Costs and Strategies in Minority Recruitment for Osteoporosis Research*

Abstract

To meet expectations for the participation of minority populations in research, we committed to enroll 140 minority subjects in addition to a random sample of Olmsted County, Minnesota residents (90% white) for a study of risk factors for age‐related bone loss and fractures. We successfully enrolled 597 additional minority subjects but encountered specific problems with respect to identification of potential subjects, recruitment, obtaining informed consent, transportation to the study site, and collecting study data. These problems were resolved by observing the tenets of outreach to a diverse study population, namely (1) understand the target population; (2) establish explicit recruitment goals; (3) agree on research plans between study staff and minority communities; (4) continuously evaluate the recruitment process; and (5) maintain lines of communication. Success depended especially on the recruitment of cultural advisors from the different ethnic groups. These special efforts increased the recruitment cost substantially; the total expense of $122,000 for recruiting 550 Asian, Hispanic, and Somali subjects was almost 5‐fold higher than the $26,000 required to recruit 699 mostly white study subjects from the population who were contacted by mail. Although it is not impossible to recruit minority subjects, investigators (and grant reviewers) should recognize that significant resources are required to gain access to ethnic communities for research. These results should contribute to more realistic budgets for recruiting minority subjects into clinical research studies.

INTRODUCTION

THE NATIONAL Institutes of Health (NIH) recently reemphasized guidelines for the inclusion of ethnic minorities in research.⁽¹⁾ Neither cost nor inconvenience is an acceptable excuse for failure to satisfy these dictates,⁽²⁾ yet recruiting minority subjects is a challenging task.⁽³⁾ The process requires deep cultural insight, the creation of a multicultural investigative team, careful planning and continuous evaluation, and a considerable amount of resources. The purpose of this report is to review our experience in recruiting for studies of age‐related bone loss and fractures. We identify key barriers encountered in the process of recruiting ethnic minority subjects, few of whom had ever heard of osteoporosis, and estimate the resources required to overcome these obstacles. If investigators fail to anticipate these additional costs in budgeting for their studies, the NIH guidelines effectively become an unfunded mandate.

UNDERREPRESENTATION OF MINORITIES IN RESEARCH

It has long been appreciated that subsets of the population may differ with respect to risk factors for disease or response to treatment, but few data were available because studies tended to focus on white men.⁽⁴⁾ To overcome this problem, in 1986, the NIH established policies to increasethe representation of women and minorities in research.^{(5, 6)} When an audit revealed that these groups were still being excluded, these policies were reemphasized,^{(7, 8)} along with explicit instructions to NIH staff regarding their implementation.⁽⁹⁾ As clinical trials continued to include predominantly middle‐class white males,⁽¹⁰⁾ Congress ruled in the 1993 National Institutes of Health Revitalization Act⁽¹¹⁾ that women and members of minority groups must be included in these studies, barring some legitimate reason for their exclusion (e.g., unaffected by the disease in question). The resulting regulations dictate that NIH‐supported grantees must ensure the appropriate inclusion of women and minorities in their research.⁽²⁾ The regulations specifically state that cost is not an acceptable reason for their exclusion.

APPLICATION TO A SPECIFIC STUDY

These directives posed a particular problem for investigators at the Mayo Clinic because the region is predominantly white. As recently as 1980, the Olmsted County population was 98% white, although an influx of refugees from Southeast Asia and Somalia and a growing number of black and Hispanic residents raised the nonwhite component of the population to 5% in 1990 and to 10% in 2000. Warned that our competing NIH grant application might not be approved if minority representation were not enhanced, we committed to enrolling 140 minority subjects ≥20 years of age in addition to a cohort of 700 women and men randomly sampled from the Olmsted County population. In approving our 1994 competitive renewal, the Epidemiology and Disease Control Study Section expressed some skepticism about this, saying “… the investigators have never had experience recruiting minority subjects in their previous work. There may be different issues in recruitment of these groups that are not anticipated in the proposal.” In fact, we far exceeded our goal (Table 1). In addition to 699 randomly sampled Olmsted County residents, of whom 15 were minority subjects,⁽¹²⁾ we have enrolled over 597 additional minority subjects to date, including residents originally from Vietnam, Cambodia, Laos, and Somalia, as well as a group of Hispanic migrant workers. Recruitment is ongoing for black residents of the community.

