Abstract
Introduction:
Individual demographic data and socioeconomic status (SES) factors from Census block group data may help define groups with disadvantaged access to clinical trials.
Methods:
Individual demographic data from the Aurora Cancer Registry and SES factors corresponding to the Census block group of the patient’s address were studied for a six-year period ending July 31, 2019.
Results:
The final study cohort included 39968 patients (enrolled = 772, and not enrolled = 39196). In univariate analysis, significantly fewer patients older than age 65 (p < 0.001) and fewer men (p < 0.001) were enrolled in clinical trials. Socioeconomic factors found to be significant during univariate analysis included: low household income (p < 0.001), percentage below the poverty line (p < 0.001), low percentage home ownership (p = 0.006), unemployment (p = 0.003), absence of a college degree (p = 0.037) and absence of a high school degree (p = 0.007). In multivariate analysis, patients older than age 65 were less likely to participate in a trial (odds ratio 0.574, p < 0.001) and men were less likely to participate (odds ratio = 0.703, p < 0.001). Only 1.4% of the variance in clinical trial participation was accounted for demographic and SES factors.
Conclusions:
The only groups with disadvantaged access to clinical trials in our institution were the elderly and men. Whether demographic or SES factors are related to accrual rates of clinical trials in other geographic regions or in other types of research studies warrants further investigation.
Keywords: Clinical Trials as a Topic, Minority Groups, Healthcare Disparities, Demography, Socioeconomic Factors
Introduction:
The National Cancer Institute (NCI) designed the NCI Clinical Oncology Research Program (NCORP) to increase the availability of clinical trials in the community and concomitantly to increase trial opportunities to underserved populations. Institutions who earn NCORP grants are judged on total accrual of patients and inclusion of minority groups. These appropriate and well-meaning efforts are hindered by the assumption that minority equates with underserved. Further confusion occurs when endpoints are conflated; a minority group may be disadvantaged for one outcome but not for all outcomes. To clarify, for outcomes such as mortality, obesity and accrual to clinical trials, a minority group could be disadvantaged for some of the outcomes but not for others. Regarding accrual to clinical trials, the situation is further hindered by the absence of a definition of adequacy of accrual. Research has been additionally hindered by poorly chosen data sets that do not include information about patients who did not participate in clinical trials.
Minority populations may be defined by demographic factors, age, sex, race and ethnicity. Socioeconomic status (SES) information about individuals is not available but income, wealth, education, employment and crowding is available for the location where a patient resides. The combination of individual-level factors with group-level factors determined by where the person lives is more descriptive than a simple analysis of demographic or socioeconomic data separately.1,2 Census block groups are contiguous areas of 600 to 3,000 people defined by the United States Census Bureau. Estimates from Census block groups are commonly used to define SES factors at the group level.3
A second issue is the definition of adequacy of accrual. We reasoned that if accrual of one group is equal to the complementary group, defined as not the first group, then neither group is underserved. If there is a difference in accrual, then the group with less accrual is underserved.
The last issue is identification of the group to be studied. In order to determine the adequacy of accrual, one must know the number of patients who participated in a trial and the number of patients who did not participate in a trial for two complementary groups (e.g., men and women, or people <65 years of age and those ≥65 years of age). There is no data set, to our knowledge, that has this information. The best information available compares accrual from one data set, commonly NCI accrual information, and another data set, commonly Surveillance, Epidemiology, and End Results (SEER) Program data.4–9 In these studies, data was commonly normalized with a third data set, Census data. For many of these studies, the dates of the NCI data were different than the SEER data. These trials appeared to include only treatment studies and no symptom and control trials.
To study this subject, six years of data in the Aurora Cancer Registry was queried for clinical trial participation by demographic and SES factors.
Methods:
Patients in the Aurora Health Care Cancer Registry with a date of first contact between August 1, 2013 to July 31, 2019 were included. For patients with more than one cancer within that six-year period, the date of the first cancer was used. For patients on more than one study, the date of the first study was used.
Demographic categories include age, sex, race and ethnicity are defined by the United States Census Bureau. The National Cancer Database requires this documentation in cancer registries. Race is defined as White, Black, Asian, Native American or Alaskan, Hawaiian or Pacific Islander. Ethnicity is defined as Hispanic or not.
