Race, affordability and utilization of supportive care in ovarian cancer patients

Abstract

Objective:

Lack of access to supportive care (SC) among cancer patients have been well documented. However, the role of affordability in this disparity among ovarian cancer (OC) patients remain poorly understood.

Methods:

Patients with OC between 2008–2015 were identified from the SEER-Medicare dataset. Racial disparities in utilization of SC medications within the 6 months of OC diagnosis among patients with Medicare Part D coverage was examined. Multivariable log-binomial regression models were used to examine the associations of race, affordability and SC medications after adjusting for clinical covariates among all patients and separately among patients with advanced-stage disease.

Results:

The study cohort included 3,697 patients: 86% non-Hispanic White (NHW), 6% non-Hispanic Black (NHB), and 8% Hispanic. In adjusted models, NHB and Hispanic patients were less likely to receive antidepressants compared to NHW patients (NHB: aOR 0.46; 95% CI 0.33–0.63 and Hispanic: aOR 0.79; 95% CI 0.63–0.99). This association persisted for NHB patients with advanced-stage disease (aOR 0.42; 95% CI 0.28–0.62). Patients dual enrolled in Medicaid were more likely to receive antidepressants (overall: aOR 1.34; 95% CI 1.17–1.53 and advanced-stage: aOR 1.29; 95% CI 1.10–1.52). However, patients residing in areas with higher vs. lower proportions of lower educated adults (overall: aOR 0.82; 95% CI 0.70–0.97 and advanced-stage: aOR 0.82; 95% CI 0.68–0.99) were less likely to receive antidepressants.

Conclusions:

Black OC patients and those living in lower educated areas were less likely to receive antidepressants as SC. Given the importance of post-primary treatment quality of life for cancer patients, interventions are needed to enhance equitable access to SC.

INTRODUCTION

Ovarian cancer (OC) is the leading cause of gynecologic cancer deaths in the United States (US) and the fifth leading cause of cancer deaths among women.^1,2 Although advancements in treatment strategies have improved OC survival in the US over the last decade, ³ racial disparities persist. Black women have worse 5-year overall survival compared to White women (29.6% vs. 40.1%, respectively), in part attributed to Black women being less likely to receive guideline-recommended treatment.⁴

A key component of recommended treatment in OC involves adequate supportive care (SC) extending from diagnosis through post-treatment care, including the management of physical and psychological symptoms such as pain, fatigue, nausea, insomnia, poor appetite, and psychosocial stress. Recent reports highlight that lack of access to SC remains a challenge for OC patients, particularly in managing physical and psychological symptoms associated with cancer.^5–6 Patients often report that at least one SC need remains largely unmet after receipt of first-line treatment, which can contribute to decreased quality of life.⁷ Unmet needs may be even more pronounced in minoritized racial groups. For instance, among patients with stage IV breast cancer, Black women were less likely to receive treatment for depression, anxiety, and insomnia compared with White patients.⁸ Similar findings were evident in a neuro-oncology study showing decreased receipt of analgesics, antidepressants, and anxiolytics in Black patients with brain metastases compared to White patients.⁹

Healthcare affordability may be one contributor to the documented disparities in SC by race. Healthcare affordability is one of the five dimensions of healthcare access and refers to one’s ability to afford the cost of cancer care;¹⁰ it is measured using individual and area-level factors assessing income, education, occupation, and insurance status. Affordability has been well documented as a predictor of quality cancer treatment ^11–12 especially among low-income, uninsured, and under-insured cancer patients,^13–19 but no prior studies have examined the relationship between race, affordability, and SC in OC patients. In this analysis, we examine racial/ethnic disparities in utilization of SC among OC patients in the SEER-Medicare database, and analyze the role of healthcare affordability in contributing to this disparity.

