Factors associated with delaying and forgoing care due to cost among long-term, Appalachian cancer survivors in rural North Carolina

ABSTRACT

Background:

Little research exists on delayed and forgone health and mental health care due to cost among rural cancer survivors.

Methods:

We surveyed survivors in 7 primarily rural, Appalachian counties February to May 2020. Univariable analyses examined the distribution and prevalence of delayed/forgone care due to cost in the past year by independent variables. Chi-square or Fisher’s tests examined bivariable differences. Logistic regressions assessed the odds of delayed/forgone care due to cost.

Results:

Respondents (n = 428), aged 68.6 years on average (SD: 12.0), were 96.3% non-Hispanic white and 49.8% female; 25.0% reported delayed/forgone care due to cost. The response rate was 18.5%. The proportion of delayed/forgone care for those aged 18–64 years was 46.7% and 15.0% for those aged 65 + years (P < 0.0001). Females aged 65 + years (OR: 2.00; CI: 1.02-3.93) had double the odds of delayed/forgone care due to cost compared to males aged 65 + years.

Conclusion:

About one in four rural cancer survivors reported delayed/forgone care due to cost, with rates approaching 50% in survivors aged <65 years.

Impact:

Clinical implications indicate the need to: 1) ask about the impact of care costs, and 2) provide supportive services to mitigate effects of treatment costs, particularly for younger and female survivors.

KEYWORDS:

• Financial hardship
• cancer survivorship
• rural
• cost of care
• Appalachian region

Introduction

Cancer treatment costs in the US increased by 27% from 2010 to 2020 and are expected to surge to more than $245 billion by 2030 with over $183 billion in direct spending attributed to cancer care costs of which cancer survivors pay over $5.6 billion in out-of-pocket expenses [1,2]. Studies estimate that between 7-11% of survivors forgo any type of medical care due to cost [3–6]. Compared to individuals without a cancer history, cancer survivors (including long-term survivors) more often forgo care due to cost despite higher rates of insurance coverage [7,8]. Forgone medical and mental health care (i.e. not receiving provider recommended care) may complicate cancer treatment, negatively impact survival, and worsen quality of life [9,10]. Rural survivors, who account for an estimated 21% of the survivor population in the US, are more likely to lack insurance, forgo post-diagnosis health care due to cost, and report poorer health outcomes compared to urban survivors [11,12]. However, rural populations are not homogeneous, and little research exists assessing delayed and forgone care in varying populations of rural cancer survivors.

Appalachia is a diverse region spanning 13 states incorporating rural and urban communities in 5 distinct subregions [13]. The population in Appalachia and the counties included in this study exceed the national average for individuals aged 65 + and have household incomes that are approximately 20% lower than the national average with a 2% higher prevalence of residents living below the poverty line compared to the rest of the US population [13]. National research notes variation in cancer outcomes with the Appalachian region representing some of the highest incidence and mortality rates in the US [14–17].

Increased treatment costs and related adverse outcomes for cancer patients and their families have received substantial attention leading to varying definitions and terms (e.g. financial hardship, burden, and toxicity) measuring material, psychological, and behavioral aspects of this phenomenon [18–21]. Studies have also reported similar outcomes in multiple, high-income international settings with nearly universal healthcare access [22]. De Souza et al.’s COmprehensive Score for financial Toxicity (COST) measure is extensively used in cancer survivorship studies [23,24], yet it predicates responses on the last 7 days and does not assess forgone or delayed care over the past 12 months like the National Health Interview Survey (NHIS) items [25]. Addressing care continuity due to financial barriers represents an actionable point of intervention beyond the subjective experience of these barriers.

This study sought to assess delayed and forgone medical and mental health care due to cost in cancer survivors, those 6 + months post-definitive treatment, residing in priority, mostly rural, Appalachian counties of our cancer center’s catchment area. Using data from a mailed survey of cancer survivors in our electronic medical record (EMR), we aimed to: 1) determine the rate of delayed and forgone care due to cost in this sample, and 2) examine factors associated with these outcomes.

