Abstract

Purpose: To identify racial/ethnic differences in retention of older adults at 3 levels of participation in a prospective observational study: telephone, in-home assessments, and home visits followed by blood draws. Design and Methods: A prospective study of 1,000 community-dwelling Medicare beneficiaries aged 65 years and older included a baseline in-home assessment and telephone follow-up calls at 6-month intervals; at 4 years, participants were asked to complete an additional in-home assessment and have blood drawn. Results: After 4 years, 21.7% died and 0.7% withdrew, leaving 776 participants eligible for follow-up (49% African American; 46% male; 51% rural). Retention for telephone follow-up was 94.5% (N = 733/776); 624/733 (85.1%) had home interviews, and 408/624 (65.4%) had a nurse come to the home for the blood draw. African American race was an independent predictor of participation in in-home assessments, but African American race and rural residence were independent predictors of not participating in a blood draw. Implications: Recruitment efforts designed to demonstrate respect for all research participants, home visits, and telephone follow-up interviews facilitate high retention rates for both African American and White older adults; however, additional efforts are required to enhance participation of African American and rural participants in research requiring blood draws.

Recruiting and retaining research participants representative of the population being studied are a challenge for prospective observational studies but essential if research findings are to be useful to practitioners and policy makers. Moreover, reducing health disparities has been identified as a national priority (Warnecke et al., 2008), making it imperative to understand the science of inclusion for the recruitment and retention of previously underrepresented groups in research studies. Prior studies have acknowledged potential barriers to achieving appropriate representation of ethnic and racial groups in research at both the participant and institutional levels (Curry & Jackson, 2003). It is also important to recognize within-group variations in seeking ways to enhance recruitment and retention.

Several theoretical models have been suggested to guide recruitment and retention efforts targeting older minority adults. All of the models utilize concepts articulated in the health behavior model and subsequent health belief models. The health behavior model (Anderson, 1995) was conceptualized to understand the reasons why individuals access health care services and focused in large part on social structure. Predisposing characteristics, such as age, race/ethnicity, and gender, were conceptualized as working in conjunction with enabling factors (social and economic resources) and perceived need of services (individual health). The health belief model expanded the health behavior model to focus on the impact of the individual's beliefs and values in pursuing health care or actions promoting health (Rosenstock, Strecher, & Becker, 1988). In this model, human behavior was conceptualized as being dependent on the perceived value of an outcome and the person's belief that their action would result in the achievement of that outcome. These beliefs may be particularly salient in the level of participation for any person considering involvement in a research study.

Social Cognitive Theory (SCT), community-based participatory research models (CBPR), and the overarching conceptual model (AHRQ) appearing in the Evidence Report “Knowledge and Access to Information on Recruitment of Underrepresented Populations to Cancer Clinical Trials” prepared for the Agency for Healthcare Research and Quality (Ford et al., 2005) suggest ways to address awareness, recruitment, and retention, the three phases in which barriers and motivators need to be considered to ensure full representation in research (Powe & Gary, 2004). SCT aligns with the health belief models in describing a framework for human behavior as the interaction of environmental and personal factors that affect behavior. Environments may be both physical and social (Glanz, Rimer, & Lewis, 2002). The contribution of CBPR models is the introduction of localized social context to traditional epidemiology. As noted by Leung, Yen, and Minkler (2004), “ … CBPR is not a method per se but an orientation to research … ” The AHRQ model specifically addresses the recruitment process which is conceptualized as involving research study awareness, opportunity to participate, and the opportunity to accept/refuse participation. The model identifies barriers and promoters across the recruitment continuum, moderators and sociodemographic factors (e.g., culture, income, gender etc.), study design factors, and interventions that influence the recruitment and retention process.

These models have guided the development of multipronged approaches to overcome barriers to recruitment that target three levels: individuals, organizational systems (e.g., community organizations), and agents of change (e.g., church leaders). Strategies for retention are similar to those for recruitment (Yancey, Ortega, & Kumanyika, 2006) and include informative and respectful communication of the risks and benefits of the research, convenience, expression of appreciation, and maintained contact. Individual level factors that may impact willingness to participate in research include both mental and physical health. Participants with poor health may evaluate participation in research as an added burden. Additionally, specific studies may exclude patients who are not cognitively or physically able to participate. Successful communication is vital to establishing a trusting relationship between the participant and those conducting the research, both the individuals and the institutions (Yancey et al., 2006). Difficulty in understanding the informed consent process has been identified as a barrier to participation (Levkoff & Sanchez, 2003). Age may influence willingness to participate because of cohort issues, such as the African American experience with the Tuskegee Syphilis Study, or because of age-associated diseases and syndromes that limit function. Transportation availability is a facet of socioeconomic status that has been cited as a potential barrier to participation (Levkoff & Sanchez, 2003). For example, rural residents may have limited opportunities to participate in research due to lack of transportation (Park et al., 2010). Additionally, Park and colleagues (2010) found that depression was associated with transportation difficulty among older rural African Americans, potentially adding another barrier by reducing interest in research participation.

At the institutional level, a factor to be considered in recruiting older African Americans may be a personal sense of previous discrimination leading to cultural bias and lack of belief in the utility of research. Another barrier is related to trust (Levkoff & Sanchez, 2003), a factor that may be particularly salient to the participation of older African Americans. The African American history with the health care system has been overshadowed by mistrust engendered by the U.S. Public Health Service Syphilis study (i.e., the Tuskegee Syphilis Study) that occurred in Macon County, Alabama (Curry & Jackson, 2003; Stahl & Vasquez, 2004). Disparate recruitment of African Americans for clinical trials has been partially attributed to knowledge of this trial and subsequent mistrust of scientists (Brown & Topcu, 2003). An in-depth qualitative study of African American adults (Scharff, Mathews, Jackson, & Hoffsuemmer, 2010) noted that the mistrust engendered by the Tuskegee Syphilis Study may be broader than noted previously. African American participants of all ages who participated in this study identified medical research with the notion of experimentation. They also expressed the fear that personal information might be used against them. Other barriers identified included inadequate information, inconvenience, and the questionable reputation of individual researchers or institutions. Today's older African Americans have likely experienced past discrimination that may have contributed to negative feelings about interaction with the health care system (Dilworth-Anderson & Williams, 2004) and a general mistrust of American society (Moreno-John et al., 2004).

Additionally, it has been reported that African Americans are likely to have a fatalistic belief associated with illness that contributes to their avoidance of medical care and self-management (Egede & Bonadonna, 2003; Peek, Sayad, & Markwardt, 2008). Brown and Topcu (2003) found that African Americans’ attitudes toward participation in a clinical trial for cancer was not influenced by the provision of free medical care or helping scientists to learn more; the only factors influencing a positive attitude toward participation were a physician recommendation and the belief that participation would help someone else be cured.

