- Email: [email protected]
Enrollment of Racial and Ethnic Minorities in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial
o
r
i
g
i
n
a
l
c
o
m
m
u
n
i
c
a
t
i
o
n
Enrollment of Racial and Ethnic Minorities in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial Paul F. Pinsky, PhD; Marvella Ford, PhD; Eduard Gamito; Darlene Higgins; Victoria Jenkins, MEd; Lois Lamerato, PhD; Sally Tenorio; Pamela M. Marcus PhD; and John K. Gohagan, PhD
Background: Minority populations in the United States, especially blacks and Hispanics, are generally underrepresented among participants in clinical trials. Here, we report the experience of enrolling ethnic minorities in a large cancer screening trial. Methods: The Prostate, Colorectal, Lung and Ovarian (PLCO) Cancer Screening Trial is a multicenter randomized trial designed to evaluate the effectiveness of screening for the PLCO cancers. Subjects were recruited at 10 U.S. centers between 1993 and 2001. One screening center had a major special recruitment effort for blacks and another center had a major special recruitment effort for Hispanics. Results: Among almost 155,000 subjects enrolled in PLCO, minority enrollment was as follows: black (5.0%), Hispanic (1.8%) and Asian (3.6%). This compares to an age-eligible population in the combined catchment areas of the PLCO centers that was 14.0% black, 2.9% Hispanic and 5.4% Asian, and an age-eligible population across the U.S. that was 9.5% black, 6.5% Hispanic and 3.0% Asian. About half (45%) of Hispanics were recruited at the center with the special Hispanic recruitment effort. Seventy percent of blacks were recruited at two centers; the one with the major special recruitment effort and a center in Detroit whose catchment area was 20% black among age-eligibles. Blacks, Hispanics and (nonHispanic) whites were all more highly educated, less likely to currently smoke and more likely to get regular exercise than their counterparts in the general population. Conclusion: Significant efforts were made to recruit racial/ ethnic minorities into PLCO, and these efforts resulted in enrollment levels that were comparable to those seen in many recent cancer screening or prevention trials. Blacks and Hispanics were nonetheless underrepresented in PLCO compared to their levels among age-eligibles in the overall U.S. population or in the aggregate PLCO catchment areas. Key words: African Americans n Latinos n cancer n clinical investigation n screening
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
© 2008. From the Division of Cancer Prevention, National Cancer Institute, Bethesda, MD (Pinsky, Marcus, Gohagan); Henry Ford Health System, Detroit, MI (Ford, Lamerato); University of Colorado Health Sciences Center, Denver, CO (Gamito, Tenorio); University of Alabama at Birmingham, Birmingham, AL (Higgins); and Pacific Health Research Institute, Honolulu, HI (Jenkins). Send correspondence and reprint requests for J Natl Med Assoc. 2008;100:291– 298 to: Dr. Paul Pinsky, 6130 Executive Blvd., EPN 3064, Bethesda, MD 20892; phone: (301) 402-6480; fax: (301) 402-0816; e-mail: [email protected]
Introduction
M
inority populations in the United States, especially blacks and Hispanics, have historically been underrepresented among participants in clinical trials.1,2 Adequate representation of these groups in clinical trials is important in order to insure equitable distribution of the risks and benefits of participation in research, and in order to enable the detection of possible interactions of treatment effect by race/ethnicity based on biological, social or cultural factors related to race.1-3 Such inclusion is particularly important when a minority group has an increased burden of the disease being studied. Recently, efforts have been made to increase minority representation in clinical research, especially phase-III trials. In 1990, the NIH issued policy guidelines on the inclusion of minorities (and women) in clinical research. These guidelines were updated in 1993 and 2001, with the current guidelines stating that minority groups and their subpopulations must be included in all NIH-supported biomedical research projects involving human subjects, unless a clear and compelling rationale is given showing that such inclusion would be inappropriate.4 The guidelines further support efforts at the study planning stage to ascertain whether there is evidence of significant differences in the effect of the intervention among racial and ethnic subgroups of the population and, if so, to design the study so that such differences can be validly assessed. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial is an ongoing, large, multicenter, randomized controlled trial designed to evaluate the mortality benefit of screening for these four cancers in VOL. 100, NO. 3, MARCH 2008 291
Enrollment of Racial and Ethnic Minorities
subjects aged 55–74 at enrollment. Mortality (and incidence) rates in the United States for the first three of these cancers are higher in black men than in white men; rates of lung and colorectal cancer are similar for black and white women, and ovarian cancer mortality rates are slightly lower in black than white women. Mortality rates for all of these cancers are lower in Hispanics than in non-Hispanic whites. In this manuscript, we describe the enrollment of racial/ethnic minorities into PLCO. We summarize overall recruitment methods and describe in-depth special minority recruitment efforts that were employed by some of the screening centers. We compare, on both a screening center and aggregate level, the racial/ethnic breakdown of PLCO participants versus that of the ageeligible group in the general population. In addition, for whites, blacks and Hispanics, we compare the behavioral and medical history characteristics of PLCO subjects to those of an aged-matched representative U.S. survey population. Finally, we discuss some of the possible barriers to minority enrollment in the PLCO.
Methods
The PLCO trial was funded from a contract from the National Cancer Institute (NCI). A request for proposals (RFP) was issued by the NCI describing the study requirements and evaluation criteria. An ability to recruit ethnic minorities was not included as an explicit evaluation criterion; however, the RFP specified that each center’s PLCO population should be representative of the eligible population in the surrounding area. The following 10 clinical centers were originally contracted to perform screening: Georgetown University, Washington, DC; University of Pittsburgh, Pittsburgh, PA; University of Minnesota, Minneapolis, MN; Henry Ford Health System (HFHS), Detroit, MI; University of Colorado, Denver, CO; University of Utah, Salt Lake City, UT; Marshfield Clinic Research Foundation, Marshfield, WI; Washington University, St. Louis, MO, Pacific Health Center, Honolulu, HI; and Maimonides Medical Center, Brooklyn, NY.5 Problems with the Brooklyn, NY, site led to termination of that center’s contract in 1996; all subjects recruited there ended participation in PLCO at that time. Since the Brooklyn site was considered a minority recruitment site, an RFP was issued in 1997 specifically for a center that could accrue minorities. The University of Alabama at Birmingham was awarded a contract in 1997. The original sample size for PLCO was 150,000. Recruitment for PLCO began in November 1993 and was originally intended to occur over a five-year span. For several reasons, including the need to provide UAB time to enroll sufficient numbers of minorities, the recruitment period was extended to 2001, and a total of almost 155,000 subjects were enrolled. The study protocol for the PLCO trial has been 292 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
described elsewhere.5 Briefly, men and women in the screened arm of PLCO received flexible sigmoidoscopy at year 0 and year 5 for colorectal cancer screening and annual chest x-ray for four years for lung cancer screening. Men in the screened arm also received prostate-specific antigen (PSA) testing annually for six years and digital rectal exam for four years, while women in the screened arm received CA-125 annually for six years and transvaginal ultrasound for four years. Control arm subjects received usual care. Subjects are to be followed for 13 years from randomization to ascertain trial endpoints. Exclusion criteria included a history of one of the PLCO cancers and current treatment for cancer. Beginning two years from the start of the trial, a history of >1 PSA within the last three years and a history of lower GI endoscopy within the last three years were also exclusion criteria. Subjects completed a baseline questionnaire at study entry which included questions on demographics, medical history and health behaviors. In the original version of the baseline questionnaire, there was a single question on race and ethnicity. This question was worded, “Which of these best describes your race or ethnic background?” with possible responses of white nonHispanic, white Hispanic, black non-Hispanic, black Hispanic, Asian, Pacific Islander and American Indian/ Alaska native. About midway through the recruitment period, the race/ethnicity question was changed to separate out the questions of race and Hispanic background. The first question asked, “Which of these groups best describes you?” with possible responses of white, black, Asian, Pacific Islander and American Indian/Alaskan native, and the second asked, “Are you of Hispanic origin?” with responses of no or yes. For the purposes of this analysis, we defined the following racial/ethnic subgroups: white non-Hispanic, black non-Hispanic, Hispanic, Asian, Pacific Islander and American Indian/ Alaska native. Subjects classifying themselves as Asian, Pacific Islander or American Indian/Alaska native were not classified as Hispanic, regardless of how they answered the question on Hispanic origin. Recruitment was accomplished primarily through mass mailings.5 Some centers utilized other methods, such as obtaining specialty mailing lists and advertising in various local media. Four centers had special recruitment efforts for minorities. These efforts are summarized below.
