State of Geographic Access to Cancer Treatment Trials in the United States: Are Studies Located Where Patients Live?

Abstract

PURPOSE

In this study, we describe the geographic distribution of US cancer treatment trials to identify disparities and opportunities for targeted improvements in access to research for people with cancer.

METHODS

US-based phase I-III cancer treatment trials registered on ClinicalTrials.gov were tabulated for the years they were open to enrollment (2017-2022), overall and by county, and supplemented with data from the US Census Bureau, National Cancer Institute, Centers for Disease Control and Prevention, and US Department of Agriculture. We evaluated geographic differences in trial availability. We assessed 5-year trends in trials per capita and mapped 1-hour drive time areas around sites.

RESULTS

A total of 6,710 trials were open to enrollment in 2022 across 1,836 sites. Trials increased by 4%, whereas sites decreased by 3% annually per capita from 2017. Seventy percent of US counties had no reported active trials in 2022 (2,211/3,143), representing 19% of people age ≥55 years. Eighty-six percent of nonmetropolitan counties had no trials versus 44% of metropolitan counties. Trial availability varied by county-level cancer mortality and social vulnerability (an index derived from demographic and socioeconomic data from the US Census). Eighteen percent of counties without trials had oncologist care sites (n = 618). Notably, 26% of people age ≥55 years lived beyond an hour drive of a site with ≥100 trials.

CONCLUSION

Most US counties have limited to no trial offerings, a disparity magnified in counties that are nonmetropolitan, with high social vulnerability, and with high cancer mortality. Effort to facilitate diverse site participation is needed to promote equitable access to trials and to ensure patients participating in trials match the characteristics of patients who will receive interventions once approved. Counties with oncology care sites but no trials provide potential expansion areas.

INTRODUCTION

Cancer clinical trials are critical to developing new standard-of-care treatment options and improving patient outcomes. Yet, most patients are not able to participate in trials, and those who do participate are unreflective of the population living with cancer.^1-3 Where trials are located across the United States likely has a large impact on patient access and participation in research.^4,5

CONTEXT

• Key Objective

• To examine whether cancer trials are located where US residents live (in general and by demographic, epidemiologic, and social features).

• Knowledge Generated

• Many US residents live geographically removed from cancer trials. 48% of the US population age ≥55 years lived in counties with broad trial offerings (≥100 active cancer trials), 33% lived in counties with limited offerings (1-99 trials), 10% lived in counties with oncology care sites but no trials, and the remaining 9% lived in counties with neither care nor trial sites. Trial availability was even more limited in counties that were nonmetropolitan, had high cancer mortality, and/or had high social vulnerability.

• Relevance

• Spatially orienting trial and oncology care sites alongside population characteristics enables systemic prioritization and tailoring of interventions to strengthen and expand research infrastructure. Care sites located in counties with limited or no trial offerings are potential places to expand reach and diversity of clinical trial participants.

Many patients with cancer live geographically removed from oncologists, suggesting that access to research staff and infrastructure is similarly uneven. A prior State of Cancer Care in America study found that 54% of US counties had no hematology or oncology specialist (affecting 11% of the US population).⁶ Moreover, a systematic review including 8,883 patients with cancer found that nearly 55% did not have access to a relevant trial at their care site.⁷ In another study, more than 50% of patients with prostate or colorectal cancers, 46% of patients with metastatic breast cancer, and 38% of patients with non–small cell lung cancer had to drive more than an hour to reach a trial site.⁸ Increased travel time is associated with a higher risk of presenting with advanced cancer and lower enrollment in trials.⁸ Recent findings also indicate socioeconomic and racial disparities in commuting distance to trial sites.^4,5,9

ASCO routinely uses geospatial analysis techniques with publicly accessible data sources to identify oncology clinicians and practices relative to population-based indicators of need, such as cancer diagnoses, social vulnerability, and rurality.^6,10 In this study, we leveraged ASCO's infrastructure and data available from ClinicalTrials.gov to describe the national landscape of cancer treatment trials and examine the demographic and social features of US geographies by trial availability. The primary objective was to evaluate the association of number of trials in geographic regions with characteristics of the regions, including population, land area, and rurality. Analyses were conducted at various regional levels (ie, county, state, and nation) and by drive-time to trial sites. Secondary objectives evaluated for relationships between availability of a trial site with the social vulnerability, cancer incidence, cancer mortality, and oncologist presence within that geography, as well as trial characteristics and 5-year trends in number of trials per capita.

