Abstract
Background:
Pancreatic cancer has a 5-year survival of just 10%. Services such as palliative care and hospice are thus crucial in this population, yet their geographic accessibility and utilization remains unknown.
Aim:
We studied the association between rurality of patient residence and the use of palliative care and hospice.
Design, Setting, and Participants:
Cohort study of continuously enrolled fee-for-service Medicare beneficiaries age≥65 diagnosed with incident pancreatic cancer between 04/01/2016-08/31/2018 and who died by 12/31/2018.
Results:
In this decedent cohort of 31,460 patients, 77% lived in metropolitan areas, 11% in micropolitan areas, 7% in small towns, and 5% in rural areas. Patient demographics were largely similar across rurality; however, the proportion of White, Non-Hispanic patients and social deprivation was highest in rural areas and lowest in metropolitan areas. Overall, 33% of patients used any palliative care and 77% received hospice services. After risk adjustment, there were no statistically significant differences in the use of palliative care for patients residing in metropolitan versus micropolitan, small town, or rural areas. Patients in small town (OR=0.77, 95% CI:0.69–0.86) and rural areas (OR=0.75, 95% CI:0.66–0.85) had lower adjusted odds of receiving hospice care compared to patients in metropolitan areas.
Conclusions:
The use of palliative care services captured in Medicare was low, representing either underutilization or failure to accurately measure the extent of services used. While the overall level of hospice enrollment was high, patients in rural communities had relatively lower use of hospice services compared to those in metropolitan areas.
Keywords: Hospice, palliative care, rural disparities, pancreatic cancer
Introduction
Pancreatic cancer is a leading cause of cancer mortality in the United States with a 5-year survival of 10% and 40% of patients dying within two months of diagnosis.1,2 Patients with pancreatic cancer experience complex symptomatology, including pain, fatigue, malnutrition, and pancreatic exocrine insufficiency, which when combined with treatment toxicities contribute to a poor quality of life.2–6 Given this poor prognosis, palliative care and hospice services are especially important for this population; however, the accessibility and use of these services across the country are poorly understood.
The use of palliative care and hospice in patients with cancer are associated with improved quality of life, decreased utilization of acute hospital services, and lower health care costs (independent of survival time).7–14 Timely delivery of palliative care and hospice is associated with improved outcomes. For instance, early palliative care (at least 3 months before death) was associated with fewer emergency department visits and intensive care unit admissions at the end of life.4,13–18 Similarly, hospice use at least 3 days prior to death is an established measure that is endorsed by the National Quality Forum and actively employed as a quality measure in the Medicare’s Merit-based Incentive Payment System and Oncology Care Model.16,19–21 But, geography and availability of services are barriers to accessing timely care, which can be particularly challenging for patients living in rural areas.7,22
In this study, we explored the association of patient rurality of residence with the use and timing of palliative and hospice care in a cohort of fee-for-service Medicare beneficiaries who died from pancreatic cancer. We hypothesized that rural place of residence is associated with decreased use of palliative care services as well as hospice compared to patients residing in metropolitan areas. Furthermore, we hypothesized that rural patients who did use palliative and hospice care would initiate care closer to the time of death compared to metropolitan patients. Our study builds upon the existing literature by describing the association of the rurality of a patient’s residence with the use and timing of palliative or hospice care in a contemporary, national, insured cohort of all fee-for-service Medicare beneficiaries who died from pancreatic cancer.
Methods
Study Design and Cohort
We performed a retrospective cohort study of continuously enrolled fee-for-service Medicare beneficiaries aged 65 years or older diagnosed with incident pancreatic cancer after 04/01/2016 who died on or before 12/31/2018 (Figure 1) using 100% fee-for-service Medicare claims data files. Beneficiaries were identified as having incident pancreatic cancer if meeting one or more of the following criteria from a previously validated algorithm (Supplemental Methods) from 04/01/2016 through 08/31/2018, which allowed for a 6-month look-back window and at least 120 days of follow-up after diagnosis using International Classification of Disease 10 codes.23,24
Figure 1.
Forming the analytic cohort of Medicare beneficiaries who were diagnosed incident pancreatic cancer and died before December 31, 2018.
* includes all U.S.-residing fee-for-service Medicare beneficiaries with continuous enrollment in Medicare Parts A and B between October 1, 2015 and December 31, 2018 (or through death) who did not live in a nursing home, n= 26,832,741 beneficiaries.
This study was approved by the Center for the Protection of Human Subjects at Dartmouth College (STUDY00031966) on 02/11/2020. The data that are analyzed in this study are available with permission from the Centers for Medicare and Medicaid Services. Restrictions apply to the availability of these data, which were used under Data Use Agreement 55841 for this study.
