Barriers to recruiting underrepresented populations to cancer clinical trials: A systematic review

Abstract

Racial and ethnic minorities, older adults, rural residents, and individuals of low socioeconomic status are underrepresented among participants in cancer-related trials. The authors conducted a systematic review to determine the barriers to participation of underrepresented populations in cancer-related trials. Their search included English-language publications that reported original data on the recruitment of underrepresented groups to cancer treatment or prevention trials between 1966 and December 2005 in multiple electronic databases. They also hand-searched titles in 34 journals from January 2003 to December 2005 and they examined reference lists for eligible articles. Titles and abstracts were reviewed to identify relevant studies. Data on barriers to participation were synthesized both qualitatively and based on statistically significant associations with trial enrollment. Of 5257 studies that were cited, 65 studies were eligible for inclusion in the current analysis, including 46 studies on recruitment into cancer therapeutic trials, 15 studies on recruitment into prevention trials, and 4 studies on recruitment into both prevention and treatment trials. Numerous factors were reported as barriers to participation in cancer-related trials. However, only 20 of the studies reported statistically significant associations between hypothesized barriers and enrollment. The available evidence had limitations in quality regarding representativeness, justification of study methods, the reliability and validity of data-collection methods, potential for bias, and data analysis. The results indicated that underrepresented populations face numerous barriers to participation in cancer-related trials. The current systematic review highlighting the literature on recruitment of underrepresented populations to cancer trials and may be used as the evidence base toward developing an agenda for etiologic and intervention research to reduce the disparities in participation in cancer-related trials. Cancer 2008. © 2007 American Cancer Society.

The participation of racial/ethnic minorities and other underrepresented groups in clinical trials is a critical link between scientific innovation and improvements in healthcare delivery and health.1 Since 1993, the National Institutes of Health has required that all sponsored research, including cancer-related trials, ensure the inclusion of women and members of minority groups.2, 3 Between 1993 and 2002, the budget of the National Cancer Institute (NCI) doubled, and clinical trial accrual increased.4 Nevertheless, only 2.5% of cancer patients are enrolled into trials, and the following populations are underrepresented among participants in NCI-funded clinical trials: African-American men, Latinos/Hispanics, Asian and Pacific Islanders, American Indians/Alaska Natives, adolescents, older adults (aged ≥65 years), individuals who reside in rural areas, and individuals of low socioeconomic status.5 Most publications that report the results from cancer clinical trials do not describe the participants' race or ethnicity.6 This lack of diversity in randomized study populations reduces opportunities for discovering effects that may be particularly relevant to underrepresented populations6, 7 and contributes to inequitable distribution of benefits and risks of trial participation.1, 8 Some investigators have suggested that access to clinical trials should be an objective9 and a component of state-of-the-art cancer care.10

Racial differences in access to state-of-the-art care contribute to disparities in cancer mortality and survival after cancer diagnosis.4, 11, 12 However, these disparities disappear when African Americans and Caucasians receive similar cancer treatment for same-stage disease.13 A number of publications have documented key barriers to participation in cancer-related trials; however, there is a lack of systematic review and synthesis of the barriers that affect accrual of underrepresented populations to cancer-related trials. We performed a systematic review of the literature regarding barriers to participation of underrepresented populations in controlled trials of cancer prevention or treatment and investigated whether these barriers vary by sociodemographic factors, cultural factors, or both.

MATERIALS AND METHODS

Underrepresented Populations

For this review, “underrepresented populations” included adolescents, older adults (aged ≥65 years), individuals of low socioeconomic status, individuals who resided in rural areas, African Americans, Latinos/Hispanics, Asian Americans and Pacific Islanders, and American Indians/Alaska Natives.

Conceptual Framework

Our previously described conceptual model14 was modified to demonstrate the relations between factors that lead to enrollment in a clinical trial (see Fig. 1). According to this model, to accept or refuse participation in a clinical trial, an individual must be aware that the study is being conducted and must have an opportunity to participate.15 Given the opportunity to participate, an individual may seek additional information, thereby achieving greater awareness of trials and of the potential harms and benefits of participation. The conceptual framework categorizes barriers based on their predicted effects on awareness, opportunity, and the decision about participation.

Figure 1. Conceptual framework.

