Neighborhood disadvantage is associated with treatment access outcomes and survival among individuals with a primary brain tumor

Abstract

Background

Neighborhood disadvantage is linked to lower rates of healthcare access. To understand how residence affects the primary brain tumor (PBT) population, we assessed neighborhood disadvantage and population density with treatment access outcomes among a cohort of 666 adult participants with a PBT and study entry data in a large observational study at the National Institutes of Health (NIH) (NCT#: NCT02851706).

Methods

We assessed neighborhood disadvantage (measured by the area deprivation index [ADI]) and population density with symptom duration before diagnosis and time to treatment using ordinal logistic and linear regression. Kaplan–Meier survival curves were estimated by population density and ADI, overall and stratified by residential distance to the NIH, tumor grade, and age.

Results

Among 666 participants, 24% lived in more disadvantaged areas. Among the overall sample, there were no associations between ADI or population density with symptom duration, but the time to any treatment was longer for patients living in more disadvantaged neighborhoods (β = 7.78; 95% confidence interval [CI] = 0.02, 15.55), especially among those with low-grade PBTs (β = 36.19; 95%CI = 12.17, 60.20). Time to treatment was longer for those in nonurbanized areas and further from the NIH (β = 0.63; 95% CI = 0.08, 1.17). Patients living in more disadvantaged neighborhoods had higher 5-year survival compared with patients living in less disadvantaged neighborhoods (P = .02).

Conclusions

Individuals with low-grade PBTs living in more disadvantaged neighborhoods and further from NIH had a longer time to treatment. Future efforts should focus on strategies to reach patients living in disadvantaged neighborhoods.

Key Points.

• Based on the area deprivation index, 76% (n = 507) of patients enrolled in the Neuro-Oncology Branch—Natural History Study lived in less disadvantaged neighborhoods, and 24% (n = 159) lived in more disadvantaged neighborhoods.

• Individuals with low-grade primary brain tumors living in more disadvantaged neighborhoods and further from the National Institutes of Health had shorter symptom duration before diagnosis but longer time to treatment.

Importance of the Study.

Neighborhood disadvantage is linked to lower rates of healthcare access. To understand how place of residence, including neighborhood disadvantage and population density, affects the primary brain tumor (PBT) population, we assessed each with treatment access outcomes (symptom duration before diagnosis, time to treatment, and overall survival) among a cohort of adult participants with a PBT and study entry data in a large observational study at the National Institutes of Health. Among 666 participants diagnosed with PBT, there were no associations between neighborhood disadvantage or population density and symptom duration prior to diagnosis. However, individuals with low-grade PBTs living in more disadvantaged neighborhoods and further from NIH had a shorter symptom duration before diagnosis but a longer time to treatment. Future work is needed to understand healthcare access needs among those with low-grade PBTs and recruit more individuals with a PBT from disadvantaged and rural communities.

Primary brain tumors (PBTs) refer to a heterogeneous group of tumors arising from within the central nervous system. Many patients who learn they have a PBT present acutely, often in the emergency room, with distressing symptoms such as severe headaches accompanied by nausea or reduced consciousness, seizures, weakness, and cognitive symptoms; however, other individuals with PBTs may experience these and a myriad of other symptoms, including personality or behavioral changes and fatigue at varying levels of severity and for prolonged periods of time before seeking care. While they account for less than 2% of all cancers diagnosed in the United States,¹ the 5-year survival rate is 36% for malignant brain tumors and 92% for nonmalignant brain tumors.² Access to treatment and intervention, including surgery, chemotherapy, and radiation, are key to longer survival times across these cancers. Other factors relevant to longer survival times for individuals living with a PBT include younger age, female sex,² and better functional or performance status.³ Furthermore, access to experienced and specialized Neuro-Oncology centers may increase the likelihood of patients receiving biopsy, appropriate diagnosis,^4,5 timely treatment, and for malignant tumors—such as glioblastoma—the maximal safe extent of resection.^6,7 Longer time to treatment for symptomatic patients with a PBT, especially among adolescents and young adults, is associated with more long-term complications after treatment.^8,9 In another study among adults with glioblastoma, shorter time with symptoms before treatment and early surgical treatment was associated with longer overall survival times.¹⁰

