Table 3. Interventions discussed in the papers, and the author’s interpretation.

Main theme	Author, date	Intervention to improve recruitment of under-served groups	Author’s findings
Recruitment sites	Agnew et al., 2013	Authors made a conscious effort to approach culturally and socio-economically diverse groups of women throughout the United Kingdom.	Reported that included organisations represented a wide socioeconomic and educational base. No socioeconomic or ethnicity data reported.
	Kolovou et al., 2020	Recruitment took place in two geographical areas: South and West Yorkshire and Southeast Wales. Within these areas neighbourhoods of high socioeconomic deprivation (10% most deprived or 10–20% most deprived) were identified using national deprivation indices.	More than half of the participants lived in the 10- 20% most deprived areas-. Authors reported that their sample may not be representative ethnic minority communities.
	Withall et al., 2020	Three trial sites were chosen that represented urban, suburban, and semirural locations with diverse socioeconomic and ethnic characteristics. Authors over-recruited General Practices (GPs) in diverse areas to allow for an anticipated lower response rate from ethnic groups and the most deprived.	Quintile 1 (most deprived) of the Index of Multiple Deprivation (IMD) = 11.1% compared to 14.3% of the general UK population of over 65-year olds; Quintile 2 = 20.2% recruited, 17.6% in general population. Asian participants = 1.2% recruited, 2.6% in general population. African/Caribbean participants = 3.0% recruited, 1.3% in general population. Caucasian/white participants = 95.1% recruited, 95.5% in general population Other/mixed ethnicities = 0.8% recruited, 0.7% in general population. Male = 33.85% recruited, 45.6% in general population. They reported that targeted efforts could help to recruit more ethnically diverse cohorts.
	Forster et al., 2010	Researchers approached GPs in areas of lower socio-economic status first...	No socioeconomic data reported. No further comments in the discussion.
Recruitment approach	Agnew et al., 2013	Recruitment of community organisations to hold a workshop to recruit participants. Compared four types of workshops (interactive; self-management; interactive & self-management; control).	No differences in recruitment rate between groups. Authors reported that using community organisations for the recruitment of community-dwelling older women their trial was successful. Authors reported difficulties recruiting community organisations, and work is needed in this area to build relationships.
	Douglas et al., 2011	Direct referrals from health care professionals (NHS), primary and secondary care.	Largely unsuccessful, recruited 3% of total referrals (target was 25%).
		Written invitations via GPs to potential participants.	Low success (5.2% of total referrals, target 25%) response rate to letters was resource intensive.
		Written invitation via diabetes register to diabetes patients (to target their relatives).	Unsuccessful, 0 people screed via this method.
		Search of practice lists for patients meeting specific inclusion criteria.	Unsuccessful, 0 people screed via this method.
		Recruitment via research team contacts, self-referrals, and use of the ‘snowball’ effect.	One of three methods, totalling 50% of the recruitment. Author’s reported that word of mouth was particularly successful in Glasgow, and that costs per participant were minimal. The partnerships with the local South Asian organisations and individuals, and referrals by word of mouth from existing participants were the most successful strategy. Snowballing was successful – three recruited participants led to the screening of 140 others.
		Research team recruitment via visits/talks.	One of three methods, totalling 50% of the recruitment. Moderately successful but labour intensive.
		Advertising: Written articles in the press, radio interviews, leaflet and poster distribution, website and e-mail distribution lists.	Not successful in directly enrolling participants.
		Advertising: Ethnic marketing and consultancy company.	Limited success achieved by fieldwork, not mass marketing (1 screened).
		Community organisations and recruiters, assisting with recruitment for small payment.	This is one of 3 settings totalling a target of 50% recruitment. Initially unsuccessful when relying on goodwill, moderately successful when payment offered.
	Kolovou et al., 2020	The researchers recruited from a range of healthcare settings in all identified neighbourhoods. Healthcare venues: GP surgeries, community pharmacies.	Author’s reported that the healthcare settings were challenging and time-consuming to approach and set-up They suggested this was due to the hierarchy in communication, and the complex delegation of responsibilities amongst staff in healthcare settings.
