Recruitment and Retention of Underrepresented and Vulnerable Populations to Research

Abstract

Per principles outlined in the Belmont Report, research involving human subjects should minimize risks to participants and maximize benefits to participants and society. Recruitment of participants should be equitable. Once enrolled, participants have the right to withdraw at any point. Researchers must balance these principles with pressures to meet enrollment goals and, in the context of repeated-measures designs, retain participants across time. The purpose of this perspective is to describe the approach and corresponding activities for recruiting and retaining underrepresented and vulnerable populations that are the focus of a transdisciplinary academic research center. To this effort, we offer diverse disciplinary backgrounds, experience working with a wide range of populations (from infants to older adults and across multiple health conditions), and spanning a variety of research designs. Effective strategies offered include partnering with community entities, approaching potential participants where they are and at a time of readiness, using population-appropriate modes of communication and data collection, conducting study activities in familiar settings and at convenient times, maintaining frequent contact, and offering meaningful incentives. These strategies are consistent with population-specific reports found in the extant literature and underscore their cross-cutting nature, with adaptations based on participant and community partner needs and preferences.

National Institutes of Health-funded clinical research investigations are expected to include women and racial-ethnic minorities as participants unless this is inappropriate for the research question (U.S. Code, 2012). In addition, policies are in place to ensure that individuals of all ages are enrolled in research (National Institutes of Health, 2017). The rationale behind these policies, apart from the importance of inclusivity, is to allow for greater generalizability of findings. Although clearly a step in the right direction, the observed disparities in many chronic and preventable diseases suggest that there is more work to be done to ensure that advances in biomedical research are realized across the general population. Certain sectors in the American population remain underrepresented in research (e.g., people of color, women). This underrepresentation and historic neglect perpetuate inequities that sustain high risks for adverse health outcomes and diminished well-being (Braveman, 2006; Trinh-Shevrin et al., 2015).

The purpose of this manuscript is to describe the approach and corresponding activities for recruiting and retaining underrepresented and vulnerable populations to research that are the focus of a collaborative, transdisciplinary academic research center. Following the All of Us genomic-wide population assessment research program definitions, we use the term underrepresented to refer to individuals who identify as other than White and non-Hispanic, children, older adults, individuals with a physical or cognitive disability, individuals with less than a high school degree, individuals with a household income at or below 200% of the Federal Poverty Level, individuals who have not had a needed medical visit in the past year or cannot obtain or afford such care, and individuals who reside in non-metropolitan areas (Mapes et al., 2020). (See Mapes et al. for a full list including characteristics based on gender identity and sexual orientation.) Per Sivayogi (2013), vulnerable populations refer to, among others, children, pregnant women, military personnel, ethnic minorities, refugees, and individuals who are elderly, terminally ill, physically and intellectually challenged, or economically and educationally disabled. Recognizing overlap, we use the term vulnerable to describe persons who do not meet the underrepresented definition but who may be medically fragile (e.g., due to advanced disease) or at risk for poor health outcomes due to traumatic exposure such as sexual assault or military combat. We appreciate that the term vulnerable may be controversial (Katz et al., 2020). Our use should not be construed to imply that these individuals are less resilient. Instead we recognize that their health risks are imposed by contextual and structural inequities.

Previous work on recruitment and retention of underrepresented and vulnerable populations has focused primarily on studies involving people of color (Dennis & Neese, 2000; Dreer et al., 2014; Jones & Broome, 2001; Rhodes et al., 2018; Tanjasiri et al., 2015; Villarruel et al., 2006; Wallace & Bartlett, 2013; Yancey et al., 2006), people residing in rural areas (Davis et al., 2009; Parra-Medina et al., 2004), and/or those of low income (Ejiogu et al., 2011; Engler-Stringer et al., 2018). Accordingly, recruitment and retention strategies presented in these articles are population-specific. Other work has either focused on a particular health condition or centered around a specific intervention target such as weight loss or physical activity (Ashing-Giwa & Rosales, 2012; Carroll et al., 2011; Marquez et al., 2020; Stevens et al., 2017). Still others have focused on barriers (Ford et al., 2008), with less attention to how to successfully recruit and retain.

We seek to fill the aforementioned gaps by taking a more solution-focused and comprehensive approach to the topic. As faculty members within a nursing college at a large public university, we are broadly focused on health promotion and disease prevention across the translational research spectrum. While we share this common theme, we come from diverse professional backgrounds: nursing, psychology, behavioral science, public health, kinesiology, and communication. Our research investigations, moreover, involve a wide variety of participants reaching across the lifespan, sociodemographic subgroups, and health states, while employing cross-sectional, longitudinal, and experimental designs. See Table 1 for an overview of our research programs. In what follows, we use exemplars from our relatively recent work to share best practices to recruit and retain underrepresented and vulnerable populations to research. We conceptualize recruitment as steps taken to identify, approach, screen, and enroll participants, and retention as steps taken to collect follow-up data in the context of studies involving more than one assessment timepoint.

