County-level disparities in care for patients with glioblastoma

Abstract

Objective:

Racial and socioeconomic disparities in neuro-oncological care for brain tumors remain underexplored. This study aimed to analyze county-level disparities in glioblastoma multiforme (GBM) care in the United States, focusing on access to surgery and the use of adjuvant temozolomide chemotherapy and radiotherapy.

Methods:

Utilizing data from the Surveillance, Epidemiology, and End Results (SEER) 17 database, the Area Health Resources File, and the American Community Survey, this retrospective study used repeated cross-sectional data from 2010–2019 with multivariate regressions to understand the associations between county-level racial and socioeconomic characteristics, as well as the rates of surgery performed, delays in surgery, and use of adjuvant chemoradiotherapy for newly diagnosed GBM.

Results:

29,609 GBM patients from 602 different U.S. counties over a decade were included in this study. Counties with lower rates of surgery for GBM were associated with higher Black resident percentages (CE, −0.001; 95% CI, −0.002 to 0; p <0.05) and being located in the Midwest (CE, −0.132; 95% CI, −0.195 to −0.069; p<0.001) or West (CE, −0.127; 95% CI, −0.189 to −0.065; p<0.001) relative to the Northeast. Counties with delayed surgical treatment were more likely to lack neurosurgeons (AOR, 2.52; 95% CI, 1.77–3.60; p<0.001), have higher percentage of Black residents (AOR, 1.011; 95% CI, 1.00–1.02; p<0.05), and be located in the Midwest (AOR, 3.042; 95% CI, 1.12–8.24; p<0.05) or West (AOR, 3.175; 95% CI, 1.12–8.24; p<0.05). Counties with high rates of adjuvant radiation were less likely to have higher percentages of Black residents (AOR, 0.987; 95% CI, 0.980–0.995; p<0.01) and uninsured individuals (AOR, 0.962; 95% CI, 0.937–0.987; p<0.01).

Conclusions:

Counties without neurosurgeons and those with higher percentage of Black patients have delays in surgical care and demonstrate lower overall rates of surgery and adjuvant therapy for GBM. This study underscores the need for targeted interventions and policies that address structural barriers in health care access, improve equitable distribution of the neurosurgery workforce, and ensure timely and comprehensive GBM care to all populations.

Introduction

Racial and socioeconomic disparities in care for brain tumors remain understudied. Glioblastoma multiforme (GBM) is the most common and lethal primary intracranial tumor. The average life expectancy for GBM patients who undergo treatment is 12–15 months, but only four months for those who do not receive treatment, highlighting the crucial role of prompt and aggressive therapy.¹ Racial and ethnic inequities in both patient outcomes and timely access to the standard-of-care have been well-documented,^2,3 with domestic health care expenditure reaching nearly $500 billion annually attributable to such disparities.⁴

Prior studies have shown that Black patients with primary brain tumors are more likely to present as emergency or urgent admissions, and with more profound neurological deficits.⁵ Black and Medicaid patients also face higher hospital mortality rates post-craniotomy.⁵

The neurosurgery workforce is unequally distributed in the United States with regional variations in demographic constitution, socioeconomic status, unemployment rates, and number of neurosurgeons per capita.^5,7 These geographical disparities can drive delayed time to treatment and poor coordination of care. For patients with rapidly growing brain tumors, the consequences of delays in time to surgery at initial diagnosis can be especially profound, with far-reaching consequences on time to adjuvant therapy and ultimately clinical outcome.

Prior groups studying health care access for patients with brain tumors have utilized individual-level data;^6–8 however, regional and county-level characteristics and their impact on patients with GBM are poorly understood. Identifying these disparities has the potential to reveal underlying structural and socioeconomic etiologies of health care inequity in this patient population, where the need for expeditious assessment and treatment is great.

In this study, we examined drivers of disparities in the clinical care for patients with GBM by community characteristics. We analyzed the influence of county-level racial and socioeconomic characteristics on time to surgery and the provision of adjuvant chemoradiotherapy. These data revealed that significant barriers persist in comprehensive and consistent care across varied patient populations with GBM.

