Effective interventions to increase representation of under-served groups in randomised trials in UK and Ireland: a scoping literature review

Barriers

Several barriers to recruitment of under-served populations to trials were identified in the NIHR INCLUDE project (NIHR, 2020); barriers relating to physical disability, lack of effective incentives, lack of interest in research, negative financial impact, poor consent procedures, risk perception, burden and support required for participation. Other research has identified barriers specific to certain groups, for example, those lacking capacity to consent have legal barriers surrounding providing consent (Shepherd, 2020); Black African American communities are found to have less trust in research than white Americans (Corbie-smith et al., 2002); and South Asian communities in the UK experience barriers relating to health care use, language and the importance of family and community (Brown et al., 2014; Hussain-Gambles et al., 2004).

Trial design

The NIHR INCLUDE Frameworks (Gardner et al., 2022; Shepherd et al., 2022; Treweek et al., 2021) guide researchers through important questions when designing trials to help researchers think about what can be done to reduce barriers for groups that are under-served due to their ethnicity (including culture, faith, and language), experience of socioeconomic disadvantage, or due to their impaired capacity to consent. There is also guidance for including older people in health and social care research on the Trial Forge website. This might involve adjustments to trial design or include specific interventions to improve engagement between the trial team and specific ethnic minority groups. NIHR funding streams now emphasise more the need for consideration of inclusivity, but research teams and Clinical Trials Units (CTUs) may lack experience in this area and not know what interventions to put in place to improve inclusion in trials.

Strategies for improving inclusion in trials

Methodological interventions have been suggested to improve representation of under-served groups in the literature (Ismail et al., 2014; Liljas et al., 2019; Shepherd, 2021; Velzke & Baumann, 2017), and previous reviews on improving recruitment of under-served groups to trials from international studies (Bodicoat et al., 2021; UyBico et al., 2007) highlight the need for effective interventions in this area. However, the variable methodological rigor and evidence gaps indicate that further research is necessary to address this issue comprehensively. A recent review of international research (Bodicoat et al., 2021) identified evidence that cultural competency training for recruiting staff and personalising communication improved representation, but no strategy was effective across trials or populations, and they recommended a multi-faceted approach to the recruitment of under-served groups. Masood et al. (2019) undertook a review of trials that aimed to recruit South Asian populations in the UK and identified the following strategies: Adaptation of screening and outcome measures, culturally specific recruitment training, working with religious leaders, collaborating with ethnic community organisations, self-referrals and assistance from family and carers, recruitment sites in diverse areas, multilingual written invitations, translation of the participant information sheets, tape recorded participant information, choice of interview location, follow-up arrangements, linguistic matching, ethnic matching, gender matching and awareness of the cultural practices and norms. However, these strategies were not evaluated. Prior research tells us that that recruitment strategies are not recorded or reported in most trials making evaluation of used strategies impossible (O’Sullivan Greene & Shiely, 2022).

Trials aimed at general populations based on disease are more common than those focussed on a specific under-served group, and due to the number, breadth and intersectionality of under-served groups, trialists need to consider a range of under-served groups to improve inclusion in trials.

A scoping review was chosen to identify existing methodological interventions across a range of under-served groups, trial types and using various methods of evaluation to provide information on their effectiveness.

Objective. The objective of this scoping review is to identify, report, and evaluate the effectiveness of interventions aiming to improve representation of four under-served groups in trials in the UK and Ireland, as described below.

Protocol and registration

A protocol for the ACCESS project was published on the Sheffield CTRU website prior to the start of the final searches: https://www.sheffield.ac.uk/scharr/research/

This scoping review forms work package 1 of the work described in the protocol.

Patient and Public Involvement

Patients and public were not involved in the scoping review but were involved in the later stages of the ACCESS project where the results of the scoping review were presented and used to stimulate further discussion around inclusive trial methodology.

