The Interaction of Rurality and Rare Cancers for Travel Time to Cancer Care

Abstract

Purpose

Geographic access to cancer care is known to significantly impact utilization and outcomes. Longer travel times have negative impacts for patients requiring highly specialized care, such as for rare cancers, and for those in rural areas. Scant population-based research informs geographic access to care for rare cancers and whether rurality impacts that access.

Methods

Using Medicare data (2014–2015), we identified prevalent cancers and cancer-directed surgeries, chemotherapy, and radiation. We classified cancers as rare (incidence <6/100,000/year) or common (incidence ≥6/100,000/year) using previously published thresholds and categorized rurality from ZIP code of beneficiary residence. We estimated travel time between beneficiaries and providers for each service based on ZIP code. Descriptive statistics summarized travel time by rare versus common cancers, service type, and rurality.

Findings

We included 1,169,761 Medicare beneficiaries (21.9% in non-metropolitan areas), 87,399; 7.5% had rare cancers, with 9,133,003 cancer-directed services. Travel times for cancer services ranged from approximately 29 minutes (25^th percentile) to 68 minutes (75^th percentile). Travel times were similar for rare and common cancers overall (median:45 v. 43 min) but differed by service type; 13.4% of surgeries were >2 hours away for rare cancers, compared to 8.3% for common cancers. Increasing rurality disproportionately increased travel time to surgical care for rare compared to common cancers.

Conclusions

Travel times to cancer services are longest for surgery, especially among rural residents, yet not markedly longer overall between rare versus common cancers. Understanding geographic access to cancer care for patients with rare cancers is important to delivering specialized care.

Precis:

Travel times for rare cancers were similar to common overall but differed by service type and rurality. Specifically, increasing rurality disproportionately increased travel time to surgical care for rare compared to common cancers

Introduction

Over a decade ago, the National Cancer Institute (NCI) and the National Institutes of Health Office of Rare Diseases, (so named at that time) spearheaded efforts to advance research on rare cancers, yet much remains unknown about geographic access to care for individuals with rare cancers. Rare cancers, defined by the American Cancer Society (ACS) as fewer than 6 cases per 100,000 annually, account for approximately 25% of all cancers among adults in the United States (U.S.).¹ Outcomes tend to be worse for patients with rare cancers compared to more common types¹, in part because rare cancers typically are more difficult to diagnose, require access to highly specialized expertise, have less standardized treatment protocols, and often entail longer travel times to treatment compared to treatment of more common cancers.²

Travel time is an important component of geographic access to cancer care that has been shown to be associated with reduced survival, lower likelihood of clinical trial enrollment, financial hardship, and decreased quality of life.^3–5 In studies examining travel time for treatment (and/or diagnosis of) common cancers (i.e., breast, lung, colorectal, and prostate), rural patients were shown to have significantly longer travel times to cancer care than their urban counterparts.^6,7 However, there exist no population-based studies to compare travel times to cancer care for common versus rare cancers in the U.S. We do have evidence from a smaller study examining only a single rare cancer type that suggest travel time is greater for rare cancers.²

Further, the comparatively longer travel times for those with rare versus common cancers may be exacerbated for those in rural regions. Geographic proximity often determines timeliness, quality, and appropriateness of care, which are critical dimensions of access.⁸ Thus, a population-based understanding of how travel time varies for rare versus common cancers – and whether rurality interacts with rare and common cancers – is fundamental to assessing access to care and informing health care delivery models for rural patients.

This study provides a population-based, national characterization of travel times for rare versus common cancers in the Medicare population between 2014 and 2015. We quantified travel times for cancer care overall, surgical care, and chemotherapy and radiation combined. We also examined the interaction between rare versus common cancer and rurality. The main objective of this work was to better characterize travel time to treatment for those with rare cancers.