METHODS

Data Sources

ClinicalTrials.gov, an international registry of clinical trials maintained by the US National Library of Medicine, was used to identify cancer studies of interest.¹¹ Representation of US treatment trials in ClinicalTrials.gov is robust and growing because of expanding regulatory, funding, publication, and other sponsor requirements for registration.^12-15

Trial data were supplemented with geographically referenced demographic, epidemiologic, social determinants of health, and provider data from other governmental sources: population and land area information from the US Census Bureau,¹⁶ Rural Urban Continuum Codes (RUCC) maintained by the US Department of Agriculture,¹⁷ age-adjusted cancer incidence and cancer mortality rates published by the National Cancer Institute and the Centers for Disease Control and Prevention (CDC),¹⁸ provider data from US Centers for Medicare & Medicaid Services,¹⁹ and overall Social Vulnerability Index (SVI) scores from the CDC.^17-20 SVI scores are derived from 16 US Census Bureau American Community Survey (ACS) variables addressing area-level socioeconomic status, household characteristics, racial/ethnic minority status, housing type, and transportation.²⁰

Number of Trials and Temporal Trends

Data on registered cancer studies with locations in the United States (territories excluded) were extracted from ClinicalTrials.gov in JavaScript Object Notation format on July 20, 2023.¹¹ Synonyms for cancer used in the ClinicalTrials.gov search function are included in Figure 1. Using R statistical software (v 4.2.2), we selected for analysis phase I-III interventional treatment trials to reflect studies with potential to expand US Food and Drug Administration-approved indications. We excluded pediatric-only trials as pediatric patients are largely treated at specialized facilities. Variables of interest included title, sponsor type, enrollment open and closure dates, sex/age eligibility criteria, and geographic coordinates for sites listed by the study sponsor.²¹ Trials were then grouped by year from 2017 to 2022 according to whether they were open at any point in the respective year, allowing representation in multiple years.

FIG 1.

Process for selecting active oncology trials. Data source: ClinicalTrials.gov.^{8 a}Includes ClinicalTrials.gov synonyms: cancer; neoplasms, tumors, tumor, neoplasm, malignancies, cancers, oncology, malignancy, tumors, tumor, neoplasia, malignant neoplasm, malignant tumors, malignant tumor, malignant neoplasms, primary cancer, neoplastic disease, cancer NOS, neoplasias, neoplastic syndrome, primary malignant neoplasm, tumor NOS, malignant neoplastic disease, malignant tumors, malignant tumor, malignant neoplasm primary. FDA, US Food and Drug Administration; NOS, not otherwise specified.

Trials open in each of years 2017-2022 were tabulated and combined with American Community Survey (ACS) 5-year population estimates available in the trial year (eg, 2022 trial data appended with 2016-2020 population data).¹⁶ Poisson regression was performed to estimate temporal changes in the number of trials, offset by age ≥55 years population data. This offset was chosen because 80% of new cancer cases arise in the age ≥55 years population.²²

