Where Have We Been and Where Are We Going? The State of the Hematology and Medical Oncologist Workforce in America

Abstract

PURPOSE

To identify geographic areas and populations in the United States likely to experience challenges accessing cancer care.

METHODS

We tabulated the number of hematologists and medical oncologists billing Medicare nationally, by state, and by county in 2014, 2019, and 2024, and assessed trends over time, offset by the US population age 55 years and older (where 80% of new cancers are diagnosed). We identified counties with oncologist presence, those with ≥25% of oncologists in late-career phases that could face reduced oncology care coverage in the future, as well as counties adjacent and nonadjacent to counties with oncologists, and outlined differences in populations residing in those counties. We characterized the current oncologist workforce by demographics and features of where they practice, stratified by career stage.

RESULTS

As the number of oncologists increased from 12,267 in 2014 to 14,547 in 2024, the number per 100,000 population age 55 years and older decreased from 15.9 to 14.9. In 2024, 38 states had fewer oncologists per capita than in 2014, with most facing limited rural workforce coverage. Although 89% of the population lived in counties with oncologists, 68% of the population lived in counties where more than a quarter of oncologists were nearing retirement age. Early-career clinicians were less likely than their later-career counterparts to work in counties that were rural or had high rates of cancer mortality, smoking, obesity, social vulnerability, uninsurance, and homes without broadband Internet access.

CONCLUSION

Gaps in oncologist coverage across the country exist, especially among rural populations and those with high cancer burden and socioeconomic risk. Understanding the geographic coverage of oncologists could facilitate efforts to improve patient access to cancer care.

INTRODUCTION

For the first time in 2024, estimates for new cancer cases in the United States exceeded 2 million.¹ With an aging population and a growing number of people being diagnosed with and surviving cancer, demand for cancer care services continues to grow. At the same time, much attention is being paid to the challenges of providing care in the current US health care environment.^2,3 A national survey conducted by ASCO in 2023 found that 59% of oncologists reported symptoms of burnout and that one in five were considering reducing clinical hours within the year.⁴ Another study found that 21% of oncologists practicing in 2015 had left clinical care by 2022.⁵

CONTEXT

• Key Objective

• What are the geographic characteristics that can influence whether a patient with cancer has access to an oncologist near their home?

• Knowledge Generated

• Approximately 11% of adults age 55 years and older lived in the 55% of US counties without oncologists and another 68% live in counties where ≥25% of oncologists are nearing retirement age. Small portions of oncologists worked in counties that are rural (7%), and have high cancer mortality rates (4%), high smoking rates (4%), and limited household access to broadband Internet (2%); an even smaller portion of early-career oncologists work in these areas.

• Relevance

• ASCO has prioritized ensuring that all patients with cancer have access to high quality care. Key to achieving this vision is locating and contextualizing areas with gaps and vulnerabilities in geographic access to inform targeted interventions.

The distribution of oncologists throughout the country is uneven and skews toward metropolitan areas.^6-8 This has been exacerbated by recent closures and consolidations of small practices that had broader geographic reach.^9,10 Recent workforce projections predict enough hematologists and oncologists to meet 93% of cancer care demand by 2037, with vastly different trajectories for rural and nonrural communities.¹¹ Although metropolitan counties are expected to have adequate capacity by 2037 (102%), nonmetropolitan areas are expected to have limited workforce supply (29%).¹¹ Similar patterns for other clinicians could further restrict geographic access to cancer prevention, screening, treatment, and survivorship care.^11,12 However, there is also heterogeneity among rural and nonrural areas. The prevalence of cancer risk factors, incidence, and mortality vary across the country, as do social determinants of health and resources that facilitate access to care.^13,14

In 2007 and 2014, ASCO published national workforce projections of oncologist shortages.^15,16 In intervening years, major political, economic, population, clinician workforce, and health system shifts (eg, the Affordable Care Act, COVID-19 pandemic) demonstrated that workforce projections may lose relevance quickly, and hinge on factors that are difficult to anticipate in model scenarios. Rather than focus its workforce studies on projections, ASCO has prioritized understanding the places where oncologists work and where people with cancer live. The goal is to identify geographic areas and populations likely to experience challenges in accessing cancer care.