Methods

This retrospective time-series analysis employed repeated cross-sectional data from 2010 to 2019 as described below. Our study used publicly available data without personal identifiers and was determined as not human subject research by the corresponding author’s institutional review board. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

We used data from the November 2021 Research Plus Data from the Surveillance, Epidemiology and End Results (SEER) program, a resource maintained by the National Cancer Institute (NCI) that provides cancer incidence, treatment, and survival information within the United States. This dataset, selected for its comprehensive patient-level data, spanned 17 registries, representing 26.5% of the United States population,⁹ and covered the diagnosis, treatment, and survival of glioblastoma patients between 2010 and 2019, identified via the ICD-O-3 code (944). Exclusion criteria included incomplete data on diagnosis-to-treatment timeline, surgical recommendations, or receipt of radiation and chemotherapy, and cases diagnosed post-mortem. Since the SEER database does not permit the export of individual-level data with county information, we aggregated data for each county based on their Federal Information Processing Standards (FIPS) stratified by year. Our study variables extracted from this aggregated data included age adjusted GBM incidence per 100,000, median survival, average months from diagnosis to treatment, whether surgery was performed, and the administration of radiation and chemotherapy. The percentage of surgery performed represents the number of surgeries performed divided by the raw newly diagnosed GBM count in a county for a specific year. In SEER, patient survival is tracked up to two years from their month of diagnosis. Of note, the diagnosis date in the SEER database refers to the earliest clinical identification of the tumor, potentially via imaging, and not the pathological diagnosis following biopsy or surgery.¹⁰ This differentiation was important for our examination of the interval from diagnosis to surgery.

We merged the data by county and year with (1) county characteristics from the American Community Survey 5-year estimates, including census region, insurance status, income levels, and population demographics (race, ethnicity, education levels),¹¹ (2) neurosurgery workforce from the Area Health Resources File (AHRF),¹² (3) primary care Health Professional Shortage Area (HPSA) designations from the Health Services and Resources Administration (HRSA), and (4) rurality defined by 8 or 9 in the Department of Agriculture’s 2013 Rural Urban Continuum Codes.¹³ AHRF has previously been used to describe the distribution of neurosurgeons and access to neurosurgical care.⁷

Data Analysis

To assess the associations between county characteristics and our main outcome variables, we performed multivariate ordinary least squares (OLS) regressions on age adjusted GBM incidence (per 100,000), the percentage of patients who received surgery, and median survival. Each model included state and year fixed effects, controlling for the following county-year level covariates: whether a county had any neurosurgeons, census region, insurance status, median household income, population demographics, HPSA, and rurality. To further investigate disparities in treatment, we stratified counties by delayed time from diagnosis to treatment. Since SEER only codes months from diagnosis to treatment, counties with an average time from diagnosis to treatment greater than one month were designated as “delayed treatment,” based on a large multicenter cohort study with 1,033 patients, which found that not having surgery within a month of diagnosis could negatively affect overall survival.¹⁴ We then performed a multivariate logistic regression on whether a county had delayed treatment in a given year. The model included county-year level covariates and GBM incidence. To describe disparities following surgery, a multivariate logistic regression was performed on counties in the top quartile of percentage of patients who received chemotherapy or radiation following surgery for their primary tumor. We additionally analyzed the county-level characteristics of counties without neurosurgeons by performing a multivariate logistic regression on all US counties and various community-level characteristics including racial composition. Finally, we mapped counties without neurosurgeons and average time to diagnosis in 2019. All tests were two-sided, used an alpha level of 0.05, and were conducted in Stata 18.0 (StataCorp).