County Trial Status and Auxiliary Data

Site coordinates associated with trials active in 2022 were mapped using ArcGIS Pro 3.2.1 and assigned a US county through spatial join analysis. Trial county information was returned to R, each trial was counted once per county (including representation in ≥1 counties), and county-level tabulations were produced. All 3,143 US counties were then categorized as having no trials, a limited portfolio of 1-99 active trials, or a broad portfolio of ≥100 active trials. We set the threshold of ≥100 trials to represent diversity by cancer type, stage, and treatment modality and improved the likelihood of patients being eligible for open trials in their county. To account for variation in navigability of counties, we also created distinctions using county adjacency information, specifically whether a county had a broad or limited portfolio or no trials within its own borders or within a neighboring county.²³

Counties were characterized as metropolitan and nonmetropolitan according to RUCC scores. For SVI scores, cancer incidence, and cancer mortality, values were divided into quartiles across all US counties with nonmissing data; fourth quartile (Q4) counties were considered highly vulnerable and first-third quartile (Q1-Q3) counties were not. The number of oncologists treating adults per county was tabulated, consistent with prior ASCO work, and counties were split into those with an oncology care site versus not based on presence of ≥1 oncologist.^6,10

Counties by demographic and vulnerability feature were compared by degree of trial availability using chi-square tests. Statistical significance was defined by a two-sided alpha level of .05.

Drive-Time Analysis

One-hour drive-time areas were generated around each mapped 2022 trial site using ArcGIS Pro network analysis tools. The one-way threshold of an hour was set on the basis of oncology care and trial literature.^8,24 We spatially overlaid these areas with 5-year ACS estimates for people age ≥55 years by US Census tract and used ArcGIS Pro apportionment analysis to approximate the share of population residing within the 1-hour drive-time areas. The same process was applied to the subset of sites with ≥100 active trials to approximate the share of people age ≥55 years living within an hour drive of a broad portfolio site.

RESULTS

Characteristics of Oncology Treatment Trials

Of nearly 44,000 US cancer studies registered on ClinicalTrials.gov as of July 2023, 6,710 phase I-III treatment trials met the criteria for inclusion in this analysis and were open to enrollment in 2022 (Fig 1).

Thirty percent of trials active in 2022 were phase I (n = 2,005; Table 1), with a median of two reported sites per study (IQR, 1-7). A total of 3,906 trials (58%) were either phase I/II or phase II with a median of two sites per study (IQR, 1-3) and 799 (12%) were phase II/III or phase III with a median of 22 sites per study (IQR, 8-48). Overall, most trials were sponsored either by academic/hospital/health systems (n = 2,858, 43%) or by industry (n = 2,682, 40%). Another 604 trials were sponsored by the US Government (9%) and the remaining 566 trials (8%) by other groups and individuals (eg, foundations, international governmental entities, and investigators). Nine in 10 trials were open to adult participants regardless of age category; similarly, 90% of trials were offered independent of participant sex.

TABLE 1.

Oncology Trials Characteristics (2022)

Characteristic	No. (%)
Total	6,710 (100)
Phase
1	2,005 (30)
I/II	1,203 (18)
II	2,703 (40)
II/III	85 (1)
III	714 (11)
Sponsor
Academic/hospital/health systema	2,858 (43)
Governmentb	604 (9)
Industry	2,682 (40)
Otherc	566 (8)
Participant age, years
Adult (age 18-64 years)	359 (5)
Adult and older adult (age ≥18 years)	6,321 (94)
Older adult (age ≥65 years)	30 (0.4)
Participant sexd
All	5,955 (89)
Female only	421 (6)
Male only	334 (5)

NOTE. Data source: ClinicalTrials.gov.¹¹

^a Extracted from other category based on keyword search and manual review of sponsor organization names.

^b US only; consolidated from sponsor type NIH (n = 430), network (n = 163), and federal (n = 11).

^c Four hundred seventy-seven entries were listed as individuals' names (84%); the remainder were nongovernmental networks, foundations, and international governmental entities.

