|Home | About | Journals | Submit | Contact Us | Français|
The recruitment of minority and underserved individuals to research studies is often problematic. The purpose of this study was to describe the recruitment experiences of projects that actively recruited minority and underserved populations as part of The Centers for Population Health and Health Disparities (CPHHD) initiative.
Principal investigators and research staff from 17 research projects at eight institutions across the United States were surveyed about their recruitment experiences. Investigators reported the study purpose and design, recruitment methods employed, recruitment progress, problems or challenges to recruitment, strategies used to address these problems, and difficulties resulting from Institutional Review Board (IRB) or Health Insurance Portability and Accountability Act of 1996 (HIPAA) requirements. Additionally, information was collected about participant burden and compensation. Burden was classified on a three-level scale. Recruitment results were reported as of March 31, 2007.
Recruitment attainment ranged from 52% to 184% of the participant recruitment goals. Commonly reported recruitment problems included administrative issues, and difficulties with establishing community partnerships and contacting potential participants. Long study questionnaires, extended follow-up, and narrow eligibility criteria were also problematic. The majority of projects reported difficulties with IRB approvals, though few reported issues related to HIPAA requirements. Attempted solutions to recruitment problems varied across Centers and included using multiple recruitment sites and sources and culturally appropriate invitations to participate. Participant burden and compensation varied widely across the projects, however, accrual appeared to be inversely associated with the amount of participant burden for each project.
Recruitment of minority and underserved populations to clinical trials is necessary to increase study generalizbility and reduce health disparities. Our results demonstrate the importance of flexible study designs which allow adaptation to recruitment challenges. These experiences also highlight the importance of involving community members and reducing participant burden to achieve success in recruiting individuals from minority and underserved populations.
One of the regrettable hallmarks of the U.S. healthcare system is that substantial health disparities exist among population subgroups. Examples include the higher breast cancer mortality among African American women compared to Caucasian women , increased prevalence of diabetes among African Americans, Hispanics, and Native Americans [2, 3], and lower use of preventive screening among rural populations [4, 5]. Narrowing the often wide gaps in health among segments of the U.S. population will involve substantial research into the causes of these disparities and ways of preventing them.
Research intended to address health disparities requires adequate representation of those most affected by the disparities. Participating in research may decrease health disparities by giving the traditionally underserved access to the newest treatments and technologies available only through clinical trials [6, 7] and by improving the generalizability of trial results to minority and underserved populations [8–10]. Recruitment of minority populations, however, has proven to be a substantial challenge for investigators. Barriers have been identified at the patient, provider, system, and community level, and include mistrust of medical research [11–24], lack of awareness of available studies [12, 17, 20, 21, 25–28], economic burden [6, 12, 16, 28–32], and failure to meet eligibility criteria [33–35].
Suggestions have been made for frameworks that conceptualize and aid in the design of the recruitment process in order to effectively recruit members of underserved populations. Brown et al.  suggested that willingness to participate in research is determined by “awareness, acceptability, and access;” therefore, activities focused on education, generating social support from community leaders, and removing access barriers should result in improved recruitment of minorities. Paskett et al. proposed the Accrual to Clinical Trials (ACT) framework that focuses on identifying and addressing barriers at participant, provider, system, and community levels . Another conceptual framework by Brown and Topcu proposes that participation is influenced first by an individual’s desire to enroll and then by factors including their personal beliefs about the disease, past experiences with the healthcare system, and social support for their participation in the trial .
Common to all the proposed frameworks is the notion that the decision to participate is not influenced by individual-level factors alone, and that the characteristics and beliefs of others, including healthcare providers, the healthcare system, the community, and society affect an individual’s decision to participate in the trial. Thus, these frameworks suggest that the relevant communities must be involved in planning the study [17, 38–41], and to do this, understanding the culture and intracultural variation of the target community [42, 43] and the history of the community’s experience with the health care system  are important. If a study is to be successful in recruiting members of underserved and minority populations, strategies that involve all factors of the target population (i.e. “community”) must be utilized to identify and address critical cultural factors.