In this study, we examined recent trends in physician rates per capita nationally, by state, and by county, with a focus on hematologists and medical oncologists. Within the current workforce, we looked at geographic availability of oncologists in different population settings (eg, by rurality, social vulnerability, cancer incidence, and mortality rates). Finally, we examined the geographic distribution of oncologists by career stage to identify areas and populations that may be susceptible to losing access to oncologist coverage because of retirements and a lack of new oncologists moving to the area.

METHODS

Oncologist Density and Temporal Trends

Hematologists and medical oncologists providing care to adult patients in the United States were identified from Medicare Care Compare, a publicly available data source developed by the Centers for Medicare & Medicaid Services and leveraged for previous oncologist workforce analyses.^7,9,17-20 According to a recent analysis, 0.1% of nonpediatric oncologists opt out of Medicare, suggesting that nearly all would be included in Care Compare reporting.²¹ Additional data source details are available in Table 1.

TABLE 1.

Data Sources and Study Considerations

Category	Source	Data Description and Study Considerations
Oncologists	Medicare Care Compare20	Information on clinicians billing Medicare within the previous year, for years 2024, 2019, and 2014. Used to represent oncologist supply because of data currency and geographic site information.18 Only 0.1% of nonpediatric oncologists opt out of Medicare, suggesting that nearly all would be included in Care Compare reporting.21 Variables of interest included specialty, sex, medical school graduation year, practice address(es), and organization identification number(s). Race/ethnicity information not available
Population counts	American Community Survey22	5-year estimates for population age 55 years and older from the American Community Survey. Used to approximate demand for cancer services because over 80% of new cancer diagnoses occur in the population age 55 years and older and data were available consistently across years available for all US counties. Appendix 1 includes sensitivity analyses using population age 40 years and older to account for rising rates in early-onset cancer and new cancer cases (for a cancer-specific population). Oncologist data were combined with population data available at the time (eg, 2024 oncologist data with 2018-2022 estimates)
County adjacency	2024 County Adjacency File25	Lists US counties and the counties that neighbor them. Used together with oncologist county tabulations as a proxy for travel requirements but does not account for differences in county size or travel times through congested areas
Rurality	2023 RUCC27	Distinguishes metropolitan counties by the population size of their metro area, and nonmetropolitan counties by their degree of urbanization and adjacency to a metro area. For this analysis, we compared urban counties (containing urban areas of 5,000+ people; RUCC values ≤3) with rural counties (RUCC values >3), where rural counties were considered highly vulnerable
HPSA	HPSA30	2024 primary care shortage areas determined for the National Health Service Corps to help optimally distribute participants in the program. Not specific to oncology. Shortage areas were considered highly vulnerable
Cancer incidence and mortality rates	State Cancer Profiles26	5-year age-adjusted incidence (2017-2021) and mortality (2018-2022) rate estimates for all cancer sites by county. Incidence and mortality data suppressed for some counties; incidence data not available for Kansas and Indiana. Top quartile values across all counties with data were considered highly vulnerable
Social vulnerability	2022 Social Vulnerability Index28	Composite score by county to aid in allocation of government resources to vulnerable communities in the event of a natural or man-made disaster or a disease outbreak. Score is derived from 16 US Census Bureau American Community Survey variables addressing area-level socioeconomic status, household characteristics (eg, civilians with disabilities), racial/ethnic minority status, housing type, and transportation. Top quartile values across all counties with data were considered highly vulnerable
Uninsurance	2024 County Health Rankings29	Percentage of adults younger than 65 years without health insurance by county, derived from 2021 Small Area Health Insurance Estimates. Top quartile values across all counties with data were considered highly vulnerable
Smoking	2024 County Health Rankings29	Percentage of adults who are current smokers (age-adjusted) by county, derived from the 2021 Behavioral Risk Factor Surveillance System data
Obesity	2024 County Health Rankings29	Percentage of adults who report a BMI ≥ 30 kg/m2 (age-adjusted) by county, derived from the 2021 Behavioral Risk Factor Surveillance System data. Top quartile values across all counties with data were considered highly vulnerable
Broadband Internet	American Community Survey22	5-year estimates of percentage of households with no access to broadband Internet by county (2018-2022). According to the Health Resources and Services Administration, broadband is the preferred Internet source for telehealth programs.42 Top quartile values across all counties with data were considered highly vulnerable

