Rural–Urban Disparities in Mortality and Place of Death for Gastrointestinal Cancer in the United States From 2003 to 2019

Patients with gastrointestinal (GI) cancer experience high symptom burdens and complex needs that challenge healthcare infrastructure at multiple levels beginning with primary care screening and continuing to postsurvival care and end-of-life services. It is unclear whether mortality improvements from national risk factor reduction, screening uptake, and treatment advances apply equally to rural areas where gaps in healthcare funding and access to hospice care exist.¹ To determine temporal trends, we investigated rural–urban disparities in age-adjusted mortality rates (AAMRs) per 100,000 and place of death in the United States in individuals dying from GI cancers between 2003 and 2019.

We used the Centers for Disease Control and Prevention’s Wide-ranging Online Data for Epidemiologic Research database to analyze deaths from GI cancers from 2003 to 2019 across large metropolitan, small- or medium-sized metropolitan, and rural areas.^2,3 We estimated annual percentage changes (APCs) in AAMRs across all areas and percentage of all deaths occurring in medical, hospice, and nursing facilities and home. See Supplementary Methods for details.

From 2003 to 2019, there were 2,490,412 deaths from GI cancers, predominantly in large metropolitan (51.6%) followed by medium and small metropolitan (30.4%) and rural areas (17.8%) (Figure 1A). Across all areas, years, and ages, non-Hispanic blacks had higher AAMRs and women had lower AAMRs. AAMRs reduced across all areas from 45.9 in 2003 to 39.7 in 2019 (APC, −0.79; 95% confidence interval [CI], −0.83 to −0.74). However, rural areas had less annual reduction in AAMRs (APC, −0.24; 95% CI, −0.31 to −0.16) compared with medium and small (APC, −0.6; 95% CI, 0.64 to −0.56; P < .001) and large metropolitan areas (APC, −1.07; 95% CI, −1.13 to −1.02; P < .001). The absolute difference in AAMRs between rural areas and large metropolitan areas increased from 0.9 to 4.2, a 4-fold increase.

Figure 1.

(A) AAMRs per 100,000 population per year for GI cancers, (B) trends in place of death by year and (C) geographic area, and (D) association with place of death for GI cancers by geographic area.

Older individuals (aged > 65 years) were noted to have reductions in AAMRs from 264.4 in 2003 to 217.2 in 2019 (APC, −1.17; 95% CI, −1.27 to −1.06), whereas no improvements were found in younger individuals (aged 25–65 years) (AAMRs of 23.9 to 23.4 during the study period; APC, 0.09; 95% CI, −0.05 to 0.24). Although all racial and ethnic groups had a reduction in AAMRs, non-Hispanic whites had the least reduction (APC, −0.65; 95% CI, −0.72 to −0.58), particularly in rural areas (APC, −0.13; 95% CI, −0.20 to −0.05).

Nationwide trends by underlying GI cancer type are shown in Figure 1A. In Supplementary Table 1, we show a subgroup analysis for ages 25–65 and >65 years. Across all cancer types and most age groups, less improvement was found in rural areas and often an increase in mortality. For all GI cancers, worsening (esophageal and gastric), no change (colorectal cancer), or only slight improvement (pancreatic or liver/biliary, in which the latter showed less improvement compared with age > 65 years) was found in those aged 25–65 years.

Nationwide deaths were reduced in medical aid nursing facilities and increased in hospice facilities and at home (Figure 1B). Trends over time for each geographic area are noted in Figure 1C. Compared with rural areas, deaths in hospice facilities were more likely to occur than in a medical facility for large metropolitan areas (odds ratio [OR], 1.81; 95% CI, 1.81–1.83) and medium and small metropolitan areas (OR, 2.34; 95% CI, 2.33–2.35), whereas deaths in nursing facilities were less likely to occur than in a medical facility for large metropolitan areas (OR, 0.74; 95% CI, 0.74–0.74) and medium and small metropolitan areas [OR, 094; 95% CI, 0.94–0.95). Compared with rural areas, deaths at home were less likely to occur than deaths in a medical facility for large metropolitan areas (OR, 086; 95% CI, 0.85–0.86) but were similar for medium and small metropolitan areas (OR, 1.07; 95% CI, 1.07–1.08) (Figure 1D). Supplementary Table 2 shows similar trends across all underlying cancer types with no 1 type that drove these findings.