Henry Ford Health System At HFHS, the AAMEN project, a randomized trial that compared different methods of recruitment for black men, was conducted from 1995–2001.6,7 Specifically, 12,400 black men residing in southeastern Michigan and northern Ohio were contacted and found eligible to participate in PLCO. These men were randomized into three increasingly intensive intervention arms (arm VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
A to arm C) and one control arm (arm D). The interventions consisted of different approaches designed to enhance recruitment into PLCO. Arm A consisted of an enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers and a mailed baseline questionnaire/ consent form packet. Arm B consisted of the enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers, during which the baseline questionnaire was administered, followed by a mailed consent form packet. Arm C included the enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers, followed by a church-based project session, during which the consent form was read and signed and the baseline questionnaire was administered. Arm D consisted of standard PLCO recruitment procedures: participants received an introductory mailing, followed by a telephone eligibility interview conducted by trained African-American or Caucasian interviewers, followed by a mailing of the baseline questionnaire/consent form packet that was to be completed and returned by study participants. A second strategy used to increase minority representation was to locate a screening center in an area with a high black population. Shortly after the trial began, an additional screening center (Detroit Northwest Center) was opened within the Detroit city limits; according to 1990 census data, the population served by this location was 92% black.
University of Alabama at Birmingham As mentioned above, UAB was selected specifically as a minority recruitment center for blacks. UAB utilized a coordinated two-pronged approach to recruit blacks that consisted of targeted mass mailings and community outreach.8,9 Mass mailings were based on voter and motor vehicle registration lists, as well as lists of professional societies with black members; the mailing included targeted recruitment materials with endorsements from well-known local black figures. For community outreach, the center staff worked with community health advisors, ministers and others, and made presentations at senior centers, libraries, churches and other settings in predominantly black communities. Interested persons who completed cards with contact information were called and screened by telephone interviewers, with eligible persons then scheduled to attend information/consenting sessions in groups of 20–40 persons. At these sessions, participants were randomized immediately after signing the informed consent form; those assigned to the intervention arm were then scheduled for screening visits. These information/consenting sessions were conducted at the screening center and at locations in rural communities throughout central and northern Alabama. Persons needing transportation to the screenJOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
ing center were organized in groups of 5–8 and scheduled for van trips. The van was available for all screening visits throughout the trial. The original contract with UAB specified a goal of 5,000 subjects enrolled, ≥60% of which were black. This was later revised to 6,200 subjects enrolled, of which ≥1,700 (27.5%) were black.
University of Colorado Starting in 1998, about five years into the recruitment phase of the trial, this center received funding for special recruitment efforts for Hispanics. Study staff initially conducted a focus group with selected Hispanic community leaders, healthcare providers and cancer survivors. Based on the information from the focus group, informal meetings with Hispanic community members and a literature search, a recruitment plan was developed that emphasized personal contact and inclusion of the family in the recruitment process. Specific elements of the plan included building and maintaining a bilingual/bicultural staff, and ensuring that this staff would be available at every point of participant contact, from recruitment to enrollment, to trial participation. A special phone line for Spanish-speaking participants was set up, and most recruitment materials were made available in Spanish and well as English. Potential participants and their families were invited to recruitment seminars in their neighborhoods, where the trial was explained and the consent form discussed in detail, and at which participants had the option of enrolling. This center also used a one-on-one recruitment method at local clinics, community centers and churches serving the Hispanic community. At these locations, recruitment staff explained the trial to interested individuals and provided bilingual study materials, including a brochure, study application and consent form. Eligible individuals who signed the consent form were either randomized on site or at the screening center office. After randomization, an exit interview was conducted to determine any barriers that could have a negative impact on study compliance. For example, if a participant did not have access to a vehicle, free taxi or bus vouchers were offered to get them to their screening appointments. In addition to a full-time screening center recruitment staff, specially trained lay outreach workers were utilized to recruit participants within their local communities, neighborhoods and social groups (e.g., church congregation). The lay outreach workers were trained and tested to ensure that they had sufficient understanding of the trial and the ethical responsibilities of clinical trials recruitment; however, only trial staff obtained written informed consent and enrolled participants.
University of Pittsburgh This center convened focus groups and performed a telephone survey to help understand factors affecting black persons’ decisions on whether to enroll in clinVOL. 100, NO. 3, MARCH 2008 293
Enrollment of Racial and Ethnic Minorities
ical trials.12,13 Having a loved one with cancer, having knowledge or experience with clinical trials, and certain beliefs regarding the benefits and risks of clinical trial participation were associated with participation in PLCO. Partly on the basis of these findings, this center tried several approaches to boost black enrollment, including extra recruitment mailing to minority neighborhoods, presentations by PLCO recruiters at churches with large minority populations, and radio and TV interviews with the PLCO principal investigator on stations and shows targeted to blacks.
and 3.3%, respectively). In contrast, for blacks, representation was higher among women (5.5%) than among men (4.4%). Table 2 shows the racial/ethnic breakdown by screening center. Three centers had black enrollment exceeding 5%: UAB (31%), HFHS (14.3%) and Georgetown (6.2%). Only two centers had Hispanic enrollment exceeding 2%: Colorado (10.0%) and Georgetown (2.2%). For Asians, Hawaii had 43% Asian enrollment; no other center exceeded 2.5% enrollment. The majority of black subjects came from two centers, HFHS and UAB, with 46% and 25% of the total black subjects, respectively. About 25% of the black men enrolled at HFHS (and 11% of all blacks at HFHS) were recruited under the AAMEN project. Almost half (45%) of the Hispanics in PLCO were enrolled at Colorado, the center with the special Hispanic recruitment effort. About 75% of the Hispanics at Colorado were enrolled during the time period of the special recruitment effort; Hispanics comprised 22% of subjects enrolled at Colorado during this period. A majority of the Asians (83%) were enrolled at Hawaii. The Hawaii center further categorized their Asian subjects into country of origin; 62% were classified as Japanese, 12% as Chinese, 17% as “mixed” Asian and 2% as Korean. Trial wide, about 18% of Hispanics and 20% of Asians were foreign born, compared to 3% of whites and 1% of blacks. Table 2 also shows the proportion of minorities among age-eligible subjects in the primary catchment area of each screening center. For all centers except UAB, the proportion of blacks among enrolled subjects was lower than the proportion in the corresponding catchment area; at UAB, 31% of enrolled PLCO subjects were black compared to 22% in the catchment area. HFHS had 14% blacks in PLCO compared to 20% in the catchment area. Washington University and Georgetown had relatively high proportions of blacks in the catchment area (15.3% and 26.4%, respectively) but quite a bit lower proportions enrolled in PLCO (3.9% and 6.2%, respectively).
Statistical Methods We utilized data from the U.S. Census Bureau to estimate the minority population in the PLCO eligible age group (55–74) in the main catchment area of each screening center. In general, the main catchment area was defined as the counties making up the metropolitan statistical area (MSA) in which the screening center was located. Data from the 2000 Census were used.10 We used data from the National Health Interview Survey (NHIS) obtained during the period of PLCO enrollment to compare the demographics and medical histories of the PLCO population to that of the general U.S. population for each of the racial/ethnic groups (nonHispanic whites, non-Hispanic blacks, Hispanics).11 The NHIS population was age standardized to the age distribution of PLCO. Note there were too few Asians in the NHIS for meaningful results.