Methods

Data and Study Population

In the United States, the Medicare program provides health insurance to residents 65 or older, as well as to certain younger residents who are disabled, have End-Stage Renal Disease (ESRD), or have Amyoropic Lateral Sclerosis. Medicare is divided into “parts,” with Part A covering inpatient care and Part B covering outpatient care. We used Medicare enrollment and claims files from January 1, 2014 through September 30, 2015, specifically the beneficiary Denominator (enrollment), the Medicare Provider Analysis and Review (MedPAR), Outpatient, and 100% Carrier files. Inclusion criteria were cancer diagnosis and treatment during the study period (see next section for details), enrollment in Medicare Parts A and B for the entire period, age-based eligibility with no End Stage Renal Disease, ages 65–99, and no HMO enrollment. We excluded individuals enrolled in HMOs because Centers for Medicare and Medicaid Services (CMS) does not provide complete claims data for these individuals. We excluded records with missing or invalid ZIP codes and those from outside of the contiguous U.S. Attributes of beneficiaries were derived from the Denominator file, specifically: age (years), sex, and race (non-Hispanic white, other). A binary race categorization was necessitated by small numbers of observations and CMS data suppression rules. We were also limited to the method of ascertainment of race and ethnicity in CMS data; specifically, self-reported, hospital-ascribed, surname-based algorithms, or unknown. We included race as a sociopolitical construct in this analysis and not as a biological classification, which does not distinguish ethnicities within races.

Definitions of Key Variables

Cancer patient denominator, rare cancers, and cancer-directed services

Cancer patients were identified using diagnosis codes (International Classification of Diseases, Ninth Revision, Clinical Modification: ICD-9-CM) listed for 26 cancer types.^9,10 Although several definitions for rare cancers exist, we used the one endorsed by ACS and RARECARENet, the European consortium of 24 countries and 94 population-based registries.¹¹ Specifically, rare cancers were defined as those with fewer than 6 cases per 100,000 individuals per year. A complete list of cancers included a rare versus common categorization is shown in Supplemental Table 1. For those beneficiaries who had more than one cancer (24,297; 2.1%), we included them in the rare cancer group if one or both were categorized as rare. Cancer services were ascertained by ICD-9-CM procedure and Current Procedure Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes found in the MedPAR, Outpatient and Part B claims files. Cancer services were defined as cancer-directed surgical procedures (“surgical”) or chemotherapy and radiation treatment codes (“chemotherapy/radiation”). A validated set of claims codes for ascertaining chemotherapy and radiation were publicly available through the CMS contracted Research Data Assistance Center (ResDAC).¹² We used a validated set of codes for cancer-directed surgery developed for related Cancer Service Area (CSA) work as published previously (Supplemental Table 1).¹³ Multiple services on the same day in the same ZIP code were captured as a single claim-day for the service type. When the claim-day included both surgical and chemotherapy/radiation services, it was categorized as surgical.

Travel time calculation

Travel time for each cancer service was calculated as the shortest travel path in minutes for road-based travel between the population centroid of the Medicare beneficiary’s ZIP code of residence (origin) and the population centroid of the billing ZIP code (destination) for that service (facility or physician) through linkages with the Medicare Provider of Services file. Travel time was calculated using ArcGIS Network Analyst and then refined by Google Maps API.¹⁴ All travel time estimates were based on one-way travel and irrespective of any specific travel restrictions including, but not limited to, left turn restrictions and/or one-way streets (which have the potential to add to overall travel time). Attributes of the beneficiary associated with that service (e.g., age, sex, race, rurality) were ascribed to each service, along with its calculated travel time.¹⁵

Rurality

Using patient ZIP code for each cancer-related service, we classified rurality based on the Rural Urban Commuting Area (RUCA) 4-tier classification schema into: metropolitan, micropolitan, small town, and isolated.¹⁶ For beneficiaries who changed residences during the study period, we used the residential ZIP code for the calendar year of the cancer-related service.

Statistical Analysis

The primary unit of analysis was the cancer service event rather than the individual beneficiary, although the cancer service event was attributed to a rare or common cancer based on the beneficiary’s cancer type. We summarized beneficiary characteristics associated with those services and cancers. We calculated the distribution of travel time for the service events overall and by service type (i.e., surgical or chemotherapy/radiation) for all cancers and stratified by common and rare cancers. In the same way, we also calculated the number and percentage of service events in travel time categories. Travel time of 0 minutes indicated that the beneficiary lived in the same ZIP code where the service was received. We additionally stratified the travel time distributions and frequencies per categories by the 4-tier RUCA classification. Travel time difference between common and rare cancers was measured across the distributions for the percentile categories of: 5^th, 10^th, 25^th, 50^th, 75^th, 90^th, and 95^th. Chi-squared tests were used to evaluate associations between categorical variables. Two-sample t-tests (unpaired) were used to compare means for continuous variables.

