they suggested that, it was possible to make adaptations to a model without a stringent piloting phase. This could be the reason that, the previous women’s group PLA cycle trials did not have a universal adaptation process. Castro et al. (2006) suggested that, an intervention can be adapted for culture, but it must maintain its fidelity by preserving the core components (Castro, Barrera Jr et al. 2004). If the previous trial teams were making some adaptations based on their personal experience, they cannot assume that others will have the same experience. Therefore, it is possible that the women’s group PLA cycle’s fidelity could be compromised, because there are variations in the implementation process, and this supports the need for a consistent adaptation process. A generic framework for theoretical adaptation could facilitate pragmatic adaptations which could potentially save time and money, but it does need to respond to changing in the research landscape as well and this could occur in relation to funding, additional health systems/government/local restrictions and differences in ethnics committees’ requirements or individual committee members preferences. For example, the NEON study had to submit NHS ethics on three occasions before it was finally accepted, and the study team decided to change which NHS ethics committee region where they submitted the application, because they recognised that the other region had experience in participatory intervention processes and therefore, may be understand the requirements of the study. Understanding the research context and responding accordingly could be an important area to consider for future applications of the intervention as well, if they are going to be implemented in HICs that may have stricter regulatory bodies.

Step 3: Process of Adaptation

Adapting programme materials, tools and approaches that are culturally and geographically appropriate is a critical part of the design and delivery of interventions. Formative research is an activity conducted at the beginning of the intervention design process. It is the systematic collection of information about the activities and characteristics of a target population or context that can improve the trial/programme/intervention and allow the trial team to make judgements about the potential of the trial/programme/intervention effectiveness, and/or inform decisions about future design, development or implementation. It is used to gain insight into the health issue or behaviour the intervention intends to address; relevant characteristics of the target population; communication access, habits and preferences; and the main drivers of behaviour. Methods of formative research that were used in the women’s group PLA trials included a wide range of quantitative and qualitative methods depending on what research teams needed to know to design an effective intervention. Quantitative methods generate numeric data and are often designed to produce information that is statistically representative of the intended audience. Qualitative methods collect verbal, descriptive information that is often rich in detail but cannot be generalised to an entire population or intended audience. These methods can help the research teams to discover and explore themes or processes, generate illuminating and illustrative personal narratives, and uncover attitudes or ideas that are common among members of a population, but they cannot be used to determine the proportion of people in an intended audience who think or act in a particular way. Qualitative methods may be used, when the research teams have limited resources, lack formal training to collect and analyse quantitative data analysis or do not need to estimate the proportion of an attribute in the population. However, using them properly still requires particular skills and sensibilities.

Step 4: Micro-adaptations

Steps 4 suggests a set of micro-adaptations which attempt to demonstrate the complex interplay between environment and heritage and how this can affect adaptation. They are specific to the women’s group PLA cycle intervention but could potentially be applied to similar participatory models.

Considering cultural and context to support the adaptation of the micro-adaptations

Manipulating the micro-adaption to echo national or regional characteristics could assist in optimal adaptation, integration and uptake of the intervention across different settings. The results from study 1 and study 2 confirmed some of the theories that I had around the adaptation process including Resnicow’s (1998) model for adaptation, which suggested that interventions should be culturally sensitive and this could happen by adapting for surface (matching intervention components to visible characteristics) and deep structures (emulating social, psychological, environmental or historical factors) (Resnicow, Baranowski et al. 1999). An example of how I applied this theory to my research was illustrated by study 2, which further suggested that interventions should be adapted for context and culture, and that this process should be continuous because the populations are changing in response to environmental changes. It also influenced some decisions around how the women’s group PLA cycle should be adapted for that population from study 1. For example, the contents and facilitation should be in Bengali and English, and that the groups should not occur on a Friday because of Friday prayer at the mosque. The groups should be held in a public space but should not be associated with the mosque, because not all mosques allow women to enter. The groups should be smaller and local, so the women who are living in the overcrowded urban environment can reach them without paying for travel expenses or in time. All of these features will be fed into the NEON study so they can test the validity of these theoretical suggestions.

Cultural stipulations like the example described above, emerged in Chapter 3 on various occasions, but the literature presented no definition of culture or how to measure it within a population. Perhaps, you cannot succinctly define or measure culture, because it is not a constant. It is influenced by contextual influences, popular influences and heritage. The idea of defining an individual or population so that you may design for them appropriately is perhaps an illustration of where the traditional objectivist public health research paradigm clashes with the subjectivist social sciences participatory research paradigm. In Chapter 3, I introduced the concept of cultural adaptation, culture in this sense being synonymous with ethnicity. I suggested that, culture extends beyond ethnicity to include creed, colour, lived environment, ancestry and country of origin. It could be important to detail a target population, but it raises the question if research teams need to pigeonhole target populations based on their ethnicity, culture or race.

This is exemplified by Liu et al. (2012), who described a 46 typology for the adaptation of health promotion interventions for ethnic minority groups in the UK (200), but they failed to account for the differences between each sub-population. For example, they grouped all south Asian populations together and did not describe the differences between each sub-population. The closest theoretical adaptation framework to this list of micro-adaptations was published by Liu et al. (2012) (1). They suggested that the research into adapting the interventions should be led by a principal investigator of the same ethnicity (1). This could be problematic in populations that identify as bicultural. Presuming that the population would identify more with their ethnic minority status above their context may be appropriate for first generation migrants, but it might not be appropriate for those that were born in the UK, especially as the FGD participants indicated they consider themselves British Bangladeshi.

The participants of the FGDs described this shift in cultural beliefs and practices when they discussed how the intervention could be adapted for their population. The data demonstrated a paradoxical relationship between Bangladeshi heritage and their Tower Hamlets context. This indicated the process of acculturation whereby the individual or sub-population can change based on their exposure to different environmental stimuli (Berry 2005). The data from the FGDs demonstrates the heterogeneity within the Bangladeshi population and exemplifies why interventions cannot be adapted to encompass only ethnicity. Liu et al. (2012) described how to culturally adapt health promotion interventions in the UK NHS context (Liu, Davidson et al. 2012). They stratified ethnic minorities into three broad categories, including grouping all south Asian populations. The recommendations of Liu et al. stressed that further research was required to determine adaptations to sub-groups (Liu, Davidson et al. 2012). Liu et al., also suggested adapting for salient language, ethnicity, religion and culture but did not offer means of capturing those who may have been born in the UK, but identify as bi-cultural (Liu, Davidson et al. 2012).

