Ineligible, Unaware, or Uninterested? Associations Between Underrepresented Patient Populations and Retention in the Pathway to Cancer Clinical Trial Enrollment

PURPOSE:

Cancer clinical trials can benefit current and future patients; however, Black patients, rural residents, and patients living in disadvantaged areas are often underrepresented. Using an adapted version of Unger and colleagues' model of the process of clinical trial enrollment, we evaluated the relationship between underrepresented patient populations and trial end points.

METHODS:

This retrospective study included 512 patients with breast or ovarian cancer who were prescribed a therapeutic drug at the University of Alabama at Birmingham from January 2017 to February 2020. Patient eligibility was assessed using open clinical trials. We estimated odds ratios and 95% CIs using logistic regression models to examine the relationship between underrepresented patient populations and trial enrollment end points: eligibility, interest, offer, enrollment, and declining enrollment.

RESULTS:

Of the patients in our sample, 27% were Black, 18% were rural residents, and 19% lived in higher disadvantaged neighborhoods. In adjusted models, each comparison group had similar odds of being eligible for a clinical trial. Black versus White patients had 0.40 times the odds of interest in clinical trials and 0.56 times the odds of enrollment. Patients living in areas of higher versus lower disadvantage had 0.46 times the odds of enrolling and 3.40 times the odds of declining enrollment when offered.

CONCLUSION:

Eligibility did not drive clinical trial enrollment disparities in our sample; however, retention in the clinical trial enrollment process appears to vary by group. Additional work is needed to understand how interventions can be tailored to each population's specific needs.

BACKGROUND

Clinical trials offer novel treatments for patients with cancer and advances in cancer treatment for patients who will be diagnosed in the future.^1,2 In the past 10 years, the number of drugs developed for breast and gynecological cancers on the basis of clinical trials has increased, which highlights the importance of evaluating efficacy and safety for those diagnosed.^3-5 Participation in cancer clinical trials has remained low for more than 2 decades.^6,7 However, a more recent study has shown differences by practice site with overall participation in cancer clinical trials being higher at National Cancer Institute designated sites (19%) while participation is much lower for community sites (approximately 4%).⁸ In addition to practice site, health status, race and ethnicity, and socioeconomic status influence cancer clinical trial enrollment. Certain groups of patients are underrepresented in clinical trials and are thus denied the benefits of participation, which limits the ability to generalize trial results to diverse patient populations. Lack of cancer clinical trial representation has been well documented for patients who are Black or African American individuals (henceforth referred to as Black), rural residents, or patients living in disadvantaged neighborhoods.^9-13 As few as 3%-9% of Black patients with cancer enroll in trials sponsored by pharmaceutical companies and National Cancer Institute's National Clinical Trials Network.¹⁴ Additionally, about one in five rural residents enroll in trials,¹⁵ and data are lacking for patients in areas of higher disadvantage.

The practical process by which patients enroll on trials is complex. Unger et al¹⁶ proposed a model which details the linear components central to the clinical trial enrollment process. First, a clinical trial has to be available at the institution where the patient is receiving care. Following availability of a trial is patient eligibility, comprising extensive eligibility criteria.¹⁷ These criteria, known as inclusion and exclusion criteria, are most often patient laboratory measures and comorbidities (eg, human immunodeficiency virus, cardiovascular disease, and previous cancer diagnoses). Of patients with cancer, 68%-92% have at least one comorbidity in addition to their cancer diagnosis,^18-20 thus, potentially excluding these individuals from trials. Patient-reported trial interest has also been described as a driver of enrollment.²¹ Patient-reported interest exists bidirectionally between patients and providers. Trial interest may be supported through health care provider engagement in discussions with a patient about trials, with the patient considering how the trials can be incorporated into his or her personal preferences, goals, and values. If the patient is uninterested, the patient and provider can further discuss participation and reassess if the patient is truly uninterested. However, trials can ultimately only be offered to those patients who are eligible. Finally, if offered, a patient can choose whether or not to participate in the specific trial (Fig 1). Each of these steps creates a potential for differential retention of patients who are underrepresented from eventual clinical trial enrollment. Thus, this study assessed retention in the clinical trial enrollment pathway (trial eligibility, interest, offer, and enrollment) in underrepresented cancer clinical trial populations, including patients who are Black, rural residents, or living in higher disadvantaged areas.

FIG 1.

Process of enrollment onto a clinical trial by provider and systems and patient levels adapted from conceptual model by Unger et al with model-estimated predicted probabilities.