^d Classification as male or female based on biological distinctions.²¹

The number of active oncology trials increased each year from 4,791 active trials in 2017 to 6,710 in 2022 (Table 2). The number of trial sites was the highest in 2017 (n = 1,847), the lowest in 2020 (n = 1,804), and increased to 1,836 by 2022. When offset by counts of people age ≥55 years, trials increased every year, with a 4.4% average annual increase over 5 years (P < .01). The number of sites per population concurrently decreased by an average of 2.5% per year (P < .01).

TABLE 2.

No. of US Oncology Trials, 2017-2022

Year	No. of Trials	No. of Reported Sites	People Age ≥55 Years	Trials/100,000 People Age ≥55 Years
2017	4,791	1,847	84,033,105	5.7
2018	5,188	1,845	86,242,374	6.0
2019	5,520	1,808	88,479,909	6.2
2020	5,824	1,804	90,525,312	6.4
2021	6,320	1,823	92,540,210	6.8
2022	6,710	1,836	94,464,256	7.1

NOTE. Data Sources: ClinicalTrials.gov and the US Census Bureau's American Community Survey.^11,16

County Characteristics by Number of Trials

Seventy percent of counties had no reported active trials in 2022 (n = 2,211 of all 3,143 counties), representing 74% of the US land area and home to 19% of those age ≥55 years. A total of 1,593 counties (51%) representing 9% of the population of interest had no trials or active oncologist care sites while 618 counties representing 10% of the population had care sites but no trials (Fig 2).

FIG 2.

Counties by presence of cancer trials versus oncologist care sites (2022). Data sources: ClinicalTrials.gov and data.Medicare.gov.^11,19 Presence of trials means at least one cancer phase I-III trial open in 2022; presence of oncologist care site means at least one hematologist, medical oncologist, hematologist/oncologist, gynecologic oncologist, surgical oncologist, or radiation oncologist with a practice location and billing Medicare in 2022.

Thirty percent of US counties (n = 932) had active trials with a median of 49 trials per county (IQR, 14-110). A total of 276 counties had a broad portfolio of ≥100 trials (9% of all US counties). These broad portfolio counties represent where 48% of the age ≥55 years population resided (Fig 2). Another 33% of the population resided in counties with limited portfolios. The distribution of trial availability varied widely by county-level demographic and vulnerability measures, with all examined features being statistically associated with whether trials were geographically available (Fig 3A).

FIG 3.

(A) County characteristics by number of trials. Data sources: ClinicalTrials.gov, Rural Urban Continuum Codes, Social Vulnerability Index, and State Cancer Profiles.^11,17-20 (B) County characteristics by number of proximal trials (incorporating county adjacency). Data sources: ClinicalTrials.gov, Rural Urban Continuum Codes, Social Vulnerability Index, and State Cancer Profiles.^11,17-20 Q1-Q3 = first-third quartiles, Q4, fourth quartile. ^aNonmetropolitan, Q4 Social Vulnerability Index, Q4 Cancer Incidence, and Q4 Cancer Mortality counties defined as high vulnerability counties. ^bP < 0.01 in chi-square analysis.

Twenty percent of metropolitan counties had broad trial portfolios, whereas one in 50 of nonmetropolitan counties had broad trial portfolios. Eighty-six percent of nonmetropolitan counties had no trials versus 44% of metropolitan counties.

We observed another marked difference among counties by cancer mortality. Specifically, 85% of counties with the highest mortality had no trials compared with 65% of counties with lower cancer mortality (Q1-Q3 mortality counties). Two percent of highest mortality counties and 11% of lower mortality counties had broad trial portfolios. Conversely, counties with high cancer incidence rates were more likely to have broad trial portfolios (13% v 8%) and more likely to have at least one trial (46% v 24%) as compared with Q1-Q3 cancer incidence counties.

Among counties with more than one type of vulnerability (combined nonmetropolitan, high SVI, high cancer incidence, and/or high cancer mortality), 18% had at least one trial. By contrast, 30% of counties with exactly one high vulnerability category and 53% of counties with no high vulnerability designation had at least one trial (Fig 3A).