We performed a sensitivity analysis to determine whether seasonality might affect the results by analyzing observations from May-October compared to November-April (colloquially referred to as the “snowbird effect”). We report here only results from the full year, as the seasonally-stratified results did not significantly affect the findings.

All analyses were covered by a Data Use Agreement (DUA) with the CMS¹⁷ through the ResDAC at the University of Minnesota and the Committee for the Protection of Human Subjects at Dartmouth-Hitchcock Medical Center.

Results

The study population was comprised of 1,169,761 Medicare beneficiaries with cancer, of whom 87,399 (7.5%) had a rare cancer. Rare cancers were more prevalent in the younger age groups (ages 65–69, 7.7%; 85+, 6.9%; p<0.001) and in males (7.7% vs 7.2%; p<0.001). Non-white individuals made up 15.1% of the population, and non-metropolitan/rural (i.e., micropolitan, small town, and isolated) made up 22.0% (Table 1). From the overall study population of prevalent cancers, there were 9,133,003 cancer-directed services, including 744,061 surgical and 8,388,942 chemotherapy and radiation events; 7% of the total cancer services were for rare cancers (Table 2).

Table 1:

Demographic characteristics of Medicare beneficiaries associated with cancer related services for common and rare cancers.

	OVERALL		COMMON CANCERS		RARE CANCERS		p-value
	N	%	N	%	N	%
Total	1,169,761	100	1,082,362	92.5	87,399	7.5
Age group (years)	326,317	27.9	301,207	27.8	25,110	28.7	<0.001
65–69
70–74	294,684	25.2	272,277	25.2	22,407	25.6
75–79	239,892	20.5	221,805	20.5	18,087	20.7
80–84	169,935	14.5	157,680	14.6	12,255	14.0
85–89	99,979	8.5	93,130	8.6	6,849	7.8
90+	38,954	3.3	36,263	3.4	2,691	3.1
Sex	649,980	55.6	599,758	55.4	50,222	57.5
Male							<0.001
Female	519,781	44.4	482,604	44.6	37,177	42.5
Race	176,966	15.1	164,640	15.2	12,326	14.1
Other							<0.001
Non-Hispanic White	992,795	84.9	917,722	84.8	75,073	85.9
Rurality	912,894	78.0	844,859	78.1	68,035	77.8
Metropolitan							0.006
Micropolitan	132,228	11.3	122,466	11.3	9,762	11.2
Small town	71,568	6.1	66,014	6.1	5,554	6.4
Isolated	53,071	4.5	49,023	4.5	4,048	4.6

Table 2.

Cancer-directed services in the Medicare beneficiary population from 2014–2015 and summary of travel time for services.

			% of services with road-based travel time from residential to service ZIP codea
		N	>0.5 hr	>1 hr	>2 hr	>3 hr	>4 hr
Common Cancers	Surgery	716,380	69.2	28.0	8.3	4.8	3.7
	Chemotherapy/Radiation	7,761,183	72.9	31.2	10.5	6.1	4.7
Rare Cancers	Surgery	27,681	71.7	35.6	13.3	7.5	5.2
	Chemotherapy/Radiation	627,759	74.5	34.7	12.4	7.2	5.5

Notes

a – Each column provides the percentage of travel times that exceed a given threshold; the columns are not mutually exclusive and the percentages therefore do not sum to 100%.

b – The distribution of travel times (for both surgery and chemotherapy/radiation) is significantly different for common vs rare cancer services (p<0.001).

