Introduction

South Asian communities, along with other minority populations, have poorer asthma outcomes, higher rates of hospital admission, greater risk of rehospitalisation and a higher death rate compared to majority white populations.1,2,3 Asthma self-management, consisting of education, written Personalised Asthma Action Plans (PAAPs) and regular reviews (supported self-management) is known to improve health outcomes, and is recommended in national and international guidelines.4,5,6 Despite hopes that self-management offers a potential solution to address preventable health inequalities,1,5,7 there are concerns that asthma self-management interventions have produced little or no positive improvements on health outcomes for South Asians or other minority populations, further widening the gap of asthma inequalities.7,8,9,10 Possible explanations for these variations include differences in health-seeking behaviours related to health beliefs and attitudes to mainstream medicine,1,7,11 environmental or lifestyle factors,1,5,11,12 poor healthcare access and the quality of asthma care provided to these communities.13 These factors may be driven by cultural diversity, by the experience of being a minority and/or by socioeconomic status (SES). Thus, the way in which self-management is accessed and delivered to these various populations, need to be explored, and self-management strategies may need to be developed for the target population’s culture, ethnicity, SES or other needs.1,5,7

There are distinctions between the way interventions can be made relevant to a population (see Table 1). ‘Culturally modified/adapted’ interventions, are developed for a majority population and then modified for use in other ethnic groups; the core content, however, is the same. ‘Culturally targeted’ interventions are developed from a bottom-up process that considers the shared characteristics and context of a cultural group before developing an intervention. Finally, bottom-up interventions that assess and are aimed at the unique cultural characteristics and dimensions of individuals within a cultural group, with individualised intervention delivery are known as ‘culturally tailored’.14,15 Culturally targeted or tailored interventions are generally suggested to be more effective than culturally modified interventions, though the evidence for this has focussed mainly on children,14,16,17 is limited or out-dated.5,14,17,18,19

Table 1 Definition of terms
Full size table

Studies and clinical practice guidelines often indiscriminately apply findings from a majority population in a South Asian country, as relevant and applicable to South Asian minorities and majorities in other countries, despite differences in time and space of lived experiences and cultural shifts.20,21 Not only are the South Asian and Black population heterogeneous groups, but culture is fluid and continuously being shaped and reshaped across time and place, depending on an individual’s interaction with, and ability to respond to, the variability in their environment. Overlooking this ‘contextualisation’ may hinder adoption of self-management behaviour. Conversely, education aimed at cultural context enhances meaning, receptivity, relevance and processing of information by patients.20,22,23 Comprehension of a patient’s contextual realm offers a deeper understanding of the dynamic nature of cultural influences on self-management behaviour e.g., collective perceptions of asthma, familiarity with self-management and availability of, or access to, resources. This raises the question of whether poor asthma outcomes in ethnic minorities can be explained by their minority-status and/or by their relative social deprivation.6,16,19,24,25,26,27,28 These differences within a cultural group can influence the level of organisational and structural asthma inequalities faced by patients.29

This systematic review aims, in South Asian and Black communities (majority and minority populations), to (1) describe features of culturally relevant asthma self-management interventions, (2) synthesise the evidence for the effectiveness of interventions in different sociocultural contexts, and (3) identify barriers and facilitators to asthma self-management behaviour. We included interventions from South Asian countries where the population forms a majority (‘majority’ South Asian), and interventions from countries where the population forms a minority (‘minority’ South Asian; ‘minority’ African American) (see Table 1). We included studies of Black minority populations because our scoping work suggested that there was important literature, especially in African-American communities. This also allowed exploration of both the role of South Asian ethnicity, specifically versus the impact of minority/majority status on self-management outcomes.

Results

Characteristics of included trials

From a total of 3174 citations, we included 17 papers (reporting 16 trials) (see Fig. 1). The randomised control trials (RCTs) were conducted between 1995 and 2016; four South Asian trials were from India (labelled ‘majority’ South Asian),30,31,32,33 four South Asian trials were from the UK34,35,36 and one from Canada37 (labelled ‘minority’ South Asian), and nine African-American trials were from the USA (labelled ‘minority’ African American)26,27,28,29,30,31,32,33,34 (see Table 2). The overall risk of bias within trials was uncertain,30,33,37,38,39,40,41 or high.31,32,36,42,43,44,45 Three trials had low risk34,35,46 (see Table 3).