Results
Table 1 displays the demographic breakdown of the PLCO study. A total of 85.6% of subjects were (nonHispanic) white, 5.0% were (non-Hispanic) black, 1.8% were Hispanic and 3.6% Asian. A total of 3.2% of subjects did not fill out the baseline questionnaire, so their race/ethnicity is unknown. Hispanics and Asians had somewhat higher representation among the men (2.1% and 3.9%, respectively), than among the women (1.6%
Table 1. Race/ethnicity of subjects enrolled in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial
White non-Hispanic Black non-Hispanic Hispanic Asian Pacific Islander or American Indian Not stated No questionnaire1
Men (n=76,702)
Women (n=78,232)
All (n=154,934)
%
%
%
85.0 4.4 2.1 3.9
86.2 5.5 1.6 3.3
85.6 5.0 1.8 3.6
0.85
0.73
0.7
0.02 3.7
0.01 2.7
0.02 3.2
1: No baseline questionnaire filled out, so race/ethnicity is unknown
294 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
For Hispanics, Colorado had similar proportions enrolled in the study (10.0%) and in the catchment area (10.2%). A number of other centers with relatively small Hispanic populations had proportions in the study that equaled or exceeded the proportions in the catchment areas (e.g., Washington University, Pittsburgh, Marshfield, UAB and HFHS). For Asians, all centers had lower proportions enrolled in the study than in the catchment area. As seen in Table 2, the aggregate catchment area for all PLCO centers was 14.0% black, 2.9% Hispanic and 5.4% Asian in the eligible age group; this contrasts to percentages for the United States as a whole in this age group of 9.3%, 6.5% and 3.0% for blacks, Hispanics and Asians, respectively. Thus, the population around the PLCO centers was overrepresented for blacks and Asians and underrepresented for Hispanics. The enrollment percentages for PLCO of 5.0% blacks and 1.8% Hispanics show relative underenrollment of these groups as compared to the overall U.S. population; in contrast, the enrollment percentage for Asians, 3.6%, was slightly higher than the percentage in the age eligible U.S. population. Tables 3A and B show the demographics and medical history for the racial/ethnic groups in PLCO and in the NHIS. Among both men and women, whites, blacks and Hispanics enrolled in PLCO were more highly educated, less likely to smoke and more likely to exercise regularly than their counterparts in NHIS. For exam-
ple, the proportion of men with a college education was 42% for white men in PLCO versus 27% for white men in NHIS, 22% for black men in PLCO versus 12% in NHIS, and 28% for Hispanic men in PLCO compared to 10% in NHIS. In contrast, for each racial/ethnic group and among both men and women, PLCO and NHIS subjects had similar body mass index (BMI) and marital status. With respect to medical history, white men in PLCO tended to be slightly healthier than their counterparts in NHIS, while black and Hispanic men in PLCO were generally similar to their NHIS counterparts. For example, 34% of white men in PLCO compared to 42% in NHIS reported history of hypertension; for black and Hispanic men, 56% and 35% of PLCO subjects and 55% and 37% of NHIS subjects, respectively, reported hypertension. Among women, whites and Hispanics in PLCO were somewhat healthier than their counterparts in NHIS, while black women in PLCO were generally similar healthwise to black women in NHIS.
Discussion
In the PLCO Trial overall, blacks (5%) and Hispanics (1.8%) are underrepresented as compared to their percentages among the age-eligible U.S. population. For blacks, this underrepresentation was not primarily a result of the geographic locations of the PLCO centers, since in aggregate these centers had a higher proportion of blacks (among age-eligible persons) than the United
Table 2. Percent minorities among PLCO-enrolled versus PLCO-eligible population, by screening center Total in PLCO
Center U.S. total All centers Univ. of Colorado Georgetown University PHRI HFHS Univ. of Minnesota Washington University Univ. of Pittsburgh Univ. of Utah Marshfield UAB
Location
#
Black (Non-Hispanic)
Hispanic
Asian
PLCO
PLCO PLCO Eligible1 PLCO Eligible1
PLCO
PLCO Eligible1
%
%
%
%
%
%
– 154,934
– 5.0
9.3 14.0
– 1.8
6.5 2.9
– 3.6
3.0 5.4
Denver, CO
13,165
2.0
4.4
10.0
10.2
0.8
2.4
Washington, DC Honolulu, HI Detroit, MI
8,108 10,847 24,665
6.2 0.5 14.3
26.4 0.8 20.3
2.2 1.6 1.7
4.8 3.3 1.6
2.1 43.0 1.2
7.2 72.9 1.6
Minneapolis, MN
28,862
0.4
2.9
0.7
1.2
0.3
2.3
St. Louis, MO
15,042
3.9
15.3
0.8
0.8
0.4
1.1
Pittsburgh, PA2 Salt Lake City, UT3 Marshfield, WI Birmingham, AL
16,930 14,387 16,740 6,188
4.2 0.2 0.0 31.0
6.8 0.4 <0.5 22.1
0.7 1.5 0.5 0.7
0.5 4.1 <0.5 0.5
0.4 0.6 0.1 0.3
0.6 1.7 <0.5 0.4
1: PLCO-eligible population is the population aged 55–74 in the primary catchment area (generally the metropolitan area) of the center; 2: Satellite centers located in New Castle, PA, Latrobe, PA, and Steubenville, OH; 3: Satellite center located in Boise, ID
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008 295
Enrollment of Racial and Ethnic Minorities
States as a whole, 14.0% versus 9.3%. In contrast, for Hispanics, the PLCO centers in aggregate had a lower proportion than the United States as a whole—2.9% versus 6.5%; thus, the location of the centers was likely a contributing factor in the underrepresentation of Hispanics. The proportion of Asians in the aggregate PLCO catchment population—5.4%—was higher than for the United States as whole—3.0%—primarily because of the center in Hawaii, and Asians ended up being slightly overrepresented in PLCO (3.6%). Special recruitment efforts were needed to achieve the observed levels of minority enrollment for blacks and Hispanics. At the two centers with major special efforts for minority recruitment—UAB for blacks and Colorado for Hispanics—levels of minority representation in PLCO equaled or exceeded the levels in the age-eligible catchment population. For blacks, the two centers with modest special recruitment efforts, HFHS and Pittsburgh, achieved representation levels in PLCO that were about 60–70% of the levels in their catchment areas, while other centers that had large black populations (e.g., Georgetown and Washington University) and did not have special recruitment efforts achieved levels in PLCO that were only about 25% of the levels in their catchment areas. Levels of Hispanic representation in PLCO equal to that of the catchment area were observed at several centers that did not put forth any special efforts; however, these were all centers with very small percentages of Hispanics in their areas. Although proportional recruitment may be easier in these types of populations for various reasons, this will not translate into appreciable numbers of Hispanics recruited to clini-
cal trials due to the small overall numbers of subjects. Guiliano et al. have categorized the factors that serve to decrease minority participation in cancer clinical trials as being cultural, linguistic or structural.3 Cultural factors for blacks include distrust of the mainstream medical system, beliefs of fatalism or hopelessness, and a reliance on religion for healing and coping. Cultural factors, such as the use of traditional health providers, the importance of group over individual welfare and decision-making by consensus, are prevalent in many Asian-American subpopulations and may adversely impact clinical trial participation. Linguistic factors are important for Hispanics and some Asian subpopulations. Structural factors for blacks and Hispanics include poverty, lower educational levels, lack of transportation and lack of health insurance.
Efforts by the PLCO Trial to Address Cultural and Linguistic Factors The special efforts for black recruitment at UAB, HFHS and Pittsburgh focused largely on cultural factors, and especially on providing accurate information about clinical trials and overcoming a sense of mistrust about clinical trials. The issue of mistrust was addressed by promoting community outreach by trusted organizations and by having black staff and investigators at the centers available to interact one-on-one with prospective participants. These efforts helped allay any fears within the community about trial participation. Community outreach, the use of media outlets targeted to blacks, and increased personal contact with center investigators and staff served to help inform prospective participants
Table 3A. Demographics, behavior and medical history by race/ethnicity in PLCO and National Health Interview Survey, men
Current Smoker Former Smoker Never Smoker BMI >30 Regular Exercise Alcohol Use (Past Year) Education < High school College degree Married History of Diabetes History of Hypertension History of Coronary Heart Disease, MI or Stroke
White (Non-Hispanic)
Black (Non-Hispanic)
Hispanic
PLCO
NHIS1
PLCO
NHIS1
PLCO
NHIS1
%
%
%
%
%
%
11 52 37 24 85 77
18 52 30 23 57 61
25 47 28 30 73 64
30 38 32 24 39 47
15 53 32 22 82 78
19 38 43 23 40 56
7 42 84 9 34
20 27 82 13 42
23 22 64 21 56
43 12 63 20 55
21 28 76 17 35
55 10 78 20 37
16
20
17
18
13
14
1: NHIS is National Health Interview Survey; NHIS rates are age adjusted to overall PLCO population; MI: myocardial infarction
296 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
about the PLCO Trial. The increased proportional enrollment of blacks at these three centers showed that these approaches were successful in increasing the proportion of blacks enrolled in the trial. Cultural as well linguistic factors were also an important component of the special effort for Hispanic recruitment and retention at Colorado. Having a bilingual and bicultural staff and having that staff available at points of participant contact, having bilingual study materials available and including the family in the recruitment process were critical elements in this center’s successful effort to enroll significant numbers of Hispanics. In contrast, earlier efforts at directed mass mailings for blacks and Hispanics at several centers were not successful alone in increasing recruitment, even when the recruitment materials were created by minorities for minorities, because they failed to adequately address cultural concerns.