Berry (2003) believed that, there were four models of acculturation that can occur– integration, assimilation, separation and marginalisation (Berry 2003). Tower Hamlets has a large and concentrated population of British Bangladeshis, which could indicate that they might be receptive to participatory models that piggy-back from existing community structures, however, the FGD participants suggested considered themselves to be British Bangladeshi and that interactions with their environment had changed their expectations for health provision, their eating habits and their language and customs. Alecar and Deuze (2017) stated that, acculturation rates were dependent on the environment or socioeconomic status of the individuals (Alencar and Deuze 2017).

The FGD participants had all been born in Bangladesh. The older men were more conservative, whereas the younger men were more involved in the care of their children. This could be a generational difference rather than a cultural difference. Measuring rates of acculturation within Tower Hamlets could be challenging due to the diversity of the population, the intergenerational shift in wealth, culture and also due to the inter-cultural acquisition of information between the indigenous population and the Bangladeshi population in Tower Hamlets (Alencar and Deuze 2017). However, as the study 2 sample was small and employed a qualitative design, these results cannot be extrapolated to the wider Bangladeshi population of Tower Hamlets. Instead, they can be used to indicate to the NEON study that there could be a change in cultural norms and values, and this should be considered in the trail design. Resource restrictions may make it difficult for any trial team to continuously adapt and/or fully understand the degree of acculturation within this population, but there should be room for the model to respond to these changes. Applying social identity theory could help unpick the dynamics of social identities in the Bangladeshi population, which has demonstrated a conflict between cultural heritage and lived environment. The social identity theory is a socio-psychological approach, which aims to understand the interplay between self-perceptions, associated cognitive processes and social beliefs in group processes and intergroup relations (Hogg 2016). It suggests that, a specific social group can display specific collective behaviour, encourage social mobilisation and protest, have different leadership within and between groups, have social influence, and demonstrate generic groups norms (Hogg 2016). Adapting for collective social identity could be a new and proactive means of adjusting interventions to emulate the increasing acculturation and assimilation of migrant population in the UK. Equally, as the Bangladeshi population of Tower Hamlets includes individuals that were born in the UK, it raises questions about the suitability of the terms such as acculturation and assimilation because these individuals are not migrants. Being born into British culture and a British environment could affect their individual and collective identity. Social identity theory suggests that there is a salient identity hierarchy, where an individual’s commitment to delivering the behaviours associated with that identity affects the relationships within their social group (Hogg 2016). This could be further explained by the loyalty that an individual feels towards a certain part of their identity (Martin 1995), such as their nuclear family unit. Adapting for the salient social identity in the Bangladeshi population of Tower Hamlets could be an appropriate means of adapting the women’s group PLA cycle for this population as the results from Study 2 demonstrated that context and heritage both contribute to their behaviour. However, difficulties lie in measuring acculturation within this population and the NEON study team should be mindful of the high levels of heterogeneity within this population and potentially avoid making any assumptions about behaviours based on their ethnicity, religion or migrant status.

Delivery of the model within Tower Hamlets

There was a framework of norms, behaviours, and artefacts within the FGD sample illustrated the relationship between past and present experiences, and how these contributed to the participants’ cultural identity that could benefit the design of the NEON study (Hogan and Coote 2014). Schein (2010) theorised that, culture is a multidimensional entity with many layers (Schein 2010). The data from Study 2 strengthened this by outlining the paradoxical relationship between Bangladeshi heritage and British context, and how both influenced behaviours. This has implications for adaptation of the delivery of the women’s group PLA cycle for the Bangladeshi population of Tower Hamlets because it demonstrates the heterogeneity within the population. Heterogeneity was identified in Study 1 as a potential limitation. The key informants did identify that adaptations cannot be made to echo every social nuance in a population; therefore, the micro-adaptations cannot be made to each facet of the social identity of the target population. To overcome this challenge the group facilitator must have full comprehension of the facilitator manual and be able to converse in the salient language and other dialectics/languages spoken by their specific group. Within Tower Hamlets, the facilitators would need to speak English, and potentially Bengali with an understanding of Sylheti and other dialects. Previous trials operated in contexts that had heterogeneity between the clusters, but the key informants suggested that, the composition of participants within each cluster was largely homogenous. This was because the clusters were assigned, generally by village or by administrative division. Participants within each village would at least be familiar with one another and there were a lot of extended family members living in the same village. This was not the case in the Mumbai trial which was conducted in the transient and high-density slum of Dharavi. The key informants from this trial reported that, women would often visit the slum to access health services to give birth in the local areas.

Language

Adapting the model for the UK NHS context and a different exemplar means that the NEON study team are not translating materials or contents that have been used in previous trials contexts. Resnicow et al. (1999) suggested that an intervention should be adapted from surface and deep structure (Resnicow, Baranowski et al. 1999). They indicated that, one of the surface structures was language. Adapting language was supported by the results from Study 2, which demonstrated that, the population in Tower Hamlets speak Bengali and English, and the contents and materials will need to be available in both the languages. There is the potential for these to be mistranslated or for the terminology to be misinterpreted because the structure of Bengali and English are different. Addressing differences in terminology has been recognised as a potential barrier internationally to the success of interventions, which are being adapted (Pokhrel, Regmi et al. 2008, Sussman and Palinkas 2008). Sussman et al. (2018) suggested using forward and backward translating is effective to overcome potential problems. As previous trials have adapted the model iteratively, it may be that the components which need to be adapted can be tested and terminology can be discussed to optimise the translation and preserve fidelity of the health information being delivered.