Inclusion Criteria and Data Abstraction

Our search strategy for relevant articles has been reported elsewhere.15 Briefly, it involved 1) a keyword search of electronic databases from 1966 to December 2005, including the MEDLINE database, the Cochrane Central Register of Controlled Trials (Issue 1, 2003); the Cochrane Database of Methodology Reviews (CDMR); the Cumulative Index of Nursing and Allied Health Literature database (CINAHL); the Psychological Abstracts data base (PsycINFO); and the Campbell Collaboration's Social, Psychological, Educational, and Criminological Trials Register (C2-SPECTR); 2) a hand search of 34 relevant journals from January 2003 through December 2005, which was the end date for eligible publications; 3) an examination of reference lists from eligible articles; and 4) an author search in MEDLINE of selected experts known to have interests related to our questions.

To be included, an article had to be written in English and had to report original data on barriers to participation of underrepresented populations into a cancer prevention or treatment trial. Potential articles from the title and abstract reviews were abstracted fully by using a structured instrument. During the article review phase, data were abstracted twice from 270 articles by using a standardized data abstraction form and a serial article review process (Fig. 2). This involved a primary reviewer, who completed the data abstraction form, and a secondary reviewer, who checked the primary review for accuracy and completeness. Data captured on the abstraction forms included whether specific patient, provider, study design, and healthcare system level barriers were associated with statistically significant or qualitatively important decreases in enrollment to clinical trials. Both reviewers performed an independent assessment of the study's quality by using a separate quality-assessment form. Unresolved disagreements between reviewers were adjudicated by a third reviewer in the group. Data from the article review process were entered into a relational database, ProCite (ISI ResearchSoft, Berkeley, Calif). The database was used to maintain and clean the data and to create evidence tables.

Figure 2. Summary of literature search and review process.

Grading of the Total Body of Evidence

The evidence-grading scheme was based on the approach recommended by the International Grading of Recommendations Assessment, Development, and Evaluation (GRADE) Working Group.16 The evidence was downgraded for limitations in quality, important inconsistencies, imprecise/sparse data, high probability of reporting bias, or uncertainty about how or the extent to which the individuals, measures, and/or outcomes were similar to those of interest.16 Study quality was evaluated based on 5 domains: representativeness, justification of study methods, reliability and validity of data-collection methods, potential for bias, and data analysis. To calculate quality scores for each area or domain, we averaged Reviewer A and Reviewer B scores. A score ≥50% in 1 of the 5 domains was deemed “adequate”; conversely, a score from 0% to 49% in any of the 5 domains was deemed “inadequate.”

RESULTS

Of 5257 citations, 65 eligible articles served as the basis for this analysis (Fig. 2).5, 10, 17–77,83,88 The most frequent reasons for exclusion at abstract review were not reporting a controlled trial or recruitment to a controlled trial and no original data. The most frequent reasons for exclusion at article review were not addressing an underrepresented group and not reporting on a controlled trial. Because of the diversity of target populations and heterogeneity of study designs, a formal quantitative summary of pooled study effects, or meta-analysis, was not possible. However, a qualitative synthesis of the available evidence is provided below.

Study Characteristics

The study characteristics are summarized in Table 1. Greater than 50% of the 65 articles (n = 41 studies) were published in 2001 or later, 14 were published between 1996 and 2000, and 10 were published in 1995 or earlier. Most of the articles were descriptive/cross-sectional (n = 33 studies). Twenty-four articles used qualitative methods. The articles reported on recruitment in diverse settings, including hospitals (n = 43 studies) and community (n = 35 studies). Studies of accrual to cancer therapeutic trials (n = 45 studies) were more common than those of accrual to prevention trials (n = 15 studies) and chemoprevention trials (n = 4 studies). In most articles (n = 55 studies), the respondents were patients or participants. Nearly 66% of the articles (n = 43 studies) reported on barriers to accrual to an actual trial. The remaining articles (n = 22 studies) reported on barriers in the absence of an actual trial (n = 13 studies). Outcome measures for these studies included respondents' willingness to participate (n = 10 studies) or perceived barriers to participation (n = 13 studies).