Measures
Our primary exposure was rurality of patient residence, captured using zip code in Medicare claims data. Rurality was then defined using the United States Department of Agriculture 2010 Rural-Urban Commuting Area and categorized as metropolitan (1.0–3.0), micropolitan (4.0–6.0), small town (7.0–9.0), or rural (10.0–10.3).25
The primary outcomes were dichotomous measures of any palliative care utilization and any hospice care use after diagnosis with incident pancreatic cancer. Palliative care utilization was identified as any claim in Medicare Part B, Outpatient, and Hospice files containing International Classification of Disease 10 diagnosis code Z51.5 “Encounter for Palliative Care”. Primary analyses included all patients, regardless of the time from diagnosis to death.
Secondary outcomes were binary indicators for the timing of care, including any use of palliative care after diagnosis and at least 90 days before death4 and any use of hospice care after diagnosis and at least 3 days before death.19. Initial analyses of timeliness of care included all patients regardless of time between diagnosis and death. Sensitivity analyses evaluating timeliness of palliative care or hospice, limiting the analytic denominator to patients diagnosed with pancreatic cancer at least 90 days and 3 days prior to death, respectively.
Further, to contextualize a patient’s care prior to death, we also studied additional clinical care measures such as emergency department visit within 30 days of death, intensive care unit admission within 30 days of death, and chemotherapy use within 14 days of death. See Supplemental Methods for the list of codes used to identify all outcomes.
Covariate Measures
We accounted for patient age at the start of the lookback period, sex, race/ethnicity, Medicare/Medicaid full dual eligibility (an individual-level surrogate measure for low socioeconomic status), disease stage (metastatic vs. non-metastatic), diabetes, previous myocardial infarction, congestive heart failure, stroke or transient ischemic attack, chronic obstructive pulmonary disorder, end-stage renal disease, the composite CMS-Hierarchical Condition Categories risk adjustment score (a comorbidity score, where a score of 1.0 indicates a relatively health beneficiary26), and Area Deprivation Index at the ZIP+4 level in all multivariable models.26–28 Patient race was captured as reported by individual hospitals and shared in Medicare claims and categorized as non-Hispanic White, non-Hispanic Black, Hispanic, or non-Hispanic Other.29 The Area Deprivation Index is a composite index used to assess local socioeconomic deprivation, with higher scores indicating increased deprivation. This was the only characteristic with missing data for 4.4% of beneficiaries (n=1,392). Here, we imputed the area deprivation rank using the mean values of five iterations from the imputation model that included the following variables at the ZIP+4 level: proportion below federal poverty line, median household income, and percent with less than high school education.
Statistical Analysis
We assessed differences in patient demographics and clinical characteristics across rurality using descriptive statistics (means and standard deviations; medians and interquartile ranges; and proportions) and inferential statistics including one-way analysis of variance for continuous parametric variables, Kruskal-Wallis one-way analysis of variance for nonparametric data, and chi-squared test for categorical or dichotomous variables, as appropriate.
Then, we estimated unadjusted and adjusted odds ratios using logistic regressions for our primary and secondary outcomes. All models were adjusted for the covariates listed above. We assessed statistical significance using the two-tailed p≤0.05 threshold, and all analyses were performed in Stata 17.0 (StataCorp, College Station, Texas).
Results
We identified 31,460 eligible Medicare beneficiaries. The median survival time (from diagnosis to death) was 96 days (IQR: 36–231 days) and 48.6% (N=15,286) died within 90 days of diagnosis. In this cohort, 77.1% of beneficiaries lived in metropolitan areas, 11.6% in micropolitan areas, 6.7% in small towns, and 4.6% in rural areas (Table 1).
Table 1.
Characteristics of patients who died after a diagnosis of incident pancreatic cancer by rurality of patient residence, N= 31,460.