Efforts by the PLCO Trial to Address Structural Factors A number of structural factors that had the potential to impact minority recruitment in PLCO were addressed at some or all of the centers. Lack of insurance was a structural issue in PLCO because diagnostic follow-up was not provided by the trial and, thus, participants were responsible (financially and otherwise) for their own follow-up. All PLCO centers attempted to make provisions for the uninsured; however, this aspect of trial design may have been a structural barrier to recruitment of blacks and Hispanics. The location of the screening center and problems with transportation may also have been a structural
factor affecting minority recruitment at some centers. Washington University initially planned to locate their screening center within the city of St. Louis; however, before the start of the trial their proposal was amended with the screening center now located in Creve Coeur, a suburb in the middle of St. Louis County, about 10 miles west of the St. Louis city (western) boundary. Among age-eligibles, blacks made up almost 50% of the population in the city of St. Louis but only 13% of the population in St. Louis County; roughly equal numbers of blacks reside in the city and the county (note the city of St. Louis city is separate from St. Louis County). This may partially explain the low percentage enrollment of blacks. In contrast, at HFHS, an additional screening site located in a predominantly black area of Detroit was added shortly after the trial opened; 21% of screenees at this additional site were black, compared to 11% at the original screening site, which was located in suburban Dearborn. The special efforts at Colorado and UAB addressed transportation needs by providing rides to screening appointments. The Georgetown center attempted to address transportation issues by negotiating with a hospital in Washington, DC, that was located closer to black and Hispanic population centers to perform recruitment and screening for PLCO. However, for various reasons, this project did not succeed, and few subjects were ever recruited or screened at this facility. Education and lifestyle were other structural factors affecting recruitment. Blacks, Hispanics and (non-Hispanic) whites in PLCO were all more highly educated and had healthier lifestyles in terms of not smoking and
Table 3B. Demographics, behavior and medical history by race in PLCO and National Health Interview Survey, women
Current Smoker Former Smoker Never Smoker BMI >30 Regular Exercise Alcohol Use (Past Year) Education < High school College degree Married History of Diabetes History of Hypertension History of Coronary Heart Disease, MI or Stroke
White (Non-Hispanic)
Black (Non-Hispanic)
Hispanic
PLCO
NHIS
PLCO
NHIS
PLCO
NHIS1
% 9 35 56 24 85 70
% 18 30 52 21 54 49
% 16 34 50 45 72 50
% 18 24 58 41 34 24
% 10 35 55 29 82 67
% 11 20 69 27 35 30
6 30 71 6 33
19 16 65 11 43
17 26 40 20 63
41 11 36 26 66
19 25 59 12 37
59 7 53 22 45
7
11
14
13
8
10
1
1
1: NHIS is National Health Interview Survey; NHIS rates are age adjusted to overall PLCO population; MI: myocardial infarction
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008 297
Enrollment of Racial and Ethnic Minorities
exercise, than their counterparts in the general population. Thus, the factors that correlate with clinical trial participation appear to be largely the same in blacks and Hispanics as in whites. However, since educational attainment is lower on average among blacks and Hispanics as compared to whites, this was likely a factor in the underrepresentation of these groups in PLCO. Special recruitment efforts for minority populations involve additional funding. In PLCO, it was not always possible to directly ascertain the amount of funds dedicated to special minority recruitment and retention efforts at a given center. Therefore, as a crude method of gauging the costs of these efforts, we compared the per-capita cost of the trial for centers with and without special recruitment efforts, where the per-capita cost was calculated by dividing the center’s total PLCO budget (for all study years through year 13) by the number of enrolled subjects. For the six centers not involved in special recruitment efforts, the aggregate per-capita cost was $1,285. Per-capita costs at the two centers with relatively small-scale minority efforts—HFHS and Pittsburgh—were similar. However, per-capita costs were considerably higher at the two centers with extensive minority efforts—Colorado, with a per-capita cost of $1,622; and UAB, with a cost of $2,646. The higher per-capita costs at these two centers generally reflect the intensity and duration of each of the centers’ special minority recruitment and retention efforts. The PLCO trial, with 5.0% black and 1.8% Hispanic enrollment, generally equaled or exceeded the percentages of blacks and Hispanics enrolled in other recent multicenter cancer screening or prevention trials. The Prostate Cancer Prevention Trial, a randomized trial of finasteride for the prevention of prostate cancer carried out at 221 centers across the United States, had black and Hispanic percentages among enrollees of 3.8% and 2.6%, respectively.14 The STAR trial comparing tamoxifen and raloxifene for breast cancer prevention was 2.5% black and 2.0% Hispanic.15 The CARET trial for lung cancer prevention was 2.9% black and 1.5% Hispanic.16 One trial whose minority participation did substantially exceed that of PLCO was the SELECT (Selenium and Vitamin E) Prostate Cancer Prevention trial, which had a goal of 24% minorities and whose final enrollment included 15% blacks, 5% Hispanics and 1% Asians.17 The success of SELECT in achieving high minority representation resulted from a comprehensive approach to minority recruitment, which began during the planning phase of the study.17 In conclusion, the experience with PLCO suggests that overcoming cultural barriers involving lack of information and mistrust of clinical trials is critical in achieving proportional levels of minority enrollment, especially among blacks. Culturally sensitive approaches to recruitment, as well as bilingual staff and materials,
298 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
were critical in successful efforts to increase Hispanic enrollment. Efforts to overcome the structural barriers of lack of transportation, lack of health insurance, and work or family conflicts were also important. These efforts overall—and especially at the PLCO centers with special minority recruitment programs—resulted in levels of black and Hispanic enrollment that were on par with or higher than those seen in several recent largescale cancer screening or prevention trials. Despite these efforts, however, the proportions of blacks and Hispanics in PLCO were still considerably lower than their respective proportions among PLCO age-eligibles in the overall U.S. population or in the aggregate PLCO catchment areas.
References
1. Ford JG, Howerton M. The science of recruiting minority populations to screening trials. Clin Trials. 2004;1:341-342. 2. Lai GY, Gary TL, Tilburt J, et al. Effectiveness of strategies to recruit underrepresented populations into cancer clinical trials. Clin Trials. 2006;3:133-141. 3. Guiliano AR, Mokuau N, Hughes C, et al. Participation of minorities in cancer research: the influence of structural, cultural, and linguistic factors. Ann Epidemiol. 2000;10:S22-S34. 4. National Institutes of Health. NIH Guidelines on the Inclusion of Women and Minorities in Clinical Research. http://grants.nih.gov/grants/guide/ notice-files/NOT-OD-02-001.html. Accessed 10/20/06. 5. Prorok PC, Andriole GL, Bresalier RS, et al. Design of the Prostate, Lung, Colon and Ovarian (PLCO) cancer screening trial. Control Clin Trials. 2000;21:273S-309S. 6. Ford ME, Havstad SL, Davis SD. A randomized trial of recruitment methods for older African American men in the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial. Clin Trials. 2004;1:343-351. 7. Ford ME, Havstad SL, Tilley BC. Recruiting older African American men to a cancer screening trial (the AAMEN Project). Gerontologist. 2003;43:27-35. 8. Simpson NK, Johnson CJ, Ogden SL, et al. Recruitment strategies in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials. 2000;21:365S-378S. 9. Stallings FL, Ford ME, Simpson NK, et al. Black participation in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials. 2000;21:379S-389S. 10. U.S. Department of the Census. Census 2000. www.census.gov/main/ www/cen200.html. Accessed 02/12/07. 11. National Health Interview Survey 2005. www.cdc.gov/nchs/about/ major/nhis/quest_data_related_1997_forward.htm. Accessed 02/12/07. 12. Trauth JM, Jernigan JC, Siminoff LA, et al. Factors affecting older African American women’s decisions to join the PLCO Cancer Screening Trial. J Clin Oncology. 2005;23:8730-8738. 13. Jernigan JC, Trauth JM, Neal-Ferguson D, et al. Factors that influence cancer screening in older African American men and women: focus group findings. Fam Community Health. 2001;24:27-33. 14. Moinpour CM, Atkinson JO, Thomas SM, et al. Minority recruitment in the Prostate Cancer Prevention Trial. Ann Epidemiol. 2000;10:S85-S91. 15. Vogel VG, Constantino JP, Wickerham DL, et al. Effects of tamoxifen vs raloxifene on the risk of developing invasive breast cancer and other disease outcomes: the NSABP Study of Tamoxifen and Raloxifene (STAR) P-2 trial. JAMA. 2006;295:2727-2741. 16. Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med. 1996;334:1150-1155. 17. Cook ED, Moody-Thomas S, Anderson KB, et al. Minority recruitment to the Selenium and Vitamin E Cancer Prevention Trial (SELECT). Clin Trials. 2005;2:436-442. n
VOL. 100, NO. 3, MARCH 2008
r
i
g
i
n
a
l
c
o
m
m
u
n
i
c
a
t
i
o
n