Table 1. Characteristics of Studies on Recruitment of Underrepresented Populations Into Controlled Trials of Cancer Prevention or Treatment*

Study/Year	Sample size	Study setting	Target population	Targeted underrepresented population	Study design	How measured	Type of cancer studied	Type of cancer trial into which recruited
AA indicates African American; Lat/Hisp, Latino/Hispanic, CCT, concurrent controlled trials; NR, not reported; SWOG, Southwest Oncology Group; RCT, randomized controlled trial; NA, not available; NCI, National Cancer Institute; SES, socioeconomic status; CTEP, Cancer Therapy Evaluation Program; NCCTG, North Central Cancer Treatment Group; CALGB, Cancer and Leukemia Group B; UCD, University of California-Davis; PDQ, Physician Data Query; SEER, Surveillance, Epidemiology, and End Results Program. * This table summarizes eligible studies, including the names of the authors, year of publication, study sample size, targeted underrepresented population, study design, methods used to characterize barriers, type of cancer, and whether the article addressed barriers to a prevention trial, treatment trial, or both.
Adams-Campbell, 200417	235	University medical center	Patients/participants, physicians	AA	Tumor registry review	Enrollment	Breast, colorectal, head and neck, lung, prostate	Treatment
Advani, 200318	218	Hospital outpatient	Patients/participants	AA	Case-control	Surveys, phone interview	Solid/hematologic malignancies (excluding melanoma)	Treatment
Bailey, 200419	175	University medical center	Patients/participants	AA	Descriptive	Case report	Cervical	Treatment
Barofsky and Sugarbaker, 197920	103	Community, hospital inpatient	Patients/participants	AA	Qualitative	Semistructured interview	Histologically confirmed soft tissue or osteosarcoma	Treatment
Berman, 199821	435	Community	Patients/participants	Lat/Hisp, AA	Quasiexperimental	Survey, enrollment	Smoking cessation	Prevention
Bieniasz, 200322	66	Community	Patients/participants	AA women	Qualitative	Semistructured interview	Cervical (dysplasia)	Prevention
Brewster, 200223	545	Community, hospital outpatient	Patients/participants	Lat/Hisp women	CCT	Enrollment	Cervical	Prevention
Broome and Richards, 200325	34	Hospital inpatient	Patients/participants	Children, adolescents	Qualitative	Semistructured interview	Ewing sarcoma, hematologic malignancies	Treatment
Broome, 200124	34	Community, hospital inpatient, hospital outpatient	Patients/participants	Children, adolescents	Qualitative	Interview	Hematologic malignancies, solid tumors	Treatment
Brown, 200026	434	Hospital outpatient	Patients/participants	AA and Hisp women	Descriptive	Survey, interview, enrollment	Breast, cervical	Prevention
Brown and Topcu, 200327	438	Community	Patients/participants	AA	Cross-sectional, descriptive	Telephone surveys	NR	Treatment
Chen, 200028	68	Hospital inpatient	Patients/participants	Older adults	Hospital registry review, retrospective	Enrollment NR	Aggressive histology lymphoma	Treatment
Comis, 200329	887	Community	Patients/participants	Older adults (by subgroup analysis)	Cross-sectional, descriptive	Telephone interview	NR	Treatment
Cook, 200530	35,534	Community, other (SWOG)	Patients/participants	AA, Hisp, Asians	Descriptive	Focus groups	Prostate	Prevention
Diener-West, 200131	4191	Hospital inpatient, hospital outpatient, group practice, solo practice	Patients/participants	Older adults	RCT	Enrollment, face-to-face interview	Eye	Treatment
Ellington, 200632	55	Community	Patients/participants	Hisp	Qualitative	Focus groups	All	Treatment
Ford, 200433	12,400	Community	Patients/participants	AA men	RCT	Enrollment NR	Prostate, lung, colorectal	Prevention
Fouad, 200034	103	Community	Patients/participants, physicians	AA	Qualitative	Focus groups/key informant interview	NR	NR
Fouad, 200183	103	Community	Patients/participants, physicians	AA	Qualitative	Focus groups/key informant interview	NR	NA
Fouad, 200488	373,158	Community	Patients/participants	AA, American Indian, Asian/Pacific, Hisp	Descriptive	Survey	All cancers	Prevention
Grann, 200535	823	Community	Patients/participants	AA and Hisp women	Descriptive	Survey	Breast	Prevention
Green, 200036	103	Community	Patients/participants	AA	Qualitative	Focus group	NR	NR
Gross, 200540	8121	NCI Cooperative Group breast cancer trials	Patients/participants	Older adults, low SES, AA	Case-control	NCI CTEP data	Breast	Treatment
Gross, 200539	16,135	NCI-sponsored lung, breast, colorectal, prostate cancer trials	Patients/participants	Older adults	Longitudinal analysis	NCI CTEP data	Breast, colorectal, lung, prostate	Treatment
Gross, 200537	36,167	NCI-sponsored lung, breast, colorectal, prostate cancer trials	Patients/participants	Older adults	Cross-sectional, descriptive	NCI CTEP data	Breast, colorectal, lung, prostate	Treatment
Gross and Krumholz, 200538	1418 Counties	NCI-sponsored lung, breast, colorectal, prostate cancer trials	Patients/participants	Uninsured, rural, low SES	Descriptive, cross-sectional	NCI CTEP data	Breast, colorectal, lung, prostate	Treatment
Grunfeld, 200241	29	Hospital inpatient	Researchers	Clinical research associates	Qualitative	Semistructured interview	NR	Treatment
Hunter, 198742	17,773	Group practice, hospital inpatient	Patients/participants	AA	Descriptive, qualitative, retrospective	Secondary analysis of program data	All	Treatment
Jatoi, 200543	226	NCCTG	Patients/participants	Older adults	Descriptive, retrospective	NCCTG Clinical Trial registry	Nonsmall cell lung cancer	Treatment