Table 1.

Overview of Research Programs

Priority populations	Latinx youth and families
	African American women
	Racial-ethnic minority immigrants and refugees
	Women and children of low income
	Individuals with mobility impairing disabilities
	Individuals with cancer and their caregivers
	Victims and perpetrators of sexual assault
	Veterans
Recruitment sources/ sectors	Community
	Clinics/ health systems
	Schools
	Social media
Disease prevention aims	Prevent type 2 diabetes
	Prevent obesity
	Prevent cancer
	Prevent addiction to illegal drugs/ other substances
	Prevent cardiometabolic disease
	Prevent HIV and sexually-transmitted infections
	Prevent sexual violence
Health promotion aims	Promote physical activity
	Promote health screenings
	Promote HPV vaccination
	Promote sleep hygiene
	Promote healthy growth, healthy eating, and breast feeding
	Promote adaptive communication and relational well-being
	Promote emotion regulation
	Promote mental and physical health and well-being

Strategies for Effective Recruitment

Table 2 lists 1-2 studies conducted by each author and their respective research team, projects conducted within the past five years and representing populations the investigator typically works with. For each study, we briefly describe the sample, research design, and strategies deemed effective for recruitment. Below we synthesize this content.

Table 2.

Effective Recruitment Strategies

Reference	Sample	Design	Effective recruitment strategies
Castro et al. (under review)	35 Latinx males and females with pre-diabetes	Cross-sectional, mixed methods	Developed a balanced sampling frame for equity based on gender, age, and acculturation status Identified eligible individuals from a diabetes registry and contacted them by letter of invitation and email The study’s bilingual-bicultural Project Director contacted interested individuals via telephone, provided informed consent information, and invited them to an in-person session At this in-person session, individuals were read the informed consent in their preferred language. Upon signing, they could participate in the interview at that time or at another scheduled time. To establish rapport, and recognizing the vulnerability of low-acculturation individuals and low literacy in some, for data quality, we administered all instruments via interview.
Chen et al. (under review)	513 racial-ethnic minority nurses	Cross-sectional	Partnered with key nursing associations/organizations (e.g., National Association of Hispanic Nurses, Asian American/Pacific Islander Nurses Association, Philippine Nurses Association of America), particularly critical to reach racial-ethnic minority nurses. Utilized user-friendly platforms for distressed frontline workers amid the SARS-CoV-2 pandemic. Participants could access online survey via QR code or link in flyer. Included in recruitment flyer statements to thank and honor nurses’ significant contributions and efforts put into patient care during the pandemic Consulted with nurses to determine appropriate incentive
Petrov et al. (2017)	19 adolescent, female victims of sex trafficking and/or abuse at an in-patient group home	Cross-sectional	Investigators and staff completed rigorous screening process to become volunteers at the group home Partnered with CEO and directors of the group home Approached participants during regularly scheduled programming Offered needs assessment for healthy lifestyle programming Asked each participant for preferred gift within a low-denomination compensation package
Langer et al. (2018)	40 hematopoietic cell transplant recipient/caregiving partner dyads	Intervention; single-group pre-post	Partnered with provider willing to champion study Met participants where they were; scheduled approach and consent sessions through clinic to accommodate medical appointments Approached patients and partners at presumed time of readiness relative to transplantation, avoiding difficult inpatient hospitalization phase and period of extreme patient immunosuppression Couple-based intervention afforded psychosocial support not just for patients but for caregiving partners/ spouses as well; this stood out from the typical medical studies offered.
Davis et al. (2019)	503 women with elevated sexual risk history	RCT	Advertised on social media platforms and print media frequently used by young women Recruited through flyers in local venues frequented by young women, including coffee shops, beauty salons, and sexual health clinics Provided postcard-sized flyers to participants to hand out to their friends, improving word-of-mouth recruitment Recruited staff of a similar age and gender to potential participants to increase comfort level
Davis et al. (2020)	209 men with self-reported sexual assault history and heavy episodic drinking	RCT	Placed ads in venues targeting young adult drinkers, such as bars and local happy hour locales Conducted eligibility screening via online survey versus telephone, for greater sense of anonymity Offered flexible session scheduling, including weekend options Staff trained to fully describe protection of participant privacy and confidentiality Offered transportation vouchers for participants to attend in-person session
Joseph et al. (2019)	60 insufficiently physical active African American women with obesity	Pilot RCT	Used paid online social media advertisements (i.e., Facebook) that linked to study website; website provided detailed information about study and access to online screening tool. Placed paid advertisements in African American-focused newspapers and social networking websites Promptly followed-up with interested individuals (i.e., ideally within 24-48 hours of initial contact) to further describe study purpose and further screen for eligibility Created and utilized manualized telephone recruitment scripts to provide detailed overview of study procedures, setting clear participant expectations
Kim et al. (in preparation)	96 hematopoietic cell transplant recipient/ primary caregiver dyads	RCT	Collaborated with treating cancer center and providers Approached patients and caregivers during an already scheduled post-transplant outpatient visit Emphasized the dyadic aspect of the study, supporting not just patients but primary caregivers