Prompt access to healthcare and specialized Neuro-Oncology centers is a necessity not only for the diagnosis of PBTs but also for access to clinical trials. A study led by our lab and the Society of Neuro-Oncology queried healthcare providers from around the world and found that less than 30% of patients with a PBT are referred to participate in clinical trials,¹¹ and that many existing specialized Neuro-Oncology centers are located in large metropolitan areas.¹² Additionally, for the majority of patients diagnosed with a PBT, there are considerable barriers to healthcare access and clinical trial recruitment, such as the cost of treatment and travel.^12,13 This may result in some individuals being disproportionally affected, such as those with cancer who are more socioeconomically disadvantaged or living in nonurban or rural areas.¹⁴ The definition of “local” healthcare has differed in previous studies based on the geographic characteristics of the town or city, the means or time needed to travel, and the type of care being accessed.¹⁵ However, without continued access, disparities in access have the potential to affect cancer outcomes, including access to subsequent treatment after recurrence and education on continued symptom management.

For individuals facing barriers related to socioeconomic status, individual neighborhoods representative of the physical area in which they live¹⁶ can be a unit of measurement to understand the potential resources available to them. Among individuals with cancer living in disadvantaged neighborhoods, there is an increased risk of negative health outcomes, including higher mortality rates¹⁷ and worse postoperative outcomes.¹⁸ However, little is known about the effects of neighborhood disadvantage on treatment access and the health-related outcomes of individuals living with PBT. Previous studies examining neighborhood disadvantage among patients with a PBT have primarily focused on assessing the incidence and survival among individuals with glioblastoma.^18,19 To address this research gap, the aim of this study was to assess the association of neighborhood disadvantage and population density with symptom duration before diagnosis, time to first treatment (any treatment, radiation, radiation with concurrent chemotherapy, or chemotherapy and immunotherapy), and overall survival among a cohort of adults with a PBT enrolled in a large National Institutes of Health (NIH) observational trial (NCT #: NCT02851706, PI: T.S.A.). We hypothesized that individuals living in more disadvantaged neighborhoods and away from urban areas would experience longer symptom duration before diagnosis, longer time to treatment, and lower overall survival rates.

Materials and Methods

Study Population

A cohort of patients with a PBT diagnosis aged 18 years or older were identified from the National Cancer Institute Neuro-Oncology Branch’s Natural History Study (NOB-NHS). The NOB-NHS is a longitudinal, observational study that follows individuals diagnosed with a primary central nervous system tumor, who may enroll at any point along their disease trajectory including at the time of diagnosis, during treatment, recurrence, or surveillance. Eligibility is not limited to those diagnosed according to the 2021 World Health Organization Classification of Tumors of the Central Nervous System.²⁰ The NOB-NHS collects data on clinical outcomes, including information on disease presentation and patient-reported outcome questionnaires. Institutional Review Board approval was obtained for the NOB-NHS, and written consent was obtained from participants. Among NOB-NHS participants with PBT, participants living in the United States with baseline study entry data were included in the analysis (n = 666). Figure 1 details a flow diagram of the cohort inclusion criteria. Furthermore, we examined the impact of distance from the NIH categorized as participants who lived within 200 miles of the NIH or further than 200 miles from the NIH to account for differences in referral patterns and access based on distance traveled.

Figure 1.

Flow diagram of analytic study population. This study flow diagram demonstrates the number of participants with study entry data from the Neuro-Oncology Branch—Natural History Study who were excluded from statistical analysis and the corresponding exclusion criteria. *Other tumor types refer to tumors not classified as solely brain tumors such as spine tumors, extra-axial tumors, or brain tumors that have infiltrated into the spine.

Exposures

We assessed 2 exposures: (1) neighborhood disadvantage levels using the area deprivation index (ADI) (less disadvantaged [reference group], more disadvantaged) and (2) population density (urbanized [reference group], nonurbanized). The ADI is a composite measure of neighborhood deprivation or disadvantage developed by the Health Resources and Services Administration.²¹ The ADI consists of 17 socioeconomic status items centered on the 4 domains of income, education, employment, and housing quality,²¹ which have been used to calculate a publicly available ADI score based on national percentiles. The ADI score is ranked from 1 to 100, with lower scores indicating less neighborhood disadvantage. Similar to previous studies,^18,22 ADI was dichotomized into 2 groups representing more (≥ 40) and less (< 40) disadvantaged groups based on zipcode-level national quartiles. For population density, areas were categorized into urbanized or nonurbanized areas (inclusive of suburban and rural populations) using the U.S. Department of Agriculture definition, which defines urbanized areas as those with a population density of 1000 or more persons per square mile.²³