| Characteristic | Metropolitan (N=24,261) | Micropolitan (N=3,635) | Small Town (N=2,108) | Rural (N=1,456) | p |
|---|---|---|---|---|---|
|
| |||||
| Demographics | |||||
|
| |||||
| Age, mean years (SD) | 76.3 (7.4) | 75.6 (7.1) | 75.6 (6.9) | 75.8 (6.8) | <0.001 |
| Age Categories, N (%) | <0.001 | ||||
| 65–70 years | 6,358 (26.2%) | 1,034 (28.4%) | 577 (27.4%) | 388 (26.6%) | |
| 71–75 years | 5,635 (23.2%) | 892 (24.5%) | 527 (25.0%) | 351 (24.1%) | |
| 76–80 years | 5,215 (21.5%) | 772 (21.2%) | 474 (22.5%) | 348 (23.9%) | |
| 81–85 years | 3,870 (16.0%) | 559 (15.4%) | 337 (16.0%) | 220 (15.1%) | |
| > 85 years | 3,183 (13.1%) | 378 (10.4%) | 193 (9.2%) | 149 (10.2%) | |
| Female Sex, N (%) | 12,149 (50.1%) | 1,791 (49.3%) | 1,058 (50.2%) | 683 (46.9%) | 0.11 |
| Race/Ethnicity, N (%) | <0.001 | ||||
| White, Non-Hispanic | 19,863 (81.9%) | 3,237 (89.1%) | 1,843 (87.4%) | 1,342 (92.2%) | |
| Black, Non-Hispanic | 2,172 (9.0%) | 209 (5.7%) | 150 (7.1%) | 44 (3.0%) | |
| Hispanic | 1,097 (4.5%) | 107 (2.9%) | 51 (2.4%) | 17 (1.2%) | |
| Other, Non-Hispanic | 1,129 (4.7%) | 82 (2.3%) | 64 (3.0%) | 53 (3.6%) | |
| Medicare/Medicaid Dual Eligible, N (%) | 1,725 (7.1%) | 220 (6.1%) | 143 (6.8%) | 92 (6.3%) | 0.091 |
| Area Deprivation Index, median (IQR) | 36 (18–60) | 61 (44–77) | 67 (52–81) | 66 (52–78) | <0.001 |
|
| |||||
| Clinical Characteristics | |||||
|
| |||||
| Metastatic Disease, N (%) | 14,972 (61.7%) | 2,180 (60.0%) | 1,264 (60.0%) | 889 (61.1%) | 0.11 |
| Diabetes, N (%) | 6,974 (28.7%) | 1,067 (29.4%) | 649 (30.8%) | 412 (28.3%) | 0.21 |
| Myocardial infarction, N (%) | 199 (0.8%) | 34 (0.9%) | 22 (1.0%) | 15 (1.0%) | 0.57 |
| CHF, N (%) | 1,870 (7.7%) | 289 (8.0%) | 173 (8.2%) | 116 (8.0%) | 0.82 |
| Stroke/Transient Ischemic Attack, N (%) | 538 (2.2%) | 73 (2.0%) | 36 (1.7%) | 29 (2.0%) | 0.39 |
| COPD, N (%) | 2,167 (8.9%) | 366 (10.1%) | 216 (10.2%) | 158 (10.9%) | 0.005 |
| End-stage Renal Disease, N (%) | 292 (1.2%) | 40 (1.1%) | 19 (0.9%) | 13 (0.9%) | 0.45 |
| CMS-HCC Risk score, median (IQR) | 0.7 (0.4–1.2) | 0.7 (0.4–1.1) | 0.7 (0.4–1.1) | 0.7 (0.4–1.1) | <0.001 |
SD= standard deviation; IQR= interquartile range; CHF= congestive heart failure; COPD= chronic obstructive pulmonary disease; CMS-HCC= Centers for Medicare and Medicaid Services Hierarchical Conditions Categories
Patient demographics were similar across the levels of rurality in terms of sex and Medicare-Medicaid dual eligibility. There were small differences in age distribution across rurality, where the mean age was highest in metropolitan areas, 76.3 (SD 7.4) years, and lowest in micropolitan and small town areas, 75.6 (7.1) years and 75.6 (6.9) years respectively. However, the proportion of White, Non-Hispanic patients was markedly different across rurality, where it was highest in rural areas (92.2%) and lowest in metropolitan areas (81.9%, p<0.001). Likewise, social deprivation was twice as high in non-metropolitan areas compared to metropolitan areas (median Area Deprivation Index in metropolitan=36 versus micropolitan=61, small town= 67, rural= 66, p<0.001). There were no notable differences in clinical characteristics by rurality.
Use of palliative care and hospice care
In this cohort, the proportion of patients using palliative care was 32.8% (n=10,322), with statistically significant but small differences across region (Table 2). Adjusting for confounding, there was no significant difference in the use of palliative care across the levels of rurality (Table 3). The full regression model estimates are available in Supplemental Table 1.
Table 2.
First time use of hospice or palliative care among patients who died of pancreatic cancer by rurality of patient residence, N= 31,460.
| Events | Metropolitan (N=24,261) | Micropolitan (N=3,635) | Small Town (N=2,108) | Rural (N=1,456) | p |
|---|---|---|---|---|---|
|
| |||||
| Palliative Care | |||||
| Any Palliative Care, N (%) | 8,094 (33.4%) | 1,109 (30.5%) | 651 (30.9%) | 468 (32.1%) | 0.001 |
| Time from use to death, median days (IQR) | 23.0 (8.0–67.0) | 22.0 (8.0–64.0) | 20.0 (6.0–60.0) | 24.0 (7.0–78.5) | 0.052 |
| Palliative care ≥ 30 days before death, N (%) | 3,590 (44.4%) | 458 (41.3%) | 271 (41.6%) | 211 (45.1%) | 0.15 |
| Palliative care ≥ 90 days before death, N (%) | 1,532 (18.9%) | 202 (18.2%) | 106 (16.3%) | 100 (21.4%) | 0.17 |
| Hospice | |||||
| Any Hospice, N (%) | 18,842 (77.7%) | 2,825 (77.7%) | 1,571 (74.5%) | 1,081 (74.2%) | <0.001 |
| Time from use to death, median days (IQR) | 14.0 (5.0–38.0) | 16.0 (5.0–41.0) | 16.0 (5.0–44.0) | 17.0 (6.0–44.0) | <0.001 |
| Hospice ≥ 3 days before death, N (%) | 16,265 (86.3%) | 2,460 (87.1%) | 1,380 (87.8%) | 942 (87.1%) | 0.25 |
| Hospice ≥ 7 days before death, N (%) | 12,993 (69.0%) | 2,007 (71.0%) | 1,102 (70.1%) | 790 (73.1%) | 0.006 |
Table 3.