Enrollment of Racial and Ethnic Minorities in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial Paul F. Pinsky, PhD; Marvella Ford, PhD; Eduard Gamito; Darlene Higgins; Victoria Jenkins, MEd; Lois Lamerato, PhD; Sally Tenorio; Pamela M. Marcus PhD; and John K. Gohagan, PhD
Background: Minority populations in the United States, especially blacks and Hispanics, are generally underrepresented among participants in clinical trials. Here, we report the experience of enrolling ethnic minorities in a large cancer screening trial. Methods: The Prostate, Colorectal, Lung and Ovarian (PLCO) Cancer Screening Trial is a multicenter randomized trial designed to evaluate the effectiveness of screening for the PLCO cancers. Subjects were recruited at 10 U.S. centers between 1993 and 2001. One screening center had a major special recruitment effort for blacks and another center had a major special recruitment effort for Hispanics. Results: Among almost 155,000 subjects enrolled in PLCO, minority enrollment was as follows: black (5.0%), Hispanic (1.8%) and Asian (3.6%). This compares to an age-eligible population in the combined catchment areas of the PLCO centers that was 14.0% black, 2.9% Hispanic and 5.4% Asian, and an age-eligible population across the U.S. that was 9.5% black, 6.5% Hispanic and 3.0% Asian. About half (45%) of Hispanics were recruited at the center with the special Hispanic recruitment effort. Seventy percent of blacks were recruited at two centers; the one with the major special recruitment effort and a center in Detroit whose catchment area was 20% black among age-eligibles. Blacks, Hispanics and (nonHispanic) whites were all more highly educated, less likely to currently smoke and more likely to get regular exercise than their counterparts in the general population. Conclusion: Significant efforts were made to recruit racial/ ethnic minorities into PLCO, and these efforts resulted in enrollment levels that were comparable to those seen in many recent cancer screening or prevention trials. Blacks and Hispanics were nonetheless underrepresented in PLCO compared to their levels among age-eligibles in the overall U.S. population or in the aggregate PLCO catchment areas. Key words: African Americans n Latinos n cancer n clinical investigation n screening
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
© 2008. From the Division of Cancer Prevention, National Cancer Institute, Bethesda, MD (Pinsky, Marcus, Gohagan); Henry Ford Health System, Detroit, MI (Ford, Lamerato); University of Colorado Health Sciences Center, Denver, CO (Gamito, Tenorio); University of Alabama at Birmingham, Birmingham, AL (Higgins); and Pacific Health Research Institute, Honolulu, HI (Jenkins). Send correspondence and reprint requests for J Natl Med Assoc. 2008;100:291– 298 to: Dr. Paul Pinsky, 6130 Executive Blvd., EPN 3064, Bethesda, MD 20892; phone: (301) 402-6480; fax: (301) 402-0816; e-mail: [email protected]
Introduction
M
inority populations in the United States, especially blacks and Hispanics, have historically been underrepresented among participants in clinical trials.1,2 Adequate representation of these groups in clinical trials is important in order to insure equitable distribution of the risks and benefits of participation in research, and in order to enable the detection of possible interactions of treatment effect by race/ethnicity based on biological, social or cultural factors related to race.1-3 Such inclusion is particularly important when a minority group has an increased burden of the disease being studied. Recently, efforts have been made to increase minority representation in clinical research, especially phase-III trials. In 1990, the NIH issued policy guidelines on the inclusion of minorities (and women) in clinical research. These guidelines were updated in 1993 and 2001, with the current guidelines stating that minority groups and their subpopulations must be included in all NIH-supported biomedical research projects involving human subjects, unless a clear and compelling rationale is given showing that such inclusion would be inappropriate.4 The guidelines further support efforts at the study planning stage to ascertain whether there is evidence of significant differences in the effect of the intervention among racial and ethnic subgroups of the population and, if so, to design the study so that such differences can be validly assessed. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial is an ongoing, large, multicenter, randomized controlled trial designed to evaluate the mortality benefit of screening for these four cancers in VOL. 100, NO. 3, MARCH 2008 291
Enrollment of Racial and Ethnic Minorities
subjects aged 55–74 at enrollment. Mortality (and incidence) rates in the United States for the first three of these cancers are higher in black men than in white men; rates of lung and colorectal cancer are similar for black and white women, and ovarian cancer mortality rates are slightly lower in black than white women. Mortality rates for all of these cancers are lower in Hispanics than in non-Hispanic whites. In this manuscript, we describe the enrollment of racial/ethnic minorities into PLCO. We summarize overall recruitment methods and describe in-depth special minority recruitment efforts that were employed by some of the screening centers. We compare, on both a screening center and aggregate level, the racial/ethnic breakdown of PLCO participants versus that of the ageeligible group in the general population. In addition, for whites, blacks and Hispanics, we compare the behavioral and medical history characteristics of PLCO subjects to those of an aged-matched representative U.S. survey population. Finally, we discuss some of the possible barriers to minority enrollment in the PLCO.
Methods
The PLCO trial was funded from a contract from the National Cancer Institute (NCI). A request for proposals (RFP) was issued by the NCI describing the study requirements and evaluation criteria. An ability to recruit ethnic minorities was not included as an explicit evaluation criterion; however, the RFP specified that each center’s PLCO population should be representative of the eligible population in the surrounding area. The following 10 clinical centers were originally contracted to perform screening: Georgetown University, Washington, DC; University of Pittsburgh, Pittsburgh, PA; University of Minnesota, Minneapolis, MN; Henry Ford Health System (HFHS), Detroit, MI; University of Colorado, Denver, CO; University of Utah, Salt Lake City, UT; Marshfield Clinic Research Foundation, Marshfield, WI; Washington University, St. Louis, MO, Pacific Health Center, Honolulu, HI; and Maimonides Medical Center, Brooklyn, NY.5 Problems with the Brooklyn, NY, site led to termination of that center’s contract in 1996; all subjects recruited there ended participation in PLCO at that time. Since the Brooklyn site was considered a minority recruitment site, an RFP was issued in 1997 specifically for a center that could accrue minorities. The University of Alabama at Birmingham was awarded a contract in 1997. The original sample size for PLCO was 150,000. Recruitment for PLCO began in November 1993 and was originally intended to occur over a five-year span. For several reasons, including the need to provide UAB time to enroll sufficient numbers of minorities, the recruitment period was extended to 2001, and a total of almost 155,000 subjects were enrolled. The study protocol for the PLCO trial has been 292 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
described elsewhere.5 Briefly, men and women in the screened arm of PLCO received flexible sigmoidoscopy at year 0 and year 5 for colorectal cancer screening and annual chest x-ray for four years for lung cancer screening. Men in the screened arm also received prostate-specific antigen (PSA) testing annually for six years and digital rectal exam for four years, while women in the screened arm received CA-125 annually for six years and transvaginal ultrasound for four years. Control arm subjects received usual care. Subjects are to be followed for 13 years from randomization to ascertain trial endpoints. Exclusion criteria included a history of one of the PLCO cancers and current treatment for cancer. Beginning two years from the start of the trial, a history of >1 PSA within the last three years and a history of lower GI endoscopy within the last three years were also exclusion criteria. Subjects completed a baseline questionnaire at study entry which included questions on demographics, medical history and health behaviors. In the original version of the baseline questionnaire, there was a single question on race and ethnicity. This question was worded, “Which of these best describes your race or ethnic background?” with possible responses of white nonHispanic, white Hispanic, black non-Hispanic, black Hispanic, Asian, Pacific Islander and American Indian/ Alaska native. About midway through the recruitment period, the race/ethnicity question was changed to separate out the questions of race and Hispanic background. The first question asked, “Which of these groups best describes you?” with possible responses of white, black, Asian, Pacific Islander and American Indian/Alaskan native, and the second asked, “Are you of Hispanic origin?” with responses of no or yes. For the purposes of this analysis, we defined the following racial/ethnic subgroups: white non-Hispanic, black non-Hispanic, Hispanic, Asian, Pacific Islander and American Indian/ Alaska native. Subjects classifying themselves as Asian, Pacific Islander or American Indian/Alaska native were not classified as Hispanic, regardless of how they answered the question on Hispanic origin. Recruitment was accomplished primarily through mass mailings.5 Some centers utilized other methods, such as obtaining specialty mailing lists and advertising in various local media. Four centers had special recruitment efforts for minorities. These efforts are summarized below.