Menon/ Larkey et al. (2020)	419 unscreened or due for colorectal cancer screening	RCT	Hired study recruitment/implementation staff from the communities to be reached Engaged community sites in low income zip codes to obtain service population sociodemographic characteristics (age, race, ethnicity); this informed selection of sites to maximize targeted enrollment Adapted initial prospective participant contact contexts based on community site leaders’ suggestions (e.g., lunch gatherings at senior centers) Included broad types of community venues, adapting message strategies to reach the intended age group: churches, schools, community centers, residential centers Included novel public sites, such as fitness centers, food trucks, and farmer’s markets
Lee, Reese-Smith et al. (2017)	310 African American and Hispanic or Latina overweight/ obese, sedentary women	RCT	Employed multiple modes of approach/ advertising Face-to face announcements at community events and meetings (e.g., church or university campus meetings, health fairs) Advertisements or announcements with local newspapers or magazines Advertisements or announcements on a listserv or on the Internet Advertisements or announcements on television or radio
Lee et al. (2017)	89 children or their parents attending early care and education centers in neighborhoods with high proportions of individuals who identify as Hispanic	Cluster RCT	Recruited mothers of preschoolers (with child assent) from already identified sites via classroom announcements, flyers, and teacher encouragement
Reifsnider et al. (2018)	175 Mexican American mother/infant dyads	RCT	Partnered with county health department and their WIC clinics Provided presentations to WIC staff about study and socialized with them Hired promotoras (clinical health workers) who were bilingual/bicultural in English/Spanish Promotoras recruited participants when the women came for their WIC enrollment WIC staff promoted study to WIC participants and collaborated in allowing promotoras time to recruit inside and outside of clinic All recruitment materials in English/Spanish Recruitment tables prominently displayed at WIC clinics Promotoras were pro-active in recruitment and inquired of pregnant women receiving clinic services if they were interested in the study Obtained county WIC approval to recruit at additional WIC clinics to meet recruitment goal
Vander Wyst et al. (2020)	117 Latino adolescents with prediabetes	RCT	Collaborated with established and trusted community entities as co-investigators Employed bilingual/bicultural research staff and recruitment materials Cast a wide net of referral sources across multiple sectors (clinic, community, and media) Clinic efforts led by project physician with ongoing engagement Established Business Associate Agreements with health centers to share protected health information for consented participants Conducted recruitment presentations with physicians, nurses, health educators, and registered dietitians to establish relationships, provide project updates, and coordinate recruitment activities Sent monthly reports to clinics to track all patients who were referred Community efforts led by community partners (community coalitions, health fairs and churches) Media efforts advertised through Spanish language outlets and posted through social media Employed snowball sampling (word-of-mouth from past participants and families) Provided results of laboratory testing back to participants Focused on a health topic highly relevant to priority population Provided access to health promotion intervention otherwise not available Offered weekend screening and enrollment visits Offered monetary (cash) reimbursement for time and travel Provided transportation and parking
Youngstedt et al. (2021)	71 veterans with PTSD	RCT	Veteran served as a research coordinator, helped with feelings of trust and credibility Received referrals from mental health professionals at the VA Placed study ads in military newspaper Placed study flyers throughout mental health clinic at VA

Note. PTSD = post-traumatic stress disorder; RCT = randomized controlled trial; WIC = women, infants and children. Terms used to describe samples with regard to race and ethnicity differ across studies due to variations in usage and preference, and to denote specificity when possible, e.g., Mexican American.

Partnerships

Community-based participatory research (CBPR) approaches emphasize bi-directionality and equal power between academic and community partners (Wallerstein & Duran, 2010). Partnerships are genuine and committed, learning is collaborative, and a general goal is capacity building to benefit community partners (Israel et al., 1998; Wallerstein & Duran, 2010). While few research investigations can fully meet these requirements, many strive to incorporate elements of CBPR principles into their approaches. Indeed, a key strategy in our research practices has been to partner with community entities, ranging from schools to clinics and health systems, to religious and nonprofit community organizations. These partnerships facilitate access to referral and recruitment sources (sometimes to populations otherwise unreachable) and present an opportunity to build capacity in the community. An exemplar collaboration that has spanned over a decade and multiple National Institutes of Health-funded studies is a collaboration with the Ben and Catherine Ivy Foundation Center for Family Wellness. Housed within a community-based medical clinic that provides care for the uninsured and working-poor, the center is focused on health promotion and disease prevention for those at highest risk. Over the years, the collaboration has grown to include additional community partners, securing philanthropic funding for community infrastructure, hosting policy forums, fostering the next generation of community engaged scholars, and building a pipeline of bilingual/bicultural registered dietitians to address diabetes disparities in the community. Collectively, these collaborations have advanced scientific understanding of diabetes risk and prevention (Soltero et al., 2019) and created new community resources to promote health equity.