Materials and methods

Design

Our team developed a cross-sectional survey to assess engagement with physical and mental health care among mostly rural, long-term cancer survivors drawing from several standardized instruments including the NHIS [26]. The Wake Forest Health Sciences Institutional Review Board reviewed and approved this study prior to data collection or analysis (IRB00056939).

Participants

We surveyed survivors with current addresses in 7 priority, primarily rural, Appalachian counties with elevated smoking and cancer incidence rates compared to other counties in the catchment area. These counties are entirely or mostly comprised of rural census tracts corresponding to Rural-Urban Commuting Area (RUCA) Codes of 4–10 (nonmetro) as classified by the US Department of Agriculture [27]. Patients were identified through our EMR. Potential participants could have had a cancer diagnosis at any point in time and did not have to be diagnosed or treated at one of our facilities. Regular attendance at clinical appointments was not a requirement, but they had to have attended an appointment of any type at one of our facilities. Eligibility criteria included: 1) adults aged 18 + years; 2) with a cancer diagnosis (other than non-melanoma skin cancer); 3) who were 6 + months post-definitive treatment (e.g. surgery, chemotherapy, and/or radiation) or were receiving on-going maintenance therapy (e.g. hormonal therapies and immunotherapies); and 4) lived in an area with RUCA code of 4–10 and/or residents of the 7 priority counties [28]. Individuals were ineligible if they were unwilling to participate or could not read or understand English as the survey was not available in any other language.

Data collection

This survey, an introduction letter describing the purpose and procedures of the study, and a postage paid envelope were included in a brightly colored packet mailed to potential participants between February and March 2020. Respondents completed the survey online or returned the completed survey in the envelope provided between February to May 2020. We classified non-responders as those who did not complete the survey and whose packet was not returned by the post office as undeliverable. We identified 44 individuals (data not shown) who were deceased after a secondary review of the EMR or notification from packet recipients. Potential participants were randomly assigned to two incentive strategy arms prior to mailing the survey packets [29]. Survey packets in Arm 1 included a $2 bill, and respondents could opt into a drawing for one of five $50 gift cards upon survey completion. Arm 2 provided respondents with a $10 gift card upon survey completion.

Measures

Outcome variable

NHIS items [26] assessed delayed and forgone health and mental health care due to cost with the following: “During the past 12 months, was there any time when you needed medical care, but did not get it because you couldn’t afford it?”; “During the past 12 months, have you delayed seeking medical care because of worry about the cost?”; “During the past 12 months, was there any time when you needed mental health care or counseling but didn’t get it because you couldn’t afford it?”; and “During the past 12 months, was there a time when you needed a medical test but didn’t get it because you couldn’t afford it?” Responses were coded as yes (yes, for my cancer or yes, but not for my cancer) and no for each variable. A combined outcome variable classified participants as reporting any positive response to delayed and/or forgone care due to cost items and dichotomized as yes or no.

Sociodemographic variables

Items from the Health Information National Trends Survey (HINTS) assessed: age in years, sex (options included female, male, intersex, transgender, prefer not to answer), ethnicity (Hispanic, Latino/a, or Spanish origin), and race (white, black, American Indian/Alaska Native, Asian, Native Hawaiian/other Pacific Islander, or other) [30]. Age was categorized into those aged 18–64 and those aged 65 + given the differences in health insurance in these groups, where those 65 + are nearly universally insured through Medicare in the US. Responses to race/ethnicity were combined into a single variable with categories including non-Hispanic white and all others due to small sample sizes. Marital status was dichotomized as (1) those married or living as married, and (2) those who were not married (divorced, widowed, separated, or single, never been married). Education level options were recoded as: (1) less than high school, (2) a high school graduate, (3) some college or vocational school, or (4) college graduate. Total household income was categorized in ranges including <$20,000; $20,000-$34,999; $35,000-$49,999; $50,000-$74,999; $75,000+; or missing/unknown.