One-way travel times for all cancer services ranged from 29 min. at the 25^th percentile to 68 min. for the 75^th percentile (Figure 1). Median travel times for all services combined were not markedly different for common and rare cancers (43 min. v. 45 min. respectively).¹. However, there was a notable divergence among higher percentiles of travel time; for example, at the 95^th percentile, travel times were almost an hour longer each way for rare cancers compared to common cancers (276 min. v. 215 min, Figure 1). The percentage of services for common and rare cancers that exceeded a 2-hour drive time ranged from 8.3% (common) to 12.4% (rare), with rare cancers having a higher percentage across all service categories (Table 2, p<0.001). The greatest difference for rare versus common cancers was seen for surgical care. Specifically, while 13.3% of rare cancer patients had to travel over two hours (one-way) to receive care, this was only true of 8.3% of common cancer patients (Table 2, p<0.001).

Figure 1.

Distribution of 1-way travel time in minutes for rare and common cancers by type of cancer services accessed

We examined the effect of rural-urban residence on the travel times for all cancers, with a specific focus on the differential impacts between common and rare cancers. Across all categories of cancer type (i.e., common and rare combined, common only, rare only), travel times increased as residential category went from metropolitan to isolated rural (Figure 2). For all cancers combined and all cancer-directed services, the median travel time for metropolitan residents was almost 2.5-fold shorter than those in isolated rural areas (38 min. v. 90 min.). Common versus rare cancer type partially modified this urban-rural effect, particularly for surgery. Specifically, among rural patients, travel time to surgeries for rare cancer were notably longer than to common cancer surgeries (median surgical travel time: rare 100 min., common 89 min.; 90th percentile: rare, 243 min., common 189 min.) (Tables 3 a,b).

Figure 2.

Distribution of travel time differences between rare and common cancers by rurality and cancer service type

Table 3a and b.

Cancer-directed services in the Medicare beneficiary population from 2014–2015 and summary of travel time for services by rural urban commuting area (RUCA) category for: a) common cancers and b) rare cancers.

			Estimated one-way road-based travel time from residential to service ZIP code (minutes)
			Percentile of Distribution
a. Common Cancers		N	P5	P10	P25	P50 (median)	P75	P90	P95
Metropolitan	All cancer-directed services	6,800,326	10	19	27	38	56	93	183
	Surgery	565,353	7	18	26	36	52	79	132
	Chemotherapy/Radiation	6,234,973	10	20	27	38	56	94	188
Micropolitan	All cancer-directed services	867,067	13	15	43	66	99	162	256
	Surgery	78,372	12	15	39	63	92	143	211
	Chemotherapy/Radiation	788,695	13	16	44	67	100	164	265
Small Town	All cancer-directed services	468,098	20	45	61	81	113	184	285
	Surgery	41,333	18	42	60	78	109	168	245
	Chemotherapy/Radiation	426,765	21	45	62	81	114	185	293
Isolated	All cancer-directed services	341,845	45	54	67	90	131	201	357
	Surgery	31,313	43	52	66	89	124	189	303
	Chemotherapy/Radiation	310,532	46	54	67	90	132	202	365
b. Rare Cancers		N	P5	P10	P25	P50	P75	P90	P95
Metropolitan	All cancer-directed services	523,061	10	20	28	40	61	110	231
	Surgery	21,342	9	19	27	38	60	112	217
	Chemotherapy/Radiation	501,719	10	20	28	40	61	109	233
Micropolitan	All cancer-directed services	67,304	13	16	46	70	108	184	321
	Surgery	3,125	12	16	43	74	116	185	285
	Chemotherapy/Radiation	64,179	13	16	47	70	108	183	322
Small Town	All cancer-directed services	37,148	34	49	64	84	124	203	363
	Surgery	1,939	17	23	61	86	128	197	282
	Chemotherapy/Radiation	35,209	35	50	65	84	124	204	375
Isolated	All cancer-directed services	27,899	46	54	69	93	140	217	422
	Surgery	1,275	43	54	70	100	153	243	367
	Chemotherapy/Radiation	26,624	46	54	68	93	140	216	426

The percentage of cancer care occurring less than 1 hour from the beneficiary ZIP code of residence was 78% for metropolitan, 42% for micropolitan, 23% for small towns, and 16% for isolated rural (Table 4, p<0.001). Differences for common and rare cancers were negligible except for surgical care among those in metropolitan and micropolitan areas, for which the proportion of surgical procedures under 1 hour travel time was higher for common compared to rare (metropolitan: common 82%, rare 75%, p<0.001; micropolitan: common 47%, rare 38%, p<0.001) (Table 4).

