Adapting health promotion interventions to meet the needs of ethnic minority groups: mixed-methods evidence synthesis

Ethnicity

Ethnicity is a complex and contested concept. Definitions of ethnicity vary, for example, across nations, as does the nomenclature that is used. For example, in the USA, ‘race’ and ‘ethnicity’ are often used interchangeably, whereas in much of Europe, ‘migrant’ is the preferred term, this term encompassing a diverse group of people such as students, labourers and asylum seekers. 10 In the UK, ethnicity often encompasses ‘race’, whereas in other countries ethnicity is a concept of cultural affiliation and is therefore distinct from race. Despite these important differences, there is emerging scientific consensus on the preferential usage of the term ‘ethnicity’. A recent analysis of MEDLINE indexed articles, for example, showed that the use of ‘ethnicity’ has now surpassed the use of ‘race’, both in the USA and worldwide. 11 Given this movement and indeed our own substantial contributions to this debate, we preferentially use the term ‘ethnicity’ in this report. 4,12–24

Furthermore, the particular terminology used to describe broader ethnic groups may also differ between countries. In the UK, for example, the label ‘Asian’ is often used to refer to Bangladeshi-, Indian- and Pakistani-origin populations. Conversely, in the USA, ‘Asian’ is more commonly used to describe people of Chinese, Korean and Japanese origin. In other countries, nationality, citizenship, name, country of birth, parents’/grandparents’ country of birth and religion are used either alone or in combination as proxies for ethnicity. 25 A definition of the term ‘ethnicity’ as used in public health16 is found in Box 1 along with the ethnic categories and definitions,16,26 we employed when conducting this work.

Although we have chosen a particular approach to dealing with nomenclature-related issues, we recognise that the use of the term ‘ethnicity’ is contingent on history, politics and geography. Who belongs and who does not and how someone identifies himself or herself or how he or she is identified are context and situation dependent. As public health and its related fields, such as epidemiology and the social sciences, strive to detect and address patterns of disease within populations,16 ‘ethnicity’ (usefully) remains a contested term as an outcome of its location amid the tension between the search for similarities and differences. 27 Consequently, ethnicity and its derivatives, such as ‘ethnic groups’, have, we appreciate, been conceptualised in many ways.

Overwhelmingly, the contemporary literature gives primacy to self-identified ethnicity, and, although this is ethically appealing and likely to best approximate ‘ethnicity’, it is worth bearing in mind that this seemingly personal ‘choice’ is a balance of preference, identification (by others) and (self-)identity, as we rarely classify ourselves in a vacuum. Therefore, in the present study, ‘ethnicity’ is emphasised as an intersubjective activity that is contingent on a range of contextual considerations. Furthermore, conventions and self-identified ethnicity can change over time. For example, in the UK, ‘black and minority ethnic’ is a relatively new term used to include all visible and invisible minorities (including the Irish, for example). 28 However, there have been more recent moves away from using ‘ethnic minority’ and its derivatives altogether, opting instead for terms such as ‘multicultural’, ‘pluralistic’ and ‘diverse’. 28 This can be contrasted with the use of a more ‘racialised ethnicity’ in the USA, where demarcation is more likely to be based on ‘white’ versus ‘non-white’, and ‘people of colour’ is an institutionally acceptable term. 28 Similarly, colour parameters to define populations are used in Canada, where ‘visible minorities’ is a widely employed substitute for ‘race’. 28 The fact that groups of people can be moved in and out of these categories gives credence to the highly contextual nature of ‘ethnicity’; for example, Aboriginal people were once included among Canada’s ‘visible minorities’28 whereas now the term refers only to non-white immigrants.

Using the term ‘ethnicity’ without reference to the context in which it was conceptualised or the method by which the term was assigned is thus problematic and limits its analytical value28 in research. Often related to or synonymous with ethnicity is ‘culture’, which has, to an extent, now replaced ‘race’ to describe sets of seemingly innate and immutable characteristics,29 or is otherwise differentiated from ‘race’ by rather indistinct criteria. 2,9,30 Irrespective of the terminology used, there is furthermore a tendency towards ‘groupism’, in which ‘ethnically, racially and nationally named populations continue to be construed as entities and cast as actors’. 30 Groupism is evident in health-care research given how commonplace it is to compare the health of ethnic minority groups and/or describe their raised incidence or prevalence of a particular disease. 29 This tendency towards groupism, when compounded by an equal propensity to locate ill-health in individuals and their bodies31 and their cultures, suggests that we have not moved far from ‘race’ in the entrenchment of concepts and ideas associated with particular ethnic minority populations.

In this study we hoped to progress the concept of ‘ethnicity’ to one that shifts the focus from defining what an ethnic group is to defining how ethnicity is enacted,30 or, in other words, moving away from characterising ethnic groups as is towards describing ethnicity-mediated effects in different contexts, which requires an active engagement with more process-focused definitions. Furthermore, we sought to encourage thinking on culture away from it being either a liability or an asset, as this view maintains the concept of culture within the framework of immutable traits, akin to previous conceptions of race. Instead, we hoped to situate culture as one dimension alongside, for example, politics and economy, environment and context, to explain observed advantages and disadvantages. Such approaches will, we hope, contribute to advancing ethnicity as a useful analytical concept in health.

In this study we use the term ‘ethnicity’ in a generic sense to encompass a few, some, or all of the following dimensions: ancestry, culture, language, physical features and religion. 12 Figure 1 provides examples for each of the more traditional dimensions of ethnicity and illustrates how these five dimensions of ethnicity overlap to constitute ‘ethnicity’.

FIGURE 1.

Dimensions contributing to the definition of ethnicity.

Health promotion interventions

The goal of health promotion is to enhance positive health and reduce the risk of ill health by encouraging the pursuit of a number of actions;32 these actions are commonly classified as health education or prevention. An inclusive definition of disease prevention was used in this study to encompass strategies that seek to limit the manifestation of disease through risk reduction, but also to manage its consequences and/or improve prognosis when disease is present. 33 Similarly, a broad definition of education can include communication strategies to improve health literacy: knowledge, attitudes and beliefs and skills relating to health. 33 Health promotion interventions are therefore conscious and planned actions towards achieving these goals. 33 Health promotion interventions may encompass initiatives at various levels, ranging from individual-, organisational- and institutional-level initiatives34 to environmental- and policy-based initiatives. Although we were open to any approaches to health promotion interventions, the body of evidence we uncovered focused on individual-centred health education and prevention approaches as it is in relation to these approaches that the evidence is currently best developed. We acknowledge that, in the UK at least, the term ‘health promotion’ has, over recent years, largely been replaced by the term ‘health improvement’ as one of the three main activities in the public health domain.

Adaptation of health promotion interventions for ethnic minority populations

Adapting an intervention for ethnic minority populations should involve an explicit set of methods and processes that can be used to increase its appropriateness or salience. 35 Adaptation can take place at any step along the intervention pathway and is aimed at improving access to, engagement in, participation with and sustainability of an intervention. Adaptation of interventions is, for example, important to improve access to health services or promote health-enhancing behaviours for ethnic minority populations when there are major language barriers. However, a more fundamental rationale for undertaking adaptation is that the body of work underpinning the entire field of health promotion has been undertaken primarily in white European-origin populations, with ethnic minority populations rarely being studied or considered. Aspects of health promotion interventions may not therefore be optimised, suitable or even appropriate for ethnic minority populations. The fact that there exist substantial and persistent inequalities in health outcomes and status represents an additional important argument for the need to adapt interventions.

Adaptation is referred to by a wide variety of other terms, including ‘culturally appropriate’, ‘culturally sensitive’, ‘culturally tailored’, ‘culturally targeted’ or ‘modified’. The degree to which an intervention is determined to be adapted or culturally sensitive is measured by ‘the extent to which ethnic/cultural characteristics, experiences, norms, values, behavioural patterns and beliefs of a target population as well as relevant historical, environmental and social forces are incorporated in the design, delivery, and evaluation of targeted health promotion materials and program’. 35 In this study we opted for the terms ‘adapted’ or ‘modified’, as opposed to using a term prefixed by ‘culturally’, as we have defined culture as being a part of ethnicity, but not its only constituent (see Figure 1); we were thereby seeking to recognise that ethnicity can influence health through aspects other than culture. In addition, terms such as ‘targeted’ and ‘tailored’ are also sometimes used to convey the purpose of increasing appropriateness. Definitions have been proposed to clearly differentiate between these two terms, with ‘targeted’ reserved for adaptations made for a ‘defined population subgroup that takes into account characteristics shared by the subgroup’s members’, and ‘tailored’ being ‘intended to reach one specific person, based on characteristics . . . derived from an individual assessment’. 36 Although we were interested only in targeted interventions, we recognised that there may be unclear and inconsistent usage in terminology and we therefore gave consideration to interventions that may have been ‘tailored’ to an individual. Furthermore, although adaptation usually excludes interventions developed from scratch, for example through community-based participatory research, we included these interventions when there was support from UK guidelines and international systematic reviews for the underlying intervention. Therefore, adaptation can refer to an adaptation of a specific or generic intervention programme. Adapted interventions in this study can refer to a range of interventions as well, from individual to community based.

Justification for focusing on African-, Chinese- and South Asian-origin populations

Our focus on these three broad ethnic groupings was a conscious decision supported by the demographics presented in the preceding section. African-, Chinese- and South Asian-origin people account for the majority of the diversity documented within the UK, and also documented in other economically developed countries – albeit in differing proportions in different countries. Given the differences noted above in terminology used across economically developed countries to describe their ethnic populations, we opted for terminology that would capture a broad range of ethnicities. For example, ‘African origin’ can refer to an Afro-Caribbean population in the UK, an African American population in the USA and also an African population in New Zealand, as different migratory histories have dictated the varying representation of these ethnic groups in their respective countries. For ‘Chinese-origin’ populations the definition is used more or less consistently across countries, describing a population with origins in mainland China or Hong Kong. Sometimes Chinese-origin people are grouped along with other ‘Asian’ populations, although this is more commonly observed in the USA than elsewhere. For this reason we have a distinct ‘South Asian’-origin term, which encompasses people with origins in Bangladesh, India, Pakistan and Sri Lanka (see Box 1).

Chronic disease in economically developed societies

Chronic or non-communicable disease represents the largest disease burden worldwide. These long-term conditions prominently include diabetes, CVD [including ischaemic heart disease (IHD)], respiratory diseases and cancers, and they have come to surpass communicable disease as the leading causes of death. Chronic disease is estimated to account for 35 million deaths per year, representing 60% of all deaths worldwide, while 20% of chronic disease deaths occur in high-income countries. 51 Chronic disease, therefore, presents one of the biggest health challenges in these economically developed nations.

Although the prevalence of and deaths from chronic disease appear overall to be increasing in low-income nations, and in many contexts surpassing the rates of communicable disease such as human immunodeficiency virus/acquired immune deficiency syndrome (HIV/AIDS), tuberculosis and malaria,51 the trend is less consistent in economically developed societies. Sharp decreases in age-standardised mortality rates from IHD from a high in the 1960s, for example, have been documented across countries such as Australia, Canada, New Zealand, the UK and the USA. 52 Within countries, for example in the USA, the age-adjusted incidence of cancer at all sites has also significantly declined in recent years (1999–2006). 53

The numbers of people diagnosed with diabetes has been seen to increase. The USA, for example, has experienced an upward trend from 5.6 to 18.1 million cases from 1980 to 2008,54 with projections of the number of people diagnosed or undiagnosed with diabetes by 2030 being between 37.7 and 47.3 million; however, these estimates differ significantly depending on the model used. 55 In the UK there were 2.6 million people with diagnosed diabetes in 2009, representing an average prevalence of 4%. 56 We know that chronic disease increases the risk of premature death and decreases quality of life. Furthermore, having one chronic disease increases susceptibility to and risk of another, for example CVD is the main cause of mortality and morbidity for people with diagnosed diabetes, accounting for 52% of deaths among people with diagnosed type 2 diabetes. 56 The cost of chronic disease extends beyond health outcomes as it has implications for health-care systems and for families and communities. 51

Overweight and obesity, defined as having a body mass index (BMI) of 25–29.9 kg/m2 and ≥ 30kg/m2 in white European-origin populations, respectively, is inextricably linked with the above-mentioned chronic diseases and will therefore be considered alongside these key long-term conditions. Furthermore, there is debate about whether or not these BMI cut-offs should be lowered for people of South Asian origin, given their increased risk of disease at lower BMI cut-points. 57 The prevalence of adult obesity has been on the rise in many countries, albeit at different rates. In England, the prevalence of obesity is 24.5% in adults aged ≥ 16 years. 58 In the USA, the age-adjusted prevalence of obesity was found to be 33.8% in adults aged ≥ 20 years. 59 Although the prevalence of obesity is lower in Canada than in the USA, obesity prevalence has nonetheless increased on average 10 percentage points over the past 20 years. 60

Ethnic variations in rates of chronic diseases and underlying risk factors

Within economically developed nations with a large proportion of ethnic minority populations, it is recognised that members of some ethnic groups experience an elevated risk of disease in comparison with the general population and relative to other ethnic groups. For example, 10-year CVD61 and type 2 diabetes62 risks were estimated using the QRISK2 and QDScore prediction algorithms, respectively. Both algorithms were developed and validated using routinely collected data from 531 general practices across England and Wales. 61,62 QRISK2 identified risk of CVD among South Asian men and women after adjusting for deprivation, diabetes and family history,61 whereas QDScore found a significant elevated risk of type 2 diabetes by ethnicity after adjusting for factors such as age, BMI, deprivation and family history of diabetes. 62 Significant heterogeneity was observed, with Bangladeshi-origin men and women at the greatest risk for diabetes, with an adjusted hazard ratio of 4.07 [95% confidence interval (CI) 3.24 to 5.11] for women and 4.53 (95% CI 3.67 to 5.59) for men. Pakistani-origin women and men had adjusted hazard ratios of 2.15 (95% CI 1.84 to 2.52) and 2.54 (95% CI 2.20 to 2.83), respectively,62 and both Bangladeshi- and Pakistani-origin men had a significantly higher risk than Indian-origin men. Black African-origin men and Chinese-origin women were also reported to be at increased risk compared with the white reference group. 62 Lower risk was also observed: black African-origin women had an adjusted hazard ratio of 0.81 (95% CI 0.66 to 0.98) and black Caribbean-origin women an adjusted hazard ratio of 0.80 (95% CI 0.70 to 0.92). 62

Different levels of risk are therefore apparent when ethnic groups are examined independently. Identifying or belonging to an ethnic group can be associated with both better and worse health, depending on the outcome considered, and this has been highlighted by health departments and government organisations around the world.

In the UK, the Health survey for England 2004: the health of minority ethnic groups6 (HSE) represents a key document outlining the existing health inequalities for the major health outcomes and their modifiable factors. As the HSE provides survey data, the information is self-reported and actual disease rates may vary. In the USA, the Centers for Disease Control and Prevention (CDCP) Health disparities and inequalities report – United States, 20115 represents a similar effort to capture differences in health outcomes for the main chronic diseases. These documents, along with others, have captured overall trends, including the relative disadvantage of the black population in the USA. For example, black people have been noted to have the highest rate of new cases of all-cause cancer compared with white, American Indian/Alaskan Native, Hispanic and Asian/Pacific Islander populations (1992–2006), despite an observed overall decrease in new cases of cancer. 53,63

In the UK, a different distribution of cancers emerges among ethnic minority populations. Overall, ethnic minority populations are reported to have a lower risk of all-cause cancer, with increased risk for certain types of cancers, for example prostate cancer for black men; liver cancer for Asian men and women; oral and cervical cancer for Asian women aged ≥ 65 years of age; cancers of the stomach, liver and myeloma for both black men and women; and cervical cancer for black women aged ≥ 65 years. 64 Although overall rates of cancer were reported to be lower for ethnic minority populations in England, health promotion and services must acknowledge the cancer types and risk factors that are most relevant to ethnic minority as well as mainstream populations in order to deliver a more equitable health-care system.