Results

Study distribution of neurosurgical workforce and patients with GBM

During the study period, 79.4% of all US counties did not have a neurosurgeon, leaving 75,000,000 residents with limited access to neurosurgical care and the vast majority of rural counties lacking any neurosurgeons (Table 1). The distribution of counties without neurosurgeons was concentrated in the Midwest and Southern United States (Figure 1A). From the SEER 17 database, there were 29,609 patients with GBM over 10 years located in 602 different counties analyzed in our study (Figure 2).

Table 1.

Baseline characteristics of counties with and without neurosurgeons in 2019

	Counties with Neurosurgeons (659)	Counties without Neurosurgeons (2484)
Total Population	255,388,599	74,336,882
% Black, non-Hispanic	11.448 (13.104)	8.098 (14.539)
% Hispanic	12.494 (13.743)	9.078 (14.010)
% Other race, non-Hispanic	7.831 (6.345)	5.791 (9.583)
% White, non-Hispanic	80.720 (14.080)	86.111 (16.464)
% High School Diploma	27.456 (6.775)	35.672 (6.708)
% Unemployed	5.239 (1.576)	5.231 (2.873)
Median Household Income	69,222.018 (18,329.267)	55,319.888 (13,275.009)
% Uninsured	8.213 (3.982)	10.063 (5.339)
Rural	3	639
HPSA	626	2432
Northeast	113	104
Midwest	155	900
South	283	1139
West	108	341

^*Mean or number of counties are given. If mean is given the Standard Deviation is in parenthesis.

Figure 1.

(A) Counties without neurosurgeons (dark blue) in 2019 based on data from the Area Health Resources File. (B) Average time to surgery for counties listed in the SEER 17 database across 2010–2019. Darker orange counties experience longer average time from a new clinical diagnosis of GBM to surgical intervention. The mean time from diagnosis to treatment was calculated for each county, and both maps were generated using the Bing Maps visualization integration with Microsoft Excel.

Figure 2.

Flowchart showing the total number of cases extracted from the SEER 17 database and the number of cases excluded for missing data. All cases were aggregated by county for statistical analysis.

County characteristics associated with rate of and time to surgery for GBM

Counties with a higher rate of surgery performed were less likely to have a higher percentage of Black residents compared to other counties (coefficient [CE], −0.001; 95% confidence interval [CI], −0.002 to 0; p <0.05) (Table 2). These counties were also less likely to be in the Midwest (CE, −0.132; 95% CI, −0.195 to −0.069; p<0.001) than in the Northeast.

Table 2:

Impact of County-level characteristics on GBM incidence, percentage of surgery performed.

	Age-adjusted incidence rate (per 100,000)			% Surgery Performed			Median Survival
	Coef.	p-value	95% Confidence Interval	Coef.	p-value	95% Confidence Interval	Coef.	p-value	95% Confidence Interval
County without Neurosurgeons	−0.174	0.329	−0.522, 0.175	0.015	0.319	−0.014, 0.044	0.559	0.153	−0.207, 1.324
% Black, Non-Hispanic	−0.013	0.020	−0.023, −0.002	−0.001	0.026	−0.002, 0.000	0.006	0.666	−0.023, 0.035
% Hispanic	−0.019	0.008	−0.034, −0.005	0.001	0.148	0.000, 0.002	0.044	0.012	0.010, 0.078
% Other, Non-Hispanic	−0.030	0.011	−0.053, −0.007	0.000	0.869	−0.002, 0.002	0.051	0.072	−0.005, 0.108
% White [Reference]	0.000	.	.	0.000	.	.	0.000	.	.
% High School Diploma	−0.005	0.660	−0.030, 0.019	−0.001	0.353	−0.003, 0.001	−0.040	0.222	−0.105, 0.024
% Unemployed	−0.022	0.350	−0.068, 0.024	0.000	0.990	−0.005, 0.005	−0.033	0.638	−0.170, 0.104
Median Household Income	0.000	0.661	0.000,0.000	0.000	0.468	0.000, 0.000	0.000	0.011	0.000, 0.000
% Uninsured	0.023	0.179	−0.011, 0.057	−0.003	0.067	−0.007, 0.000	−0.102	0.044	−0.201, −0.003
HPSA	−0.304	0.414	−1.035, 0.426	0.030	0.339	−0.031, 0.091	2.164	0.010	0.522, 3.805
Urban [Reference]	0.000	.	.	0.000	.	.	0.000	.	.
Rural	−0.266	0.088	−0.570, 0.039	0.025	0.213	−0.014, 0.063	0.942	0.096	−0.166, 2.050
Northeast [Reference]	0.000	.	.	0.000	.	.	0.000	.	.
Midwest	−0.275	0.470	−1.021, 0.471	−0.132	0.000	−0.195, −0.069	−1.374	0.110	−3.060, 0.313
South	−0.550	0.160	−1.317, 0.217	−0.020	0.532	−0.085, 0.044	−0.632	0.469	−2.343, 1.078
West	0.129	0.738	−0.628, 0.887	−0.127	0.000	−0.189, −0.065	−2.956	0.000	−4.585, −1.326