Although multilevel analysis has been applied to many endpoints (e.g., BMI/obesity, mental health, pregnancy and birth outcome, cancer screening, cancer diagnosis and survival, self-rated health and more than a dozen other outcomes), there was no precedent for studying accrual based on area-based socioeconomic factors3. Krieger et al. identified practical measurements of socioeconomic factors for low birthweight and lead poisoning.10 These factors were found to influence use of Medicare benefits.11 These SES factors were:
Median household income, standardized to range from 0–1
Percentage of people below the federally defined poverty line
Median value of owner-occupied values, standardized to range from 0–1
Percentage of people aged 16 years or older in the labor force who are unemployed (and actively seeking work)
Percentage of people aged ≥ 25 years with at least 4 years of college
Percentage of people aged ≥ 25 years with less than a 12th-grade
Percentage of households containing one or more person per room
These seven factors include two measures of income, one measure of wealth, one measure of employment, two measures of education and one measure of crowding. In the absence of guidance regarding SES factors for trial participation, these factors were empirically evaluated for accrual to clinical trials.
The risk of ecologic fallacy, the erroneous inference of causal relation at the individual level based on grouped data, was considered. “Classic” ecologic fallacy occurs when both dependent and independent variable are grouped data.12 In this evaluation, that risk was minimized, an individual-level variable, accrual, was analyzed in conjunction with group-level, independent SES variables. The use of multilevel data, including individual data and group data, helps to avoid “individualistic fallacy,” the assumption that individual-level data are sufficient to explain social phenomena.12 Multilevel analysis helps to avoid risks of individual measures alone.13,14 Multilevel analysis is widely accepted in sociologic literature.1,2
Patient’s addresses were geocoded to determine a block group. Socioeconomic factors were attributed to a patient by Census block group from five-year estimates in the 2014–2018 American Community Survey. Patients with missing or poor match data regarding block group (e.g., post-office boxes, address ranges not found in the reference data or candidate collisions) were not included in second level SES factor calculations. To test the possibility that patients excluded from the analysis due to missing or poor geographic data may be different than those with good geographic data, a Chi-Square Test was used to examine group differences for clinical trial enrollment.
The Kolmogorov-Smirnov Test was used to check normality for the distribution of the socioeconomic factors and all were non-normally distributed; therefore, nonparametric tests were conducted when appropriate. Descriptive and frequency statistics were used to summarize the characteristics of the study population. A Mann-Whitney U test was used to compare the socioeconomic factors for enrolled and not enrolled patients. A Chi-Square Test was used to determine group differences between enrolled and not enrolled patients for categorical variables.
Multivariate logistic regression was used to examine factors associated with clinical trial participation. All potential predictive variables associated with the dependent variable during univariate analysis at the p ≤ 0.05 significance level were included in the model. The factors included in the model were age, sex, household income, below poverty level, owner-occupied home values, unemployment, college education and less than a 12th grade education. All variables were entered simultaneously along with the constant. A receiver operating characteristic (ROC) curve was used to evaluate the fit of the logistic regression model. A p-value of ≤ 0.05 was considered statistically significant.
A post hoc analysis was completed to describe accrual by cancer site. An unplanned subset analyses was done to assess the interplay of race and ethnicity and to assess differences in accrual for male-specific cancer, female-specific cancer and all other types of cancer. All comparisons were done with Chi-Square tests. The statistical analysis was performed using SPSS statistical software (version 25; IBM Corp., Armonk, NY).
Results:
Study population:
Initially, 41940 (817 enrolled in clinical trials and 41123 not enrolled) patients were identified. Census block group data was complete for 95.3% of these patients. Good matches for geographic data were found for 39968 patients. Poor matches were found in 1972 patients due to missing geographic data (n = 894) or poor geographic confidence levels (n = 1078). Chi-square test results indicated that there was no statistically significant relationship between study participation and poor or missing geographic data (45/1927 vs. 772/39,196, p=0.272); therefore, these patients were excluded. After excluding the patients with missing or poor geographic data, the final study cohort included 39968 patients (enrolled = 772, and not enrolled = 39196).