METHODS

Study Population

This was a retrospective cohort study using SEER-Medicare (Surveillance, Epidemiology and End Results Program), a linkage of two population-based sources that includes data on Medicare beneficiaries with cancer. Medicare provides health insurance for adults ages 65 and older, as well as for younger individuals with disabilities or end-stage renal disease. We included women ages 65+ who were Non-Hispanic White (NHW), Non-Hispanic Black (NHB), or Hispanic and who were diagnosed with primary OC from 2008 to 2015. Given that nearly half (46%) of all patients with OC are diagnosed at age 65 or older ³, SEER-Medicare provides a robust population-based data source with sufficient data on OC patients to examine SC use in this population. Patients were required to have at least 12 months of continuous enrollment in Medicare fee-for-service parts A and B prior to the SEER diagnosis; at least one Medicare inpatient, outpatient, or carrier claim with an International Classification of Diseases- Clinical Modification (ICD-CM) diagnosis code for OC (ICD-9-CM and ICD-10-CM diagnosis codes 183.0 or C569) within two months of the SEER diagnosis; and continuous fee-for-service Medicare enrollment in the 12 months following their diagnosis date, or until death, whichever came first. Patients were also required to have continuous enrollment in Medicare Part D for prescription medications six months following their diagnosis date or until death to assess receipt of medications. Patients were excluded if they had missing values in measures of affordability variables.

Receipt of Supportive Care Medications

Patient receipt of a SC medication was determined by Medicare Part D Prescription Drug Event claim in the six months following OC diagnosis for one of the following drug classes: antidepressants/anxiolytics, anti-fatigue medications, analgesic medications, or anti-nausea medications. Included drugs are detailed in Appendix A. Receipt of SC medications were also examined by the following drug subclasses: for antidepressants/anxiolytics: Selective Serotonin Reuptake Inhibitors (SSRIs), Selective Norepinephrine Reuptake Inhibitors (SNRIs), Monoamine oxidase inhibitors (MAOIs), atypical antidepressants, tricyclic antidepressants, and serotonin modulators; for anti-fatigue medication: psychostimulant; and for analgesics: opioid or non-opioid analgesics. The date of a patient’s earliest Prescription Drug Event claim for any SC medication served as the first date of first receipt of a SC medication.

Measures of Healthcare Affordability

Multiple measures of healthcare affordability were available in SEER-Medicare, including dual enrollment in Medicaid in the 12 months prior to OC diagnosis; census tract-level measures of socioeconomic status (SES) of the patient’s residential census tract at the time of diagnosis (median household income, percentage of adults ages 25+ with less than a high school education, and percentage of households with incomes below the poverty level); and whether the patient’s primary treatment hospital qualified for disproportionate share hospital (DSH ) payments. DSH payments are made by Medicaid to qualifying hospitals that serve a large number of Medicaid and uninsured individuals. Census tract SES characteristics were categorized into quartiles. Additional linkage was conducted to assess county-level health insurance coverage as another indicator of affordability. FIPS codes for the patient’s county and state of residence and the patient’s year of diagnosis were used to link the US Census Bureau’s Small Area Health Insurance Estimates ²⁰ to obtain the estimated percentage of county residents without health insurance in the year of the patient’s diagnosis.

Patient Demographic and Clinical Characteristics

We examined patient and clinical characteristics from SEER data, including race (NHW, NHB, or Hispanic), age at diagnosis, year of diagnosis, cancer stage at diagnosis, histology (Type I Epithelial, Type II Epithelial, Other),²¹ marital status, geographic region of residence at diagnosis (West, Northeast, Midwest, South, Other/Unknown), and residence in a metropolitan area. We used validated coding algorithms to assess patient comorbidities and to calculate the Charlson Comorbidity Score in the 12 months prior to OC diagnosis using diagnosis codes (Appendix B) from inpatient, outpatient, and carrier Medicare claims files.^22,23

Statistical Analysis

Descriptive analyses using chi-square tests, Wilcoxon rank-sum tests, and Kruskal-Wallis tests were performed to describe demographic and clinical characteristics by patient race/ethnicity. Unadjusted associations between patient race/ethnicity and utilization of SC medications in the six months following OC diagnosis were also examined in a sub-cohort of OC patients with advanced-stage at diagnosis (stage III or IV). Given the clinically aggressive nature of advanced-stage cancer, we hypothesized that this sub-cohort would have a greater symptom burden compared to patients with stage I or II OC. We estimated prevalent depression in the 12 months prior to OC diagnosis and incident depression in the 6 months after OC diagnosis using an established claims-based algorithm.²⁴