Geographic variable

A RUCA code was assigned to each respondent based on their address and corresponding census tract.

Cancer variables

Participants selected their cancer site(s) from a list that was aggregated to: breast, colorectal, gynecological, hematological, lung, melanoma, prostate, other solid tumor/unknown, or multiple sites. These categories were ranked by incidence with the top three cancer sites (breast, prostate, colorectal), multiple cancer sites, and all other single types recoded to “other/unknown”. Respondents provided the month and year of their diagnosis to calculate a time since diagnosis, dichotomized as <5 years, or 5 + years. Survivors reported their cancer treatment status as planned, current, finished, or unknown, and maintenance therapy status (yes, no, or other/unknown).

Health literacy variable

Adequate health literacy was assessed with the following: “how confident are you filling out medical forms by yourself?” using a validated measure [31]. The responses were grouped into adequate health literacy (extremely/quite a bit) and inadequate health literacy (somewhat/a little bit/not at all) categories.

Patient Reported Outcomes Measurement Information System (PROMIS) variables

PROMIS is a National Institutes of Health (NIH)-funded project that provides item banks for valid and reliable measures of mental, physical, and social patient reported outcomes (PROs) in the general population and among individuals with chronic conditions [32]. Our survey included items from the PROMIS Global-10 that measured the general mental and physical health of our respondents with standardized scores (T-scores) where higher values indicate better overall mental and physical wellbeing. We also included a pain intensity score that included one item asking: “in the past 7 days, how would you rate your pain on average?” with responses ranging from 0 (no pain at all) to 10 (worst imaginable pain).

Statistical analysis

Univariable analyses examined the sample distribution and prevalence of delayed and forgone care by sociodemographic, cancer-related, health literacy, and PROMIS variables. We used Chi-square or Fisher’s exact tests to compare the proportional differences among these variables by age group and delayed/forgone care. Variables with P-values of less than 0.2 were included in the regression models. Multivariable logistic regressions assessed the odds of delaying/forgoing health and mental health care. Tests of statistical significance were two-sided with a P-value of 0.05. All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results

Sample distribution and stratification by age group

The sample consisted of 428 survey respondents, with a mean age of 68.6 years (standard deviation [SD]: 12.0) and an average time since first cancer diagnosis of 9.9 years (SD: 8.3). The response rate of eligible surveys was 18.5% (data not shown). Female respondents represented 49.8% of the sample, and 96.3% identified as non-Hispanic white, reflective of the racial/ethnic distribution of the general population in this area (Table 1). Respondents were divided primarily between micropolitan areas (populations less than 49,999) with RUCA codes of 4–6 (52.1%) and small towns (populations less than 9,999) with RUCA codes of 7–10 (43.2%); a small proportion of respondents living in one of the priority counties had addresses in metropolitan census tracts (urbanized areas) with RUCA codes of 1–2 (4.7%). Approximately one third of participants (34.6%) were college graduates, and 9.8% had less than a high school education. Respondents had diverse household income levels with 17.8% reporting <$20,000, and 23.8% disclosed household incomes of $75,000 + . Over two-thirds of the sample (67.1%) reported being married or living as married.

Table 1. Sample distribution of respondents stratified by age group (N = 428).