Factoring in county adjacency, 678 counties (22%) had no open trials within the county or an adjacent county, representing 3% of the age ≥55 years population and 27% of land area. Meanwhile, 1,168 counties (37%) had a broad portfolio of open trials within or in an adjacent county and were home to 77% of the age ≥55 years population. The remaining 20% of the age ≥55 years population were distributed among the 1,297 counties with limited trial availability within or in an adjacent county. Counties stratified by demographic and vulnerability category had significant and meaningful differences by trial status factoring in adjacency (Fig 3B). For example, counties with multiple high vulnerability categories were 4 times more likely to have no proximal trials (33% vs. 8%) and nearly a third as likely to have broad portfolio trial sites in proximity (20% v 55%) as compared with counties with no evaluated high vulnerability category.

Populations Within an Hour Drive of a Trial

Approximately 89 million residents age ≥55 years lived within an hour drive of an oncology trial site (94% of the 95 million age ≥55 years US population; Fig 4A). Approximately 70 million residents age ≥55 years lived within an hour drive of a site with a broad oncology trial portfolio (74% of the population). Figure 4B displays the distribution of broad-portfolio sites against tract-level age ≥55 years population.

FIG 4.

(A) Areas within an hour drive of a limited or broad portfolio trial site.^a Data sources: ClinicalTrials.gov and the US Census Bureau's American Community Survey.^{11,16 a}Limited portfolio site defined as having 1-99 active oncology trials in 2022 (n = 1,523 sites). Broad portfolio site defined as having ≥100 active oncology trials in 2022 (n = 313 sites). (B) Areas within an hour drive of a broad portfolio trial site^b versus people age ≥55 years by census tract. Data sources: ClinicalTrials.gov and the US Census Bureau's American Community Survey.^{11,16 b}Broad portfolio site defined as having ≥100 active oncology trials in 2022 (n = 313 sites). (C) Percentage of people age ≥55 years who live within an hour drive of a broad portfolio trial site^c by state. Data sources: ClinicalTrials.gov and the US Census Bureau's American Community Survey.^{11,16 c}Broad portfolio site defined as having ≥100 active oncology trials (n = 313 sites). Colors determined by data quintiles.

In 31 states, an estimated 90% or more of age ≥55 years residents lived within an hour drive of a trial site (Fig 4C, Appendix Table A1, online only). In Alaska and Wyoming, a minority of age ≥55 years residents lived within an hour of a trial site (47% and 44%, respectively). When considering only broad-portfolio sites, Connecticut, District of Columbia, Illinois, Michigan, New Jersey, Rhode Island, and Wisconsin had trials situated within an hour of ≥90% of their age ≥55 years population, whereas 12 states, Alabama, Alaska, Arkansas, Louisiana, Maine, Mississippi, New Hampshire, New Mexico, North Dakota, South Dakota, West Virginia, Wyoming, had a majority of residents living beyond an hour drive of a broad-portfolio site (Fig 3B, Appendix Table A1).

DISCUSSION

This study found disparities in geographic access to cancer treatment trials across the United States, underscoring the critical need to bring clinical trials closer to patients.²⁵ The results revealed that more than 70% of US counties lacked a single active cancer treatment trial in 2022. Less than 9% of counties had a broad portfolio of ≥100 trials, a threshold intended to represent a diversity of trial offerings by cancer type, stage, and treatment modality. Access varied widely at the state and county levels. These findings are troubling because ASCO and many other leading cancer organizations have identified access to clinical trials as an essential part of high-quality care.^26-29

To our knowledge, our study is the first to systematically examine trial locations in relation to oncologist care sites, thus quantifying a driver of geographic disparities in trial offerings and identifying geographic areas with potential for increased trial participation. Counties with no cancer care or research sites (see white in Fig 2) have limited short-term potential to expand research opportunities given the lack of oncologists in the area. By contrast, counties with care sites that currently have limited or no research infrastructure (highlighted in orange and light purple in Fig 2) represent places with the potential to expand research because of the presence of oncologists. Strategies to reduce existing barriers to running trials will be needed to engage these clinicians in research, however, such as addressing complex trial designs, regulatory burdens, resource needs, and training and support of providers.