Outcomes

The outcomes of interest were (1) symptom duration, defined as time with symptoms prior to diagnosis: less than 6 months with symptoms, 6 months to 1 year with symptoms, and more than 1 year with symptoms, (2) time to treatment, defined as time in months from PBT diagnosis and first surgery to the time of any first treatment, including radiation, radiation and chemotherapy, or chemotherapy and immunotherapy, and (3) overall survival, defined as time from PBT diagnosis to death. Symptom duration was self-reported by patients. Information on treatment history (time to treatment) was collected from patient medical records, and treatment could have been completed at either the NIH or the patient’s local healthcare system.

Statistical Analyses

Frequency and percentages were reported for categorical variables. Means and standard deviations (SDs) were reported for continuous variables. Descriptive characteristics were compared by distance to the NIH using independent sample t-tests and Pearson’s chi-squared tests. To assess the association of population density and neighborhood disadvantage with symptom duration, we estimated odds ratios (ORs) and 95% confidence intervals (CI) using multivariable ordinal logistic regression models adjusted for age and sex, overall and by distance to the NIH. To assess the association of population density and neighborhood disadvantage with time to treatment, we estimated beta coefficients and 95% CIs using multivariable linear regression models adjusted for age and sex, overall and by distance to the NIH.

To assess overall survival, we used the Kaplan–Meier estimator of survival probabilities. For overall survival, follow-up time began at cancer diagnosis until death, end of follow-up, or the end of study (August 31, 2022). We assessed differences in survival by population density and neighborhood disadvantage using the log-rank test, and survival was stratified by distance to the NIH, tumor grade (low, high), and age at cancer diagnosis (<50, ≥50 years). For all analyses, we conducted sensitivity analyses limited to participants diagnosed with glioblastoma (N = 242). Two-sided P-values were reported with a significance set at .05. Stata version 17 was used.²⁴

Results

There were 666 participants with a PBT who were seen at the NIH. At study entry, 7% of the study sample reported that they were on a clinical trial at the National Cancer Institute, which consisted of radiation and chemotherapy. Sixty percent (n = 402) of participants lived within a short distance, while 40% (n = 264) lived long distance to the NIH. Of these participants, 60% (n = 401) lived in urbanized areas, and 40% (n = 265) of participants lived in nonurbanized areas. Participant ADI scores ranged from 1 to 92 with a mean of 26.96 and SD of 21.54. Based on ADI, 76% (n = 507) of patients lived in less disadvantaged neighborhoods, and 24% (n = 159) lived in more disadvantaged neighborhoods. The map presented in Figure 2 shows the distribution of NOB-NHS participants, distinguishing those living in more or less disadvantaged neighborhoods across the United States. The median age at cancer diagnosis was 44.22 years (SD 15.19; range: 5–79), and those who lived closer to the NIH were slightly older (46.20 ± 15.01 vs. 41.19 ± 14.99 years) compared to those who lived further from the NIH. Participants were predominantly non-Hispanic (n = 599; 90%) and White (n = 544; 82%), married or living with a partner (n = 451; 68%), employed (n = 326; 49%), or retired (n = 112; 17%), and had received either a bachelor’s (n = 225; 34%) or advanced degree (n = 246; 37%). Most participants had high-grade tumors (3 or 4: n = 467; 70%) and had undergone surgery (n = 648; 97%) (Table 1). The 4 most common tumor types in the sample were glioblastoma (n = 242; 36%), astrocytoma (n = 142; 21%), oligodendroglioma (n = 82; 12%), and ependymoma (n = 44; 7%), comprising 78% of the sample. The sociodemographic and clinical characteristics of the subset of participants with glioblastomas are presented in Supplementary Table 1. In univariate analysis, those living in more disadvantaged neighborhoods were younger (P < .001), more likely to be female (P < .001), reported lower income (P < .001), and did not to have an advanced degree (P < .001). Importantly, although not statistically significant, those living in less disadvantaged neighborhoods were less likely to have low-grade tumors (n = 109; 22%) compared to those living in more disadvantaged neighborhoods (n = 49; 31%).

Figure 2.

Map of cohort participants by neighborhood disadvantage levels. Approximate locations of study participants aggregated to larger (2-digit Zip Code Tabulation) geographic areas. Regions with Area Deprivation Index (ADI) values below the first national quartile are designated as advantaged and the remainder as disadvantaged. Participants in the blue advantage group correspond to the “less advantaged,” and the red disadvantage group corresponds to the “more disadvantaged” group.