Univariate analysis:
Study participation by individual and group characteristics is summarized in Table 1. Patients 65 or older were less likely than patients younger than age 65 to participate in clinical trials (38.3% vs. 61.7%, p < 0.001). Men were significantly less likely to participate in clinical trials than women (37.6% vs. 62.2%, p < 0.001). There was no significant relationship between study participation for race (p = 0.260) or ethnicity (p = 0.065).
Table 1.
Characteristics of the Study Population (N = 39968)
| Total | Enrolled | Not Enrolled | ||
|---|---|---|---|---|
| Characteristic | n (%) | 772 (1.9) | 39196 (98.1) | p-value |
| Race | 0.260 | |||
| White | 35994 (90.1) | 711 (92.1) | 35283 (90.0) | |
| Black | 3141 (7.9) | 51 (6.6) | 3090 (7.9) | |
| Other | 623 (1.6) | 9 (1.2) | 641 (1.6) | |
| Missing | 210 (0.5) | 1 (0.1) | 209 (0.5) | |
| Ethnicity | 0.065 | |||
| Hispanic or Latino | 1227 (3.1) | 15 (1.9) | 1212 (3.1) | |
| Not Hispanic or Latino | 38625 (96.6) | 757 (98.1) | 37868 (96.6) | |
| Missing | 116 (0.3) | - | 116 (0.3) | |
| Sex | <0.001 | |||
| Male | 18737 (46.9) | 290 (37.6) | 18447 (47.1) | |
| Female | 21212 (53.1) | 480 (62.2) | 20732 (52.9) | |
| Missing | 19 (0.0) | 2 (0.3) | 17 (0.0) | |
| Age | <0.001 | |||
| Under 65 | 19326 (48.4) | 476 (61.7) | 18850 (48.1) | |
| 65 and Older | 20642 (51.6) | 296 (38.3) | 20346 (51.9) | |
| Household Income, median (IQR)* | 0.23 (0.13) | 0.25 (0.13) | 0.23 (0.14) | <0.001 |
| % Below Poverty Level, median (IQR) | 7.0 (10.7) | 6.3 (8.9) | 7.0 (10.7) | <0.001 |
| Owner-Occupied Values, median (IQR)* | 0.17 (0.11) | 0.18 (0.11) | 0.17 (0.11) | 0.006 |
| % Unemployed, median (IQR) | 3.3 (4.7) | 2.8 (4.2) | 3.3 (4.7) | 0.003 |
| % College Educated, median (IQR) | 26.3 (19.7) | 27.2 (22.0) | 26.3 (19.7) | 0.037 |
| % Less Than 12th Grade Education, median (IQR) | 4.7 (6.3) | 4.2 (5.5) | 4.8 (6.3) | 0.007 |
| % Crowded Household, median (IQR) | 0.0 (2.2) | 0.0 (2.1) | 0.0 (2.2) | 0.521 |
Standardized to range from 0 – 1.
The absence of a difference in accrual by race and ethnicity was unexpected. A post hoc analysis was done to study combinations of race and ethnicity (Appendix 1). Non-Hispanic White patients accounted for most of the accruals, (698 accruals, 90.4%). This was not significantly different (p=0.078) than Hispanic (15 patients, 1.9%), Non-Hispanic Black patients (51 patients, 6.6%) and other races, not Hispanic (eight accruals 1%). Comparisons of the combined groups did not reach statistical significance (p = 0.078).
A higher median household income was associated with a greater likelihood of participation in a trial (0.25 vs. 0.23, p < 0.001) and a lower percentage of patients below the poverty level was associated with trial participation (6.3 vs. 7.0, p <0.001). A higher median of owner-occupied property values was associated with a higher likelihood of trial participation (0.18 vs. 0.17, p = 0.006). A lower percentage of unemployment was associated with trial participation (2.8 vs. 3.3, p = 0.003). The percentage college educated was associated with trial participation (27.2 vs. 26.3, p = 0.037). Less than a high school degree was associated with failing to participate in clinical trials (4.2 vs 4.8, p = 0.007). Crowding as judged by more than two people per room was the same for those who participated and those who did not (0.0 vs. 0.0, p = 0.521).
Multivariate logistic regression analysis:
When holding all other variables constant, regression odds ratios were calculated (Table 2). Patients age 65 or older were less likely to participate in trials (OR = 0.57, 95% CI = 0.495 – 0.665, p < 0.001). Men were less likely than women to participate in trials (odds ratio = 0.703, 95% CI = 0.606 -- 0.815, p < 0.001). None of the other factors were found to be associated with clinical trial participation in multivariate analysis.