Supplementary Table 3 shows sociodemographic differences in place of death. Compared with non-Hispanic whites, deaths in the home hospice facility, and nursing facility were less likely than a medical facility for Asian American and Pacific Islander, Native American and Alaskan Native, Hispanic, and non-Hispanic black groups. These racial and ethnic groups also had a lower percentage of deaths in hospice or nursing facilities and a higher percentage of deaths in medical facilities. Similar trends held for older (>65 years) compared with younger (25–65 years) individuals and men compared with women.

Barriers at multiple levels may explain widening disparities in morality between rural and urban areas. Rural areas have higher rates of obesity, tobacco use, and secondhand smoke exposures implicated in the pathogenesis of GI cancers⁴ and lower screening rates for colorectal cancer,⁵ the predominant GI cancer amenable to early screening and treatment. Regionalization of health care may result in reduced proximity to high-volume surgical hospitals and to centers that provide guideline-concordant medical or radiotherapy or postsurvival cate.^1,6 Further, rural areas lack convenient access to specialized oncologic care, ensuring access to new treatments and clinical trials, and palliative care and end-of-life services.¹ Of particular concern was the minimal improvement or worsening in mortality among younger patients across most cancer types. This is consistent with recent national trends. Although diet, obesity, smoking, and alcohol consumption may be implicated, a complete understanding of causes remain unknown and are likely multifactorial.⁷

We notes disparities in end-of-life care in rural areas, which face unique constraints in delivering hospice care with lower Medicare reimbursement, staffing shortages, and high operational costs.¹ Although hospice services may be provided in one’s home or a nursing facility (both places where higher odds of death in a rural area were noted), end-of-life care at home requires significant caregiving and financial investment even with the Medicare hospice benefit, whereas care in a nursing home is often not provided by trained providers in a dedicated setting.⁸ These inequities are compounded by higher levels of poverty in rural areas, with a larger proportion of self-employed individuals (resulting in lower rates of employee-sponsored health insurance and paid medical leave) whose incomes do not qualify for Medicaid.

Finally, we also found nationwide differences across minority groups where non-Hispanic whites were least likely to die in a medical facility and most likely to die in a hospice facility. Although differences in place of death may underlie cultural and spiritual beliefs that dictate understanding of and uptake of palliative care, considerable evidence confirms that minority populations are less likely to be knowledgeable about advanced care planning, receive poorer communication in documentation of treatment preference, and more likely to have shorter lengths of hospice stay.⁹

Our data are limited by lack of information on social determinants of health that can dictate access to health-modifying resources and details on diagnoses, patient preferences, rate of functional decline, and caregiver support Further, differences within ethnic groups including Asian American and Pacific Islander exist, and further investigation was limited by lack of disaggregated data. Other limitations include potential misclassification in death certificate documentation.

Despite overall mortality improvement in GI cancers, there are geographic disparities in rural areas, a trend that has likely been aggravated during the pandemic because of facility closures and reduced oncologic diagnostic and treatment services.¹⁰ Interventions ensuring access to care across the care continuum (eg, tele-oncology/hospice, virtual tumor boards), increasing funding for young-onset GI cancers, and long-term investments in palliative and hospice care infrastructure (eg, capacity building, better reimbursement) are required. Further, with a population that grows more diverse in its minority composition, a tailored approach to increasing education on advanced care planning and end-of-life care for all cultural groups is needed.

Abbreviations used in this paper:

AAMR

age-adjusted mortality rate

APC

annual percentage change

GI

gastrointestinal