Multivariable log-binomial regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) to assess the associations between patient race/ethnicity and affordability measures (dual enrollment in Medicaid, median household income, low-educated census tract, low-income census tract, percentage in county without health insurance, and receiving treatment at a hospital eligible for disproportionate share payments) with receipt of SC medications in the six months following OC diagnosis.. Models were conducted first with adjustment for clinical and demographic characteristics only (age at diagnosis, stage at diagnosis, tumor histology, patient comorbidities, marital status, geographic region of residence, and metropolitan residence) and subsequently with additional adjustment for the affordability measures. All regression analyses were additionally conducted stratified by race and also limited to patients who were diagnosed at stages III/IV only. Stratified analysis was used to test potential effect modification for affordability measures by patient race/ethnicity and geographic region. Collinearity in the model variables was assessed using the variance inflation factor (VIF) method. Statistical significance was established using two-sided tests with an alpha level of 0.05. All statistical analyses were conducted using SAS version 9.4 (Cary, NC, USA).

RESULTS

Study Population and clinical characteristics

The study cohort included 3,697 OC patients ages 65 years or older diagnosed between 2008 and 2015. Overall, 86% of patients were NHW, 6% were NHB, and 8% were Hispanic (Table 1). The majority (68%) of patients were diagnosed with advanced-stage OC (stage III/IV). There was no significant difference in mean age at diagnosis; however, NHB patients had a higher comorbidity burden compared to NHW and Hispanic patients.

Table 1.

Baseline clinical and sociodemographic characteristics of OC patients by race (N=3,697)

Variable	NH-White	NH-Black	Hispanic	p-value*
N (%)	3,188	227	282
Age at OC diagnosis, Mean (SD)	75.8 (6.7)	75.0 (5.8)	75.1 (6.2)	.17
Year of diagnosis				.74
2008	410 (12.9%)	25 (11.0%)	37 (13.1%)
2009	363 (11.4%)	28 (12.3%)	32 (11.3%)
2010	367 (11.5%)	26 (11.5%)	33 (11.7%)
2011	311 (9.8%)	31 (13.7%)	32 (11.3%)
2012	384 (12.0%)	29 (12.8%)	37 (13.1%)
2013	426 (13.4%)	34 (15.0%)	33 (11.7%)
2014	449 (14.1%)	32 (14.1%)	41 (14.5%)
2015	478 (15.0%)	22 (9.7%)	37(13.1%)
Tumor stage at dx				.09
I	497 (15.6%)	33 (14.5%)	38 (13.5%)
II	259 (8.1%)	19 (8.4%)	21 (7.4%)
III	1,329 (41.7%)	79 (34.8%)	109 (38.7%)
IV	824 (25.8%)	79 (34.8%)	81 (28.7%)
Unknown	279 (8.8%)	17 (7.5%)	33 (11.7%)
Histology				< .001
Type I epithelial	462 (14.5%)	25 (11.0%)	40 (14.2%)
Type II epithelial	2,606 (81.7%)	176 (77.5%)	230 (81.6%)
Other	120 (3.8%)	26 (11.5%)	12 (4.3%)
Married	1,449 (45.5%)	40 (17.6%)	90 (31.9%)	< .001
Geographic region				< .001
Midwest	439 (13.8%)	30 (13.2%)	<11
Northeast	714 (22.4%)	64 (28.2%)	45 (16.0%)
South	457 (14.3%)	68 (30.0%)	<11
Unknown/NA	302 (9.5%)	26 (11.5%)	<11
Residence in metropolitan area	2,630 (82.5%)	205 (90.3%)	262 (92.9%)	< .001
Median Charlson Comorbidity Score (IQR)	2.0 (1.0, 3.0)	3.0 (2.0, 5.0)	2.0 (1.0, 4.0)	< .001
Patient comorbidities
Myocardial infarction	140 (4.4%)	19 (8.4%)	<11	.01
Hypertension	2,381 (74.7%)	211 (93.0%)	224 (79.4%)	< .001
Peripheral vascular disease	543 (17.0%)	59 (26.0%)	50 (17.7%)	.003
Congestive heart failure	358 (11.2%)	53 (23.3%)	43 (15.2%)	< .001
Dementia	65 (2.0%)	<11	<11	.25
Cerebrovascular disease	458 (14.4%)	39 (17.2%)	28 (9.9%)	.05
Chronic obstructive pulmonary disease	766 (24.0%)	70 (30.8%)	68 (24.1%)	.07
Rheumatologic disease	208 (6.5%)	15 (6.6%)	27 (9.6%)	.15
Peptic ulcer disease	88 (2.8%)	<11	<15	.21
Mild liver disease	481 (15.1%)	37 (16.3%)	56 (19.9%)	.10
Renal disease	273 (8.6%)	45 (19.8%)	26 (9.2%)	< .001
Any Diabetes	763 (23.9%)	110 (48.5%)	121 (42.9%)	< .001
Diabetes with complications	169 (5.3%)	42 (18.5%)	49 (17.4%)	< .001
Hemiplegia or paraplegia	18 (0.6%)	<11	<11	.26