This gratifying result can be attributed to the fact that the Mayo Clinic had already initiated efforts to facilitate minority recruitment. In 1994, a task force recommended the creation of an Office for Diversity in Clinical Research, to provide investigators with advice on appropriate protocol designs, and establishment of the Mayo Medical Center Diversity Committee, to foster new educational and treatment programs in the community and to develop a cultural awareness program for Mayo staff. However, the crucial step in our own success was the organization of a minority recruitment team, wherein leaders from each ethnic population were enlisted as advisors to assist in planning and recruitment efforts. This was helpful because recruitment strategies must be culturally sensitive,^{(13, 14)} yet the minority population is quite diverse,^(10,15–18) and study designs must acknowledge the traditions, norms, values, and beliefs of each community.^{(19, 20)} Moreover, while it is clear that cultural, language, and economic barriers affect research participation by all minority subjects,^{(15, 16)} it is the barriers to recruitment of black and Hispanic subjects that have received the most attention in the literature.^(10,21–24) Little was known about such barriers for the Southeast Asian and Somali populations.⁽²⁵⁾

ISSUES IN RECRUITING DIVERSE STUDY POPULATIONS

The minority recruitment team devised culturally sensitive approaches to identification, recruitment, and examination of subjects from different minority populations in conjunction with our ongoing studies of osteoporosis in the predominantly white population of Olmsted County. Numerous operational problems had to be overcome, however. Solutions required an in‐depth knowledge of the different ethnic groups, as well as a substantial amount of ingenuity and flexibility. Compared with the cost of studying the majority population, considerable additional resources were incurred for recruitment. These are summarized in Table 1 and delineated in the sections that follow.

Identification

Using the resources of the Rochester Epidemiology Project,⁽²⁶⁾ the Olmsted County population (73% urban; 90% white) can be enumerated, and our primary study cohort was identified by drawing a random sample of the general population stratified by sex and decade of age (20–29,…, ≥ 80 years).⁽¹²⁾ However, a random sample of this population produces few minority subjects by chance alone. Consequently, community leaders from each minority group were hired to serve as cultural issue advisors, contact persons, and interpreters. They were particularly helpful in identifying potential participants. Thus, to enroll as many Olmsted County residents of Vietnamese heritage as possible, we began with a list of names created by Tosomeen et al.⁽²⁷⁾; this was combined with a search for common Vietnamese last names in the Mayo patient registry and comparable telephone directory listings and individuals known to the advisor. Of 1021 potential participants, we were only able to contact 360. Of these, 93 were ineligible (64 were not Vietnamese, 12 were pregnant or breastfeeding, 16 were disabled, and 1 had died); 31 other Vietnamese volunteers heard about the study and asked to be included. For the Cambodian and Laotian groups, we used lists accumulated by the advisors combined with directory listings for common Cambodian and Laotian last names. Of 420 names obtained, 316 were successfully contacted. Because the Southeast Asian population, like other migrant groups, is a mobile one, the main reason for failure to make contact was that the person had moved. For the Somali population, 86 names were obtained from a list created for a Somali prenatal care project combined with names presented by a community leader and interpreter for the Mayo Clinic International Department. We were able to contact 70 Somali women, of whom 66 enrolled in the study. All attempts to enroll Somali men were ineffective. The other black subjects were found by working through a local church with strong ties to that community, with assistance from the cultural advisors. Potential Hispanic subjects were also identified by the respective cultural advisor who was well known in the community of seasonal workers, most of whom were used by local canning companies. The resulting minority subjects were younger on average than the population sample (mean age, 43.6 vs. 55.1 years), but this was mainly because of the different age distributions of the underlying minority and white populations (10% vs. 27% ≥50 years of age).

Recruitment

Mayo guidelines dictate that an explanatory recruitment letter precede any telephone contact, and this strategy was used to contact the subjects in the population cohort. However, we were persuaded by the cultural advisors that this was not feasible for some minority populations because of the language barrier. Even if the contact letter were translated into the appropriate language, some potential subjects have low reading skills and might have had difficulty understanding the implications of participation in a research project. Thus, it was necessary to employ bilingual advisors to call or otherwise contact (some have no telephone) potential subjects to explain the study, using a telephone dialogue approved in advance by Mayo's Institutional Review Board. The advisors reassured potential participants that confidentiality would be maintained at all times, and they assisted in building trust with individual subjects. The total costs of the services provided by the cultural advisors, including study planning, identification of potential study subjects, translation of the consent forms and result letters, and assistance in escorting and interpreting for subjects on the day of the study are shown in Table 2.