Abbreviations: HPSA, health professional shortage areas; RUCC, Rural-Urban Continuum Codes.

National provider data sets from Care Compare were downloaded in March 2024, April 2019, and April 2014 to allow for assessment over time. Eligible oncologists for our study had hematology, medical oncology, or hematology/oncology as their primary specialty, as well as those with a primary specialty of internal medicine and a secondary specialty of hematology, medical oncology, or hematology/oncology. We used geospatial techniques to assign county information to each oncologist address and tabulated the number of oncologists nationally, by state, and by county for each year of data. Oncologists with practice addresses in multiple counties in a year were assumed to split time evenly between counties (eg, 0.5 full-time equivalent in each county).

Tabulated oncologist data were combined with 5-year population estimates from the US Census Bureau.²² We defined oncologist density as the number of oncologists per 100,000 population age 55 years and older, the age group where 80% of new cancers are diagnosed, and performed sensitivity analyses using number of new cancer cases and population age 40 years and older to account for rising incidence of early-onset cancers (Appendix 1; Appendix Tables A1 and A2, online only).^23,24 Oncologist density was compared over time nationally, by state, and by county. We also assessed density trends among rural- and urban-based clinicians.

County Characteristics

To understand the current state of geographic access to oncologists, we categorized counties by oncologist presence in 2024 and presence in adjacent counties.²⁵ Oncologist medical school graduation year was used to approximate career stage, consistent with previous studies.^5,19 Oncologists were stratified as early-career (≤16 years from graduation), mid-career (17-29 years from graduation), or late-career (≥30 years from graduation; ie, nearing retirement age). We then identified counties with ≥25% oncologists in late-career stage, noting the ≥25% threshold could lead to clinically meaningful consequences if retirements were not met with replacements by other clinicians.

Oncologist coverage by county was supplemented with area-level population characteristics including age-adjusted cancer incidence and mortality rates, rurality, social vulnerability, rates of uninsurance, smoking, and obesity, as well as household access to broadband Internet (Table 1).^22,26-29

Oncologist Characteristics

Oncologists practicing in 2024 were stratified by career stage (described above) and characterized according to Care Compare variables and area-level data including sex, practice size, and practice in high vulnerability and health professional shortage areas (HPSAs; Table 1).^22,26-30 Practice size, on the basis of the number of oncologists per organization in Care Compare, was grouped as 1-3, 4-9, 10-29, or ≥30. The 2% of oncologists not associated with an organization were included in the 1-3 group. If oncologists practiced in multiple locations, they were assumed to split time evenly between practice locations.

Statistical Analysis

Descriptive statistics, including chi-square tests, were used to compare oncologist and US county characteristics.

Separate Poisson regression models were performed nationally, for each state, and for each county with oncologist data in 2014, 2019, and 2024 to estimate temporal changes in the number of oncologists offset by population size. Regression coefficients were used to assess increases or decreases within the region and summarized using descriptive statistics.

Sample sizes were large, leading to power to detect small differences and effect sizes. Therefore, we focus interpretation on estimated relative or absolute effect sizes, rather than P values. Table footnotes provide some guidance on interpretation of effect sizes.

Statistical analyses were conducted in R 4.2.2; geospatial analyses were conducted in ArcGIS Pro 3.2.2.