^*determined using chi-square test

Column percentages used

NH-White = Non-Hispanic White

NH-Black = Non-Hispanic Black

SD = standard deviation

IQR = interquartile range

OC = ovarian cancer

Bold indicates p<0.05

Healthcare affordability

At the time of diagnosis, 57% of Hispanic patients and 53% of NHB patients were dual enrolled in Medicaid compared to 14% of NHW patients (Table 2). About 20% of NHW patients were in the lowest income quartile, while the majority of NHB patients (63%) and nearly half of Hispanic patients (47%) were in the lowest income quartile. Additionally, NHB (50%) and Hispanic (51%) patients were more than twice as likely as NHW (21%) patients to live in a census tract in the highest quartile of adult residents without a high school degree. Similarly, compared to NHW patients, NHB and Hispanic patients were more likely to live in a census tract in the highest quartile of households below poverty and in counties with a higher average percentage of uninsured residents.

Table 2.

Baseline patient measures of healthcare affordability at time of ovarian cancer diagnosis by patient race (N=3,697)

Variable	NH-White	NH-Black	Hispanic	p-value*
N (%)	3,188 (86.2)	227 (6.1)	282 (7.6)
Dual enrolled in Medicaid	435 (13.6)	120 (52.9)	161 (57.1)	< 0.001
Zip code median household income				< 0.001
$5,299 – $26,469	641 (20.1)	142 (62.6)	132 (46.8)
$26,470 – $36165	804 (25.2)	48 (21.1)	53 (18.8)
$36,166 – $50,838	808 (25.3)	23 (10.1)	52 (18.4)
$50,839 – $200,014	935 (29.3)	14 (6.2)	45 (16.0)
Zip code in highest quartile: Adults 25+ < high school education	676 (21.2)	113 (49.8)	144 (51.1)	< 0.001
Zip code in highest quartile: households below poverty line	653 (20.5)	133 (58.6)	117 (41.5)	< 0.001
County: % residents without health insurance (SD)	13.8 (5.1)	15.7 (4.9)	16.1 (5.2)	< 0.001
Hospital eligibility for disproportionate share payments	2,387 (74.9)	195 (85.9)	242 (85.8)	< 0.001

^*determined using Wilcoxon rank sum tests

NH-White = Non-Hispanic White

NH-Black = Non-Hispanic Black

Median household income measured at census tract level. Education and poverty measured at census tract level in highest quartile. Average percentage of residents without health insurance measured at county level.

Column percentages used

IQR = interquartile range

SD = standard deviation

Bold indicates p<0.05

Supportive care utilization

Approximately 89.5% of NHW, 87.2% of NHB and 90.1% received at least one SC medication within six months of OC diagnosis; there was no statistically significant difference by race (Table 3). In unadjusted analyses, a lower proportion of NHB (15%) and Hispanic (25%) patients received antidepressant/anxiolytic medication, specifically serotonin-selective reuptake inhibitors (SSRI), than NHW (30%) patients (Table 3). No difference was observed with the use of medication for fatigue, but Hispanic (83%) and NHB (79%) patients were more likely to receive analgesic therapy compared to NHW (76%) patients. NHB patients were less likely to receive anti-nausea medications (48.5%) compared to Hispanic (59.6%) and NHW patients (58.6%) (Table 3). Among patients with advanced-stage disease, similar patterns were observed; there was no racial difference overall in SC medication, but NHB (12%) and Hispanic (25%) patients were less likely to receive antidepressants compared to NHW patients (31%) (Appendix C). To better contextualize differences in receipt of antidepressants by race, we estimated depression diagnoses in claims data: no significant differences were observed in rates of prevalent depression in the 12 months prior to OC diagnosis (NHW: 13.0% vs. NHB: 12.3% vs. Hispanic: 11.0%; p=.61), but significantly lower incident depression in the 6 months after OC diagnosis was observed in NHB patients (7.9% vs. NHW: 14.6% vs. Hispanic: 15.6%; p=.02).