Henry Ford Health System At HFHS, the AAMEN project, a randomized trial that compared different methods of recruitment for black men, was conducted from 1995–2001.6,7 Specifically, 12,400 black men residing in southeastern Michigan and northern Ohio were contacted and found eligible to participate in PLCO. These men were randomized into three increasingly intensive intervention arms (arm VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
A to arm C) and one control arm (arm D). The interventions consisted of different approaches designed to enhance recruitment into PLCO. Arm A consisted of an enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers and a mailed baseline questionnaire/ consent form packet. Arm B consisted of the enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers, during which the baseline questionnaire was administered, followed by a mailed consent form packet. Arm C included the enhanced recruitment letter followed by a telephone eligibility interview conducted by trained African-American interviewers, followed by a church-based project session, during which the consent form was read and signed and the baseline questionnaire was administered. Arm D consisted of standard PLCO recruitment procedures: participants received an introductory mailing, followed by a telephone eligibility interview conducted by trained African-American or Caucasian interviewers, followed by a mailing of the baseline questionnaire/consent form packet that was to be completed and returned by study participants. A second strategy used to increase minority representation was to locate a screening center in an area with a high black population. Shortly after the trial began, an additional screening center (Detroit Northwest Center) was opened within the Detroit city limits; according to 1990 census data, the population served by this location was 92% black.
University of Alabama at Birmingham As mentioned above, UAB was selected specifically as a minority recruitment center for blacks. UAB utilized a coordinated two-pronged approach to recruit blacks that consisted of targeted mass mailings and community outreach.8,9 Mass mailings were based on voter and motor vehicle registration lists, as well as lists of professional societies with black members; the mailing included targeted recruitment materials with endorsements from well-known local black figures. For community outreach, the center staff worked with community health advisors, ministers and others, and made presentations at senior centers, libraries, churches and other settings in predominantly black communities. Interested persons who completed cards with contact information were called and screened by telephone interviewers, with eligible persons then scheduled to attend information/consenting sessions in groups of 20–40 persons. At these sessions, participants were randomized immediately after signing the informed consent form; those assigned to the intervention arm were then scheduled for screening visits. These information/consenting sessions were conducted at the screening center and at locations in rural communities throughout central and northern Alabama. Persons needing transportation to the screenJOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
ing center were organized in groups of 5–8 and scheduled for van trips. The van was available for all screening visits throughout the trial. The original contract with UAB specified a goal of 5,000 subjects enrolled, ≥60% of which were black. This was later revised to 6,200 subjects enrolled, of which ≥1,700 (27.5%) were black.
University of Colorado Starting in 1998, about five years into the recruitment phase of the trial, this center received funding for special recruitment efforts for Hispanics. Study staff initially conducted a focus group with selected Hispanic community leaders, healthcare providers and cancer survivors. Based on the information from the focus group, informal meetings with Hispanic community members and a literature search, a recruitment plan was developed that emphasized personal contact and inclusion of the family in the recruitment process. Specific elements of the plan included building and maintaining a bilingual/bicultural staff, and ensuring that this staff would be available at every point of participant contact, from recruitment to enrollment, to trial participation. A special phone line for Spanish-speaking participants was set up, and most recruitment materials were made available in Spanish and well as English. Potential participants and their families were invited to recruitment seminars in their neighborhoods, where the trial was explained and the consent form discussed in detail, and at which participants had the option of enrolling. This center also used a one-on-one recruitment method at local clinics, community centers and churches serving the Hispanic community. At these locations, recruitment staff explained the trial to interested individuals and provided bilingual study materials, including a brochure, study application and consent form. Eligible individuals who signed the consent form were either randomized on site or at the screening center office. After randomization, an exit interview was conducted to determine any barriers that could have a negative impact on study compliance. For example, if a participant did not have access to a vehicle, free taxi or bus vouchers were offered to get them to their screening appointments. In addition to a full-time screening center recruitment staff, specially trained lay outreach workers were utilized to recruit participants within their local communities, neighborhoods and social groups (e.g., church congregation). The lay outreach workers were trained and tested to ensure that they had sufficient understanding of the trial and the ethical responsibilities of clinical trials recruitment; however, only trial staff obtained written informed consent and enrolled participants.
University of Pittsburgh This center convened focus groups and performed a telephone survey to help understand factors affecting black persons’ decisions on whether to enroll in clinVOL. 100, NO. 3, MARCH 2008 293
Enrollment of Racial and Ethnic Minorities
ical trials.12,13 Having a loved one with cancer, having knowledge or experience with clinical trials, and certain beliefs regarding the benefits and risks of clinical trial participation were associated with participation in PLCO. Partly on the basis of these findings, this center tried several approaches to boost black enrollment, including extra recruitment mailing to minority neighborhoods, presentations by PLCO recruiters at churches with large minority populations, and radio and TV interviews with the PLCO principal investigator on stations and shows targeted to blacks.
and 3.3%, respectively). In contrast, for blacks, representation was higher among women (5.5%) than among men (4.4%). Table 2 shows the racial/ethnic breakdown by screening center. Three centers had black enrollment exceeding 5%: UAB (31%), HFHS (14.3%) and Georgetown (6.2%). Only two centers had Hispanic enrollment exceeding 2%: Colorado (10.0%) and Georgetown (2.2%). For Asians, Hawaii had 43% Asian enrollment; no other center exceeded 2.5% enrollment. The majority of black subjects came from two centers, HFHS and UAB, with 46% and 25% of the total black subjects, respectively. About 25% of the black men enrolled at HFHS (and 11% of all blacks at HFHS) were recruited under the AAMEN project. Almost half (45%) of the Hispanics in PLCO were enrolled at Colorado, the center with the special Hispanic recruitment effort. About 75% of the Hispanics at Colorado were enrolled during the time period of the special recruitment effort; Hispanics comprised 22% of subjects enrolled at Colorado during this period. A majority of the Asians (83%) were enrolled at Hawaii. The Hawaii center further categorized their Asian subjects into country of origin; 62% were classified as Japanese, 12% as Chinese, 17% as “mixed” Asian and 2% as Korean. Trial wide, about 18% of Hispanics and 20% of Asians were foreign born, compared to 3% of whites and 1% of blacks. Table 2 also shows the proportion of minorities among age-eligible subjects in the primary catchment area of each screening center. For all centers except UAB, the proportion of blacks among enrolled subjects was lower than the proportion in the corresponding catchment area; at UAB, 31% of enrolled PLCO subjects were black compared to 22% in the catchment area. HFHS had 14% blacks in PLCO compared to 20% in the catchment area. Washington University and Georgetown had relatively high proportions of blacks in the catchment area (15.3% and 26.4%, respectively) but quite a bit lower proportions enrolled in PLCO (3.9% and 6.2%, respectively).
Statistical Methods We utilized data from the U.S. Census Bureau to estimate the minority population in the PLCO eligible age group (55–74) in the main catchment area of each screening center. In general, the main catchment area was defined as the counties making up the metropolitan statistical area (MSA) in which the screening center was located. Data from the 2000 Census were used.10 We used data from the National Health Interview Survey (NHIS) obtained during the period of PLCO enrollment to compare the demographics and medical histories of the PLCO population to that of the general U.S. population for each of the racial/ethnic groups (nonHispanic whites, non-Hispanic blacks, Hispanics).11 The NHIS population was age standardized to the age distribution of PLCO. Note there were too few Asians in the NHIS for meaningful results.