There are limitations associated with translating materials in a locality where several languages are spoken. Kleinman et al. (1978) suggested that, the process of translating an intervention into the language spoken within the context can be difficult, when there are differences in cultural understandings with respect to idioms and explanatory models, which are aimed at initiate a discussion around health or behaviour change (Kleinman, Eisenberg et al. 2006), but did not suggest how to overcome language barriers. The FGD participants suggested speaking Bengali, English and other Bengali dialects, but not everyone was able to read each language. The key informants suggested using a well-trained group facilitator who could speak the salient local language and other languages/dialects to deliver the intervention. This is a cost-effective and practical means of overcoming translation issues. Sussman et al. (2018) stressed the importance of translating the intervention into the appropriate language, literacy level and terminology when translating interventions between nations (Sussman, Baezconde-Garbanati et al. 2018). Linguistic accessibility includes adjusting for language, which includes adapting written or spoken information so that it is comprehensible to participants with different cognitive abilities. This can include simplification of scientific language, focusing on main areas only, using a simple oral or written structure so that the disseminated information is clear, focused and accessible. The data from Study 1 demonstrated that, it was beneficial to disseminate information to the groups using a variety of techniques that are preferred by the target population. This could reflect the target population's preferred method of communication: storytelling, poetry, and literature; hands-on/interactive learning; testimonials; face-to-face. They also suggested that, they did not want to sit around talking; they wanted activities, demonstrations or games to initiate discussions in the groups. Factors such as literacy levels and comprehension of health information delivered by an intervention was described by Sussman et al. (2018) (Sussman, Baezconde-Garbanati et al. 2018), but they only referred to these obstacles at a national level and mainly focused on LICs. The women’s group PLA cycle intervention does not necessarily need the participants to be literate to participate, but written materials may be issued at the end of each session. All the FGD participants were literate but did suggest that, they knew people in their community who were not. Levels of education can vary within each context, and this can affect the comprehension of health messages (Sussman, Baezconde-Garbanati et al. 2018). Appreciating that the education levels within the UK NHS context may vary and that some participants may need pictorial tools as well as written information is a key adaptation for the context.

Creating a Community

Providing a cohesive network similar to the rural villages that previously received the women’s group PLA cycle intervention in India, Nepal, Malawi and Bangladesh could assist with the embedding of the intervention within the Bangladeshi population of Tower Hamlets. Therefore, the intervention for the Tower Hamlets would need to reconceptualise a community in this context. This could be done by designing a value proposition, which appeals to the target group (Bangladeshi population, who have an infant between 0-2 years, which are living in Tower Hamlets). Establishing the drivers that are causing these individuals to practice sub-optimal feeding practices and reframing them to reflect what they value could help form a new community around the intervention. This research has demonstrated that community is potentially a critical component that facilitates the success of the women’s group PLA cycle. One of the core components of the women’s group PLA cycle intervention are the philosophical principles of Paolo Freire’s critical conscientious(1972), which encourage participants to critically examine their environment together as a group (Freire 1970). Arguably, Tower Hamlets is not a traditional community because of the urban context, autonomy of individuals and the transient nature of the context. Nonetheless, it still boasts a large and concentrated population of British Bangladeshis. RotheramBorus et al., (2012) also suggested that, targeting minority populations within their locality with appropriate marketing, making the intervention culturally appropriate and providing incentives could be a way to increase participation in interventions (Rotheram-Borus, Swendeman et al. 2012). In Tower Hamlets, it is not the visibility of the Bangladeshi population that is problematic, but understanding how to engage them in health services and how to encourage them to question their environment to promote the Freire model of change (Freire 1972). The participants described Tower Hamlets as a radically different context, where they felt less capable of making changes. They also suggested that, people within the Tower Hamlets context had higher standards and they were unsure if the participatory action approach would be attractive to people who could visit the doctor instead. This raises questions about the possibility of bringing the target population together to encourage collective action within an urban environment where they have less ability to encourage institutional change. Parallels can be drawn between the situation in the Mumbai trial context Dharavi (More, Bapat et al., 2008) and that in the Bangladeshi population of Tower Hamlets. The population of Tower Hamlets exudes similar features through their Bangladeshi heritage, and the FGD participants did recognise these as traits, that bonded individuals within their context. They suggested that, they could tell if someone was Bangladeshi by their physical features such as hair and skin colour, and practices such as their religion. Nonetheless, physical features and faith are not the only defining attributes of a community. A community could be based around the people with similar socioeconomic status or similar interests. For the purpose of creating a community for Tower Hamlets, it may be better to base this around a group of people with similar interests such as child health. This was supported by the results of Study 2, where the female FGD participants articulated that they would be interested in attending groups that centred on infant growth and development. They also suggested that, this was enough to encourage attendance of new and experienced mothers as it was not confined to complementary feeding practices. The participants in Study 2 reported feeling unable to share their problems with their neighbours in Tower Hamlets. There was unease amongst the participants at the thought of sharing their problems in public. The male participants did voice concern, but it was the female participants that were particularly anxious about the women’s group format.