Table 3.

Prevalence of supportive care utilization within 6 months of OC diagnosis by race (N=3,697)

	NH-White N = 3,188 (%)	NH-Black N = 227 (%)	Hispanic N = 282 (%)	p-value*
Any supportive care medication	2,852 (89.5%)	198 (87.2%)	254 (90.1%)	.53
Any Antidepressants	942 (29.5%)	33 (14.5%)	69 (24.5)	<0.001
SSRI	650 (20.4)	23 (10.1)	44 (15.6)	<0.001
SNRI	149 (4.7)	<11	15 (5.3)	.10
MAOI	<11	<11	<11	.79
Atypical	107 (3.4)	<11	<11	.84
Tricyclic	37 (1.2)	<11	<11	.55
Serotonin modulator	107 (3.4)	<11	<11	.96
Fatigue medications
Psychostimulants	17 (0.5)	<11	<11	.26
Any Analgesics	2,432 (76.3)	179 (78.9)	233 (82.6)	.04
Opioids	2,309 (72.4)	169 (74.4)	220 (78.0)	.11
Non-opioids	2,176 (68.3)	161 (70.9)	214 (75.9%)	.02
Any anti-nausea medications	1,867 (58.6%)	110 (48.5%)	168 (59.6%)	.01

^*determined using Kruskal Wallis tests

SSRI = selective serotonin reuptake inhibitor

SNRI = serotonin-norepinephrine reuptake inhibitor

MAOI = monoamine oxidase inhibitor

Bold indicates p<0.05

Healthcare affordability and utilization of supportive care medication

In multivariable log-binomial regression, no associations were found between measures of affordability and overall utilization of SC medications, anti-fatigue medications, analgesics, or anti-nausea medications in analyses of all OC patients with early or advanced-stage disease; differences in receipt by race were explained by differences in patient clinical characteristics (data not shown). However, differences in use of antidepressants/anxiolytics by race and affordability were observed (Table 4). In fully adjusted models, NHB patients and Hispanic patients were 54% and 21%, respectively, less likely to receive antidepressants compared to NHW patients (NHB: aOR 0.46; 95% CI 0.33–0.63 and Hispanic: aOR 0.79; 95% CI 0.63–0.99). Among advanced-stage patients, this association persisted for NHB patients compared to NHW patients (NHB: aOR 0.42; 95% CI 0.28–0.62); however, there was no significant difference between Hispanic and NHW patients. Further, OC patients dual enrolled in Medicaid were 34% more likely to receive antidepressants (aOR 1.34; 95% CI 1.17–1.53) than those not dual enrolled. This association persisted among the sub-cohort of advanced-stage patients dual enrolled in Medicaid (aOR 1.29; 95% CI 1.10–1.52). There was also an association between census tract-level education and receipt of antidepressants; both all-stage and advanced-stage OC patients living in areas with the highest proportion of adults with low educational attainment were 18% less likely to receive antidepressants relative to patients in areas with higher educational attainment (all: aOR 0.82; 95% CI 0.70–0.97 and advanced-stage: aOR 0.82; 95% CI 0.68–0.99). There was no association between use of antidepressants and other measures of affordability, including median household income, percentage of households in poverty, percentage of uninsured residents, and hospital eligibility for disproportionate share payments. No model covariates were excluded due to collinearity, and no significant differences in associations by race were observed in stratified models (data not shown), though numbers of patients in NHB-only and Hispanic-only models were small and the analyses were likely underpowered. There was also no evidence of effect modification by geographic region (data not shown).