Counties with delayed time to initial surgery were more likely to be in a county without neurosurgeons (Adjusted Odds Ratio [AOR], 2.52; 95% CI, 1.77–3.60; p<0.001) and have higher percentage of Black residents (AOR, 1.011; 95% CI, 1.00–1.02; p<0.05) compared to other counties (Table 3). The counties with delayed treatment were also more likely to be located in the Midwest (AOR, 3.042; 95% CI, 1.12–8.24; p<0.05) compared to counties in the Northeast. Counties with delayed time to treatment had a lower GBM incidence rate (AOR, 0.928; 95% CI, 0.898–0.959; p<0.001). Figure 1B shows the distribution of average time from diagnosis to treatment in 2019, which reveals that delays in treatment are correlated with counties lacking neurosurgeons (Figure 1A).

Table 3:

Characteristics of counties with delayed treatment time

Counties with delayed treatment time	Adjusted Odds Ratio	p-value	95% Confidence Interval
County without Neurosurgeons	2.521	0.000	1.767, 3.596
% Black, Non-Hispanic	1.011	0.046	1.000, 1.021
% Hispanic	0.983	0.031	0.968, 0.998
% Other, Non-Hispanic	0.972	0.060	0.945, 1.001
% White [Reference]	1.000	.	.
% High School Diploma	1.009	0.472	0.985, 1.034
% Unemployed	0.981	0.401	0.937, 1.027
Median Household Income	1.000	0.016	1.000, 1.000
% Uninsured	1.014	0.341	0.985, 1.044
HPSA	2.078	0.127	0.811, 5.322
Urban [Reference]	1.000	.	.
Rural	1.371	0.083	0.960, 1.958
Northeast [Reference]	1.000	.	.
Midwest	3.042	0.029	1.123, 8.239
South	1.405	0.516	0.504, 3.919
West	3.175	0.029	1.124, 8.968
GBM age-adjusted incidence per 100k	0.928	0.000	0.898, 0.959

County characteristics associated with use of adjuvant therapy after surgery

Counties with a higher percentage of Black residents (AOR, 0.987; 95% CI, 0.980–0.995; p<0.01), uninsured individuals (AOR, 0.962; 95% CI, 0.937–0.987; p<0.01), and counties without a neurosurgeon (AOR, 0.767; 95% CI, 0.601–0.977; p<0.05) or those located in a rural area (AOR, 0.343; 95% CI, 0.252–0.466; p <0.001) were less likely to have high adjuvant radiation use (Table 4). Counties with greater GBM rates per 100,000 people (AOR, 1.15; 95% CI, 0.241–4.01; p<0.001) were more likely to fall into this quartile. Counties in the Midwest were less likely to have high use of adjuvant radiation compared to counties in the Northeast (AOR, 0.416; 95% CI, 0.230–0.752; p <0.01).