ASCO and other stakeholders are working to address many of these barriers by encouraging broadened eligibility criteria in trial design,^30,31 using pragmatic and decentralized trial elements,³² and enabling research participation in home and community settings via telemedicine^33,34 and remote patient monitoring.³⁵ ASCO has also led initiatives aimed at streamlining elements of clinical trials engagement for sites including clinical trial feasibility assessments, workload assessments, contract negotiations, and centralizing coverage analyses.^36-38 Efforts like the NCI Community Oncology Research Program have expanded clinical trials in community settings and alleviated some financial barriers to site trial engagement,³⁹ but broad reform is needed to make clinical research a viable option for sites with fewer financial resources. As it stands, small practices with limited accrual opportunities do not attract industry attention and cannot receive sufficient payment from NCI trials to offset the substantial cost of conducting research. Indeed, the counties with broad research portfolios (highlighted in dark purple in Fig 2) cluster around cancer centers and other major multispecialty oncology groups.

Sites with broad research portfolios, especially those serving vulnerable populations, may have best practices to share with practices currently offering limited or no trials. Nevertheless, even these sites with broad portfolios may not be enrolling representative patient populations. Using the methodology in this study, ASCO now has the ability to systematically identify research and practice sites in underserved areas to gather information on barriers to research participation, to develop tools, and to share resources in a tailored fashion.

Critical to understanding geographic access to research is evaluating trial availability and drive times for populations historically excluded from research. We found that more than 50% of adults age ≥55 years, the population most likely to receive cancer diagnoses,²² lived in counties with limited or no oncology trials and 26% lived more than an hour from a site with a broad portfolio of ≥100 trials. For disadvantaged populations,^40,41 we found that access to trials is most limited in counties that are nonmetropolitan, have high social vulnerability, and have high cancer mortality. Specifically, one of five metro counties had a broad portfolio of trials versus one of 50 nonmetropolitan counties. Similarly, 82% of counties with more than one measure of social vulnerability had no trials, and 85% of counties with the highest cancer mortality had no trials. These results are consistent with previous research finding poorer geographic access and participation in trials among patients from rural and less affluent counties.^42,43 Other research has found that patients with metastatic cancer, a population most likely to benefit from new agents and treatment approaches tested in clinical trials, routinely are required to drive over an hour to access clinical trials.⁸

In contrast to the other dimensions of vulnerability, counties with increased demand for trials because of high cancer incidence rates tended to have greater trial availability. This is consistent with findings that metropolitan counties have higher cancer incidence than nonmetropolitan ones (yet lower mortality).⁴⁴ However, incidence rates of smoking-related cancers and cancers preventable through screening are higher in nonmetropolitan versus metropolitan areas. While this study focused on characterizing the overall trial landscape, future analyses may explore nuances based on cancer type, investigational modality, and additional area-level social determinates such as race and ethnicity.

Furthermore, this study found that the geographic distribution of trials is not improving over time. The number of active oncology trials slightly increased between 2017 and 2022 at an annual rate of 4% when adjusting of population levels. However, the number of sites per population decreased on average by 3% per year. Thus, although there are more trials opening every year, they are conducted at fewer sites. This finding may be mitigated if the research community is able to engage more oncology sites in trials.