Table 4:

Association^a of race and healthcare affordability with receipt of antidepressants within 6 months of diagnosis

	All Stages (N = 3,697)		Advanced Stage (N = 2,501)
Parameter	Model 1	Model 2	Model 3	Model 4
	Clinicalb	Clinicalb + Affordability	Stage 3/4 Clinicalb	Stage 3/4 Clinicalb+ Affordability
Race
NH-White (Ref)	-	-	-	-
NH-Black	0.46 (0.33–0.63)	0.46 (0.33–0.63)	0.37 (0.24–0.56)	0.42 (0.28–0.62)
Hispanic	0.82 (0.66–1.02)	0.79 (0.63–0.99)	0.82 (0.63–1.06)	0.80 (0.61–1.03)
Dual enrolled in Medicaid		1.34 (1.17–1.53)		1.29 (1.10–1.52)
Median household income
$50,839 – $200,014 (Ref)	-	-	-	-
$36,166 – $50,838		1.02 (0.88–1.18)		1.08 (0.92–1.27)
$26,470 – $36,165		0.88 (0.75–1.03)		0.90 (0.75–1.08)
$5,299 – $26,469		0.94 (0.78–1.13)		0.96 (0.77–1.19)
Adults 25+ < high school education
Lowest 3 quartiles (Ref)
Highest quartile		0.82 (0.70–0.97)		0.82 (0.68–0.99)
Households in poverty
Lowest 3 quartiles (Ref)
Highest quartile		0.93 (0.80–1.09)		0.99 (0.83–1.19)
% without health insurance		1.01 (0.99–1.02)		1.01 (1.00–1.03)
Hospital eligibility for disproportionate share payments		1.07 (0.94–1.22)		1.07 (0.93–1.24)

^aLog-binomial regression analysis modeling receipt of any supportive care medication in the 6 months following ovarian cancer diagnosis.

^bAll models adjusted for age, stage, histology, marital status, geographic region of residence, metropolitan residence, year of diagnosis, and patient comorbid conditions

NH-White = Non-Hispanic White; NH-Black = Non-Hispanic Black

Median household income, education and poverty measured at census tract level. Average percentage of residents without health insurance measured at county level. Advanced stage defined as stage III-IV at diagnosis

Adjusted odds ratios (aOR) and 95% CI reported

DISCUSSION

In this retrospective cohort study, we evaluated the association between race/ethnicity and affordability with utilization of SC among older women with OC in SEER-Medicare. In adjusted analyses, we found no associations between race/ethnicity or healthcare affordability with receipt of anti-fatigue medications, analgesics, or anti-nausea medications. However, significant associations between various measures of affordability and receipt of antidepressants were observed; patients who were dual enrolled in Medicaid were more likely to receive antidepressants, and patients living in areas with low educational attainment were less likely to receive antidepressants. These associations persisted among patients with advanced-stage OC. Additionally, NHW patients were more likely to receive antidepressants than NHB and Hispanic patients, even after adjusting for measures of healthcare affordability. Notably, nearly 1 in 3 NHW patients received antidepressants, compared to only roughly 1 in 7 NHB patients.

We observed that dual enrollment in both Medicare and Medicaid was associated with greater likelihood of receiving antidepressants. Dual enrollment is traditionally an indicator of low healthcare affordability as Medicare patients who qualify for full Medicaid benefits are required to meet strict low-income and poverty thresholds.²⁵ Prior studies have documented associations between dual enrollment and poor outcomes in gynecologic cancer.^26,27 However, our findings of greater utilization of anti-depressants in this population highlights dual eligibility may be enhancing access to SC via additional associated benefits (e.g., assistance with premiums and medication co-pays) that contributes to higher affordability for these patients.²⁵ Future studies should explore the mechanisms by which dual eligibility may enhance access to anti-depressants for cancer patients.

Prior studies have reported racial differences in receipt of SC among other cancer types. For instance, among breast cancer patients, Black women were less likely to use psychotropic medications to treat conditions such as depression, anxiety, and insomnia.⁸ Similarly, racial differences have been documented among patients with lung cancer and brain cancer metastases.^9,28 Our findings among OC patients were similar, highlighting a racial difference in the use of psychotropic medications that was even stronger among advanced-stage patients. Historically, patients from minority racial/ethnic groups have been less likely to utilize mental health services. In addition to lack of access, this disparity may be due to culturally informed attitudes, negative perceptions, and stigma related to mental health conditions and their treatment.^29–31 Lack of formal diagnoses, or misdiagnoses by healthcare providers may also contribute; though prevalent depression prior to OC diagnosis did not significantly differ by race, incident depression rates after OC diagnosis were lower in NHB compared to NHW patients, a pattern that is also observed in the general population.^32–35 While we found that NHB and Hispanic patients were more likely to have a dual enrollment in Medicaid, live in areas with higher rates of lower-educated adults, and have lower affordability measures, racial/ethnic differences in receipt of antidepressants persisted in fully adjusted models despite controlling for affordability measures. In addition to cultural attitudes and stigma regarding mental health and differential diagnosis of depression by race, other domains of healthcare access (e.g., availability of mental health specialists and accessibility to those specialists) may contribute to racial differences in SC utilization and should be further studied.