RESULTS

US Oncologist Trends, 2014-2024

There were 14,547 hematologist/oncologists billing Medicare in 2024 (Fig 1), representing a national oncologist density of 14.9 (per 100,000 adults age 55 years and older). Although the number of oncologists increased steadily, the number per capita decreased from a density of 15.9 in 2014 to 15.1 in 2019 to the current 14.9 (P < .01). Thirty seven percent of oncologists identified as female in 2024 (n = 5,384), up from 34% in 2019 (n = 4,525) and 30% in 2014 (n = 3,665). Although the number of late-career oncologists has increased modestly over time from 4,401 in 2014 to 4,690 to 4,748 in 2024, the share relative to the workforce has declined from 36% to 35% to 33% during the time frame.

FIG 1.

Flow diagram of 2024 oncologists for study analysis.

Approximately three quarters of oncologists practiced in a single county (74%), 19% practiced in two, and 7% practiced in three or more counties. Four percent of oncologists worked across state boundaries (n = 653). A total of 1,011 oncologists (7%) practiced in rural counties in 2024 for an oncologist density of 6.5. Meanwhile, nearly 16 million adults age 55 years and older (16% of the population) lived in rural counties. The remaining 13,536 oncologists practiced in urban counties (92%) for a density of 16.6, more than twice the rate in rural counties (P < .001). The number of oncologists per capita decreased in urban counties (P < .001) and remained stable in rural counties compared with 2014.

Oncologist density and trends varied by state (Fig 2). Wyoming had the fewest oncologists in 2024 (n = 21) and a density of 11.8. California had the most oncologists (n = 1,573) and a density of 14.8 in 2024. By density, Nevada had the lowest value of 7.6 and Washington DC had the highest value of 49.3 in 2024, although Washington DC also serves patients from Maryland and Virginia. Oncologist density by state was highly correlated with density calculations using new cancer case counts and population age 40 years and older (Appendix 1).

FIG 2.

Oncologist workforce characteristics, 2024, nationally and by state. ^aOncologist density defined as the number of oncologists per 100,000 adults age 55 years and older.

Thirty-eight states had lower oncologist density in 2024 compared with a decade earlier, while the remaining 12 states and Washington DC had higher density (Figs 2 and 3). Six states—Arizona, North Carolina, Utah, Vermont, Virginia, and Wyoming—had their highest oncologist density in 2024; the rest experienced declines in oncologist density compared with 2014 and/or 2019. Delaware had the largest negative density difference from 2014 to 2024 with five fewer oncologists per 100,000 residents. It had 37 oncologists in 2014 and 36 in 2024, but a corresponding 38% increase in residents age 55 years and older.

FIG 3.

2024 and 2019 oncologist density^a differences relative to 2014 by state. ^aOncologist density defined as the number of oncologists per 100,000 adults age 55 years and older.

State Characteristics, 2024

Oncologist demographics varied by state (Fig 2). Only four states—Alabama, Oregon, South Dakota, and Utah—had fewer than a quarter of oncologists nearing retirement age. Conversely, Alaska, Hawaii, Montana, Nevada, North Dakota, and Rhode Island had ≥40% of oncologists nearing retirement age. North Dakota had the lowest percentage of oncologists identifying as female (18%), while Maine and Maryland had the highest percentage (45% in both states). Wyoming, with 70% of its residents age 55 years and older living in rural counties, had the highest percentage of oncologists working in those counties (63%). By contrast, despite having 10% of its population in rural areas, Nevada had ≤1% of oncologists in those areas. Oncologist density by state was highly correlated with density calculations derived from new cancer cases and the population age 40 years and older (Appendix 1).

County Characteristics by Oncologist Presence and Adjacency, 2024

Fewer than half of US counties had oncologists present (45%, n = 1,420); however, counties with oncologist presence were home to 87 million adults age 55 years and older (89% of the population; Fig 4). Of these, 478 counties had less than a quarter of oncologists nearing retirement age (21% of the population), while 942 counties had more than a quarter of oncologists nearing retirement age (68% of the population; Fig 4).