Kaluzny, 199344	NR	Community, hospital outpatient, group practice	Patients/participants, physicians	NR	Descriptive, qualitative	NCI patient log, interviews from site visits	NR	Treatment
Kemeny, 200345	154	Community	Patients/participants, physicians	Older adults	Case-control	Patient interviews, self-administered written questionnaires for physicians	Breast	Treatment
Kemp, 198446	1022	Community	Patients/participants	Older adults (by subgroup analysis)	Descriptive, cross-sectional	Face-to-face interview, survey	Breast, bone	Treatment
Kimmick, 200547	101 Symposium, 29 seminar, 500 monthly e-mails, 800 monthly mailings	CALGB institutions (community and academic hospitals)	Physicians and healthcare providers	Older adults	RCT	Enrollment	Cancer in general (breast, prostate, leukemia, lymphoma, melanoma, genitourinary)	Treatment
Kornblith, 200248	156	Hospital inpatient	Physicians	Older adults	Cross-sectional, descriptive	Questionnaire	Breast	Treatment
Krailo, 199349	2788	Hospital inpatient, hospital outpatient	Patients/participants	Children, adolescents	Hospital registry review, qualitative	Enrollment NR	All	Treatment
Lara, 200150	276 Patients, 18 physicians	Hospital outpatient clinic	Physicians	Older adults	Descriptive	Survey	All	Treatment
Lara, 200551	596	University medical center	Physicians	AA, Asian/Pacific Islander, Hisp	Descriptive	Survey	All	Treatment
Lee, 199952	165	Hospital inpatient	Patients/participants	AA men	Descriptive, cross-sectional	Survey	Prostate	Prevention
Lerman, 199453	271	Community	Patients/participants	AA women	Descriptive, cross-sectional	Structured telephone interview	Breast	Prevention
Lewis, 200354	59,300	NCI cooperative groups	Patients/participants	Older Adults	Cross-sectional, descriptive, retrospective	Secondary analysis of national database (NCI, CTEP, PDQ, SEER)	NR	Treatment
Ling, 200055	1206	Hospital inpatient, hospital outpatient	Patients/participants	Older adults (by subgroup analysis)	Qualitative	Survey	NR	Treatment
Mandelblatt, 200556	450 (Intervention, n = 232; control, n = 218)	Community health facility	Patients/participants	Hisp/Lat women	RCT	Intent to participate	Breast	Prevention
Martel, 200457	23 Providers, 216 patients	University medical center	Patients, physicians	Older adults	Descriptive	Survey	All	Treatment
McCaskill-Stevens, 199958	89	Community	Physicians	AA	Qualitative	Workshops with discussions and questionnaires	All	Prevention, treatment
Melisko, 200559	297 patients, 127 physicians	Community health facility, hospital outpatient clinic, private and academic medical centers	Patients, physicians	Minorities	Descriptive, cross-sectional	Mailed survey for physicians, face-to-face survey for patients	Breast	Treatment, (Complementary and Alternative Medicine)
Millon-Underwood, 199360	220	Community	Patients/participants	AA	Descriptive, cross-sectional	Survey	NR	Prevention, treatment
Moinpour, 200061	NR	Hospital outpatient, University medical center	Patients/participants, researchers	AA and Hisp men	RCT	Enrollment NR	Prostate	Prevention
Mouton, 199762	80	Community	Patients/participants	AA women	Descriptive, cross-sectional	Telephone survey	Breast, colorectal	Prevention
Murthy, 200463	75.215	NCI-sponsored lung, breast, colorectal, prostate cancer trials	Patients/participants	Minorities (AA, Hisp, Asian/Pacific Islanders, American Indian/Alaskan Native), older adults	Cross-sectional, descriptive	Secondary analysis of NCI Cancer Clinical Data Update Service	Breast, colorectal, lung, prostate	Treatment
Outlaw, 200010	52	Hospital inpatient	Physicians, data managers	AA	Qualitative	Survey	NR	NR
Paskett, 200264	360	Community	Physicians	Rural	CCT	Survey, enrollment NR	Breast, colorectal	Treatment
Paterniti, 200565	59 Patients observed	University medical center	Patients/participants, families/caregivers	Asian Americans	Qualitative	Field observation	All	Treatment
Pinto, 200066	73	Community	Physicians	AA	Qualitative	Focus groups	NR	Prevention, treatment
Randall-David, 200167	37	Community	Patients/participants	Rural	Qualitative	Focus groups	All	Treatment
Roberson, 199468	28	Community	Patients/participants	AA, American Indian or Alaskan Native, Lat/Hisp	Qualitative	Telephone survey	NR	Treatment
Robinson, 199669	46	Community	Patients/participants	AA men	Qualitative	Focus groups	Prostate	Treatment
Sateren, 20025	24,332	All patients from NCI-funded trials between April 1998 and April 1999	Patients/participants	AA	Cross-sectional, descriptive, retrospective	Secondary analysis of national database (US Census, SEER)	All	Treatment
Sears, 200370	558	Community	Patients/participants	AA women	RCT, retrospective	Enrollment NR	Breast	Treatment
Simon, 200471	319	Hospital inpatient	Patients/participants	AA	Descriptive	Survey	Breast	Treatment
Spaight, 198472	23	Hospital inpatient	Physicians	Older adults	Qualitative	Structured interview	NR	Treatment
Tercyak, 200673	209	Hospital tumor registry	Patients/participants	Adolescents	Descriptive	Telephone interviews	All	Survivorship
Thornquist, 199174	1284	Community	Patients/participants	Older adults (by subgroup analysis)	Cohort, prospective	Enrollment NR	Lung	Prevention
Trauth, 200575	299	Community	Patients/participants	AA women	Qualitative	—	Lung, colorectal, ovarian	Prevention
Twelves, 199876	4688	Hospital inpatient, hospital outpatient, solo practice	Patients/participants	Older adult women (by subgroup analysis)	Retrospective, tumor registry review	Enrollment NR	Breast	Treatment
Woods, 200277	120	Hospital outpatient	Patients/participants	AA	Qualitative	Focus groups/interview	NR	Prevention, treatment