Association between rurality of patient residence and any use of palliative care or hospice.
| Patient Residence | Any palliative care use | Any hospice use | ||||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| Unadjusted | Adjusted | Unadjusted | Adjusted | |||||
|
| ||||||||
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
|
| ||||||||
| Metropolitan | 1.00 (ref) | - | 1.00 (ref) | - | 1.00 (ref) | - | 1.00 (ref) | - |
| Micropolitan | 0.88 (0.81 – 0.95) | 0.001 | 0.93 (0.86 – 1.01) | 0.09 | 1.00 (0.92 – 1.09) | 0.943 | 0.93 (0.85 – 1.02) | 0.106 |
| Small Town | 0.89 (0.81 – 0.98) | 0.02 | 0.98 (0.88 – 1.08) | 0.638 | 0.84 (0.76 – 0.93) | 0.001 | 0.77 (0.69 – 0.86) | <0.001 |
| Rural | 0.95 (0.84 – 1.06) | 0.338 | 1.01 (0.90 – 1.14) | 0.807 | 0.83 (0.73 – 0.94) | 0.002 | 0.75 (0.66 – 0.85) | <0.001 |
OR= odds ratio; CI= confidence interval
Adjusted models included the following covariates: age, sex, race/ethnicity, Medicare/Medicaid full dual eligibility, disease stage (metastatic vs. not), diabetes, previous myocardial infarction, congestive heart failure, stroke or transient ischemic attack, chronic obstructive pulmonary disorder, end-stage renal disease, the composite CMS-Hierarchical Condition Categories risk adjustment score, and ADI.
A majority of patients (77.3%, n=24,319) received some hospice care prior to death, with slight variations across rurality. The highest rates of hospice use were in metropolitan and micropolitan areas, at 77.7% of patients; small town and rural areas had slightly lower rates of hospice at 74.5% and 74.2% respectively (p<0.001, Table 2). Controlling for confounding factors, small town (OR 0.77; 95% CI 0.69–0.86, p<0.001) and rural residence (OR 0.75; 95% CI 0.66–0.85, p<0.001) was associated with a significantly lower odds of using hospice care as compared to urban patients (Table 3). The full regression model output is available in Supplemental Table 2.
Evaluation of timeliness of palliative care and hospice
Among the 10,322 patients who received palliative care, the median time from palliative care use to death was 23 days (7–67 days). The proportion of patients receiving palliative care by 30 and 90 days before death are shown in Table 2; these rates did not vary significantly across rurality. Among all patients who survived at least 90 days following diagnosis (N=16,174), 12% (N=1,940) received palliative care at least 90 days before death. Figure 2 shows the results of this measure stratified across the levels of rurality. After risk adjustment, there were no statistically significant differences in receipt of palliative care at least 90 days before death for micropolitan and small town patients, in comparison with patients living in metropolitan areas (Table 4). However, patients in rural areas had a higher odds of receiving palliative care at least 90 days before death (OR 1.29, 95% CI 1.03–1.62, p=0.027).
Figure 2. Proportion of patients who used (A) palliative care or (B) hospice, by rurality of patient residence.
Blue bars indicate the overall proportion using that service at any time between diagnosis and death. Green bars indicate the proportion using that service within the time period for the respective measure: at least 90 days before death for palliative care and at least 3 days before death for hospice. Note, for timeliness indicators, the denominator patient population was limited those who survived the minimum number of days to qualify for the timeliness measure, e.g., at least 90 days survival from diagnosis for palliative care measure (N=16,174) and 3-days for hospice (N=31,123).
Table 4.
Association between patient rurality and timeliness indicators for hospice and palliative care use. Note the patient population was limited those who survived the minimum number of days to qualify for the timeliness measure, e.g., at least 90 days survival from diagnosis for palliative care measure (N=16,174) and 3 days for hospice (N=31,123).
| Patient Residence | Palliative care at least 90 days before death | Hospice use at least 3 days before death | ||||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| Unadjusted | Adjusted | Unadjusted | Adjusted | |||||
|
| ||||||||
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
|
| ||||||||
| Metropolitan | 1.00 (ref) | - | 1.00 (ref) | - | 1.00 (ref) | - | 1.00 (ref) | - |
| Micropolitan | 0.89 (0.76 – 1.04) | 0.15 | 0.92 (0.78 – 1.09) | 0.33 | 1.03 (0.96 – 1.11) | 0.44 | 0.97 (0.89 – 1.05) | 0.44 |
| Small Town | 0.82 (0.67 – 1.01) | 0.065 | 0.91 (0.73 – 1.13) | 0.40 | 0.93 (0.85 – 1.02) | 0.13 | 0.87 (0.79 – 0.96) | 0.008 |
| Rural | 1.17 (0.94 – 1.46) | 0.15 | 1.29 (1.03 – 1.62) | 0.027 | 0.91 (0.81 – 1.02) | 0.11 | 0.86 (0.76 – 0.96) | 0.010 |
OR= odds ratio; CI= confidence interval
Adjusted models included the following covariates: age, sex, race/ethnicity, Medicare/Medicaid full dual eligibility, disease stage (metastatic vs. not), diabetes, previous myocardial infarction, congestive heart failure, stroke or transient ischemic attack, chronic obstructive pulmonary disorder, end-stage renal disease, the composite CMS-Hierarchical Condition Categories risk adjustment score, and ADI.
