Comparison of Recruitment Strategies for Engaging Older Minority Adults: Results From Take Heart

Abstract

Background

Few published studies report lessons learned for recruiting older adults from racial/ethnic minority, low SES communities for behavioral interventions. In this article, we describe recruitment processes and results for Take Heart, a randomized controlled trial testing the effectiveness of an adapted heart disease self-management program for primarily African American, urban, low SES adults 50 years or older living in Detroit.

Methods

Older adults were recruited via community-based (CB), electronic medical record (EMR), and in-person hospital clinic (HC) methods. Recruitment processes, demographic characteristics of enrolled participants, yield and cost, lessons learned, and best practices for each method are described.

Results

Within 22 months, 1,478 potential participants were identified, 1,223 were contacted and 453 enrolled, resulting in an overall recruitment yield of 37%. The CB method had the highest yield at 49%, followed by HC at 36% and EMR at 16%. Of six CB approaches, information sessions and flyers had the highest yields at 60% and 59%, respectively. The average cost of recruiting and enrolling one participant was $142.

Conclusions

CB, EMR, and HC methods each made important contributions to reaching our recruitment goal. The CB method resulted in the highest recruitment yield, while EMR had the lowest. Face-to-face interaction with community members and hiring a community health worker were particularly useful in engaging this population. Further research is needed to confirm these findings in urban, minority, low SES populations of older adults.

By the year 2050, racial and ethnic minorities will become the majority of the older adult population in the United States (1). Despite this projection, recruitment of older people of color for behavioral and clinical health research remains a challenge (2,3). Older African Americans, in particular, are less likely to participate in clinical trials than whites (4,5). Because this group experiences disproportionate chronic disease burden (1,3), effective strategies are needed to ensure their adequate representation in research that tests interventions for chronic disease self-management.

Factors impeding recruitment of older African American adults to research studies include poor health, cognitive impairment, transportation problems, isolation, and low health literacy levels (6,7). Other common barriers are mistrust of research based on negative past experiences, fear of exploitation, and concern about possible harm (4,5). Furthermore, a recent study indicates that African Americans are less likely to be invited to participate in research than their white counterparts (8). Research demonstrates that healthy older adults who are mobile, have fewer conditions, and take less medication are overrepresented in research compared with “real-world older adults” (9,10) who are in poorer health, are less independent, have multiple comorbidities and lower socioeconomic status (SES) (1,10). The underrepresentation of vulnerable older adults in health research contributes to a lack of evidence needed to address the very health disparities we seek to eliminate.

This article describes our process for screening, enrollment, and recruitment for Take Heart, a randomized controlled trial testing the effectiveness of a community-based self-management program for older adults with heart disease or cardiovascular risk factors (11). We implemented community-based (CB) outreach, electronic medical record (EMR) review, and hospital clinic-based (HC) recruitment methods. First, we describe study goals and processes for screening and enrollment. Next, for each recruitment method, we describe: (a) recruitment activities, recruitment flow numbers, eligibility, refusal rates and reasons for refusal; (b) characteristics of enrolled participants; and (c) yield and cost.

Method

Study Design and Eligibility Criteria

All project activities were approved by the University of Michigan (UM), Detroit Medical Center (DMC), and Wayne State University (WSU) Institutional Review Boards. Take Heart (TH) is a RCT, with an intervention (education) group and a control (usual care) group. Block randomization lists were generated using Sealed Envelope, an online randomization application (12). Intervention participants received a 7-week program consisting of five 2.5-hour in-person group sessions and two 1-hour telephone sessions; both facilitated by a trained health educator (see Ramsay and colleagues (11) for details about the TH curriculum).

TH participants had to: (a) be ≥50 years of age; (b) report one or more of the following diagnoses: atrial fibrillation, myocardial infarction, valvular disease, pulmonary hypertension, angina, congestive heart failure, or peripheral vascular disease, or at least two cardiovascular risk factors: high cholesterol, high blood pressure, diabetes, chronic kidney disease (Stage 3 or 4) or smoke tobacco; and (c) be able to participate in group and telephone education sessions.