In the USA, the age-adjusted mortality rate for coronary heart disease (CHD) was higher for black men and women than for men and women from other ethnic groups (white, Asian/Pacific Islander, American Indian/Alaskan Native). 5 In the UK, a different trend was observed. For any CVD (including IHD/CHD/coronary artery disease and stroke), black African men had the lowest prevalence and Irish men the highest; for women, those from the general population had the highest prevalence and Bangladeshi women had the lowest. 6 For IHD only, South Asian (Bangladeshi-, Indian- and Pakistani-origin) people had higher mortality rates than the general population, and black Caribbean- and black African-origin people had lower mortality rates than the general population. 6 Within the South Asian-origin population, further differences and disparities were observed. 6

For obesity, black females aged ≥ 5 years consistently have had the highest prevalence of obesity in the USA from 1988 to 2008. 5 Furthermore, higher obesity rates persisted across income classes for black females whereas an inverse relationship between obesity rates and family income was observed for white and Mexican populations (2005–8). 5 In the UK, black Caribbean- and Irish-origin men were more likely to be obese and Bangladeshi- and Chinese-origin men were less likely to be obese than men in the general population. Black Caribbean-, black African- and Pakistani-origin women had a higher prevalence of obesity than the general population, whereas Chinese-origin people had a much lower prevalence. 6

In the USA, differences in the prevalence of diabetes between black and Hispanic populations and the white population were observed,5 with black females having the highest age-specific prevalence of diabetes. 65 In the UK, similar health disparities were observed, with diabetes prevalence higher in Bangladeshi- and Pakistani-origin men than in general population men. Similarly, Pakistani-origin women had a higher prevalence of diabetes than general population women. 6

From the above summary of ethnic differences in health outcomes and risk factors it is clear that ethnic health disparities have been documented and compared in many economically developed countries. Documenting and comparing trends represents an initial essential step towards the goal of improving the health of the entire population. In striving to reduce these disparities it is vital to capitalise on the opportunities presented by preventive health approaches to chronic disease, which have the potential to greatly improve health outcomes and reduce health inequalities for present and future generations.

Justification for focusing on smoking cessation, physical activity and healthy eating

Chronic diseases can to a large extent be prevented and once present managed through the modification of shared risk factors, in particular unhealthy diet, physical inactivity, tobacco use and alcohol abuse. 66 These are commonly referred to as individual ‘lifestyle factors’. Controlling for these risk factors, for example, has the potential to eliminate 80% of CVD and type 2 diabetes and 40% of cancers. 51 The potential health gains of a preventive approach provide the justification to focus on smoking cessation, physical activity and healthy eating in our study. Alcohol abuse was not within the scope of this study (see Study scope).

These lifestyle risk factors, however, operate within the complexity of everyday life, in which food choices, living conditions, educational achievement and access to health services51 may not be entirely within the control of the ‘at-risk’ individual. Where an individual or group finds themselves in relation to these social determinants, among other determinants such as income, social status, gender and culture,26 can increase their susceptibility or enhance their resiliency to these modifiable risk factors. One significant factor is ‘ethnicity’, which can, as noted above, encompass dimensions of ancestry, culture, language, physical features and religion (see Figure 1). 12 The broad reach of ethnicity and its intersection with social and contextual factors that can enable or constrain behaviour needs to be taken into consideration when planning health promotion interventions aiming to prevent the onset of or minimise the impact of chronic diseases.

Prioritising preventive health care

A preventive health approach is warranted given that the underlying causes of chronic disease are to a large extent avoidable through the elimination of shared risk factors such as smoking, physical inactivity and unhealthy eating. 66 Consequently, international, national and local responses have been initiated to support a preventive health-care approach. The World Health Organization (WHO), for example, produced the 2008–2013 Action plan for the global strategy for the prevention and control of noncommunicable diseases,66 along with the creation of the NCDnet (global Noncommunicable Disease Network), a voluntary collaboration comprising United Nations agencies, intergovernmental organisations, academic institutions, research centres, non-governmental organisations and the private sector working towards the prevention and control of long-term conditions. 66

At a national level, countries have responded to the WHO agenda and created strategies to combat chronic disease. The CDCP describes chronic disease as the ‘public health challenge of the 21st Century’67 and highlights the huge demand that chronic disease places on the US economy, for example estimated annual expenditures of US$313.8B (2009), US$89.0B (2007) and US$116B (2007) for CVD, cancer and diabetes, respectively. 67 The CDCP actively promotes the preventive approach, with a focus on obesity (physical activity and healthy eating), smoking and alcohol as the most effective and cost-effective methods to tackle chronic disease. 67

Similarly, in England (2002), the direct cost of treating obesity was estimated to be between £45.8M and £49.0M, and between £945M and £1075M for treating the consequences of obesity,68 which included CVD and diabetes. For the Scottish population, the equivalent costs were estimated as £175M to NHS Scotland,69 the majority of the costs being associated with the consequences of obesity, most notably type 2 diabetes (£48M) and hypertension (£38M). Primary prevention is thus rightly a priority strategy for the tackling of chronic disease in the UK,69,70 given that almost a quarter of adults in England aged ≥ 16 years have been classified as obese, with only one-quarter of men and women reporting meeting the government’s ‘5-a-day’ guideline of consuming five or more portions of fruit and vegetables a day. 70 Other economically developed countries, including Canada,71 New Zealand72–74 and Australia,75 are similarly making the prevention of chronic disease a health policy priority. In Canada, an investment of C$300M over 5 years has been made towards disease prevention. 71 Equally, in European countries, recognition of the health and economic burden of chronic disease and the importance of a co-ordinated preventive health approach76 has led to the establishment of the European Chronic Disease Alliance77 to influence the policies of European member states towards a unified prevention approach to chronic disease. A corresponding alliance was formed in Australia to monitor and advocate for a national preventive health strategy. 78

The majority of these strategies, in reflection of the broader literature, also recognise the contribution of chronic disease to inequalities in health. 77,79–83 The CDCP, for example, underscores how chronic disease disproportionately affects ethnic minority populations and has called for equity in the provision of health promotion programmes that are accessible and target the social determinants of health. 67 Creating effective, appropriate preventive health/health promotion programmes for ethnic minority groups is thus now seen as a key strategy in reducing ethnic disparities in health.

Evidence of effective preventive health care for ethnic minority populations

The preventive health approach includes health promotion campaigns, interventions and services to decrease tobacco use, increase physical activity and improve diet to reduce and control chronic diseases and the approaches adopted are often steered by international guidelines and national health policies. General recommendations for type, dose and duration of physical activity to improve health are usually made without specific recommendations for gender, ethnicity or income group. Evidence-based UK guidelines, for example, include four methods to increase physical activity for the population,79 although a new framework has been established to drive local physical activity activities84 (see Chapter 4). Despite these strategies there is low adherence to recommended levels of physical activity for health85 in most ethnic minority groups compared with adherence in the general population, except for the black Caribbean-origin population. 6 Low prevalence of physical activity has most strikingly been noted in Bangladeshi- and Pakistani-origin women compared with the general population. 6 Given that a higher prevalence of obesity and diabetes clusters within these South Asian-origin groups, a more concerted effort is required to determine the differential needs of, and identify approaches to increasing physical activity in, these population groups. 85

A review of the effectiveness of interventions to promote healthy eating in people from ethnic minority groups80 found limited research in the UK; the majority of evidence came from the USA. A more recent review of the healthy eating of UK South Asians81 reported that there had been an increase in research that could help develop effective interventions, but with little attempt to synthesise this evidence for use by health practitioners. The Welsh Assembly Government undertook a review of the health promotion evidence base for ethnic minority groups and, regarding healthy eating, concluded that the messages may not be appropriate or understood by a high percentage of the population. 82 Recent evidence-based UK guidelines on preventing type 2 diabetes, including increasing physical activity and improving healthy eating in high-risk groups, have demonstrated some consideration for ethnic minority populations. 83

In the USA, a main driving force directing preventive strategies is the Healthy people 201086 initiative, now superseded by Healthy people 2020. 87 Although emphasising that additional effort is needed to reach subgroups according to age, gender, socioeconomic position and ethnicity, there is also acknowledgement that the evidence is lacking to inform more definitive recommendations for what works in these subpopulations. Federal initiatives such as a National Tobacco Quitline network and public–private initiatives such as insurance coverage for tobacco treatment have been proposed to promote smoking cessation,88,89 while evidence-based guidance provides key recommendations for the treatment of tobacco dependence. 90 There remain, however, health disparities in tobacco-related health outcomes, as indicated by the differential prevalence of current smokers. 86 Although prevalence is comparable between white (non-Hispanic) and black (non-Hispanic) populations, prevalence is low among the Asian (non-Hispanic) population. 91 For those reporting multiple ‘races’, however, the prevalence is higher than in the white (non-Hispanic) population. 91 Beyond disparities in prevalence, there are other pronounced differences in usage, environmental contact, commencement, treatment and health outcomes related to tobacco use. 92

In the UK, the NHS Stop Smoking Service is the primary vehicle for delivering cessation services. There is, however, documented evidence of lower use of the Stop Smoking Service by ethnic minority groups, which may be a consequence of these services not being linguistically or culturally appropriate or sensitive. 93 Guidance that highlights best practice in tobacco prevention and cessation for ethnic minority communities can help to support Stop Smoking Service managers working at the front line within their catchments, as it is commonly believed that the services need to be made more accessible, rather than tailored per se. 93,94 These additional efforts are needed to address smoking health disparities, particularly among Bangladeshi-origin men who have the highest prevalence of cigarette smoking in England, followed by white Irish-origin men and women and Pakistani-origin men and black Caribbean-origin women. 6 All of these groups have higher smoking prevalence than the general population. 6 Among Bangladeshi-origin women, chewing tobacco is much more prevalent than cigarette smoking6 and potentially represents a gap in the current services as smokeless tobacco falls outside the official scope of stop smoking programmes. Therefore, in the case of smokeless tobacco cessation, targeted services may be appropriate based on a local needs assessment, as there may be potential difficulties with access. 93

When determining whether adapted mainstream services or specially developed, culturally sensitive services would be most appropriate, needs assessments are essential to draw out local knowledge and attitudes towards health, as well as draw on local assets and resources. In particular, how ethnicity relates to eating practices, physical activity and smoking is contingent on many inter-related factors such as age, gender, culture, education, income, family structure and immigration status, in the context of the local environment. Insight into these issues and its application to preventive services is likely to increase successful uptake and implementation of health promotion activities and strategies. 95

Furthermore, even if services were believed to be culturally sensitive, the reasons for their underuse are multifaceted and may include aspects such as location and available hours of services; appropriateness to language, gender, culture and religious norms; and preferences for and linkages with other community services and institutions. Overall, research suggests that, if services demonstrated appreciation for the importance of diversity without reverting to stereotypes, provided respectful and supportive staff, increased accessibility to information on services and created a safe environment without discrimination, they were reported as being acceptable by ethnic minority people. 96

Study scope

The field of ethnicity and health is broad and far-reaching, as has been illustrated by the above introductory note. In the context of this report, we will henceforth be focusing on the three main populations of interest to this study, that is, African-, Chinese- and South Asian-origin populations, along with the three areas of health promotion important for the prevention and/or reduction of chronic diseases, namely, smoking cessation, increasing physical activity and improving healthy eating. Literature on other ethnic populations, such as the Hispanic population in the USA and the indigenous populations in the USA, Canada, New Zealand and Australia, although important and potentially very useful in informing learning about adapting interventions, was not within the remit of this particular piece of work. Furthermore, as noted above, although alcohol is another modifiable risk factor salient for chronic disease prevention and control, it was purposefully excluded, as there is a lack of consensus across countries and communities as to what constitutes an acceptable/unacceptable alcohol intake and how this differs from high-risk alcohol use. On the other hand, the recommendations for smoking, physical activity and healthy eating and their associated health outcomes are relatively clear-cut and agreed upon.

Furthermore, the primary purpose of this report and its research components is to support a conceptual and methodological study of the issue of adapting interventions to meet the needs of ethnic minority populations. This purpose is in direct response to an Health Technology Assessment (HTA) commissioning brief (see Appendix 1) that sought research on how to adapt interventions to meet the needs of ethnic minority groups. The brief works with the assumption that adaptation should be pursued as the best way forward to reduce the observed inequities for some health conditions among particular ethnic minority populations. 9

Phase 1 objectives

1. A. To identify and summarise information on health promotion interventions for smoking cessation, physical activity and healthy eating that are of proven effectiveness for use in populations at large.

2. B. To assess to what degree ethnic minority populations are considered within the evidence base for smoking cessation, physical activity and healthy eating health promotion interventions known to be effective in populations at large.

3. C. To identify health promotion interventions for smoking cessation, increasing physical activity, and improving healthy eating that have been adapted to meet the needs of African-, Chinese- and South Asian-origin ethnic minority populations, describe the adaptation approaches used and assess the clinical effectiveness, cost-effectiveness, feasibility, acceptability and equity of these adapted approaches.

Phase 2 objective

1. D. To understand the rationale for and processes and experiences of adapting health promotion interventions for ethnic minority populations and to summarise lessons learned.

Phase 3 objective

1. E. To identify which adapted interventions are ready for implementation and establish which interventions require further research.

The research questions arising from these objectives are discussed in subsequent chapters (see Chapters 4–8). Figure 2 outlines the relationship between the three phases, five objectives (A–E) and 10 questions and maps them onto the chapter(s) in which they are discussed in more detail.

FIGURE 2.

Relationship between project phases, objectives and questions.

The original scientific rationale for this study is included in Appendix 2. The relatively few amendments made to aspects of work outlined in the original scientific rationale are detailed in Appendix 3. These in the main reflected our responses to the unexpectedly large volume of literature identified.

User engagement

User engagement was undertaken throughout our project through the inclusion of lay members on our Independent Project Steering Committee; in addition, we held two user conferences, one to launch the study and one to share and discuss our preliminary findings. The details of our user engagement approach and conferences are found in Chapter 3.

UK guidelines and international systematic reviews

Our review of UK guidelines and international systematic reviews was designed to examine what health promotion interventions have proven effectiveness in the general population for smoking cessation, increasing physical activity and improving healthy eating. Details of effectiveness for the general population were extracted from guidelines and systematic reviews and a summary framework of effective interventions was created for behaviour change in general and for each of the health promotion topics of interest (see Chapter 4). These frameworks were further compared with the adapted interventions to identify where there was overlap and to identify research gaps in which promising evidence-based interventions for the general population have not yet been adapted for ethnic minority populations (see Chapter 8).

This same body of literature was also searched to determine whether or not recommendations and interventions made by UK guidelines and international systematic reviews were equally applicable to ethnic minority populations (see Chapter 5).

Systematic review of adapted interventions

We then conducted a systematic review of health promotion interventions to identify smoking cessation, physical activity and healthy eating interventions that have been adapted for African-, Chinese- and South Asian-origin populations. From the included studies we identified the adaptation approaches utilised in these interventions (see Chapter 6). The included studies were also assessed for clinical effectiveness and cost-effectiveness in relation to health behaviour change and associated outcomes.

Qualitative interviews

In parallel with the systematic review we carried out qualitative interviews with key researchers and health promoters who had been involved in developing, delivering and evaluating adapted health promotion interventions to supplement the systematic review of published literature. In-depth semi-structured interviews were conducted either by telephone or face-to-face and were recorded and transcribed. Transcripts were coded and thematically analysed to identify overarching themes (see Chapter 7).

Realist synthesis

Finally, we synthesised the findings from the above phases of work – that is, the user engagement conferences, the review of UK guidelines and international systematic reviews, the systematic review of adapted interventions and qualitative interviews with researchers and health promoters – using the explanatory framework offered by realistic evaluation102 to understand better how adapted interventions work and to identify interventions ready for implementation and prioritise future research (see Chapter 8).

Ethics

Ethical considerations were outlined in the original scientific rationale (see Appendix 2) and an application was submitted to the School of Health in Social Science Research Ethics Committee for the qualitative interviews (see Appendix 4).