	Full sample		Aged 18–64 years		Aged 65 + years
	n	%	n	%	n	%
Age (years)
18–64	135	31.5	135	100.0	–	–
65+	293	68.5	–	–	293	100.0
Sex
Female	213	49.8	83	61.5	130	44.4
Male	215	50.2	52	38.5	163	55.6
Race/ethnicity
Non-Hispanic white	412	96.3	126	93.3	286	97.6
Other	16	3.7	9	6.7	7	2.4
Rural Urban Commuting Area Code
1–2	20	4.7	8	5.9	12	4.1
4–6	223	52.1	71	52.6	152	51.9
7–10	185	43.2	56	41.5	129	44.0
Education level
<High School	42	9.8	13	9.6	29	9.9
High School	110	25.7	30	22.2	80	27.3
Some college or vocational	128	29.9	42	31.1	86	29.4
College graduate	148	34.6	50	37.0	98	33.5
Household Income level
<$20,000	76	17.8	27	20.0	49	16.7
$20,000-$34,999	64	15.0	17	12.6	47	16.0
$35,000-$49,999	48	11.2	17	12.6	31	10.6
$50,000-$74,999	56	13.1	24	17.8	32	10.9
$75,000+	102	23.8	36	26.7	65	22.5
Missing/unknown	82	19.2	14	10.4	68	23.2
Marital status
Married or living as married	287	67.1	89	65.9	198	67.6
Not married	141	32.9	46	34.1	95	32.4
Cancer site
Breast	86	20.1	32	23.7	54	18.4
Prostate	92	21.5	12	8.9	80	27.3
Colorectal	31	7.2	12	8.9	19	6.5
Multiple sites	67	15.7	14	10.4	53	18.1
Other/Unknown	152	35.5	65	48.2	87	29.7
Time since diagnosis
<5 years	133	31.1	50	37.0	83	28.3
5 + years	295	68.9	85	63.0	210	71.7
Cancer treatment status
Planned	27	6.3	2	1.5	25	8.5
Current	45	10.5	11	8.2	34	11.6
Finished	339	79.2	120	88.9	219	74.7
Other/Unknown	17	4.0	2	1.5	15	5.1
Maintenance therapy
Yes	65	15.2	28	20.7	37	12.6
No	346	80.8	101	74.8	245	83.6
Other/Unknown	17	4.0	6	4.4	11	3.8
Adequate health literacy
Extremely/quite a bit	314	73.4	101	74.8	213	72.7
Somewhat/a little bit/not at all	114	26.6	34	25.2	80	27.3
PROMIS T-scores (M, SD)^a
Mental health	50.2	9.2	48.0	9.9	51.3	8.8
Physical health	47.8	9.7	47.1	9.8	48.1	9.6
Pain intensity	2.9	2.8	3.1	3.0	2.8	2.7

^a PROMIS = Patient-Reported Outcome Measurement Information System; M = mean; SD = standard deviation

Prostate (21.5%), breast (20.1%), and colorectal (7.2%) cancers were the most common self-reported cancer sites (Table 1). Most respondents received their diagnosis more than 5 years prior to the administration of the survey (68.9%), had finished their cancer treatment (79.2%), and were not receiving maintenance therapy (80.8%). Most participants reported adequate health care literacy (73.4%). PROMIS T-scores averaged 50.2 (SD: 9.2) for mental health, 47.8 (SD: 9.7) for physical health, and 2.9 (SD: 2.8) for pain intensity.

For respondents aged 18–64 years (31.5%), there were more females (61.5%) compared to males (38.5%) than the overall study sample (Table 1). Although prostate cancer was the second ranked cancer site in this age group at 8.9%, it was lower than the general study sample (21.5%). A greater proportion of respondents aged 65 + years (8.5%) had planned cancer treatment compared to those aged 18–64 years (1.5%). Those aged 65 + reported lower rates of maintenance therapy (12.6%) compared to younger survivors (20.7%). The distribution of the other variables for younger survivors was consistent with the larger study sample.

Prevalence of cancer and non-cancer delayed and forgone care

Very few survivors reported delayed and forgone care related to cancer (Table 2: ns = 1-5). Respondents replying “Yes, but not for my cancer” ranged widely (ns = 4-42) per item. These responses were collapsed into a yes (combined) composite due to the few cancer-related responses. In the full sample, 16.2% of respondents reported any type of delayed medical care with only 5 respondents (1.2%) reporting delayed medical care specifically due to their cancer. Forgone care in the yes (combined) composite ranged from 6.0% for medical care, 6.4% for forgone mental health care, 6.8% for forgone medical tests, to 12.9% for forgone medications. Overall, 25.0% of the sample reported some form of delayed and/or forgone care due to cost.