Overview of Recruitment Goals and Steps

The overall recruitment goal for the TH study was to enroll 450 participants. Our two study partners for recruitment were the Detroit Area Agency on Aging (DAAA) and the DMC, a large hospital and health care system. The University of Michigan (UM) TH staff, including a Project Director, Project Coordinator, two research assistants and a Community Health Worker (CHW), with help from the DAAA outreach liaison, carried out CB recruitment activities. UM TH staff conducted HC recruitment methods at two DMC clinics; EMR methods were implemented by the DMC in conjunction with UM TH staff.

Recruitment Materials

All recruitment materials followed guidelines from the Center for Disease Control’s “Simply Put Guide to Creating Easy to Understand Materials” (13) that were used previously with older African Americans (14). Plain language was used to describe heart-related topics (eg, valvular disease was described as a “leaky valve”). Materials included flyers, posters, a banner, handouts, and 3-D heart and artery models for display.

Community-Based Recruitment

A CHW was hired and trained to enhance CB recruitment efforts. The CHW was a Detroit resident of similar racial/ethnic background as the priority community with extensive outreach experience in the region. The CB method involved six approaches: information sessions, health fairs, community events, flyer distribution, community referrals from DAAA, and word of mouth. The CHW hosted the information and flyer-distribution sessions and represented the project at health fairs and community events.

TH staff also hosted information sessions about heart health and the TH study at senior housing communities, public and private housing buildings, recreation centers, churches and other community organizations. Depending on the event setup, potential participants were either screened for the study onsite, left their contact information, or contacted project staff at their convenience.

In addition to the locations selected for information sessions, flyers were posted at other community sites (eg, libraries, hair salons, restaurants). The CHW contacted and built rapport with staff at each site and then delivered flyers in person to be posted or distributed directly to older adults. The DAAA outreach liaison also incorporated TH as part of their menu of program options at scheduled presentations (community referrals). Finally, TH staff encouraged and gave extra flyers to both potential and enrolled participants, to spread the word to others (word of mouth).

Electronic Medical Record Recruitment

EMR recruitment involved the identification of potentially eligible DMC patients by age and diagnosis via searches conducted by the study’s medical advisor, a cardiologist at DMC. An information packet with a letter signed by the clinics’ providers was mailed to each patient on the list. One to two weeks later, TH staff called potential participants to gauge interest and when appropriate, screen for eligibility.

Hospital Clinic Recruitment

TH staff visited two DMC clinics on a weekly basis to recruit participants. Each patient in the waiting room received a brief overview of TH, a flyer, and the opportunity to share his or her contact information if interested. TH posters were displayed in each clinic and TH staff provided information sessions about the study to clinic providers.

Screening, Consent, Enrollment, and Randomization

TH staff attempted to contact and screen all potential participants (ie, those for which we had contact information) for interest and/or eligibility. Potential participants were considered contacted if they were screened eligible or ineligible, and/or refused participation. If eligible, participants were told what to expect if they enroll, and were given time to ask questions. Those who refused participation were asked if they would be willing to share their age, race, ethnicity, and reason for not participating. Next, all eligible and interested participants completed the informed consent process verbally or by mail. The baseline telephone interview (BL) was then scheduled. Once completed, participants were considered enrolled in the study, randomized to the intervention or control group, and mailed a $20 gift card (Figure 1).

Figure 1.

Take Heart screening, consent, enrollment, and randomization process.

Recruitment Yield and Costs

Recruitment yield is the percentage of contacted participants (ie, were screened for eligibility, and/or refused participation) who enroll in the study. Direct costs were estimated for recruitment and enrollment efforts for each of the three methods, and separately for each of the six CB approaches. Recruitment-related direct costs included materials (eg, flyers, envelopes, stamps); staff time including fringe benefits; staff transportation to and from recruitment sites; attendance fees for community events; and staff time for telephone calls made before participant enrollment. Enrollment-related direct costs included staff time needed to complete the BL interview (1 hour per participant on average). We combined the direct costs for recruitment and enrollment efforts by method, allowing us to provide an average cost per enrolled participant. (See Supplementary Appendix A for details.)

Statistical Analyses

Descriptive statistics were generated to describe the demographics of participants recruited for the study. Chi-square tests were used to compare proportions across the recruitment methods, enrollment status, gender, and age groups. Within-group differences in yield were examined using chi-square tests. All analyses were done using SAS 9.4 (15).