Table 2.
Predictors of Clinical Trial Participation
| Factors | Odds Ratio | 95% CI | p-value |
|---|---|---|---|
| Sex (Reference = Female) | 0.70 | 0.61 – 0.82 | < 0.001 |
| Age Group (Reference = Under 65) | 0.57 | 0.50 – 0.67 | < 0.001 |
| Household Income* | 1.56 | 0.48 – 5.05 | 0.459 |
| Below Poverty Level | 0.76 | 0.27 – 2.13 | 0.600 |
| Owner-Occupied Home Values* | 1.17 | 0.31 – 4.38 | 0.812 |
| Unemployment | 0.24 | 0.04 – 1.70 | 0.154 |
| College Education | 0.98 | 0.49 – 1.94 | 0.944 |
| Less Than 12th Grade Education | 0.60 | 0.15 – 2.39 | 0.469 |
Note: The dependent variable in this analysis is clinical trial participation so that 0 = Not Enrolled and 1 = Enrolled.
Standardized to range from 0 – 1.
The overall model fit of the predictors was questionable (−2 Log Likelihood = 7413.27, Nagelkerke R2 = 0.014) yet statistically significant [X2 (8) = 94.91, p < 0.001]. The Nagelkerke R2 statistic takes into consideration the sample size and this statistic is 0.014, which can be interpreted to mean that these predictors combined explain approximately 1.4% of the variance in clinical trial participation is accounted for by the model. This suggests that approximately 98.6% of the variance in clinical trial participation remains unexplained. The ROC analysis results showed the area under the curve (AUC) is 0.601 (95% confidence interval .580 - .622), p < 0.001). An AUC = 0.60 suggests that the model is poor for separating enrolled patients from not enrolled patients and the findings should be interpreted cautiously.
Post hoc analysis of accrual by cancer site:
Breast cancer had the highest proportion of accruals (39.0%) and the highest cancer site accrual rate (4.5%) (Table 3). Accrual was significantly different for cancers that occur in women (i.e., breast and gynecologic cancers), cancers that occur in men, and cancers that happen in both men and women (p < 0.001). Results displayed in Table 4. Accrual for organ sites that are in both men and women were analyzed separately. For this subgroup, results indicated that there was a significant relationship between sex and trial participation with men more likely than women to participate in clinical trials (60.4% vs. 39.1%, p = 0.005). Results in Table 5.
Table 3.
Clinical Trial Participation and Cancer Site (N = 39968)
| Total | Enrolled | Not Enrolled | ||
|---|---|---|---|---|
| All Cancer Sites | n (%) | 772 (1.9) | 39196 (98.1) | Site Accrual Rate (%) |
| Site | ||||
| Head & Neck | 1097 (2.7) | 2 (0.3) | 1095 (2.8) | 0.2 |
| Digestive System | 5622 (14.1) | 94 (12.2) | 5528 (14.1) | 1.7 |
| Respiratory / Chest | 4340 (10.9) | 94 (12.2) | 4246 (10.8) | 2.2 |
| Blood & Bone Marrow | 2229 (5.6) | 76 (9.8) | 2153 (5.5) | 3.4 |
| Bone & Connect / Soft Tissue | 505 (1.3) | 2 (0.3) | 503 (1.3) | 0.4 |
| Skin | 2087 (5.2) | 9 (1.2) | 2078 (5.3) | 0.4 |
| Breast | 6728 (16.8) | 301 (39.0) | 6427 (16.4) | 4.5 |
| Female Genital | 3115 (7.8) | 36 (4.7) | 3079 (7.9) | 1.2 |
| Male Genital | 5576 (14.0) | 64 (8.3) | 5512 (14.1) | 1.1 |
| Urinary System | 3049 (7.6) | 40 (5.2) | 3009 (7.7) | 1.3 |
| Brain & CNS | 2476 (6.2) | 28 (3.6) | 2448 (6.2) | 1.1 |
| Endocrine | 1153 (2.9) | 3 (0.4) | 1150 (2.9) | 0.3 |
| Lymphatic System | 1557 (3.9) | 22 (2.8) | 1535 (3.9) | 1.4 |
| Unknown Primary | 388 (1.0) | 1 (0.1) | 387 (1.0) | 0.3 |
| Other / Ill-Defined | 46 (0.1) | - | 46 (0.1) | 0.0 |
Table 4.