In addition to knowledge acquisition and capacity building, pragmatic support for partners in the form of funding can be extremely helpful. For example, if a study can provide a sub-contract with a community-based agency where space can be used under a “rental agreement,” that agency may be more accommodating and see this as a good-faith effort to contribute resources, rather than simply coming in to gather data without sensitivity to the agency’s operating expenses. Similarly, staff members within partnering agencies or clinics are more able to facilitate recruitment or other project activities if they are appropriately compensated for their effort through partial salary support, and if institutional leadership is supportive of their involvement and finds value in the project.

Approach Timing and Context

The context in which potential participants are approached is critical, as is the timing. A successful strategy across our research programs has been to approach potential participants during “regularly scheduled programming” both literally and metaphorically, in other words, where they are. We attend face-to-face community events such as health fairs or church gatherings frequented by the population, or present the study at program events (Larkey et al., 2008). For example, in studies involving persons undergoing hematopoietic cell transplantation, a rigorous treatment for hematologic malignancies, we have had success presenting a brief overview of the study at a required pre-transplant class for patients and their family caregivers held at the transplant center (Langer, Porter, et al., 2018). To optimize such contacts, when possible, having the partner make the initial approach promotes a sense of trust. This strategy, however, has to be weighed in the context of avoiding the appearance of coercion (e.g., if treating physicians present studies to their patients). It may be more suitable to have providers mention the study, then bring in someone not directly involved in services provision to describe the study. Alternatively, supportive care providers may present the study, e.g., a dietician if the study involves a nutritional intervention.

It is also critical, when approaching potential participants where they are, to be sensitive to the reason that they are there. Medical and other needs must come first. Relatedly, it is important to approach potential participants at a presumed time of readiness. Are they at a point in treatment, for example, known to be especially difficult or stressful? Can research activities be delayed given the population’s needs? If the study is an intervention, it is helpful to consider the stages of change model that describes five stages of readiness: precontemplation, contemplation, preparation, action, and maintenance (DiClemente & Prochaska, 1998). This framework can inform not just intervention development but point of approach, recognizing of course that individual differences will exist.

The manner in which an initial approach is made also matters. Studies that prioritize persons of Latinx ethnicity, for example, have benefitted from having recruitment materials in both Spanish and English. In-person approaches at community events may make sense for certain populations, although online approaches may be preferable for young adults. Eligibility screening that requires the collection of sensitive information is best conducted using modalities that feel more removed, such as online. Regardless of modality, it is critical that screening staff be trained in the collection of this information such that they respond appropriately to any concerns or distress during the screening process. Lastly, prompt follow-up to initial expressions of interest, ideally from a relatable staff member (someone of the same race, ethnicity and/or age group), is important to capitalize on initial interest and to convey time sensitivity and responsiveness.

Advertising Modality and Placement

Community-based recruitment typically requires advertising. Tailoring advertisements to the population of interest involves attention not just to the wording and look of the materials but to the modality and placement. Studies on veterans, for example, have had success with advertisements in military newspapers. For younger adults and increasingly a large segment of the general population, social media is an effective way to cast a wide recruitment net. Although these avenues afford significant reach, they can require a great deal of personnel resources to follow up with the high response rate.

Despite the rise in web-based recruitment, hard copy flyers placed in relevant community settings and tailored to the population of interest still have utility. For example, in a study recruiting young adults who consume alcohol (Davis et al., 2020), flyers placed in local bars were effective. Similarly, placing flyers in wig shops set up especially for cancer patients has helped us reach breast cancer survivors (Larkey et al., 2015). Snowball sampling has also been a useful strategy where current or past participants are able to represent the study in authentic ways to their network based on their own experience (Vander Wyst et al., 2020).

Consent Process

Consent sessions, be they in person, by phone, or web-based, must cover certain elements of the study, namely the purpose, procedures, risks, and benefits. One approach has been to create manualized scripts and content. These materials are typically vetted by the IRB and thus are well-crafted and easy to understand. A script also ensures consistent messaging and can ease burden on recruiters. Another approach is to adapt delivery of consent content based on participant preference and/or learning style. Some individuals may benefit from clear visuals to aid understanding of key timepoints or activities. Others may prefer to receive study information in Spanish versus English. Some low-acculturated Spanish speaking adults have low literacy skills. In this case, an in-person reading of the informed consent is important to avoid any misunderstandings.