METHODS

Study Design and Participants

This retrospective cohort study included patients with breast or ovarian cancer who received oncology care at the University of Alabama at Birmingham (UAB) from January 2017 through February 2020. Inclusion criteria included an incident breast or ovarian cancer diagnosis staged I-IV, age 18 years or older, female sex, complete address information, complete insurance data, and were prescribed a therapeutic drug. Patients who identified as a race other than Black or White were excluded. Cancer clinical trials were trials testing therapeutic drugs and were open to enrollment at UAB between 2016 and 2020. This study was approved by UAB's Institutional Review Board (300001910).

Outcomes

Eligible for a clinical trial.

Eligibility criteria for available cancer clinical trials were abstracted from ClinicalTrials.gov and OnCore, a Clinical Trial Management System used to track protocols and participants throughout the trial period. In our previous research, we abstracted all inclusion and exclusion criteria for these trials and found a substantial amount of heterogeneity among eligibility criteria.¹⁷ Using these eligibility criteria, we determined if patients were eligible to enroll onto at least one trial after treatment start date.

Interest in clinical trial participation.

Patient-reported outcomes (PROs) were captured in clinic via a treatment planning survey. This survey is typically administered to new patients receiving cancer treatment within the institution around the time of treatment decision, although some patients may not be new to the system and still receive the survey. These PRO instruments were created in response to the Centers for Medicare & Medicaid Innovation Center Oncology Care Model, which is specific to the Medicare population. Therefore, a higher proportion of patients with Medicare were surveyed. Patients were asked if they were interested in learning about or participating in a clinical trial and reported yes or no.

Offered a clinical trial.

Patient chart abstraction from the electronic health record (EHR) was performed to determine if a patient was offered to enroll onto a clinical trial (yes or no).

Enrollment onto a clinical trial.

Patient's enrollment status was determined via the EHR. Patient was considered as having ever or never enrolled.

Exposures

Race.

Race (Black or White) was self-reported by the patient, and the information was abstracted from the EHR.

Rurality.

Rurality or urbanicity of patient residence was determined using the Rural-Urban Commuting Area (RUCA) codes. RUCA codes map a census block group 12-digit Federal Information Processing Standard Publication codes to rural or urban designations.²²

Neighborhood disadvantage.

Area Deprivation Index (ADI) determines neighborhood disadvantage on the basis of resident census block group income, education, employment, and housing quality. ADI is scored from 1%-100%, with higher percentages representing higher neighborhood disadvantage. According to Kind et al, ADI scores of 1%-85% represent neighborhoods of lower disadvantage and 86%-100% represent higher disadvantage.^23,24

Patient Demographics and Clinical Characteristics

Patient age at diagnosis, self-reported sex, cancer type (breast or ovarian), and insurance status (private, Medicaid, Medicare, or, none) were abstracted from EHR data. Insurance status was determined at treatment initiation.

Statistical Analysis

Descriptive statistics were calculated using frequencies and percentages for categorical variables and medians and interquartile ranges for continuous variables. Differences in underrepresented patient group characteristics were calculated using measures of effect size such as Cohen's d (ie, the standardized mean difference; small: 0.2, medium: 0.5, and large: 0.8) or Cramer's V—which is based on the chi-square statistic. V of 0.1 is considered a small effect, 0.3 a medium effect, and 0.5 a large effect when comparing across two categories; 0.1 a small effect, 0.25 a medium effect, and 0.4 a large effect when comparing across more than two categories.²⁵ Odds ratios (ORs), predicted probabilities, and 95% CIs were estimated using five separate multivariable logistic regressions evaluating odds of clinical trial eligibility, interest, offer, enrollment, and declining enrollment. Interest in clinical trial participation only included those who answered the clinical trial interest question. Models estimating odds of offering a clinical trial and enrollment included only those eligible. The model estimating odds of declining enrollment included only those who were both eligible and offered a trial. All models contained age at diagnosis, race, cancer type, insurance status, RUCA, and ADI. These variables were selected as they were believed to be confounders. Analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC).

RESULTS

Sample Characteristics

A total of 512 patients with complete data were included in our study (Appendix Table A1, online only), 27% of patients were Black, 18% were living in rural areas, and 19% were living in neighborhoods of higher disadvantage (Table 1). Overall, the median age of our sample was 59 years at diagnosis, and 37% were Medicare beneficiaries. Patients with breast (70% of sample) and ovarian cancer (30%) were similar. However, women with ovarian compared with breast cancer were more often enrolled in Medicare insurance (49% v 32%, V = 0.22) and had higher median age (63 v 58; d = 0.43; Appendix Table A2, online only). Patients ineligible (n = 177) and patients who were eligible (n = 335) for clinical trial enrollment were compared; ineligible patients had higher median age at diagnosis (61 v 58; d = 0.30; Appendix Table A3, online only). Appendix Figure A1 (online only) shows the exclusion cascade.

