Effective Recruitment and Retention in Community-Engaged Research with Predominantly African American Participants in Neighborhoods of High Poverty

Background and Purpose: Low research participation and retention rates hamper understanding and addressing health and social disparities that adversely affect minority and disadvantaged communities, who continue to be underrepresented in research.  Multiple barriers to recruitment, enrollment and retention have led to the label of "hard-to-reach," implicitly blaming non-participants.  Little research has identified best strategies, but there is suggestive evidence for the importance of ten factors (Yancey, Ortega & Kumanyika, 2006 DOI: 10.1146/annurev.publhealth.27.021405.102113).  We proposed that attention to these factors, combined with careful matching of intervention and comparison communities, would achieve a minimum enrollment and retention goal of n=200 per group for a quasi-experimental study with three measurement points (baseline, 12 months and 24 months) and result in highly comparable intervention and comparison groups.

Methods: We recruited and enrolled participants from 11/2013 through 5/2014.  Recruitment focused on seven census tracts in two SC counties (4 intervention, 3 comparison; combined population=19,117; 28%-62% of households <federal poverty income; 43%-100% African American residents.) Because geographic boundaries may not match residents' self-identified neighborhoods, we extended eligibility by one mile past the core tracts' boundaries, if a tract had at least 16% households below poverty (S.C.'s poverty level). The study (evaluation of a community-initiated food hub) was powered to detect to small effects (Cohen's d=.28), with power=.80, α=.05., and minimum n/group=200. To offset attrition, the goal was 560 (280 per group), an extra 40%. We layered active and passive recruitment and retention strategies and trust-building approaches, based on guidance from the literature (Yancey, Ortega, Kumanyika, 2006 DOI: 10.1146/annurev.publhealth.27.021405.102113), community partners and community outreach recruiters.  For retention, we sent a holiday card and three project newsletters between 7/2014 and 4/2015.  We implemented a six-week retention call and mailing protocol (min. 1 phone attempt, max. 19 attempts) from 7/2014 to 10/2014). We used the same strategies in both communities, which provided a natural replication. We graphed enrollment (x axis) by time (y axis) and plotted implementation of recruitment strategies to identify spurts and dormant periods in both groups; compared groups on baseline sociodemographic characteristics (t-tests and chi-squared tests); and tracked retention, withdrawal and loss to followup over 16 months. 

Results: Of n=928 inquiries, we screened n=863 (458 intervention, 405 comparison).  Of 666 eligibles, n=532 completed interviews (n=269 intervention, n=263 comparison), achieving 96% and 94% of the expanded goal of 280/group, respectively.  There were no group differences by gender (p=.14), education (p=.10), income (p=.21), female-headed households (p=.55), or self-rated health (p=.06).  Significant differences of relatively small magnitude were age (mean=50±15 versus 54±14, p<.01), households with children (38% versus 28%, p<.01), and African Americans (89% versus 96%, p<.01).  By 3/2015, contact had been maintained with 90% (88% intervention, 93% comparison); 477 were still in the study (89.7%) (236 intervention, 241 comparison), three withdrew (.56%), two deceased (.38%) and 50 lost to followup (9.4%). 

Conclusions and implications for practice: Culturally tailored, community-involved recruitment and retention strategies led to socio-demographically comparable intervention and comparison groups, achievement of enrollment goals and very high retention at 16 months among participants usually designated as "hard-to-reach."