While it can be advantageous to capitalize on initial interest, it is also important to allow time as needed for individuals to decide whether to participate. Eligible individuals may need and want to consult with family members at home, or to take more time to digest. If conducting an in-person consent, offer to leave the room and let family members discuss the pros and cons of participation in private, especially if the study involves family members. In recruiting dyads, it has been helpful to acknowledge up front that both parties may not want to participate and that neither should feel coerced to do so. Alternatively, there may be scenarios whereby private consents are indicated, for example, in studies involving survivors of interpersonal violence.

Staff Characteristics

An effective strategy in many of our research programs has been to hire research staff from the priority population or community of interest. Having as front-facing members of the research team individuals who are perceived as culturally competent can facilitate comfort with study personnel and activities, e.g., a veteran for a study recruiting veterans, promatoras/es (lay health workers) for a study recruiting Latinas. These individuals will likely communicate using sociocultural norms which may provide potential participants with a greater level of ease. In the case of non-English speaking participants, bilingual and bicultural research coordinators can help facilitate comprehension of study documents. Involving students from the priority population as team members can help achieve the same aims and establish a pipeline for diversifying the scientific workforce.

Incentives

Incentives are often mentioned in the recruitment process, in flyers and other advertisements, in consent documents, and can motivate participation. Financial incentives are most common and can help to offset perceived costs of participation, either from effort put forth or time taken away from work or family commitments to participate. Cash is almost always appreciated though some institutions do not allow cash incentives or may require documentation of social security numbers to process payments. Regardless of local institutional policy, the IRS considers payments greater than $600 in a calendar year as reportable income, which requires additional paperwork. These forms may present a barrier to entry for undocumented populations. As an alternative to cash, grocery store and “super store” gift cards, the promise of diapers for mothers of infants or books and toys for young children fill a pragmatic need and may be particularly incentivizing to those of lower income.

Incentives need not necessarily be financial. Offering transportation vouchers or parking in the recruitment process can defray perceived travel burdens. Intervention studies may offer medical or supportive care for a condition of concern, or may provide results of laboratory tests. For some participants, these treatments or tests might not otherwise be possible. Lastly, offering to involve family members in interventions or making it clear at the outset that family members may accompany participants to study sessions can go a long way to easing participation, providing free benefits, and thereby facilitating recruitment.

Strategies for Effective Retention

Table 3 describes effective strategies for retention in the context of studies involving more than one assessment timepoint.

Table 3.