TABLE 1.

Patient Demographics and Clinical Characteristics (N = 512)

Differences in Clinical Trial Enrollment by Race

In adjusted analyses, Black patients had similar odds of trial eligibility and, although not statistically significant, had 0.61 times the odds of being offered a trial when compared with White patients (95% CI, 0.34 to 1.10; Table 2). Additionally, Black patients had 60% lower odds of interest in clinical trial participation (OR, 0.40; 95% CI, 0.18 to 0.90) and 44% lower odds of trial enrollment (OR, 0.56; 95% CI, 0.32 to 0.98) compared with White patients. Black patients were also directionally more likely to decline enrollment when offered a trial (OR, 1.44; 95% CI, 0.65 to 3.19; Table 2), when compared with White patients, although the differences were not statistically significant.

TABLE 2.

Model-Estimated Odds Ratios and 95% CIs for Eligible (N = 512), Interest (n = 185), Offered (n = 335), Enrolled (n = 335), Declined Enrollment When Offered (n = 223) Models

Differences in Clinical Trial Enrollment by RUCA Status

Compared with patients living in urban areas, patients living in rural areas had similar odds of trial eligibility, interest, offer, enrollment, and declining enrollment onto a clinical trial (Table 2).

Differences in Clinical Trial Enrollment by ADI Status

Patients living in neighborhoods of higher disadvantage compared with patients living in neighborhoods of lower disadvantage had similar odds of eligibility, interest, and being offered a trial (Table 2). Patients living in neighborhoods of higher disadvantage had 54% lower odds of enrolling onto a trial (OR, 0.46; 95% CI, 0.24 to 0.89), with 29% and 47% enrolling for patients in higher versus lower areas of disadvantage, respectively (Fig 2). When offered to participate, they had 3.40 times the odds of declining enrollment (95% CI, 1.44 to 8.00) compared with patients living in lower disadvantaged areas (Table 2). Model‐estimated predicted probabilities and 95% CI for all models can be found in Appendix Table A4 and Appendix Table A5, online only.

FIG 2.

Model-estimated predicted proportions for trial eligibility (N = 512), interest (n = 185), offer (n = 335), and enrollment (n = 335). Independent variables of interest for each model include age at diagnosis, race, cancer type, insurance type, RUCA, and ADI. ADI, Area Deprivation Index; RUCA, Rural-Urban Commuting Area.

DISCUSSION

This study assesses the full process of clinical trial participation from interest to enrollment for groups of patients who are considered underrepresented in clinical trials. The multidecade issue of clinical trial accrual has the potential to exacerbate clinical outcomes (eg, disease, disability, and death) in drugs not typically tested in underrepresented patient populations.^26,27 We expected and found Black patients and patients in areas of higher disadvantage had lower odds of enrollment; however, we found that patients living in rural areas had similar odds of enrollment. We identified that different patient populations had differing odds of retention along the trial enrollment process. Although 35% of our sample was ineligible for an available clinical trial, trial eligibility did not differ by race, residence, or neighborhood disadvantage. Although eligibility did not appear to drive clinical trial enrollment disparities in our sample, eligibility remains an issue given that about a third of patients were found to be ineligible for an available study within our institution. Encouragingly, the National Cancer Institute's Experimental Therapeutics Clinical Trials Network and National Clinical Trials Network have updated and will continue to evaluate and expand appropriate eligibility criteria proposed by ASCO and Friends of Cancer Research.^28-30