Clinical Trial Participation and Type of Cancer Trial (N = 39968)
| Total | Enrolled | Not Enrolled | ||
|---|---|---|---|---|
| Characteristic | n (%) | 772 (1.9) | 39196 (98.1) | p-value |
| Type of Trial | <0.001 | |||
| Male-Specific Cancer | 5576 (14.0) | 64 (8.3) | 5512 (14.1) | |
| Female-Specific Cancer | 9843 (24.6) | 337 (43.7) | 9506 (24.3) | |
| All Other Types of Cancer | 24549 (61.4) | 371 (48.1) | 24178 (61.7) |
Table 5.
Clinical Trial Participation and Sex for All Non-Sex-Related Types of Cancer (N = 24549)
| Total | Enrolled | Not Enrolled | ||
|---|---|---|---|---|
| Characteristic | n (%) | 371 (1.5) | 24178 (98.5) | p-value |
| Sex | 0.005 | |||
| Male | 13100 (53.4) | 224 (60.4) | 12876 (53.3) | |
| Female | 11433 (46.6) | 145 (39.1) | 11288 (46.7) | |
| Missing | 16 (0.1) | 2 (0.5) | 14 (0.1) |
Discussion
Older age was consistent with poor accrual by univariate and multivariate analysis. These findings are consistent with every other study published.4–6,15 Cancer is a disease of the elderly16 and this suggests that there are opportunities for geriatric studies and studies with entry criteria designed to include the elderly at our institution. This may be a general opportunity applicable to other institutions.
Men were less likely to participate in clinical trials by univariate and multivariate analysis. The differences may be due to greater accrual to breast and gynecologic trials for women than for prostate cancer in men. Evaluation of the accrual to clinical trials for cancers found in both sexes demonstrated higher accrual for men than for women implying that men are at least as likely to participate in a trial if available. These findings are in contrast to other evaluations that found trial participation to be approximately equal between sexes.4,7 This information suggests that there may be opportunities for accrual of men at our institution if we had the studies to offer.
Race was not significantly associated with accrual in univariate analysis and did not qualify for multivariate analysis. Information in the literature about race and accrual are mixed. Black patient participation in clinical trials was found to be comparable to white participation in some evaluations5,6 while others found Black patients to be underrepresented.7–9 Asian patients were found to be adequately represented in one analysis8 and underrepresented in another.7 The accrual ratio was significantly better for Alaskans and Native Americans and for Hawaiians and Pacific Islanders in one analysis. The results were statistically significant, but the numbers were small.8
Ethnicity came close but did not reach statistical significance in univariate analysis and did not qualify for multivariate analysis. Hispanic ethnicity was associated with less accrual in some analyses7,8 but not in another.6 A literature search suggested that when offered the opportunity to participate in research, Black and Hispanic people, participate in trials at the same ratio as white people.17
Socioeconomic factors determined by Census block group were significant in univariate analysis for all but people per room (crowding). None were significant in multivariate analysis. Other investigators noted accrual to be better in areas with a higher socioeconomic profile.7 Unger was able to collect individual-level socioeconomic data, income and education, by use of an internet-based treatment decision tool. Lower-income patients were less likely to participate in clinical trials.18,19
In this dataset the relationship between demographic and SES factors was small, estimated to account for only 1.4% of the variance in clinical trial participation in our model. This implies that although availing trials to all socioeconomic groups is important, approximately 98.6% of the variance in trial participation may be attributed to other factors.
Speculation about other factors relevant to accrual is part of a larger literature on barriers to trial participation. These barriers are generally categorized as patient/community, physician/provider and site/organizational influences.20–25 Lack of available trials and ineligibility for a trial probably account for three quarters of the explanation for failure to participate and physician/patient factors may be very important in the remaining quarter.26 Structural, infrastructural and procedural barriers to opening oncology clinical trials may be important.27
Rigorous testing of these theorized barriers is not commonly available.22 Few rigorously conducted studies have tested interventions to address challenges to clinical trials accrual.20 There are many outcomes that can be assessed by sex, age, race, ethnicity and area-based SES factors. Our data suggests that for the outcome, clinical trials participation, that only sex and age are important at our institution. This may be true nationally and perhaps efforts to accrue men and patients older than 65 may slightly increase overall accrual.