Table 4:

Characteristics of counties with increased use of post-surgery, adjuvant radiation therapy

High Adjuvant Radiation	Adjusted Odds Ratio	p-value	95% Confidence Interval
County without Neurosurgeons	0.767	0.032	0.601, 0.977
% Black, Non-Hispanic	0.987	0.001	0.980, 0.995
% Hispanic	0.997	0.567	0.988, 1.007
% Other, Non-Hispanic	1.000	0.000	1.000, 1.000
% White [Reference]	1.000	.	.
% High School Diploma	0.973	0.004	0.956, 0.992
% Unemployed	1.031	0.073	0.997, 1.067
Median Household Income	1.000	0.000	1.000, 1.000
% Uninsured	0.962	0.003	0.937, 0.987
HPSA	1.080	0.789	0.616, 1.892
Urban [Reference]	1.000	.	.
Rural	0.343	0.000	0.252, 0.466
Northeast [Reference]	1.000	.	.
Midwest	0.416	0.004	0.230, 0.752
South	1.836	0.037	1.037, 3.249
West	1.139	0.661	0.636, 2.039
GBM age-adjusted incidence per 100k	1.150	0.000	1.129, 1.172

Counties with a higher percentage of Black residents (AOR, 0.985; 95% CI, 0.979–0.992; p<0.001), counties without neurosurgeons (AOR, 0.667; 95% CI, 0.556–0.799; p<0.001), and those located in rural areas (AOR, 0.422; 95% CI, 0.341–0.523; p <0.001) were less likely to be in a county with high adjuvant chemotherapy use compared to other counties (Table 5). However, those with higher GBM rates per 100,000 people (AOR, 1.23; 95% CI, 1.206–1.249; p<0.001) were more likely to fall into this quartile.

Table 5:

Characteristics of communities with high use of post-surgery, adjuvant chemotherapy use

High Adjuvant Chemotherapy	Adjusted Odds Ratio	p-value	95% Confidence Interval
County without Neurosurgeons	0.667	0.000	0.556, 0.799
% Black, Non-Hispanic	0.985	0.000	0.979, 0.992
% Hispanic	0.996	0.324	0.989, 1.004
% Other, Non-Hispanic	1.000	0.187	1.000, 1.000
% White [Reference]	1.000	.	.
% High School Diploma	0.969	0.000	0.955, 0.983
% Unemployed	1.022	0.101	0.996, 1.050
Median Household Income	1.000	0.000	1.000, 1.000
% Uninsured	0.993	0.421	0.976, 1.010
HPSA	1.565	0.023	1.064, 2.304
Urban [Reference]	1.000	.	.
Rural	0.422	0.000	0.341, 0.523
Northeast [Reference]	1.000	.	.
Midwest	0.881	0.525	0.596, 1.303
South	1.269	0.248	0.847, 1.899
West	0.736	0.127	0.496, 1.091
GBM age-adjusted incidence per 100k	1.227	0.000	1.206, 1.249

County characteristics associated with having no neurosurgeons

We analyzed all 3,143 counties in the US to identify characteristics of counties without neurosurgeons in 2019. Counties in rural areas were significantly more likely to have no neurosurgeons (AOR, 50.037; 95% CI, 32.316–77.476; p<0.001) (Table 6). Counties with a high proportion of Black and Hispanic individuals were not more likely to have no neurosurgeons.

Table 6.

Characteristics of Counties without Neurosurgeons

Counties	Adjusted Odds Ratio	p-value	95% Confidence Interval
% Non-Hispanic, Black	0.963	0.000	0.959, 0.967
% Hispanic	0.976	0.000	0.972, 0.980
% Non-Hispanic, Other	0.951	0.000	0.944, 0.959
% White [Reference]
% High School Degree (or Equivalent)	1.275	0.000	1.264, 1.287
% Employed	1.057	0.000	1.036, 1.078
Median Household Income	1.000	0.259	1.000, 1.000
% Uninsured	1.072	0.000	1.057, 1.087
HPSA	0.521	0.000	0.421, 0.645
Rural [with urban as reference]	50.037	0.000	32.316, 77.476
Northeast [Reference]
Midwest	0.411	0.002	0.238, 0.712
South	1.258	0.435	0.707, 2.240
West	1.747	0.103	0.893, 3.418