Table 2. Sample distribution and prevalence of delayed and/or forgone care stratified by age group (N = 428).

	Full sample		Aged 18–64 years		Aged 65 + years
	n	%	n	%	n	%	P
Delayed medical care
Yes, for my cancer	5		2		3
Yes, but not for my cancer	64		42		22
Yes (combined)	69	16.2	44	32.8	25	8.6	<0.0001
No	357	83.6	90	67.2	267	91.4
Didn’t need it	1	0.2	1		0
Forgone medical care
Yes, for my cancer	1		1		0
Yes, but not for my cancer	24		16		8
Yes (combined)	25	6.0	17	13.1	8	2.8	<0.0001
No	393	92.3	113	86.9	280	97.2
Didn’t need it	8	1.9	4		4
Forgone medications
Yes, for my cancer	5		1		4
Yes, but not for my cancer	50		26		24
Yes (combined)	55	12.9	27	20.0	28	9.6	0.003
No	371	86.9	108	80.0	263	90.4
Didn’t need it	1	0.2	0		1
Forgone mental health care
Yes, for my cancer	3		2		1
Yes, but not for my cancer	22		18		4
Yes (combined)	25	6.4	20	15.8	5	1.9	<0.0001
No	364	85.5	107	84.3	257	98.1
Didn’t need it	37	8.7	8		29
Forgone medical tests
Yes, for my cancer	5		4		1
Yes, but not for my cancer	23		16		7
Yes (combined)	28	6.8	20	15.2	8	2.9	<0.0001
No	385	90.4	112	84.9	273	97.2
Didn’t need it	13	3.1	3		10
Delayed and/or forgone care^a
Yes	107	25.0	63	46.7	44	15.0	<0.0001
No	321	75.0	72	53.3	249	85.0

^a A combined outcome variable was calculated based on any positive response to the delayed and forgone variables compared to respondents who did not delay or forgo care for any reason

Participants aged 18–64 years experienced significantly higher rates of delayed and/or forgone care due to cost for all items compared to their counterparts aged 65 + years (Table 2). The prevalence of delayed medical care was 32.8% among those aged 18–64 years compared to 8.6% for respondents aged 65 + years (P < 0.0001). The prevalence of forgone medical care for younger survivors (13.1%) was more than 4.5 times that of older survivors (2.8%; P < 0.0001). Respondents aged 18–64 years (20.0%) were twice as likely to report forgone medications compared those aged 65 + years (9.6%; P = 0.003). Younger participants aged 18–64 years (15.8%) reported forgone mental health care at over 8 times the proportion of older participants aged 65 + years (1.9%; P < 0.0001). The proportion of forgone medical tests was 15.2% for respondents aged 18–64 years compared to 2.9% for respondents aged 65 + years (P < 0.0001). The proportion of delayed/forgone care due to cost for participants aged 18–64 years was 46.7% compared to 15.0% for those aged 65 + years (P < 0.0001).