Launch conference

This conference was held in London, England in June 2009. Speakers included members of the project team and international experts in ethnicity and health research from the UK, the USA and Europe. A total of 81 delegates attended the conference from a wide variety of stakeholder organisations. Interactive workshops allowed smaller group discussion of topics pertinent to the research methods, including approaches to multiple behaviour change; the role of culture/acculturation in interventions for ethnic minority communities; reflecting on community-adapted programmes; and adaptation in non-health contexts. The programme for this conference is available in Appendix 5. This conference was reported in the British Medical Journal (www.bmj.com/content/338/bmj.b2326.extract).

Dissemination conference

Our second user engagement event was a project dissemination conference held in Edinburgh, Scotland in February 2011. This event was held towards the end of the project and provided us with the opportunity to share, discuss and debate our preliminary conclusions, and refine these in the light of participant feedback. A total of 71 delegates attended the conference from a wide variety of stakeholder organisations – some of the participants were attendees at the first conference. In addition to the presentation of project findings, interactive workshops allowed smaller group discussion and feedback on the project outputs including methodological reflections on complex systematic reviews; theoretical considerations on the principles of adaptation; translation of study research findings into practice; and identifying key future research themes. The programme for the conference can be found in Appendix 7.

Background

A key assumption underpinning this project is that health promotion interventions found to be effective in the general population are, if appropriately adapted, likely to also prove effective in ethnic minority populations. We therefore began our work by seeking evidence-based recommendations for health promotion interventions found to be effective in the general population.

Objectives

We sought to identify and summarise information on health promotion interventions for smoking cessation, increasing physical activity and improving healthy eating that are recommended for use in populations at large.

Methods

We undertook a systematic review of the evidence for health promotion interventions for smoking cessation, increasing physical activity and improving healthy eating in UK guidelines and these searches were supplemented by additional scrutiny of international systematic reviews. Relevant guidelines were identified from the Clinical Evidence, National Institute for Health and Clinical Excellence (NICE) and Scottish Intercollegiate Guidelines Network (SIGN) databases. We also searched for international systematic reviews from established databases of evidence, namely The Campbell Collaboration, The Cochrane Library, the Database of Abstracts of Reviews of Effects (DARE) and the National Institute for Health Research Health Technology Assessment (NIHR HTA) databases, in order to identify additional and more recent evidence that had not yet been incorporated into the guidelines. Two reviewers independently searched for and identified the literature and extracted data on recommended interventions for the general population. We created summary tables of the evidence identified according to health promotion topic and also a table for behaviour change interventions in general (not specific to any topic). These tables were then further categorised by six Tannahill-based intervention ‘strategies and activities’ (i.e. environment, policy, setting based, techniques, provider and resource).

Results

We identified 15 relevant guidelines and 111 systematic reviews. Most of the evidence we identified was targeted at individuals. There were considerably stronger recommendations for smoking cessation than for interventions aiming to increase physical activity and/or improve healthy eating. The strongest evidence and guidance in relation to smoking cessation was for pharmacological treatments and approaches aiming to promote the use of these pharmacological therapies. In contrast, there was relatively little in the way of strong or consistent guidance in relation to interventions that were successful for promoting changes in diet and exercise patterns that were sustained over time, or those directed at societal level-mediated change.

Conclusions

We have identified a number of guidelines and systematic reviews that have summarised and sought to provide guidance on how best to promote smoking cessation, increase physical activity and improve healthy eating. This evidence is strongest and most consistent in relation to pharmacological treatments for smokers and approaches aimed at increasing use of these products. Based on this work we suggest the need for improved communication between guideline developers and research commissioners with a view to catalysing research into finding effective and sustainable interventions for promoting physical activity and improving healthy eating. There is, in particular, a need for more evidence on the providers and resources that can meaningfully increase physical activity and improve healthy eating, and, more generally, for policy- and environmental-based interventions for smoking cessation, increasing physical activity and improving healthy eating.

Overview of methods

The UK has some of the most well-established and respected guideline development procedures in the world. These guidelines therefore represented the principal source of evidence that we sought to identify and scrutinise. Because of their lengthy and rigorous development processes, guidelines, however, inevitably remain a step behind the evidence and are thus less likely to contain as up-to-date information as is contained within the systematic review literature on which they predominantly draw. We therefore supplemented our guideline-based searches with data derived from international systematic reviews of evidence.

Searches for guidelines and systematic reviews

We searched for and retrieved any pertinent guidelines on interventions for smoking cessation, increasing physical activity and improving healthy eating in the Clinical Evidence, NICE and SIGN databases using the following broad key search terms:

• diet

• exercise

• healthy eating

• obesity

• physical activity

• smoking

• tobacco.

We also conducted searches in The Campbell Collaboration (Campbell), The Cochrane Library (Cochrane), the DARE databases and the NIHR HTA database for systematic reviews and health technology assessments (HTAs) of interventions for smoking cessation, increasing physical activity and improving healthy eating from January 1950 to April 2009 using the same broad search terms detailed above. HTA reports use systematic methods110 to provide independent research information about the effectiveness, costs and broader impact of health-care treatments and tests for those who plan, provide or receive care in the NHS and henceforth in this report will be referred to under systematic reviews.

Selection of eligible guidelines and systematic reviews

Two researchers (JJL and ED) independently screened and selected guidelines and systematic reviews based on predefined inclusion/exclusion criteria (Table 2), first by scrutinising titles and then, if necessary, the abstracts and full text of manuscripts. Disagreements regarding inclusion/exclusion of papers were resolved through discussion. If consensus could not be reached a third reviewer (AS) was consulted. The full text of all eligible records was retrieved; 12 full-text papers were translated into English from Danish, French, Norwegian, Spanish and Swedish.

Data extraction and synthesis

We extracted relevant evidence from the guidelines and systematic reviews onto a customised data extraction form (see Appendix 9). We extracted statements pertaining to the effectiveness of smoking cessation, increasing physical activity and improving healthy eating interventions, the quality of the evidence (when reported) and the quality assessment tool used (when reported). These evidence statements were compiled in two tables (see Appendix 10) where they remain attributed to the guideline(s) or systematic review(s) from which they were derived. The tables of evidence statements therefore incorporated two levels of evidence – guidelines that have been developed through a collaborative scrutiny of the evidence, as in the case of NICE and SIGN guidelines, and systematic reviews, which have been subject to quality control by peer reviewers and journal editors.

Subsequently, the evidence statements were organised in a summary framework of effective interventions according to the health promotion topic: health behaviour in general (not specific to one health promotion topic), smoking cessation, increasing physical activity and improving healthy eating (see Tables 3–6). We adapted the revised Tannahill model of health promotion (Box 2)111 to categorise the evidence statements according to six broad categories of ‘strategies and activities’ (Box 3).

Environment applied to interventions that alter the social, economic and physical space (as opposed to place), with the potential to affect a more general population. Policy included organisational, local or national policies. Setting based represented specific places or communities identified as effective locations to deliver an intervention, and which often target ‘captive audiences’, such as children at schools or employees at worksites. Techniques described the wide-ranging types and formats of interventions that have been found to be effective, including education and learning, but it also captured behavioural techniques and other elements that can enhance the effectiveness of a variety of interventions. Provider also included persons singled out as important partners in the intervention delivery process. Resource encompassed the materials or products given to participants as the intervention. Unlike in the Tannahill model, provider represented a specific type of resource, referring to the specialised personnel or services designated for intervention delivery.

These categories referred to the ‘strategies and activities’ utilised in health promotion interventions (not to the interventions themselves) and are overlapping and additive in that combinations of ‘strategies and activities’ may be used in any one intervention. These categories facilitated the examination, and cross-comparison, of such a large and diverse body of evidence for smoking cessation, increasing physical activity and improving healthy eating.

Summary framework of effective ‘strategies and activities’

The evidence statements extracted from these included guidelines and systematic reviews were compiled in tables (see Appendix 10) and then synthesised into our summary framework according to health topic and the six categories of health promotion ‘strategies and activities’ discussed above. The following four tables represent the summary framework to guide behaviour change in general (see Table 3), smoking cessation (see Table 4), increasing physical activity (see Table 5) and improving healthy eating (see Table 6). Each table will be described in turn.

Strengths and limitations

The four summary tables (see Tables 3–6) succinctly summarise the current evidence available for effective ‘strategies and activities’ for behaviour change in general, smoking cessation, increasing physical activity and improving healthy eating. In the systematic reviews we searched for compelling evidence of effectiveness that had not yet found its way into the guideline literature; in contrast, we deliberately excluded interventions that have not yet accumulated a sufficient body of evidence of effect. Thus, for example, we excluded a review of nicotine vaccines239 and cytisine240 for smoking cessation, a review of eHealth interventions for physical activity and/or diet241 and a review of clinical counselling for increasing physical activity. 242 The summary tables have therefore captured the most up-to-date and relevant evidence for smoking, physical activity and healthy eating health promotion interventions.

Because of the large volume of evidence identified, and the fact that this was one part of a much larger mixed-methods study (see Chapter 2), we were unable independently to quality appraise each guideline and systematic review and instead relied on the quality control measures in place for producing and publishing guidelines and systematic reviews. When reported we extracted whether or not the quality of the evidence and the recommendations was appraised and the appraisal tool used (see Appendix 10). To an extent this approach represented a limitation. In the case of contradictory evidence statements (e.g. if an intervention was reported as effective in one systematic review and not effective in another), we were unable to investigate the reasons underpinning this difference in evidence. In such cases we opted to describe both evidence statements.

Our review of guidelines was limited to UK-based bodies, which is a further potential limitation. The above described ‘strategies and activities’ thus provide a detailed and comprehensive overview of the UK recommendations; however, as the UK guidelines draw on international evidence and because, in addition, we identified international systematic reviews, the state of the evidence described should also on the whole reflect the international evidence base in relation to these three areas of enquiry. In the absence of also interrogating guidelines from other countries (which we were unable to do because of resource constraints), however, we cannot be sure.

The definition of health promotion adopted for this study111 was adapted to produce the six categories of ‘activities and strategies’; this has a potential limitation in that the categories are overlapping and not mutually exclusive. In addition, not all interventions could easily be assigned directly to a level of health promotion intervention. For example, policies can be implemented at the organisational or population level and providers could deliver interventions individually or in groups. In support of this approach these categories are placed across a spectrum, from individual- to population-level ‘strategies and activities’, which loosely corresponds to their scope for adaptation. Despite these potential limitations it was overall still helpful to have an organising framework to describe these strategies and actions to allow the examination, and cross-comparison, of such a large and diverse body of evidence.

Background

A number of interventions have been developed to aid smoking cessation, promote physical activity and improve healthy eating for use in the general population. The effectiveness of these interventions in improving health outcomes for ethnic minority populations is, however, unclear.

Objectives

We sought to assess to what degree ethnic minority populations are considered within the evidence base for smoking cessation, physical activity and healthy eating health promotion interventions known to be effective in populations at large.

Methods

Two reviewers independently searched UK guidelines and international systematic reviews of interventions for smoking cessation, increasing physical activity and improving healthy eating. Guidelines were identified from the Clinical Evidence, National Institute for Health and Clinical Excellence (NICE) and Scottish Intercollegiate Guidelines Network (SIGN) databases. Systematic reviews were identified from The Campbell Collaboration, The Cochrane Library, the Database of Abstracts of Reviews of Effects (DARE) and the National Institute for Health Research Health Technology Assessment (NIHR HTA) database. We searched the guidelines and systematic reviews using predefined terms relating to ethnicity. Two reviewers independently extracted the statements made in relation to ethnicity, initially grouping these according to shared meaning and/or concept; these data were then thematically analysed.

Results

We identified 15 guidelines and 111 systematic reviews that fulfilled our eligibility criteria. A total of 12 (87%) guidelines and 66 (59%) systematic reviews had one or more terms relating to ethnicity. All 12 of these guidelines containing key terms made statements relating to ethnicity; however, only 41 of the 66 systematic reviews with key terms made any informative statements relating to ethnicity and none of the systematic reviews conducted subgroup analyses by ethnic group. Five main themes emerged from the thematic analysis of statements: (1) acknowledging diversity; (2) identifying evidence gaps; (3) observing differential effects of interventions; (4) taking action to adapt interventions; and (5) improving research, reporting and analysis.

Conclusions

UK guidelines and international systematic reviews provided little specific guidance with regard to which health promotion interventions are effective (or ineffective) for ethnic minority populations or how interventions with proven effectiveness in the general population can be adapted to be effective for ethnic minority populations. There is a need to consider more explicitly the evidence in relation to ethnic minority populations in the context of guidelines and systematic reviews of key health promotion interventions.

Themes relating to ethnicity in UK guidelines and international systematic reviews

The evidence statements from the 12 guidelines were grouped under 16 generic statements (Table 8). From the generic statements, five themes were identified and these are summarised in Box 4. No new themes emerged from the evidence statements extracted from the systematic reviews that were not already encompassed under the five themes. Table 8 therefore represents these two sources of evidence.

Strengths and limitations

A key strength was our systematic approach to assessing what consideration has been taken of ethnic minority populations in the high-level evidence and recommendations for health promotion interventions for smoking cessation, increasing physical activity and improving healthy eating. To our knowledge, this is the first time such an endeavour has ever been undertaken with the inclusion of international systematic review.

The 2 years between the literature search (April 2009) and the production of this report (June 2011) mean that additional guidelines and systematic reviews are likely to have been published or updated during this time and this is therefore a potentially important limitation. For example, the two SIGN guidelines (SIGN 8236 and 69200) that have been included in this study have since been superseded by a newer guideline (SIGN 115243). Similarly, other reviews, such as the recently published interventions to reduce pre-diabetes risk factors for ethnic minority populations,244 have been used by NICE to develop Public Health Guidelines. An examination of these recent reviews has not, however, revealed any definitive evidence on how to improve the health of ethnic minority populations that has not already been included in systematic reviews identified in this study.

Background

Health promotion interventions are potentially important strategies to reduce the morbidity and mortality associated with smoking, physical inactivity and poor diet. Adaptation of interventions with proven effectiveness in the general population may help to increase their acceptability to and effectiveness in ethnic minority populations.

Objectives

We sought to identify health promotion interventions for smoking cessation, increasing physical activity, and improving healthy eating that have been adapted to meet the needs of African-, Chinese- and South Asian-origin ethnic minority populations; describe the adaptation approaches used; and assess the clinical effectiveness, cost-effectiveness, feasibility, acceptability and equity of these adapted approaches.

Methods

We systematically searched Applied Social Sciences Index and Abstracts (ASSIA), BIOSIS, The Campbell Collaboration, Cumulative Index to Nursing and Allied Health Literature (CINAHL), The Cochrane Library, EMBASE, ISI Web of Science, Latin American and Caribbean Health Sciences Literature (LILACS), MEDLINE, PsycINFO and NHS Evidence Specialist Collection for Ethnicity and Health (SCEH) to identify relevant conceptual work, systematic reviews and empirical investigations of adapted interventions. Study quality was independently assessed by two reviewers using the Critical Appraisal Skills Program (CASP) and Effective Public Health Practice Project (EPHPP) tools. Study findings were descriptively summarised and then thematically synthesised.

Results

Our searches identified a total of 48,740 potentially eligible studies, of which 173 reports satisfied our inclusion criteria. We identified 12 theoretical papers, seven systematic reviews and 107 empirical investigations (reported in 154 papers). The majority of adapted intervention studies took place in the USA and were conducted with African American populations, predominantly women. All studies conducted with Chinese-origin populations took place in the USA while studies with South Asian-origin populations were conducted in Australia, New Zealand and the UK. Multicomponent interventions targeting physical activity and healthy eating were the most common focus of attention, followed by smoking cessation interventions. We were able to describe and classify a spectrum of approaches used to adapt these interventions for ethnic minority populations. Of the 107 experimental studies identified, only nine directly compared an adapted intervention with a standard approach. One of these nine studies found evidence in support of the adapted intervention; five studies reported no differences; one study was in favour of the standard approach; and the remaining two studies reported no difference for some outcomes with other outcomes favouring the standard intervention. No studies reported a cost-effectiveness analysis. Through studying the adapted approaches we were able to describe the range of methods employed and use this to construct a 46-item Typology of Adaptation. These adaptation approaches were then mapped on to different stages of the research cycle to produce an eight-stage Programme Theory of Adapted Health Promotion Interventions.