The majority of the articles (n = 35 studies) reported data on African-American participants; fewer studies focused on Latinos/Hispanics (n = 12 studies), Asian and Pacific Islanders (n = 6 studies), and American Indians/Alaska Natives (n = 4 studies). Whereas some articles reported on the accrual of older adults (n = 19 studies), only 4 articles focused exclusively on adolescents or children. We did not synthesize the evidence on barriers by trial type (eg, Phase I/II vs Phase III trials), because most of the studies did not specify trial phase.

The eligible studies identified 150 distinct barriers to accrual to cancer-related trials, including 124 barriers to accrual to therapeutic trials only, 26 barriers to accrual to prevention trials only, and 32 barriers to accrual to both therapeutic and prevention trials. The studies reported 8 barriers to awareness, 88 barriers to the opportunity to participate, and 44 barriers to acceptance of enrollment into a cancer-related trial. Among these, 79 barriers were relevant at the patient level, 37 were relevant at the provider level, 25 were relevant at the study design level, and 8 were relevant at the healthcare systems level. The most frequently reported barriers were 1) mistrust of research and the medical system (n = 20 studies), 2) perceived harms (n = 18 studies), 3) costs of participating (n = 17 studies), 4) patient demographics (n = 16 studies), 5) availability of transportation (n = 15 studies), 6) lack of education about clinical trials (n = 13 studies), 7) fear (n = 10 studies), 8) time commitment (n = 9 studies), and 9) family issues (n = 8 studies) (see Fig. 3).

Figure 3. Number of studies reporting on patient barriers to clinical trial enrollment (n = 45 studies). CTs indicates clinical trials.