The models suggest several methods to enhance participation in research; however, literature addressing these issues is recent and focuses mainly on initial recruitment (Yancey et al., 2006). Strategies to enhance recruitment include matching study participants with research staff (Levkoff & Sanchez, 2003), personalized letters, and financial incentives (Yancey et al., 2006). Additional strategies suggested for retention include prompt payment of incentives, maintaining interviewer–participant interactions, providing a toll-free number to contact study personnel, and flexibility to accommodate participant schedules (Yancey et al., 2006).

It should be noted that two recent review articles suggest that participation of underserved racial/ethnic groups in research studies has increased (Yancey et al., 2006) and may have equalized (Wendler et al., 2006). Recently, others have been successful in recruiting for survey research but have had less success recruiting for more invasive procedures or clinical trials (Scharff et al., 2010).

The University of Alabama at Birmingham (UAB) Study of Aging had a purposeful design to recruit and retain equal numbers of African American and White community-dwelling older adults in three levels of research: an in-home interview, telephone follow-up interviews, and participation in laboratory research. Although recruitment and retention methods were designed before the dissemination of many of the study results reported above, efforts to achieve participation at all three levels focused on promoting individual awareness of the research opportunity as well as minimizing barriers to participation, strategies congruent with CBPR and the AHRQ model. Specific factors addressed but not measured included integrating the social context though pilot testing the questionnaire among African American and White older adults. Interviewers were trained by local minority investigators to be respectful and to communicate clearly during all interactions with participants. Interviewers were instructed to treat participants as “partners,” to show appreciation and to stress the importance of participants’ contribution. Bringing the research to the participants at their convenience was part of an effort to bridge the institution/participant dichotomy as well as eliminate the need for transportation. The purpose of these analyses was to identify racial/ethnic differences in retention of participants at the three different levels of participation and to identify factors associated with participant retention and participation after four years of follow-up.

Methods

Research Design

In spite of the study being conducted in Alabama, where knowledge of the Tuskegee Syphilis study was expected to be widespread, we hypothesized that conducting face-to-face interviews in participants’ homes would address issues related to trust and the willingness of research staff to bring the research to the participant thereby resulting in increased enrollment at the same time reducing barriers related to transportation availability. Personal identification with the study and study staff was considered a necessary component for retention for the telephone follow-up component of the study. The study design allowed for retention of African American and White community-dwelling older adults at three different levels of participation after four years of follow-up: (1) a telephone interview, (2) an in-home interview and assessment, and (3) a blood draw performed by certified phlebotomists in their homes.

The Sample.—

Baseline recruitment for the UAB Study of Aging began in December 1999 and ended February 2001. A random sample of Medicare beneficiaries aged 65 years and older living in five counties of central Alabama were selected from a list of Medicare beneficiaries provided by the Centers for Medicare and Medicaid Services (CMS), stratified by county, race, and sex. Specific rural counties were selected for their proximity to UAB and proportion of African American and White residents. In 2000, the population aged 65 years and older in Alabama was 80% White, 60% female, and 9% rural (United States Census, 2000). By design, rural, African American, and male older adults were oversampled to achieve a balanced sample (50% African American, 50% male, and 51% rural). Persons who remained community-dwelling four years following enrollment were eligible for the second in-home visit and the collection of a blood sample.

Data Collection.—

The investigative team included African American researchers from psychology, medicine, and nursing who were involved in the study design and development of the questionnaire. Additionally, portions of the questionnaire were developed based upon qualitative interviews (N = 49) conducted among African American and White respondents (Sawyer & Allman, 2010) by African American and White interviewers, including one of the authors of this paper, singly and in mixed pairs. Participants in these previously completed qualitative interviews were selected to represent an equal number of African American and Whites at all levels of function measured by life-space mobility. Several African American participants reflected that they were impressed by the research team's dedication in making home visits. Findings were incorporated into the final protocol and questionnaire design of the UAB Study of Aging. For example, we found that participants had difficulty with a numerical scale for pain intensity; thus, verbal descriptors were used thereafter. The interview format has remained essentially the same throughout the period of data collection for the UAB Study of Aging.

As required by the CMS, letters were ultimately sent to 3,100 persons, followed by telephone calls to set an appointment for an in-home assessment. Within counties four race/sex strata were established and a computer-generated random listing of names was developed for each strata. Letters were sent on a rolling basis until recruitment goals from each specific county/race/sex strata were reached. Of 2,188 persons contacted by telephone, 1,000 participants were enrolled (43.7% of the African Americans and 47.9% of the Whites). Persons were ineligible if they reported that they did not live in one of the five study counties, were under the age of 65, or were unable to communicate on the telephone to make arrangements for the in-home assessment. Participants were invited to have another family member or friend present when the interviewer came to their homes.

Interviews were conducted by male and female interviewers trained by the research team; 10% were conducted by African American interviewers. Although there was no attempt to match interviewer and participant by ethnicity, data were collected to determine if a match had occurred. Participants scheduled for interviews were told that their interviewers would have picture identification from the University. At the baseline interview, no financial incentives were offered, but participants were given a pen and magnet with the toll-free study telephone number to call if they had any concerns or to report health events that may have occurred. A follow-up call 1–2 weeks following the in-home interview asked about participant satisfaction with the interview process and informed participants that they would be called at six-month intervals for the duration of the study. Follow-up interviews were scheduled at six-month intervals following the initial in-home interview.

As much as possible, all telephone follow-up interviews were conducted by the same research interviewer who attempted to call during the scheduled month and the subsequent month. Participants were asked to designate a convenient time to call, which was used by the interviewer. If requested, interviews were conducted at night, on weekends, and early morning hours. During the second month of no response, the interviewer called person or persons designated as contacts at the in-home interview. At the end of that period, if participant status was still undetermined, a letter requesting contact information was mailed to the last known address. A stamped return envelope was provided.

We planned a second in-home interview and assessment in which participants were asked to participate in a blood draw after the conclusion of the 48-month follow-up telephone interview conducted by the research interviewer (White) who had talked to the participant over the four years of the study. Participants were invited to participate in a four-year in-home interview and assessment lasting approximately 2 hr, similar to the one completed at baseline between 1999 and 2001.

For the second in-home interview, participants were informed that $20 would be offered in appreciation for their time and that they would have the opportunity to receive results of laboratory tests at no charge if they provided a blood sample. Participants willing to complete the in-home interview were informed that they would be called by the interviewer who would conduct the in-home interview to schedule an appointment.

Only community-dwelling participants were eligible for the in-home visit at Year 4. If a participant declined the in-home visit, he or she was asked if we could continue to call at six-month intervals. The previously completed qualitative study and the baseline and follow-up UAB Study of Aging protocols were approved by the UAB Institutional Review Board (IRB).

Training sessions were held for interviewers at the time of the initial recruitment and the four-year in-home assessments. All interviewers completed IRB Training with a focus on the ethical conduct of research. Initially, a full-day group training session was provided with supplementary one-on-one training sessions as new interviewers were hired. In addition to gaining familiarity with the assessment instruments, communication techniques and skills to put persons at ease were discussed. Other topics included the different situations and living environments that interviewers might encounter, particularly in disadvantaged communities and households. Interviewers were instructed to be polite and respectful at all times.