Table 4.

Percentage of cancer services by service category and common v. rare, in one-way travel time categories.

		% travel time <1 hour					% travel time 1–2 hours					% travel time 2–4 hours					% travel time >4 hours
	Cancer Type	Isolated	Small Town	Micropolitan	Metropolitan		Isolated	Small Town	Micropolitan	Metropolitan		Isolated	Small Town	Micropolitan	Metro-politan		Isolated	Small Town	Micropolitan	Metropolitan
Surgery	All combined Common Rare	17	25	46	82		56	55	39	13		21	15	11	2		7	5	4	3
		17	25	47	82		56	55	39	13		20	15	11	2		6	5	4	3
		15	24	38	75		48	47	39	16		27	21	17	5		10	7	6	5
Chemotherapy/Radiation	All combined Common Rare	16	23	42	78		54	54	41	14		22	16	12	4		8	7	6	4
		16	23	42	78		54	54	41	14		22	16	12	4		8	7	5	4
		16	21	38	75		52	53	41	16		24	19	14	4		9	8	7	5

The main differences in travel time for rare v. common cancers by rurality were seen in the upper ends of the travel time distributions, that is, above the median (50^th percentile) and beyond 4 hours. Although these more extreme travel times pertain to the right-hand tail of the distribution, in absolute terms, that still equates to 13,467 rural beneficiaries with common cancers in this time period, and 1,342 rural beneficiaries with rare cancers (for >4 hours travel time) (Supplemental Table 2).

We also quantified the difference in travel time between rare and common cancers across higher travel time percentiles. In general, travel times were typically greater for rare cancers across travel time percentiles, including when stratified by rurality and type of service. Conversely, the travel time for patients with rare cancers to receive chemotherapy and radiation treatment were consistently longer than those with common cancers, which was particularly exaggerated in micropolitan communities. (Fig. 2).

Discussion

This study provides the first national comparison of travel times for common versus rare cancers overall and by type of cancer services and rurality. We found that median travel time for cancer-directed services did not differ notably for common and rare cancers, but the distributions of travel time did. For travel times greater than the median, treatment of rare cancers had markedly longer travel times than common cancers, particularly for surgical care. Travel time for all cancer care increased with increasing rurality. The proportion of surgical care under 1-hour one-way travel time was greater for common cancers compared to rare cancers. Differences in travel time for urban compared to rural patients were modified by cancer type based on service type. Surgical care had longer travel times for rare than common cancers in rural areas. For chemotherapy and radiation, rural patients consistently had longer travel times for rare compared to common cancers, with differences greater than for urban patients. Longer travel times for chemotherapy and radiation translate into substantially greater travel times over the course of care compared to surgery, as these treatment cycles require multiple visits over several weeks to a number of months. Together, these results demonstrate a greater travel time for rare cancers v. common for surgical care. For the vast majority of patients, increasing rurality exacerbates the travel burden for those with rare cancers more than common cancers. The empirical evidence of travel time for common and rare cancers by cancer service type and rurality is important in understanding the impacts on patients of geographic access to cancer care, as well as in informing tailored care delivery models.

Rare cancers are a heterogeneous group of neoplasms that share important non-biological characteristics, such as limited evidence, difficulty in diagnosis, lack of expertise, issues in quality of care, discrepancies in outcomes and high rates of migration to receive care.¹⁸ Very little is known about geographic access to care for rare cancers, and much of what we do know comes from the European RARECARE project, based on data provided by 90 cancer registries in 22 European countries.¹⁹ European Reference Networks (ERN) of the RARECARE project have a general consensus on the hub and spoke model for care, with reference (or referral) centers as the hubs and local collaborating facilities as the spokes.^18,20 While such a centralized model allows for geographic coverage of specialized expertise by higher volume providers, the need for travel to hubs can impose a dramatic burden on patients’ quality of life and financial security.²⁰ Further, the centralized model may be more practicable in Europe than the U.S. given that transportation systems in Europe are more heavily based on public transportation in contrast to the heavy reliance on single occupancy vehicles in the U.S.