Scoping review strategy

A scoping review was conducted according to Joanna Briggs Institute (JBI) methodology guidance for scoping reviews (Peters et al., 2020) to ensure a rigorous, transparent and trustworthy evidence synthesis to explore and summarise the literature across a range of under-served groups and trial design. A literature review was undertaken on trials that evaluated interventions to improve the representation of under-served groups. An initial scoping exercise using the Online Resource for Research in Clinical triAls database (ORRCA) was undertaken to explore the relevant literature on improving representation of under-served groups in clinical trials. Following this, the search strategy was developed in consultation with the collaborator group. Based on the scoping exercise and collaborator experience, the search focussed on the following under-served groups: minority ethnic groups, socioeconomically disadvantaged groups, those with impaired capacity to consent and older people. We focussed on four under-served groups, as commonality and intersectionality of under-served groups means focussing on one under-served group is unlikely to be sufficient in making trials more inclusive.

Eligibility criteria

Inclusion criteria were:

Exclusion criteria were:

Search strategy

The search strategy (Extended data: Appendix 1 (Biggs, 2024)) was comprised of search string relating to randomised controlled trials, a search string for papers conducted in the UK and Ireland, and a string related to recruitment, retention and inclusion. Search strategies were developed for each under-served group and combined with these.

Information sources

CD searched PubMed for published papers on 29^th November 2021. We also included one paper identified through a preparatory search whilst developing the search string, this was not identified in the final search due to the addition of the RCT and UK and Ireland filters.

Quality of the included studies

We did not perform a formal assessment of the quality of included studies in line with the recommendations for scoping reviews (Peters et al., 2020).

Study selection process

The titles and abstracts of the papers were reviewed independently by KB and CD using Mendeley. The full text publications were retrieved and screened by both KB and CD, and they had regular meetings to discuss the interventions being identified and data extraction.

Data items and charting

The following data were extracted from the included papers: Author, date, background/conclusions, methods, population, researchers’ definition of the population, trial description, disease area, host trial intervention, methodological intervention, rationale for intervention, implementation, recruitment figures, retention figures, qualitative findings, costs of the intervention, author discussion around effectiveness of interventions, and author recommendations. Data were extracted into a spreadsheet for all included papers by KB and CD independently and discussed.

Synthesis of results

The methodological interventions identified during data extraction were explored by KB and organised into themes based on categories in a previous UK review (Masood et al., 2019). The interventions and themes identified in the included papers were discussed at a collaborator meeting. The previous categories were related to interventions to recruit an ethnic minority group, therefore some needed to be amended or widened to cover the additional under-served groups included in this review, some of the categories were split to provide additional detail and additional categories were added.

Included paper characteristics

Characteristics of the included studies are presented in Table 2, this includes details of the trial(s) the methodological interventions were evaluated in or for.

Under-served population. All seven included papers described interventions to improve recruitment of an under-served group to clinical trials in the UK or Ireland and evaluated them. One trial concerned the recruitment of South Asian participants, two trials concerned the recruitment of socioeconomically disadvantaged participants, with one of these also focused on older people. Three trials were solely focussed on older people and one trial was concerned with recruiting people with aphasia (who may have impaired capacity to consent).

Study methods. The included papers include one randomised evaluation, a randomised evaluation across five UK trials looking at the £100 incentive and non-randomised evaluation of the Consent Support Tool, which was a well-designed mixed methods study of a tool to support recruitment. The remaining papers retrospectively report recruitment or retention strategies with no comparative evaluation. These five papers were ‘lesson learnt’ papers reporting on one trial. They either compared recruitment settings or discussed a range of methods used in their trial and apart from where recruitment settings were compared, the interventions discussed did not have a comparator.

Trial characteristics

Trial design. One paper evaluated a methodological intervention across five trials, all two-arm RCTs, and one paper evaluated a tool to be used in a two-arm pilot trial. The remaining papers all reported on methodological interventions from an RCT, one four-arm cluster RCT, one three-arm RCT and three two-arm RCTs.

Trial population. The trials’ populations included participants with cardiovascular conditions, hypertension, and Aphasia, participants with untreated urinary incontinence, pre-diabetes, and populations without an existing diagnosis, such as elderly participants at risk of mobility disability, improving nutrition in older people, and cancer awareness. Number of participants recruited ranged from 34–777.

Trial interventions. Interventions included public health strategies, group-based interventions, speech and language therapy, dietary consultation, nutrition supplements and drugs.

Interventions to improve recruitment of under-served groups

The papers reporting on ‘lessons learnt’ from a trial (Agnew et al., 2013; Douglas et al., 2011; Forster et al., 2010; Kolovou et al., 2020; Withall et al., 2020) discussed more than one methodological intervention, with 48 discussed in total. Each intervention is listed in Table 3 and categorised into nine main themes.

