Health System and Beneficiary Costs Associated With Intensive End-of-Life Medical Services

IMPORTANCE Despite recommendations to reduce intensive medical treatment at the end of life, many patients with cancer continue to receive such services. OBJECTIVE To quantify expected beneficiary and health system costs incurred in association with receipt of intensive medical services in the last month of life. DESIGN, SETTING, AND PARTICIPANTS This retrospective cohort study used data collected nationally from Medicare and the Veterans Health Administration for care provided in fiscal years 2010 to 2014. Participants were 48937 adults aged 66 years or older who died of solid tumor and were continuously enrolled in fee-for-service Medicare and the Veterans Health Administration in the 12 months prior to death. The data were analyzed from February to August 2019. EXPOSURES American Society of Clinical Oncology metrics regarding medically intensive services provided in the last month of life, including hospital stay, intensive care unit stay, chemotherapy, 2 or more emergency department visits, or hospice for 3 or fewer days. MAIN OUTCOMES AND MEASURES Costs in the last month of life associated with receipt of intensive medical services were evaluated for both beneficiaries and the health system. Costs were estimated from generalized linear models, adjusting for patient demographics and comorbidities and conditioning on geographic region. RESULTS Of 48937 veterans who received care through the Veterans Health Administration and Medicare, most were white (90.8%) and male (98.9%). More than half (58.9%) received at least 1 medically intensive service in the last month of life. Patients who received no medically intensive service generated a mean (SD) health system cost of $7660 ($1793), whereas patients who received


Introduction
2][3] Much of that difference is no doubt because of unavoidable costs of serious illness. 4However, for patients with cancer, it is often possible to predict when intensive medical services have lost much of their potential benefit. 4For that reason, the National Academy of Medicine and the American Society of Clinical Oncology (ASCO) recommend a reduction in use of intensive medical services at the end of life, noting it is at odds with the focus on palliation and reduction in patient suffering that should characterize health care at this time. 5,6spite such recommendations, patients continue to receive intensive medical services in the last month of life. 7,8Intensive services at the end of life are not linked to better outcomes, [9][10][11] are associated with poorer patient quality of life, 12,13 and are considered undesirable by many patients. 14,15An ancillary consequence of poor-quality end-of-life care is its cost, both to the health care system and to patients, who often bear nontrivial cost-sharing.Medical care is the leading cause of personal bankruptcy in the United States, and insurance is not sufficiently protective against patient financial consequences; most persons experiencing medical bankruptcy were insured at the time of their illness. 16w studies have tried to quantify the financial consequences associated with these end-of-life intensive medical services.In the present study, we evaluate the additional costs incurred for patients who receive intensive medical services at the end of life, from both a health system and a beneficiary perspective.Although the medical community is no doubt aware that costs increase as use of health services increases, our goals in the present study are (1) to quantify the magnitude of that association, including for specific intensive medical services, and (2) to shed light on patient financial responsibility for medically intensive end-of-life services.
We focus the present study on patients dying of cancer, for 5 reasons.First, almost 40% of people in the United States will develop cancer at some point in their lives 17 ; and cancer accounts for nearly 1 in 4 US deaths. 18Second, patient financial consequences for cancer care are especially high, 19 with almost half of patients reporting cancer care-related financial strain, 20,21 and many patients forgoing or discontinuing cancer treatment partly for financial reasons. 21,22Third, literature indicates death from solid tumor may be easier to prognosticate than death from other chronic illness. 4urth, the National Quality Forum (NQF)-endorsed ASCO guidelines for appropriate end-of-life care are premised on the assumption that death from cancer is able to be anticipated; therefore, there is a growing consensus that cancer-related death can often be forecast and that care decisions can be made based on that understanding.Fifth, in recognition of the high costs and limited health benefit of some cancer care, ASCO has recommended physicians discuss the value-the costs and likely outcomes-of treatment strategies with their patients.Yet physicians often have limited knowledge of the costs of care, 23,24 making informed discussions challenging if not impossible.The present study quantifies the financial consequences of medically intensive end-of-life services and provides physicians with reference estimates that may be of use.