The weakness of our investigation is that we only had 772 unique patients enrolled in clinical trials at a single institution. The widespread applicability of these findings requires analysis of other data sets at other institutions or, ideally, the combination of multiple data sets. The strengths of our analysis is that the data for those who participated and those who did not participate came from the same data set.
In summary, groups can be identified based on demographic and SES census block group data. Underserved groups can be identified by comparing complementary groups such as men and women or age less than 65 and age 65 or order. This analysis requires a data set that has the number of patients accrued and the number of patients not accrued. Finally, our data suggested that the effect of demographic and SES factors on total accrual may be small. This finding is provocative and inclusion of all demographic and SES groups is important. This observation will require further investigation in other geographic regions and in other data sets.
Appendix 1.
Clinical Trial Participation and Composite Race/Ethnicity (N = 39968)
| Total | Enrolled | Not Enrolled | ||
|---|---|---|---|---|
| Characteristic | n (%) | 772 (1.9) | 39196 (98.1) | p-value |
| Race/Ethnicity | 0.078 | |||
| Non-Hispanic White | 34775 (87.0) | 698 (90.4) | 34077 (86.9) | |
| Hispanic or Latino | 1227 (3.1) | 15 (1.9) | 1212 (3.1) | |
| Non-Hispanic Black | 3117 (7.8) | 51 (6.6) | 3066 (7.8) | |
| Other Non-Hispanic Race | 600 (1.5) | 8 (1.0) | 592 (1.5) | |
| Missing | 249 (0.6) | - | 249 (0.6) |
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Competing Interests:
None of the authors have any competing interests to declare.
Contributor Information
Thomas Saphner, Aurora NCORP, Advocate Aurora Health, 960 N 12th St., Milwaukee, Wisconsin 53233.
Lisa Robinson, Aurora Clinical Data Registries, Advocate Aurora Health, 960 N 12th St., Milwaukee, Wisconsin 53233.
Neha Glandt, Aurora NCORP, Advocate Aurora Research Institute, 960 N 12th St., Milwaukee, Wisconsin 53233.
References:
- 1.Diez-Roux AV. Multilevel analysis in public health research. Annu Rev Public Health. 2000;21:171–192. [DOI] [PubMed] [Google Scholar]
- 2.Zahnd WE, McLafferty SL. Contextual effects and cancer outcomes in the United States: a systematic review of characteristics in multilevel analyses. Ann Epidemiol. 2017;27(11):739–748 e733. [DOI] [PubMed] [Google Scholar]
- 3.Arcaya MC, Tucker-Seeley RD, Kim R, Schnake-Mahl A, So M, Subramanian SV. Research on neighborhood effects on health in the United States: A systematic review of study characteristics. Soc Sci Med. 2016;168:16–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hutchins LF, Unger JM, Crowley JJ, Coltman CA Jr., Albain KS. Underrepresentation of patients 65 years of age or older in cancer-treatment trials. N Engl J Med. 1999;341(27):2061–2067. [DOI] [PubMed] [Google Scholar]
- 5.Lewis JH, Kilgore ML, Goldman DP, et al. Participation of patients 65 years of age or older in cancer clinical trials. J Clin Oncol. 2003;21(7):1383–1389. [DOI] [PubMed] [Google Scholar]
- 6.Tejeda HA, Green SB, Trimble EL, et al. Representation of African-Americans, Hispanics, and whites in National Cancer Institute cancer treatment trials. J Natl Cancer Inst. 1996;88(12):812–816. [DOI] [PubMed] [Google Scholar]
- 7.Sateren WB, Trimble EL, Abrams J, et al. How sociodemographics, presence of oncology specialists, and hospital cancer programs affect accrual to cancer treatment trials. J Clin Oncol. 2002;20(8):2109–2117. [DOI] [PubMed] [Google Scholar]
- 8.Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race-, sex-, and age-based disparities. JAMA. 2004;291(22):2720–2726. [DOI] [PubMed] [Google Scholar]