Among patients who survived at least 3 days following cancer diagnosis (N=31,123), 67.6% received hospice 3 or more days before death, and this percentage did not vary across rurality (p=0.13, Figure 2). Among the 24,319 patients who used any hospice care, 86.5% (n=21,047) started hospice 3 or more days before death, and this also did not vary across rurality (p=0.25, Table 2). On unadjusted analyses, we found that the odds of receiving hospice care at least 3 days before death was not significantly different across rurality (Table 4). However, after adjusting for patient demographics and clinical factors, compared to metropolitan areas, patients in small town (OR 0.87, 95% CI 0.79–0.97, p=0.008 and rural areas (OR 0.86, 95% CI 0.76–0.96, p=0.010) had reduced odds of receiving hospice care at least 3 days before death.
Additional clinical care measures
Almost half of all patients had an emergency department visit within 30 days of death, with the highest rates in metropolitan areas (52.0%), followed by small town (49.9%), micropolitan (49.0%) and rural areas (46.9%, p<0.001) (Supplemental Table 3). Similarly, the highest proportions of intensive care unit admissions were in metropolitan areas (23.3%), and lowest in rural areas (16.4%, p<0.001); however, the median time from intensive care unit admission to death was similar across all regions (p=0.84). Chemotherapy use within 14 days of death was low overall, at 1.6%; this ranged from 0.5 – 2.0% across geographic regions but did not differ significantly (p=0.31).
Discussion
In this study of 31,460 Medicare beneficiaries age 65+ who died following an incident pancreatic cancer diagnosis, we found low use of palliative care overall and lack of timely palliative care in many of those who received it at all. Patients in rural communities had a significantly higher odds of receiving palliative care at least 90-days before death as compared to those in metropolitan areas. However, we also observed high overall rates of hospice use (77.3%), of which 86.5% was initiated at least 3 days before death. Patients in small towns and rural areas had significantly lower odds of receiving hospice care as compared to those in urban settings. Furthermore, aggressiveness of care measures such as any emergency department use and intensive care unit admission in the last 30 days of life were quite high at 51% and 22% respectively, but chemotherapy in the last 14 days of life was only 1.6%, with no differences by rurality.
Use of palliative and hospice care
Our findings are consistent with prior studies of pancreatic cancer cohorts. Early reports demonstrate a similarly low (6.8% overall) utilization of palliative care among Medicare beneficiaries with pancreatic cancer, but with an upward trend between 2000–2009.5 Likewise, reports note overall rates of hospice use similar to our study. A 2000–2011 Surveillance, Epidemiology, and End Results-Medicare linked study found that 70% of patients with pancreatic adenocarcinoma used hospice.22 A similar evaluation of pancreatic cancer patients who received pancreatectomy from 2013–2017 found that 74% of patients received hospice services prior to death.30 These last two studies noted similar disparities in hospice use by rurality.22,30,31 While previous studies also suggested an increase in aggressiveness of care measures among rural patients, we did not observe the same trend despite finding similar overall rates of emergency department visits and intensive care unit admissions.11,32
Our rurality findings could be due to differential coding for services and thus better capture of palliative care for rural patients. Rural patients disproportionately travel to urban centers for surgical and oncologic services compared to those in suburban areas. These urban centers may be more likely to include explicitly coded visits with palliative care providers, therefore capturing care provided better than other facilities. However, it also could reflect receipt of palliative care consultation for rural patients. Such services could continue through time of death without formal transition to services that are captured as hospice in this cohort, thus also explaining the concurrent lower odds of hospice care for rural patients identified in our study as well as the existing literature.22,30
Our overall low percentage of patients receiving palliative care could be related to coding and measurement of palliative care services in claims data. Administrative claims data does not capture all the means of delivering palliative care. Using natural language processing of electronic medical records, other authors have found higher rates of palliative care use; 68% of encounters of patients with known advanced pancreatic cancer were associated with at least one palliative care process such as code status clarification, goals of care discussions, referral to a palliative care specialist, and hospice assessment, which is much higher than the 33% of any use of palliative care we identified in Medicare claims.12 Thus, further research into building claims-based algorithms to identify palliative care will be important for future studies.