The Centers for Population Health and Health Disparities (CPHHD) initiative, funded by the National Institutes of Health (NIH), focuses on identifying and addressing health disparities among a variety of minority and underserved populations. The Centers each use Community-Based Participatory Research (CBPR) methodology, a strategy that involves the community in every facet of the research, in at least one of their projects. The goal of this paper is to describe the recruitment experience of the Centers in this initiative through March 31, 2007 and to provide examples of strategies employed to aid in recruiting underserved populations to research studies.
A total of twenty-one research projects involving the active recruitment of participants from minority and underserved populations and conducted through the CPHHD initiative at eight institutions in the U.S. were identified and included in this review of recruitment strategies. The projects described were still actively recruiting, and results represent information collected through March 31, 2007. Principal investigators and research staff from each project provided information using a standardized data collection form about the study purpose and design, recruitment methods employed, recruitment progress as of March 31, 2007, problems or challenges to recruitment, and problems resulting from Institutional Review Board (IRB) or Health Insurance Portability and Accountability Act of 1996 (HIPAA) requirements. Information about how these problems were addressed was also solicited from the Centers. Investigators at The Ohio State University CPHHD pilot tested the data collection form used in this study. The results were compiled and inspected for commonalities in the Centers’ experiences, and percentages were calculated to describe the frequency of the responses overall. Additionally, information describing participant burden and compensation for each participatory research project was obtained from each Center.
For four projects, recruitment occurred within another project at that center (e.g. the Boston Puerto Rican Center on Population Health and Health Disparities consisted of four separate projects, but only one overall instance of active recruitment). Thus, there were only 17 projects that implemented active recruitment strategies. Percentages were calculated using the total number of active recruitment instances (n=17) as the denominator. For simplicity, these instances of recruitments are referred to as “projects” in the reporting of results. Participant burden was categorized into three groups, high, medium, and low. High burden included tasks such as an extended follow-up and several medical tests, while medium burden included follow-up visits with minimal medical testing, such as a blood sample. Low participant burden was defined as participating in a one-time survey.
Following is a brief description of each Center and the project(s) included in this paper; this information is also summarized in Table 1. All projects were approved by their respective Institutional Review Boards.
The primary focus of the Center for Interdisciplinary Health Disparities Research at the University of Chicago was to study social, behavioral, and biological reasons for disparities in breast cancer between African American and Caucasian women. The Social Environment, Stress, and Health project was a community-based longitudinal study of African American women in Chicago with a new diagnosis of breast cancer. The purpose of this project was to study how factors related to the patients’ community, social situation, behaviors, and biological characteristics relate to breast cancer. Study participants were age 18 or older, African American, and non-Hispanic Caucasian women and provided data through in-person interviews.
Those enrolled in the study were followed for a combined twelve months, and were interviewed twice over two non-consecutive days at baseline, six months, and twelve months. Furthermore, blood samples were collected from all participants, and subjects consented to have previously excised tumors examined. Compensation was $30 per interview, for a potential total compensation of up to $180.
The primary focus of this center was to examine nutritional, genetic, and psychosocial factors that contribute to poor health outcomes in a longitudinal cohort of Hispanic men and women. To address this goal, this center conducted four research projects, all focusing on Puerto Rican men and women aged 45–74. Project 1 was a 2-year, prospective cohort study that investigated temporal associations between psychosocial stressors and certain disease outcomes, specifically diabetes and hypertension. Project 2 was a qualitative sociological evaluation in a representative subset of the cohort population aimed at understanding contextual sources of stress and adaptive mechanisms of adjusting to this stress. Project 3 examined two separate cohort subsets through intervention studies to evaluate the effectiveness of vitamin supplementation, food coupons and nutrition education, and social support over a two-year span. Project 4 investigated the relationship between these stressors, chronic disease outcomes, and genetics. Among a cohort subset, gene variants were determined, and responses to the varying interventions were compared.
Participants for all projects had a total follow-up length of two years. Only two interviews, however, were conducted—one at baseline and one at the end of the two-year follow-up. Phone contact was maintained every six months to administer a life events inventory. A blood sample was also collected at baseline and at the two-year follow-up. No other medical tests were performed. Compensation was $50 for both the baseline and two-year follow-up interview for a potential total of up to $100.