Table 1.

Descriptive Characteristics of Primary Brain Tumor Patients (n = 666)

	Overall Cohort (n = 666)		Short Distance (<200 miles) to NIH (n = 402)		Long Distance (200+ miles) to NIH (n = 264)		P Value
	N	%	N	%	N	%
Age at cancer diagnosis, mean (SD)	44.22	15.19	46.20	15.01	41.19	14.99	<.001
Sex							.30
Female	279	41.89	162	40.30	117	44.32
Male	387	58.11	240	59.70	147	55.68
Race							<.001
Asian	38	5.71	30	7.46	8	3.03
Black or African American	46	6.91	38	9.45	8	3.03
Native Hawaiian or Pacific Islander	2	0.30	1	0.25	1	0.38
White	544	81.68	310	77.11	234	88.64
Other	10	1.50	7	1.74	3	1.14
American Indian or Alaska Native	2	0.30	0	0.00	2	0.76
Missing	24	3.60	16	3.98	8	3.03
Ethnicity							.19
Not Hispanic/Latino	599	89.94	360	89.55	239	90.53
Hispanic/Latino	51	7.66	35	8.71	16	6.06
Missing	16	2.40	7	1.74	9	3.41
Income							.045
<$50 000	88	13.21	45	11.19	43	16.29
$50 000–$149 000	152	22.82	86	21.39	66	25.00
≥$150 000	114	17.12	81	20.15	33	12.50
Prefer not to answer	50	7.51	29	7.21	21	7.95
Missing	262	39.34	161	40.05	101	38.26
Employment Status							.49
Employed	326	48.95	202	50.25	124	46.97
Retired	112	16.82	71	17.66	41	15.53
Unemployed	219	32.88	123	30.60	96	36.36
Missing	9	1.35	6	1.49	3	1.14
Education Level							<.001
High school or less	84	12.61	56	13.93	28	10.61
Associate degree or any college	103	15.47	58	14.43	45	17.05
Bachelor degree	225	33.78	113	28.11	112	42.42
Advanced degree	246	36.94	170	42.29	76	28.79
Missing	8	1.20	5	1.24	3	1.14
Marital Status							.53
Single	206	30.93	118	29.35	88	33.33
Married/Partner	451	67.72	278	69.15	173	65.53
Missing	9	1.35	6	1.49	3	1.14
Karnofsky Performance Scale							.22
90–100	440	66.07	258	64.18	182	68.94
50–80	183	27.48	120	29.85	63	23.86
Missing	43	6.46	24	5.97	19	7.20
Tumor grade							.20
Grade 1/2	158	23.72	84	20.9	74	28.03
Grade 3/4	467	70.12	293	72.89	174	65.91
No tissue diagnosis	24	3.60	15	3.73	9	3.41
Not assigned	17	2.55	10	2.49	7	2.65
Recurrences							.008
Yes	400	60.06	225	55.97	175	66.29
No	266	39.94	177	44.03	89	33.71

Unemployed includes self-reported unemployment due to disability, medical leave, student, volunteer, and homemaker.

Abbreviation: NIH, National Institutes of Health.

Symptom Presentation and Duration

The most common presenting symptoms were neurologic dysfunction (n = 309; 46%), followed by headaches (n = 221; 33%) and seizures (n = 216; 32%). Other symptoms reported at the presentation included cognitive dysfunction (n = 155; 23%), fatigue (n = 18; 3%), nausea/vomiting (n = 54; 8%), pain (n = 2; 0.3%), gastrointestinal/autonomic symptoms (n = 2; 0.3%), and behavioral or mood changes (n = 15; 2%). More patients with high-grade PBTs presented with neurologic dysfunction (n = 241; 52%) compared to patients with low-grade PBTs (n = 68; 43%) and more headaches (n = 176, 38% vs. n = 45, 28%), while more patients with low-grade PBTs presented with seizures compared to patients with high-grade PBTs (n = 60, 38% vs. n = 156, 33%).

There were no associations observed between population density or neighborhood disadvantage and symptom duration prior to diagnosis, overall or by distance to the NIH (Figure 3). Among those with low-grade tumors, participants living in more disadvantaged neighborhoods and living long distances to NIH had shorter symptom duration (OR = 0.30, 95% CI = 0.12, 0.75) compared with participants living in less disadvantaged neighborhoods. These findings were not found among participants with high-grade PBTs or among the subset of those with glioblastoma (Supplementary Tables 2 and 3).