Results
Table 1 displays the demographic breakdown of the PLCO study. A total of 85.6% of subjects were (nonHispanic) white, 5.0% were (non-Hispanic) black, 1.8% were Hispanic and 3.6% Asian. A total of 3.2% of subjects did not fill out the baseline questionnaire, so their race/ethnicity is unknown. Hispanics and Asians had somewhat higher representation among the men (2.1% and 3.9%, respectively), than among the women (1.6%
Table 1. Race/ethnicity of subjects enrolled in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial
White non-Hispanic Black non-Hispanic Hispanic Asian Pacific Islander or American Indian Not stated No questionnaire1
Men (n=76,702)
Women (n=78,232)
All (n=154,934)
%
%
%
85.0 4.4 2.1 3.9
86.2 5.5 1.6 3.3
85.6 5.0 1.8 3.6
0.85
0.73
0.7
0.02 3.7
0.01 2.7
0.02 3.2
1: No baseline questionnaire filled out, so race/ethnicity is unknown
294 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
For Hispanics, Colorado had similar proportions enrolled in the study (10.0%) and in the catchment area (10.2%). A number of other centers with relatively small Hispanic populations had proportions in the study that equaled or exceeded the proportions in the catchment areas (e.g., Washington University, Pittsburgh, Marshfield, UAB and HFHS). For Asians, all centers had lower proportions enrolled in the study than in the catchment area. As seen in Table 2, the aggregate catchment area for all PLCO centers was 14.0% black, 2.9% Hispanic and 5.4% Asian in the eligible age group; this contrasts to percentages for the United States as a whole in this age group of 9.3%, 6.5% and 3.0% for blacks, Hispanics and Asians, respectively. Thus, the population around the PLCO centers was overrepresented for blacks and Asians and underrepresented for Hispanics. The enrollment percentages for PLCO of 5.0% blacks and 1.8% Hispanics show relative underenrollment of these groups as compared to the overall U.S. population; in contrast, the enrollment percentage for Asians, 3.6%, was slightly higher than the percentage in the age eligible U.S. population. Tables 3A and B show the demographics and medical history for the racial/ethnic groups in PLCO and in the NHIS. Among both men and women, whites, blacks and Hispanics enrolled in PLCO were more highly educated, less likely to smoke and more likely to exercise regularly than their counterparts in NHIS. For exam-
ple, the proportion of men with a college education was 42% for white men in PLCO versus 27% for white men in NHIS, 22% for black men in PLCO versus 12% in NHIS, and 28% for Hispanic men in PLCO compared to 10% in NHIS. In contrast, for each racial/ethnic group and among both men and women, PLCO and NHIS subjects had similar body mass index (BMI) and marital status. With respect to medical history, white men in PLCO tended to be slightly healthier than their counterparts in NHIS, while black and Hispanic men in PLCO were generally similar to their NHIS counterparts. For example, 34% of white men in PLCO compared to 42% in NHIS reported history of hypertension; for black and Hispanic men, 56% and 35% of PLCO subjects and 55% and 37% of NHIS subjects, respectively, reported hypertension. Among women, whites and Hispanics in PLCO were somewhat healthier than their counterparts in NHIS, while black women in PLCO were generally similar healthwise to black women in NHIS.
Discussion
In the PLCO Trial overall, blacks (5%) and Hispanics (1.8%) are underrepresented as compared to their percentages among the age-eligible U.S. population. For blacks, this underrepresentation was not primarily a result of the geographic locations of the PLCO centers, since in aggregate these centers had a higher proportion of blacks (among age-eligible persons) than the United
Table 2. Percent minorities among PLCO-enrolled versus PLCO-eligible population, by screening center Total in PLCO
Center U.S. total All centers Univ. of Colorado Georgetown University PHRI HFHS Univ. of Minnesota Washington University Univ. of Pittsburgh Univ. of Utah Marshfield UAB
Location
#
Black (Non-Hispanic)
Hispanic
Asian
PLCO
PLCO PLCO Eligible1 PLCO Eligible1
PLCO
PLCO Eligible1
%
%
%
%
%
%
– 154,934
– 5.0
9.3 14.0
– 1.8
6.5 2.9
– 3.6
3.0 5.4
Denver, CO
13,165
2.0
4.4
10.0
10.2
0.8
2.4
Washington, DC Honolulu, HI Detroit, MI
8,108 10,847 24,665
6.2 0.5 14.3
26.4 0.8 20.3
2.2 1.6 1.7
4.8 3.3 1.6
2.1 43.0 1.2
7.2 72.9 1.6
Minneapolis, MN
28,862
0.4
2.9
0.7
1.2
0.3
2.3
St. Louis, MO
15,042
3.9
15.3
0.8
0.8
0.4
1.1
Pittsburgh, PA2 Salt Lake City, UT3 Marshfield, WI Birmingham, AL
16,930 14,387 16,740 6,188
4.2 0.2 0.0 31.0
6.8 0.4 <0.5 22.1
0.7 1.5 0.5 0.7
0.5 4.1 <0.5 0.5
0.4 0.6 0.1 0.3
0.6 1.7 <0.5 0.4
1: PLCO-eligible population is the population aged 55–74 in the primary catchment area (generally the metropolitan area) of the center; 2: Satellite centers located in New Castle, PA, Latrobe, PA, and Steubenville, OH; 3: Satellite center located in Boise, ID
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008 295
Enrollment of Racial and Ethnic Minorities
States as a whole, 14.0% versus 9.3%. In contrast, for Hispanics, the PLCO centers in aggregate had a lower proportion than the United States as a whole—2.9% versus 6.5%; thus, the location of the centers was likely a contributing factor in the underrepresentation of Hispanics. The proportion of Asians in the aggregate PLCO catchment population—5.4%—was higher than for the United States as whole—3.0%—primarily because of the center in Hawaii, and Asians ended up being slightly overrepresented in PLCO (3.6%). Special recruitment efforts were needed to achieve the observed levels of minority enrollment for blacks and Hispanics. At the two centers with major special efforts for minority recruitment—UAB for blacks and Colorado for Hispanics—levels of minority representation in PLCO equaled or exceeded the levels in the age-eligible catchment population. For blacks, the two centers with modest special recruitment efforts, HFHS and Pittsburgh, achieved representation levels in PLCO that were about 60–70% of the levels in their catchment areas, while other centers that had large black populations (e.g., Georgetown and Washington University) and did not have special recruitment efforts achieved levels in PLCO that were only about 25% of the levels in their catchment areas. Levels of Hispanic representation in PLCO equal to that of the catchment area were observed at several centers that did not put forth any special efforts; however, these were all centers with very small percentages of Hispanics in their areas. Although proportional recruitment may be easier in these types of populations for various reasons, this will not translate into appreciable numbers of Hispanics recruited to clini-
cal trials due to the small overall numbers of subjects. Guiliano et al. have categorized the factors that serve to decrease minority participation in cancer clinical trials as being cultural, linguistic or structural.3 Cultural factors for blacks include distrust of the mainstream medical system, beliefs of fatalism or hopelessness, and a reliance on religion for healing and coping. Cultural factors, such as the use of traditional health providers, the importance of group over individual welfare and decision-making by consensus, are prevalent in many Asian-American subpopulations and may adversely impact clinical trial participation. Linguistic factors are important for Hispanics and some Asian subpopulations. Structural factors for blacks and Hispanics include poverty, lower educational levels, lack of transportation and lack of health insurance.
Efforts by the PLCO Trial to Address Cultural and Linguistic Factors The special efforts for black recruitment at UAB, HFHS and Pittsburgh focused largely on cultural factors, and especially on providing accurate information about clinical trials and overcoming a sense of mistrust about clinical trials. The issue of mistrust was addressed by promoting community outreach by trusted organizations and by having black staff and investigators at the centers available to interact one-on-one with prospective participants. These efforts helped allay any fears within the community about trial participation. Community outreach, the use of media outlets targeted to blacks, and increased personal contact with center investigators and staff served to help inform prospective participants
Table 3A. Demographics, behavior and medical history by race/ethnicity in PLCO and National Health Interview Survey, men
Current Smoker Former Smoker Never Smoker BMI >30 Regular Exercise Alcohol Use (Past Year) Education < High school College degree Married History of Diabetes History of Hypertension History of Coronary Heart Disease, MI or Stroke
White (Non-Hispanic)
Black (Non-Hispanic)
Hispanic
PLCO
NHIS1
PLCO
NHIS1
PLCO
NHIS1
%
%
%
%
%
%
11 52 37 24 85 77
18 52 30 23 57 61
25 47 28 30 73 64
30 38 32 24 39 47
15 53 32 22 82 78
19 38 43 23 40 56
7 42 84 9 34
20 27 82 13 42
23 22 64 21 56
43 12 63 20 55
21 28 76 17 35
55 10 78 20 37
16
20
17
18
13
14
1: NHIS is National Health Interview Survey; NHIS rates are age adjusted to overall PLCO population; MI: myocardial infarction
296 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008
Enrollment of Racial and Ethnic Minorities
about the PLCO Trial. The increased proportional enrollment of blacks at these three centers showed that these approaches were successful in increasing the proportion of blacks enrolled in the trial. Cultural as well linguistic factors were also an important component of the special effort for Hispanic recruitment and retention at Colorado. Having a bilingual and bicultural staff and having that staff available at points of participant contact, having bilingual study materials available and including the family in the recruitment process were critical elements in this center’s successful effort to enroll significant numbers of Hispanics. In contrast, earlier efforts at directed mass mailings for blacks and Hispanics at several centers were not successful alone in increasing recruitment, even when the recruitment materials were created by minorities for minorities, because they failed to adequately address cultural concerns.