The Institute for Health Research and Policy Centers at the University of Illinois at Chicago applied, primarily, the elements of the Berkman and Glass model [44, 45] to address racial disparities surrounding the diagnosis, treatment, and prognosis of breast cancer through elements of multidisciplinary research and community-based partnerships. The Institute undertook four projects. Project 1: Neighborhood and Individual Effect on Stage at Diagnosis evaluated both individual and neighborhood effects on breast cancer stage at diagnosis. Additively; Project 2: Social Network Effects on Breast Cancer Prognosis evaluated the effect of social networks on breast cancer prognosis; Project 3: Breast Cancer Delay in Black, Hispanic and White Women addressed delay in breast cancer diagnosis and treatment among African American, Hispanic, and Caucasian women; and Project 4: Mediators of Ethnic Disparity in Breast Cancer Prognosis examined mediators and ethnic disparities in breast cancer prognosis. All four projects were cross-sectional studies and included African American, Hispanic, and Caucasian women aged 21–79. These multilevel projects were largely community-based, and focused on social, behavioral, and environmental issues.
Each project conducted at the University of Illinois at Chicago was a one-time, cross-sectional interview. Only Projects 1 and 4 collected a blood sample. Compensation for Projects 1 and 4 was, initially, $50 for the one-time interview. That amount, however, was increased for patients enrolled on or after February 1, 2007 to $100 for the baseline interview and an additional $50 for a blood sample. Projects 2 and 3 offered $25 for the completion of the survey, however, Project 2 stopped providing compensation for patients enrolled after January 31, 2007.
The Center for Population Health and Health Disparities at The Ohio State University (OSU) focused primarily on addressing the high rates of cervical cancer incidence and mortality in Appalachian Ohio. Partnering with the University of Michigan and using CBPR with the framework of the Social Determinants of Health model [44, 45], the OSU Center conducted three projects under the title Community Awareness, Resources and Education (CARE). All three projects examined cervical cancer in women from all races and ethnicities aged 18 or older. Project 1 investigated multi-level risks associated with Pap smear utilization and consisted of two phases. The first phase was a cross-sectional study analyzing risk factors associated with cervical cancer incidence, while the second phase examined the effectiveness of a lay health educator intervention to increase Pap smear utilization through a quasi-experimental, randomized control trial. Similarly, Project 2 evaluated the effectiveness of a lay health educator intervention in promoting smoking cessation in a quasi-experimental trial. Project 3 was a case-control study that evaluated the contribution of human papilloma virus (HPV), controlling for multilevel behaviors, to the incidence of cervical abnormalities in Appalachian Ohio.
Participants in Phase 1 of Project 1 responded to a one-time eligibility interview, and compensation was $10. The second phase of Project 1 involved two in-person visits and two phone calls from a lay health advisor to deliver a Pap test intervention for those randomized to the intervention group—one of each at baseline and one of each at the end of a 12–14 month follow-up. Compensation was $10 for responding to the baseline interview and another $10 for completing the 12–14 month follow-up for a total potential compensation of up to $20.
Participants in Project 2 were followed for 12 months and completed a baseline interview and interviews at three, six, and 12 months. Additionally, a saliva sample was collected. Compensation was $10 for completion of the baseline and 12 month interviews, and $15 for completion of interviews at three and six months. Total potential compensation for Project 2 was $50. Those randomized to the intervention group also received up to six visits by a trained lay health advisor who delivered the smoking cessation intervention over an eight-week period.
Eligible participants in Project 3 were contacted at baseline and assessed a second time once they were selected for inclusion in the study. Participants provided a blood sample, a second cervical smear and an HPV test sample during a routine Pap test. Participants were compensated $10 for the baseline survey, and another $10 if and when the second assessment was completed.
The RAND Corporation Center for Population Health and Health Disparities primarily studied the impact of neighborhoods on overall health, and how neighborhoods diffuse effects to the individual. Through community research, RAND also aimed to improve public policy regarding population health. The study consisted of four projects, however, only the first project involved recruiting community participants and was considered in this analysis. RAND conducted a cross-sectional study that focused primarily on determining the role of parks in physical activity and overall health. The study targeted all ages, all races, and both males and females, and examined the impact of developing and renovating recreational facilities on physical activity and other health outcomes. Participants were interviewed only once, and no compensation was provided.