The informed consent for the 48-month in-home visit included a brief explanation of the laboratory tests that were part of the study, “cholesterol, blood cell counts, vitamin levels, and measures of blood sugar and proteins,” and notified participants that the blood draw was optional. Participants designated their willingness to provide a blood sample, to have their blood preserved for future aging research, and if the participant wanted to be notified if the blood was used for future research. Participants who consented to the blood sample were told that they would be contacted by a qualified experienced person trained in drawing blood to set up a mutually convenient time to draw a fasting blood sample. Participants were informed that they would receive the results and be notified of any abnormalities.

In-home interviews lasting approximately 2 hr were conducted by seven different White women; laboratory samples were collected by both African American and White personnel, although detailed information was not available on the race or ethnicity of these personnel. Personnel from the James A. Pittman General Clinical Research Center at UAB, a contract research nurse, and a private company providing home health care nursing services were responsible for collecting the blood specimens. Participants were told that they would be called within two weeks of the in-home interview to complete a dietary assessment and provide feedback on the interview process. Payment was sent by check through the UAB system within two weeks of the interview.

Measures

Sociodemographic characteristics included self-reported age, race, gender, and residence. Socioeconomic variables included education and income. A priori, residence was classified by county population as urban (county population more than 100,000) and rural (county population less than 21,000) at the time of initial enrollment.

Physical and mental health were evaluated by life-space, a comorbidity count, and a cognitive assessment. Participants’ perception of lifetime discrimination was also ascertained. Baseline and 48-month data were compared for differences in life space, comorbidity, income, and cognition.

Life Space.—

The UAB Study of Aging Life-Space Assessment (LSA) measures mobility based on the distance through which a person reports moving during the four weeks preceding the assessment (Baker, Bodner, & Allman, 2003). Questions establish movement to five specific life-space levels ranging from within one's dwelling to beyond one's town. Frequency of movement and use of assistance (from equipment or persons) are assessed. For each life-space level, persons were asked how many days within a week they attained that level and if they needed help from another person or from assistive devices to move to that level. A life-space score was calculated based upon the life-space level, the degree of independence in achieving each level, and the frequency of attaining each level. Scores ranged from 0 to 120 with higher levels representing greater mobility. Interclass correlation of the LSA scores obtained within two weeks of each other previously was reported to be .96 (Baker et al., 2003); within this sample (N = 1,000), the LSA had a Cronbach's alpha of .81.

Comorbidity.—

A count of comorbidities (0–12) was based on the number of medical conditions included in the Charlson Comorbidity Index that were verified for each participant without regard to the severity of the conditions (Charlson, Pompei, Ales, & MacKenzie, 1987). Respondents were asked if a physician had told them they had specific medical conditions including congestive heart failure, myocardial infarction, valvular heart disease, peripheral artery disease, hypertension, diabetes, chronic lung disease (COPD), kidney failure, liver disease, nonskin cancer, neurological disease, and gastrointestinal disease. A condition was considered verified if the participant reported taking a medication for the disease or condition, if the primary physician returned a questionnaire indicating that the participant had the condition, or if the condition was listed on a hospital discharge summary within the previous three years.

Cognition.—

A cognitive assessment based on orientation to time and place of residence, immediate and short-term memory, attention, and language use had scores ranging 0–30 with higher scores representing greater cognition (Folstein, Folstein, & McHugh, 1975; Pfeiffer, 1975). This test consisted of commonly used questions to assess cognition. The 30 items had a Cronbach's alpha of .89.

Self-reported income included nine categories ranging less than $5,000 annually to more than $50,000 annually. As previously reported (B. R. Williams, Baker, Allman, & Roseman, 2006), baseline income categories for persons with missing responses for reported income (n = 166) were imputed from responses to a question about perceived income adequacy. The four possible responses to the question about income adequacy corresponded to mean income ranges as follows: “income is not enough to make ends meet” was equivalent to $5,000–$7,999; “income gives you just enough to get by on” was equivalent to $8,000–$11,999; “income keeps you comfortable but permits no luxuries” was equivalent to $16,000–$19,999; and “income allows you to do more or less what you want” was equivalent to $30,000–$39,000. For persons with responses for both the income category and the income adequacy items, the income variables correlated at r = .64 (p < .001).

Education was coded as completion of six or fewer years of school (1), 7 through 11 years of school (2), high school graduation (3), and post-high school training or education (4).

A history of lifetime perceived discrimination was determined by asking “Over your lifetime, how often have you experienced discrimination because of your race or skin color?” Responses were coded as occasionally, often, very often, and always. Any response of discrimination was coded as perceived discrimination. A history of perceived discrimination and baseline interview by an African American interviewer were examined as potential predictors of subsequent participation among the African Americans.

Analyses

Interviewers entered data on forms that were first reviewed for completeness and then scanned and entered directly in SPSS data files. Acceptable ranges were predetermined, and all forms were visually verified during the scanning process. For example, in asking about the number of months participation within a year in which a specific activity occurred, a response higher than 12 would not be accepted. Recruitment data were merged with the interview forms before analyses were completed. SPSS (16) was used for all analyses. Factors potentially presenting barriers to participation were examined to determine if the level of participation differed by race/ethnicity, sex, urban/rural residence, and age groups (less than 75 years of age compared with persons aged 75 years and older). Descriptive statistics (chi-square and analysis of variance) were used to examine potential differences in level of participation for sociodemographic factors, mental and physical health factors, and socioeconomic factors (including income, education, and history of previous perceived discrimination). Models were constructed based on the premise that sociodemographic factors would be particularly salient, with effects from measures of physical and mental health (life space, comorbidity, and cognition) as well as socioeconomic factors. The study design attempted to mitigate effects of socioeconomic factors by bringing the research team to the participant, thus models were constructed in the order just described. Models were hierarchical, that is, first sociodemographic factors were tested, then health measures were added to the model, and in the final step, socioeconomic factors were included. Logistic regression was used for analyses with significance defined at the p < .05 level.

Results

Recruitment outcomes for the baseline interview have been reported previously (Allman, Sawyer, & Roseman, 2006). Figure 1 traces the cohort from enrollment through participation at four years. At baseline, the study consisted of both African American and White adults, aged 65–106 years (mean age = 75.3; SD = 6.7). Recruitment rates suggested that our hypothesis of personal in-home interviewing would optimize recruitment was supported in part, although recruitment varied by gender, race, and urban/rural residence. Our results indicated that African American recruitment was less successful in urban areas and more successful in rural areas. Overall, 47% of contacted men and 45% of contacted women were recruited. Among potential African American participants, recruitment rates were 35% for urban males, 36% for urban females, 55% for rural males, and 55% for rural females. In comparison, recruitment rates for Whites were 50% for urban males, 46% for urban females, 52% for rural males, and 44% for rural females.

Figure 1.

Participation in the University of Alabama at Birmingham Study of Aging: baseline to four years.

Figure 1.

Participation in the University of Alabama at Birmingham Study of Aging: baseline to four years.