Figure 3.

Association of population density and area deprivation index (ADI) with symptom duration among patients diagnosed with primary brain tumors, overall and by distance traveled to National Institutes of Health (NIH), and stratified by tumor grade. Panel (a) demonstrates the association of population density with symptom duration using odds ratios among patients diagnosed with primary brain tumors, overall and by distance traveled to NIH. Panel (b) demonstrates the association of neighborhood disadvantage levels with symptom duration using odds ratios among patients diagnosed with primary brain tumors, overall and by distance traveled to NIH. CI = confidence interval, OR = odds ratio, Ref = reference group. Models were adjusted for age (continuous) and sex (male/female). Short distance is defined as living <200 miles from NIH, and long distance is defined as living 200+ miles from NIH.

Time to Treatment

There was no difference in time to treatment for participants based on population density in relation to those who received any treatment, radiation alone, or chemotherapy and immunotherapy. Longer time to treatment with radiation and chemotherapy was observed among those living in nonurbanized areas and longer distance to the NIH (β = 0.63; 95% CI = 0.08, 1.19). For time to receipt of any treatment, we observed that patients living in more disadvantaged neighborhoods had longer time to treatment (β = 7.78; 95% CI = 0.02, 15.55); and similarly, although not significant, longer time to any treatment receipt was seen among those living in more disadvantaged neighborhoods and living long distance to NIH (β = 11.17; 95% CI = −0.99, −24.34) (Table 2). Among those with low-grade tumors, patients living in more disadvantaged neighborhoods had an even longer time to treatment (β = 36.19; 95% CI = 12.17, 60.20). There were no associations between time to treatment and neighborhood disadvantage among individuals with high-grade tumors. Among the subset of participants with glioblastoma, there were no associations observed between population density or neighborhood disadvantage and time to any treatment type, either among the overall group of participants with glioblastoma or by their distance to the NIH (Supplementary Table 4).

Table 2.

Association of Population Density and ADI With Time to Treatment Among Patients Diagnosed with Primary Brain Tumors, Overall and by Distance Traveled to NIH, and Stratified by Tumor Grade