Conclusions

We identified a large body of studies predominantly from the USA examining adapted interventions in African American populations, but relatively little evidence in relation to other ethnic minority populations. This body of evidence has been used to develop a novel 46-item Typology of Adaptation and eight-stage Programme Theory of Adapted Health Promotion Interventions, which will, we hope, assist researchers in more systematically reflecting on approaches that can be employed to adapt interventions and also support the improved reporting of these studies of adapted interventions. There is at present a small inconclusive body of evidence examining the effectiveness of adapted compared with standard intervention approaches. Further studies of well-theorised adapted compared with standard interventions are needed and it is important that these trials assess long-term outcomes and cost-effectiveness considerations.

Search strategy

We searched the following 11 databases for literature on health promotion interventions for smoking cessation, increasing physical activity and improving healthy eating for African-, Chinese- and South Asian-origin populations: ASSIA, BIOSIS, The Campbell Collaboration, CINAHL, Cochrane Central Register of Controlled Trials, published in The Cochrane Library, EMBASE, ISI Web of Science, LILACS, MEDLINE, PsycINFO and NHS Evidence SCEH.

We used the search terms and strategy described in Appendix 12; a more detailed specimen search strategy for the MEDLINE database is provided in Appendix 13. The terms and search strategies used were adapted accordingly for each database. There were no restrictions placed on the language of publication. The searches were carried out between March and June 2009 and included literature dating from 1950.

A grey literature search was conducted using Google, Google Scholar and NHS Evidence (the first five pages), along with ClinicialTrials.gov and current controlled trials (www.controlled-trials.com) for ongoing trials. Additional grey literature was referred to us by the experts who we contacted for our qualitative interviews.

Inclusion and exclusion criteria

Summarised below are the review inclusion and exclusion criteria in relation to the populations, interventions, outcomes and types of studies.

Data extraction strategy

Two reviewers (JJL and ED) independently screened all titles and abstracts identified by the search strategy to assess inclusion/exclusion. Full papers of potentially eligible studies were retrieved and assessed by the same two reviewers based on the above criteria. Any differences of opinion between the two reviewers were resolved by discussion or, when necessary, by a third reviewer (AS). Reasons for exclusion were documented and the list of excluded studies is located in Appendix 14.

Two reviewers (either JJL/ED for physical activity and healthy eating studies or JJL/CW for smoking cessation studies) independently extracted data using a tailored data extraction form (see Appendix 15). Data were extracted on:

• country of study and study setting (urban, rural, community, school)

• study design, inclusion/exclusion criteria

• participant demographics [sample number, age, gender, socioeconomic status (SES)]

• ethnicity and how ethnicity was assessed

• description of intervention – level of intervention (population, group, individual), type of intervention (focus – health risk/disease, duration, setting), selection criteria, sampling, randomisation, characteristics of comparison groups or control groups (as relevant to study design)

• type of adaptation(s) used

• outcome measures (these varied considerably according to the study design and focus)

• analytical techniques (where relevant)

• outcomes

• relevant process evaluations

• conclusions.

The extracted data were summarised in data tables and checked for discrepancies by a third reviewer (SK), and then further checked and edited by the original reviewers (JJL and ED). Discrepancies were resolved through discussion.

Quality assessment strategy

Two authors (either JJL/ED or JJL/CW or ED/SK) independently assessed the methodological quality of included studies using a suite of appropriate assessment tools appropriate to the study design. As recommended by the Cochrane Public Health Group in their Guidelines for Systematic Reviews of Health Promotion and Public Health Interventions,247 we used the CASP checklist for reviews to assess the quality of adapted systematic reviews;248 the quality appraisal tool developed by the EPHPP to assess intervention and observational studies;249 and the CASP checklist for qualitative research for qualitative studies. 250 Disagreements were resolved by discussion and, if necessary, with the involvement of a third reviewer (AS).

Methods of analysis and synthesis

Both quantitative and qualitative data were extracted to assess how interventions were designed and adapted, and to assess feasibility, acceptability, clinical effectiveness, cost-effectiveness and equity.

Because of the wide range of health promotion topics, populations, interventions, time frames and outcome measures studied, it was neither desirable nor possible to undertake meta-analyses. Instead, the findings of each study were descriptively summarised and then this body of work was thematically synthesised.

Given the focus of this work, specific attention was given to extracting the behaviour change theories and adaptations employed in the interventions. When studies reported using a specific theory, this information was extracted; however, many papers did not report the use of a theory or, when they did report this, it was unclear how theory had informed the study design, intervention content or study outcomes.

Summary of the literature identified

A total of 48,740 records were identified. A breakdown of the number of records identified from each of the 11 databases is provided in Table 9. The bulk of the records was identified from MEDLINE and EMBASE.

After deduplication, 36,047 records remained. Titles were screened and 35,422 records were subsequently excluded as not relevant. Abstracts and full-text articles were obtained and assessed for the remaining 625 records. A further 452 records, after being assessed for eligibility according to our inclusion/exclusion criteria, were subsequently excluded as they were not relevant to the interventions, populations, health promotion topics or outcomes of interest, or they were not deemed to contain sufficient information on adaptation. A total of 12 conceptual/theoretical papers, 7 systematic reviews and 154 reports of 107 empirical investigations were included in the final review (Figure 8).

FIGURE 8.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram to identify literature on adapted health promotion interventions.

Summary of conceptual/theoretical papers

It is suggested that theory-based interventions are in general more effective than interventions that are not underpinned by theory. 251 ‘Theories’ are broadly defined here as hypotheses, models or frameworks to help clarify thinking about situated realities. 252 A theory can help explain and/or predict outcomes by explicating the relationships between variables posited to be important,251 for example why some people do or do not make health behavioural changes. A theory is characterised by abstraction rather than specificity for any context and should allow for testing in different contexts. 251 A theory can be broadly categorised as either explanatory- or change-oriented, posited to either understand or model how change takes place, although these are not mutually exclusive. 251

Theory in public health interventions has been usefully classified along a continuum, that is, studies that are informed by theory, by applying a theory, by testing theory and by building/creating theory. 253 The use of ‘theory’, however, has been variable, with few studies applying or testing theories,253 which represents the foundation work before theory building can occur, and this thus reveals a gap in the field. 251 Although theories for behaviour change dominate the health promotion literature, our focus here was more directed as we gave consideration to the ‘theoretical’ literature in the sense that the studies discussed a model, framework or approach to adapting health promotion interventions for ethnic minority populations. Each theory was first briefly summarised (see Table 10) and the overarching themes are presented in Box 5; in this, we highlight the different points of emphasis between the health promotion for ethnic minority populations and that for the general population.

Davis et al. 254 proposed an Afrocentric approach for therapists to treat obese and overweight African American women. This approach incorporates cultural strengths with recognition and address of multiple oppressions (e.g. racism, sexism and stigma of obesity and overweight). The framework promotes developing trust, treating the client as the expert, taking personal histories, tailoring treatment and considering cultural components, such as tapping into social support networks and recognising potential strengths in the clients’ environment. Davis et al. 254 posit that African American women are likely to operate outside of mainstream ideals of thinness and, as such, identification with African American culture may offer protections against, for example, stigma in African American women who are overweight or obese, while greater acculturation may be harmful in this respect. When framing weight loss interventions for African American women, it is thus posited that weight loss for aesthetic reasons should not be emphasised as the main goal. Furthermore, individual-oriented activities promoted by interventions such as healthy eating or lifestyle changes may be perceived to detract from family activities (in which the definition of ‘family’ includes adopted and extended family members, as well as the nuclear family). Furthermore, the type of communication strategy used could influence what is deemed appropriate; for example, a preference for improvisation and oral communication (compared with written) could render interventions with a self-monitoring component unsuccessful. Recognition of the importance of biculturalism, acculturation and degree of black identity and their moderating effects on the enablers and barriers are also acknowledged in this approach.

Airhihenbuwa255 presented empowerment education for a health education and prevention model, with group education, efficacy and action at the core of this theory. Empowerment here goes beyond controlling environmental forces and draws on Freire’s notion256 of empowerment to engage both the interventionist and the target groups in the joint production of meaning, value and knowledge as a central aim of the health promotion intervention. A ‘failure to involve communities in the production of knowledge based on their own cultures, histories, and experiences doomed an otherwise logical health promotion strategy’. For example, utilising storytelling for health promotion draws on African oral tradition; however, applying this technique requires a consideration of both the storyteller(s) and the listener(s), with both groups being equally important to the meaning-making process. The educational and cultural background of both members engaged in the dialogue needs clarification, as the reasons why storytelling is effective may be a function of strong cultural traditions, or a function of social inequalities producing low literacy levels.

Hugo257 proposed a model for grading media appropriateness and cultural sensitivity of communication strategies used in health education for a culturally diverse community. As health education is largely reliant on the success of messages, the message needs to reflect ‘relevancy of information, the particular sociocultural context in which communication takes place, as well as the suitability of media used in message transfer’. It is therefore the method or type of technology used in context that is thought to produce effective communication and generate the appropriate response (i.e. behaviour change). Furthermore, media acculturation, defined as modifying the communication message to increase acceptability and reduce confusion, is another crucial aspect, bearing in mind that ‘appropriateness by nature is not at all a fixed quality’. The proposed method assesses both the appropriateness of the media and sociocultural sensitivity and can contribute to the design of health education materials. The programme outcome, whether the learning concept is abstract or concrete, media preferences of instructor and learners, current level of knowledge, preferred ‘media codes’ (e.g. showing colour, motion) and potential resource limitations are considered when developing communication strategies.

Davis et al. 258 described a tool kit for health and resilience in vulnerable environments (THRIVE) to aid communities to identify and increase health-improving factors and reduce ethnic disparities. The tool kit underscores the contribution of oppression and racism and addresses these factors by directing attention to both risk and resilience. Risk describes the ‘characteristics or circumstances that increase the likelihood that people within the community will experience poor health and safety outcomes’, whereas resilience describes the ‘ability to thrive despite the presence of risk factors’. Health promotion has, it is argued, overemphasised risk reduction while neglecting to see that resilience can buffer the negative effects of risk factors. Even when resilience is explored, it is usually considered as an individual quality rather than a potential community attribute. THRIVE assesses risk and resilience at four levels: built environment (e.g. safety, availability of spaces, accessibility to transportation), social capital (e.g. trust, community engagement and behaviour and gender norms), services and institutions (e.g. availability of public and community-based services) and structural factors (e.g. employment and advertising).

Ashley259 presented a framework to guide the needs assessment, programme planning and implementation and evaluation of health promotion and disease prevention interventions for African American communities. At the needs assessment stage, the framework recommends building relationships, involving community members throughout the intervention process, collaborating with formal and informal community leaders and working with community organisations such as churches to help access the community and form a planning committee to understand residents’ health needs. Data collection instruments should be pre-tested with a target group, culturally appropriate and matched for both language and literacy level. During planning a detailed account of the historical, economic, political and traditional or culture-specific issues for the community should be collected. Furthermore, it is important to assess whether there are competing priorities or socially sanctioned or acceptable norms, such as bigger body sizes (discussed above). Interventions should be based on scientific theories that have been tested in the target population, such as social learning theory260 and theory of planned behaviour change261 in African American communities. During implementation interventions should be pilot tested to determine cultural appropriateness and acceptability. Furthermore, given the difficulty of retention, ‘non-traditional sites and captive audiences . . . [such as] churches, Laundromats, nail shops, beauty salons, barber shops, family and class reunions, social clubs and organisations’ should be considered, as well as the use of a small group discussion format, which has been found to be effective. Evaluation and dissemination should be presented in clear, jargon-free language.

Resnicow et al. 35 introduced a framework for developing culturally sensitive interventions using the concepts of surface and deep structures. Surface structures are the ‘observable, “superficial” characteristics of target population’, while deep structures are the cultural, social, historical, environmental and psychological forces. Addressing surface structures is perceived to increase ‘receptivity’ and ‘acceptance’ and is achieved by delivering messages and programmes through appropriate channels and settings, using people, places, language, music, food brands, locations and clothing preferred by or familiar to the target population. On the other hand, addressing deep structures is thought to increase ‘salience’ and efficacy. For African American communities, for example, deep structures can include religion/spiritualism, expressiveness, preference for verbal communication, connection to ancestors and history, centrality of community and family and their potential use as motivators, and intuition and experience over empiricism, as well as stressors and responses to these stressors. Addressing deep structures is likely to require pre-testing and involve directed questions on perceived differences between the target group and the mainstream group and preferences for intervention content and format while being simultaneously aware of within-group differences. For example, for African Americans, promoting the spiritual or psychological benefits of weight loss may be effective. Resnicow et al. 35 suggested that an important means of articulating whether surface or deep structures are relevant is to determine individuals’ degree of ethnic identity. Therefore, even Afrocentric programmes may actually be culturally insensitive depending on participants’ ethnic identity, and therefore what constitutes an ‘adapted’ baseline intervention is unknown. Although the framework was developed for African American and Hispanic/Latino populations, it is suggested that this may be equally applicable to other populations.

James262 proposed the PEN-3 health education model that places culture as the central focus for both health promotion and disease prevention, such that culture is both an explanatory model and an action model. The PEN-3 model has three central dimensions: a ‘diagnosis’ of who should be targeted for health education and the most appropriate channel to reach that target group; a ‘diagnosis’ of target groups’ perception of health information, barriers and enablers; and the identification of the enabling and constraining culturally appropriate health behaviours. The PEN-3 model utilises findings from a focus group study about food with African Americans to illustrate the application of the model. In this study, women were identified as central targets for dietary change. Differing responsibilities among these women (i.e. with or without children) translated into different experiences and relationships with their weight and food. Discussion of dietary changes focused on ‘Soul Food’, a traditional form of African American cuisine. The women concluded that the food items themselves were not unhealthy, rather, it was the methods of food preparation that increased the fat content, for example. Although individuals recognised that Soul Food may be unhealthy, they also articulated the difficulty in changing such a historically and contemporarily significant cultural artefact. Furthermore, Soul Food was sometimes perceived as resistance to mainstream culture and connected with feelings of belonging, comfort and tradition. Mainstream guidance, such as the food guide, was perceived not to be applicable to the African American population. In light of this work, James proposed that health behaviours should be examined according to whether they are ‘long term’, that is, rooted in tradition (e.g. Soul Food), or ‘short term’, that is, more recent manifestations of cultural and/or socioeconomic idiosyncrasies (e.g. consumption of fast food). Making a distinction between eating healthy food as giving up tradition and eating healthy food to reduce fast food consumption was one recommended approach to dietary change. The PEN-3 model emphasised that ‘culture’ is an ever-changing dynamic force in individuals’ lives.

Fitzgibbon and Sánchez-Johnsen263 described successful strategies for low-income African American women for CVD and cancer risk reduction and the development of culturally competent programmes to address their specific needs. They recommended the stages of change model,264 a model originally designed to address addictive behaviours such as smoking cessation, but increasingly used for dietary modification and in programmes targeting ethnic minority populations. They suggested that ‘dietary change interventions present unique challenges to researchers because their main focus is replacement and modification, not cessation, as in smoking or drug abuse’. Their discussions also encompassed logistical considerations for encouraging recruitment: having safe and easily accessible settings in familiar locations and developing relationships with trusted, community and peer leaders, as well as considering the importance of the extended rather than the nuclear family. These strategies were important in the context of multiple recruitment challenges, including fear of exploitation by researchers and various obligations, roles and day-to-day stressors that may render health-care important only when there is presence of symptoms. As programmes are developed and implemented, a range of cultural factors should be considered, including acculturation, ethnic identity and religion. Focus groups should be conducted to determine the preferred intervention format, cultural norms and literacy and educational levels (not just reading level, but also the amount of material that could be read/digested). For dietary programmes ethnic foods should be encouraged. Programme retention presented unique challenges such as transient populations, gang activity and poverty-related stress. To address these challenges, retention strategies such as weekly contact calls, multiple incentives (monetary more effective than vouchers/coupons), the use of peer leaders and tracking participants by obtaining multiple telephone numbers and the contact details of close friends and family can be employed.