Effective Retention Strategies

Reference	Sample	Study design	Effective retention strategies
Langer et al. (2018)	107 cancer survivor/ partner dyads	Smartphone-based ecological momentary assessment (twice-daily assessments for 14 days)	Piloted smartphone application to assess usability and acceptability; altered design and timing based on feedback Met face-to-face with participants to support download of smartphone app required to complete twice-daily assessments, extremely helpful for older and less tech-savvy participants; fully explained process. Sent couples home with user-friendly guide with clear visuals Remotely monitored data completion; contacted non-responders to help troubleshoot. Incentivized data completion per assessment plus extra if completed > 85% of assessments
Joseph et al. (2019)	60 insufficiently physical active African American women with obesity	Pilot RCT	Consent process included manualized orientation session (structured PowerPoint presentation outlining study purpose and procedures, expectations of study participants, and what participants could expect from research team) Contacted participants bi-weekly to inquire about any issues/ problems encountered with study smartphone app Compensated participants for parking and other transportation-related costs incurred (i.e., bus or light rail fare) for attending assessment sessions Scheduled follow-up assessments on days/times convenient for participants, including early morning and weekends Delivered cash incentives at the time of assessment Sent text and email reminders for scheduled study assessments
Kim et al. (in preparation)	96 hematopoietic cell transplant recipient/ primary caregiver dyads	RCT	Maintained strong rapport with participants at baseline and provided clear explanations of follow-up plans Administered assessments remotely via REDCap, easing participant burden Sent weekly emails and phone calls for reminders and to maintain current contact information Provided in-person orientation to facilitate use of web-based intervention Incentivized data completion per assessment
Lee, Reese-Smith et al. (2017)	310 African American and Hispanic or Latina overweight/obese, sedentary women	RCT	2-week run-in phase that included the completion of a packet of questionnaires, health behavior monitoring logs, and attendance at a pre-randomization session to determine commitment At pre-randomization orientation, asked participants to consider importance of the study and possible reasons for dropout Asked to withdraw before randomization if not willing to be randomized
Lee et al. (2017)	89 children or their parents attending early care and education centers in neighborhoods with high proportions of individuals who identify as Hispanic	Cluster RCT	Sent weekly infographic-style newsletters to parents describing their child’s activities in school and family-friendly home and community activities that could enrich in-school learning
Menon/ Larkey et al. (2020)	419 unscreened or due for colorectal cancer screening	RCT	Contacted both intervention and control group participants (nearly) weekly by phone to set goals or investigate barriers to continuing through steps of the study to obtain colorectal cancer screening Adjustments were made to address individual participants’ insurance and clinic status (e.g., if no medical home, recommended nearby federally qualified health center) Addressed changes in medical reimbursement and immigration policies (e.g., new laws affecting racial-ethnic profiling) by building relationships, remaining transparent, asking no questions regarding social security ID, continuing to provide services
Reifsnider et al. (2018)	175 Mexican American mother/infant dyads	RCT	Maintained contact with participants through texts and phone calls Conducted all study visits in the home, negating participant travel Contacted participants one day before home visit to remind and confirm time Trained promatoras conducted all study visits in participant’s preferred language Promotoras created friendships with families, often invited to family celebrations Provided cash incentive at each visit Provided non-cash incentive (book, toy, etc.) to intervention group participants Mailed birthday cards to children Entered active participants in a $50 gift card raffle every 6 months (those remaining in 3-year study) Hosted fiesta at end of study (food and fellowship) Honored completers with “graduation” certificates
Soltero et al. (2018)	160 Latino adolescents with obesity	RCT	Maintained regular contact by phone, email, or text based on participant preference Sent automated text messages for birthdays and holidays Allowed family household members to attend intervention sessions Provided childcare during intervention Sent automated appointment reminders 1 week, 3 days, and the evening prior to scheduled appointments; included images for visit instructions and parking directions Offered tiered incentive program that included intervention participation and completion of all study-related visits and procedures Provided updated information on incentives as intervention goals were met
Youngstedt et al. (2021)	71 veterans with PTSD	RCT	Weekly assessments allowed for frequent interaction Compensated participants at multiple time points Connected regularly with participants via telephone, provided encouragement When possible, the same research team member communicated with a given participant throughout the study

Note. PTSD = post-traumatic stress disorder; RCT = randomized controlled trial. Terms used to describe samples with regard to race and ethnicity differ across studies due to variations in usage and preference, and to denote specificity when possible, e.g., Mexican American.

Orientation

It is imperative to set clear expectations regarding study participation from the start. The study must be explained in ways that are cogent and easy to comprehend. Visuals and laminated cards that illustrate key timepoints and “asks” can augment written and oral consent information provision. When possible, guide participants through any potentially challenging study activities in person. For example, for studies involving accelerometers, demonstrate to participants how the device works and set clear reminders for when devices need to be charged or when data downloads must be triggered. For studies involving smartphone applications, guide participants through download of the app to their phones. For some older adults, this has meant helping them to generate an email address for the study. Provide user-friendly guides with pictures for participants to take home and consult as needed. We have found that this sort of “front-loading” can go a long way to ensuring comfort with technology and avoiding frustration down the road.

Assessment of Fit

Some studies in our center have utilized a run-in phase prior to randomization. This allows the opportunity for both the participant and research team to determine if the study and expectations for participation are a good fit for the participant prior to randomization and deployment of staff and other resources necessary for intervention delivery. For example, in a physical activity promotion study for African American women at risk for cardiometabolic disease (Joseph et al., 2019), participants were asked to wear a study-provided accelerometer for 7 days as part of the baseline assessment. Participants unable to meet compliance expectations (operationally defined as wearing the accelerometer for at least 10 hours per day for 4 of the 7 days) were given a second 7-day trial. If compliance expectations were still unmet, a mutual decision, achieved through respectful and thoughtful dialogue between the participant and study staff, was made that the study was not a good fit for the participant at that time.

Frequent Contact

Consistent contact with participants helps to sustain interest in and commitment to the study. It lets them know what activities will happen when, and may provide a caring exchange at a time of need. For chronically underserved populations, particularly those having pervasive mistrust of research or outside authority figures, frequent contact and continued expressions of care can build much needed trust. The preferred method of contact may differ within or between populations so it is helpful to ask at the time of enrollment what mode participants prefer and to use that method throughout. Young adults may prefer automated text messages whereas middle- and older-aged adults may prefer email reminders or phone calls.