Participants were called on the telephone 1–2 weeks after the baseline and the 48-month in-home assessments to give them the opportunity to express concerns about their interviewer and/or the study. Most participants reported enjoying the home visit and many asked if there would be additional in-home visits. Although respect was not formally measured, these two calls were designed to monitor participant satisfaction with the in-home interview experience and demonstrated that the research team valued and respected the participant's assessment of the interaction with interviewers.

At the 48-month follow-up, 217 (114 African Americans and 103 Whites) persons were deceased and 7 (3 African Americans and 4 Whites) had withdrawn. We were unable to contact or determine vital status of 43 persons. These 43 persons had completed fewer than six follow-up telephone interviews. In contrast, 92.4% of persons contacted at 48 months had been contacted at every interview period (eight possible; range six to eight interviews for each participant). The number of completed interviews was not associated with type of interview at four years or willingness to participate in the blood draw. One hundred and nine persons were willing to be followed only by telephone interviews; of these, 6 were in a nursing home and provided information themselves, 12 were in a nursing home with a proxy informant, and 29 were community dwelling with a proxy informant.

Mean age at baseline of persons contacted was 74.3 years, 48% African American, 54% female, and 50% rural. There were significant differences between those contacted and not contacted for age, race, cognitive status, and education. Those unable to be contacted were older (mean age = 77.0), 65% African American, 51% female, and had lower scores on the cognitive assessment (22.4/30 vs. 25.8/30). Six hundred and twenty-four participants completed the in-home interviews, and 408 (65.4%) provided a blood sample.

Bivariate analyses indicated that the participants willing to complete the 48-month in-home interview were younger, with less comorbidity, better functional health indicated by baseline life space, lower comorbidity, and better cognition (see Table 1). The oldest person completing the 48-month in-home assessment as well as providing a blood sample was 94 years.

Table 1.

Comparison of Participant Characteristics by Interview Type at 48 Months

Participant characteristics In-home interviews, N = 624 Telephone but not in-home, N = 109 p Value 
Age (range) 73.8 (65–95) 77.2 (66–97) <.001 
African American, n (%) 303 (49) 52 (48) .870 
Female 332 (53) 63 (58) .375 
Rural 306 (49) 60 (55) .248 
Comorbidity count, M (SD2.0 (1.4) 2.4 (1.7) .005 
    0, n (%) 85 (13.6) 13 (11.9)  
    4+, n (%) 84 (13.4) 29 (26.6)  
Baseline life-space, M (SD69.8 (22.8) 54.8 (24.4) <.001 
    0–29.9, n (%) 30 (4.8) 16 (14.7)  
    30–59.9, n (%) 158 (25.3) 48 (44.0)  
    60–89.9, n (%) 277 (44.4) 31 (28.4)  
    90+, n (%) 159 (25.5) 14 (12.8)  
Education category, M (SD2.7 (1.1) 2.4 (1.1) .014 
    Less than 6th, n (%) 116 (18.6) 29 (26.6)  
    Greater than high school, n (%) 189 (30.3) 26 (23.9)  
Cognitive evaluation (0–30), M (SD) 26.1 (3.9) 24.1(5.5) <.001 
    <24, n (%) 129 (20.7) 40 (36.7)  
Income category, M (SD) 3.88 (2.4) 3.15 (2.1) .003 
    Less than $8,000, n (%) 130 (20.8) 26 (23.8)  
    More than $50,000, n (%) 62 (9.9) 4 (4.6)  
African American perceived discrimination (N = 355), n (%) 192/303 (63.4) 28/52 (53.8) .125 
Participant characteristics In-home interviews, N = 624 Telephone but not in-home, N = 109 p Value 
Age (range) 73.8 (65–95) 77.2 (66–97) <.001 
African American, n (%) 303 (49) 52 (48) .870 
Female 332 (53) 63 (58) .375 
Rural 306 (49) 60 (55) .248 
Comorbidity count, M (SD2.0 (1.4) 2.4 (1.7) .005 
    0, n (%) 85 (13.6) 13 (11.9)  
    4+, n (%) 84 (13.4) 29 (26.6)  
Baseline life-space, M (SD69.8 (22.8) 54.8 (24.4) <.001 
    0–29.9, n (%) 30 (4.8) 16 (14.7)  
    30–59.9, n (%) 158 (25.3) 48 (44.0)  
    60–89.9, n (%) 277 (44.4) 31 (28.4)  
    90+, n (%) 159 (25.5) 14 (12.8)  
Education category, M (SD2.7 (1.1) 2.4 (1.1) .014 
    Less than 6th, n (%) 116 (18.6) 29 (26.6)  
    Greater than high school, n (%) 189 (30.3) 26 (23.9)  
Cognitive evaluation (0–30), M (SD) 26.1 (3.9) 24.1(5.5) <.001 
    <24, n (%) 129 (20.7) 40 (36.7)  
Income category, M (SD) 3.88 (2.4) 3.15 (2.1) .003 
    Less than $8,000, n (%) 130 (20.8) 26 (23.8)  
    More than $50,000, n (%) 62 (9.9) 4 (4.6)  
African American perceived discrimination (N = 355), n (%) 192/303 (63.4) 28/52 (53.8) .125 

Life space, comorbidity, income, and cognition at the time of the 48-month in-home interview were evaluated. These values correlated with baseline values (0.660 for life space, 0.807 for comorbidity, 0.847 for income, and 0.742 for cognition; p < .001 for all). For consistency, baseline values were used for comparisons and in all models. Bivariate results indicated that persons willing to provide a blood sample were less likely to be African American, female, or rural and more likely to have better cognitive function, higher education, and greater income. These comparisons are noted in Table 2.

Table 2.

Comparison of Participant Characteristics for Completion of In-Home Interviews With and Without Blood Draws