	Overall Cohort (N = 666)			Short Distance (<200 miles) to NIH (n = 402)			Long Distance (200 +miles) to NIH (n = 264)
	N (%)	β	95% CI	N (%)	β	95% CI	N (%)	β	95% CI
Received any treatment (n = 481)
Population density
Urbanized	290 (60.29)	Ref (0)	—	190 (64.41)	Ref (0)	—	100 (53.76)	Ref (0)	—
Nonurbanized	191 (39.71)	0.96	−5.83, 7.74	105 (35.59)	−3.78	−11.76, 4.20	86 (46.24)	5.42	−6.84, 17.68
ADI
Less disadvantaged	361 (75.05)	Ref (0)	—	262 (88.81)	Ref (0)	—	99 (53.23)	Ref (0)	—
More disadvantaged	120 (24.95)	7.78	0.02, 15.55*	33 (11.19)	−3.85	−16.16,8.47	87 (46.77)	11.17	−0.99, 24.34
Received any treatment, low grade (n = 158)
Population density
Urbanized	41 (53.25)	Ref (0)	—	24 (66.67)	Ref (0)	—	24 (58.54)	Ref (0)	—
Nonurbanized	36 (46.75)	15.45	−8.86, −39.75	12 (33.33)	18.09	−5.21, 41.39	17 (41.46)	−1.85	−43.03, 39.33
ADI
Less disadvantaged	48 (62.34)	Ref (0)	—	32 (88.89)	Ref (0)	—	25 (60.98)	Ref (0)	—
More disadvantaged	29 (37.66)	36.19	12.17, 60.20***	<5	DS	DS	16 (39.02)	32.14	−8.59, 72.86
Received any treatment, high grade (n = 46)
Population density
Urbanized	240 (61.07)	Ref (0)	—	160 (63.75)	Ref (0)	—	80 (56.34)	Ref (0)	—
Nonurbanized	153 (38.93)	−2.13	−8.65,4.39	91 (36.25)	−6.25	−15.01, 2.51	62 (43.66)	4.67	−5.01, 14.36
ADI
Less disadvantaged	304 (77.35)	Ref (0)	—	223 (88.84)	Ref (0)	—	81 (57.04)	Ref (0)	—
More disadvantaged	89 (22.65)	−2.32	−10.03, 5.38	28 (11.16)	−7.60	−21.22, 6.02	61 (42.96)	0.56	−9.16, 10.28
Received radiation and chemotherapy (n = 284)
Population density
Urbanized	172 (60.56)	Ref (0)	—	120 (62.83)	Ref (0)	—	41 (44.09)	Ref (0)	—
Nonurbanized	112 (39.44)	−0.15	−3.43, 3.14	71 (37.17)	−0.02	−4.94, 4.90	52 (55.91)	0.63	0.08, 1.19*
ADI
Less disadvantaged	227 (79.93)	Ref (0)	—	170 (89.01)	Ref (0)	—	57 (61.29)	REF (0)	—
More disadvantaged	57 (20.07)	−0.29	−4.33, 3.76	21 (10.99)	1.72	−6.09, 9.54	36 (38.71)	0.08	−0.50, 0.65
Received radiation alone (n = 78)
Population density
Urbanized	55 (70.51)	Ref (0)	—	34 (77.27)	Ref (0)	—	21 (61.76)	Ref (0)	—
Nonurbanized	23 (29.49)	5.55	−16.49, 27.60	10 (22.73)	−5.11	−36.94, 26.73	13 (38.24)	11.64	−22.27, 45.56
ADI
Less disadvantaged	56 (71.79)	Ref (0)	—	38 (86.36)	Ref (0)	—	18 (52.94)	Ref (0)	—
More disadvantaged	22 (28.21)	−10.04	−33.42, 13.33	6 (13.64)	−27.01	−68.04, 14.02	16 (47.06)	−12.03	−48.17, 24.11
Received chemotherapy and immunotherapy (n = 32)
Population density
Urbanized	15 (46.88)	Ref (0)	—	3 (33.33)	Ref (0)	—	12 (52.17)	Ref (0)	—
Nonurbanized	17 (53.13)	6.33	−7.39, 20.04	6 (66.67)	1.01	−0.55, 2.58	11 (47.83)	8.87	−10.64, 28.38
ADI
Less disadvantaged	21 (65.63)	Ref (0)	—	8 (88.89)	Ref (0)	—	13 (56.52)	Ref (0)	—
More disadvantaged	11 (34.38)	−4.40	−19.18, 10.37	<5	DS	DS	10 (43.48)	−9.32	−29.77, 11.14

Models were adjusted for age (continuous) and sex (male/female). Short distance is defined as living <200 miles from NIH, and long distance is defined as living 200+ miles from NIH.

Abbreviations: ADI, area deprivation index; CI, confidence interval; DS, data suppressed due to n < 5; NIH, National Institutes of Health; OR, odds ratio; Ref, references; Ref, reference group.

Bold indicates significance. *P <.05; *P <.01; ***P <.001.

Overall Survival

There were no differences in overall survival by population density (P = .28) as 5-year survival was 66% (n = 109 deaths, 95% CI = 59.89, 70.52) for patients in urbanized areas compared with 63% of those living in nonurbanized areas (n = 82 deaths, 95% CI = 56.17, 69.13) (Figure 4A). For neighborhood disadvantage, participants living in more disadvantaged neighborhoods surprisingly had higher 5-year survival (n = 36 deaths; 72%, 95% CI = 63.27, 79.02) compared with participants living in less disadvantaged neighborhoods (n = 155 deaths, 62.02%, 95% CI = 57.03–66.60; P = .02) (Figure 4B). For both population density and neighborhood disadvantage, there were no significant differences in overall survival when stratified by distance to the NIH. There were no significant differences based on population density or neighborhood disadvantage in overall survival when stratified by age, tumor grade, or among those with a glioblastoma (Supplementary Figure 1A and 1B).

Figure 4.

Survival distributions by (A) population density and (B) area deprivation index, overall and stratified by distance to National Institutes of Health (NIH). Panel (a) demonstrates the survival distributions of participants diagnosed with primary brain tumors based on population density, overall and by distance traveled to NIH. Panel (b) demonstrates the survival distributions of participants diagnosed with primary brain tumors based on neighborhood disadvantage levels, overall and by distance traveled to NIH. P indicative of log-rank tests of survival distribution differences. Local/short distance is defined as living <200 miles from NIH, and long distance is defined as living 200+ miles from NIH.