Community-led versus health professional-led health intervention delivery

The previous women’s group PLA cycle interventions utilised a variety of methods to disseminate information regarding neonatal mortality in each groups (Prost, Colbourn et al. 2013). The NEON study team should consider input from health professionals and the target population, when they design their tools. Local women could be trained as group facilitators to assist the groups through the PLA cycle. With the support of trained supervisors, who could oversee the groups, monitor the learning objectives for each session and suggest activities that could be used as strategies to combat neonatal mortality in each context. The key informants suggested that group facilitators were local mothers who were respected within their community. They suggested that, literacy was not an absolute necessity, although it was more practical to have a literate facilitator. The important trait was the individual’s credibility within the group. The FGD participants agreed that, they would seek advice from a respected individual. Being respectable was an attribute that they all considered very important. They wanted to seek advice from a respectable individual, and they wanted to be seen as respected so others would seek advice from them. The male and female participants respected different people in their community. The male participants considered a medical professional to be a respectable person and the females said that, they would consider an experienced Bangladeshi mother from Tower Hamlets to be respectable. The men also suggested that, they should be referred by their GP to the adapted women’s group PLA cycle intervention. Although this could be an adaptation to the intervention that needs to occur in Tower Hamlets, and it challenges the participatory nature of the process. It is possible that, the intervention will need to become less participatory if it is to be embedded within the formal NHS services within Tower Hamlets. Populations in previous trial contexts did not have the same access to or trust in health facilities as they do in Tower Hamlets, therefore a health professional-led model is not appropriate. The female participants considered different people to be respectable. First time mothers would seek advice from their mothers and mothers-in-law when they had their first child and more experienced mothers considered experienced mothers to be respectable as they were able to give advice on infant care. If the mother/mother-in-law was from Bangladesh, then they would listen to their advice, but not always practice it as they felt it was “old fashioned”. Unlike the male participants, the females did not want to seek advice from a medical professional, particularly the health visitors, for two reasons: they thought that they might be viewed as a bad parent and they often received conflicting advice. The women’s group PLA cycle literature suggested that the ‘respected’ peer-led facilitation was a core component (Prost, Colbourn et al. 2013). This could have implications for the delivery of the adapted women’s group PLA intervention because it demonstrates that, men and women respect different individuals or institutions in the Tower Hamlets. The male community facilitators indicated that they wanted to be linked to the NEON project. They suggested that people would trust them if they knew that they were “working” for UCL. The female community facilitators were not as concerned with being associated with UCL. The male community facilitators found it challenging to recruit men for the FGD in Study 2. The male participants suggested that they did not have time to attend the groups because they had “other social and economic commitments.” Hypothetically, the male community facilitators may have found that, recruiting men would be easier; if he was attached to a familiar institution and that an identification card would help them in demonstrating their position. The female participants were recruiting females, and they did not suggest that, they required the support of the institution to do so. Potentially this is because the women in this setting were recruited by someone they trusted, which made them feel comfortable attending informal groups. The response from the female participants is consistent with the literature around cultural adaptation, particularly Falicov (2009) who suggested that interventions should be run by staff of the same ethnicity as the target population (Falicov 2009) and Liu et al. (2012) also believed that, the principal investigator in each intervention should be of similar ethnicity to the target population (Liu, Davidson et al., 2012). However, neither of them considered that additional factors such as socioeconomic status or gender may influence an individual’s or a group’s ability to interact with an intervention. By considering intersectionality, we may be able to adapt for differences between male and female participants, individuals that consider themselves bicultural and those that do not, and other phenomena such as migrant populations beginning to engage with their new environment.

Framing the Topic to encourage buy-in in Tower Hamlets

In the reverse innovation literature, Depasse and Lee (2013) advocate for an emphasis on innovation crossover, when innovations pass from LICs to early adopters of innovations in HICs (DePasse and Lee 2013). Harris et al. (2015) suggested that, emphasising the frugality of the intervention could make it more appealing to HIC health systems (Harris, Weisberger et al. 2015), but neither of them have addressed how to support uptake of the intervention when it is implemented within the context. Unpicking what makes an intervention attractive to the community it is targeting could encourage buy-in. One-way to support buy-in could be possible to create a values position for the intervention that recognises the main challenges in the implementation context. In the initial seven trials, the key informants suggested that neonatal and perinatal mortality were chosen as exemplars in previous trial contexts for three reasons:

They are a global health priority as identified by the WHO

They are considered a health priority by national governments and ministries of health

There is demonstrable evidence to suggest that neonatal and perinatal mortality was a problem within each community in the context

Each of these areas helped to boost support for the intervention, which allowed it to become embedded within the community and accepted by healthcare providers. However, previous trial contexts often had little to no healthcare provision; therefore, any incremental changes would have an impact on trial outcomes. Where previous trial participants may have seen rapid decreases in the outcomes, it may take longer to show an impact on nutrition-related health in the Tower Hamlets Bangladeshi population. Chronic illness prevention requires a sustained long-term behaviour change and there is no quick fix. Addressing chronic illnesses rather than acute illness could require long-term behavioural change and this could affect the longevity, contents and delivery of the intervention further. There will need to be a variety of topics to keep the participants engaged over the length of time that is required to show an impact on obesity rates. This raises questions about engagement, retention and sustainability of an intervention that is using a participatory approach to address chronic illness. This could require specific short-term outcomes so that behaviour is being reinforced regularly (Michie, Atkins et al., 2014). Framing the topic to create buy-in from the target population is only one part of supporting innovation dissemination. Harris et al. (2016) believed that, the process of reverse innovation is complex, with both the HIC consumers and the commissioners of healthcare requiring persuasion to adopt innovations from LICs (Harris, Bhatti et al., 2016). Framing this intervention as an alternative to current health services could promote buy-in from the health sector. Harris et al. (2016) suggested promoting interventions from LICs as a frugal alternative to current services in a bid to increase adoption by HIC health systems (Harris, Bhatti et al., 2016). Both are the methods of increasing support for the adapted women’s group PLA cycle within the Tower Hamlets context. Tower Hamlets is a densely populated and heterogeneous environment, where not everyone is aware of their neighbour’s issues. One challenge that could influence buy-in and uptake of the groups was addressed in theme 3 apprehension about sharing experiences with unfamiliar faces. This theme highlighted worries expressed by some of the female participants about being judged by their neighbours, strangers or other people’s husbands. This was a similar concern in the Mumbai trial in Dharavi. Pregnant women would visit Dharavi to access maternity services. The women and their families were visiting from across India; they would not always speak the same language, share the same customs or be the same religion as their neighbours. This created barriers to accessing pregnant women because of the length of the duration of their stay in Dharavi and also created barriers to share experiences because of insecurities about being judged by strangers. The women’s group PLA cycle has previously been implemented in contexts with little to no access to healthcare. In areas where healthcare was available, it may be costly, unregulated or limited by resources such as drug stock-outs and out-dated equipment. Therefore, any incremental change that was made by the groups in these contexts would have an impact on the outcomes of the trial. In Tower Hamlets, there is health care that is regulated, free and accessible. On a local level, having access to good quality health care that is managed by a central body and is standardised across the UK that could make it difficult to advocate for change at this level. This raises questions about the ability of the women’s group PLA cycle to encourage social change through collective action in the Tower Hamlets context. Potentially, the social change at a local level may be challenging but changing family or personal habits of groups of locals may be more feasible. Equally, if the provision of health care in Tower Hamlets is remaining constant and the rates of nutrition-related ill health are increasing, then it could be that a preventative intervention that operates out with the realms of the NHS could be successful. Other factors that could prohibit collective action could be the lack of prominent and singular community structure within Tower Hamlets which could be utilised to piggy-back the intervention.