Discussion

This study provides a county-level analysis of disparities in GBM care. Prior studies in this domain have focused on individual-level data, and have revealed the impact of race on rates of surgery, access to high-volume hospitals, readmissions, adjuvant radiation provision,^6,15–18 and the impact of insurance coverage on survival outcomes.¹⁹ It was previously shown, for instance, that when examining individual-level data, Black patients do not have significantly different odds of surgical resection for GBM in the “post-Stupp era.”⁸ However, the county-level data in the current study reveal that racial composition of a community is associated with significant differences in time to surgery and provision of adjuvant therapies. These findings highlight potential divergences between individual and community-level experiences with GBM treatment. The discrepancy suggests that while the standardization of the Stupp protocol²⁰ may help mitigate disparities at an individual level, structural and systemic barriers at the community level may continue to influence access to care.

Counties without neurosurgeons, and counties with high proportions of Black, Hispanic, and uninsured individuals were less likely to be in the top quartile of counties with patients receiving radiotherapy and chemotherapy after surgery. Several studies have demonstrated that workforce disparities in particular are associated with socioeconomically disadvantaged communities,^7,21 and that underrepresented minority patients have to travel longer distances for neurosurgical care.⁵ These differences also may confound coordination among the sphere of relevant health professionals and specialties (i.e., timely and referrals to oncologists and radiation oncologists).

Our analysis revealed that the incidence of the disease within a community is associated with the recommendation and application of the standard-of-care therapies. Counties with higher GBM incidence rates demonstrated a higher likelihood of falling into the top quartile for adjuvant radiation and chemotherapy provision post-surgery, suggesting that the disease burden within a community may play a significant role in access to treatment. Potential explanations for this include differences in local familiarity with the disease, resource allocation, and presence of established care pathways. Lower disease burden may result in less demand for providers in specialties like neurosurgery as well as multidisciplinary health care facilities providing the latest treatment advancements.

Given that the absence of neurosurgeons was found to be a predominant predictor of delays in care and poor access, we examined the county-level characteristics associated with having no neurosurgeons. Our finding that rural counties are substantially more likely to lack neurosurgery services illuminates major geographic disparities in health care access. Barriers such procedural specialist shortages, underinvestment in health care infrastructure, lower population densities, large geographic distances, and rural hospital closures hamper the maintenance of specialized neurosurgical services.^22,23

In 1975, the Study on Surgical Services for the United States report proposed a guideline of 1 neurosurgeon per 100,000 population.²⁴ A more recent study found that the ratio of neurosurgeons to total US population was 1:65,580 in 2013, but this ratio varied dramatically by state.²⁵ In our study, we considered the impact of having no neurosurgeons in a county, which has a mean population of approximately 100,000 in the US.²⁶ The lower rates and delays of both surgical and adjuvant GBM care in counties lacking at least 1 neurosurgeon align with the 1975 guideline. Furthermore, our analysis found that counties with greater than or equal to 1 neurosurgeon per 100,000 population were significantly less likely to have delayed time to treatment (data not shown). This suggests that the 1 neurosurgeon per 100,000 population recommendation is valid, but studies are needed that capture more granular data, including the number of brain tumor-specific neurosurgeons.

As we conceptualize solutions to the issues raised by this study, the role of the neurosurgical workforce emerges as a key driver of change. The underrepresentation of diversity within the neurosurgical workforce itself has been noted,^27,28 and addressing this could potentially impact broader disparities in neurosurgical care. A recent study showed that, only 5.1% of MD neurosurgical residents identified as Black and 6.4% as Hispanic, and only a single neurosurgical resident in an annual cohort who identified as American Indian.²⁹ Evidence suggests that Black physicians are more likely to practice in communities with a higher percentage of Black constituents.³⁰ Data published in the Association of American Medical Colleges (AAMC) from 2018–2019 have shown a higher percentage of matriculants from underrepresented groups in medicine plan to practice in underserved areas.³¹ Bolstering the neurosurgical workforce in rural communities should also include enhanced loan forgiveness and training partnerships between rural, community hospitals and larger tertiary medical centers through rural training tracks.