Consenting

White and black subjects were provided with a consent form written in English, allowed to ask questions, and then to participate in the study if they so chose. The process was more involved and time consuming (20–40 minutes) for the other minority subjects, who were provided with a copy of the consent form in English as well as a separate copy translated into their own native language. Because of the legalistic text and the low literacy skills of some of the non‐English speaking population, a verbal explanation of patient rights contained in the consent form was provided. After this explanation, the cultural advisor, acting as interpreter, read the ethnic consent form aloud. A member of the investigative team answered all ensuing questions, and participants were then allowed to sign either the ethnic, the English, or both consent forms.

Transportation

For long‐term residents of the community, it was sufficient to schedule a date for the examination. However, members of the minority population may have difficulties obtaining transportation and childcare services, affecting both their access to health care and their adherence to clinic appointments and research protocols.^{(10, 28)} Therefore, a member of the minority community was hired to provide transportation to and from the study site (Table 2). This was feasible because most of the subjects resided in the central city of Rochester; transportation could pose greater difficulties in a more rural population. Subjects were called the evening before the clinic visit to set a specific pick‐up time, and the cultural advisor accompanied the driver at all times to diminish subjects' fears of riding with an unfamiliar face, to respect cultural traditions concerning women riding alone, and to overcome language barriers. During the ride to the Mayo General Clinical Research Center (GCRC), the advisor greeted the subjects, answered additional questions about the research process, and tried to reduce subjects' anxiety. The driver subsequently acted as navigator as the participants moved through the various stations (sample collection, medical history, dietary history, and bone density measurement) at the GCRC.

Data Collection

White and black subjects were scheduled individually for the baseline examination. For the non‐English speaking subjects, GCRC visits were organized for groups of six to seven subjects to encourage families and friends to come together, to reduce fears, to increase social interactions, and to reduce costs. In accordance with cultural norms, minority participants were greeted and briefed on the morning events, and tea was served to create a friendly atmosphere. After the consent procedure, a lecture followed that was designed to give the minority participants some basic information about osteoporosis risk factors and prevention. This took approximately 1 h and was well received. Simultaneous translation was supported by visual aids and written materials in the appropriate language. A subsequent group tour of the bone densitometry device, accompanied by the interpreter, minimized the language barrier between technician and participant and eliminated the need for an interpreter when the subjects were later scanned.

Along with an interpreter, a member of the investigative team then interviewed each participant privately to collect pertinent medical history data. Interestingly, the rapport established during the medical history interview allowed subjects to openly discuss personal health concerns. Likewise, dietitians were assisted by an interpreter while interviewing non‐English speaking participants about dietary intakes using a modified tool focused on the ethnic diet. At the end of the examination, an ethnic lunch was catered by a local restaurant. This proved to be very important for the non‐English speaking participants, who appreciated the chance to socialize with each other and with project staff. The cost associated with these additional activities is also noted in Table 2.

DISCUSSION

As a result of these efforts, our overall study population is 57% female and 47% nonwhite, both in excess of comparable proportions in Olmsted County (51% and 10%, respectively). This is important because the random population sample produced only 15 minority subjects,⁽¹²⁾ which is obviously insufficient for any detailed analysis. Instead, the large number of Southeast Asian subjects allowed us to show that much of their apparent deficit in bone density was an artifact of smaller bone size compared with whites.⁽²⁹⁾ Similarly, correction for bone size eliminated the apparent deficit in bone density among Somali compared with white women but did not reduce differences between black women and white women or Somali immigrants.⁽³⁰⁾ This, along with differences observed in size‐corrected bone density among the various Southeast Asian subgroups, reinforces the notion that particular races (e.g., “Asian”) should not be viewed as homogeneous populations with respect to their bone metabolism but rather that efforts be made to explore differences in the component ethnic groups.⁽³¹⁾ Even after adjusting for bone size, 32% and 9% of Southeast Asian women and men, respectively, would be considered to have osteoporosis at the femoral neck according to World Health Organization criteria.⁽²⁹⁾ This further indicates a need for appropriate educational interventions for minority populations and for physicians to pursue diagnosis and treatment efforts to prevent osteoporosis‐related disabilities in these individuals.