FIG 4.

Counties by career stage of the oncology workforce.

Most of the 1,724 counties without oncologists had oncologists in neighboring counties (n = 1,517, where 10% of adults age 55 years and older lived), while the 207 remaining counties had no oncologists in adjacent counties (486,000 adults age 55 years and older, or 1% of the population; Fig 4). The distributions of adults age 40 years and older and of people newly diagnosed with cancer in counties by availability of an oncologist closely mirrored that of adults age 55 years and older (Appendix Table A2).

Oncologist presence varied by county population characteristic (Fig 5). Rural status was notable, where 67% of urban counties contained oncologists compared with 32% of rural counties. Most counties with lower cancer mortality rates had oncologists compared with 32% of counties with high mortality rates. Counties with high rates of households without broadband access were less than half as likely to have oncologist presence compared with counties with lower rates of households without broadband (20% v 54%). For every category evaluated, the share of counties with ≥25% of oncologists in late-career stage surpassed the share of counties with more early- and mid-career oncologists.

FIG 5.

County characteristics by career stage of the oncology workforce. Q1-Q3, first-third quartiles; Q4, fourth quartile; SVI, Social Vulnerability Index.

In many cases, the most vulnerable counties lacked oncologists both within or nearby (Fig 5). Notably, 10% of rural counties had no oncologists within or adjacent compared with 0% of urban counties. Similarly, counties with high rates of uninsured populations were twice as likely as those without high vulnerability status to have no oncologists nearby (12% v 5%).

Oncologist Characteristics and Features of Where They Practice by Career Stage, 2024

A total of 3,855 oncologists were in early phases of their career in 2024 (27%), while 5,916 were in mid (41%) and 4,786 were in late stages (33%; Table 2). Meaningful differences by career stage were evident in almost every personal, practice, and area-level population characteristic we evaluated.

TABLE 2.

Oncologist Characteristics by Career Stage

	Overall, Frequency (%)	Early-Career, Frequency (%)	Mid-Career, Frequency (%)	Late-Career, Frequency (%)
Total	14,539 (100.0)	3,855 (100.0)	5,916 (100.0)	4,768 (100.0)
Sexa,b
Female	5,383 (37.0)	1,752 (45.5)	2,502 (42.3)	1,129 (23.7)
Male	9,156 (63.0)	2,103 (54.6)	3,414 (57.7)	3,639 (76.3)
Practice sizea,b
1-3 oncologists	1,226 (8.4)	107 (2.8)	388 (6.6)	731 (15.3)
4-9 oncologists	2,169 (14.9)	480 (12.5)	892 (15.1)	797 (16.7)
10-29 oncologists	3,542 (24.4)	889 (23.1)	1,511 (25.5)	1,142 (24.0)
30+ oncologists	7,602 (52.3)	2,379 (61.7)	3,125 (52.8)	2,098 (44.0)
Rural statusa,b
Urban	13,528 (93.1)	3,681 (95.5)	5,504 (93.0)	4,343 (91.1)
Rural	1,011 (7.0)	174 (4.5)	412 (7.0)	425 (8.9)
HPSA shortage areaa,b
No	8,018 (55.5)	2,335 (59.7)	3,255 (55.1)	2,428 (52.4)
Yes	6,440 (44.5)	1,579 (40.3)	2,652 (44.9)	2,209 (47.6)
Cancer incidence
Q1-Q3 cancer incidence	12,287 (86.7)	3,288 (87.5)	5,011 (86.8)	3,988 (85.9)
Q4 cancer incidence	1,882 (13.3)	468 (12.5)	761 (13.2)	653 (14.1)
Cancer mortalitya,b
Q1-Q3 cancer mortality	13,783 (96.2)	3,691 (97.2)	5,611 (96.2)	4,481 (95.4)
Q4 cancer mortality	543 (3.8)	107 (2.8)	219 (3.8)	217 (4.6)
Social vulnerabilitya,b
Q1-Q3 SVI	9,229 (63.5)	2,502 (64.9)	3,833 (64.8)	2,894 (60.7)
Q4 SVI	5,310 (36.5)	1,353 (35.1)	2,083 (35.2)	1,874 (39.3)
Uninsurancea,b
Q1-Q3 uninsured	12,570 (86.5)	3,414 (88.6)	5,102 (86.2)	4,054 (85.0)
Q4 uninsured	1,969 (13.5)	441 (11.4)	814 (13.8)	714 (15.0)
Smokinga,b
Q1-Q3 smoking	14,012 (96.4)	3,765 (97.7)	5,704 (96.4)	4,542 (95.3)
Q4 smoking	527 (3.6)	90 (2.3)	212 (3.6)	226 (4.7)
Obesitya,b
Q1-Q3 obesity	13,770 (94.7)	3,714 (96.3)	5,601 (94.7)	4,456 (93.4)
Q4 obesity	769 (5.3)	141 (3.7)	315 (5.3)	312 (6.6)
Broadband accessa,b
Q1-Q3 no broadband	14,305 (98.4)	3,820 (99.1)	5,826 (98.5)	4,659 (97.7)
Q4 no broadband	234 (1.6)	35 (0.9)	90 (1.5)	109 (2.3)