Regardless of preferred method, some research teams in our center send holiday or birthday cards in the mail (or in the case of bereavement, condolence cards). These mailings provide an opportunity to engage with participants and to make sure that the home address on file is correct; low-income populations may move frequently because of unstable housing. The rise in cell phone ownership and use may have lessened concerns for maintaining contact; however, minorities and individuals from low-income populations are more likely to have intermittent cell service and/or to change their contact number frequently (Pew Research Center, 2019). We have also found it helpful to collect multiple contacts from participants including friends and extended family members who may be able to facilitate recontact in the event that the research team is unable to do so. Other forms of contact include a study newsletter, website blog postings or, once the study is completed, a summary of results. We find that all can be meaningful to participants, connecting them to the larger purpose and progress of the study, and showing them that their contributions are being put to good use. Aggregate-level data conveyed in layperson terms can also help to normalize thoughts, feelings, or behavior patterns.

The need for frequent contact applies to community and clinic-based partners as well. Keep partners appraised regarding study progress and solicit their input to address any obstacles or barriers. Further, once data collection is complete, enlist partners to help interpret results and to draw implications. This can inform local practice or programs and provide additional insight into expansion or adaptation of the research.

Timing and Mode of Study Activities

A recurrent theme across our collective research efforts is a clear need to be flexible in the timing of study activities. Participants may work full-time or have multiple part-time jobs and/or have child-care needs. When children are the priority population, school schedules are an additional consideration. Others may be asked to participate because of their caregiving status but the very nature of that status renders leaving the house or stepping away from their loved one difficult. It is incumbent upon researchers to ease and facilitate participation by working around participants’ schedules. Offer evening and weekend options for study activities if possible, recognizing the need to also consider and accommodate the schedules of community partners or partnering entities, e.g., clinic hours. However, when possible, setting a regular schedule is helpful for staffing and helps establish commitment to the study.

Just as recruitment is best conducted by meeting participants where they are, so too is retention. Study visits conducted in the home setting may be preferable for some. For example, in a study of mother-infant dyads, poor attendance at clinic-based intervention sessions to prevent obesity was the impetus for a subsequent project offering home-based intervention visits instead. This was well-received as it negated needs for transportation and childcare for any children not in the study (Reifsnider et al., 2013). It is still important, though, even with home-based sessions, to flexibly schedule sessions at times convenient for families.

Technology can be used to great advantage as evident in the growing number of ehealth and mhealth interventions. This is highly appropriate for some populations who are both comfortable with the modality and have access to the technology. Young adults, for example, often prefer to complete assessments via cellphone. While cellphone ownership is almost ubiquitous in the US (Pew Research Center, 2019), access to data or WIFI may be limited or transient. Increasingly, research-ready apps and devices allow for data acquisition and retention in the absence of WIFI or require only periodic charging or uploading. Clearly, the simplest approach is the one that asks as little as possible from participants. Pilot-testing, optimally with user-centered designs, is critical to understand what is acceptable and compelling for the priority population, including look, feel, and timing of prompts. Alternatively, consider familiar or low-tech options, such as paper-and-pencil surveys for those who might prefer these modalities and for whom delivery modality might make the difference between completing a given study activity and not doing so. For some low literacy populations, even a paper-and-pencil approach may yield unreliable data when a survey is simply handed out. Reading surveys aloud to these individuals may be necessary to obtain accurate responses.

Incentives

Just as incentives play a key role in recruitment, they also play a role in retention. Timing of payments for data completion matters. We have found it helpful across studies to pay per assessment and in real-time versus in a delayed fashion. This provides immediate gratification. Tiered incentives for assessment completion, if approved by the IRB of record, can maximize retention. This may mean offering a greater incentive for the last follow-up assessment (e.g., $25 for first assessment and $40 for second assessment) or offering a bonus for a certain level of adherence. An additional factor to consider is whether compensation for participation should be adjusted for inflation over the course of a study. Taking a 5-year NIH R01 as an example, the time between initial IRB-approval to final enrollment may span 4 years and although many institutions provide year-to-year salary increases for personnel to adjust for inflation, participant compensation likely remains unchanged.

Non-monetary rewards can also be effective. One study hosted a fiesta for participants to celebrate completion, and handed out “graduation” certificates. Another study in process is offering “swag bags” with population-relevant content, for example, diapers and sleep masks for mothers of infants enrolled in a study on sleep. If commercial physical activity trackers are used as part of the research (i.e., Fitbits), can the study afford to let participants keep them? If investigations involve a health screening, can those results be conveyed to participants in ways that are not harmful or that do not compromise integrity of the research? Such information may be viewed as highly valuable to individuals and families, especially to those who might not have regular access to care. A more distal benefit that some participants appreciate is the mere act of contributing to understanding of a health condition of great personal relevance. This notion of research altruism (Carrera et al., 2018) may be especially relevant for those at later stages of life (Dahlin-Ivanoff et al., 2019) or disease (Prakash et al., 2018).