Discussion

To our knowledge, this is the first study to explore the impact of neighborhood disadvantage and population density with treatment access-related outcomes, including symptom presentation and overall survival, in a diverse cohort of patients with a PBT. Although we found no associations between neighborhood disadvantage or population density with symptom duration prior to diagnosis among the overall study sample, patients with low-grade PBTs living in more disadvantaged neighborhoods and further from NIH had a shorter symptom duration before diagnosis but overall a longer time to treatment.

As demonstrated by the sample’s report at the time of presentation, common symptoms across the sample were neurologic dysfunction, headaches, and seizures. Participants from more disadvantaged neighborhoods may lack access to healthcare, leading them to seek out care at emergency rooms for seizures and severe headaches, paradoxically leading to shorter time to imaging studies and diagnosis,^25–27 especially for those living long distances or traveling from rural areas to care.²⁶ This association reported above was not found in those with more malignant tumors, which may represent a sampling bias, as malignant tumors are associated with acute symptoms, rapid progression, and shorter survival times. Patients in our sample with prolonged presentation time and delayed treatment may not have survived long enough to be referred for specialty care, especially as among patients with glioblastoma shorter symptom duration prior to diagnosis and earlier treatment is linked to longer survival times.¹⁰ Perla and colleagues¹⁸ reported in a 5-year retrospective analysis in two datasets that individuals with glioblastoma from more disadvantaged neighborhoods had decreased odds of receiving treatment such as gross total resection, chemotherapy, radiation, and even participating in clinical trials; but the authors also reported no difference in survival times among patients in these datasets, reflecting the limited treatment options and short time to progression and death among these patients. Familiarity with and availability of state-of-the-art diagnostic technologies and standardized treatments for these patients, or other factors not measured in analyses, may also be factors not accounted for which may have impacted our results.

Among our sample, neither living in more disadvantaged neighborhoods nor nonurbanized areas were associated with shorter overall survival time among adult individuals with a PBT. As noted above, this result is consistent with the report by Perla and colleagues in patients with glioblastoma, but inconsistent with the larger cancer literature as lower socioeconomic status,²⁸ including among individuals with lung,¹⁷ pancreatic,²⁹ ovarian,³⁰ primary central nervous system lymphoma,³¹ and among individuals who have delayed access to treatment,³² is usually associated with higher mortality and lower survival rates.^28,33 While PBTs are more common both in higher socioeconomic status countries² and counties within the United States,³⁴ we would have expected to see similar effects among our sample related to barriers in healthcare access and socioeconomic factors, leading to worse survival outcomes. In addition, living in less populated nonmetropolitan areas is associated with higher mortality rates across cancer types.²⁸ Rural residence, in particular, has been linked to decreased access to treatment for patients with a PBT, as well as worse survival outcomes.^35,36

We expected to see similar effects related to barriers in healthcare access related to socioeconomic factors that could lead to worse survival outcomes; however, we did not see these trends among our sample, either based on neighborhood disadvantage or nonurbanized/rural residence. This relationship may not have been reflected in our results as while 40% of our sample lived in nonurbanized areas, this measurement included both rural and suburban residents who may have had better access to health services than a solely rural patient population. Our sample may not have experienced the same barriers as rural dwellers in relation to being referred for and able to obtain specialty care.

There were no survival differences based on distance traveled to the NIH or population density, which may support the idea that most patients in this sample had the resources to travel longer distances to access treatment at the NIH. Other neuro-oncologic studies among patients with a PBT have reported mixed results on how socioeconomic status affects overall survival among patients with a PBT,^18,34,36,37 but one review found that studies with large samples may be more likely to report that patients with a PBT and lower socioeconomic status have shorter overall survival times.¹⁹ The smaller sample size of our study and the heterogenous PBT types included in our analysis do not mirror the tumor or sample characteristics of other PBT studies examining neighborhood disadvantage, which have largely consisted of patients with glioma or glioblastoma.^18,19,38 This may explain why individuals from less disadvantaged neighborhoods had shorter survival times. However, our subanalyses examining tumor grade and a subset of 242 patients with glioblastoma saw no differences in survival times among the two-neighborhood disadvantage groups.