NOTE. The table includes 14,539/14,547 = 99.9% of oncologists with medical school graduation year listed. Oncologists with practice sites in multiple areas had full-time equivalent values apportioned accordingly.

Abbreviations: HPSA, health professional shortage areas; SVI, Social Vulnerability Index.

^a P<.001 by chi-square analysis.

^b Scientifically meaningful finding where percentages had ≥1.5 relative and/or ≥3 absolute differences between categories.

Nearly half of early-career oncologists identified as female (46%) compared with 42% among mid-career and 24% among late-career oncologists. Early-career oncologists were half as likely as mid-career oncologists (3% v 7%) and a fifth as likely as late-career oncologists (3% v 15%) to work in small practices, with ≤3 oncologists. Conversely, they were more likely to work in large practices with ≥30 oncologists (62%) than those in mid-career (53%) or late-career (44%).

Few oncologists were practicing in areas of high vulnerability and there was wide variation by career stage (Table 2). Early-career oncologists were less likely to practice at a rural site (5%) than mid-career (7%) and late-career (9%) oncologists. Thirty-five percent of early-career versus 39% of late-career oncologists worked in counties with high social vulnerability. Similarly, 41% of early-career versus 48% of late-career oncologists practiced in HPSA. Meanwhile, only 3% of early-career, 4% of mid-career, and 5% of late-career oncologists worked in counties with high cancer mortality rates. Oncologists were underrepresented in counties with high levels of smoking (4%), obesity (5%), and uninsurance (14%), especially among early-career oncologists (2%, 4%, and 11%, respectively, for high smoking, obesity, and uninsurance counties). Only 2% of counties with limited household availability of broadband Internet had oncologist coverage.

DISCUSSION

Our study identified 15 practicing hematologists and medical oncologists for each 100,000 residents age 55 years and older nationally, with varying concentration by state and county. Approximately 11% of adults age 55 years and older lived in the 55% of US counties without oncologists. Most of these adults had oncologist availability in a neighboring county, leaving only 1% of the population geographically isolated from oncology care. However, even adults with oncologists nearby may face shortages in the future—68% of the population lived in counties where more than a quarter of oncologists were nearing retirement age.

By enriching oncologist practice location data with underlying population characteristics, we were able to contextualize gaps in geographic access to oncology care. Counties with high cancer mortality rates were twice as likely to not have oncologists practicing in or around them compared with counties with lower mortality rates. Similar trends were observed by county rurality, uninsurance prevalence, and smoking status. Thus, some of the most vulnerable populations are likely to experience the most challenges accessing care.