Results

Total Recruitment Activities by Method

For the CB method, TH staff conducted 23 information sessions, attended 14 health fairs and 8 community events and distributed approximately 1,050 fliers to over 130 community sites. The DAAA outreach liaison included TH in over 100 of their regularly scheduled presentations. TH staff mailed 530 information packets to patients identified via the EMR method and attended 12 clinic sessions to recruit via the HC method.

Overview of Recruitment Flow

Across all TH recruitment methods, 1,478 potential participants were identified as the recruitment pool, meaning either they provided their contact information or we received it through the EMR method. Of these, 1,223 (82%) were successfully contacted to determine eligibility and/or interest. Of these, 425 (35%) refused participation, 220 (18%) were ineligible and 578 (47%) were eligible and interested. Of those that were eligible, 125 (22%) did not enroll because we were subsequently unable to contact them.

Ineligible and Refused by Recruitment Strategy

Of all contacted participants who were ineligible, 69% were recruited from CB methods, 24% from EMR methods, and 7% from HC methods. Of those who refused participation, 64% were recruited from EMR methods and 32% from CB methods.

Characteristics of Enrolled Participants

Of those who were contacted, eligible and interested in participating, 453 enrolled and were randomized to control (n = 225) or intervention (n = 228) groups. The recruitment goal of 450 was met within 22 months, 2 months earlier than planned. For every enrolled participant, three potential participants were contacted. The staff completed an average of 18 phone calls to screen and enroll participants (including unanswered calls and voicemails). See Supplementary Appendix B for a flow chart of recruitment and enrollment results.

Enrolled participants were primarily female (74%), African American (86%) and reported an annual household income of $15,000 or less (59%). About 1/3 of participants were in their 50s, and 41% were between 60 and 69 years old. Sixty-four percent reported having completed some college (43%) or graduated from college (21%) (Table 1). There were no significant demographic differences by enrollment status or recruitment method.

Table 1.

Take Heart Demographic Profile of Enrolled Participants

Characteristic	All Enrolled Participants (n = 453)
Gender
Female	334 (73.9)
Race
Black/African American	381 (85.8)
Multiracial (2+ races reported)	41 (9.2)
White	21 (4.7)
Native American/Alaskan Native	1 (0.002)
Age
50s	130 (28.9)
60s	183 (40.7)
70s	101 (22.4)
80s or older	36 (8.0)
Education
<High school	53 (11.8)
High school/GED	110 (24.4)
Vocational/tech/some college/associates	194 (43)
College graduate+	94 (20.8)
Income
<$15,000	250 (58.7)
$15,001—–$40,000	134 (31.5)
$40,001+	42 (9.9)
Employment
Retired	227 (50.2)
Disability/social security income	95 (21)
Unemployed	85(18.8)
Employed	45 (10)
Marital status
Single (including widowed, divorced)	392 (87.3)

Note: Values are expressed as n (%).

In regards to cardiovascular-related health conditions, 53% of participants reported having one or more heart disease diagnoses; the remainder had risk factors only. The proportion with diagnosed heart disease was highest in the EMR sample (66%), followed by HC (63%) and CB (50%). The most prevalent heart disease diagnoses across all enrolled participants were myocardial infarction (MI, heart attack), congestive heart failure (CHF) and angina. Participants enrolled via EMR reported a 10%–15% higher prevalence of MI, CHF, and atrial fibrillation than those enrolled via CB or HC methods. The most prevalent risk factors among all enrolled participants were high blood pressure (91%), high cholesterol (78%), and diabetes (43%) (Table 2).

Table 2.

Number of Enrolled Take Heart Participants Who Reported Heart Disease and Heart Disease Risk Factor Diagnoses, by Recruitment Method

Diagnosisa	Participants Enrolled by Community-Based Method (n = 362) (%)	Participants Enrolled by Hospital Clinic Method (n = 27)	Participants Enrolled by EMR Mailings Method (n = 64)	All Enrolled Participants (n = 453)
Heart diseaseb
Myocardial infarction	56 (15)	5 (19)	18 (28)	79 (17)
Congestive heart failure	51 (14)	5 (19)	22 (34)	78 (17)
Angina	61 (17)	5 (19)	11 (17)	77 (17)
Atrial fibrillation	61 (17)	5 (19)	20 (31)	56 (12)
Peripheral vascular disease	40 (11)	3 (11)	10 (16)	53 (12)
Valvular disease	25 (7)	1 (4)	8 (13)	34 (8)
Pulmonary hypertension	13 (4)	2 (7)	3 (5)	18 (4)
Risk factor
High blood pressure	329 (91)	24 (89)	61 (95)	414 (91)
High cholesterol	285 (79)	14 (52)	55 (86)	354 (78)
Diabetes	154 (43)	7 (26)	36 (56)	197 (43)
Smoke tobacco	110 (30)	7 (26)	16 (25)	133 (29)
Chronic kidney disease, Stage 3 or 4	21 (6)	2 (7)	3 (5)	26 (6)

Note: Values are expressed as n (%).