The challenge of accurately capturing palliative care services also has important policy implications. First, current estimates of services provided and any unmet need influences efforts to ensure an adequate specialty palliative care workforce, especially in rural areas. Additionally, incomplete capture of services already provided also hinders evaluation of specific barriers to palliative care in rural areas and other marginalized populations. Finally, the operationalization of palliative care quality indicators may be limited due to inconsistent and incomplete capture of palliative services actually provided in communities, especially by primary care or oncology specialists.
Timeliness of palliative and hospice care
Interpreting our findings on the timeliness of receipt of palliative and hospice care is more challenging. Although several societies such as the World Health Organization, American Society of Clinical Oncology, the American Cancer Society, and the Institute of Medicine recommend early initiation of palliative care processes along with treatment for pancreatic cancer, uniform consensus on the definition of timely palliative care is lacking.33 Based on a review of the literature, several studies report on a 3-month or 90-day minimum use of palliative care to define early initiation, which is the definition we used in this study.15,16,33 We found that 49% of our cohort died within 90 days, and were thus ineligible to qualify for timely palliative care if the threshold is set at 3-months. Therefore, a 90-day window for evaluating the timely use of palliative care might be inappropriate for this population. Among patients who receive any palliative care in previous studies, about one-third received it in a timely fashion which is quite different to the 12% of patients receiving palliative care within 90-days in our cohort.4,15 Despite the variation, the overall recorded use of timely palliative care for pancreatic cancer patients remains discouragingly low and the geographic variations identified were still notable, namely an increased odds for palliative care within 90 days of diagnosis for rural versus metropolitan-residing patients.
Though guidelines exist for integrating quality palliative and hospice care into cancer care, barriers remain for the provision of this care. Rural residents have poorer geographical access to hospice, perhaps serving as a reason for their decreased use.34 Challenges due to low patient volume, increased medication costs, and insufficient Medicare reimbursement have also been cited as barriers to providing hospice care in rural areas.34 Lastly, physician recruitment and retention and high turnover of community physicians in rural areas can complicate the referral process and care coordination, a challenge applicable to both palliative care and hospice services.34 Emerging delivery models, including telehealth, and community-based partnerships to train local healthcare workers can potentially bridge this gap to offer palliative and end-of-life care in rural health care settings, thereby expanding access of improved quality of life and health outcomes from these programs to rural cancer patients.35 Further study is needed to understand the receipt and timing of palliative care and hospice care along with the overall continuum of end-of-life services in rural populations.
While timeliness measures as proposed through the American Society of Clinical Oncology Quality Oncology Practice Indicators and the National Quality Forum exist for hospice care use, standardized reporting remains inconsistent throughout published literature. Several studies have previously evaluated early versus late initiation of hospice care, but there is inconsistency in the definitions used for these terms.30 In general, we found that the results on the timeliness of hospice use are encouraging, with studies reporting that the majority (83–85%) of hospice use occurred at least 3 days prior to death, corroborating our own results that 85% of patients using hospice received it in a timely manner.22,30 Including all patients who died from pancreatic cancer, however, over 30% failed to receive hospice at least 3 days prior to death, highlighting opportunities to improve.
Limitations
Our study builds on existing literature with the use of a contemporary national sample, and further contributes insight into the role of patient rurality in these associations. Nevertheless, our findings should be considered in the context of the limitations of our data. Although the median age of pancreatic cancer diagnosis is 70 years, by using Medicare claims data limited to older patients, our findings might not be generalizable to younger populations of cancer patients. Second, we are unable to account for cancer stage at diagnosis. However, given the poor prognosis for pancreatic cancer and our inclusion of only patients that died makes this limitation less critical. Lastly, while palliative care can be delivered in many ways, if it was not billed to Medicare, then it was not captured in our analysis, leading to a likely undercounting of services. For example, patients that receive palliative care through their medical oncologist or primary care provider without an explicit coding for “encounter for palliative care” are not counted as receiving palliative services. We attempted to address this issue by identifying claims that included a palliative care provider as evidence of palliative care utilization, but this did not meaningfully change the absolute utilization.
Conclusion
Palliative and hospice care are an essential component of comprehensive cancer care delivery for pancreatic cancer. Of Medicare beneficiaries who die from pancreatic cancer, we report only 33% of patients received palliative care with no variation by patient location. Over 75% of beneficiaries received hospice care but the odds of receiving hospice care were significantly lower for small town and rural beneficiaries compared to metropolitan patients. Rural residence remains a barrier for access to quality end-of-life care for cancer patients. These findings provide critical national appraisal of hospice use and potential limitations in measurement of palliative care use. Targeted efforts to improve the utilization of and ability to identify timely provision of palliative care and hospice across all populations and geographies with a focus on low-resource rural settings are needed to ensure equitable end-of-life care.
Supplementary Material
Funding Sources:
The authors disclosed receipt of financial support for the research, authorship, and/or publication of this article by the National Cancer Institute (NCI R01 CA248470).
Footnotes
Disclosures: We have no conflicts of interest to disclose.