		Community venues: libraries, social clubs, sheltered housing, homeless service centre, community centres and churches.	Community settings had higher percentages of unemployed and self-employed participants, education, employment, ethnicity and deprivation did not differ between settings. Using community settings for recruitment (in addition to healthcare venues) allowed for the recruitment of participants who are not regular visitors to healthcare settings.
	Forster et al., 2010	Recruitment through GPs. GPs in areas of lower socio-economic status were approached first.	Writing directly to potential participants via GPs was the most successful recruitment strategy (195 participants recruited, 90% of total recruitment).
		Recruitment through Barnsley Metropolitan Borough Council.	3 (1.4%) participants recruited. Authors did not request ethical approval to send reminder letters, but suggested that they may have helped the recruitment rate after the initial contact letter.
		Advertising: Recruitment through posters in community groups and 2 advertisements were placed in the local newspaper.	7 (3.2%) participants recruited. No further comments in discussion.
		Recruitment through interviews about the trial by two local radio stations.	0 participants recruited. No further comments in discussion.
		Recruitment through a stand in a local supermarket ASDA and market.	4 (1.8%) participants recruited. No further comments in discussion.
		Recruitment through presentations to a range of groups including the Women’s Institute and Friendship groups.	1 (0.5%) participant recruited. No further comments in discussion.
		Snowballing via participants.	7 (3.2%) participants recruited. No further comments in discussion.
	Withall et al., 2020	Primary care (letters from GPs). GPs were recruited via the UK Clinical Research Network (CRN).	GP practices were the most productive recruitment route (Recruited 682 participants (87.8% of total recruitment)). Some GP practices in diverse areas were already involved in other research that was aiming for a diverse sample and were unable to participate.
		Third-sector organisations: community groups, social enterprises and sheltered housing facilities.	Sheltered housing, recruited = 8, (1.02%) Community partners, recruited =12, (1.5%) Found presentations, relationship building, and meetings with community groups and established partners added only small numbers of participants, while requiring considerable staff resources. But they did find these relationship-based approaches supported recruitment within diverse communities.
		Word-of-mouth, and snowball techniques (friends, relations, or spouses of invitees).	Recruited = 23 (3%). No further comments in discussion.
		Advertising: A supplementary low cost (£726) public relations (PR) campaign.	5.4% of total recruitment figures. £17.29 cost per recruit. No further comments in discussion.
Community engagement	Kolovou et al., 2020	Lay advisors were trained to deliver the intervention and helped with recruitment. To support recruitment they communicated with key stakeholders, identified eligible venues, liaised with local gatekeepers, organised recruitment days, and recruited participants.	The lay advisors thought community recruitment was successful because there was a lot of people visiting the community venues, the visitors had free time on site and were more willing to hear about the trial, and “older visitors” enjoyed talking to the lay advisors. There was no discussion around the impact of the lay advisors delivering the intervention.
	Withall et al., 2020	Local community groups, charities, and the public sector facilitated events to explain and discuss the study with their service users and issued written invitations. A close working relationship was established to achieve this.	Not discussed in relation to raising awareness, used also for recruitment (see above).
Communication between study team and participants	Douglas et al., 2011	The study employed three South Asian bilingual dietitians, two had extensive work experience in the recruitment area	No discussion around the impact of employing bilingual staff.
	Kolovou et al., 2020	Participants were told in advance that the researcher would call from a number from a specific area code.	Not specifically mentioned in discussion. Commented on high retention rates at 2 weeks (90.5%) and 6 months (85.0%).
		Participants were given a general timeframe for their follow-ups.
		Emphasis was placed on the lay advisors’ affiliation with the University (to increase trust).
	Forster et al., 2010	Strategies were put in place to help participants with reading and writing difficulties, such as getting help from partners and relatives and researchers.	This support required extra time which had to be planned for. The authors found that encouragement and reassurance were especially important in help in participants complete the task.