In the U.S., patients with rare cancers travel longer for care than those with common cancers, although the extent is not known at a population level. One study in the Pacific Northwest examined geographic access to care for 391 patients who were seen at the Seattle Cancer Care Alliance with the rare cancer Merkel Cell Carcinoma (MCC).²¹ Those patients were similar in age to this study (mean age 67 years; SD ± 11 years) and showed a median 813-mile one-way trip.²¹ Our study had a much shorter median one-way travel time to cancer care overall (45 min.), likely due to including the whole spectrum of rare cancers and analyzing across the full geographic extent of the continental U.S. There are no previous studies that examine rare cancers in the context of urban-rural differences for comparison, but our findings of a direct association between travel time to cancer care and rurality are consistent with prior studies.^6,7

We have limited insight from the patient perspective on perceived travel burden to care for common versus rare cancers, but a cross-sectional national survey in the Netherlands showed notable differences in health care experiences for these groups.²² For example, patients with common cancers were more likely to choose a hospital close to home than rare cancer patients, resulting in longer travel distances for patients with rare cancer. However, patients with rare cancer also showed greater willingness to travel to receive this specialized care.²¹ Previous studies in head and neck cancer patients also found a greater willingness to travel longer distances to have access to higher quality cancer care for their rare cancer.^22,23 These findings fit the travel time patterns we demonstrate with this study and seem to be particularly notable for surgical cancer care in our study population. Specifically, rare cancers had higher proportions of surgical services delivered across increasing travel times compared to chemotherapy/radiation in rare or common cancers. This result is likely due to the relative dearth of surgical oncologists relative to medical oncologists and radiation oncologists. According to the American Society of Clinical Oncology (ASCO) 2020 Workforce Information System report, there were about 25 times more medical oncologists and hematologist/oncologists, and almost 10 times more radiation oncologists than surgical oncologists practicing in the U.S.²⁴ Additionally, rare cancers that are solid tumors may be more likely than common solid tumors to require specialized surgical oncologists, rather than other surgeons.

Several key strengths of our study include the ability to: 1) take a full contiguous U.S. population-based approach for individuals aged 65 and older through the use of national Medicare claims; 2) examine travel times based on actual utilization of cancer services overall and separately for surgery and chemotherapy/radiation; and 3) evaluate the impact of increasing rurality on travel time differences for rare and common cancers. However, we acknowledge several limitations to our study. For example, the proportion of rare cancers in our study population is lower than typically reported.²⁵ One likely reason for this is due to age distributions of rare cancers, many of which are types found in younger individuals.^25,26 We also ascertained prevalent, not incident cancers. It is worth noting that there is no single accepted definition of rare cancers. The European RARECARE network and the ACS use <6/100,000/year, while the NCI uses <15 cases/100,000/year. Differing definitions may have a large impact on how we understand health care delivery for rare cancers, particularly related to centralized v. less centralized care delivery models. With the more liberal definition of rare cancers at <15 cases/100,000/year, it may be useful to take a gradated approach (such as rare, intermediate, and common) rather than use a threshold. Thus, some cancers (such as bladder) may be in a more intermediate range, based on incidence in the U.S., which could have implications for volumes of similar patients seen at health systems and hospitals and the concomitant need to travel for expert care. Additionally, our travel time calculations quantify one-way travel times which demonstrates the lowest bound of the “travel burden” for those seeking cancer services for rare and/or common cancers. Lastly, our study assumes that individuals within the study area are traveling via single occupancy vehicles to reach services. However, it is possible that the patients utilized public transportation or need to rely on a caretaker to take them due to lack of available transportation.

In summary, we show that travel time to cancer services for rare cancers is notably longer than common cancers, particularly for surgical care and for rural individuals. This study provided foundational and conservative estimates of travel time for rare and common cancers by service type and rurality, which are important to report, given the lack of any national estimates to inform ongoing discussions of access to cancer care. The sparse evidence on health care access and utilization for rare cancers may become even more important as we expand beyond traditional definitions of “rare cancer” as some common cancers are reclassified into multiple rare subtypes based on molecular tumor profiles.^27,28