Conducting research with ClinicalTrials.gov data has notable strengths but also limitations. Despite registration requirements, some trials inevitably still fail to register.⁴⁵ Of particular importance to this study is accuracy of individual trial locations. ClinicalTrials.gov registration requires that sponsors list individual sites where a participant may be enrolled. Because data entry is manual, sponsors may be incentivized to group sites at the highest organizational level, rather than provide granular information.⁴⁶ Little research exists about the completeness of individual sites reported to ClinicalTrials.gov. Previous analyses have found data quality discrepancies in other ClinicalTrials.gov data fields across select trials when compared with other trial documents (eg, regulatory review, publication).⁴⁷ Under-reporting of sites could bias our findings toward lower trial counts at the county level. Because our approach included all phase I-III cancer trials across 5 years and employed broad categories of availability, however, we think the data support a full picture of available research sites.

Additional limitations to this research include the 100-trial threshold used to define a broad portfolio which was based on expert consensus, not validated on types of cancer trials included or patient-level participation outcomes. Cancer incidence data are unavailable for Kansas, Minnesota, Nevada, and Virginia in the national data set. We conducted analysis of trial status comparisons with and without these states for the other vulnerability dimensions and found identical conclusions, limiting the impact of the missing data. While county as the geographic unit of interest is an imperfect proxy for access, this was somewhat mitigated by including county adjacency and drive-times as additional measures of geographic access, acknowledging that drive-time assumes availability of an automobile or driver.

Ultimately, regulators have identified enhancing diversity of trial participants by race/ethnicity, age, sexual and gender minorities, and other sociodemographic characteristics as a key regulatory goal.^48-50 Expanding geographic access to trials is one strategy for achieving this. Moving forward, additional research is needed on the importance of geographic access to patient participation in research in different contexts (eg, disease, stage, access to transportation). It is also critical that the research community meets its ethical obligation to improve accuracy of reporting in ClinicalTrials.gov, which could have the dual benefit of strengthening research on trial access and providing patients, families, and clinicians with comprehensive lists of available trials. Altogether the oncology community should prioritize improving geographic access to research and ensuring all patients with cancer can participate in trials over the course of their care.

APPENDIX

TABLE A1.

Trials, 5-Year Trend, and Population within an Hour Drive of a Trial Site by State