Fig. 1
figure 1

PRISMA flow diagram

Full size image
Table 2 Overview of study characteristics of included trials
Full size table
Table 3 All included paper findings as reported and the decisions underpinning the harvest plots
Full size table

Participant characteristics: The ‘majority’ population in the South Asian trials comprised of Indians,30,31,32,33 whereas ‘minority’ South Asian trials included Indians,37 and mixed subcultures (e.g., Bangladeshi, Pakistani, Indian or Sri Lankan).34,35,36 All Black population trials studied the African-American minority population in the USA.38,39,40,41,42,43,44,45,46 Most trials (fourteen studies) did not define ethnicity; only three ‘minority’ South Asian trials defined ethnicity according to self-identification or language spoken.34,35,37 All trials aimed interventions at asthma patients (whether this was children, adolescents, adults or elders).30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46 In addition, some trials also targeted parents,30,32,38,46 trained African-American coaches and/or residents,38,46 or healthcare professionals (clinicians and nurses).30,32,34,35,36

Study setting: All ‘majority’ South Asian trials were based in tertiary care hospitals.30,31,32,33 In contrast, ‘minority’ South Asian trials were conducted in primary care,35,36 or a combination of community, primary care and hospital (secondary/tertiary) settings.34,37 Similarly, the African-American trials were conducted in various settings: primary or secondary schools,40,41,45 tertiary care hospitals,39,42 emergency department43 and three trials used a combination of settings; community, school and hospital (secondary/tertiary).38,44,46

Geographical area and socioeconomic status: Among the ‘minority’ trials that specified the demographic location of patients, these were described as urban in six trials34,35,36,40,41,46; and one African-American trial was conducted in mixed urban and rural areas.43 Eight trials were described as from economically deprived or low-income areas,34,35,38,39,40,41,45,46 and two ‘minority’ trials (South Asian and African American) were conducted in low/middle-class areas.36,44

Intervention characteristics: Table 2 describes intervention characteristics. All interventions included patient education, though the approach, method of delivery and content varied. Examples included education-sessions or classes,30,32,33,35,36,38,39,40,41,42,43,44,45,46 training for patients,30,32,34,35,38,45,46 and healthcare professionals, coaches or residents,30,32,34,35,36,38,46 education in written,31,32,33,35,39,43,44 or video format,35,37,42 education in the form of social support,46 or a local education/promotional campaign.38 Twelve out of 17 interventions were delivered by healthcare professionals,30,32,34,35,36,38,39,40,41,42,43,44,45,46 five of whom were specifically trained for the project.30,32,35,42,43 Three interventions from minority countries were delivered in South Asian languages by healthcare professionals or research facilitators,35,36,37 two ‘majority’ South Asian trials had written materials in Hindi or Tamil,30,33 and two USA interventions were delivered by trained African American lay people or university staff who were residents in the community.38,46 Intervention duration ranged from 40 minutes to 1 year and follow-up lengths ranged from 1 month to 3 years (see Table 3 for details on the latter).

Strategies for reinforcing knowledge or self-management behaviours included follow-up classes,36,45 nurse clinics34,35,39,41,44,45 and written materials.42,43 Most trials described other intervention characteristics used alongside education,30,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46 including the use of written PAAPs in all South Asian trials (majority and minority)30,31,32,33,34,35,36,37 and some African-American trials,41,45,46 provision of emergency oral corticosteroid courses,34 asthma medication/therapy,30,32,34,36,39,42,44,45 placebo inhalers to practice technique,43 asthma diary/workbook,30,32,33,42 peak flow monitoring,30,34,36,37,39,41,42,44,45 medication counselling33 and access to free asthma organisation helplines.42 In seven trials, intervention strategies were based on specific guidelines, e.g., National Institutes of Health, National Heart Lung and Blood Institute, Global Initiative for Asthma (GINA) and Scottish Intercollegiate Guideline Network (SIGN).33,34,35,36,39,44,46 Usual care for the control groups varied,30,31,32,33,34,35,36,39,40,41,42,43,44,46 including illustrative leaflets,37 routine education classes,45 and recruiting similar neighbourhood areas to the intervention sites.38