Theoretical and Conceptual Strengths of this research

Data collection techniques

This research project involved both key informants who had adapted the intervention for their specific contexts and members of the Bangladeshi population of Tower Hamlets who suggested how they would like the intervention adapted for their context. As such, it provides a balanced view of how the intervention could be successfully adapted for the UK NHS context and the Bangladeshi population of Tower Hamlets. This provides health policy makers, commissioners of health and health service designers (international and local) with a framework that can support the translation of the women’s group PLA cycle intervention between LICs and HICs. Secondly, the topic guides in both Study 1 and Study 2 offered questions that further allowed the participants to demonstrate their attitudes and beliefs surrounding adaptation. For the key informants specifically, the topic guide was designed with prompts to encourage them for examining their experience of adaptation. This facilitated a richer account that was able to build on the adaptations that had been described in the literature. The participants in the FGDs were prompted using visual aids to encourage a more engaging conversation around adaptation. The visual aids sparked conversations around delivery methods, logistics, materials and contents of the intervention. This strengthened the methodology because it made the session more interactive which made the participants feel more relax. It also increased engagement, as the groups could discuss each picture with the researcher and the group facilitator. Strength in this methodology was the presence of the group facilitator who could speak English, Bengali and Sylheti, who had received training from the researcher on the session, training on how to facilitate FGDs and was familiar with the NEON project. This was strength within the project, as it means that, the FGDs were managed well and that everyone could access the information.

Strengths in relation to Policy

This thesis not only highlights the need for generic theoretical adaptation frameworks so that interventions can be adapted within contexts, it highlights how this could be effective by going further to demonstrate how the framework could be applied to a population, context and exemplar (Bangladeshi population of Tower Hamlets targeting infant nutrition). Developing a generic framework for the adaptation of an intervention could assist with the global translation of the intervention and, thus, support the bi-directional flow of knowledge, interventions and learning from LICs to HICs (Syed, Dadwal et al., 2013). This would potentially save policymakers the cost of carrying out feasibility studies to provide future frameworks for adaptation of participatory interventions that are being adapted between LICs and HICs. This thesis shows how potential policy makers could encourage interventions to move through the innovation pathway. Depasse and Lee’s (2013) (DePasse and Lee, 2013) a model for reverse innovation in healthcare described a unique pathway where innovations spread from LICs to HICs. The framework considered their recommendations of encouraging the stakeholders to endorse innovations by including external influences in a context that Influences Adaptation. Adapting the intervention to adopt a national government’s priorities could create buy-in from macro stakeholders and also could facilitate the adoption of interventions from LICs by HIC adopters. This was also demonstrated in the data from the key informant interviews, where most key informants described targeting neonatal mortality because it was a priority for governments and for global policy makers. Specifically within the Tower Hamlets context, this thesis demonstrates that barriers and limitations highlighted in the literature (acculturation, assimilation and biculturalism (Sam and Berry, 2010) are affecting the Bangladeshi population of Tower Hamlets. The population is being influenced by both their cultural Bangladeshi heritage and their Tower Hamlets lived environment. This has implications for policy makers because it suggests that materials, contents and delivery need to encompass both domains for their British Bangladeshi identity, which includes, for example, offering nutritional advice for infants that includes European and South Asian options and hosting groups in English but having a facilitator that can speak Bengali and any relevant dialects. This is a new perspective on cultural adaptation and goes beyond the recommendations of Liu et al. (2012) adapting health promotion interventions for ethnic minority groups by exploring the interplay between the two influential domains of heritage and environment (Liu, Davidson et al. 2012). It could assist the policy makers to design future campaigns legislation and provide recommendations for British minority ethnic groups

Theoretical and Conceptual Limitations of this research

Steps were built into the methodology of this project (see Study 1 Methods and Study 2 Methods) to optimise the credibility, auditability and fittingness of the results. Despite these efforts, some limitations still existed in the theory and conceptualisation of the thesis, the research strategy and data quality. The following section will discuss some of the limitations of this thesis.

Concept of reverse innovation

The tenets of reverse innovation are still poorly defined. The literature demonstrated that HICs and LICs could benefit from the global acquisition of knowledge, but it did not offer a framework to demonstrate how an intervention could be moved between LICs to HICs. It also did not offer insights into how an intervention can be successfully integrated within a new context. This thesis has attempted to create the first theoretical framework for the reverse innovation of a participatory intervention. This framework has yet to be tested and the intervention has yet to be implemented within the UK NHS context and, therefore, it remains theoretical. This means that, although the intervention has theoretically been subjected to reverse innovation, it still needs to be piloted within the UK NHS context. It however, demonstrates that, this intervention could potentially be subject to reverse innovation. Therefore, further research is required to pilot and implement the intervention within the context to examine the real-time barriers and limitations of adapting an intervention developed in a LIC into the UK NHS context. There could be potential barriers and limitations that have perhaps not been identified in the literature due to the novel nature of the field of reverse innovation.

Concept of health intervention adaptation

There were several theoretical frameworks that suggest how to adapt interventions in the health intervention adaptation literature, but none expressed specifically how to adapt an intervention between HIC and LIC contexts (Barrera Jr and Castro 2006, Barrera Jr, Castro et al. 2013). This thesis sampled frameworks from cultural adaptation (Castro, Barrera Jr et al. 2004), adapting for ethnic minorities (Liu, Davidson et al. 2012, Liu, Davidson et al. 2016) and implementation science (Liu, Davidson et al. 2012) to inform a base for the theoretical adaptation framework for the women’s group PLA cycle. This base was then supplemented with data from the key informant interview (Study 1). Measuring whether an intervention has been sufficiently culturally adapted can be problematic because the target population may be different from the members that were used to input on adaptation of the curriculum, materials, graphics, films etc. Healey et al., (2017) questioned the ability to determine whether observed results were pertinent to effective adaptation or whether it was another aspect of an intervention unless the control and intervention groups were identical (Healey, Stager et al. 2017). An additional barrier is the cost of designing and evaluating culturally adapted interventions to determine if they are effective in the wider target population. The Bangladeshi population of Tower Hamlets has high levels of heterogeneity; they may differ in their cultural norms and practices. The qualitative data that was collected during this project cannot be extrapolated to the wider population because of the nature of qualitative research.