The Center for Population Health and Health Disparities at the University of Pennsylvania concentrated on studying, evaluating, and disseminating information related to the interaction of biological, psychological, social, and clinical factors surrounding racial differences in prostate cancer. The center conducted two separate projects to examine prostate cancer. The Biological and Behavioral Predictors of Prostate Cancer project was a case-control study that examined the biological and behavioral predictors of prostate cancer recurrence in African-American and white men aged 35–90. The Determinants of Ethnic Differences in Quality of Life Following a Prostate Cancer Diagnosis project examined the psychological, cultural, and environmental determinants of ethnic differences in quality of life following prostate cancer diagnosis in non-Hispanic African American and white men aged 18–80 through a prospective longitudinal, CBPR study that was conducted collaboratively with the Philadelphia Chapter of the National Black Leadership Initiative on Cancer.
The Biological and Behavioral Predictors of Prostate Cancer project involved a one-time interview. Though follow-up is ongoing, it was clinically based via medical records. Only a subset of patients provided a blood sample. No other medical tests were required and no compensation was provided.
The Determinants of Ethnic Differences in Quality of Life Following a Prostate Cancer Diagnosis project followed participants for one year and involved four interviews, one at baseline and one at three, six, and 12 months. No blood samples were collected. To be eligible, however, men must have had a biopsy confirmed diagnosis of prostate cancer (though biopsy was not required for the study protocol). Compensation was $10 per interview for a potential total of up to $40 upon study completion.
The Center for Population Health and Health Disparities at the University of Texas Medical Branch (UTMB) managed two separate projects. The Environmental Risk, Coping, and Mexican Health Project, was a cross-sectional community study that examined the impact of environmental risk on health outcomes. This project enrolled both men and women aged 25 or older of all races, and involved a community-based, cross-sectional survey with subsequent follow-up interviews. The Liberty County Community-based Cancer Control Project, was a community-based longitudinal study that focused mainly on cancer incidence and prevalence in neighborhoods experiencing an influx of Hispanics. This project furthered the UTMB Center’s overall focus on the phenomenon of overall lower cancer incidence and prevalence, lower prevalence of overall disease, and lower associated mortality in neighborhoods that experience Hispanic immigration. This project enrolled both men and women of all races and ages in an attempt to understand this protective effect which has become known as the “Hispanic Paradox.”
The Environmental Risk, Coping, and Mexican Health Project followed participants for approximately five months. Investigators interviewed participants at baseline and at the end of follow-up. Participants provided a blood sample at the baseline interview. Additionally, height, weight, and blood pressure measurements were taken at baseline and at follow-up. Compensation was $15 for the baseline interview, $30 for providing a blood sample, and $15 for completing the follow-up questionnaire. Total potential compensation for participation was up to $60.
The Liberty County Community-based Cancer Control Project was a one-time, cross-sectional study. No biological samples were taken, though the project encouraged the community to increase screening adherence to mammograms and prostate-specific antigen (PSA) tests. No compensation was provided.
The Wayne State University Center of Urban and African American Health focused research efforts on understanding how the environmental, individual, biological, and genetic mechanisms operating in the surrounding Detroit community affect chronic disease conditions and their precursors. The Center conducted three individual projects.
The Exploring Changes in Experiences and Lifestyles (EXCEL) project was a randomized controlled trial that focused on African-American women aged 18–70 and the effects of lifestyle factors on breast cancer outcomes. This project was a 52 week study that conducted a baseline interview and interviews at three, six, and 12 months. A blood sample and Dual X-Ray were required. Participants received $15 and a $5 parking voucher per office visit. Total potential compensation was up to $100.