(1) Features of culturally relevant interventions. In line with our definition and that in previous literature,14,15 we did not find any culturally tailored interventions, and only two of seventeen trials evaluated culturally targeted interventions.31,37 Behera et al.31 (‘majority’ South Asian trial at high risk of bias) provided a targeted written self-care booklet in Hindi (including a PAAP) developed collaboratively from patient knowledge, relevant literature and expert advice. Poureslami et al.37 (‘minority’ South Asian trial at unclear risk of bias) developed educational videos in collaboration with community members and healthcare professionals. The educational videos included three intervention possibilities (i.e., scientific knowledge, community opinions/narratives or a combination of both), that incorporated cultural beliefs and attitudes, e.g., cultural gestures, humour, storytelling and social interaction styles appropriate for Punjabi Indians. The aim was to facilitate patients’ trust in the community member and/or clinician who delivered the intervention.37 Both interventions were piloted in focus groups to improve clarity, relevance and acceptability and were refined before evaluation. These trials were not classified as culturally tailored because they were delivered to the specified cultural group without distinguishing or measuring individual cultural differences within that group.31,37

Both trials significantly improved knowledge. Poureslami et al.37 improved adherence to physician instructions on medication and inhaler use, and Behera et al.31 reported reduced symptoms, hospital admissions and use of breathing exercises during acute attacks. Although, the former trial achieved significant findings on all outcomes for Punjabi Indians, the Chinese population (who were studied as a parallel group with their own culturally targeted intervention) performed even better. The authors considered that this may be related to participant demographics; the Punjabi Indians were older and less educated than the Chinese community.37

In contrast, 15 out of 17 interventions were found to be culturally modified.30,32,33,34,35,36,38,39,40,41,42,43,44,45,46 They used strategies such as adapting existing interventions or materials for the target ethnic group,32,35,39,44 e.g., an African-American training video was re-recorded with South Asian actors,35 and ethnically relevant images were used such as African-American celebrities.34,35,42 Other studies applied interventions to several ethnic groups without considering cultural differences; thus, providing written or oral education (e.g., classes, PAAPs and workbooks) translated from English to the target participant language or using bilingual educators, without adjusting intervention content.33,34,35,36 However, the distinction between modified, tailored and targeted interventions is not clear-cut. Both culturally targeted interventions also incorporated some modified components,31,37 e.g., adaptation of language in PAAPs to meet the target population needs.31

(2) Effectiveness of interventions in different sociocultural contexts. In the harvest plot (Fig. 2 and Table 3), the four outcome categories (i.e., unscheduled care, asthma control, process and behavioural), are plotted for the three ethnic groups, ‘majority’ South Asian, ‘minority’ South Asian and ‘minority’ African American.47 The harvest plots show that the interventions in the ‘majority’ South Asian trials were effective, though notably they were all based in tertiary care settings potentially serving a relatively severe asthma population (thus with greater potential for improvement).30,31,32,33 In addition, risk of bias, was either high,31,32 or unclear,30,33 and two of these trials had short follow-up periods (1 and 4 months).30,33

Fig. 2
figure 2

Harvest plots illustrating the effectiveness on clinical, process and behavioural outcomes of self-management interventions across different ethnic groups and social contexts. To determine the overall effectiveness of trials, plots were placed under each category (unscheduled care, asthma control, process or behavioural), according to whether findings were positive (i.e., interventions, which were significantly effective in the intervention group), negative (i.e., interventions, which were significantly effective in the control group), or outcomes that had no impact between groups.50 The colours of the plots in the graph represent the study length (long and/or short), the height of the bars represent the sample size and the icon on the top of the bars represent the overall risk of bias within studies

Full size image

In contrast, trial outcomes from studies involving both ‘minority’ communities were inconsistent, though more trials were at a low risk of bias,34,35,46 in contrast to ‘majority’ trials. In the ‘minority’ South Asian trials, most of the outcomes did not show significant benefit.34,35,36 The exceptions were improved quality of life in a trial at high risk of bias,36 and in another study improved self-efficacy at 3 months, which was not sustained at 12 months.35 Similarly, in ‘minority’ African-American trials (all but one were at high or unclear risk of bias),46 most interventions were ineffective,38,40,41,42,43,45 or inconsistent.39,40,41 In addition, one trial at unclear risk of bias had a negative impact on unscheduled care.45 Three trials had positive outcomes (unscheduled care and behavioural),41,44,46 of which one trial was at a low risk of bias.46

(3) Identified barriers and facilitators to self-management in included trials.A range of barriers and facilitators to asthma self-management were identified and differentiated according to ethnicity and sociocultural context (Illustrated in Fig. 3). Key findings were that:

  • Across both ethnic groups and all social contexts, barriers included insufficient knowledge and understanding of asthma and related factors31,36,37,43; facilitators included providing self-management education,31,32,37,39,44,45 and support from healthcare professionals (with continuity of care).31,32,37,41,44