Research Design

Ideally, this project should have included piloting of the adapted women’s group PLA cycle intervention. This could have allowed for a practical, rather than hypothetical, assessment of whether the adaptations made to the intervention were feasible in the context. Although a stakeholder panel that included medical professionals and NHS managers were engaged throughout the process, piloting the full PLA cycle was beyond the realms of this project. Presenting the fully adapted model to a stakeholder panel would have given more strength to the study in the context of embedding within the NHS. Certain parts of the adapted intervention did, however, undergo pre-piloting to assess the acceptability and appropriateness of the adapted content, delivery and materials within the Tower Hamlets context. Meetings 2-4 were tested in a pre-pilot phase which brought together women from the Bangladeshi population of Tower Hamlets and were facilitated by a female group facilitator who had been trained by the NEON team. It was recognised during this phase that the female group facilitator would need more training, so that they were able to adequately deliver the intervention without compromising the core components.

Sampling: Key Informants

In Study 1 there were three areas that had the potential to incur bias. First, the key informants were selected purposely: this has the possibility of introducing selection bias into the results generated. Second, although I tried to have equal representation from the entire seven Women’s Group PLA Cycle cluster-RCTs, some trials had a higher representation those others. This was due to my ability to engage potential key informants and to the availability of potential key informants over the data collection period. Finally, the majority of the key informants occupied senior positions within the trial team: this meant that their experience of adapting the model was from a strategic rather than an operational stance. Advising on research strategy (monitoring and evaluation) is important, particularly when there are time constraints, however, discussing the programmatic and operational issues that affected the implementation where the teams may have given a more comprehensive overview of adaptation at site level. Exploring the experiences of the implementation team could have illuminated the specific ontological principles of that process and how they overcame barriers and limitations in context.

Sampling: Focus Group Discussions

A limitation in the FGD sampling was based on the recycling of participants from earlier phases in the NEON study. Most of the female participants were involved in earlier research with the NEON study. This gave them a wider understanding of women’s group PLA cycle intervention, which allowed them to make more specific suggestions. The FGDs were all conducted in English by a facilitator who could speak English and Bengali. The information sheet and the consent form were available in English and Bengali, but there were no participants that spoke solely in Bengali. All the participants had originated from Bangladesh with time in the UK ranging from 9-45 years. This is a limitation because the data does not include the adaptations for individuals who have newly arrived in the UK (3 years) or for individuals who do not speak fluent English.

The challenges of taking a multidisciplinary approach

This research was funded by the Collaboration in Leadership for Allied Health Research and Care (CLAHRC), there for it was conducted to feed into an actual intervention that was going to be physically implemented in Tower Hamlets. This made working within a multidisciplinary team potentially more challenging than normal because there was a tension exploring the theoretical adaptation and thinking long-term about the practical implementation of the adapted intervention. I want to highlight some of the personal challenges working at the intersection of several disciplines including health behaviour, public health, anthropology, psychology, sociology and epidemiology.

The challenges of conducting a project with a multidisciplinary supervisor team

My supervisors had different ideas about how my thesis should be structured and the key themes that I should address. Initially, it took many iterative conversations to align our research priorities. From the outset, I wanted to focus on the process of reverse innovation because I had identified the unique nature of translating an intervention developed in a LIC into a HIC’s health system, but some members of my supervisor team were very focused on the terminology which is still interchangeable in the reverse innovation literature. For this context, my supervisor team consisted of three supervisors from different disciplines:

Clinical Professor of Integrated Community Child Health – Great Ormond Street Institute of Child Health

Principal Research Fellow– Department of Applied Health Research

Senior Lecturer in Epidemiology–Institute for Global Health

It was evident to me that the challenges lay in our understanding of each other’s thematic knowledge and technical expertise. The clinician wanted me to focus on the sub-optimal feeding practices and high-levels of nutrition-related ill-health in the Bangladeshi population, which I later realised that, it could inform the cultural and contextual adaptation of the women’s group PLA cycle intervention. This supervisor was very cautious about allowing me to begin community data collection. The Principal Research Fellow, however, wanted me to focus on qualitative research methodology, which they believed would optimise my research design. They encouraged me to conduct a community mapping exercise community data collection, so I could get insights into the contextual factors that were influencing health behaviours. Unfortunately, I was unable to conduct a community mapping exercise because I had been advised not to go into Tower Hamlets until my second year. My supervisors and I found it challenging to align the aims and objectives of this thesis because we were trying to balance the objectivist clinical and epidemiological views of traditional public health research with some of the subjectivist views of cultural adaptation and reverse innovation that were required to theoretically adapt this intervention. The epidemiologist wanted me to focus on the women’s group PLA cycle trial and encouraged me to go overseas to collect data in India so that I could better understand previous trial contexts. Another challenge occurred, when I wanted to conduct a three-part literature review, each part I believed to be critical to the reverse innovation of the women’s group PLA cycle. I believed that, a literature review on reverse innovation would help me in identifying the challenges faced in terminology (which I had experienced myself). The literature review on health intervention adaptation could give me grounding on the principles of adaptation and also fill in the gaps from the reverse innovation literature around how to adapt and implement an intervention in a new context. The final part of the literature review gave me the theoretical overview of the previous applications of the women’s group PLA cycle intervention and how it had been adapted for previous trial contexts. Some of my supervisors found this a logical pathway, but others did not understand why I needed the three-part literature review, as they believed I just required focusing on the results of the previous cluster-RCTs. This is where I found a tension between those that favoured the objectivist and those that favoured the subjectivist approach to research. The supervisor that supported the subjectivist view accepted my choice to explore context and culture through reverse innovation and the women’s group PLA cycle, and how to overcome any challenges by considering health intervention adaptation, and also encouraged me to explore qualitative methodology. Additionally, I decided to split this thesis into two studies, because I recognised from the literature that, to understand how to adapt the intervention for the new context, I required in determining how and why it had been adapted in previous trial contexts. Initially, there was resistance from the quantitative supervisors because they felt that the literature was sufficient to tell me whether it had been adequately adapted. I also wanted to understand the adaptation process that led to a successful trial. Ultimately, I believed that a multidisciplinary approach would support research rigour and offer a balance between subjectivist and objectivist approaches, because it can bring the best of both paradigms. It just takes a several conversations in the research design phase to align aims and objectives. In the end, I strongly favoured using a qualitative methodology throughout my thesis, because I wanted to capture the nuances around a population’s interaction with context. I would explore the published literature on the cluster RCTs and supplement it with my own qualitative research that addressed previous adaptation and qualities for theoretical future adaptation

Challenges in integrating the concept of reverse innovation and health intervention adaptation in public health.