State	No. of Trials	No. of People Age ≥55 Years	Trials/100,000 People Age ≥55 Years	5-Year Trenda	No. of People Age ≥55 Years Who Live Within an Hour Drive From a Trial Site	Percentage of People Age ≥55 Years Who Live Within an Hour Drive From a Trial Site	No. of People Age ≥55 Years Who Live Within an Hour Drive From a Broad Portfolio Site	Percentage of People Age ≥55 Years Who Live Within an Hour Drive From a Broad Portfolio Site
Texas	2,835	6,808,352	42	Increasing	6,370,578	94	4,425,397	65
California	2,635	10,384,172	25	Increasing	9,935,306	96	8,391,491	81
New York	2,549	5,806,129	44	Increasing	5,634,724	97	5,033,273	87
Florida	1,876	7,187,308	26	Increasing	7,004,167	97	5,730,255	80
Pennsylvania	1,654	4,141,819	40	Increasing	4,077,794	98	3,191,973	77
Ohio	1,553	3,596,039	43	Increasing	3,581,942	100	3,099,283	86
Massachusetts	1,542	2,071,881	74	Increasing	2,035,931	98	1,841,599	89
Illinois	1,340	3,647,179	37	Increasing	3,617,384	99	3,465,124	95
Tennessee	1,321	1,993,985	66	Increasing	1,822,334	91	1,281,758	64
New Jersey	1,285	2,654,895	48	Increasing	2,628,971	99	2,485,458	94
Michigan	1,262	3,110,876	41	Increasing	3,025,592	97	2,799,723	90
Maryland	1,234	1,741,936	71	Increasing	1,690,591	97	1,454,882	84
North Carolina	1,221	3,029,426	40	Increasing	2,945,866	97	1,877,843	62
Missouri	1,088	1,859,314	59	Increasing	1,645,791	89	1,543,154	83
Washington	1,079	2,116,150	51	Increasing	1,947,414	92	1,761,268	83
Arizona	1,070	2,126,209	50	Increasing	1,823,957	86	1,581,263	74
Colorado	1,019	1,515,474	67	Increasing	1,326,225	88	1,214,940	80
Georgia	1,007	2,744,364	37	Stable	2,551,382	93	1,692,714	62
Virginia	988	2,417,756	41	Increasing	2,335,124	97	1,600,612	66
Minnesota	983	1,639,143	60	Increasing	1,591,845	97	1,300,120	79
Oregon	875	1,277,518	68	Increasing	1,066,414	83	752,552	59
Wisconsin	799	1,801,967	44	Increasing	1,768,519	98	1,624,595	90
Kentucky	680	1,324,824	51	Increasing	1,228,682	93	687,799	52
Connecticut	675	1,126,005	60	Stable	1,116,478	99	1,083,656	96
Indiana	642	1,924,394	33	Stable	1,870,940	97	1,303,933	68
Utah	635	647,683	98	Increasing	563,339	87	465,081	72
South Carolina	602	1,580,202	38	Stable	1,505,998	95	864,961	55
Kansas	551	834,296	66	Increasing	793,463	95	629,064	75
Oklahoma	550	1,111,196	50	Increasing	906,425	82	715,247	64
Alabama	526	1,481,051	36	Stable	1,281,533	87	411,087	28
Nebraska	523	544,015	96	Stable	472,613	87	385,537	71
Louisiana	514	1,323,135	39	Increasing	1,210,341	91	586,887	44
District of Columbia	501	156,244	321	Stable	156,244	100	156,244	100
Iowa	484	954,127	51	Increasing	920,674	96	601,497	63
Arkansas	394	894,076	44	Increasing	688,221	77	343,369	38
Nevada	362	853,238	42	Stable	782,013	92	739,884	87
South Dakota	320	263,569	121	Stable	195,434	74	83,896	32
Montana	261	349,886	75	Stable	218,456	62	184,624	53
New Hampshire	250	456,682	55	Decreasing	434,766	95	168,911	37
New Mexico	242	637,707	38	Decreasing	412,972	65	293,137	46
Delaware	238	317,289	75	Stable	310,948	98	191,001	60
Idaho	237	492,541	48	Stable	392,735	80	300,193	61
Mississippi	235	857,648	27	Stable	691,406	81	257,302	30
Maine	226	487,329	46	Stable	397,699	82	196,653	40
Rhode Island	226	331,279	68	Increasing	328,343	99	327,913	99
West Virginia	218	618,566	35	Stable	511,266	83	94,164	15
North Dakota	200	210,642	95	Stable	138,195	66	84,038	40
Alaska	199	181,485	110	Stable	86,096	47	69,666	38
Hawaii	193	442,220	44	Stable	351,150	79	269,325	61
Vermont	149	216,470	69	Stable	195,235	90	113,612	52
Wyoming	112	174,565	64	Stable	76,165	44	33,926	19

NOTE. Data sources: ClinicalTrials.gov and the US Census Bureau's American Community Survey.^11,16

^a Trend determined by Poisson regression, offset by population age ≥55 years population, where p <0.05.

AUTHOR CONTRIBUTIONS

Conception and design: M. Kelsey Kirkwood, Caroline Schenkel, David M. Waterhouse, Ishwaria M. Subbiah, Laura A. Levit

Administrative support: M. Kelsey Kirkwood, Dominique C. Hinshaw, Ishwaria M. Subbiah, Laura A. Levit

Collection and assembly of data: M. Kelsey Kirkwood, Suanna S. Bruinooge, Laura A. Levit

Data analysis and interpretation: M. Kelsey Kirkwood, Dominique C. Hinshaw, Suanna S. Bruinooge, David M. Waterhouse, Jeffrey M. Peppercorn, Ishwaria M. Subbiah, Laura A. Levit

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

State of Geographic Access to Cancer Treatment Trials in the United States: Are Studies Located Where Patients Live?

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).