  • In ‘minority’ trials, even though language barriers were accounted for,36,37 a barrier identified for South Asians, was insufficient consideration of individual learning styles related to age,36,37 gender36,37 and level of education.37 In a ‘minority’ African-American trial, culturally/age specific self-management strategies (e.g., gaming) were identified as a facilitator.45

  • A facilitator that occurred frequently in studies involving South Asians across both majority and minority settings was providing culturally and linguistically appropriate educational materials. Language barriers were not an issue for ‘minority’ African Americans.31,36,37

  • Some barriers and facilitators were specific to one of the two ethnic groups or social context. For instance, facilitators for ‘majority’ South Asian trials included generic self-management strategies,30,31,32 e.g., use of PAAPs,30 written reinforcement,31 and practising preventative behaviour.32 One African-American trial observed that stressors (e.g., neighbourhood violence), interfered with generic self-management strategies such as relaxation and breathing exercises in adolescents.45 Similarly, three African American trials incorporated discussions of managing common stressors in daily African American lives as a facilitator, because this allowed individuals to focus on asthma.42,45,46 Another African-American trial identified social support as a facilitator.46

Fig. 3
figure 3

Summary of identified barriers and facilitators to asthma self-management in interventions across different groups

Full size image

Discussion

Main findings

We identified seventeen RCTs, most at unclear or high risk of bias, which tested asthma self-management interventions for South Asian or African-American communities. Education was a component of all interventions, but content, mode of delivery and additional strategies varied.30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46 Only two interventions were culturally targeted,31,37 in contrast to 15 culturally modified interventions,30,32,33,34,35,36,38,39,40,41,42,43,44,45,46 and no culturally tailored interventions. Trials based in South Asian countries,30,31,32,33 appeared to be more effective than those delivered to minority populations (for both South Asians and African Americans),34,35,36,38,39,40,41,42,43,44,45,46 though with the caveat that none of the ‘majority’ population trials were at low risk of bias and targeted populations were from tertiary care hospitals (in whom it may have been easier to demonstrate health benefits due to more severe asthma).34,35,46 Hence, it is unclear whether culture or minority-status of an ethnic group influences the variance in self-management outcomes. Education with on-going professional support was identified as a facilitator to asthma self-management in all groups.31,32,37,39,44,45 Other facilitators included focussing on individual learning styles in minority communities,45 culturally and linguistically appropriate education for minority and indigenous South Asians,31,36,37 generic self-management strategies in ‘majority’ South Asian communities,30,31,32 and strategies for dealing with stress and social support in African-American populations.42,45,46

Interpretation of findings in relation to previously published literature

A previous systematic review14 concluded that a culturally targeted intervention48 (in line with the definitions of this review) was more effective than generic programmes in improving asthma outcomes, and revealed that most interventions were culturally modified. We found only two culturally targeted interventions,31, 37 suggesting that this recommendation has not been adopted, hence progress in this area of research has not advanced. This may be due to the expensive and lengthy nature of developing targeted or tailored interventions compared to the ease of adapting or re-testing modified interventions,14, 17 however, in the long-term culturally targeted or tailored interventions may be more cost-effective. Trials have typically considered ethnic groups as homogenous, e.g., they do not consider variation among smaller subcultural groups of South Asians or African Americans, or the influence of acculturation in minority communities, potentially important for designing interventions.34,35,36, 38,39,40,41,42,43,44,45,46 The two culturally targeted trials also included some modified characteristics, e.g., language adaptation for PAAPs, so the distinctions between culturally relevant interventions is not absolute. This is supported by a previous systematic review,19 which found interventions labelled as targeted or tailored also incorporated modified features, e.g., community/participatory approach to smoking cessation. It may be that modification of certain proven asthma self-management strategies, e.g., PAAPs, together with customising by culturally specific elements is an optimal approach.

Targeted trials customise the development of interventions to a cultural group rather than just adjusting the content. For instance, interventions developed collaboratively with target groups helped existing self-management strategies to be linguistically and culturally relevant.7,16,31,37 This can be further understood as aiming at deep structures, e.g., cultural beliefs, norms, lifestyles, environmental and social contexts, which aid receptivity of information and behaviour change. The Person-Based Approach49 to intervention development suggests that comprehension of user perspectives and contexts based on qualitative studies at every stage of development is central to customisation. In contrast, modifying surface structures to observable traits, e.g., language, ethnicity, food and clothing, may influence information processing but not behaviour change (a common characteristic of modified interventions).23 For instance, two ‘minority’ South Asian trials modified interventions according to language with mostly ineffective outcomes, suggesting merely focussing on language modifications is insufficient for their needs.35,36 However, more rigorous trials are needed, as both targeted interventions had either high or unclear risk of bias.31,37