The core principles of public health are based on policy development, assessment and assurance (America 2019). Arguably, public health is a predominately objectivist field that applies quantitative data collection methods to generate statistics, which can help to monitor the diseases within a population, but in diverse populations it could be challenging to control disease and support wellness if the health services are not tailored for the populations they are serving. Therefore, there could be a need for qualitative information to help adapt health services and tailor them towards the patient’s needs. For example, for policy development it is challenging to base a new policy solely on quantitative data. Data can help you in identifying an area that may need extra resources, but it cannot provide insights into the culture or context that are driving the issue. It is similar for assurance: quantitative data can monitor who is adhering to the policies and can help evaluate effectiveness and accessibility but cannot explain why a policy is not effective. This chapter has attempted to present evidence for supporting the tension existing between the need to adequately theoretically adapt the intervention and compromised adaptation processes due to research restraints (time, funding, personnel and documentation of previous adaptation process). It has tried to demonstrate that, adaptation was occurring, but that it was not always recognised as a formal part of the trial life cycle. I have presented some well-documented challenges, which are related to international translation of interventions particularly around when to involve a diverse target population (Palinkas, Allred et al. 2005), working with local stakeholders {Palinkas, 2005 #832}(Palinkas, Allred et al. 2005) and working towards a common goal (Brownson 2017). Sussman et al. (2008) documented that, the process of translating interventions between contexts can be compromised due to lack of control over the adaptation process (Sussman and Palinkas 2008), but they did not consider that a framework for adaptation which could facilitate this process controlled adaptation. They suggested that, programme developers needed to culturally adapt the intervention (Sussman and Palinkas 2008), but did not consider which components could be adapted and which could not. The purpose of this research was to supply evidence that could potentially support future uptake of the intervention when it is piloted, and this is what led me to attempt to create a theoretical adaptation framework that can be used to assist the NEON study team to adapt and evaluate the intervention for Tower Hamlets. The women’s group PLA cycle had evidence to support that, it was efficacious within a trial setting (Prost, Colbourn et al. 2013), however, it did not have the evidence to support implementation in Tower Hamlets and there was no evidence to optimise its uptake by the Bangladeshi population. The theoretical generic adaptation framework for the reverse innovation of the women’s group PLA cycle could facilitate this process.

Conclusion

As a potential reverse innovation, the women’s group PLA cycle, if adequately adapted, could offer a cost-effective alternative to current health-professional led models, which are working with the Bangladeshi population of Tower Hamlets. Further research by the NEON study team needs to be conducted to test the theoretical adaptation process that I have presented as the intervention’s feasibility within the new context. If the model is to generate cost-savings, it should prove efficacious in a trial setting and effective within a real-time setting. This research has contributed to knowledge and practice in five ways. First, I have explored the process of reverse innovation and the challenges specifically relating to the individuals, institutions and information that could potentially be experienced by the women’s group PLA cycle model. Second, I examined health intervention adaptation and presented examples of how cultural and contextual adaptation has been informed and implemented. This has offered a theoretical understanding of where adaptation to the women’s group PLA cycle could support the reverse innovation process. Third, I have explored the evolution of the women’s group PLA cycle and its previous applications in a bid to understand how it has previously been implemented in the trial contexts. Fourth, I conducted interviews with key informants that were involved in the implementation or monitoring and evaluation of the RCTs. This helped me to further explore adaptation to the model before, during and after implementation. It also helped me to determine the difference between pragmatic and evidence-based adaptation. This further allowed me to form a potential understanding around where and how the model had previously been adapted and use this to inform the topic guide for the FGDs with the Bangladeshi population of Tower Hamlets. Fifth, through an iterative dialogue during these FGDs, I was able to gather information that could assist with the theoretical adaptation of the women’s group PLA cycle model so that it could potentially undergo reverse innovation into this context.

Final reflections

Globalisation and changing populations are putting new demands on existing health systems. In an increasingly diverse UK NHS context, health care planners should consider the needs of these populations and how these can be supported. They could optimise inclusion within the health systems by supporting services that can cater to the needs of a diverse population. This may require alternative approaches to optimise engagement of ethnic minority populations within the health system. Adapting a peer-led participatory intervention that has proven efficacious across seven trial contexts could offer such an alternative to the current health professional-led NHS models. This could encourage uptake and continuous engagement in health services. This thesis has attempted to demonstrate that there are innate complexities in translating an intervention between LICs and HIC. This research has implications for future policy because it suggests that interventions could be adapted from LICs to the UK NHS if they undergo a systematic adaptation process. It also suggests that participatory models in the presence of a UHC may find it harder to recruit and retain the participants. Therefore, the values proposition needs to be carefully curated based on insights from the target population. It highlights the need for consistent and appropriate information generation, analysis and integration within a model, so that it can respond to environmental changes. Finally, it demonstrates that, minority populations in the UK may identify with both their cultural heritage and their lived environment and this has implication for adapting interventions in the future as the populations see themselves more embedded within their environment and their culture was fluid and has the potential to change. My PhD research experience made me capable of reflecting on some of my own biases towards where knowledge emerges. Personally, I would have always considered myself to be open-minded and willing to try new approaches. Upon reflecting on whether or not I would opt for a reverse innovation in place of current NHS interventions, I think I would have had some concerns about the validity of that intervention. After completing the narrative review on Reverse Innovation, I realised that, innovation is a continuum, not only in terms of its evolution, but also in terms of how it travels through the world and changes along the way. This thesis allowed me to understand why it is important to elevate knowledge and expertise from LICs and to think about why we need to decolonise global health if we want to provide stronger and equitable health for all. Moving forward, I would like to use my knowledge and revised outlook to work in partnership with individuals from other contexts so that we can develop innovation that will improve the lives of the most vulnerable. Previously, I would have advocated for health professional-led initiatives because of my experience in public health and my own interactions with the NHS. Working with community facilitators from the Bangladeshi population of Tower Hamlets, I was able to gain insights into how peers could leverage their existing community status to gain the trust of the participants and how they were able to lead groups of men and women in their community. The men and women were also extremely candid about their own issues, often commenting on how current affairs such as Brexit were going to lead to increased racial discrimination towards their community. It made me realise the value of working in partnership with the community and promoting self-efficacy within these communities. It also made me think that a participatory model could potentially be viable in the UK and, if adequately trained; peers could mobilise and deliver specific health information straight to the communities. These communities know what their problems are and may just need some additional support to elevate their confidence and help them integrate within the current health system. I lived in Tower Hamlets on Halcrow Street for one year from 2012 to 2013. I remember the white British market vendors on Whitechapel High Street shouting in Bengali to the Bangladeshi vendors. It seemed like a diverse and prospering community in the midst of a chaotic central London location. Working on this thesis, I was able to build relationships with individuals that, though they lived in the same areas as me during that period, I would probably never even have spoken to, and that has added immense value to my experience. Drinking tea and eating biscuits in the Flower and Dean Community Centre in Tower Hamlets and speaking to mothers, aunties, uncles, fathers, grandmothers and grandfathers, I was able to get a glimpse of what life was like for the Bangladeshi community of Tower Hamlets. I am very grateful for the opportunity and for the lessons that I have learned and will carry with me throughout the rest of my career. I will end this thesis with an Islamic Proverb which is one of the NEON study community facilitators that told me when we were talking about the importance, brilliance and special nature of cross-cultural partnerships: Take a deeper dive into American foreign policy in the Persian Gulf with our additional resources.