	Withall et al., 2020	Provided funds for translators.	Figures around translation not reported.
		Researchers aimed to build rapport and trust during telephone and face-to-face contacts. Telephone contact was prompt (within 3–4 days) and friendly.	Research staff thought this was one of the critical success factors.
Incentives	Jennings et al., 2015	£100 incentive mentioned in invitation letter (not mentioned in letter for control group, but still given to participants).	Mentioning the £100 incentive did lead to more people to respond positively to an invitation letter (6.9% increase) and resulted in slightly more randomised patients, however, this effect was marginal. The incentive payment did not attract older or participants living in the most deprived areas. Even where a significant improvement was observed, it was not a cost-effective recruitment method.
	Kolovou et al., 2020	Participants were offered a £10 High Street shopping voucher after completing their baseline questionnaires and a £5 voucher for completing the 6-month follow-up.	Authors reported high retention rates of recruited participants at the 2-weeks (90.5%) and 6-months (85.0%) follow-up points. Participants were offered a financial incentive at recruitment that may have impacted on their willingness to take part. The lay advisors highlighted the value of the participant’s contribution to research by participating, which authors thought may have improved trust and reciprocity.
	Forster et al., 2010	Participants were notified about a £100 completion bonus after displaying initial interest in the study..	8/217 people dropped out overall. Authors thought the incentive may have helped with retention.
	Withall et al., 2020	Participants were informed at the time of recruitment that they would receive shopping vouchers - £15 for each of the 6-, 12-, and 24-month follow-up visits.	Retention not reported. Not discussed further.
Flexibility	Withall et al., 2020	Delivery of multiple face-to-face screening sessions, the research team gave participants date/time choices and reimbursed travel expenses for assessments.	Retention figures not reported but authors reported that trial staff thought this was one of the critical success factors.
	Kolovou et al., 2020	Flexibility in recruitment methods: pre-booked appointments or opportunistic recruitment. Lay advisors adopted opportunistic recruitment, approaching individuals in a community or healthcare setting.	Recruitment was restricted to weekdays and working hours which may have impacted on the recruitment of working adults. Authors noted that community venue staff may have inadvertently biased recruitment.
		Participants were asked their preferred time (weekend/weekday,morning/afternoon/evening) and method of contact (phone call, text message, e-mail, post) for their follow-ups.	Authors reported high retention rates at the 2-weeks (90.5%) and 6-months (85.0%) follow-up. A personalised flexible follow-up approach may have allowed for added trust and reciprocity between the participants and the researcher lay advisors.
Participant-facing written materials	Kolovou et al., 2020	All patient-facing study materials were written in line with national average literacy levels and were reviewed by the trial’s Patient and Public Involvement group prior to recruitment.	Retention rates were high, indicating people were fully informed of what the trial involved. Authors reported that a study limitation was the lack of translated materials and language support, as this contributed to limited recruitment of people with limited English language.
	Withall et al., 2020	Patient documentation: In the pilot, the Participant Information Sheet (PIS) was sent out on receipt of a response/enquiry form but this was changed in the main trial to save time and effort. The invitation letter was changed to provide a much more noticeable required participant profile in a large, prominent text box, and the PIS was sent with the invitation letter.	The response rate to the initial invitation letter was 8% lower in the main trial than the pilot study but a much higher proportion of responders progressed through telephone screening to face-to-face assessments compared to the pilot study (43.3% vs 27.2%), indicating that more eligible candidates responded. Authors reported that making eligibility clear, prominent and in plain language helped with this.
Inclusion criteria	Withall et al., 2020	The change of trial’s inclusion criteria to SPPB 4–9 from 4–8 widened the participant pool to include frail and prefrail populations were eligible for the trial.	Figures before and after this change were not provided. Authors reported that this positively impacted inclusion rates at the face-to-face screenings.
Consent process	Jayes & Palmer, 2014	Consent Support Tool (CST). A tool to facilitate the involvement of people with communication disorders.	The results show that the CST can be used to accurately identify the best information style for participants.