Comparison of delayed and forgone care by age group

Female respondents accounted for significantly higher rates of delayed/forgone care due to cost compared to males for those aged 18–64 years (54.2% v. 34.6%; P = 0.026) and those aged 65 + years (20.8% v. 10.4%; P = 0.014; Table 3). Only those aged 18–64 years reported significant differences of delayed/forgone care by RUCA code with residents in areas with codes 1–2 (75.0%) reporting the highest rates followed by those with codes 4–6 (52.1%) and codes 7–10 (35.7%) reporting lower rates (P = 0.047). Among respondents aged 65 + years, survivors with household incomes of $20,000 to $34,999 experienced the highest rates of delayed/forgone care at 27.7% (P = 0.018). Average mental health PROMIS T-scores were significantly higher for both age groups who did not delay/forgo care due to cost (Aged 18–64 years: 44.9, SD: 10.2 v. 50.7, SD: 8.8, P = 0.001; aged 65 + years: 48.6, SD: 9.2 v. 51.7, SD: 8.6, P = 0.028). Average physical health PROMIS T-scores followed a similar pattern (Aged 18–64 years: 44.0, SD: 9.8, v. 49.7, SD: 9.2, P = 0.001; aged 65 + years: 44.4, SD = 44.0 v. 48.7, SD: 9.6, P = 0.006). Respondents aged 65 + years who had delayed/forgone care due to cost reported higher pain intensity scores (3.7, SD: 2.8) compared to those who did not (2.6, SD: 0.2; P = 0.010). Education level was not a significant factor for delayed/forgone care in either age group.

Table 3. Comparison of survey responses by delayed/forgone care due to cost (yes/no) and stratified by age group (N = 428).

	Aged 18–64 years						Aged 65 + years
	Delayed/forgone care^a						Delayed/forgone care^a
	Total	Yes	%	No	%	P	Total	Yes	%	No	%	P
Sex						0.026						0.014
Female	83	45	54.2	38	45.8		130	27	20.8	103	79.2
Male	52	18	34.6	34	65.4		163	17	10.4	146	89.6
Race/ethnicity						0.733						0.283
Non-Hispanic white	126	58	46.0	68	54.0		286	42	14.7	244	85.3
Other	9	5	55.6	4	44.4		7	2	28.6	5	71.4
Rural Urban Commuting Area Code						0.047						0.107
1–2	8	6	75.0	2	25.0		12	0	0.0	12	100.0
4–6	71	37	52.1	34	47.9		152	19	12.5	133	87.5
7–10	56	20	35.7	36	64.3		129	25	19.4	104	80.6
Education level						0.655						0.277
<High School	13	7	53.8	6	46.2		29	6	20.7	23	79.3
High School	30	13	43.3	17	56.7		80	16	20.0	64	80.0
Some college or vocational	42	17	40.5	25	59.5		86	11	12.8	75	87.2
College graduate	50	26	52.0	24	48.0		98	11	11.2	87	88.8
Household Income level						0.570						0.018
<$20,000	27	14	51.9	13	48.1		49	10	20.4	39	79.6
$20,000-$34,999	17	9	52.9	8	47.1		47	13	27.7	34	72.3
$35,000-$49,999	17	8	47.1	9	52.9		31	6	19.4	25	80.6
$50,000-$74,999	24	12	50.0	12	50.0		32	5	15.6	27	84.4
$75,000+	36	12	33.3	24	66.7		66	5	7.6	61	92.4
Missing/unknown	14	8	57.1	6	42.9		68	5	7.4	63	92.6
Marital status						0.198						0.926
Married or living as married	89	38	42.7	51	57.3		198	30	15.2	168	84.8
Not married	46	25	54.3	21	45.7		95	14	14.7	81	85.3
Cancer site						0.173						0.691
Breast	32	17	53.1	15	46.9		54	10	18.5	44	81.5
Prostate	12	2	16.7	10	83.3		80	9	11.3	71	88.8
Colorectal	12	7	58.3	5	41.7		19	2	10.5	17	89.5
Multiple sites	14	5	35.7	9	64.3		53	10	18.9	43	81.1
Other/Unknown	65	32	49.2	33	50.8		87	13	14.9	74	85.1
Time since diagnosis						0.634						0.595
<5 years	50	22	44.0	28	56.0		83	11	13.3	72	86.7
5 + years	85	41	48.2	44	51.8		210	33	15.7	177	84.3
Cancer treatment status						0.734						0.949
Planned	2	1	50.0	1	50.0		25	3	12.0	22	88.0
Current	11	6	54.5	5	45.5		34	4	11.8	30	88.2
Finished	120	56	46.7	64	53.3		219	35	16.0	184	84.0
Other/Unknown	2	0	0.0	2	100.0		15	2	13.3	13	86.7
Maintenance therapy						0.725						0.490
Yes	28	15	53.6	13	46.4		37	6	16.2	31	83.8
No	101	45	44.6	56	55.4		245	38	15.5	207	84.5
Other/Unknown	6	3	50.0	3	50.0		11	0	0.0	11	100.0
Adequate health literacy						0.100						0.273
Extremely/quite a bit	101	43	42.6	58	57.4		213	29	13.6	184	86.4
Somewhat/a little bit/not at all	34	20	58.8	14	41.2		80	15	18.8	65	81.3
PROMIS T-scores (M, SD)^b
Mental health		44.9	10.2	50.7	8.8	0.001		48.6	9.2	51.7	8.6	0.028
Physical health		44.0	9.8	49.7	9.2	0.001		44.4	8.7	48.7	9.6	0.006
Pain intensity		3.6	3.0	2.7	3.0	0.076		3.7	2.8	2.6	0.2	0.010