As mentioned in the context of recruitment, providing travel vouchers and parking can offset transportation concerns. This is equally true as the study goes on. Any perceived barrier is a risk to continued participation. It is incumbent upon the research team to recognize and remove these barriers to the extent possible. Time costs for low-income versus high-income populations vary dramatically. In addition to disparities in income, complexities to consider that may make it more inconvenient for low-income populations to participate in research include a higher likelihood of unpredictable/unstable work schedules, lack of paid time off, childcare, and transportation challenges. Therefore, how to determine “time-cost” for participating in a research study is not straightforward and researchers could benefit by seeking input from community advisory boards to balance issues of coercion and undue influence in the face of social determinants of health. It may be prudent to consider a sliding scale for compensation whereby low-income populations receive higher compensation rates to overcome additional barriers to participation (Williams & Walter, 2015).

Discussion

Overlap with extant literature

As noted previously, much of the prior literature around recruitment and retention has focused on a single population. The strategies described herein, based on experience with multiple populations, are in many ways similar to those noted elsewhere. Overlapping themes include the critical role of community partners and gatekeepers; similarity between front-facing study personnel and community members; use of clear, engaging, and tailored recruitment materials; convenience and familiarity in location of study activities; flexibility in timing of activities; frequent contacts and reminders in preferred modalities; consent processes allowing time for decision-making; and incentives ranging from cash and gift cards to transportation vouchers, childcare, certificates of achievement, and population-relevant gifts (Bailey et al., 2004; Berry et al., 2013; DeFrank et al., 2019; Di Noia et al., 2019; McNeely & Clements, 1994; Rhodes et al., 2018; Wise & Cantrell, 2019). Incremental compensation has also been deemed effective by others (DeFrank et al., 2019; Di Noia et al., 2019), as has a non-monetary approach, namely tapping into altruistic motives (Di Noia et al., 2019). The extent of this overlap lends support for our recommended practices and speaks to their cross-cutting nature, with recognition of course that tailoring is built into specific recommendations. As noted by Rhodes et al. (2018), creativity and persistence are needed among both study team members and partners to effectively recruit and retain underrepresented populations to research. Creative examples include offering gift teasers prior to enrollment (Rhodes et al., 2018) and sending “missing you” letters to non-responders (McSweeney et al., 2009). It is important to share not just study results but study approaches as well. These details are often excluded from journal articles.

Limitations

The purpose of this manuscript was to offer recruitment and retention strategies deemed successful across a wide variety of study populations, spanning infants to older adults, and across a wide range of research designs and methodologies. In taking this broad approach, we admittedly lose some contextual and methodological detail. It was beyond the scope of this paper to provide specific recruitment and retention rates and to describe the different populations in detail with respect to baseline sociodemographic or other characteristics. Also, studies presented in Tables 2–3 represent exemplars based on studies conducted within the past five years and illustrate populations representative of our larger research programs. Thus, they do not represent an exhaustive list of all the research conducted by the authors. A third limitation is that we did not address engagement. We see engagement as a process that is critically important for intervention research, where adherence and fidelity are maximized to facilitate uptake, relevance, and efficacy. Readers are referred elsewhere for strategies to enhance engagement and sustainability of interventions within diverse populations (Castro & Yasui, 2017).

Future Directions

Focused efforts to recruit and retain populations historically underrepresented in research are of paramount importance to enhance generalizability of findings as well as to promote health equity. Future research focused on leveraging technology and social networks to enhance recruitment and retention may further close the disparities gap. An emerging area that warrants additional attention is the importance of non-traditional/ non-academic dissemination to better translate research findings into impact and action in diverse communities and settings. Researchers could consider hosting town-halls or community forums with participants and stakeholders (e.g. practitioners, advocacy groups, and policy makers) to discuss research findings and their implications within the local context where the work was carried out. In addition to these high-touch dissemination avenues, developing infographics and other user-friendly products to supplement research papers may enhance the accessibility of research findings for the lay public. Better messaging of science may not only enhance the potential impact of research but also increase its perceived value within the general public.

Implications for Public Health Nursing Practice

Public health nurses (PHNs) are experts in population health; an important aspect of providing population health is promoting health equity (Reifsnider, 2021). PHNs may benefit from strategies outlined in this manuscript to reach and maintain connection with community members who experience health disparities. When conducting campaigns for vaccine advocacy, for example, partnerships are critical to reach underrepresented and vulnerable populations. When providing psycho-education for persons at high risk for disease, it is essential to ensure that barriers to entry are addressed. Best practices described herein for retention, moreover, such as maintaining frequent contact using population-appropriate modalities, may be used to support disease management. It is also important to be attuned to the language, cultural traditions, and norms of a given population (Chen et al., 2007; Wilson & Neville, 2009). Although PHNs may not be familiar with the tenets of CBPR, their connections to community groups, grassroots organizations, and local businesses present an opportunity to apply these basic principles to promote health equity.