The Obesity, Nitric Oxide, Oxidative Stress and Salt Sensitivity Study (ONOSS) was a randomized controlled trial of African-American men and women aged 45 or older aimed at studying the effects of obesity, nitric oxide, oxidative stress, and salt sensitivity on cardiovascular disease risk. The project was a 78 week study and involved a baseline interview and interviews at three, six, 12, and 18 months. A blood sample was required of participants, as well as a Dual X-Ray, a gastric lavage, and an exercise test. Compensation was $40 each for the baseline interview and interviews at 12 and 18 months, plus an additional $80 dollars at each time point (baseline, 12 and 18 months) if a lavage was conducted. Compensation for interviews completed at three and six months was $15 each. A $5 parking voucher was also provided for each interview. Total potential compensation for participation was up to $415.
The final project, the Women’s Healthy Lifestyle Study (WHLS) was a randomized controlled trial that explored the effects of changes in experiences and lifestyle on cardiovascular disease. WHLS targeted African-American women aged 18 or older. The project was a 33 week study and involved 16 clinic visits and 17 phone calls. Participants provided a blood sample and consented to a Dual X-Ray, absorbitiometry and SphygmoCor® measurements, and ambulatory blood pressure management. Participants were compensated $15 per office visit and given a $5 parking voucher. Total potential compensation was up to $320.
The CPHHD projects varied in both disease and population focus, as previously described. The total number of participants to be recruited varied substantially among projects, as well (Table 2), ranging from 100 for the Women’s Healthy Lifestyle Study at Wayne State University to a maximum of 2,400 for three different projects among different Centers. The median number of participants to be recruited among all included projects was 500, and more than one-quarter of the projects sought to recruit at least 1,000 participants.
The CPHHD studies recruited participants using a variety of sites, sources and invitation procedures (Table 3). Most of the projects used multiple sites when recruiting participants, with 58.8% of the projects using at least two referral sites and 11.8% using at least three. The most common referral sites were community and hospital clinics. Nine of the seventeen projects reported using community clinics as a referral site for their participants. Similarly, hospital clinics were also important for recruitment, with six projects recruiting participants from patients seen in hospital clinics. Other common referral sources were community outreach (five), cancer registries (four), and hospital records (three). Less common referral sites were from another ongoing study (two), public records (two), random sampling of individuals through a multistage cluster design (two), parks and households (one), churches (one), and civic events (one).
Most projects also used multiple sources for referrals of potentially eligible participants. The majority of projects (82.3%) used at least two referral sources, while two projects reported using four referral sources. Ten projects (58.8%) reported using self-referral by potential participants (i.e. participants responded to a brochure, flyer, or other type of community posting). Research staff and clinic personnel were also important, with seven and five projects utilizing these sources, respectively. Six projects also reported using hospital or medical records as a referral source. Other less commonly used sources were from another ongoing study (two), location in the randomly selected census block (two), mass media (two), community coalitions (one), and cardiac rehabilitation programs (one).
Slightly over half (52.9%, nine) of the projects used only a single method of inviting potentially eligible participants to join their study, while six studies reported using two forms of invitations, and two studies reported using three. All but one of the projects (94.1%) reported using an in-person invitation procedure; that is participants were invited to participate through face-to-face encounters with clinic and/or research staff. Centers also commonly reported using mailed invitations either with (four) or without (four) telephone follow-up to invite potentially eligible subjects.
The total recruitment goal, accrual goal for recruitment through March 31, 2007, and the actual number recruited through that same date were reported for each project (Table 2). Fourteen of the seventeen (82.4%) projects with stated accrual goals for March 31, 2007 had achieved at least 75% of goal by this date. Recruitment to the March 31, 2007 goal ranged from 52.3% (158 recruited with a goal of 302) for phase 2 of project 1 at The Ohio State University to 184.0% (1,380 recruited with a goal of 750) for the Biological and Behavioral Predictors of Prostate Cancer Project at the University of Pennsylvania. Overall, six of the seventeen projects (35.3%) met or exceeded their accrual goal for the end of March, 2007.