We also examined workforce trends over time. Over the past decade, oncologist density per capita has declined modestly overall and to varying degrees in 38 states because of population growth outpacing workforce expansion. Meanwhile, team-based care models—facilitated by greater reliance on advanced practice providers (nurse practitioners and physician assistants)—are increasingly common in oncology. These models may help address some challenges created by decreasing oncologist density, but further research is needed to understand how they can best facilitate patient access to cancer care.

Some have called for training additional oncologists to improve access to care, especially with aging population and growth in number of individuals requiring cancer care.^31,32 Training slots are slow and expensive to expand, and must be balanced against needs across medicine.³³ For this approach to be successful, new slots would need to be pursued in tandem with placement incentives to minimize access gaps. A cautionary tale can be found in the surge in training and subsequent oversaturation of emergency care doctors in some markets.³⁴

Policy change may be needed to improve retention and productivity of oncology clinicians. ASCO committed to “Drive healthy clinical and research work environments that lead to fulfillment for oncology professionals” in its 2023 five-year strategic plan and recently released a suite of well-being research papers including one on institutional approaches to promoting professional satisfaction.^4,35-37

Additionally, we found that early-career oncologists were less likely than their older counterparts to work in counties that are rural, have high cancer mortality rates, high smoking and obesity prevalence, and low levels of broadband Internet, suggesting that areas that already have scarce oncologist coverage may face declining capacity in the future. Compared with late-career oncologists, early-career oncologists are more likely to be in large practices (66% v 44% worked in practices with 30 or more oncologists) and work in urban areas (93% v 87%). Implementing policy solutions that help recruit and retain clinicians in rural and underresourced communities may be a strategy for broadening geographic access to cancer care. Approaches including visa waivers, loan forgiveness, and Medicare bonuses under the HPSA program have demonstrated success in primary care and could be adapted and further studied in cancer care. Efforts are also underway to increase oncologists' exposure to rural and community practice during training.³⁸ Telemedicine could expand the geographic reach of care,³⁹ and data on the regulatory flexibilities adopted during COVID-19 pandemic indicate that quality cancer research can be maintained with the use of such technologies.^40,41 However, regulation, reimbursement, and licensure issues are in flux.^40,41 Moreover, our research found areas of the country that had shortages of oncologists that also lack home access to broadband Internet, suggesting that telemedicine and remote technologies may not be available to all patients who could benefit.

These findings can inform state-level planning. Nevada, Idaho, and Oklahoma, for instance, have many times fewer oncologists per capita relative to areas such as Washington, DC, and Massachusetts. They have also experienced declines in oncologist density, have many late-career clinicians, and have limited capacity to serve their rural residents.

This study has limitations. Medicare Care Compare, our data source for oncologist supply, is current and encompasses all care sites.¹⁸ However, Care Compare contains limited variables and necessitated making several assumptions in our analysis. We assumed that each oncologist provided full-time patient care, treated similar numbers of patients, and evenly split their time between practice sites. Additionally, using medical school graduation year as a proxy for career stage could lead to misclassification (eg, someone who graduated later in life might be classified as mid-career rather than late-career). Additional considerations for all data sources can be found in Table 1. Ultimately, our high-level view does not allow for detailed understanding of patient travel and care utilization patterns that can vary across geographic settings, which should be examined in future research.

Although this analysis focused on the medical oncology and hematology workforce, high-quality cancer care is dependent on patient access to multispecialty and multidisciplinary care teams across ambulatory and hospital-based settings of care. Future research to examine workforce trends for other essential clinicians involved in cancer care delivery, including nurse practitioners, physician assistants, and specialists in radiation oncology, surgical oncology, and gynecologic oncology is warranted.

Overall, on the basis of geographic disparities in patients' access to oncologists, ASCO will continue to monitor the oncologist workforce to track national and local developments. This will influence ASCO's advocacy around local and system-level changes that affect the workforce. In addition, the methodologies used in this study provide ASCO with mechanisms to identify practices located in underserved areas, including in rural counties and those with high proportions of vulnerable populations, to target needs assessments, resource and quality improvement support, and community-appropriate advocacy. Ensuring that all patients with cancer have access to the highest-quality care is a top ASCO priority.