Efforts by the PLCO Trial to Address Structural Factors A number of structural factors that had the potential to impact minority recruitment in PLCO were addressed at some or all of the centers. Lack of insurance was a structural issue in PLCO because diagnostic follow-up was not provided by the trial and, thus, participants were responsible (financially and otherwise) for their own follow-up. All PLCO centers attempted to make provisions for the uninsured; however, this aspect of trial design may have been a structural barrier to recruitment of blacks and Hispanics. The location of the screening center and problems with transportation may also have been a structural
factor affecting minority recruitment at some centers. Washington University initially planned to locate their screening center within the city of St. Louis; however, before the start of the trial their proposal was amended with the screening center now located in Creve Coeur, a suburb in the middle of St. Louis County, about 10 miles west of the St. Louis city (western) boundary. Among age-eligibles, blacks made up almost 50% of the population in the city of St. Louis but only 13% of the population in St. Louis County; roughly equal numbers of blacks reside in the city and the county (note the city of St. Louis city is separate from St. Louis County). This may partially explain the low percentage enrollment of blacks. In contrast, at HFHS, an additional screening site located in a predominantly black area of Detroit was added shortly after the trial opened; 21% of screenees at this additional site were black, compared to 11% at the original screening site, which was located in suburban Dearborn. The special efforts at Colorado and UAB addressed transportation needs by providing rides to screening appointments. The Georgetown center attempted to address transportation issues by negotiating with a hospital in Washington, DC, that was located closer to black and Hispanic population centers to perform recruitment and screening for PLCO. However, for various reasons, this project did not succeed, and few subjects were ever recruited or screened at this facility. Education and lifestyle were other structural factors affecting recruitment. Blacks, Hispanics and (non-Hispanic) whites in PLCO were all more highly educated and had healthier lifestyles in terms of not smoking and
Table 3B. Demographics, behavior and medical history by race in PLCO and National Health Interview Survey, women
Current Smoker Former Smoker Never Smoker BMI >30 Regular Exercise Alcohol Use (Past Year) Education < High school College degree Married History of Diabetes History of Hypertension History of Coronary Heart Disease, MI or Stroke
White (Non-Hispanic)
Black (Non-Hispanic)
Hispanic
PLCO
NHIS
PLCO
NHIS
PLCO
NHIS1
% 9 35 56 24 85 70
% 18 30 52 21 54 49
% 16 34 50 45 72 50
% 18 24 58 41 34 24
% 10 35 55 29 82 67
% 11 20 69 27 35 30
6 30 71 6 33
19 16 65 11 43
17 26 40 20 63
41 11 36 26 66
19 25 59 12 37
59 7 53 22 45
7
11
14
13
8
10
1
1
1: NHIS is National Health Interview Survey; NHIS rates are age adjusted to overall PLCO population; MI: myocardial infarction
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 100, NO. 3, MARCH 2008 297
Enrollment of Racial and Ethnic Minorities
exercise, than their counterparts in the general population. Thus, the factors that correlate with clinical trial participation appear to be largely the same in blacks and Hispanics as in whites. However, since educational attainment is lower on average among blacks and Hispanics as compared to whites, this was likely a factor in the underrepresentation of these groups in PLCO. Special recruitment efforts for minority populations involve additional funding. In PLCO, it was not always possible to directly ascertain the amount of funds dedicated to special minority recruitment and retention efforts at a given center. Therefore, as a crude method of gauging the costs of these efforts, we compared the per-capita cost of the trial for centers with and without special recruitment efforts, where the per-capita cost was calculated by dividing the center’s total PLCO budget (for all study years through year 13) by the number of enrolled subjects. For the six centers not involved in special recruitment efforts, the aggregate per-capita cost was $1,285. Per-capita costs at the two centers with relatively small-scale minority efforts—HFHS and Pittsburgh—were similar. However, per-capita costs were considerably higher at the two centers with extensive minority efforts—Colorado, with a per-capita cost of $1,622; and UAB, with a cost of $2,646. The higher per-capita costs at these two centers generally reflect the intensity and duration of each of the centers’ special minority recruitment and retention efforts. The PLCO trial, with 5.0% black and 1.8% Hispanic enrollment, generally equaled or exceeded the percentages of blacks and Hispanics enrolled in other recent multicenter cancer screening or prevention trials. The Prostate Cancer Prevention Trial, a randomized trial of finasteride for the prevention of prostate cancer carried out at 221 centers across the United States, had black and Hispanic percentages among enrollees of 3.8% and 2.6%, respectively.14 The STAR trial comparing tamoxifen and raloxifene for breast cancer prevention was 2.5% black and 2.0% Hispanic.15 The CARET trial for lung cancer prevention was 2.9% black and 1.5% Hispanic.16 One trial whose minority participation did substantially exceed that of PLCO was the SELECT (Selenium and Vitamin E) Prostate Cancer Prevention trial, which had a goal of 24% minorities and whose final enrollment included 15% blacks, 5% Hispanics and 1% Asians.17 The success of SELECT in achieving high minority representation resulted from a comprehensive approach to minority recruitment, which began during the planning phase of the study.17 In conclusion, the experience with PLCO suggests that overcoming cultural barriers involving lack of information and mistrust of clinical trials is critical in achieving proportional levels of minority enrollment, especially among blacks. Culturally sensitive approaches to recruitment, as well as bilingual staff and materials,
298 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
were critical in successful efforts to increase Hispanic enrollment. Efforts to overcome the structural barriers of lack of transportation, lack of health insurance, and work or family conflicts were also important. These efforts overall—and especially at the PLCO centers with special minority recruitment programs—resulted in levels of black and Hispanic enrollment that were on par with or higher than those seen in several recent largescale cancer screening or prevention trials. Despite these efforts, however, the proportions of blacks and Hispanics in PLCO were still considerably lower than their respective proportions among PLCO age-eligibles in the overall U.S. population or in the aggregate PLCO catchment areas.
References
1. Ford JG, Howerton M. The science of recruiting minority populations to screening trials. Clin Trials. 2004;1:341-342. 2. Lai GY, Gary TL, Tilburt J, et al. Effectiveness of strategies to recruit underrepresented populations into cancer clinical trials. Clin Trials. 2006;3:133-141. 3. Guiliano AR, Mokuau N, Hughes C, et al. Participation of minorities in cancer research: the influence of structural, cultural, and linguistic factors. Ann Epidemiol. 2000;10:S22-S34. 4. National Institutes of Health. NIH Guidelines on the Inclusion of Women and Minorities in Clinical Research. http://grants.nih.gov/grants/guide/ notice-files/NOT-OD-02-001.html. Accessed 10/20/06. 5. Prorok PC, Andriole GL, Bresalier RS, et al. Design of the Prostate, Lung, Colon and Ovarian (PLCO) cancer screening trial. Control Clin Trials. 2000;21:273S-309S. 6. Ford ME, Havstad SL, Davis SD. A randomized trial of recruitment methods for older African American men in the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial. Clin Trials. 2004;1:343-351. 7. Ford ME, Havstad SL, Tilley BC. Recruiting older African American men to a cancer screening trial (the AAMEN Project). Gerontologist. 2003;43:27-35. 8. Simpson NK, Johnson CJ, Ogden SL, et al. Recruitment strategies in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials. 2000;21:365S-378S. 9. Stallings FL, Ford ME, Simpson NK, et al. Black participation in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials. 2000;21:379S-389S. 10. U.S. Department of the Census. Census 2000. www.census.gov/main/ www/cen200.html. Accessed 02/12/07. 11. National Health Interview Survey 2005. www.cdc.gov/nchs/about/ major/nhis/quest_data_related_1997_forward.htm. Accessed 02/12/07. 12. Trauth JM, Jernigan JC, Siminoff LA, et al. Factors affecting older African American women’s decisions to join the PLCO Cancer Screening Trial. J Clin Oncology. 2005;23:8730-8738. 13. Jernigan JC, Trauth JM, Neal-Ferguson D, et al. Factors that influence cancer screening in older African American men and women: focus group findings. Fam Community Health. 2001;24:27-33. 14. Moinpour CM, Atkinson JO, Thomas SM, et al. Minority recruitment in the Prostate Cancer Prevention Trial. Ann Epidemiol. 2000;10:S85-S91. 15. Vogel VG, Constantino JP, Wickerham DL, et al. Effects of tamoxifen vs raloxifene on the risk of developing invasive breast cancer and other disease outcomes: the NSABP Study of Tamoxifen and Raloxifene (STAR) P-2 trial. JAMA. 2006;295:2727-2741. 16. Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med. 1996;334:1150-1155. 17. Cook ED, Moody-Thomas S, Anderson KB, et al. Minority recruitment to the Selenium and Vitamin E Cancer Prevention Trial (SELECT). Clin Trials. 2005;2:436-442. n
VOL. 100, NO. 3, MARCH 2008