The interventions identified in the included trials are listed below, grouped into themes guided by the type of intervention or strategy and by previous reviews. There was considerable overlap across the papers, particularly in relation to recruitment settings.

Two studies discussed recruitment sites, which refers to where the sites are geographically, and how diverse the areas are, choosing sites with the targeted population demographics (Kolovou et al., 2020; Withall et al., 2020) and one study targeted culturally and socio-economically diverse groups of women (Agnew et al., 2013).

Five studies reported on one or more recruitment settings, the recruitment pathway and where participants were recruited from. Settings included:

One paper used community engagement and lay advisors for delivery of the intervention (Kolovou et al., 2020) and another used community engagement to raise awareness of the health condition and research (Withall et al., 2020)

Communication between researchers and participants was mentioned in four papers: one employed bilingual staff to deliver the intervention and recruit (Douglas et al., 2011), one highlighted affiliation with the university and provided further information to improve trust (Kolovou et al., 2020), one allowed/provided support to complete questionnaires (Forster et al., 2010) and one provided funds for translators and built rapport with participants through friendly telephone and face-to-face contacts (Withall et al., 2020).

Financial incentives were discussed in four papers (Forster et al., 2010; Jennings et al., 2015; Kolovou et al., 2020; Withall et al., 2020). Flexibility in appointments for recruitment (Kolovou et al., 2020) and follow-up (Kolovou et al., 2020; Withall et al., 2020) was discussed in two papers. Two papers mentioned the design of their invitation letter (Withall et al., 2020), or patient information sheet, where they had patient and public input and adhered to national average literacy levels (Kolovou et al., 2020). Widening inclusion to more frail patients to increase the recruitment of older people was reported in one paper (Withall et al., 2020) and the Consent Support Tool was evaluated to see whether the tool could determine the level of information needed for the participant to consent (Jayes & Palmer, 2014).

These interventions were grouped into the following categories, adapted from categories in two previous reviews (Bodicoat et al., 2021; Masood et al., 2019): Recruitment sites and settings, Community engagement, Communication between study team and participants, Incentives, Widening the inclusion criteria, Flexibility, Patient documentation, and Consent process. Categories that were not in the previous reviews were: incentives, widening the inclusion criteria, advertising as a recruitment method, and using the Consent Support Tool to facilitate the consent process.

Table 3 provides further details of the interventions described in each paper, and the authors’ evaluation of the intervention. A table of the main and sub-themes identified is in Extended data: Appendix 2 (Biggs, 2024).

Evaluation of methodological interventions to improve recruitment of under-served groups

Not all interventions were evaluated in the included papers; six papers reported recruitment data, with five of these reporting the impact of different recruitment settings on recruitment. One paper found no difference in recruitment between four types of workshops (varying in content and participant interaction) (Agnew et al., 2013), and one showed few differences in ethnicity, education or employment between participants recruited from a healthcare or a community setting (Kolovou et al., 2020). Three other papers compared a number of settings, with one reporting 73% of their screened participants being recruited via the research team contacts, snowball sampling, talks and through community organisations, and these were the most successful methods for recruiting South Asian participants (Douglas et al., 2011). Two studies reported that writing to participants via their GPs was the most successful in recruiting older participants (Forster et al., 2010; Withall et al., 2020), with one of these pointing out that more targeted efforts might increase ethnic and SE diversity (Withall et al., 2020).

Jennings et al. (2015) used an RCT to evaluate their methodological intervention across five trials. The authors found a 6.9% increase in positive responses to the invitation letter when the £100 incentive was mentioned, but this did not affect the age of those responding or increase the number of socioeconomically disadvantaged participants. Other studies that mentioned an incentive did not assess them, though one mentioned that it may have contributed to recruitment (Kolovou et al., 2020) and the other that it may have contributed to high retention rates (Forster et al., 2010). These two trials were the only ones that reported retention rates (Forster et al., 2010; Kolovou et al., 2020), which were high in both.

Jayes & Palmer (2014) did not evaluate their intervention within an existing trial but used a mixed methods approach to evaluate the Consent Support Tool (CST) that would be used in trial recruitment. They found that the tool successfully identifies the appropriate information to give the participant based on their aphasia and can be used in the trial consent process.