References

Alencar, A. and M. Deuze (2017). "News for assimilation or integration? Examining the functions of news in shaping acculturation experiences of immigrants in the Netherlands and Spain." European journal of communication32(2): 151-166.

Asad, A. L. and T. Kay (2015). "Toward a multidimensional understanding of culture for health interventions." Social science & medicine144: 79-87.

Barrera Jr, M. and F. G. Castro (2006). "A heuristic framework for the cultural adaptation of interventions." Clinical Psychology: Science and Practice13(4): 311-316.

Barrera Jr, M., et al. (2013). "Cultural adaptations of behavioral health interventions: A progress report." Journal of Consulting and Clinical Psychology81(2): 196.

Bolam, A., et al. (1998). "The effects of postnatal health education for mothers on infant care and family planning practices in Nepal: a randomised controlled trial." Bmj316(7134): 805-811.

Brownson, R. C. (2017). Dissemination and implementation research in health: translating science to practice, Oxford University Press.

Castro, F. G., et al. (2004). "The cultural adaptation of prevention interventions: resolving tensions between fidelity and fit." Prevention science : the official journal of the Society for Prevention Research5(1): 41-45.

DePasse, J. W. and P. T. Lee (2013). "A model for ‘reverse innovation’in health care." Globalization and health9(1): 1.

Fottrell, E., et al. (2013). "The effect of increased coverage of participatory women's groups on neonatal mortality in Bangladesh: A cluster randomized trial." JAMA Pediatr167(9): 816-825.

Harris, M., et al. (2016). "Does the Country of Origin Matter in Health Care Innovation Diffusion?" Jama315(11): 1103-1104.

Harris, M., et al. (2015). "'They hear "Africa" and they think that there can't be any good services'--perceived context in cross-national learning: a qualitative study of the barriers to Reverse Innovation." Global Health11: 45.

Healey, P., et al. (2017). "Cultural adaptations to augment health and mental health services: a systematic review." BMC health services research17(1): 8.

Hogan, S. J. and L. V. Coote (2014). "Organizational culture, innovation, and performance: A test of Schein's model." Journal of Business Research67(8): 1609-1621.

Houweling, T. A., et al. (2011). "The effect of participatory women's groups on birth outcomes in Bangladesh: does coverage matter? Study protocol for a randomized controlled trial." Trials12(1): 1.

Manandhar, D. S., et al. (2004). "Effect of a participatory intervention with women's groups on birth outcomes in Nepal: cluster-randomised controlled trial." The Lancet364(9438): 970-979.

Michie, S., et al. (2014). "The behaviour change wheel: a guide to designing interventions." Needed: physician leaders: 26.

More, N. S., et al. (2008). "Cluster-randomised controlled trial of community mobilisation in Mumbai slums to improve care during pregnancy, delivery, postpartum and for the newborn." Trials9(1): 7.

Nahar, T., et al. (2012). "Scaling up community mobilisation through women's groups for maternal and neonatal health: experiences from rural Bangladesh." BMC Pregnancy Childbirth12: 5.

Palinkas, L. A., et al. (2005). "Models of research-operational collaboration for behavioral health in space." Aviation, space, and environmental medicine76(6): B52-B60.

Prost, A., et al. (2013). "Women's groups practising participatory learning and action to improve maternal and newborn health in low-resource settings: a systematic review and meta-analysis." The Lancet381(9879): 1736-1746.

Rath, S., et al. (2010). "Explaining the impact of a women's group led community mobilisation intervention on maternal and newborn health outcomes: the Ekjut trial process evaluation." BMC International health and human rights10(1): 25.

Rotheram-Borus, M. J., et al. (2012). "Disruptive innovations for designing and diffusing evidence-based interventions." American Psychologist67(6): 463.

Schein, E. H. (2010). Organizational culture and leadership, John Wiley & Sons.

Sussman, S. and L. A. Palinkas (2008). "Reflections on international translation of evidence-based health behavior programs." Evaluation & the health professions31(3): 323-330.

Syed, S. B., et al. (2013). "Reverse innovation in global health systems: towards global innovation flow." Global Health9: 36.

Tripathy, P., et al. (2011). "Community mobilisation with women's groups facilitated by Accredited Social Health Activists (ASHAs) to improve maternal and newborn health in underserved areas of Jharkhand and Orissa: study protocol for a cluster-randomised controlled trial." Trials12: 182.

Tripathy, P., et al. (2016). "Effect of participatory women's groups facilitated by Accredited Social Health Activists on birth outcomes in rural eastern India: a cluster-randomised controlled trial." Lancet Glob Health4(2): e119-128.