^a A combined outcome variable was calculated based on any positive response to the delayed and forgone variables compared to respondents who did not delay or forgo care for any reason.

^b PROMIS = Patient-Reported Outcome Measurement Information System; M = mean; SD = standard deviation

Regression analyses of delayed and forgone care due to cost

Multivariable logistic regressions stratified by age group included any univariable association with a P-value of 0.2 or less (Table 3) and excluded household income due to missing values and individuals RUCA codes of 1–2 due to the small sample size (Table 4). Respondents aged 18–64 years with RUCA codes of 7–10 (OR: 0.42; CI: 0.20-0.89) had significantly lower odds of delayed/forgone care compared to those residing in areas with RUCA codes of 4-6. Younger survivors with higher mental health PROMIS T-scores (OR: 0.93; CI: 0.90-0.97) also experienced significantly lower odds of delayed/forgone care compared to younger survivors with lower mental health scores.

Table 4. Odds of delayed/forgone care due to cost (the combined outcome variable) stratified by age group (N = 428).

	Aged 18–64 years		Aged 65 + years
	Odds Ratio	95% Confidence Interval	Odds Ratio	95% Confidence Interval
Sex
Female	–	–	2.00	1.02-3.93
Male	Ref.		Ref.
Rural Urban Commuting Area Code
4–6	Ref.		Ref.
7–10	0.42	0.20-0.89	2.02	1.03-3.95
PROMIS T-scores^a
Mental health	0.93	0.90-0.97	–	-
Physical health	–	–	0.95	0.92-0.99

^a PROMIS = Patient-Reported Outcome Measurement Information System

Among participants aged 65 + years, females (OR: 2.00; CI: 1.02-3.93) had double the odds of reporting delayed/forgone care compared to males (Table 4). Respondents with RUCA codes 7–10 (OR: 2.02; CI: 1.03-3.95) experienced increased odds of reporting delayed/forgone care due to cost compared to those with RUCA codes 4-6. Respondents with higher physical health PROMIS T-scores (OR: 0.95; CI: 0.92-0.99) had reduced odds of reporting delayed/forgone care due to cost compared to their counterparts.

Discussion

This study assessed the prevalence of delayed/forgone medical care of any type due to cost in a sample of mostly rural, Appalachian, long-term cancer survivors and factors associated with these outcomes. A quarter of these survivors reported delayed/forgone care of some type but not primarily due to cancer. Nationally representative studies estimate 7%−10% of survivors forgo care due to cost [3–6]. Our sample reported higher rates of delayed/forgone compared to national estimates that report between 8.5%−21.2% delayed/forgone care for those aged 18–64 compared with 25.0% in our sample and 12.4%−24.7% for those aged 65 + compared with 46.7% in our sample [3,4,6]. This study represents an important first step to delineate cancer and non-cancer related delayed/forgone care due to cost among long-term survivors that has not been described extensively in previous literature [19,33].