Black patients had lower odds of enrolling onto a clinical trial, which is consistent with prior studies.^14,31-34 In our study, this could be associated with clinical trial interest, as Black patients had lower odds of interest in clinical trial participation when compared with White patients. There are many reasons Black patients may be less likely to enroll, including historical events, systemic racism, and policies that have adversely affected Black patients who have existed and continue to exist in our health care system.^9,35 The historical and current issues surrounding Black patients have likely caused mistrust in the health care system, which may also influence Black patients having lower odds of enrolling onto clinical trials. Additionally, the perceived lack of interest may be the result of lack of exposure to clinical trial knowledge, potentially by the lack of the adequate education materials given to the patient by the provider or system along with internet resources not often presented in lay language.^36,37 Our results suggest the need for further education about cancer clinical trials as an important target for improving disparities in clinical trial enrollment. One potential approach to engaging Black patients in discussions surrounding clinical trials is the use of navigators to provide education about clinical trials. Previous studies have shown that navigators have the ability to assist the clinic workflow and increase beneficial outcomes including clinical trial participation.^38-40 Findings by Fouad et al³⁹ showed that navigators were able to increase retention of Black patients with cancer in clinical trials when compared with those who enrolled without the assistance of a navigator. Navigation has the potential to overcome both specific barriers to trial participation and increase global knowledge of clinical trials. Another key strategy to improving trial participation is increasing physician engagement. Although patients may view the system (and the concept of clinical trials) as more tangential, prior studies have found that patients trust their own doctor's treatment decisions.^41,42 The clinician can play a major role in overcoming the low participation of Black patients by focusing on engaging in conversations surrounding risks and benefits of trial participation and offering trials to eligible patients.⁴³ This sentiment is supported both by our study results and the meta-analysis by Unger et al,⁴⁴ which demonstrated that when offered to participate, Black patients with cancer similarly enroll onto clinical trials when compared with White patients.

In contrast, patients living in higher disadvantaged areas had similar levels of trial interest and eligibility but lower odds of enrolling on a trial. Our study found that this was driven, at least in part, by patients declining participation. Disadvantaged patients may be less likely to enroll on trials because of the perceived financial and logistical barriers of enrollment. A study by Advani et al⁴⁵ of patients receiving cancer treatment found that education and income are important associations with participation. Additionally, patients may believe that on the basis of their income level they may not be able to afford participation.^46,47 Although insurance covers standard-of-care components of qualified clinical trials, patients may still have to pay medical costs (eg, copay, coinsurance, and deductible) and nonmedical costs (eg, transportation and accommodations). As of January 2022, Medicaid is also required to cover additional health care expenses in connection with qualified clinical trials.⁴⁸ Future interventions surrounding insurance and health education are needed in addition to interventions that provide resources for the monetary and time costs that are often uncompensated for clinical trial participants. Finally, there are also methodological approaches that can be implemented into clinical trial design to increase inclusion of underrepresented patients in cancer clinical trials, including decentralized trial designs and synthetic control arms.^49-52 Implementing PROs in relation to the clinical trial enrollment process and participation would be beneficial to understand the patient experience and patient perceptions.

This paper should be considered in light of limitations. For patients who enrolled, we did not assess if a patient withdrew from the clinical trial, which does not fully capture patients who agreed but did not ultimately participate for other reasons. In addition, offer and decline of clinical trial relied on documentation within the EHR, which may be missed because of some physicians' lack of documentation in clinic notes and likely underestimates this group of patients. Furthermore, our sample is of patients with breast or ovarian cancer seeking care at an academic medical center in the Southeastern United States and, therefore, cannot be applied to all populations who may have unique logistical barriers to participation.

In conclusion, retention in the pathway for clinical trial enrollment differed for different underrepresented patient populations. Although eligibility was similar across patient groups, overall lack of interest in clinical trials appeared to be a driver for Black patients, while patient declining enrollment was a driver for patients living in higher disadvantaged areas. Furthermore, additional work is needed to understand how interventions can be tailored to the clinical trial process to guide more equitable clinical trial participation.

APPENDIX

TABLE A1.

Patient Demographics and Clinical Characteristics by Patients Included (n = 512) and Excluded (n = 6,636) in This Study

TABLE A2.

Patient Demographics and Clinical Characteristics by Breast Cancer Diagnosis (n = 359) Versus Ovarian Cancer Diagnosis (n = 153)

TABLE A3.

Patient Demographic and Clinical Characteristics by Patients Ineligible (n = 177) and Patients Eligible for Enrollment on Clinical Trial (n = 335)

TABLE A4.

Model-Estimated Predicted Probabilities and 95% CIs for Eligible (N = 512), Interest (n = 185), Offered (n = 335), and Enrolled (n = 335) Models

TABLE A5.

Model-Estimated Predicted Probabilities and 95% CIs for Patients Who Were Offered a Trial but Declined Enrollment (n = 223)

FIG A1.

Study exclusion cascade.

DATA SHARING STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

AUTHOR CONTRIBUTIONS

Conception and design: Nicole E. Caston, Jaclyn Wall, Courtney Williams, Andres Azuero, Gabrielle B. Rocque

Financial support: Gabrielle B. Rocque

Administrative support: Gabrielle B. Rocque

Collection and assembly of data: Nicole E. Caston, Fallon Lalor, Jaclyn Wall

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Ineligible, Unaware, or Uninterested? Associations Between Underrepresented Patient Populations and Retention in the Pathway to Cancer Clinical Trial Enrollment

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).