Participant characteristics In-home interview and blood draw completed, N = 408 In-home interview only, N = 216 p Value 
Age 73.6 (65–94) 74.2 (65–90) .223 
African American, n (%) 174 (43) 129 (60) <.001 
Female, n (%) 204 (50) (59) .027 
Rural, n (%) 184 (45) (56) .007 
Comorbidity count (baseline), M(SD1.9 (1.4) 2.1 (1.4) <.001 
    0, n (%) 60 (14.7) 25 (11.6)  
    4+, n (%) 51 (12.5) 33 (15.2)  
Baseline life space, M(SD72.5 (22.3) 64.9 (23.1) .101 
    0–29.9, n (%) 14 (3.4) 16 (7.4)  
    30–59.9, n (%) 86 (21.1) 72 (33.3)  
    60–89.9, n (%) 186 (45.6) 91 (42.1)  
    90+, n (%) 122 (29.9) 37 (17.1)  
Education category, M (SD2.8 (1.1) 2.5 (1.1) .004 
    Less than 6th, n (%) 66 (16.2) 50 (23.1)  
    Greater than high school, n (%) 137 (33.6) 52 (24.1)  
Cognitive evaluation (baseline), M (SD26.6 (3.6) 25.3 (4.3) <.001 
    <24, n (%) 70 (17.2) 59 (27.3)  
Income (baseline), M (SD4.15 (2.3) 3.38 (2.4) <.001 
    Less than $8,000, n (%) 72 (17.6) 58 (26.8)  
    More than 50,000, n (%) 43 (10.5) 19 (8.8)  
African American perceived discrimination (N = 303), n (%) 111/174 (63.8) 81/129 (62.8) .476 
Participant characteristics In-home interview and blood draw completed, N = 408 In-home interview only, N = 216 p Value 
Age 73.6 (65–94) 74.2 (65–90) .223 
African American, n (%) 174 (43) 129 (60) <.001 
Female, n (%) 204 (50) (59) .027 
Rural, n (%) 184 (45) (56) .007 
Comorbidity count (baseline), M(SD1.9 (1.4) 2.1 (1.4) <.001 
    0, n (%) 60 (14.7) 25 (11.6)  
    4+, n (%) 51 (12.5) 33 (15.2)  
Baseline life space, M(SD72.5 (22.3) 64.9 (23.1) .101 
    0–29.9, n (%) 14 (3.4) 16 (7.4)  
    30–59.9, n (%) 86 (21.1) 72 (33.3)  
    60–89.9, n (%) 186 (45.6) 91 (42.1)  
    90+, n (%) 122 (29.9) 37 (17.1)  
Education category, M (SD2.8 (1.1) 2.5 (1.1) .004 
    Less than 6th, n (%) 66 (16.2) 50 (23.1)  
    Greater than high school, n (%) 137 (33.6) 52 (24.1)  
Cognitive evaluation (baseline), M (SD26.6 (3.6) 25.3 (4.3) <.001 
    <24, n (%) 70 (17.2) 59 (27.3)  
Income (baseline), M (SD4.15 (2.3) 3.38 (2.4) <.001 
    Less than $8,000, n (%) 72 (17.6) 58 (26.8)  
    More than 50,000, n (%) 43 (10.5) 19 (8.8)  
African American perceived discrimination (N = 303), n (%) 111/174 (63.8) 81/129 (62.8) .476 

The numbers of participants for each level of follow-up and study participation are shown by interviewer and participant pairings at baseline in Table 3; comparisons between interviewer/participant pairs were not significant for either African Americans or Whites. These results highlight that African American participants were less likely to participate in the laboratory procedures at four years. We examined the association of the race/ethnicity of the interviewer to the participant to determine if matching impacted participants’ willingness to consent to blood draws. The data suggest that the race/ethnicity of the interviewer did not impact the willingness of African Americans to participate in the blood draws (Table 3). Multivariable logistic regression models were developed in stages, first including age, race (African American was the referent category), gender (female referent category), and rural (referent category) versus urban residence. After including life space and income in the models, multivariable logistic regression predicting being contacted and interviewed at 48 months showed that those contacted had significantly higher income and cognition at baseline (Table 4).

Table 3.

Participation by Baseline Pairing of Interviewer and Participant

 Contacted at 4 years, N = 718 4 year in-home interview, N = 614 4 year blood draw, N = 401 
White interviewer, n (%) 
    White participant 335/348 (96.3)a 287/335 (85.7)a 213/287 (74.2)b 
    African American participant 309/333 (92.8)a 261/309 (84.5)a 152/261 (58.2)a 
African American interviewer, n (%) 
    White participant 37/39 (94.9)a 31/37 (83.8)a 19/31 (61.3)a 
    African American participant 37/41 (90.2)a 35/37 (94.2)a 17/35 (48.6)c 
 Contacted at 4 years, N = 718 4 year in-home interview, N = 614 4 year blood draw, N = 401 
White interviewer, n (%) 
    White participant 335/348 (96.3)a 287/335 (85.7)a 213/287 (74.2)b 
    African American participant 309/333 (92.8)a 261/309 (84.5)a 152/261 (58.2)a 
African American interviewer, n (%) 
    White participant 37/39 (94.9)a 31/37 (83.8)a 19/31 (61.3)a 
    African American participant 37/41 (90.2)a 35/37 (94.2)a 17/35 (48.6)c 

Notes. Interviews conducted by an Hispanic interviewer are excluded (22 at baseline); Means in the same column that share a superscript do not significantly differ; means in the same column that do not share superscripts differ at p < .001 in the Tukey’s honestly significant difference comparison.

Table 4.

Model Predicting Telephone Participation Versus No Participation at 48 Months

 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.945* (0.903–0.989) 0.974 (0.926–1.024) 0.977 (0.927–1.029) 
African American 0.544 (0.284–1.042) 0.972 (0.453–2.086) 1.603 (0.710–3.620) 
Female 1.143 (0.613–2.129) 1.141 (0.577–2.258) 1.358 (0.668–2.761) 
Rural 0.544 (0.285–1.041) 0.609 (0.313–1.184) .869 (0.430–1.758) 
Life space  1.007 (0.991–1.024) 0.998 (0.981–1.015) 
Comorbidity  1.082 (0.870–1.346) 1.066 (0.858–1.325) 
Cognition  1.108** (1.033–1.188) 1.092* (1.012–1.179) 
Education   1.061 (0.702–1.604) 
Income   1.402** (1.119–1.756) 
 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.945* (0.903–0.989) 0.974 (0.926–1.024) 0.977 (0.927–1.029) 
African American 0.544 (0.284–1.042) 0.972 (0.453–2.086) 1.603 (0.710–3.620) 
Female 1.143 (0.613–2.129) 1.141 (0.577–2.258) 1.358 (0.668–2.761) 
Rural 0.544 (0.285–1.041) 0.609 (0.313–1.184) .869 (0.430–1.758) 
Life space  1.007 (0.991–1.024) 0.998 (0.981–1.015) 
Comorbidity  1.082 (0.870–1.346) 1.066 (0.858–1.325) 
Cognition  1.108** (1.033–1.188) 1.092* (1.012–1.179) 
Education   1.061 (0.702–1.604) 
Income   1.402** (1.119–1.756) 

Notes.CI = confidence interval; OR = odds ratio.

*p < .05. **p < .01.

The contrast between persons having only telephone contact and participants willing to be interviewed in their homes was significant for age when considering only age, race, gender, and residence. When measures of baseline function were added to the model, the odds of African Americans were 2.2 more likely to be interviewed in person at 48 months. Higher life space and better cognition were also factors predicting who would have an in-home interview (see Table 5). Significant factors predicting participation in the blood draw following the in-home assessment included being White and urban (Table 6) when all factors were considered. The fully adjusted models for telephone participation, in-home assessments, and home visits with blood draws had Hosmer and Lemeshow goodness of fit p values of .559, .138, and .551, respectively. These values suggested that all three models adequately fit the study data.

Table 5.