One lessons learnt paper (Kolovou et al., 2020) asked lay advisors about their experience of recruitment. They reported that recruitment was successful due to higher footfall in the community venues, and potential participants having time to ask questions and enjoying discussing the research. Lay advisors reported that minimising burden for recruiting centres was helpful in recruiting venues, but not being able to pay for venue hire was a barrier. Another study (Withall et al., 2020) eliciting research staff’s views about recruitment methods reported that friendly contact, rapport building, flexible screening, follow-up appointments and reimbursement for travel were key to the success of the study.

The other interventions were not formally evaluated across the studies: recruitment sites, design of patient materials, patient and public involvement (PPI), employing bilingual staff, flexibility in recruitment appointments, flexible follow-up, and communication between study team and participants. However, Withall et al. (2020) reported that the following elements were key to successful recruitment of their elderly population: invitations and advertising using lay language and providing a good definition of the study, amending inclusion criteria to more frail participants, and prompt and friendly contact to build rapport between researchers and participants. They also commented that although talks at community venues were resource intensive and only recruited a few participants, they supported recruitment of diverse participants.

Although not formally evaluated, three studies reported that involvement with community groups helped with recruitment (Douglas et al., 2011) and there were recommendations to start this work as early as possible, developing a relationship with community leaders who can access, via networks, newsletters or venues, the underserved people you need.

In addition to the findings above, three papers provided further recommendations for recruiting and retaining their included populations. In relation to recruiting elderly populations, Withall et al. (2020) recommends more accurate targeting to improve response rates and reduce costs, but that in large-scale RCTs, these should be in addition to large-scale approaches, such as mailouts. They found their internal pilot useful in fine-tuning recruitment methods and that building rapport and trust was important as potential participants passed through the screening process. Forster et al. (2010) recommends minimising respondent burden to maximise response rates at the recruitment stage of a trial and they felt support for completion of participant documentation, either from friends and family, or researchers, and reassurance were important in helping participants complete tasks. In relation to recruiting adults experiencing socioeconomic disadvantage, Kolovou et al. (2020) states that future studies might benefit from community engagement and recruitment through communities and local gatekeepers. They suggested that more work is needed on how to include groups who were underrepresented in their trial: men, ethnic minority communities and adults from socioeconomically disadvantaged areas in part- or full-time employment.

What did we find?

Seven papers were identified for inclusion in this scoping review, demonstrating that published empirical evidence exists to support trialists in the UK and Ireland to improve representation of four key under-served groups: people from minoritised ethnic communities, people experiencing socioeconomic disadvantage, older people, and people with impaired capacity to consent.

The seven included papers reported various interventions that we categorised into nine broad themes, six of which had been reported in previous UK reviews: Recruitment sites, recruitment settings, community engagement, communication between study team and participants, flexibility, and patient documentation. Interventions specific to the papers included in this scoping review were: incentives, consent processes, widening inclusion criteria, and advertising campaigns. Only the financial incentive, consent support tool and recruitment settings were evaluated.

The only randomised evaluation was of a £100 incentive mentioned in the in the invitation letter, which improved positive response rates to the invitation letter, although there were no differences in the age or number of people from the most deprived areas between those who were offered the incentive, and those not (Jennings et al., 2015). There was only one other pre-planned evaluation (Jayes & Palmer, 2014) which showed the Consent support Tool could be used in the consent process for people with communication issues.

What does this mean for trialists in the UK and Ireland?

The lack of evaluation identified in the review means we cannot draw firm conclusions about successful interventions for improving inclusion in trials for these under-served groups. The papers suggest that having different recruitment pathways can be helpful in recruiting diverse under-served groups. Community recruitment can be beneficial for some under-served groups, but is resource intensive, and consent support processes can be used to aid consent.

The studies are context-specific, and interventions shown to be effective for one under-served group may not be effective for others. Several settings and under-served groups were included in this review, and most trials adopted more than one recruitment pathway. As a minimum, it would be helpful for trialists to report the recruitment rates by under-served group if using more than one recruitment pathway or method in reports. Pre-planned assessments of recruitment and retention methods are encouraged so that good (and bad) practice can be shared and learnt from.

How does this compare to previous reviews?