Improving access to neurosurgical care in underserved communities also requires federal and state intervention. The American Association of Neurological Surgeons (AANS) has emphasized the need to improve neurosurgery workforce distribution and called on Congress to increase the number of funded residency slots to address physician shortages in the United States.³²

Expanding telemedicine could also help improve access to neurosurgical care for populations away from tertiary care hospitals. Since the COVID-19 pandemic, telemedicine has greatly expanded access to subspecialty care with the Centers for Medicare and Medicaid Services (CMS) easing interstate regulations and boosting telemedicine reimbursement.³³ Tele-neurosurgery can help triage patients with suspected GBM, reduce patient travel times and costs for post-operative visits, and inform local providers about transferring to a facility with neurosurgical services.^34,35

The use of adjuvant chemoradiotherapy was underutilized in communities with uninsured patients. This particular population faces severe barriers to accessing health care since they have low rates of preventive care screenings,³⁶ are less likely to see doctors for health concerns,³⁷ and have higher rates of emergency department utilization compared to insured patients.³⁸ Disruptions in insurance coverage reduce cancer treatment adherence and overall cancer survival.³⁹ State Medicaid expansion and efforts to improve transitional coverage for patients with cancer could offer an additional layer of financial security for patients seeking affordable, quality access to care for GBM.

This study sheds light on community-level disparities GBM treatment and care, but several limitations must be acknowledged. The retrospective design, coupled with the use of a large national database, inherently results in the lack of granular data. As such, this limits our ability to discern the intricate clinical and patient factors influencing care decisions. We also recognize that our analyses suffer from the absence of additional outcome data such as progression free survival due to SEER database limitations and county-level nature of the study, which prevents individual-level outcomes analysis.

Furthermore, the SEER data is limited both numerically and geographically, posing a potential challenge to its generalizability. The SEER database captures data from a maximum of 17 registries, but does not fully represent the vast racial, socioeconomic, and geographic contexts across the United States. Consequently, the conclusions herein are confined to the regions represented in these registries. Additional work utilizing more comprehensive nationwide data would enhance our understanding of the disparities in GBM care.

Another critical factor potentially biasing our findings is that many neurosurgeons in each country may not treat brain tumors. This fact may skew the data, altering the landscape of treatment availability and, consequently, the results of our study. The Area Health Resources File only provides counts of total neurosurgeons by county and does not provide breakdown by neurosurgical sub-specialty. Future studies would benefit from more comprehensive data collection and granular patient-level information to better understand the complexity of the factors influencing the diagnosis, treatment, and outcomes of brain tumors.

Our analysis also lacks data on time to radiation or chemotherapy after surgery. This is because the SEER database, our primary data source, only includes information on the first form of treatment, which is typically surgery. We also did not examine the county-level workforce of neuro-oncologists and radiation oncologists due to data specificity limitations, a crucial component that could greatly influence the provision of adjuvant therapy. In addition, the SEER database does not document reasons why surgery was not performed, leaving us unable to discern if the lack of surgery was due to patient choice, surgical infeasibility, or structural barriers impeding surgical provision.

Conclusions

This study unravels county-level disparities in the timely surgical treatment and use of adjuvant therapies for GBM in the United States. Counties without neurosurgeons and those with higher percentage of Black patients also have significant delays in surgical care and lower overall rates of surgery and adjuvant therapy. The geographic distribution of neurosurgeons also impacts patient outcomes. As we strive towards equitable health care, policies and interventions that address these systemic barriers, increase diversity in the neurosurgery workforce, and ensure timely access to comprehensive GBM care may mitigate these disparities.