Model Predicting Home Versus Telephone Participation at 48 Months

 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.924*** (0.896–0.954) 0.956* (0.923–0.989) 0.956* (0.924–0.990) 
African American 1.124 (0.740–1.706) 2.223** (1.327–3.725) 2.283** (1.310–3.978) 
Female 0.859 (0.565–1.307) 1.035 (0.648–1.653) 1.059 (0.660–1.700) 
Rural 0.801 (0.528–1.215) 0.831 (0.537–1.286) 0.844 (0.531–1.341) 
Life space  1.024*** (1.012–1.035) 1.023*** (1.011–1.035) 
Comorbidity  0.917 (0.797–1.054) 0.919 (0.799–1.058) 
Cognition  1.074* (1.015–1.136) 1.079* (1.015–1.147) 
Education   0.933 (0.707–1.230) 
Income   1.044 (0.915–1.190) 
 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.924*** (0.896–0.954) 0.956* (0.923–0.989) 0.956* (0.924–0.990) 
African American 1.124 (0.740–1.706) 2.223** (1.327–3.725) 2.283** (1.310–3.978) 
Female 0.859 (0.565–1.307) 1.035 (0.648–1.653) 1.059 (0.660–1.700) 
Rural 0.801 (0.528–1.215) 0.831 (0.537–1.286) 0.844 (0.531–1.341) 
Life space  1.024*** (1.012–1.035) 1.023*** (1.011–1.035) 
Comorbidity  0.917 (0.797–1.054) 0.919 (0.799–1.058) 
Cognition  1.074* (1.015–1.136) 1.079* (1.015–1.147) 
Education   0.933 (0.707–1.230) 
Income   1.044 (0.915–1.190) 

Notes.CI = confidence interval; OR = odds ratio.

*p < .05. **p < .01. ***p < .001.

Table 6.

Model Predicting Participation in Blood Draw at 48 Months

 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.988 (0.960–1.017) 1.001 (0.971–1.032) 1.001 (0.971–1.033) 
African American 0.497***(0.353–0.698) 0.638* (0.430–0.946) 0.631* (0.412–0.966) 
Female 0.676* (0.481–0.951) 0.715 (0.493–1.037) 0.724 (0.497–1.055) 
Rural 0.634** (0.452–0.890) 0.654* (0.461–0.928) 0.648* (0.450–0.933) 
Life space  1.007 (0.998–1.017) 1.007 (0.998–1.017) 
Comorbidity  0.953 (0.843–1.077) 0.955 (0.845–1.080) 
Cognition  1.041 (0.988–1.097) 1.048 (0.990–1.110) 
Education   0.938 (0.751–1.171) 
Income   1.011 (0.912–1.121) 
 Model 1, OR (95% CIModel 2, OR (95% CIModel 3, OR (95% CI
Age 0.988 (0.960–1.017) 1.001 (0.971–1.032) 1.001 (0.971–1.033) 
African American 0.497***(0.353–0.698) 0.638* (0.430–0.946) 0.631* (0.412–0.966) 
Female 0.676* (0.481–0.951) 0.715 (0.493–1.037) 0.724 (0.497–1.055) 
Rural 0.634** (0.452–0.890) 0.654* (0.461–0.928) 0.648* (0.450–0.933) 
Life space  1.007 (0.998–1.017) 1.007 (0.998–1.017) 
Comorbidity  0.953 (0.843–1.077) 0.955 (0.845–1.080) 
Cognition  1.041 (0.988–1.097) 1.048 (0.990–1.110) 
Education   0.938 (0.751–1.171) 
Income   1.011 (0.912–1.121) 

Notes.CI = confidence interval; OR = odds ratio.

*p < .05. **p < .01. ***p < .001.

Models were run separately for the African American participants and showed minimal differences from the models in Tables 4–6. Among African Americans, only rural residence significantly reduced the likelihood of participation in the blood draw. In models adding the baseline interviewer characteristics and history of perceived discrimination, neither were significant predictors of subsequent retention or level of participation. It should be noted that our findings do not preclude perceived discrimination as a barrier to initial recruitment.

Discussion

The level of subject retention achieved in the UAB Study of Aging (with 95% of survivors participating in a four-year follow-up telephone interview) emphasizes the importance of respecting all participants and highlights the value of participant identification with study personnel. Only 7 out of 1,000 participants (less than 1%) withdrew from the UAB Study of Aging, indicating a level of comfort with the research and research staff. Careful telephone follow-up and ascertainment of mortality over four-years (21.7%) ensured the ability to document participant eligibility for follow-up. As previously suggested by Yancey and colleagues (2006), retention was also strongly influenced by recruitment procedures of this observational cohort study.

The odds of African American participation in the four-year in-home assessment were 2.3 times greater than for White participants after controlling for function. This provided additional support to our hypothesis that willingness of research personnel to travel to participants’ homes and established relationships with research staff would increase comfort and trust among participants and improve retention. These results also suggest that physical disability may be responsible for lower retention of African Americans in research. Despite the willingness of African Americans to be interviewed in their homes, the odds of African Americans permitting a blood draw were nearly 40% lower than for Whites. Although our procedures fit nicely with the AHRQ model, we missed opportunities suggested by CBPR models that might have enhanced participation in the blood draw portion of our research. Such efforts would have involved active participation by the community in designing and conducting that portion of research (O’Fallon & Dearry, 2002).

This study supports the theoretical frameworks highlighting barriers and facilitators for inclusion in research. Potential participants need to be aware of the opportunity and have a reason to enroll. Initial recruitment for the UAB Study of Aging did not include a financial incentive, indicating that participants enrolled for other reasons. However, initial recruitment was based on consideration of participant burden, establishing trust, and eliciting a desire to benefit future older adults. By bringing the research to the participants, we hoped to reduce opportunity barriers and promote participation. The consent form noted that “you may not personally benefit from taking part of this study, but your participation may lead to knowledge that will help others.” Participants were invited to discuss their participation with family members and others before the interviewer came to the home. In some cases, participants had invited one of their children to review the consent form before the parent signed it.

Even though the UAB Study of Aging was designed prior to current research on recruitment and retention, it incorporated many of the procedures now recommended to enable inclusion of underrepresented groups. Travel to participant homes showed the willingness of research staff to reduce the burden of participation in terms of time and transportation. The face-to-face contact also allowed for identification with the study personnel. Interviewers were trained in communication skills and ways to make the interview a pleasant experience. Participants were given the opportunity to evaluate the interviewer who came to their home as a way to ensure that interviewers were respectful and to demonstrate that participant feelings were important. Provision of a toll-free number to contact research staff for any concerns or to report health-related events in their lives was another attempt to bridge the gap between the institution and the individual participants.

Of the 733 participants who continued in the study after four years of follow-up, a majority had completed all follow-up interviews, showing the importance of repeated contact at the convenience of the participant. Additionally, all follow-up interviews were conducted by the same research interviewer as frequently as possible. Timely completion of follow-up interviews most likely contributed to the establishment of a relationship with the participants and increased retention over four years. By the end of the four-year period, the interviewer reported that several participants said they had “been waiting for your call.”