Keller et al. 265 used treatment theory to address issues of recruitment and retention of minority women participants in an exercise programme to reduce obesity. They identified the Harvard Matching Model of Recruitment266 and a model by Levkoff et al. ,267 which assessed matching at three levels: macro (community agencies vs academic institutions), mediator (gatekeepers vs research team) and micro (participants vs interviewers). These papers were more focused on recruitment whereas the present paper considered retention, which can be improved by considering critical inputs, mediating processes and substantive participant characteristics in intervention design. Treatment theory looks for the causal mechanism between the intervention and its outcomes. It further highlights the significant variables and proposes a way to measure them and control sources of external influences. The critical inputs for increased retention were identified as convenience, supportive surroundings and cognitive factors such as self-efficacy. Mediating processes included, for example, recognition that caring for their family was contingent on their own positive health status. Three types of participant characteristics were posited to influence intervention outcomes: personal characteristics, which were the measurable features such as demographics; cognitive processes and health knowledge, attitudes and beliefs – either illness or health related, such as degree of illness and associated symptoms or disabilities; and personal resources, such as social support and coping mechanisms.

Sorensen et al. 268 developed a conceptual model that tackled the interconnected pathways between risk-related behaviours leading to increased cancer morbidity and mortality and the complex sociocontextual factors mediating these behaviours. The model includes sociodemographic characteristics (e.g. social class, ethnicity, gender) as inputs into a larger social context, which includes interpersonal, organisational, neighbourhood/community and societal factors. These sociocontextual factors, which are modifiable, are in turn mediated by sociocontextual mechanisms, such as social norms, social support and organisational environment, as well as by individual factors, such as self-efficacy, intentions to perform behaviour and skills. This model therefore targets interventions at the level of mediating mechanisms that are amenable and actionable. Interventions are thus targeted at multiple risk factors while being sensitive to the sociodemographic characteristics of participants and their broader social contexts. This model attempts to draw out the salient characteristics of these multilayered sociocontextual factors as they relate to ethnicity and integrate them into intervention design. For example, one intervention study sought to target modifying factors salient to low-income, multiethnic populations, such as barriers to physical activity, by, for example, providing free gym passes, working to create walking paths with employers and creating opportunities to access community resources. Social support was further enhanced by working with employers to maintain healthy and safe working environments.

Appel et al. 269 addressed the disproportionate mortality and morbidity burden of African American women and developed a model that takes into account the wide variety of daily contextual factors influencing the health status of this population. The model is structured on ‘the interrelationship between contextual risk factors, rational choice theory, and opportunity cost’. Contextual risk factors were defined as those circumstances outside the control of women and could lead to both social isolation – a lack of access to health-care information and hesitancy to interact with the health-care system – and social incongruence – ‘an inconsistency between individuals’ social, educational, lifestyle, and monetary status they were able to achieve’. Appel et al. 269 argued that contextual factors become risk factors when they conflict with women’s capacity to modify cardiovascular risk factors. This model suggested that rational choice theory is useful for understanding how women make particular choices that can negatively impact on their health outcomes. Rational choice theory has four main factors that impact on African American women’s health choices, including a hierarchy of preferences (lifestyle, beliefs, attitudes, habits that shape behaviour); opportunity costs that encourage unhealthy choices (such as difficult access to healthy foods); and institutional constraints (such as a lack of culturally competent African American physicians). The fourth factor is an aggregation mechanism that proposes the replication of these factors in a given population – when faced with similar experiences and needs there is a tendency to make similar choices. These risk factors then create increased opportunity costs for African American women wanting to engage in healthier behaviours. Advanced practice nurses will, it is envisaged, play important roles in implementing interventions for communities living in this socially isolated environment of scarcity. The central role of the advanced practice nurses is, it is suggested, to work with community leaders to reduce opportunity costs and institutional constraints and provide education. Reducing these costs and constraints can be achieved through projects such as providing childcare, bus passes and car pools. Minimising opportunity costs within a community-based framework is proposed to empower women to make positive health choices.

Burnet et al. 270 developed a theoretical model focused on preventing diabetes in high-risk minority youth. This model situated behaviour change as constrained by a number of barriers including knowledge and beliefs (e.g. perceived difficulties in modifying cooking to include lower fat methods), attitudes (e.g. enjoyment in eating unhealthy foods), social norms (e.g. acceptance of larger body sizes) and the environment (e.g. a lack of safe recreational areas). Minority populations such as African Americans face difficulty in accessing health care, have little experience with research as few studies are directed at them and may distrust the medical community because of past abuses, which affects participation in health programmes. Taking stock of these factors, Burnet’s model is based on an underlying conceptual framework that situates positive behavioural skills as most effectively emerging from experienced-based learning to cultivate self-efficacy. Behaviours are cognitively targeted by social learning theory,260 which hypothesises that individuals must believe that their current behaviours are detrimental to their health and that changing them will lead to a positive outcome expectation. In addition, the individual must feel able to make those changes, and value the incentives for changing. The health belief model271 contributes to this modelling and suggests that behavioural change must also be considered subjectively acceptable. The broader environmental conditions influencing behaviour are also considered in the form of the theory of planned behaviour272 and the ecological model. 273,274 Within interventions, self-efficacy should be promoted through achievable performance goals and verbal encouragement. Normative beliefs that hinder self-efficacy must be targeted through community members and leaders. Thus, interventions must target both cognitive and environmental conditions of minority populations, within community contexts to increase self-efficacy.

Table 10 summarises the salient features of these explanatory theories and models.

It can be seen from this summary that the theoretical literature on how to modify health promotion interventions to meet the needs and preferences of ethnic minority populations is rather piecemeal, and contains a mixture of both explanatory and predictive theories, featuring some, but not all, of the three ethnic minority populations of interest to our current study. Whether different approaches apply to different ethnic groups or whether there are consistent principles that can apply across ethnic groups is still under investigation. Nonetheless, overall this body of work directs our attention to some overarching themes or messages for practice that may be relevant across ethnic groups, and are summarised in Box 5. Furthermore, as most of these papers do not present theories per se, there is room to develop a more all-encompassing approach, drawing on both these themes identified here and the findings from the critique of systematic reviews and primary studies presented in the remainder of this chapter.

Summary of systematic reviews

The seven systematic reviews that met our eligibility criteria covered various health topics including smoking cessation,246,275 physical activity/fitness,245,276 obesity,277 diabetes278 and CVD. 279 Most of the reviews undertook narrative analysis and synthesis of the included studies; however, two reviews had sufficient data to undertake meta-analyses. The studies are summarised in Table 11 and are discussed in this chapter by health topic. More detailed descriptions of these systematic reviews are found in Appendix 16, while quality assessments are included in Appendix 17.

Summary of empirical studies of adapted interventions

In total, 107 studies (reported in 154 papers) met our inclusion/exclusion criteria and were fully extracted and quality appraised. A summary of the distribution of these papers across ethnic minority populations and health promotion topics is outlined in Table 12. The bulk of the included studies focused on African-origin populations for physical activity and healthy eating (n = 50) and smoking cessation (n = 19). Detailed one-page summaries of all 107 studies are included in Appendix 18. A summary of the quality scores of the included studies is reported in Appendix 19. We first use the entire corpus of 107 studies to consider the feasibility, acceptability and equity of adapted interventions. In-depth discussion of direct comparison studies, adapted studies with additional components and studies reporting cost information is presented later in this chapter. These studies have the potential to inform deliberations on the effectiveness of adapted interventions.

Typology of adaptation

We extracted the adaptations reported by each study. We then compared and contrasted these adaptations and identified 46 mutually exclusive types of adaptation. Collectively, these 46 adaptations represent the breadth and depth of adaptations undertaken in the 107 included studies, which span all three health promotion topics and three ethnic populations of interest (Table 13). The number of adaptations used within each study varied and different combinations of adaptations were apparent. The main adaptations undertaken in each study are described in the one-page study summaries (see Appendix 18) and are also summarised in Appendix 20 to allow for comparison of adaptations used across studies.

Assessing the feasibility, acceptability and equity of adapted interventions

Explicit discussions of the feasibility, acceptability and equity of adapted interventions were limited in the systematic reviews discussed earlier (see Summary of systematic reviews). In addition, therefore, we searched the primary studies for relevant insights.

Effectiveness

The vast majority of the 107 studies had effective outcomes. This is including studies that reported effectiveness for any of the outcomes measured, at any time point and sometimes within subgroups of the participants. These findings are encouraging; however, the design of these studies does not allow any assessment of the extent to which the adaptation contributes to their effectiveness. To determine whether or not adapted interventions are effective for ethnic minority populations we therefore identified the studies that directly compared an adapted approach with a non-adapted intervention. These studies were mostly RCTs or CCTs, as a control group was essential to draw conclusions on the comparative effectiveness of adapted interventions. The hypothesis was that interventions that had been adapted would be more effective than interventions that had not been adapted (i.e. standard or usual care).

Cost-effectiveness

Thirteen studies reported on cost data and one study conducted a cost–utility analysis. A summary of these studies is provided in Table 17. The level of detail and degree of formal economic evaluation differed across these studies. Two of these 13 studies reported on the cost of recruitment and the remaining 11 on the cost of intervention materials and delivery. No studies carried out a cost-effectiveness analysis.

Programme Theory of Adapted Health Promotion Interventions

The purpose of this strand of work was first to synthesise the available evidence on whether or not adapted interventions were more effective than non-adapted intervention, and then to assess whether or not adaptation was cost-effective for uptake by health services. Although no definitive evidence supports the assumption that adapted interventions are more effective for ethnic minority populations, this is likely to be because of the infancy of the evidence base. What is clear is that more studies are needed that compare an adapted intervention with a standard intervention of comparable frequency, intensity and materials/methods used. Given that we identified only nine (out of 107) studies that used a comparative design this suggests that we still lack a sufficient body of evidence and we are therefore unable to provide conclusive recommendations for or against adapted interventions for ethnic minority populations. Overall, the body of identified studies suggests that adaptation increases acceptability of the intervention within the targeted population, measured by process outcomes such as trust in the health message, programme satisfaction and increased programme usage. The onus is therefore on researchers and target populations alike to develop more sophisticated health promotion interventions and programmes that are better evaluated so that we are able to discern the effect of the basic interventions from that of the adaptations. For improved evaluation we also need better outcome measurement tools to differentiate between the lack of effect of adaptations compared with the lack of sensitive measures to detect difference.

One way forward in the pursuit of a rigorous evidence base for adapted interventions and programmes and development of the means to be able to detect difference is the adoption of a standardised approach, which would increase comparability between interventions. Given the diverse range of studies we have identified, it is imperative that future studies be reported in such a way that they can be discussed in a systematic and comparative manner. Thus, from the body of literature we have analysed, we propose a generic Programme Theory of Adapted Health Promotion Interventions (Figure 9) as one strategy to facilitate comparison. The theory is in no way prescriptive and it is certainly not intended to be followed without critical reflection; however, it does provide an initial scaffold to support one’s thinking on how to develop and report on adapted health promotion interventions as a primary goal, and also to further the undertaking of comparative work. Clearly, not all programmes can incorporate study designs with control groups, but being able to discuss findings in a way that can be compared and contrasted with findings from other intervention conditions, in other studies, can generate learning principles across the spectrum of health promotion topics and ethnic minority groups. This theory represents the health promotion intervention cycle as a generic model, which is likely to capture some (if not all) of the steps usually undertaken in a wide range of health promotion interventions. Given that it has been developed using a specific body of identified literature used for this review, we will restrict our current discussion of the theory to interventions that target smoking, physical activity and healthy eating behaviours, although it is likely to have wider applicability.

FIGURE 9.

Programme Theory of Adapted Health Promotion Interventions.

In addition to prompting thinking on improving the reporting of studies, this programme theory can also be viewed as an approach to improve the development of interventions. Other work in this area of ethnicity and health has highlighted the need to examine every stage of an intervention pathway for potential intervention-generated inequalities. 376 This present programme theory offers a possible way to address, then, those potential inequalities. From the 46 items outlined in our Typology of Adaptation (see Table 13) we can begin identifying which of these adaptations may be required when developing an intervention for a particular population and health topic and subsequently map these adaptations onto the Programme Theory of Adapted Health Promotion Interventions to help with study planning. Figure 10 provides a schematic representation of some of the possible adaptations mapped onto the appropriate stages of the intervention. This method represents both a systematic and a transparent way to plan as well as report on adapted interventions, as many of the studies we reviewed often lacked clarity on whether a particular action represented a conscious effort to adapt or an artefact of working with specific ethnic populations. To be able to detect these differences it is necessary first to identify the adapted techniques and strategies that are effective for health promotion. A more comprehensive mapping of the 46 items onto the Programme Theory of Adapted Health Promotion Interventions is offered in Table 18. This is by no means exhaustive or definitive; however, it does draw on the synthesised literature and should be used as a working document that requires further testing. Furthermore, discussion of how to consider adaptations at each stage along the intervention pathway is provided in Chapter 8.

FIGURE 10.

Programme Theory of Adapted Health Promotion Interventions with examples of adaptations at each stage.

Strengths and limitations

A key strength of the systematic review presented in this chapter is its comprehensiveness; we examined a wide range of population groups and health promotion topics to identify the most contemporaneous research on the adaptation of interventions for ethnic minority populations. Our search strategy was not restricted by language and we included interventions from any nation. One limitation observed was the unequal geographical spread of the literature as the vast majority of studies originated from the USA and focused on African American populations. This reflects the state of the research and was thus out of our control; the concern is that this offers little in the way of direct evidence in relation to non-African American populations.

We were also limited by the quality of the studies. Because we did not limit the study type, many interventions were adapted but the studies did not tell us about the relative effectiveness of adapted interventions as there was no comparison group. The inclusion of diverse study designs added to the heterogeneity observed in the study populations, topics and techniques such that we were unable to meta-analyse any results. Furthermore, the quality of reporting of adaptations or adapted components varied dramatically across studies and subsequently, for some studies, it was difficult to extract this information. Information on equity and SES was also not often reported or linked with intervention findings; we were therefore unable to explore these factors further within the scope of our study.

There was also heterogeneity between health promotion interventions that were related to additive behaviours (e.g. increasing physical activity and improving healthy eating) and those promoting subtractive behaviours (e.g. smoking cessation), which posed difficulties in conceptualising how best to compare and analyse these health promotion interventions. Smoking cessation also differs in that once achieved for a number of years it appears easier to maintain, whereas increasing physical activity and improving healthy eating require more ongoing effort to sustain behaviour change; it is therefore a cause for concern that there is a lack of long-term studies to assess the sustainability of physical activity and healthy eating interventions.

Lastly, the review is limited to the search dates and we recognise that since the search was undertaken further reviews in this area have been published. This more recent literature includes, for example, the review by Netto et al. 13 on principles for adapting health promotion interventions for ethnic minority populations and the review from the National Institute for Health and Clinical Excellence on interventions to reduce risk factors for pre-diabetes among UK adults from black and minority ethnic groups. 244 Studying this more recent literature, however, suggests that the evidence has not progressed in any major way and thus we believe that the material presented in this chapter continues to provide a comprehensive overview of the state of the current evidence.

Background

Because of the space limitations of many journal publications and the tendency to report normative accounts of research studies, much of the information on the process of adapting health promotion interventions for ethnic minority populations may not be accessible through published reports. This information may, however, be more forthcoming from detailed discussions with researchers and health promoters with particular experience of working with ethnic minority populations. Qualitative research is especially useful to understand processes and experiences and to gather in-depth knowledge to get beneath public accounts. The rationale for this qualitative component was to extend our understanding of how adapted interventions are delivered and why they were successful (or not) for smoking cessation, increasing physical activity and improving healthy eating for African-, Chinese- and South Asian-origin ethnic minority populations.