In those studies desiring to recruit a multi-racial/ethnic population and reporting specific accrual results for each race/ethnicity, no clear pattern in terms of recruitment success was observed. For example, the Biological and Behavioral Predictors of Prostate Cancer Project at the University of Pennsylvania reported recruiting 1,149 Caucasian participants, more than two times their March 31, 2007 accrual goal of 500; yet they had recruited only 200 African American participants, eighty percent of their March 31, 2007 accrual goal of 250. Likewise, the Environmental Risk, Coping and Mexican Health Study at UTMB experienced somewhat greater success in recruiting non-Hispanics (1,297 to a March 31, 2007 accrual goal of 800 (162.1%)) than in recruiting Hispanics (1,458 to a March 31, 2007 accrual goal of 1,600 (91.1%)). Conversely, projects 1 and 4 at the University of Illinois at Chicago reported greater success in recruiting minorities (297 African Americans to a March 31, 2007 goal of 270 (110.0%), and 78 Hispanics to a March 31, 2007 goal of 96 (81.3%)) than they did in recruiting Caucasian participants (185 to a March 31, 2007 goal of 270 (68.5%)).
The projects reported a wide variety of problems and challenges encountered during recruitment (Table 4). One of the most common types of problems reported was related to the coordination with clinics used for recruitment; 52.9% of the projects reported at least one problem in this category. Coordination problems with clinics included difficulty in identifying a clinic willing to participate in recruitment, as well as coordination with the clinic’s administration. Investigators noted that the clinic staff was already busy and lacked the staffing requirements necessary for adequate recruitment and for carrying out study activities. Further, one project reported that physicians and staff often did not make study recruitment a priority in their daily activities. Recruitment was also hindered by the lack of space available to research staff in the clinics, making on-site interviewing of potential participants difficult. Finally, one project also noted that the clinic restricted the research staff’s access to the charts of potentially eligible patients.
Other common challenges to recruitment were related to administrative issues. Over half of all projects (58.8%) reported at least one problem/challenge that fell into this category. Projects reported poor cooperation from local hospitals and physicians as well as poor communication between researchers and clinic personnel. One Center reported significant difficulty in acquiring their state cancer registry data which were to be used for recruitment because of difficulties encountered in trying to obtain a contract with the state health department. Another project reported problems in establishing partnerships within the communities they were targeting for recruitment. Staffing problems were also commonly reported, including difficulties in managing research staff working away from the university and not having enough personnel to adequately manage recruitment.
Many projects (47.6%) also reported problems with eligible participants who refused to respond, failed to show up for scheduled appointments, or disregarded follow-up attempts made by mail and/or telephone. Further, many individuals were reportedly either unwilling to participate or were initially willing but later changed their minds. When participants did schedule clinic appointments, these were frequently cancelled. Investigators also reported difficulty in locating potential participants, due to the targeted population being either highly transient or spread across a large geographic area. One Center also reported that potential participants were mistrustful of research and lacked an understanding about the potential benefits of research to their community.
Characteristics of the projects themselves also presented challenges to recruitment. Two Centers reported that completion of a lengthy study questionnaire was a barrier to recruitment, while another reported that the long duration of follow-up was a barrier. Problems with the eligibility criteria were also reported, including criteria that were too narrow and inconsistent with the culture of the target population, exclusion of many participants due to high body mass index (BMI), and simply having a lower rate of eligible individuals than initially estimated. Two Centers reported that the timing of their recruitment was shortly after diagnosis or surgery for the condition under study, causing a burden for potential participants, and hindering recruitment. Other commonly reported problems were related to the busy schedules of potential participants and the limited time available for participant recruitment. UTMB’s Environmental Risk, Coping and Mexican Health Study experienced extremely unique problems related to the occurrence of local disasters, including an explosion at a nearby chemical plant and the landfall of Hurricane Katrina.
Ten of the projects (58.8%) reported problems related to IRB approval or restrictions (Table 5), with the time required to obtain IRB approval being the most common problem reported. Both the time needed for preparation of the IRB application and the time needed to obtain approval were often longer than anticipated, leading to delays in beginning recruitment. Projects also reported that their IRBs were overburdened and restricted the number of amendments that could be submitted at one time. Other problems included needing additional certificates of confidentiality or consents, difficulties in coordinating IRB approval at multiple institutions, and problems related to the use of non-university affiliated clinics and clinic staff for recruitment.