Familiarity with research staff no doubt helped in establishing a trusting relationship so that the four-year in-home visit was accepted, even though it would mean meeting with a different interviewer. Persons who did not want to participate in the in-home visit at four years were more likely to be in poor health or ineligible for the in-home visit because they were no longer community dwelling in the study counties or needed a proxy respondent.

A prior history of perceived discrimination did not appear to impact willingness to interact with research staff after initial enrollment nor influence participation in the laboratory portion of the study. It is notable that older age and function predicted type of follow-up interview (telephone vs. in-home) but that once participants committed to the in-home interview these were not factors predicting who would provide blood samples. Our findings indicate that although interviewer/participant ethnic or racial matching might enhance recruitment, other interviewer characteristics can supersede this matching. Indeed, after controlling for function, in comparison to Whites, the African American participants were even more likely to invite our interviewers into their homes.

However, even with such positive retention, not all participants were willing to provide a blood sample. Rural residence and African American race were independent predictors of not being willing to have blood drawn, even when an in-home service was offered. Although it is likely that most of our participants would have been aware of the Tuskegee Syphilis Study, we think that it is unlikely that they had friends or family members that were involved because our participants were recruited from counties in Alabama that are more than 150 miles from Tuskegee. Nonetheless, this knowledge may have contributed to unwillingness to participate in laboratory procedures. More recent research indicates that mistrust of the medical establishment may be more widespread than previously thought and that a general lack of knowledge contributes to misunderstandings about research (Scharff et al., 2010).

Bivariate analyses indicated that the persons not participating in the blood draw had lower levels of education, but this was not a significant factor in the multivariable analyses. It should be noted that our measure of education, years of formal school attendance, is only one factor in estimating the quality of education (Crowe, Clay, Sawyer, Crowther, & Allman, 2008) and that education in rural areas may not have been equivalent to urban educational opportunities.

These results are similar to findings by Brown and Topcu (2003) in which African Americans were less likely to participate to obtain free medical care or to help scientists acquire knowledge. With the focus of current research on genetic biomarkers, the lower participation of the African American participants is disturbing. Reasons may parallel the lower level of influenza vaccinations among African Americans in the Veterans Affairs Healthcare System (Straits-Tröster et al., 2006)—that is, that participants reported not wanting the flu shot based on community word-of-mouth admonitions against it. Nearly 20% of all participants reported a fear of needles as a reason for not getting an immunization (Johnson, Nichol, & Lipczynski, 2008), and African American college women were more likely to report fear as a reason for not donating blood (Shaz et al., 2009). Even though the laboratory results from our study were not genetic, the refusal rates were higher for African Americans, possibly due to similar concerns of trust that have been noted by other studies of blood donations for genetic research (M. M. Williams et al., 2010).

Additional discussion with African Americans and rural residents is needed to identify reasons for nonparticipation in the blood draw component of the study. It may be that the in-home interviewers were not seen as part of the health care system and that reluctance to interact with medical personnel explained lower rates of participation in the blood draw. The mistrust of the utility of laboratory values and feelings that the tests would be of no personal benefit may have been a barrier as well as the inconvenience of providing a fasting blood draw. Bussey-Jones and colleagues (2010) surveyed participants in a cancer study about their willingness to provide an additional blood sample and found that African Americans were less likely to provide a blood sample, even though they had previously participated in a genetic epidemiology study. The focus group discussions to explore research participation by African Americans (M. M. Williams et al., 2010) noted that previous research in low-risk studies did not increase participation in research involving more invasive procedures. They suggest using strategies of CBPR to overcome mistrust, in particular thoughts that results would be used for reasons other than those disclosed by the researchers.

Implications and Future Directions

In summary, we have found that utilizing recruitment and follow-up methods that emphasize respect for all participants and identification with study personnel will facilitate recruitment and retention of both African American and White community-dwelling older adults. However, additional research is needed to understand and overcome factors that may limit African American retention and participation in research that involves blood draws for laboratory testing. Our data suggest that rural Whites as well as African Americans may be less likely to participate in research involving blood draws. Our results do not address what these barriers may be; qualitative research is needed to discover why our participants refused to have clinical procedure performed, even when the procedure was brought to them. Further research is needed to optimize participation and retention of underrepresented older adults in research studies, with specific attention to potentially unique facilitators and barriers in urban and rural settings.

Funding

The study described was supported by Award Number R01 AG16062, “Mobility Among Older African Americans and Whites—the University of Alabama at Birmingham Study of Aging;” P30AG031054, the Deep South Resource Center for Minority Aging Research, from the National Institute on Aging; and General Clinical Research Center grant M01 RR-00032, from the National Center for Research Resources.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Aging or the National Institutes of Health. The authors gratefully acknowledge the efforts of our interviewers, especially Ms. Teresa D. Tennyson.

References

Allman
RM
Sawyer
P
Roseman
JM
The UAB Study of Aging: Background and insights into life-space mobility among older Americans in rural and urban settings
Aging Health
 , 
2006
, vol. 
2
 (pg. 
417
-
429
doi:10.2217/1745509X.2.3.417
Anderson
RM
Revisiting the behavioral model and access to medical care: Does it matter?
Journal of Health and Social Behavior
 , 
1995
, vol. 
36
 (pg. 
1
-
10
doi:10.2307/2137284
Baker
PS
Bodner
EV
Allman
RM
Measuring life-space mobility in community-dwelling older adults
Journal of the American Geriatric Society
 , 
2003
, vol. 
51
 (pg. 
1610
-
1614
doi:10.1046/j.1532-5415.2003.51512.x
Brown
DR
Topcu
M
Willingness to participate in clinical treatment research among older African Americans and Whites
The Gerontologist
 , 
2003
, vol. 
14
 (pg. 
62
-
72
doi:10.1093/geront/43.1.62
Bussey-Jones
J
Garrett
J
Henderson
G
Moloney
M
Blumenthal
C
Corbie-Smith
G
The role of race and trust in tissue/blood donation for genetic research
Genetics in Medicine
 , 
2010
, vol. 
12
 (pg. 
116
-
121
doi:10.1093/geront/43.1.62
Charlson
ME
Pompei
P
Ales
KL
MacKenzie
CR
A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation
Journal of Chronic Disease
 , 
1987
, vol. 
40
 (pg. 
373
-
383
doi:10.1016/0021-9681(87)90171-8
Crowe
M
Clay
OJ
Sawyer
P
Crowther
MR
Allman
RM
Education and reading ability in relation to differences in cognitive screening between African American and Caucasian older adults
International Journal of Geriatric Psychiatry
 , 
2008
, vol. 
23
 (pg. 
222
-
223
doi:10.1002/gps.1902
Curry
L
Jackson
J
The science of including older ethnic and racial group participants in health related research
The Gerontologist
 , 
2003
, vol. 
43
 (pg. 
15
-
17
doi:10.1093/geront/43.1.15
Dilworth-Anderson
P
Williams
SW
Recruitment and retention strategies for longitudinal African American caregiving research. The Family Caregiving Project
Journal of Aging and Health
 , 
2004
, vol. 
16
 (pg. 
137S
-
156S
doi:10.1177/0898264304269725
Egede
LE
Bonadonna
RJ
Diabetes self-management in African Americans: An exploration of the role of fatalism
Diabetes Educator
 , 
2003
, vol. 
29
 (pg. 
105
-
115
doi:10.1177/014572170302900115
Folstein
MF
Folstein
S
McHugh
PR
Mini-mental state: A practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
 , 
1975
, vol. 
12
 (pg. 
189
-
198
)
Ford
JG
Howerton
MW
Bolen
S
Gary
TL
Lai
GY
Tiburt
J
, et al.  . 
Knowledge and access to information on recruitment of underrepresented populations to cancer clinical trials. Evidence report/technology assessment no. 122 (Prepared by the Johns Hopkins University Evidence-based Practice Center under Contract No. 290-02-0018). AHRQ Publication No. 05-E019-2
 , 
2005
Rockville, MD
Agency for Healthcare Research and Quality
Glanz
K
Rimer
BK
Lewis
FM
Health behavior and health education. Theory research and practice
 , 
2002
San Francisco
Wiley & Sons
Johnson
DR
Nichol
KL
Lipczynski
K
Barriers to adult immunization
American Journal of Medicine
 , 
2008
, vol. 
21
 (pg. 
S28
-
S35
doi:10.1016/j.amjmed.2008.05.005
Leung
MW
Yen
IH
Minkler
M
Community-based participatory research: A promising approach for increasing epidemiology's relevance in the 21st century
International Journal of Epidemiology
 , 
2004
, vol. 
33
 (pg. 
499
-
506
doi:10.1093/ije/dyh010
Levkoff
S
Sanchez
H
Lessons learned about minority recruitment and retention from the Centers on Minority Aging and Health Promotion
The Gerontologist
 , 
2003
, vol. 
43
 (pg. 
8
-
16
doi:10.1093/geront/43.1.18
Moreno-John
G
Gachie
A
Fleming
CM
Nápoles-Springer
A
Mutran
E
, et al.  . 
Ethnic minority older adults participating in clinical research: Developing trust
Journal of Aging and Health
 , 
2004
, vol. 
16
 (pg. 
93S
-
123S
doi:10.1177/0898264304268151
O’Fallon
L
Dearry
A
Community-based participatory research as a tool in advance environmental health sciences
Environmental Health Perspectives
 , 
2002
, vol. 
110
 