This scoping review shows that limited evidence is available when assessing what interventions can be done to improve the recruitment of four under-served groups in the UK and Ireland but that letters via the GP seem effective for recruitment of older people, and community engagement and lay advisors can aid recruitment of South Asian populations. A previous review reported strategies in the US for improving inclusion but stated methodological rigor was variable and there were significant evidence gaps (UyBico et al., 2007). A more recent review (Bodicoat et al., 2021), including papers from the US, which reported that no strategy for recruitment was successful across populations and that several methods should be used when recruiting under-served groups. This is also evident in this review where comparisons of recruitment methods were made, with each method recruiting some people and one author highlighting that targeted efforts in recruiting older participants might also improve the recruitment of ethnic minorities.

A previous UK-based review examined the recruitment of South Asians to trials and reported several strategies for recruitment, though did not evaluate them (Masood et al., 2019). One strategy reported was to use lay advisors from the community to help with recruitment, which Douglas (Douglas et al., 2011) found that community engagement and lay advisors were the best method of recruitment of the South Asian community in their trial.

Strengths and limitations

The main strength of our review is its focus. A number of previous reviews have focussed on ethnic minority groups, and on more general ‘under-served’ (Bodicoat et al., 2021), ‘vulnerable’ (UyBico et al., 2007) or ‘socially, culturally, or financially disadvantaged’ groups (Bonevski et al., 2014), whereas we chose to focus this review on four specific under-served groups. We identified papers specifically relating to ethnic minority groups, socioeconomic disadvantage, older people and people with impaired capacity to consent, which allowed us to explore interventions used in different populations. This is important when considering more than one under-served group, which trialists should be doing. There are some common features to the barriers for under-served groups, and further barriers due to intersectionality of under-served groups, that trialists should work to overcome. We also focussed searches to the UK and Ireland to make the findings relevant to the healthcare system, and population in these countries. We used methods to support the systematic approach (Peters et al., 2020) and have reported in line with the PRISMA-SCR reporting guidance for scoping reviews (Tricco et al., 2018).

The review only identified seven papers, which could be due using only one database, and one paper was identified from other sources. Limiting the scoping review to the UK and Ireland limited the number of relevant papers, as there is a work relating to improving inclusivity in trials outside the UK and Ireland (Bonevski et al., 2014; UyBico et al., 2007) which could provide important lessons even if effectiveness could not be translated to a UK and Ireland setting.

We did not include patient and public involvement in the review, so we are interpreting these papers as health researchers, and although some of the researchers may also be members of under-served groups, we did not focus on including lived experience in developing the question for review or interpreting the findings. We are involving PPI in the rest of the ACCESS project.

The interventions identified in this review were included and discussed by the paper authors as they were considered effective in recruiting or retaining their target populations, but there was often no assessment of effectiveness undertaken.

The papers were included due to their focus on a particular under-served group, but the older populations were not necessarily considered as an under-served group in the paper. One of the trials recruiting older people had an age limit of 85, and so excluded part of the older population that is often under-served.

Recommendations and future research

There is a clear need for the interventions undertaken by trial teams with the aim of improving the recruitment of under-served groups to trials to be evaluated. Without rigorous evaluation, trialists are undoubtedly investing time and money into methods that either 1) do not have an effect, 2) have a harmful effect(s) that remains unreported and/or 3) have a beneficial effect(s) that again, remains unreported. Lack of evaluation and reporting means that others cannot build on potential successes to both replicate evaluations in other trial contexts and fine-tune interventions to optimise their effects. Ultimately, this contributes to research waste.

In line with Brown’s 2014 review, we recommend undertaking nested methodological studies within randomised controlled trials to provide this evidence. There are a number of Studies Within A Trial (SWAT (Treweek et al., 2018; Treweek et al., 2020a)) listed on the SWAT repository (https://www.qub.ac.uk/sites/), that could take account of the recruitment of under-served groups by collecting and reporting the relevant demographics. The evaluation of interventions focused on improving recruitment of under-served groups is one of the top priorities for recruitment methodology research (Healy et al., 2018).

As mentioned above, there is also a need for trialists to report on interventions that are currently being adopted with the aim of improving the diversity of participant populations. Retrospective ‘lessons learnt’ papers, although considered lower evidence than pre-planned evaluations, would be a good start to improving the evidence base for potential interventions and could lead to further effectiveness research in relation to inclusion to trials.