^aMany participants reported more than one diagnosis; thus, the columns (recruitment method) do not add up to the total number enrolled via that method but rather, the total number of diagnoses reported by participants enrolled by each method.

^b53% of enrolled participants reported having a diagnosis of at least one type of heart disease (66% of EMR, 63% HC, 50% CB); the remaining 47% that did not have a heart disease diagnosis reported a diagnosis of at least two of the five risk factors.

Yield and Cost

Overall recruitment yield was 37%. The CB method had the highest yield at 49%, followed by hospital clinic at 36% and EMR at 16%. Of the six CB approaches, information sessions and flyers were the most fruitful at 60% and 59%, respectively (Table 3). When comparing yield by demographic characteristics, we found significant differences within age and gender groups (ps < .010). Among participants under the age of 70, yield was highest for flyer distribution (47%, p =. 03). Women were most likely to enroll via word of mouth at 51% (p = .03).

Table 3.

Take Heart Recruitment and Enrollment Yield

Method	No. Contacted	No. Enrolled	Yield (%)
Community-based	744	362	48.7
Health fair	69	24	34.8
Community event	149	50	33.6
Information session	128	77	60.2
Flyer	239	141	59.0
Word of mouth	43	21	48.8
Referral from community partners	116	49	42.2
Hospital clinic	76	27	35.5
EMR (mailings)	403	64	15.9
Totals	1,223	453	37.0

The total direct cost of all recruitment and enrollment efforts for participants who enrolled in the RCT was $64,151. The costs of implementing each recruitment method, including telephone calls are summarized in Table 4. Across all methods, the average cost of recruiting and enrolling one participant was $142. The HC method had the highest cost per enrolled participant at $248, followed by overall CB methods at $165, and EMR at $138. Of the CB approaches, health fair recruitment was the highest at $436 per enrolled participant, while community referrals had the lowest cost at $60 per participant. Information sessions and flyers, the two CB approaches with the highest yield, cost $153 and $91 per enrolled participant, respectively. It is important to note that the cost calculations are based on the unique aspects of our study’s recruitment logistics (eg, length of travel for staff conducting recruitment, staff hourly pay), which may vary significantly from study to study. Supplementary Appendix A provides a framework for estimating recruitment costs of a given study.

Table 4.

Take Heart Recruitment and Enrollment Costs

Recruitment Method	Cost of Implementation ($)a	Recruitment and Enrollment Phone Calls ($)	Total Ptsb Enrolled	Average Cost per Enrolled Ptc ($)
Community-based	26894.20	21756.20	362	164.66
Health fair	9019.43	1442.40	24	435.91
Community event	6395.43	3005.00	50	188.01
Information session	7163.91	4627.70	77	153.14
Flyer	4315.43	8474.10	141	90.71
Word of mouth	0.00	1262.10	21	60.10
Referral from community partners	0.00	2944.90	49	60.10
Hospital clinic	5064.43	1622.70	27	247.67
EMR (mailings)	4966.95	3846.40	64	137.71
Totals: all methods	36925.58	27225.30	453	141.61

Note: Total direct recruitment and enrollment costs = $64150.88.

^aSee Supplementary Appendix A for more details.

^bPts = participants.

^cPt = participant.

Discussion

Given the underrepresentation of minority older adults in health research, it is critical to report lessons learned about the recruitment process to guide future work. The goal of this article is to describe our process for screening, enrollment, and recruitment for Take Heart, a randomized controlled trial of a self-management intervention for African American adults with heart disease and risk factors in Detroit. Our multi-pronged recruitment strategy included three methods: community-based, hospital clinic, and EMR mailings. Across the three methods, our overall yield was 37%. Several factors may have contributed to this relatively high yield compared to similar published studies (11% (2) and 17% (16)). First, we adopted inclusive screening criteria that likely increased eligibility among our priority population (eg, age 50+ rather than 65+; relatively broad inclusion criteria including common cardiovascular risk factors). Although a number of chronic disease self-management programs are available in the Detroit area, Take Heart is the only one that focuses on heart disease. This may have increased interest given the high prevalence of heart disease and related risk factors among Detroit adults. Finally, Take Heart provides transportation to attend group sessions, which may have increased ability to participate (17,18).