Only three Centers reported having recruitment problems related to HIPAA requirements (Table 5). The projects at OSU reported difficulty due to differing interpretations of HIPAA by the researchers and staff at each participating clinic, as well as restrictions from HIPAA that would only allow contact with clinic patients through clinic staff. UTMB’s Liberty County Community-based Cancer Control Project experienced similar challenges due to HIPAA restrictions that disallowed direct contact of participants by the research staff, and required adherence to strict privacy guidelines to prevent even perceived breaches of privacy. One project at the University of Pennsylvania also reported HIPAA-related problems, specifically that recruitment was prevented due to lack of private space in which to interview patients.
Efforts to address recruitment problems were as diverse as the problems themselves (Table 4) and focused on issues at the different levels described in the ACT framework. The most common attempted solution was community outreach, with 71.4% of all projects reporting at least one solution in this category. Attempts at community outreach included establishing partnerships with community representatives and organizations, soliciting help from a previously established community advisory board, holding meetings with cancer registrars and physicians, sponsoring frequent, in-depth presentations to groups of potentially eligible individuals, emphasizing the benefits of the research to the community, and providing incentives and personalized recruitment materials.
Protocol changes in response to recruitment difficulties were also commonly employed (eight projects). In some cases the eligibility criteria or recruitment goals were revised, while in other cases the timing of enrollment or the length of the study was modified. Modifications were also made to recruitment procedures, including extending study area boundaries, implementing on-site recruitment, and establishing a toll-free number to encourage follow-up from participants. Other protocol changes included offering extra assurances of confidentiality, changing the questionnaire content or the timing of administration, and providing gift certificates as incentives for participation.
Changes to study management were implemented by nine projects in response to recruitment challenges. Such changes included developing computerized data tracking and reminder systems. One project retrained their interviewers in how to present the study to potential participants, while others increased their personnel effort devoted to recruitment. Multiple projects increased the number and frequency of meetings held with key players in recruitment, including providing updates to clinics about recruitment at their site. Other efforts to address recruitment problems included enlisting the help of additional physicians or hospitals, and hiring additional personnel to support recruitment efforts. Some projects also extended their work hours and provided home visits to accommodate the busy schedules of potential participants. UTMB’s Environmental Risk, Coping and Mexican American Health Study addressed their disaster-related problems by delaying their scheduled interviews while the community dealt with the plant explosion and by ensuring the security of their data during Hurricane Katrina.
Seven (41.2%) of the projects that actively recruited participants had a medium participant burden. Six (35.3%) had low participant burden, and only four (23.5%) projects had high participant burden. None of the projects with high participant burden reached their March 31, 2007 accrual goal, while 28.6% of projects with medium burden and 66.7% of projects with low participant burden reached their March 31, 2007 accrual goals. Median total potential compensation for projects with high, medium, and low participant burden was $250, $50, and zero dollars, respectively. Most projects (nine, 52.9%) provided participants with some compensation but less than $100. Four projects (23.5%), all with low participant burden, provided no compensation, and four projects (23.5%), all with high participant burden, provided compensation in excess of $100. These results are summarized in Table 6.
The Centers for Population Health and Health Disparities initiative focuses on understanding and addressing various health disparities in different populations that suffer a disproportionate disease burden. This report summarizes both the recruitment issues surrounding minority and underserved populations served by the CPHHD and the efforts made to address these emerging recruitment problems. Consistent with previous research, strategies for reaching recruitment goals used by almost all of the Centers were similar to those suggested in the ACT framework . These strategies include using multiple recruitment sites and sources, using members of the target population in planning efforts, educating the population about the importance of the study, improving staff sensitivity, providing personal, culturally appropriate invitations to participate in studies, and providing either compensation or benefit to community members for their participation [20, 46]. By using a recruitment framework, effective strategies may be incorporated at multiple levels (individual, provider, healthcare system, community) to improve recruitment. The rate of recruitment as of March 31, 2007 varied from 52% to 184% of the recruitment goals for that date, though some sites reported either much lower or much higher recruitment of underserved and minority populations.