References
- 1.Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer Statistics, 2021. CA: a cancer journal for clinicians. Jan 2021;71(1):7–33. doi: 10.3322/caac.21654 [DOI] [PubMed] [Google Scholar]
- 2.Adsersen M, Chen IM, Rasmussen LS, et al. Regional and age differences in specialised palliative care for patients with pancreatic cancer. BMC Palliative Care. 2021/December/20 2021;20(1):192. doi: 10.1186/s12904-021-00870-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rasmussen LS, Fristrup CW, Jensen BV, et al. Initial treatment and survival in 4163 Danish patients with pancreatic cancer: A nationwide unselected real-world register study. European journal of cancer (Oxford, England : 1990). Apr 2020;129:50–59. doi: 10.1016/j.ejca.2020.01.015 [DOI] [PubMed] [Google Scholar]
- 4.Michael N, Beale G, O’Callaghan C, et al. Timing of palliative care referral and aggressive cancer care toward the end-of-life in pancreatic cancer: a retrospective, single-center observational study. BMC Palliative Care. 2019/January/28 2019;18(1):13. doi: 10.1186/s12904-019-0399-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bhulani N, Gupta A, Gao A, et al. Palliative care and end-of-life health care utilization in elderly patients with pancreatic cancer. J Gastrointest Oncol. 2018;9(3):495–502. doi: 10.21037/jgo.2018.03.08 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Shavers VL, Harlan LC, Jackson M, Robinson J. Racial/Ethnic Patterns of Care for Pancreatic Cancer. Journal of Palliative Medicine. 2009/July/01 2009;12(7):623–630. doi: 10.1089/jpm.2009.0036 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sheffield KM, Boyd CA, Benarroch-Gampel J, Kuo YF, Cooksley CD, Riall TS. End-of-life care in Medicare beneficiaries dying with pancreatic cancer. Cancer. Nov 1 2011;117(21):5003–12. doi: 10.1002/cncr.26115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Connor SR, Pyenson B, Fitch K, Spence C, Iwasaki K. Comparing hospice and nonhospice patient survival among patients who die within a three-year window. J Pain Symptom Manage. Mar 2007;33(3):238–46. doi: 10.1016/j.jpainsymman.2006.10.010 [DOI] [PubMed] [Google Scholar]
- 9.Obermeyer Z, Makar M, Abujaber S, Dominici F, Block S, Cutler DM. Association between the Medicare hospice benefit and health care utilization and costs for patients with poor-prognosis cancer. JAMA. 2014;312(18):1888–1896. doi: 10.1001/jama.2014.14950 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pyenson B, Connor S, Fitch K, Kinzbrunner B. Medicare cost in matched hospice and non-hospice cohorts. J Pain Symptom Manage. Sep 2004;28(3):200–10. doi: 10.1016/j.jpainsymman.2004.05.003 [DOI] [PubMed] [Google Scholar]
- 11.Jang RW, Krzyzanowska MK, Zimmermann C, Taback N, Alibhai SMH. Palliative Care and the Aggressiveness of End-of-Life Care in Patients With Advanced Pancreatic Cancer. JNCI: Journal of the National Cancer Institute. 2015;107(3)doi: 10.1093/jnci/dju424 [DOI] [PubMed] [Google Scholar]
- 12.Udelsman BV, Lilley EJ, Qadan M, et al. Deficits in the Palliative Care Process Measures in Patients with Advanced Pancreatic Cancer Undergoing Operative and Invasive Nonoperative Palliative Procedures. Annals of Surgical Oncology. 2019/December/01 2019;26(13):4204–4212. doi: 10.1245/s10434-019-07757-2 [DOI] [PubMed] [Google Scholar]
- 13.Bakitas M, Lyons KD, Hegel MT, et al. Effects of a palliative care intervention on clinical outcomes in patients with advanced cancer: the Project ENABLE II randomized controlled trial. JAMA. 2009;302(7):741–749. doi: 10.1001/jama.2009.1198 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bakitas MA, Tosteson TD, Li Z, et al. Early Versus Delayed Initiation of Concurrent Palliative Oncology Care: Patient Outcomes in the ENABLE III Randomized Controlled Trial. J Clin Oncol. May 1 2015;33(13):1438–45. doi: 10.1200/jco.2014.58.6362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hui D, Kim SH, Roquemore J, Dev R, Chisholm G, Bruera E. Impact of timing and setting of palliative care referral on quality of end-of-life care in cancer patients. Cancer. Jun 1 2014;120(11):1743–9. doi: 10.1002/cncr.28628 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Earle CC, Park ER, Lai B, Weeks JC, Ayanian JZ, Block S. Identifying Potential Indicators of the Quality of End-of-Life Cancer Care From Administrative Data. Journal of Clinical Oncology. 2003/March/15 2003;21(6):1133–1138. doi: 10.1200/JCO.2003.03.059 [DOI] [PubMed] [Google Scholar]