Objectives

To understand the processes and experiences of, as well as the rationale for, adapting health promotion interventions for ethnic minority populations and to summarise lessons learned.

Methods

We conducted semi-structured interviews with researchers and health promoters adapting health promotion interventions for smoking cessation, increasing physical activity and improving healthy eating aimed at African-, Chinese- and South Asian-origin ethnic minority groups around the world. Interviews were transcribed and two researchers independently coded the interview transcripts using NVivo 8 (QSR International, Doncaster, Australia). Thematic analysis was undertaken, this being informed by the literature reviewed in earlier phases of the work while also remaining open to more emergent issues.

Results

Twenty-six interviews were conducted. Findings both complemented and extended the existing literature on adapting interventions for ethnic minority populations. Participants discussed how adaptations to promote healthy behaviours for ethnic minority populations could occur at every stage of the research process, from conceptualisation to dissemination, but they in particular highlighted the importance of sustainability considerations. Newer insights emerged including the need for a greater recognition of the lived dimensions of ethnicity, these including dynamic relationships with particular spaces and places, intergroup heterogeneity and previous experience with other health interventions that impacted on how new adapted interventions were received. These insights are generally unaccounted for in conventional descriptions of ethnicity. Recognition of what we have framed as the ‘contextual elements of ethnicity’ can facilitate process-oriented thinking on ethnicity and its potential impact on health behaviour. Finally, interviewees discussed the benefits of informal research networks, which constrained or enabled access to unpublished findings and ‘lessons learned’, these being readily available in the USA but less available elsewhere.

Conclusions

Researchers have employed a broad array of approaches in adapting health promotion interventions for ethnic minority populations, these mapping onto the Typology of Adaptation and the Programme Theory of Adapted Health Promotion Interventions we have developed. In developing and delivering these interventions, participants emphasised the need to understand and work with a range of relevant contextual considerations, and, given the complexity of this task, the need to be able to draw on the experiences of colleagues is critical. This engagement was more easily achieved in the USA, where critical mass of researchers appeared to exist, in contrast to, for example, Australia, where researchers and health promoters felt more isolated.

Ethics

Ethical approval was obtained from the School of Health in Social Science Research Ethics Committee at the University of Edinburgh (see Appendix 4).

Sampling

We employed purposive, maximum diversity sampling aiming to recruit 20–30 participants; based on our previous experience with undertaking a number of related qualitative studies, we anticipated that these numbers of interviews would allow us to achieve saturation on the issues at the heart of our enquiry. We primarily wished to sample researchers delivering and evaluating innovative interventions for smoking cessation, physical activity and healthy eating for African-, Chinese- and South Asian-origin groups. We were, in addition, also interested in hearing the perspectives of individuals involved in delivering the intervention, such as counsellors and health educators/promoters. To ensure that we had a maximum variation sample we devised a three-by-three recruitment approach (Figure 11), which was used to inform sampling and recruitment decisions. Participants with different roles in health promotion interventions and programmes were approached for interviews with the explicit aim of capturing a range of opinions and perspectives.

FIGURE 11.

Sampling matrix for qualitative interviews.

Participant selection and recruitment

Guided by the sampling matrix (see Figure 11) we initiated four interview recruitment waves. Each wave satisfied a different sampling goal, namely piloting, generating interest, maximising diversity and ensuring representativeness. A description of the recruitment waves and the corresponding numbers of participants approached and who participated is presented in Table 19. Recruitment wave two was particularly innovative as participants were recruited from a sampling frame derived from studies included in the systematic review (see Chapter 6). We continued to interview until we reached saturation,378 that is, when additional interview data no longer generated any major new insights. Although no formal assessment for saturation was used, saturation became apparent as we iteratively coded interview transcripts, such that no new NVivo codes were generated with the last few interviews.

We supplied an information sheet explaining the purpose of the study (see Appendix 21) to all potential participants and allowed them sufficient time to consider participation and ask questions. We provided a link to our project website (www.chs.med.ed.ac.uk/ethnicity-and-health/) and our contact details should those approached wish to request more information, and also contact details for someone outside of our project should they have questions they wished to ask an independent person. Written consent (see Appendix 22) was obtained from all consenting participants (both researcher and participant signed the consent form) and all participants completed a demographics questionnaire (see Appendix 23).

Data generation

Semi-structured interviews with participants were conducted either by telephone or face-to-face. Semi-structured interviews provided an opportunity to build rapport, elicit detailed accounts and explore new topics as they arose while keeping focus on the topic of interest. The interviews were structured around an interview guide, which was first piloted and refined through expert consultation (see Appendix 24). Interviews were conducted in English by either JJL or ED or jointly by both researchers and were digitally recorded and transcribed by the project secretary (FA). All transcripts were subsequently checked for accuracy and completeness by one of the two researchers (JJL, ED).

Data analysis

Analysis was comparative and iterative to synthesise the body of included studies. 379,380 A coding frame was developed drawing on both the theoretical concepts and emerging information and applied to the corpus of data for thematic extraction (and the subsequent realist evaluation, discussed in Chapter 8). The identified themes from the theoretical literature (see Box 5) were key concepts framing this analysis. As these themes were mirrored in the empirical literature, we therefore anticipated that they would also be apparent in the interviews and thus they informed the development of some of the deductive codes. Furthermore, the systematic review (see Chapter 6) gave rise to 46 adaptations derived from the empirical studies – these too further informed our deductive code generation. These findings together and along with the discussion of ethnicity in Chapter 1 constituted what we knew on this topic before conducting the interventions. We anticipated being able to locate and follow some of these common threads in the interview data; this was expected given that our sampling frame was in part populated by authors of the empirical studies included in our systematic review (see Chapter 6). We therefore anticipated uncovering some of these earlier discussed ideas. These qualitative findings were further important to contextualise and illustrate how adaptations and intervention stages work under the realist framework introduced and discussed in Chapter 8.

The coding frame (see Appendix 25) included both deductive codes generated through familiarity with the body of literature12,13,16,26,35,254,255,257–259,262,263,265,268–270 as described above (see Chapters 5 and 6) and inductive codes generated from an initial read-through of interview data. This hybrid approach381 has the potential to yield a rigorous thematic analysis as new insights and emerging themes are accommodated alongside themes already identified from the literature. The coding frame was independently piloted by two coders (JJL and ED) and disagreements were resolved by discussion or, when necessary, a third coder (AS). The two authors independently coded all transcripts using the NVivo 8 qualitative data analysis software. Multiple independent coding allowed for discussion of new and emerging insights and, importantly, the transparent documentation of how analysis developed. 382

The two coders grouped and mapped the identified codes with attention to how and what factors were considered to be important in the adaptation of health promotion intervention for ethnic minority groups. Thematic analysis383 was used to analyse text: 65 inductive and deductive codes were generated from the five topic areas and were grouped into conceptual clusters (see Appendix 26). These clusters formed 24 basic themes, which were further grouped into 11 organising themes. In light of our overall study question, namely ‘How can interventions be adapted to be more effective for ethnic minority groups?’, these themes were further abstracted into three global themes to answer this central question. The organisation of the codes into this hierarchical thematic network is described more fully in Appendix 26. This systematic method of building up the themes was grounded in the text-linked codes and helped to identify consensus or conflict across themes, as well as the absence of expected themes,384 to explore how adapted interventions worked or were thought to work.

Reflexive considerations

Our interest in interviewing researchers and health promoters with considerable experience in delivering adapted interventions meant that we were, in some instances, ‘studying up’. Research, and particularly interviews, usually take place between people of unequal socioeconomic status (SES). Participants in health-care studies, for example, are usually patients or subjects in trials, while those conducting the research are health professionals with institutionally sanctioned degrees. In contrast, ‘studying up’ refers to conducting research with participants who have a higher social/economic status and/or are members of bureaucratic or institutional bodies. ‘Studying up’ can help uncover the logic behind how and why interventions are designed and delivered, as well as help deconstruct the ideologies that both enable and constrain the success of adapted interventions. 385

The background of the team, with wide-ranging interests and expertise, aided recruitment as members initially reached out to those in their networks. This multidisciplinary and multiethnic research team provided thoughtful suggestions in grouping the deductive and inductive codes, cautioning against generalisations, and particularly in illuminating the boundaries of an analysis focused on ethnicity, to suggest considerations of SES, age and education.

The ethnic positionings of the two interviewers (JJL identifies as a Chinese Canadian who has also lived and worked in Asia, Africa and Europe, and ED identifies as Celtic European who has lived and worked in New Zealand for the past 10 years) positively facilitated identification and familiarity with the researchers interviewed. However, as the majority of our interviews were with researchers identifying as African American, although we were able to appreciate the historical situated experiences they were drawing on, we were largely unable to grasp the gravity of such a history that has been compounded by contemporary health-care and structural issues. At times these unfamiliarities widened our social distance and perhaps prevented us from ‘digging deeper’.

In analysis, our collective backgrounds in social sciences and public health framed our approach and likely directed our attention to the influence of social, structural and environmental factors on ethnicity and health over, for example, cognitive factors. Furthermore, there was a tension between two of the researchers (JJL and ED) over the use of the term ‘community’, with one opting for a more critical perspective and the other for a more practical perspective. This tension is likely apparent in our use and disuse of ‘community’ and who it can represent in our research.

Overview of findings

We identified three main ‘global’ themes: constructing ethnicity, adapting interventions and sustaining adapted interventions.

Constructing ethnicity emerged as a new insight in terms of broadening the thinking on ethnicity from its more conventional dimensions (i.e. religion, language, culture, physical features and ancestry) as described in Chapter 1. Furthermore, although demographic variables such as age, gender and SES interact with ethnicity, these distinctions were maintained so as to retain the utility of ‘ethnicity’ as a concept and not render it all encompassing. We found that expanding the concept of ‘ethnicity’ to also include contextual elements (e.g. social environments, physical space, past exposures to health research and services, diverse life experiences including stress) was particularly valuable for consideration on how to adapt interventions for ethnic minority populations.

The theme of adapting interventions discussed the kinds of adaptations undertaken and considerations around implementing the adapted intervention. These adaptations and considerations reinforced the adaptations identified in the systematic review (see Table 13). This was expected as many of the interview participants were, as noted above, recruited from a sampling frame derived from studies included in the systematic review. Many of the identified themes complemented what was already known about working with ethnic minority populations, for example the importance of community engagement and building trusting relationships,97 providing incentives to individuals or organisations for recruitment and retention,386 adapting messages and materials,35 matching personnel and using preferred methods to deliver messages and materials257 and, finally, building linkages with existing organisations. 387

The final global theme, namely sustaining adapted interventions, was an unexpected finding as we did not anticipate the degree of in-depth discussion in relation to the practicalities of delivering adapted interventions and sustaining this field of research (e.g. funding, formalising the intervention, recognition of work, staff safety and capacity building).

More detailed findings on each of these overarching themes are presented in the following sections in which each of the three global themes is elaborated on and discussed in turn by drawing on its constitutive organising and basic themes (italicised). A schematic diagram precedes each global theme to outline the hierarchical and associated relationships between these themes. Verbatim quotes are provided to further illustrate and support the three overarching themes [the speaker is identified by their participant (P) number, e.g. P1]. In the following sections we use the term ‘researchers’ to refer to the participants of this study, and the term ‘participants’ to refer to those for whom the intervention was delivered.

Global theme 1: constructing ethnicity

The process of constructing ethnicity incorporated commonly used population demographics such as age, gender and socioeconomic factors, which were, generally speaking, acknowledged to intersect with ethnicity in both expected and sometimes surprising ways. Constructing ethnicity also drew on familiar dimensions of ethnicity such as ‘culture’, food, language, religion/spirituality and ancestry, which were discussed in Chapter 1. Finally, in considering ethnicity in the practice of delivering adapted interventions for ethnic minority groups, constructing ethnicity also stimulated discussions of participants’ exposure to health care and research, setting-based considerations in which space and place are prominently featured, participants’ social environment and researchers’ recognition and acknowledgement of differences. Consequently, population demographics, conventional ethnicity and contextual ethnicity represent the three organising themes contributing to the global theme of constructing ethnicity (Figure 12). Together, these three subthemes have the potential to shed light on ethnicity as a dynamic process rather than a limited set of immutable characteristics.

FIGURE 12.

Global theme 1: constructing ethnicity.

Global theme 2: adapting interventions

We classified researchers’ descriptions of implementing adapted interventions for ethnic minority groups (regardless of the ethnic group targeted) into the following six organising themes: development, engagement, materials, methods, description and linkages (Figure 13). Not all researchers provided insights for each theme as some tended to focus on, for example, engagement, whereas others more readily discussed materials and methods. These themes contributed to the global theme of adapting interventions. The interviews interpreted together revealed the additional considerations and frequent negotiations that required attention at different stages of the intervention. These considerations and negotiations shaped and mediated what adaptations and programme components were feasible and acceptable and, therefore, ultimately possible to deliver.

FIGURE 13.

Global theme 2: adapting interventions.

Global theme 3: sustaining adapted interventions

Sustaining an adapted intervention involved interpreting intervention findings, reflecting on what was learned and what could be taken forward, and fostering the capacity of the workforce as steps towards continuity. The three organising themes of evaluation, insights and strengthening the workforce thus contributed to the global theme of sustaining adapted interventions (Figure 14). Discussions of how to disseminate project findings, sustain the programme of research and manage project personnel demonstrated the constraints and challenges faced by the researchers we interviewed as well as revealing the rewards and motivators for continuing their work.

FIGURE 14.

Global theme 3: sustaining adapted interventions.

Background

Increasing ethnic diversity and the persistence of health inequalities in the UK and elsewhere challenges health systems to produce evidence-based guidelines for effective health promotion interventions for ethnic minority populations. Relevant evidence should ideally come from randomised controlled trials (RCTs) of health promotion interventions in the target population, but adaptation of existing health promotion interventions that have been found to be effective in the general population offers a potential short- to medium-term solution to fill the current evidence void.

Objectives

In the context of promoting smoking cessation, physical activity and healthy eating in African-, Chinese- and South Asian-origin populations, we sought to identify which adapted interventions are ready for implementation and establish which interventions require further research.

Methods

We undertook a realist synthesis of our earlier phases of work, that is, the discussion groups in the two user engagement conferences, the review of guidelines and systematic reviews for health promotion interventions in the general population, the systematic review of adapted interventions and the in-depth interviews with academics and health promoters with experience of working with relevant ethnic minority populations. Explanatory frameworks to interpret and synthesise this body of work were sought by two researchers independently searching for and considering recurring patterns or demi-regularities in the evidence, and then reaching consensus through discussion. In particular, we aimed to describe which intervention contexts moderated adaptation mechanisms to impact on intervention outcomes. We utilised the six categories of ‘strategies and activities’ used in health promotion interventions to compare the findings from our review of UK guidelines and international systematic reviews in relation to the general population with those from our systematic review of adapted interventions. From this comparative work we identified which ‘strategies and activities’ were advocated by both guidelines and international systematic reviews, and which also featured in interventions adapted for ethnic minority populations. Through explicating this evidence base and interpreting it through the construction of context–mechanism–outcome (CMO) configurations, we reflected on interventions that are potentially ready for implementation in ethnic minority populations and those that require further research.

Results

We were able to identify contexts that either enhanced or diminished the effectiveness of adaptations at each stage of the Programme Theory of Adapted Health Promotion Interventions. Four exemplary in-depth case studies on group-based interventions, communication strategies, behaviour patterns and pharmacotherapy, and social support were derived to illustrate what adaptation approaches are likely to be effective for whom and in which circumstances. There was no clear evidence that any adapted intervention approach is ready for rolling out, but the ‘lowest hanging fruit’ in this respect was found to be in relation to pharmacotherapy-based interventions for promoting smoking cessation, which are likely to be effective for ethnic minority populations if the salience and acceptability to target populations can be increased. This effort could, we hypothesised, be achieved through paying close attention to underlying beliefs and values, the potential impact on family members and the provision of appropriate linguistic and peer-to-peer support, emphasising the benefits of relatively short-term drug therapy. There is also the need to create greater opportunities for incorporating exercise and active lifestyles into everyday life through interventions that alter the built inner-city environments in which many ethnic minority populations tend to be most concentrated; this may involve more easily accessible (by walking) amenities and safe locations for individuals, families and communities to increase physical activity. Interventions should also test changing the availability of foods served at community organisations and religious institutions as well as ensuring that any dietary modification or recommendation should prioritise substitution rather than avoidance of commonly consumed and perhaps culturally relevant foods. Furthermore, use of peer and lay advisors/educators and appropriately trained dietitians working with families, and in particular household cooks, has the potential to effect sustained changes in healthy eating habits. The RESET (relevance, evidence base, stages of intervention, ethnicity and trends) decision-making tool developed from this work will, we hope, aid the planning of this future research.