Common problems and challenges to recruitment were reported, with the most prevalent problem being related to the clinics where participants were being recruited. Specifically, researchers experienced difficulty working with clinic staff and physicians to implement recruitment procedures. These findings suggest that although many research studies rely on physicians to refer potentially eligible patients, physicians themselves can act as barriers to recruitment. Many physicians lack awareness of ongoing research projects [18, 20, 47, 48], worry that they will lose control over their patients’ medical care and damage the doctor-patient relationship by referring patients to a study [6, 18, 20, 48, 49], believe research activities would take up too much of their time [12, 16, 49], and view clinical trials as complex and providing care that is inferior to standard treatment . Specific to minority recruitment, physician-level barriers further include having few minority patients in their practices , and perceptions that minority patients are less likely to want to participate in research .
Other common problems included administrative issues, forming partnerships with communities, staffing changes, and difficulties contacting potential participants. Characteristics of the research studies themselves also posed recruitment problems including long questionnaires, long follow-up duration, and narrow eligibility criteria. Notably, the recruitment rate appeared to vary by the amount of participant burden for each project, while compensation seemed to have little impact on the rate of recruitment. While this study was unable to test this empirically, future studies should directly examine the relationship between the level of burden placed on participants and their corresponding likelihood of study participation and overall rate of accrual. The majority of projects reported difficulties with IRB approvals that caused delays in starting recruitment. Conversely, HIPAA requirements did not negatively impact most of the projects.
As part of the CPHHD initiative, efforts to address recruitment problems included community outreach, amending the study protocol, and modifying recruitment procedures, questionnaires, and management strategies. These strategies were implemented individually by some Centers, and either incrementally or concurrently by other Centers. Some of these same strategies have resulted in successful recruitment in previous studies [34, 52–68].
As with any study, there are several limitations and strengths in summarizing the Centers’ approach to participant recruitment. Limitations include difficulties comparing Centers with different study designs, variation in population size and the number recruited across the Centers, and the fact that the participants were recruited from sites that differed in both demographic makeup and geographic location. Additionally, Centers differed in their reporting methods, with some Centers reporting recruitment information for each research project and other Centers reporting information for the entire Center. Specifically, these limitations prevent us from explicitly determining what exact strategies worked in the context of CPHHD recruitment.
While this study examined accrual goal attainment percentages, implemented strategies were not specifically tested, either within or between Centers, because study designs and recruitment methods varied so dramatically across projects. Compounding this difficulty was the fact that recruitment plans themselves demand flexibility over time and in reality no control groups could have been established to compare attempted recruitment solutions. Thus, as described by Larkey et al., while many of the strategies implemented may have improved minority and underserved accrual to the projects, this analysis of the CPHHD Centers recruitment strategies cannot distinguish which recruitment efforts were most effective . Furthermore, not meeting recruitment goals may be a function of varying research designs, rather than similar recruitment problems related to underserved and minority populations.
Although vast differences between the Centers can be considered limitations to a degree, they have also illuminated the immense diversity of the recruitment strategies occurring at the Centers. This study provides a broad overview of problems faced by investigators attempting to recruit minority and underserved populations in clinical research with the overall goal of reducing health disparities and increasing the generalizability of clinical trial results. Common problems associated with participant recruitment have emerged and served as a catalyst for the CPHHD investigators to be flexible in their strategies to meet the various study recruitment goals. The importance of being able to modify recruitment strategies after the study begins and adhere to community-based, multilevel frameworks, while maintaining scientific integrity is an important lesson learned by this evaluation of participant recruitment across the CPHHD centers. Finally, as with any study that attempts to actively recruit participants, careful attention should be paid to the amount of burden placed on and compensation provided to those participating in the research. Future studies should build upon these findings, and look beyond unique and similar challenges and attempted solutions to such challenges between varying minority and underserved recruitment efforts to examine the effectiveness of specific recruitment techniques. These observations must be considered in future research studies in order to make serious attempts to reduce health disparities.
P50 ES012395, P50 ES012382, P50 ES012383, P01 AG02323394, P50 CA1065631, P50 CA015632, P50 CA106743, P50 CA105641
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.