Of the six CB approaches implemented by Take Heart, tailored information sessions and flyer distribution had the highest yields. Both of these approaches involved personal connections with community partners. Results of the few published articles that include a comparison of CB recruitment approaches are mixed. One study recruited Chinese American immigrant older adults and compared four CB approaches, concluding that active, in-person recruitment at community venues was more effective than approaches that required participants to call in on their own, such as posting flyers and advertisements (16). Others compared CB approaches for older adults with limited mobility, concluding that mass mailings or personal letters resulted in the highest yield while CB promotional events had the lowest yield (2,19). Further research is needed to identify the strategies that are most effective for particular populations. For studies using CB recruitment approaches, we recommend hiring and training CHWs from the priority community to carry out these efforts, rather than research staff who do not have a deep understanding of and connection with the community (also reported in (17,18,20,21).

EMR recruitment had the lowest yield for Take Heart, at 16%, and accounted for 64% of the refusals. Our study population may be less likely to respond to a physician endorsed, mailed letter followed by “cold” calls, compared with “warmer,” tailored CB methods that create a connection with project staff or partner organizations (16,22,23).

Unlike those recruited via CB and HC methods, patients recruited via EMR did not express initial interest in participating before being contacted, a factor that almost certainly decreased yield. A higher percentage of those enrolled via EMR had a heart disease diagnosis compared with those enrolled via the CB method. EMR may thus be particularly useful for recruiting older adults with diagnosed conditions (as opposed to risk factors only) and/or advanced disease (as found in (2,22,24)). Based on findings reported in other studies, EMR methods may have resulted in a higher yield if the CHW conducted phone calls to patients as opposed to research staff (25). HC recruitment had a higher yield than EMR, likely because it involves a face-to-face connection, similar to CB approaches. However, the small sample recruited by this method limits generalizability.

The average cost of recruiting and enrolling one participant in our study was $142. Comparable behavioral trials report significantly higher costs per enrolled participant at $494 (18) and $1080 (26). Comparison among studies, though, is difficult because so many factors are at play, including number of staff, hourly pay, and location of recruitment sites. In our study, the cost of CB approaches ranged from $60 (word of mouth, community referral) to $326 (health fair). Based on our experience, the time and resources devoted to tailored, face-to-face CB methods was worthwhile (also reported in (17,24)). If more publications include details about the recruitment process, results, and costs, future researchers will be better able to plan, estimate, and compare across studies.

Conclusion

Recruitment of vulnerable and marginalized groups of older adults remains one of the biggest challenges in health research (27). Sharing lessons learned will increase capacity to engage older adults in intervention research designed to decrease the disparate burden of disease faced by our most vulnerable populations.

Funding

This work was supported by the National Institute on Aging of the National Institutes of Health (R01AG047203). The content in this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Authors’ Contributions

J.E.R. is the project director of Take Heart. J.E.R. conceived the idea for this article, took the lead on writing it and developing the data analysis plan. She coordinated all feedback from team members and kept the article on the agreed upon submission timeline. C.K.H. is the project coordinator of Take Heart. C.K.H. drafted the methods section, assisted in conducting the recruitment cost analysis, and created tables and figures for the article. She also incorporated edits and formatted the article for submission. M.R.J. is the co-investigator of Take Heart. M.R.J. helped to fine-tune the article’s goals and overall content, and edited multiple drafts. R.R.C. is the Research Associate Take Heart. R.R.C. conducted a literature review for this article and helped format figures for submission. Additionally, R.R.C. collected much of the recruitment and enrollment data reported in this article. K.L.A. is the Epidemiologist/Consultant of Take Heart. K.L.A. conducted the data analysis for this article. She also assisted in drafting and editing the results section. C.M.C. is the Principal Investigator of Take Heart. C.M.C. helped to shape the overall article, ensure adherence to the overarching goals, fine-tune the discussion and make edits throughout the drafting process.

Conflict of Interest

None reported.