Conclusions

The appropriate adaptation of health promotion interventions is a complex undertaking with an, as yet, inadequate evidence base. This in part stems from the fact that, apart from a few notable exceptions, we still know relatively little about how best to effect clinically important and sustained lifestyle changes in the general population, let alone for ethnic minority populations. Relatively simple adaptations to improve the salience, acceptability, trust and use of smoking cessation therapies are likely to yield important benefits in the short to medium term, but, given the current paucity of evidence in relation to ethnic minority populations, this work needs to proceed in an evaluative context. Effecting sustained changes in exercise and healthy eating is likely to prove challenging, involving, among other things, environmental-level interventions that have been developed through a detailed understanding of the values, practices and aspirations of the target population.

Rationale and overview of the realist approach

We adopted a realist approach to synthesise the large and heterogeneous body of evidence from the preceding phases of work (see Chapters 3–7) in an attempt to understand which adaptation approaches are most likely to be effective in which populations, in what circumstances and why. We sought to develop a rich, multiperspective, contextualised interpretation and synthesis of our mixed-methods data.

Realist evaluation has previously been used in health research as a method for systematically reviewing primary studies (often in the context of studying complex interventions). These reviews have, for example, synthesised large and disparate bodies of evidence on internet-based medical education389 and provided a parallel analysis to a Cochrane review on the efficacy of school feeding programmes. 390 These reviews accommodated and often sought out heterogeneous studies (in design, population and topic, for example) for analysis. By searching for underlying theory in interventions and focusing on explaining the ‘demi-regularities’ or patterns observed in the body of evidence, it is often possible to generate important new insights into underlying explanatory mechanisms. 391 There is also the potential to elaborate on a body of evidence ‘to create an abstract model of how and why programmes work, which then can be used to provide advice on implementation and targeting of any novel incarnation of the intervention’. 392 Articulating how interventions are thought to work and how they actually work can thus help in contributing to the building of an evidence base that is sensitive to contextual considerations and, therefore, more likely to approximate ‘real-life’ circumstances and generate practical and applicable conclusions for research, practice and policy. 393 Data generated during the earlier stages of this project (see Chapters 3–7) were thus purposefully selected and used to derive a number of CMO configurations (illustrated in Figure 15: the mechanisms were the adapted interventions that we examined in different contexts to consider how their likely effectiveness was enhanced or diminished).

FIGURE 15.

Mapping the data generated from the different research methods to a realist CMO configuration.

Realist framework

A realist approach to data is not a review technique but a review logic,394 meaning that it is driven by an underlying framework rather than by set procedures. Each review will therefore require its own specific methods in line with the key steps of identifying and testing key theories and uncovering the likely mechanisms of interventions to hypothesise and test how interventions might work (or not). The key overall steps in this process are outlined in Box 9. A central goal is to maximise learning using the most appropriate methods available. We therefore modified these key steps as the original steps were developed for the synthesis of primary studies and we required steps to synthesise data generated from multiple method sources. Our adapted framework for realist synthesis is outlined in Box 10.

Data extraction and analysis relating to the different methods are discussed more fully in their appropriate chapters (see Chapters 3–7) and only a summary is offered here. In addition to the usual data extraction methods outlined for each specific chapter, two reviewers independently extracted any additional significant or ‘realist insights’. This means, for example, that when the reviewers extracted data from empirical studies for the systematic review (in Chapter 6), traditional information such as population, study type and outcome measures were recorded, as well as additional insights reported by the authors on why their intervention was thought to be successful or not. Synthesis of these data was achieved through repeated discussion among the same two researchers (JJL and ED) who searched, reviewed, interviewed, coded, extracted and analysed the datasets from the preceding chapters.

Reflexive considerations

Given the large amount of data collected over the course of the project, we frequently discussed the additional insights, or ‘nuggets’, found in the datasets; this process was facilitated by the close working proximity of the principal researchers and their complementary academic and practical backgrounds (social science and public health) and regular discussions with a third researcher (AS). Early in the project, insights discovered were discrete and fleeting and as such we struggled to detect initial patterns. Towards the end of the systematic review of adapted interventions (see Chapter 6), however, patterns began to emerge from the data. Only after the interview data (see Chapter 7) were coded and discussed did these patterns take on more tangible forms as the interview data helped to direct our attention to what the researchers working in this field perceived as significant to their work. Uncovering realist insights thus involved an iterative process – returning to the same piece of data with a slightly different perspective each time.

When initial CMO configurations (such as the ones presented in the examples, Boxes 11–14) were presented and discussed among the investigators and at the second user conference (see Chapter 3) they were generally well received. The research users felt that the example presented at the conference (see Box 11) offered valuable contextual information that was grounded in the evidence, but without being prescriptive. They did, however, caution an interpretation exclusively focused on ethnicity and suggested that SES might be an important confounder. This feedback broadened our perspective; however, determining whether SES or not was indeed a salient factor, or whether it acted alone or in concordance with ethnicity,376 was beyond the scope of the study.

What adapted interventions work for whom and in what contexts?

Our realist synthesis drew on data from 15 UK guidelines and 111 international systematic reviews of relevant health promotion interventions for the population in general; our systematic review of adapted interventions, which comprised 12 theoretical papers, seven reviews and 107 empirical studies of adapted interventions; 26 in-depth qualitative interviews with experts in adapting health promotion interventions; and seven discussion groups with research and health service users to identify, test and explore how adapted interventions work or not in the form of CMO configurations. A full list of the identified CMO configurations is included in Appendix 27. The findings presented should be considered a ‘work in progress’ of CMO configurations hypothesising which adaptations are likely to be effective for whom and in what circumstances. As evidence emerges we will be better placed to further test and, if necessary, refine these CMO configurations. Exerts from Appendix 27 are included in Tables 21–25 and are discussed below to illustrate some of the configurations identified for each adaptation mechanism at different stages of the Programme Theory of Adapted Health Promotion Interventions (see Chapter 6). Some of the eight proposed stages have been combined to facilitate more coherent discussion than might otherwise have been achieved.

Contextualised ‘real-world’ examples of how adapted interventions work

We developed a few illustrative examples to exemplify the main issues that have emerged from the above analysis, and to offer contextual richness to the main focus of this report.

Interventions that are ready for implementation and interventions that need more research

We drew on the six categories of ‘strategies and activities’ derived from the Tannahill definition of health promotion interventions111 (see Chapter 4) to compare the findings from the review of UK guidelines and international systematic reviews with the findings from our systematic review (see Chapter 6).

Adapted interventions that are ready for implementation

Given that our review of guidelines and systematic reviews for the general population (see Chapter 4) found relatively little robust evidence in support of interventions effecting clinically important and sustained behavioural changes in relation to exercise and diet in the general population, there is understandably at present little clear evidence in relation to ethnic minority populations (see Chapter 5). Our systematic review of adapted interventions reported in Chapter 6 similarly found no clear evidence of how best to impact on these lifestyle factors in relation to our target populations. Therefore, based on this work, there is no clear evidence of any intervention for promoting exercise or healthy eating that is ready for widespread implementation in UK ethnic minority populations.

Although the overall evidence base is more encouraging for interventions aiming to promote smoking cessation in the general population (see Chapter 4) – particularly in relation to individual support and drug-based interventions – there is still uncertainty about how best to adapt interventions in a clinically effective and cost-effective way in relation to ethnic minority populations (see Chapters 5 and 6) and so here too, we contend, there is at present no adapted intervention approach that is ready for rolling out throughout the NHS.

Adapted interventions for research

Although there are no adapted interventions that are ready for implementation, there are a number of important and promising areas in need of further research; these are considered below. Interventions were considered candidates for research if they had evidence of effect in the general population but had not been identified as having been trialled in an adapted form for our populations of interest; in addition, interventions that we did identify as adapted in our review and which had demonstrated evidence of effect from ‘strong’ quality studies were considered suitable for pursuing with further study.

User engagement

In Chapter 3 we discussed our experiences with user engagement through, among other things, national launch and dissemination conferences. Members of the project team and invited international experts in ethnicity and health research spoke at, and led on, workshops at the initial 1-day launch conference. This conference was attended by a variety of stakeholders from government and community organisations. The main feedback emphasised the importance of incorporating ecological models of health promotion and the need to expand our breadth of vision beyond individual factors contributing to health. Delegates suggested, for example, that we should consider interventions delivered outside of traditional health-care settings and examine those delivered by commercial organisations. In response to this suggestion we ensured that we remained alert to potentially relevant evidence on policy, resource-based and environmental interventions, even if evidence in these areas was less robust than that for individual-focused behavioural interventions (see Chapter 4).

Delegates also expressed their wish for us to foster linkages among them as they felt that there was much that could be gained through sharing of local, national and international experience. This was pursued through the creation of a project website (www.chs.med.ed.ac.uk/ethnicity-and-health/), which allowed the sharing of project materials and presentations through circulating contact details of interested participants, and the creation of an informal network in relation to health promotion for ethnic minority communities. The network and relationships that developed served as a welcome source of information and advice that we were able to draw on during subsequent phases of the research.

Our dissemination conference was held towards the end of the project and facilitated the sharing and discussion of our preliminary conclusions. Delegates were encouraged to reflect on, challenge and debate our findings. Although there was considerable support to further develop this field of enquiry, delegates cautioned against drawing simplistic conclusions from the available evidence, pointing to, for example, the need to thoroughly consider the effects of socioeconomic status (SES) and reflect on the possible impact of unmeasured, confounding influences. In the realist synthesis presented in the preceding chapter (see Chapter 8) we took on board the feedback received, highlighting how attempts at interventions need to stem from a detailed appreciation of the target community, the heterogeneity within communities and the ways in which interventions are likely to be received and interpreted. The idea of developing a theory of adaptation was also raised and initial encouraging discussions on how to take this forward took place in one of the workshops (see Table 1).

Extracting evidence on effective health promotion interventions from UK guidelines and international systematic reviews

In Chapter 4 we identified the evidence on effective health promotion interventions for the population as a whole for behavioural change in general and, more specifically, for smoking cessation, increasing physical activity and improving healthy eating. We then created summary frameworks of the effective ‘strategies and activities’ for each health promotion topic (see Tables 3–6). These summary frameworks were organised according to six categories, which spanned the spectrum of health promotion intervention levels, from an individual to a population level, with corresponding variations in scope for adaptation (see Figure 5). These summary frameworks enabled us to identify both saturated and sparser areas of evidence across the health promotion topics of interest and approaches used. The evidence base identified was considerably stronger for smoking cessation than it was for either promoting physical activity or changing dietary practices, for which there is still considerable uncertainty as to how best to promote clinically relevant changes that can be sustained over any length of time. In particular, there was a lack of empirical evidence from studies investigating the effectiveness of population-based interventions such as food pricing policy or environmental modifications (e.g. introducing cycling routes). There is an obvious tension here as the latter are far more difficult to evaluate using rigorous designs of the kind that guideline developers and systematic reviewers preferentially draw on.

Individual-centred behavioural approaches thus dominated the body of evidence we identified. The strongest and most consistent evidence of effectiveness was for smoking cessation interventions, which tended to be individual-level strategies, such as resource provision and provider interactions, aiming to increase uptake of and compliance with pharmacological interventions. In contrast, the as yet minimal evidence on activities and strategies associated with promoting physical activity and healthy eating was more likely to draw on population-level strategies, such as enacting environmental changes. We hypothesised that this difference in focus is related to the differing needs of interventions attempting to reduce a particular behaviour for a relatively short period of time (e.g. tackling the nicotine cravings associated with smoking cessation) compared with interventions seeking to stimulate particular behaviours (i.e. increasing physical activity and changing healthy eating) over a sustained period of time. This insight, together with the feedback received from our launch stakeholder conference, resulted in a continued search for evidence and experiences of adaptation of population-level interventions for promoting physical activity and health eating (see Figure 5). We believe that this approach was justified and appropriate as the need for clear evidence on the effectiveness of adapted interventions is, from a public health and clinical perspective, particularly important in relation to non-pharmacological interventions.

Consideration of ethnicity in UK guidelines and international systematic reviews

The high-level evidence synthesis of UK guidelines and international systematic reviews was then further scrutinised in Chapter 5 to assess the extent to which the particular needs of ethnic minority populations were considered within the evidence base for smoking cessation, physical activity and healthy eating health promotion interventions that are known to have some evidence of effectiveness and, consequently, are recommended for use in the general population. Although most of the UK guidelines (12 of 15 guidelines) and just under two-thirds of the international systematic reviews (66 of 111 reviews) made at least some reference to ethnicity and its related terms, acknowledging the existence (and by implication the needs) of ethnic minority populations (see Table 7), none of the guidelines or systematic reviews reported sufficient evidence to determine whether recommended interventions were more or less effective for ethnic minority populations. Consequently, it is at present not possible to determine from such evidence syntheses whether interventions of proven effectiveness and recommended for the general population are equally, more or less effective for ethnic minority populations.

The 12 guidelines and 66 systematic reviews commenting on ethnicity were further probed to extract the evidence statements made in relation to ethnic minority groups. All 12 guidelines and only 41 of 66 systematic reviews made evidence statements pertaining to ethnicity in the context of the health promotion topics of interest. These statements were grouped according to shared meaning and/or concept and then thematically analysed to identify five overarching themes: the acknowledgment of diversity within populations; identifying gaps in the evidence of which interventions are effective for ethnic minority populations; observing differential effects of interventions for different ethnic groups; reporting that action should be taken to adapt interventions; and, lastly, recommending improved research, reporting and analysis as a response to the absence of definitive evidence for interventions that are effective for ethnic minority populations (see Table 7). Importantly, we found no guidance in these high-level syntheses on how best to adapt the health promotion interventions being recommended to ensure that they were acceptable to and effective in ethnic minority populations.

Systematic review of adapted health promotion interventions

To complement the summary and critique of the guidelines and systematic review literature (see Chapters 4 and 5), we conducted a systematic review to identify and analyse the evidence for smoking cessation, physical activity and healthy eating health promotion interventions that had been adapted for African-, Chinese- and South Asian-origin populations (see Chapter 6). We identified and separately analysed three types of evidence in our review: theoretical/conceptual literature on adapted health promotion interventions for our health topics; systematic reviews of adapted health promotion interventions for the ethnic minority populations of interest to us; and relevant primary studies investigating the effectiveness of adapted health promotion interventions.

We identified 12 theoretical studies35,249–259,262,263,265,268–271 that offered concepts and models/theories in relation to adapting health promotion interventions for African-, Chinese- and South Asian-origin populations (see Table 10). These studies provided seven important contributions to frame our thinking around the similarities and differences between interventions adapted for ethnic minority groups and interventions for the general population (see Box 5). One contribution suggested that women were identified as central figures for behaviour change as both targets and agents of change, and this insight suggested an important and possibly greater need to consider gender influences when planning health promotion for ethnic minority populations compared with health promotion for the general population. 253,262,263,265,269 It was also suggested that the goals underpinning or motivating behaviour change in ethnic minority populations should account for the role of individuals within the context of families and communities to a greater degree than is often considered when planning health promotion interventions for the general population. 35 The development of communication strategies to better match the needs and preferences of ethnic minority populations was also emphasised, although this approach has increasingly been recognised as an important strategy in health promotion interventions for the general population. 254,255,257

These and the other overarching themes from the theoretical studies discussed in Chapter 6 provided a comparative foundation from which to reflect on the systematic reviews and primary studies we identified (discussed below). These themes also yielded helpful insights in relation to understanding the context–mechanism–outcome (CMO) pathways that we constructed in the context of our realist synthesis of the evidence uncovered (discussed in Chapter 8 and below).