Our study corroborated studies that have found higher rates of delayed/forgone care in younger cancer survivors [5,6]. Still, the results of this study indicated that some older survivors, who may be managing multiple age-related comorbidities, remained vulnerable to delayed/forgone care due to cost, especially medications, despite Medicare coverage [34].

Our findings are consistent with previous literature describing higher rates of delayed/forgone care among females, despite having higher rates of health care engagement [35]. Qualitative findings have also noted the impact of financial hardship/toxicity due to cancer treatment among rural, female survivors [36,37]. Future studies should examine these differences in greater detail as we cannot determine if our participants experienced more acute financial difficulties compared to males in the sample or if they had more opportunity to delay/forgo care due to their increased rates of engagement with health care providers. This study provides evidence that cancer care is not the primary factor driving forgone care, at least in this sample.

The National Cancer Institute (NCI) has called for greater investment in catchment specific rural cancer research given the multilevel barriers to optimal care faced by residents of these areas [38–41]. Aligned with this goal, researchers have examined interventions to assess forgone care and mitigate barriers to cancer care due to cost [42–46]. Financial navigation is a promising strategy to address financial concerns in the oncology setting, with the potential to connect rural patients with financial assistance resources. Trained navigators knowledgeable about resources including prescription assistance, free or reimbursed travel expenses, lodging, financial arrangements for treatment services, and other resources that integrate clinical resources as well as partnering community organizations are potential sources to mitigate financial barriers resulting in delayed and forgone care. Rural oncology practices face challenges in implementing financial navigation including consistent assessment of financial needs, and inconsistent availability of available resources and eligibility [46–48]. Travel distance to providers and limited resources act as barriers to navigation interventions addressing the underlying causes of delayed and forgone care [47]. Continued relationships between navigators and survivors are also critical to support long-term survivors and require follow up outside of the clinical environment [49].

Limitations

Potential participants for this study were selected from 7 priority, primarily rural, Appalachian counties in North Carolina due to their higher rates of smoking and cancer incidence, but these counties are not necessarily representative of rural counties in general or other Appalachian areas in the US. Our study had a relatively large sample to analyze despite occurring at the beginning of the COVID-19 pandemic. Comparisons of respondents to non-respondents found no differences among the two groups [29].

Conclusion

One quarter of rural longer-term cancer survivors reported delaying and/or forgoing care in the prior year primarily unrelated to their cancer diagnosis, with rates approaching 50% in survivors under 65 years of age. Survivors rarely associated delayed/forgone care explicitly to cancer. This study presents findings on rural, Appalachian survivors with clinical implications indicating the need to: 1) ask survivors about delayed/forgone care initially and many years following cancer diagnosis and treatment, and 2) provide supportive services that mitigate delayed/forgone care due to cost to the extent possible for all types of medical and mental health care, particularly for younger and female survivors throughout their cancer survivorship journey.

Financial support

This study was supported by grant, 3P30CA012197-43S2, from the National Cancer Institute. Dr. Falk was supported by grant, T32CA122061, Training Grant in Cancer Prevention and Control from the National Cancer Institute (NCI). Dr. Morris was supported by an NCI K00 fellowship, K00CA245799. The authors wish to acknowledge the support of the Wake Forest Baptist Comprehensive Cancer Center Biostatistics Shared Resource, supported by the NCI’s Cancer Center Support Grant award number P30CA012197. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NCI.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Due to the nature of this research, participants of this study did not agree for their data to be shared publicly. Deidentified data are available upon request from the investigators.

Additional information

Funding

This work was supported by National Cancer Institute: [Grant Number 3P30CA012197-43S2].