Similarly, some ‘majority’ South Asian interventions were modified from generic programmes rather than developed for their own community.30,32,33 For example, Ghosh et al.32 a trial from India, adapted self-management strategies from an intervention from Colorado, USA.50,51 Trials from diverse sociocultural contexts and different cultural groups demonstrate the potential pitfalls of extrapolating findings from one context and applying it to another.16,20,21 A possible explanation for ‘majority’ South Asian trials incorporating culturally modified strategies may be that international clinical guidelines for respiratory diseases,30,31,32 e.g., GINA,6 promote a generic model of self-management interventions with evidence and examples from high-income populations and recommendation of adaption to low or middle-income countries (LMICs).27 While remaining true to the core evidence-based features of supported self-management presented in guidelines, intervention developers also need to deliberate on the principles of cultural relevance to the targeted local community, rather than depending on translation.52 For LMICs, this may be challenging due to the lack of resources, training and manpower, as well as public health priorities and models of care focusing on communicable rather than long-term conditions.27,28,53 GINA guidelines acknowledge these difficulties, but do not offer specific guidance on providing targeted or tailored self-management;54 in contrast to the advice about cost-effective options for diagnosis and treatment in LMICs.6,28

Conceptualising culture with its interaction with context offers new avenues of comprehending the role of culture in health. Apart from better outcomes in ‘majority’ South Asian trials based in tertiary care settings compared to ‘minority’ communities,30,31,32,33 poor reporting with limited descriptions of SES,30,31,32,33,37,38,39,42,44,45 and diversity of trial settings,34,35,36,37,38,39,40,41,43,44,45,46 meant we were unable to draw conclusions about associations between outcomes and contextual data. This is an important point as variations in SES within a culture has been suggested to determine health outcomes, e.g., restrictions in accessing services.29 In LMICs such as India, tertiary care may currently be the only practical setting for delivering asthma self-management interventions due to lack of community-based clinical and research expertise, as well as social and financial barriers that result in under-diagnosis, under-treatment and limited treatment availability. In the absence of adequately resourced primary care, it is common for individuals in these populations (particularly for children) to only access healthcare during exacerbations, rather than receiving preventative care.28,53

Strengths and limitations of this study

To our knowledge, this review is one of few studies analysing the effectiveness of South Asian or African-American asthma self-management interventions. By identifying barriers and facilitators across two different ethnic groups and sociocultural contexts, our review can inform the customisation of interventions.21,32,35 We included seventeen trials, though the exclusion criteria of requiring separate outcome data for the specific groups of interest may have restricted the number of articles included in the final analysis; identification of more culturally targeted and even some tailored trials would have been informative. Limited resources precluded duplicate selection of papers, but we undertook a ten percent reliability check of the selection process. Risk of bias assessment was duplicated and data extraction was fully checked by a second reviewer.

Further, limited descriptions of the studies made it difficult to know how the interventions were developed or on what they were based on, particularly in the ‘majority’ South Asian trials.30,32,33 In addition, few authors responded to our request for further information. This meant that one of the targeted trials was excluded from the harvest plot analysis because data on between group differences were missing.37 Additionally, some harvest plot decisions relied upon sub-group analyses, which reduce study power and thereby could have increased the potential for null findings.34,36,43 However, primary outcomes were prioritised and, for clarity, inconsistent findings were indicated by hatched bars to limit over interpretation.35,39 Subjectivity in assessing the outcomes for the harvest plot was minimised by specifying predefined criteria that were replicable, and all the judgements were checked by at least two reviewers. Additionally, even though harvest plots are a good technique of illustrating heterogeneous findings and can be personalised to the requirements of the review, they may neglect some important outcomes that cannot be reported in the plots and overemphasise others.4,55