We identified seven systematic reviews245,246,275–279 of adapted health promotion interventions for ethnic populations, of which the majority (six of seven reviews)245,246,276–279 reported on interventions delivered for African American populations (see Table 11). Findings from these systematic reviews were largely inconclusive, revealing mixed effects. Two reviews245,278 suggested that adaptation of interventions increased their effectiveness, but they were unable to identify which specific adaptations were associated with this improved effectiveness. On the other hand, a review on smoking cessation246 suggested that adaptation increased the effectiveness of health promotion interventions in the short term, but with no demonstrated effects in the longer term. The remaining four reviews275–277,279 did not report on whether or not adapted interventions increased effectiveness, attributing this to the lack of available primary research and the preliminary nature of some of the included studies, which were still ongoing at the time that the review was conducted. There was therefore no conclusive evidence from systematic reviews specifically focusing on adapted interventions for African-, Chinese- and South Asian-origin populations on whether adaptation increases the effectiveness of health promotion interventions or indeed which adaptations are most important to undertake for promoting the health of ethnic minority populations.

In addition, we identified 107 empirical intervention studies (reported on in 154 papers) on smoking cessation, physical activity and healthy eating health promotion interventions adapted for African-, Chinese- and South Asian-origin populations (see Appendix 18). Overwhelmingly, the majority of these interventions (91 of 107 studies) were delivered for African American populations (see Table 12). From these studies we were able to extract the range of approaches to adaptation reported to create a novel 46-item Typology of Adaptations (see Table 13), developed to code all reported adaptations in these studies.

Although 107 studies were identified, only nine studies293–295,306,341,347–352 directly compared an adapted intervention with an unadapted version of the same intervention (see Table 14). These relatively few studies offered the most direct evidence on the effectiveness of adaptation. Eight293–295,306,341,347,348,350–352 of these nine intervention studies were conducted with African American populations (the ninth was with Chinese-origin people). The results of these direct comparison studies were mixed, providing no conclusive evidence of whether or not adaptation increased the effectiveness of health promotion interventions. To further test the assumption that adaptation increases the effectiveness of interventions for ethnic minority populations, we used our Typology of Adaptations (see Table 13) in subsequent realist synthesis to broadly detect which adaptations were used more often in effective interventions.

Given the absence of definitive evidence on whether or not adaptation(s) increased the effectiveness of interventions, we then looked to the entire body of studies to examine the feasibility, acceptability, related process measures and equity of the included adapted interventions. We found that resource-intensive interventions, interventions that required additional individual tailoring and interventions with limited/inadequate funding tended to reduce the feasibility of successfully delivering the adapted intervention. In contrast, the use of and/or linkages with existing organisations or familiar settings were, overall, found to increase the feasibility of delivering adapted interventions. In terms of acceptability, interventions developed in collaboration with the targeted community and interventions that offered a degree of flexibility in their structure and delivery were found to be the most acceptable, whereas randomised controlled designs, interventions perceived to be too complex, competing priorities and costs to participate were all found to decrease acceptability. Intervention settings (e.g. churches and schools), criteria for participation (e.g. parental participation) and resources required for participation (e.g. computers and telephone access) were reported to influence the equity of access to adapted interventions.

There was limited reporting of costs associated with conducting adapted interventions in the 107 studies identified in the systematic review and an absence of rigorous economic evaluation of interventions. Only 13 studies285–289,292,299,302,323,327,328,334,342–345,358,365–373 reported cost data and only one study374 performed a cost–utility analysis; no studies reported on cost-effectiveness (see Table 17).

Finally, these adaptations were observed to occur throughout the stages of an intervention, from development to dissemination, and we correspondingly developed an eight-stage Programme Theory of Adapted Health Promotion Interventions (see Figure 9), which is designed to be used alongside the Typology of Adaptations (see Figure 10). This Typology of Adaptations (see Table 13) was further mapped onto the eight intervention stages identified in the Programme Theory of Adapted Health Promotion Interventions (see Table 18). Our Programme Theory of Adapted Health Promotion Interventions aims to provide a tool for the systematic and transparent consideration of approaches to planning and reporting on adapted interventions.

Qualitative interviews with researchers and health promoters who have adapted interventions for ethnic minority populations

We supplemented the findings of the systematic review with 26 in-depth interviews with purposively selected researchers and health promoters involved in adapting and delivering health promotion interventions (see Chapter 7). A thematic analysis identified three main themes: constructing ethnicity within adapted interventions; considering how interventions are adapted; and sustaining these adapted interventions. Constructing ethnicity represented a new insight in terms of broadening thinking on ethnicity from its more conventional dimensions (i.e. religion, language, culture, physical features and ancestry), as well as demographic variables such as age, gender and SES, to include contextual elements (e.g. the influence of social environments, particularly, in recent immigrant communities, physical spaces that facilitated or hindered participation, past experiences with health research and services and diverse life experiences including stress) (see Figure 12). The theme of adapting interventions (see Figure 13) identified many of the adaptations extracted from the empirical studies examined in the systematic review in Chapter 6, which were subsequently incorporated into the Typology of Adaptations (see Table 13). The reinforcement of adaptations identified in the systematic review in the qualitative interviews was somewhat expected as a number of the interview participants were recruited from a sampling frame derived from studies included in the systematic review, but we also found that our typology featured more recent approaches that have yet to gain traction within the wider literature, for example the use of peer leaders who are not only ethnically matched but also matched in terms of having successfully changed their behaviour (adaptation number 6); interventions that address health behaviour patterns found in the target populations (adaptation number 26); and purposefully maintaining an open intervention environment (adaptation number 42). The last theme, namely sustaining adapted interventions (see Figure 14), was an unexpected finding as we did not anticipate the in-depth discussions in relation to the practicalities of delivering adapted interventions and sustaining this field of research (e.g. in terms of funding, formalising the intervention, getting recognition for the work, ensuring staff safety and capacity building) (see Appendix 26). Finally, we observed a potential geographical variation in the support networks available to researchers delivering adapted interventions for ethnic minority populations. Researchers in the USA, in particular, identified an informal network accessible for sharing ideas and collaborating on projects, whereas researchers in other geographical areas such as New Zealand and Australia reported less access to support and interactions, and tending to work somewhat in isolation. UK researchers were somewhere between these extremes.

Realist synthesis

In Chapter 8, our realist synthesis drew on data from seven group discussions with research and health service users attending our conferences; 15 UK guidelines; 111 international systematic reviews; a review of 12 theoretical papers, seven systematic reviews and 107 empirical studies on adapted interventions; and a further 26 qualitative interviews to identify, test and explore realist insights of how adapted interventions work or do not work in the form of CMO configurations. Data from each of the preceding phases of work contributed to the context, mechanism and/or outcome constituents of a CMO configuration. Using the Programme Theory of Adapted Health Promotion Interventions, along with the 46 adaptations mapped onto to the relevant stages of intervention (see Table 18), we constructed CMO configurations relating to the different adaptations in order to describe how adaptations function as mechanisms in different contexts to produce varying outcomes.

Through this analysis we began to uncover why interventions work in different contexts and what factors may moderate their effectiveness. At the initial stages of intervention conception and planning, taking stock of appropriate study designs and behaviour change models as well as conducting exploratory work with a wide range of target community members were highlighted. At the intervention promotion and engagement/recruitment stages it was noted that interventions promoted through a target community’s preferred method of communication showed mixed results in terms of success; participants’ preferred formats for communication may have facilitated uptake but not necessarily increased retention. Recruiting from appropriate settings, however, appeared to be more successful. This was owing, in part, to the existence of established social networks and existing activities. There was also evidence in support of the ethnic matching of research facilitators to the target population; however, in some instances the recruiter’s experience and sensitivity were perhaps more important than their ethnicity.

The majority of CMO configurations were linked to the implementation/delivery stage and highlighted how adaptations needed to accommodate the ‘lived experiences’ of ethnic minority participants, for example acknowledging travel patterns, the timing and location of interventions and preference for different kinds of facilitators. Furthermore, the CMO configurations revealed that social support, derived from family and friends or group-based formats, was far more complex than anticipated, with preliminary findings suggesting that the preference for social support may be contingent on a participant’s health status (e.g. diabetes) and desire for privacy. Culturally matched intervention materials were another often cited adaptation for ethnic minority populations. In the studies we examined, the effectiveness of these materials was observed to differ in relation to participants’ perceived ethnic identity and, as such, required a greater accommodation of the heterogeneity that exists within these populations. For retention, telephone follow-up was reported to be effective if access to a telephone was consistent and continuous. The provision of transportation to the intervention site, or locating the intervention site in an already familiar and convenient location, was also associated with successful retention. For evaluation and outcome stages, external factors such as life stressors, social networks and substantive barriers (e.g. employment priorities) were all reported to moderate intervention effect. Additionally, during evaluation, ensuring the appropriateness and importance of the outcomes to the target population was a critical consideration.

Following this examination of what adaptations work for whom and in what circumstances, four illustrative examples provided the contextual ‘real-life’ trajectory of adapted interventions. These examples elaborated on the CMO configurations underpinning potential successes and failures for group-based interventions, communication strategies, behaviour patterns and pharmacotherapy, and social support-oriented interventions (see Boxes 11–14).

We then compared the ‘strategies and activities’ recommended for the general population for smoking cessation, increasing physical activity and improving healthy eating (see Chapter 4) with the ‘strategies and activities’ used in adapted interventions (see Chapter 6) to consider which interventions are ready for implementation and which interventions require further research. At present there are no adapted intervention approaches that are ready for rolling out throughout the NHS; however, we have been able to suggest which interventions should be prioritised for further research. These research priorities include increasing awareness of, access to and use of effective pharmacotherapies and potentially utilising a greater range of providers alongside peer or lay providers for smoking cessation interventions; personalised or group-based support (depending on the health status of the targeted participants) for diet and exercise interventions; and, in the context of exercise, group-based initiatives with an emphasis on social and health benefits along with setting-based (e.g. religious sites, community organisations, established institutions) solutions.

Finally, we extracted the principal findings from this synthesis and developed a tool to assist researchers and health promoters in the development of adapted interventions. This decision-making tool, called RESET (i.e. Relevance, Evidence base, Stages of intervention, Ethnicity and Trends), is described in more detail in Box 15. The RESET decision-making tool can be used alone but is recommended for use along with other key outputs from this research, namely the 46-item Typology of Adaptation and the eight-item Programme Theory of Adapted Health Promotion Interventions. For example, one could take an intervention that has been recommended for more research and then use the RESET tool to decide if this would be an appropriate intervention to deliver for a particular target population and, if so, consider how best to adapt it. Although we have begun to validate this through the qualitative work with an international group of experts (see Chapter 7), it should be noted that this is, at present, still a preliminary tool built up from the different research components undertaken in this report; we hope in due course to formally test and if necessary refine this further before making it available online for widespread use (see Chapter 8).

Adapting health promotion interventions for ethnic minority populations

Over the past two decades, support for adapting health promotion interventions for ethnic minorities has been gathering momentum. 9,13 The rationale for conducting this body of work is the assumption that adapted or culturally sensitive (or culturally tailored, based or adapted)35 interventions will be more effective for ethnic minority populations than mainstream interventions. 9 Although culture is commonly invoked to explain health inequalities, it is seldom operationalised, and its pathways and mechanisms rarely explicated. 416 Greater testing of culturally sensitive interventions to demonstrate health benefits for ethnic minority populations has therefore been championed. 35 Reducing health disparities for ethnic minority populations requires considerable thinking around the complex issues of culture, context and theory. 251 More specifically, the role of culture must itself be tested. In this work, ‘ethnicity’, which was conceptualised as encompassing culture, was tested as an operationalisable feature of adapted interventions.

There have been previous attempts to synthesise the available evidence on adapted health promotion interventions for ethnic minority populations. 245,246,275–279 Although these reviews have varied in the populations and health topics considered, they have predominantly originated from the USA with a focus on African American populations. The findings of our systematic review echoed the larger evidence base such that the majority of the studies we identified (91 of 107 studies) were undertaken with African American populations.

In earlier work that informed the current research, Netto et al. 13 identified five common principles of adaptation for health promotion for Chinese-, Indian- and Pakistani-origin populations. The principles proposed were using community resources to publicise the interventions and increase accessibility; identifying and addressing barriers to access and participation; developing communication strategies sensitive to language use and information requirements; working with cultural or religious values that promote or hinder attitudinal and behavioural change; and accommodating various degrees of cultural identification. This study has built and expanded on the earlier work through the addition of African-origin populations, which dramatically increased the number of studies included for systematic review and also allowed us to draw on a far more diverse body of quantitative and qualitative evidence.

Through this work we have been able to elaborate on these five principles of adaptation by producing a much richer typology of approaches to adaptation, which will, we hope, assist researchers and health promoters to identify the specific adaptations that they might wish to consider. The five principles outlined by Netto et al. 13 map onto some of the adaptations presented in Table 13, and sometimes onto more than one adaptation mechanism. For example, the first principle, using community resources to publicise the interventions and increase accessibility, maps onto adaptation number 10 (utilise local/respected religious/spiritual leaders) and adaptation number 11 (collaboration with ethnic-specific institutions and professional organisations). The third principle, developing communication strategies sensitive to language use and information requirements, encompasses adaptation numbers 12–21 (see Table 13) as they all deal with modifying communication materials. One main difference between the two adaptation resources is that, although the five principles of adaptation provide a rationale for each action, for example using community resources ‘to publicise the interventions and increase accessibility’,13 the 46 adaptations developed in this study are not necessarily associated with a particular purpose, and these can therefore potentially be used at any stage of an intervention. In this flexible format, these adaptations relate to a Programme Theory of Adapted Health Promotion Interventions to assist in deciding at what stage of the intervention one needs to consider and undertake these adaptations.

Recommendations for further research

Through this work we have been able to identify a number of priority areas for future research:

• Guideline developers and systematic reviewers of health promotion interventions should be encouraged to make clearer the population from which the evidence is derived so that those wishing to apply the evidence can assess it for relevancy; it is important that any areas of caution/evidence gaps are highlighted so that these can be filled.

• There is currently a lack of evidence-based models and tools to guide researchers and practitioners on how to best adapt health promotion interventions and, thus, the models and tools we have developed here should be tested and, if necessary, refined.

• A key research recommendation is the need for more high-quality research that compares, for example, carefully adapted interventions for promoting exercise with near equivalent standard approaches to facilitate assessment of the effectiveness of the adaptation(s). Furthermore, there should be improved reporting of the intervention and any adaptation undertaken.

• There is also a need for cost-effectiveness to be studied, possibly using mathematical modelling techniques such as Markov modelling, to examine the real end points of interest and link these with the costs of the interventions.

• As noted above we have uncovered a body of work on other populations (e.g. Hispanic and indigenous populations) that could usefully be synthesised to inform more international deliberations. If pursued, it would then be useful to synthesise this body of work with the findings in this report.

• We suggest that all multiethnic societies need to be contributing to this important research agenda, particularly as work in one geographical area is unlikely to be easily exportable elsewhere. Over-reliance on research in the USA is likely to undermine the initiation of appropriate scholarship and research elsewhere.

• Our qualitative work revealed that there are many researchers and practitioners breaking new ground and that, in view of the uncertainties in the evidence base and challenges inherent in undertaking this work, they would value peer support of the kind that is now available in the USA. A UK-based academic network could readily meet this need for the UK and indeed possibly also many European academics. Further, links should be established between this network and US-based networks that have a longer history of working on adapted interventions.

• Finally, given that the prevalence of long-term conditions is projected to soar and populations are likely to become increasingly ethnically diverse, it is crucial that this evidence synthesis on adapted health promotion interventions for ethnic minority populations is periodically updated as we strive to deliver equitable health services and outcomes for all.