Notably, we did not observe differences by race in receipt of analgesics after adjusting for clinical characteristics. Inadequate management of cancer-related pain in ethnic minorities has been documented in other studies, highlighting a lower likelihood of receiving analgesics, or receiving doses that were less potent than guideline standards and thus more likely to be under-medicated for their pain.^36–41 Our study differs from these previous studies in type of malignancy and the sex of the population studied. Furthermore, the previous studies account for other methods of pain relief in cancer care including radiation and spinal procedures which were not examined in our study; it is possible that there are differences in rates of receipt of other methods of pain relief that our study does not capture.

There are several strengths of this analysis. We used a high-quality data source with a relatively large sample size to better understand delivery of SC to OC patients--a research area not previously examined but crucial to improving cancer care. SC utilization was defined based on Medicare claims, limiting potential exposure misclassification due to recall bias. Our study also has several limitations. To account for oral prescription medications, our study only included patients continuously enrolled in Medicare Part D— reducing our sample size and excluding approximately 43% of patients, some of whom may have had lower healthcare affordability. Our findings may not generalize to younger OC patients or patients outside of SEER’s geographic catchment areas.⁴² In addition, our study population is limited to NHW, NHB, and Hispanic OC patients, which limits our ability to make conclusions about other racial/ethnic groups. The SEER-Medicare database cannot fully capture the clinical decision-making involved in screening patients before prescribing various SC medications or patient preferences for treatment. Though we were able to estimate the prevalence of depression using a claims—based algorithm, this method is less sensitive and may underestimates true rates of diagnoses.⁴³ Furthermore, we did not evaluate the limits to patients’ out-of-pocket Medicare Part D spending over a specific time frame, such as only receiving a fixed number of prescriptions every month or every year. Although beyond the scope of this analysis, a better understanding of these factors will better illustrate patients’ financial burdens in affording prescription medication, particularly in the context of other non-cancer comorbidities.

In conclusion, substantive racial differences exist in antidepressant SC utilization among older OC patients, and while some measures of healthcare affordability are significant predictors of this outcome, affordability does not fully explain the observed racial disparity. Future studies are needed to evaluate perceived unmet SC needs using metrics that assess pain, psychosocial distress, fatigue, and nausea, while further characterizing associations between these measures, affordability and other healthcare access domains by race. A deeper understanding of these relationships can lead to strategies that mitigate financial barriers and ensure affordable SC treatment for all OC patients.

Figure 1. Prevalence of antidepressant use and analgesic for all stage and late-stage ovarian cancer patients, by race.

Late-stage defined as stage III-IV at diagnosis.

Novelty and Impact:

We evaluated the role of healthcare affordability in utilization of supportive care medications for depression, fatigue, pain, and nausea among Non-Hispanic White, Non-Hispanic Black, and Hispanic ovarian cancer patients. Non-Hispanic Black patients and those living in lower educated areas were less likely to receive antidepressants, highlighting a critical need for clinical strategies to address this gap.

KEY MESSAGE.

Racial differences exist in antidepressant SC utilization among older OC patients, and healthcare affordability does not fully explain the observed disparity. A deeper understanding of the mechanisms contributing to these relationships can inform strategies to mitigate financial barriers and ensure equitable SC treatment for all OC patients.

Abbreviations:

95% CI

95% confidence interval

aOR

adjusted odds ratio

OC

ovarian cancer

SC

supportive care

NHW

Non-Hispanic White

NHB

Non-Hispanic Black

Data Availability Statement:

The SEER-Medicare database is owned and managed by the National Cancer Institute. Information on how to obtain these data is available at: https://healthcaredelivery.cancer.gov/seermedicare/obtain/