Barriers to Awareness, Opportunity, and Acceptance of Enrollment

Barriers to clinical trial awareness

Table 2 lists the factors that were associated with a statistically significant decrease in reported enrolment (Fig. 4). Three articles18, 51, 75 reported an association between lack of awareness and reduced participation in trials. In addition, one of those studies reported lack of knowledge about the origins of cancer as a barrier to enrollment.75 Qualitatively, in the 18 studies that reported on barriers to awareness, lack of education regarding clinical trials was the most frequently reported barrier (n = 13 studies). Additional awareness barriers included lack of culturally appropriate information (n = 6 studies), cancer knowledge (n = 5 studies), and physician awareness of trials (n = 4 studies)

Figure 4. Number of studies reporting on barriers to clinical trial awareness (n = 18 studies).

Table 2. Factors Associated With a Decrease in Clinical Trial Enrollment: Awareness, Opportunity, and Decision-Making Barriers

Author and year	Barriers to enrollment
	Awareness	Opportunity	Decision making
Advani, 200318	Patient barriers: Lack of education about clinical trials	—	Patient barriers: Perceived harms of clinical trial participation; chance of side effects; distance to clinic
Bieniasz, 200322	—	—	Patient barrier: Intervention characteristics
Brown and Topcu, 200327	—	Patient barriers: African American older age; lower income	—
Diener-West, 200131	—	Patient barriers: Age; living near treatment center	—
Ford, 200433	—	Patient barriers: Unlisted telephone number; age	—
Grann, 200535	—	Patient barriers: SES; African American race	—
Gross, 200540	—	Patient Barriers: SES (reside in high poverty zip code)	—
Gross, 200539	—	Patient barriers: Poverty; study design barriers: no. of comorbid conditions	—
Gross, 200537	—	Study design barriers: Comorbidity exclusion; age exclusion	—
Gross and Krumholz, 200538	—	Provider barriers: Index of managed care competition	—
Kemeny, 200345	—	Patient barriers: Age; study design barriers: disease stage, no. of comorbidities	—
Lara, 200551	Patient barriers: Lack of education about clinical trials	Patient barriers: Race (African American or Asian); age (18–24 y); income < $25,000	—
Martel, 200457	—	Provider barriers: Referral source (ie, surgeons)	—
Melisko, 200559	—	—	Patient barriers: Transportation; time commitment; loss of income; loss of control
Murthy, 200463	—	Patient barriers: Race (African American; Hispanic)	—
Sateren, 20025	—	Patient barriers: Lack/inadequate health insurance; being a black man aged 30–59 y; Asian adults and Latino/Hispanic adults	—
Simon, 200471	—	Patient barriers: Lack/inadequate health insurance	—
Thornquist, 199174	—	Patient barriers: Age (50–54 y; 65–69 y)	—
Trauth, 200575	Patient barriers: Lack of knowledge about origins of cancer; lack of education about clinical trials	—	—
Twelves, 199876	—	Patient barriers: Age	—

Barriers to opportunity to participate

Sixteen articles reported an association between lack of opportunity and reduced participation in trials.5, 27, 31, 33, 35, 37–40, 45, 51, 57, 63, 71, 74, 76 Most of those articles reported an association between specific sociodemographic characteristics, such as older age, socioeconomic status, and racial/ethnic minority status, along with lack/inadequate health insurance and reduced enrollment (Table 2). Additional opportunity barriers include study design barriers, namely, comorbid conditions and age-based exclusion,37, 39, 45 lack of provider referral,57 and the extent of managed care competition.38 Qualitatively, the most frequently reported barriers to the opportunity to participate related to provider attitudes (n = 19 studies) and patient eligibility criteria (n = 18 studies); exclusion criteria, such as comorbid conditions (n = 11 studies) and functional status (n = 9 studies); patient/provider communication or the provider's method of presenting information about the trial to the patient (n = 7 studies); and the provider's data-collection costs or burden (n = 6 studies). In some studies, demographic factors, such as age (n = 8 studies), race/ethnicity (n = 6 studies), and not being married (n = 1 study), in themselves, were associated with reduced opportunity to participate (Fig. 5).

Figure 5. Number of studies reporting on barriers to clinical trial opportunity (n = 34 studies).

Barriers to acceptance of enrollment

Three articles reported an association between specific factors and the decision not to participate in trials (Fig. 6).18, 22, 59 These barriers included the perceived harms of clinical trial participation, loss of control (uncertainty about treatment allocation), the nature of the intervention, the time commitment, loss of income, and transportation. Qualitatively, the most frequently reported barriers to acceptance of participation were mistrust of research and the medical system (n = 20 studies), direct and indirect costs of participation (n = 17 studies), transportation (n = 14 studies), time commitment (n = 10 studies), fear (n = 9 studies), family considerations (n = 8 studies), and factors related to the patient-provider relationship or the perception of the provider (n = 10 studies). Whereas mistrust was the most frequently reported barrier factor in the eligible studies, none of the studies reported a statistically significant association between mistrust and reduced enrollment.