Suppl. 2
(pg. 
155
-
159
)
Park
NS
Roff
LL
Sun
F
Parker
MW
Klemmack
DL
Sawyer
P
, et al.  . 
Transportation difficulty of black and white rural older adults
Journal of Applied Gerontology
 , 
2010
, vol. 
29
 (pg. 
70
-
88
doi:10.1177/0733464809335597
Peek
ME
Sayad
JV
Markwardt
R
Fear, fatalism and breast cancer screening in low-income African American women: The role of clinicians and the health care system
Journal of General Internal Medicine
 , 
2008
, vol. 
23
 (pg. 
1848
-
1853
doi:10.1007/s11606-008-0756-0
Pfeiffer
E
A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients
Journal of the American Geriatric Society
 , 
1975
, vol. 
23
 (pg. 
433
-
441
)
Powe
NR
Gary
TL
Beech
BM
Goodman
M
Clinical trials
Race and research: Perspectives on minority participation in health studies
 , 
2004
Washington, DC
American Public Health Association
(pg. 
61
-
78
)
Rosenstock
IM
Strecher
VJ
Becker
MH
Social learning theory and the health belief model
Health Education Quarterly
 , 
1988
, vol. 
15
 (pg. 
175
-
183
doi:10.1177/109019818801500203
Sawyer
P
Allman
RM
Fry
PS
Keyes
CLM
Resilience in mobility in the context of chronic disease and aging: cross-sectional and prospective findings from the UAB Study of Aging
New frontiers of resilient aging, life strengths and wellness in late life
 , 
2010
New York
Cambridge University Press
(pg. 
310
-
339
)
Scharff
DP
Mathews
KJ
Jackson
P
Hoffsuemmer
J
More than Tuskegee: Understanding mistrust about research
Journal of Health Care for the Poor and Underserved
 , 
2010
, vol. 
21
 (pg. 
879
-
897
doi:10.1353/hpu.0.0323
Shaz
BH
Demmons
DG
Crittenden
CP
Carnevale
CV
Lee
M
Burnett
M
, et al.  . 
Motivators and barriers to blood donation in African American college students
Transfusion and Apheresis Science
 , 
2009
, vol. 
41
 (pg. 
191
-
197
doi:10.1016/j.transci.2009.09.005
Stahl
SM
Vasquez
L
Approaches to improving recruitment and retention of minority elders participating in research. Examples from selected research groups including the National Institute on Aging's Resource Centers for Minority Aging Research
Journal of Aging and Health
 , 
2004
, vol. 
16
 (pg. 
9S
-
17S
doi:10.1177/0898264304268146
Straits-Tröster
KA
Kahwati
LC
Kinsinger
LS
Orelien
J
Burdick
MB
Yevich
SJ
Racial/ethnic differences in influenza vaccination in the Veterans Affairs Health Care System
American Journal of Preventive Medicine
 , 
2006
, vol. 
31
 (pg. 
375
-
382
doi:10.1016/j.amepre.2006.07.018
United States Census
Census 2000 Summary File 1: Alabama
 , 
2000
 
Warnecke
RB
Oh
A
Breen
N
Breen
N
Gehlert
S
Paskett
E
, et al.  . 
Approaching health disparities from a population perspective: The National Institutes of Health Centers for Population Health and Health Disparities
American Journal of Public Health
 , 
2008
, vol. 
98
 (pg. 
1608
-
1615
doi:10.2105/AJPH.2006.102525
Wendler
D
Kington
R
Madans
J
Van Wye
G
Christ-Schmidt
H
Pratt
LA
, et al.  . 
Are racial and ethnic minorities less willing to participate in health research?
Public Library of Science Medicine
 , 
2006
, vol. 
3
 pg. 
e19
  
Williams
BR
Baker
PS
Allman
RM
Roseman
JM
The feminization of bereavement among community-dwelling older adults
Journal of Women and Aging
 , 
2006
, vol. 
18
 (pg. 
3
-
18
doi:10.1300/J074v18n03_02
Williams
MM
Scharff
DP
Mathews
KJ
Hoffsuemmer
JS
Jackson
P
Morris
JC
, et al.  . 
Barriers and facilitators of African American participation in Alzheimer disease biomarker research
Alzheimer Disease & Associated Disorders
 , 
2010
, vol. 
24
 (pg. 
S24
-
S29
doi:10.1097/WAS.0b013e3181fl4a14
Yancey
AK
Ortega
AN
Kumanyika
SK
Effective recruitment and retention of minority research participants
Annual Review of Public Health
 , 
2006
, vol. 
27
 (pg. 
1
-
28
doi:10.1146/annurev.publhealth.27.021405.102113

Author notes

Decision Editor: Letha A. Chadiha, PhD, MSW