Our sample’s survival time may also have been affected by the fact that all individuals with a PBT on the NOB-NHS were research study participants. Seventy-six percent of our sample consisted of individuals from less disadvantaged neighborhoods, which is a higher percentage than similar studies but may also reflect the pattern of who is referred to the NIH and the NOB-NHS from the greater United States. Finally, while we did not see differences in survival among our sample based on age or tumor grade, in the univariate analysis comparing those living in more or less disadvantaged neighborhoods the groups did differ based on sex, age, and tumor grades—although tumor grade differences were not statistically significant. Those living in more disadvantaged areas were more likely to be female, younger, and have lower-grade tumors, which are all factors associated with longer survival time among individuals with a PBT.² We may have been underpowered to detect important differences in survival time based on these variables.

Limitations

Limitations of this study include the descriptive and retrospective nature of the data, which means causation cannot be inferred. The results from our sample may not be generalizable to other individuals with a PBT outside of the United States as supportive and social programs available to patients with cancer and those with lower socioeconomic status vary by country, and ADI is a measure specific to the United States. Due to the cross-sectional nature of the data, we were unable to assess if patients’ residencies changed after diagnosis, and patients’ preferences in their residential areas before and after tumor diagnosis were not assessed. In addition, our sample consisted of a diverse group of tumors inclusive of both low- and high-grade PBTs, which are associated with differing survival times and treatment options. To address this limitation, we conducted sensitivity analyses to assess differences in outcomes based on tumor grade and among a subset of 242 patients with glioblastoma in the cohort. Importantly, as many PBTs are considered rare cancers, the inclusion of diverse tumor types which affect patient health-related quality of life is paramount to understand the experiences of individuals living with a PBT. Finally, the NIH is a unique treatment center where all patients are enrolled in a clinical trial or observational study, and all treatment costs are covered for associated care received at the NIH. Patients are referred to the NIH for research or clinical trials that may include additional treatment and second opinions that are not offered in their local areas. The generalizability of our results to those not enrolled in clinical or research studies may be limited, and our sample may not be representative of the population of patients with PBTs as a whole. For example, we found that the majority of those in the NOB-NHS were less disadvantaged, according to the ADI.

Conclusions

The field of Neuro-Oncology should consider how to reach more patients living in disadvantaged neighborhoods with diverse PBTs, including those that have malignant and nonmalignant tumors and patients with diversity in socioeconomic factors. Learning from research with other cancer populations that have already explored similar questions related to treatment access and neighborhood disadvantage may be helpful in determining how to increase the number of participants living in disadvantaged neighborhoods with a PBT in Neuro-Oncology clinical trials. Such strategies may include community-based recruitment, utilizing culturally relevant recruitment materials (ie, flyers and social media posts) and offering reimbursement for travel and parking.^39,40 Future studies examining neighborhood-level socioeconomic status among the PBT population should examine tumor types outside of glioblastoma and gliomas to determine how ADI may affect access and clinical outcomes among patients with these tumor types. In our study, we did not assess individual-level income as many participants prefered not to disclose their incomes. Future directions should further assess how neighborhood-level resources, such as those measured by the ADI, compare to strain related to changes in individual-level socioeconomic status because of the PBT diagnosis and treatment. Finally, among the PBT population, other social determinants of health affecting treatment access and patient-reported outcomes such as quality of life, symptom burden, cognitive, and mood-related symptoms, should be investigated to identify targets and sensitive outcome measures for interventions.

Supplementary material

Supplementary material is available online at Neuro-Oncology Practice (https://academic.oup.com/nop/).

Conflict of interest statement

None declared.

Funding

This research was supported in part by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, and a National Institutes of Health Center for Cancer Research Health Disparities Award. The Natural History Study project is supported by Intramural Project 1ZIABC011768-03 (T.S.A.). M.L.S. is a postdoctoral fellow supported by the National Cancer Institute’s Intramural Continuing Umbrella for Research Experiences (iCURE) program.

Authorship statement

Study contribution and design: M.L.S., J.B.V., O.C., Y.K., E.V., M.G., and T.S.A. Data collection: M.L.S., J.B.V., O.C., Y.K., Z.K., E.V., H.M., M.G., and T.S.A. Analysis and interpretation of results: M.L.S., J.B.V., O.C., Y.K., Z.K., E.V., H.M., M.G., and T.S.A. Draft manuscript preparation: M.L.S., J.B.V., O.C., Y.K., Z.K., E.V., H.M., M.G., and T.S.A.

Data Availability

Data are available upon reasonable request with appropriate privacy protections and data sharing agreements due to ethical restrictions.