Figure 6. Number of studies reporting on barriers to acceptance (n = 36 studies).

Effects of Demographic Factors

Older age, (n = 13 studies), lower socioeconomic status (n = 8 studies), and racial/ethnic minority status (n = 6 studies), in themselves, were reported as barriers to recruitment. These demographic factors related most frequently to study design barriers, such as exclusion by age restriction (n = 6 studies), comorbid conditions (n = 11 studies), functional status (n = 8 studies), and study duration and/or complexity of study procedures (n = 5 studies). In the recruitment of adolescents, parental influence was reported as a factor that influenced acceptance of participation (n = 4 studies). Across underrepresented populations, it was difficult to separate race/ethnicity and socioeconomic status from the barriers themselves.

Effects of Language and Cultural Factors

Thirteen articles reported that cultural factors, including lack of culturally relevant education about clinical trials (n = 7 studies), particularly in native languages of target populations (n = 6 studies), enhanced communication barriers. However, none of those articles reported on statistically significant associations between these factors and reduced trial enrollment. Moreover, the heterogeneity of the evidence and the overlap among underrepresented populations (eg, older African-American adults) limited our ability to synthesize the evidence regarding whether and how specific barrier factors vary across cultural contexts.

Quality of Studies

We assessed the quality of the 65 studies based on each of 5 criteria: representativeness, justification, bias, reliability and validity, and statistical analysis (Fig. 7). In terms of representativeness, 66% of the articles (n = 39 studies) clearly described the participants and settings. For the justification of study methods, 60% of the articles (n = 39 studies) provided an adequate rationale based on a conceptual model, a statement of the research question and objectives, and an adequate justification of the target population. The survey instruments and data-collection methods used in the majority of studies (n = 56 studies) showed limitations in reliability and validity. Eighteen studies used adequate measures to protect against bias or confounding. Only 8 studies provided adequate information to assess the quality of their data analysis.

Figure 7. Percentage of studies showing varying quality of evidence. N/A indicates not applicable.

The overall quality of the evidence on barriers to participation was assessed as low because of limitations in the quality and precision of eligible articles, lack of strong evidence within the studies, the high probability of reporting bias, and lack of a quantitative estimate of effect size.16 Although the studies were heterogeneous and contained > 627,044 participants in total, lack of information to answer our study questions was most prominent for Latino/Hispanics, American Indians/Alaska Natives, Asian and Pacific Islanders, adolescents, and rural populations.

DISCUSSION

Our systematic review suggests that numerous factors affect participation of underrepresented populations in cancer-related trials. However, few of the eligible articles reported a statistically significant association between specific barriers and clinical trial enrollment (Table 2). Consistent with our hypothesized conceptual framework, these barriers may influence recruitment outcomes through their effects on awareness of trials, the opportunity to participate, and the decision to accept or refuse participation.

The most commonly reported barrier factors relate to the opportunity to participate in trials; and among opportunity factors, individual characteristics, such as age, socioeconomic status, racial/ethnic minority status, and comorbid conditions, are associated most consistently with reduced clinical trial enrollment. Because opportunity barriers largely reflect protocol design as well as the process of study implementation, investigators play a major role in determining the extent to which trials are accessible to underrepresented groups.78 If such opportunity barriers affect underrepresented groups disproportionately, then they likely contribute to disparities in participation rates. However, taken together, the studies that demonstrate an association between opportunity barriers and reduced enrollment provide limited information about potential mechanisms that lead to the observed disparities in enrollment on the basis of race/ethnicity. Clearly, exclusion criteria based on age and comorbid conditions constitute barriers to enrollment of older adults. Nonetheless, the mechanisms that reduce opportunities for racial/ethnic minority populations are less evident, and they cannot be separated easily from the effects of low socioeconomic status. Our qualitative synthesis of the literature offers a wide array of hypothesized barriers that await confirmation as determinants of reduced opportunity for enrollment for racial/ethnic minority populations. However, until the effects of such barriers to enrollment are assessed quantitatively, uncertainty will persist about the mechanisms of reduced trial enrollment opportunities for racial/ethnic minorities.