Conclusions and implications for future research, policy and practice

Asthma self-management interventions delivered in South Asian and African-American minority communities were less effective than interventions delivered in indigenous populations in South Asia, though the design/conduct of the latter studies meant that they were at greater risk of bias. Additionally, most trials from India are not designing interventions to their community, instead they are following guideline recommendations from studies in high-income countries. Studies that improve understanding of sociocultural contexts, allow a deeper appreciation of customising interventions and how to prevent inequalities in self-management behaviour, both are needed to inform international asthma guidelines. Targeted or tailored intervention development does not exclusively include collaboratively developed components customised to beliefs and needs of the target ethnic group, but may also include adaption of existing resources. Intergroup subcultural heterogeneities, cultural changes over generations (due to acculturation) and individual learning styles, add to the complexity of self-management behaviour and all need to be explored further. Rigorous trials of culturally targeted or tailored interventions are needed. Moreover, there needs to be standard recommendations on how trials verify participant ethnicity/culture, as only three ‘minority’ South Asian trials defined ethnicity according to self-identification or language spoken and culture was not considered and/or perceived to be synonymous to ethnicity.34,35,37

Methods

The review protocol is registered with the PROSPERO database (registration number CRD42015020174). We followed the procedures described in the Cochrane handbook for systematic review of interventions.56

Search strategy

Our key search terms were ‘asthma’ ‘AND’ ‘self-management’ ‘AND’ ‘population’ (including terms for South Asian and Black communities as summarised in Table 4 (detailed in Supplementary Appendix 1). We searched for RCTs on eight electronic databases (Medline, EMBASE, Web of Science, PsycINFO, Scopus, Elsevier Science Direct, Cochrane Library including Cochrane Airways Group Register of Trials and Google Scholar), three research registers in [February 2015] (PROSPERO, The University of York’s Centre for Reviews and Dissemination, and the Clinical Trials Database), manually searched relevant journals (Patient Education and Counselling, Health Psychology and Ethnicity and Health), and searched reference lists of identified systematic reviews. The search was not confined by publication year or language.

Table 4 Search strategy terms
Full size table

Inclusion and exclusion criteria

We included RCTs evaluating self-management interventions delivered to South Asian or Black asthma patients, the parents/carers of children with asthma, lay or healthcare professionals who care for people with asthma from these communities. The search included populations of all ages and in any country. Black African Americans, were included because they are from another well-studied minority population, with experience of socioeconomic deprivation, and our scoping of literature suggested there was a relatively large evidence base. Outcomes of interest were clinical (e.g., unscheduled care and asthma control),57 process, behavioural (e.g., knowledge and medicine adherence). We excluded studies that did not specify their population (e.g., trials using broad terms when describing their population such as ‘West Indians’ and ‘Asians’), and trials of multiple ethnic populations that did not provide separate asthma outcome data for the ethnic groups of interest (see Fig. 1; The PICO strategy is summarised in Table 5).

Table 5 PICO search strategy
Full size table

Study selection

A PRISMA diagram was used to report the number of studies identified, the screening process and the final list of included studies (see Fig. 1). All titles, abstracts and full texts were screened by one reviewer (S.A.), and a random 10% by two other reviewers (L.S., H.P.). Disagreements were resolved by discussion and the inclusion/exclusion criteria clarified as necessary.

Data extraction and risk of bias

A standardised Cochrane data extraction sheet was modified for this study.58 All data extraction was completed by one reviewer (S.A.) and independently checked by a second reviewer (K.H.). Discrepancies were resolved by discussions between reviewers and the wider team (L.S., H.P.), until consensus was achieved. Trial authors were contacted by email to clarify any missing, unclear or additional data required. If contact with the author failed, the uncertainty was noted on the data extraction form. The Cochrane EPOC Risk of Bias Assessment Checklist,59 was used to evaluate bias in included studies. This was independently coded by two researchers (S.A., K.H.), and any discrepancies were resolved by another researcher (L.S.).

Analysis

We anticipated that studies would be too heterogeneous for meta-analysis, and, therefore, used a narrative synthesis, illustrating key findings on trial effectiveness with a harvest plot.55 Harvest plots allow visual representation of the findings of a narrative synthesis (comparable to Forrest plots in a meta-analysis), facilitating comparison across studies.55 They enable identification of interesting patterns among varying outcomes, and may highlight the strongest or most inconsistent evidence, areas of possible concern, and gaps in the research. If there were various outcomes in one category (e.g., the asthma control category might include symptom scores, symptom-free days, or days off work/school with a range of significant and non-significant results), the overarching outcome was determined according to predefined criteria (see note to Table 3), applied and agreed by three researchers (S.A., H.P. and/or L.S.).55 Sizes of lines and colour hatchings were used to illustrate features of the trial according to a defined convention (see summary in footnote to Fig. 2 and detailed description in Table 3). Barriers and facilitators were identified from data and/or interpretations of study authors.

Data availability

All included papers are published; no further data are available. Requests for further information should be addressed to the corresponding author.

Disclaimer

The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.