- 17.Maltoni M, Scarpi E, Dall’Agata M, et al. Systematic versus on-demand early palliative care: results from a multicentre, randomised clinical trial. European Journal of Cancer. 2016/September/01/ 2016;65:61–68. doi: 10.1016/j.ejca.2016.06.007 [DOI] [PubMed] [Google Scholar]
- 18.Bevins J, Bhulani N, Goksu SY, et al. Early Palliative Care Is Associated With Reduced Emergency Department Utilization in Pancreatic Cancer. American Journal of Clinical Oncology. // 2021;44(5):181–186. doi: 10.1097/COC.0000000000000802 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Rosenblum R, Huo R, Scarborough B, Goldstein N, Smith CB. Comparison of Quality Oncology Practice Initiative Metrics in Solid Tumor Oncology Clinic With or Without Concomitant Supportive Oncology Consultation. Journal of Oncology Practice. 2018/December/01 2018;14(12):e786–e793. doi: 10.1200/JOP.18.00380 [DOI] [PubMed] [Google Scholar]
- 20.Wasp GT, Alam SS, Brooks GA, et al. End-of-life quality metrics among medicare decedents at minority-serving cancer centers: A retrospective study. Cancer medicine. Mar 2020;9(5):1911–1921. doi: 10.1002/cam4.2752 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Meier DE. Increased access to palliative care and hospice services: opportunities to improve value in health care. The Milbank Quarterly. 2011;89(3):343–380. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nipp RD, Tramontano AC, Kong CY, Hur C. Patterns and predictors of end-of-life care in older patients with pancreatic cancer. Cancer medicine. Dec 2018;7(12):6401–6410. doi: 10.1002/cam4.1861 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Bronson MR, Kapadia NS, Austin AM, et al. Leveraging linkage of cohort studies with administrative claims data to identify individuals with cancer. Medical care. 2018;56(12):e83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Setoguchi S, Solomon DH, Glynn RJ, Cook EF, Levin R, Schneeweiss S. Agreement of diagnosis and its date for hematologic malignancies and solid tumors between medicare claims and cancer registry data. Cancer causes & control : CCC. Jun 2007;18(5):561–9. doi: 10.1007/s10552-007-0131-1 [DOI] [PubMed] [Google Scholar]
- 25.Service UER. 2010. Rural-Urban Commuting Area (RUCA) Codes. Accessed November 9, 2022. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes/documentation/
- 26.Li P, Kim MM, Doshi JA. Comparison of the performance of the CMS Hierarchical Condition Category (CMS-HCC) risk adjuster with the charlson and elixhauser comorbidity measures in predicting mortality. BMC Health Services Research. 2010/August/20 2010;10(1):245. doi: 10.1186/1472-6963-10-245 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Kind AJH, Buckingham WR. Making Neighborhood-Disadvantage Metrics Accessible — The Neighborhood Atlas. New England Journal of Medicine. 2018;378(26):2456–2458. doi: 10.1056/NEJMp1802313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Health UoWSoMP. Area Deprivation Index Version 3.0. Accessed August 1, 2020. https://www.neighborhoodatlas.medicine.wisc.edu/
- 29.Eicheldinger C, Bonito A. More accurate racial and ethnic codes for Medicare administrative data. Health Care Financ Rev. Spring 2008;29(3):27–42. [PMC free article] [PubMed] [Google Scholar]
- 30.Paredes AZ, Hyer JM, Tsilimigras DI, et al. Hospice utilization among Medicare beneficiaries dying from pancreatic cancer. Journal of surgical oncology. Sep 2019;120(4):624–631. doi: 10.1002/jso.25623 [DOI] [PubMed] [Google Scholar]
- 31.Paredes AZ, Hyer JM, Palmer E, Lustberg MB, Pawlik TM. Racial/Ethnic Disparities in Hospice Utilization Among Medicare Beneficiaries Dying from Pancreatic Cancer. Article. Journal of Gastrointestinal Surgery. 2021;25(1):155–161. doi: 10.1007/s11605-020-04568-9 [DOI] [PubMed] [Google Scholar]
- 32.Lees C, Weerasinghe S, Lamond N, Younis T, Ramjeesingh R. Palliative Care Consultation and Aggressive Care at End of Life in Unresectable Pancreatic Cancer. Current Oncology. 2019;26(1)doi: 10.3747/co.26.4389 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Hui D, Hannon BL, Zimmermann C, Bruera E. Improving patient and caregiver outcomes in oncology: Team-based, timely, and targeted palliative care. CA: a cancer journal for clinicians. Sep 2018;68(5):356–376. doi: 10.3322/caac.21490 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Charlton M, Schlichting J, Chioreso C, Ward M, Vikas P. Challenges of Rural Cancer Care in the United States. Oncology (Williston Park, NY). September/19 2015;29 [PubMed] [Google Scholar]
- 35.Bakitas MA, Elk R, Astin M, et al. Systematic Review of Palliative Care in the Rural Setting. Cancer Control. 2015;22(4):450–464. doi: 10.1177/107327481502200411 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.