Despite these limitations, the eligible articles suggest that there are multiple potential ways to improve opportunities for enrollment of underrepresented populations, including less rigid study designs and developing systems that facilitate the participation of healthcare providers. However, the steps taken by researchers to increase enrollment too often will prove inadequate in the absence of an infrastructure to support trials.80–82 Moreover, given the prominence of inadequate health insurance as a barrier, policy-level interventions are needed urgently to address the costs of trial participation.

Our qualitative synthesis of the literature illuminates multiple hypothesized barriers to awareness, including lack of trial dissemination and targeting of clinical trial enrollment opportunities on the basis of the cultural background and literacy level of the target population. However, only 3 studies demonstrated an association between lack of awareness (specifically, lack of education about clinical trials) and enrollment. On a qualitative basis, several articles suggested culturally targeted education as a strategy for increasing trial participation10, 32, 33, 51, 65, 75, 83; however, none of the eligible articles reported an association between culturally targeted education strategies and trial enrollment. This question needs to be addressed in future studies.

Similarly, many of the eligible articles identified psychosocial and cultural factors in the decisional balance regarding acceptance or refusal of participation in cancer-related trials. However, only 3 articles reported an association between specific factors and refusal to participate in trials. It is noteworthy that, whereas mistrust was the most frequently reported barrier factor in the eligible articles, none of the 20 articles that identified mistrust as a barrier factor reported a statistically significant association between mistrust and reduced enrollment. In this respect, the disconnect between the qualitative data and quantitative findings reflects, in part, the lack of standardized, quantitative measures of trust as applied to studies of determinants of trial enrollment: None of the 20 articles that reported trust as a barrier factor used a quantitative measure of trust.

The available literature suggests the need to improve the quality of studies on the recruitment of underrepresented populations. Among the articles on recruitment to cancer-related trials, as reported recently, we identified few randomized controlled trials or concurrent controlled trials.87 Likewise, in the current article, we report a rich qualitative literature with limited quantitative evidence on barriers to enrollment. There is an urgent need for hypothesis-driven, quantitative studies with enrollment into a trial as the primary outcome variable. Such an approach would help validate observations made in qualitative studies and would facilitate the eventual application of meta-analytic techniques toward identifying the most critical barriers and the most promising intervention strategies.

Limitations

The available evidence presents limitations in study quality with regard to 1) representativeness, 2) justification of study methods, 3) reliability and validity of data-collection methods, 4) the potential for bias/confounding; and 5) data analysis. Our search strategy included strict eligibility criteria, and some articles that contained potentially relevant evidence were not included in this review.9, 84–86 We used a standardized approach to generate the body of evidence that is presented in this review. However, because of the heterogeneity of study designs, we did not use formal meta-analytic techniques. In addition, many of the articles that were included did not facilitate a distinction between early-phase clinical trials and Phase III clinical trials.

For this review, we combined the results from both prevention and therapeutic trials. In reviewing prevention and treatment trials separately,15 we observed that there were similar barriers of awareness, opportunity, and acceptance for both prevention and therapeutic trials. Although, individually, many of the studies limited our ability to reach firm conclusions, the sum of the evidence and its consistency across multiple studies allow us to construct a conceptual framework for barriers to participation in cancer-related trials (Fig. 8).

Figure 8. Application of evidence to conceptual framework. SES indicates socioeconomic status; CTs, clinical trials.

Conclusions

In conclusion, the underrepresented populations in cancer-related trials face numerous barriers to participation. The number and complexity of the reported barriers to participation suggest that there is an urgent need for well-designed studies with appropriate methods and measures to establish key determinants of awareness, opportunity, and acceptance of participation. Knowledge of such determinants would provide the basis for the design of evidence-based intervention strategies directed to participants/patients, providers, institutions, and communities. Our systematic review highlights the state of the science on recruitment of underrepresented populations to cancer trials. It summarizes reported determinants of enrollment of underrepresented populations in cancer-related trials. It can be used as the evidence base toward developing an agenda for etiologic and intervention research to reduce the disparities in participation in cancer-related trials. This urgent need should be given a high priority.

Acknowledgements

This article is based on research conducted by Johns Hopkins Evidence-based Practice Center under contract to the Agency for Healthcare Research and Quality (contract no. 290-02-0018), Rockville, Md, with funding and scientific support provided by the National Cancer Institute, National Institutes of Health. The authors of this article are responsible for its contents, including any clinical and treatment recommendations. No statement in this article should be construed as an official position of the Agency for Healthcare Research and Quality, the National Cancer Institute, the National Institutes of Health, or of the United States Department of Health and Human Services.