APPENDIX 1. DATA SELECTION AND SENSITIVITY ANALYSES FOR DEMAND FOR OUTPATIENT ONCOLOGY SERVICES

Three data points were considered to approximate national-, state-, and county-level demand in this study (Appendix Table A1).

We defined oncologist density as the number of oncologists per 100,000 adults age 55 years and older and performed sensitivity analyses to determine how well it captured the geographic distribution of oncologist density on the basis of the population age 40 years and older and on the basis of average annual new cancer cases. We assessed models for linearity using F-statistics and found them suitable at P < .001. We calculated Spearman correlations for nonparametric data comparing oncologist density calculations per demand proxy across all US states (n = 51) and counties (N = 3,144). We also compared the proportion of US population (age 55 years and older, age 40 years and older, and new case counts) who lived in counties with geographic access to oncologists.

The number of oncologists per 100,000 adults age 55 years and older had a strong positive correlation with the number of oncologists per 100,000 adults age 40 years and older across all states (r = 0.980) and all counties (r = 0.999). Among the 48 states and Washington, DC and the 93% of US counties with new cancer case estimates (n = 2,933), oncologist density on the basis of the population age 55 years and older was also strongly and positively correlated with oncologist density on the basis of new cancer cases (r = 0.958 for states; 0.999 for counties).

Furthermore, the distribution of population residing in counties by proximity to oncologists is similar (Appendix Table A2).

Study findings were robust to sensitivity analyses, and modifying the selection of demand proxy would not have changed the study conclusions.

TABLE A1.

Data Source Considerations for Cancer Care Demand

Data	Source	Considerations
Population age 55 years and older	US Census American Community Survey22	Advantages  80% of new cancer diagnoses occur in the age ≥55 years population  Uniformly available for all US counties and over time Disadvantages  includes populations who will not develop cancer
Population age 40 years and older	US Census American Community Survey22	Advantages  92%-96% of new cancer diagnoses occur in the age ≥40 years population  Accounts for rise in early-onset cancers  Uniformly available for all US counties and over time Disadvantages  Includes populations who will not develop cancer
Average annual incident cancer cases	NCI/CDC State Cancer Profiles26	Advantages  Compared with the overall population, patients within their first year of a cancer diagnosis are more reflective of patients who receive oncologist care Disadvantages  Data suppressed for some counties and not available for counties in Indiana and Kentucky  Oncologists also care for patients beyond a year from diagnosis

Abbreviation: NCI, National Cancer Institute.

TABLE A2.

Distributions of Population and Annual New Cancer Cases Residing in Counties by Proximity to Oncologists

County Status	Population Age 55 Years and Older (%)	Population Age 40 Years and Older (%)	Annual New Cancer Cases (%)
Oncologists, <25% late-Career	20,064,447 (21)	33,313,551 (21)	352,958 (20)
Oncologists, ≥25% late-Career	66,578,072 (68)	109,379,423 (69)	1,186,154 (69)
No oncologist, oncologists in adjacent county	10,186,858 (10)	15,763,928 (10)	181,444 (10)
No oncologist, no oncologist in adjacent county	485,746 (0.5)	747,519 (0.5)	7,889 (0.5)

AUTHOR CONTRIBUTIONS

Conception and design: M. Kelsey Kirkwood, David D. Chism, Helen M. Parsons, K. Robin Yabroff, Laura A. Levit

Administrative support: M. Kelsey Kirkwood

Provision of study materials or patients: M. Kelsey Kirkwood

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

Data analysis and interpretation: M. Kelsey Kirkwood, Erin P. Balogh, Melissa K. Accordino, Elizabeth Garrett-Mayer, Helen M. Parsons, Manali I. Patel, K. Robin Yabroff, 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

Where Have We Been and Where Are We Going? The State of the Hematology and Medical Oncologist Workforce in America

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).