These results were possible because the minority recruitment effort observed the tenets of reaching out to a diverse study population: (1) understand the target population; (2) establish explicit recruitment goals; (3) agree on research plans between study staff and minority communities; (4) continuously evaluate the recruitment process; and (5) maintain lines of communication.⁽³⁾ However, the operationalization of these principles depended on participation and endorsement by key community leaders who served as advisors and interpreters. These bilingual individuals with strong ties to the minority populations were recruited through networking in the community and by job announcements posted at the Mayo Clinic. The fact that they were involved in the study themselves helped minimize anxiety among potential participants and allowed the advisors to give a firsthand account of the procedures. Hiring an additional member of the ethnic community as an escort and interpreter also contributed to reducing language barriers and enhancing patient compliance with scheduled appointments.

Recruitment costs were greatest for the Somali group ($417 per subject), followed by Hispanic migrant workers ($240) and the Laotian ($221) and Cambodian ($211) groups; the Vietnamese had the lowest cost per subject ($148). All of these differed substantially from the cost of $38 per subject for whites, who were recruited by mail. The high cost for the Somali group was caused mainly by the small number participating in each educational session (an average of three per session vs. six per session for the Vietnamese). The group sessions were designed to make the research experience more attractive and to reduce the cost of translation, transportation, and special arrangements by distributing the fixed costs per session across all participants. This related especially to the didactic educational presentations, which were vital for providing some mutual benefit to the participants⁽¹⁾ by increasing their basic knowledge about osteoporosis, current prevention strategies, including the calcium content of their own diet, and the implications of a longer life expectancy in the United States. Although this is an effective option to increase minority participation in clinical studies,⁽³²⁾ a minimum number of participants per session is required to maximize the additional investment.

Of particular importance in designing minority studies is information about the needs of the target populations, as well as alternative strategies for contacting them and providing clear information about the study process. We had the flexibility to work with one ethnic group at a time and therefore were able to focus on the specific cultural traditions of that population. Our strategy also provided an opportunity to gain experience in organizing study procedures, build on previous experiences, and adjust to the immediate needs of each ethnic group. In addition, in‐service training for the GCRC staff served to develop awareness of ethnic traditions and to decrease anxiety among the research team concerning language barriers and cultural differences. Providing the subjects with transportation, ethnic food, educational materials in the appropriate language, and direct contact with health professionals created an atmosphere that people enjoyed and valued. Thus, it is not impossible to recruit minority subjects, but additional efforts are needed. Specifically, investigators (and grant reviewers) should appreciate the fact that significant resources are required to gain access to ethnic communities, involve minority leaders in planning and recruitment, coordinate logistics, and offer appropriate incentives. In this study, the recruitment cost per minority subject averaged $180 more than that for the population sample, and the total expense of $122,000 for recruiting 550 Asian, Hispanic, and Somali subjects was almost 5‐fold higher than the $26,000 required to recruit 699 mostly white study subjects from the general population. These data should contribute to more detailed planning and more realistic budgets for recruiting minority subjects into clinical research studies.

Acknowledgements

This work was supported by research grants AR27065, AG04875, and RR00585 from the National Institutes of Health, U.S. Public Health Service. The authors thank the following cultural advisors and interpreters for their invaluable assistance with recruiting: Mrs. Nga Edmonson and Mr. Han Hoang (Vietnamese); Mr. Ponneareay Peng and Mrs. Alice Sinnas (Cambodian); Mr. Oukhane Phrachansisourath, Mrs. Vanphen Phrachansisourath, and Miss Samone Phrachansisourath (Laotian); Dr. Mahamoud A. Jimale and Mrs. Udbi Wallin (Somali); Mrs. Ea Porter and Mrs. Amy White (black); and Mrs. Blanca Pineda (Hispanic). We also thank Mrs. Veronica Gathje, Mrs. Celia Wright, Mrs. Margaret Holets, Dr. Anna Georgiopoulos, Dr. Jana Rivera, and Dr. Lisa Buenger for help with data collection; Ms. Sara Achenbach for assistance with data analysis; and Mrs. Mary Roberts for help in preparing the manuscript.

References

Author notes