- 9.Pang HH, Wang X, Stinchcombe TE, et al. Enrollment Trends and Disparity Among Patients With Lung Cancer in National Clinical Trials, 1990 to 2012. J Clin Oncol. 2016;34(33):3992–3999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, Carson R. Choosing area based socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning: The Public Health Disparities Geocoding Project (US). J Epidemiol Community Health. 2003;57(3):186–199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bonito AJ BC, Eicheldinger C, Carpenter L. Creation of New Race-Ethnicity Codes and Socioeconomic Status (SES) Indicators for Medicare Beneficiaries. Final Report, Sub-Task 2. (Prepared by RTI International for the Centers for Medicare and Medicaid Services through an interagency agreement with the Agency for Healthcare Research and Policy, under Contract No. 500-00-0024, Task No. 21). 2008. [Google Scholar]
- 12.Krieger N Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. American Journal of Public Health. 1992;82:703–710. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Soobader M, LeClere FB, Hadden W, Maury B. Using aggregate geographic data to proxy individual socioeconomic status: does size matter? American journal of public health. 2001;91:632–636. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Subramanian SV, Jones K, Kaddour A, Krieger N. Revisiting Robinson: the perils of individualistic and ecologic fallacy. Int J Epidemiol. 2009;38:342–373. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Freedman RA, Foster JC, Seisler DK, et al. Accrual of Older Patients With Breast Cancer to Alliance Systemic Therapy Trials Over Time: Protocol A151527. J Clin Oncol. 2017;35(4):421–431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.U.S. Department of Health and Human Services CfDCaPaNCI. U.S. Cancer Statistics Working Group. U.S. Cancer Statistics Data Visualizations Tool, based on 2019 submission data (1999–2017). 2020.
- 17.Wendler D, Kington R, Madans J, et al. Are racial and ethnic minorities less willing to participate in health research? PLoS Med. 2006;3(2):e19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Unger JM, Hershman DL, Albain KS, et al. Patient income level and cancer clinical trial participation. J Clin Oncol. 2013;31(5):536–542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Unger JM, Gralow JR, Albain KS, Ramsey SD, Hershman DL. Patient Income Level and Cancer Clinical Trial Participation: A Prospective Survey Study. JAMA Oncol. 2016;2(1):137–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Denicoff AM, McCaskill-Stevens W, Grubbs SS, et al. The National Cancer Institute-American Society of Clinical Oncology Cancer Trial Accrual Symposium: summary and recommendations. J Oncol Pract. 2013;9(6):267–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hillyer GC, Beauchemin M, Hershman DL, et al. Discordant attitudes and beliefs about cancer clinical trial participation between physicians, research staff, and cancer patients. Clin Trials. 2020;17(2):184–194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Heller C, Balls-Berry JE, Nery JD, et al. Strategies addressing barriers to clinical trial enrollment of underrepresented populations: a systematic review. Contemp Clin Trials. 2014;39(2):169–182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Denson AC, Mahipal A. Participation of the elderly population in clinical trials: barriers and solutions. Cancer Control. 2014;21(3):209–214. [DOI] [PubMed] [Google Scholar]
- 24.Comis RL, Miller JD, Aldige CR, Krebs L, Stoval E. Public attitudes toward participation in cancer clinical trials. J Clin Oncol. 2003;21(5):830–835. [DOI] [PubMed] [Google Scholar]
- 25.Lara PN Jr., Higdon R, Lim N, et al. Prospective evaluation of cancer clinical trial accrual patterns: identifying potential barriers to enrollment. J Clin Oncol. 2001;19(6):1728–1733. [DOI] [PubMed] [Google Scholar]
- 26.Unger JM, Vaidya R, Hershman DL, Minasian LM, Fleury ME. Systematic Review and Meta-Analysis of the Magnitude of Structural, Clinical, and Physician and Patient Barriers to Cancer Clinical Trial Participation. J Natl Cancer Inst. 2019;111(3):245–255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dilts DM, Sandler AB. Invisible barriers to clinical trials: the impact of structural, infrastructural, and procedural barriers to opening oncology clinical trials. J Clin Oncol. 2006;24(28):4545–4552. [DOI] [PubMed] [Google Scholar]