Methods
Consistent with ASCO/NQF measures, we evaluated total costs of care in the last month of life for patients who did or who did not receive the following medically intensive services: 2 or more emergency department visits; chemotherapy; a hospital admission without an intensive care unit (ICU) stay; an ICU stay; or hospice for fewer than 3 days in the last month of life.We evaluated both oral and intravenous chemotherapy.Of note, ASCO/NQF include metrics for chemotherapy use in the last 14 days of life.6][27] Recognizing that the specific definition of "end of life" is subjective, and with a goal of being more inclusive, we similarly chose to evaluate chemotherapy in the last 30 days of life.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology

JAMA Network Open | Geriatrics
Health System and Beneficiary Costs for Intensive End-of-Life Medical Services (STROBE) reporting guideline for cohort studies.This study was approved by and received a waiver of informed consent from the Stanford University Institutional Review Board because the research involved no more than minimal risk to the participants and the waiver would not adversely affect the rights and welfare of the participants.

Study Population
Cohort members were veterans who died of cancer in fiscal years 2010 to 2014 and were identified using Surveillance, Epidemiology, and End Results-Medicare International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes for underlying cause of death from National Death Index death certificate data. 28,291][32] We made 2 adjustments to ASCO/NQF metrics to increase the likelihood that physicians were better able to anticipate their patients' deaths: we limited the cohort to persons dying of solid tumor and to those who had an ICD-9 code for cancer for at least 6 months in administrative data.These criteria have the effect of providing more conservative estimates of intensive medical services at the end of life than would strict ASCO/NQF criteria.
Cohort members were aged 66 years or older and were continuously enrolled in fee-for-service Medicare in the 12 months prior to death; the latter criterion allowed for full capture of use.4][35][36][37] Medicare data were included in this analysis because other work indicates excluding Medicare substantially underestimates total costs of veteran care. 38We excluded patients who were also enrolled in Medicaid owing to the unavailability of those cost data; our cohort was limited to patients for whom we had complete capture of costs and utilization.

Costs
The health system costs of intensive medical services in the VA were obtained from VA Managerial Cost Accounting data that we linked to inpatient, outpatient, and pharmacy administrative utilization data.Beneficiary costs in the VA are not directly available and were assigned using VA national office guidance, based on patients' utilization of outpatient primary care, specialty care, and days of inpatient service use and enrollment priority. 39For example, patients who have an enrollment priority of 7 or 8 are eligible for copayments in VA.Patients in priority group 8 have the highest copayments; if these patients received inpatient care, their cost-sharing was $10 per day plus approximately $1200 for the first 90 days, and $600 for the next 90 days.Medicare health system and beneficiary costs and utilization were obtained from the MEDPAR, Outpatient, Carrier, Durable Medical Equipment, Hospice, Home Health and Part D administrative files.Beneficiary costs are a separate variable in the claims data present in these files.We also included care that was delivered in the community and paid for by the VA, known as Fee-Basis care.Taken together, these represent the entirety of costs incurred by patients in their last month of life.Medicare data are reimbursements for care provided, whereas VA data are cost estimates based on activity-based cost accounting.
Patient costs in both systems are beneficiary expectations of payment, rather than patient copayments because the latter are not available in the Medicare or VA research data.They therefore represent the starting point for patient copayments, rather than final patient copayments; yet they remain the most comprehensive data available for research purposes.For simplicity, we refer to all as costs.Costs were inflation adjusted to 2014 dollars using the personal consumption expenditures index, 40 as recommended by the Second Panel on Cost-Effectiveness in Health and Medicine. 41

Statistical Analysis
We estimated the association between receipt of medically intensive services and costs in the last month of life using a generalized linear model.A modified Park test and a Box-Cox regression recommended a gamma distribution with a log link function. 42Models were adjusted for age as a

JAMA Network Open | Geriatrics
Health System and Beneficiary Costs for Intensive End-of-Life Medical Services categorical variable, race, the cancer category representing the underlying cause of death, and Elixhauser comorbidities. 43Models were conditioned on geographic area (hospital referral region) to account for both geographic differences in the practice of end-of-life care 9 and geographic variation in wages; this method allows for interpretation of the cost differences between medically intensive and nonmedically intensive services within geographic regions.Models were used to estimate the additional costs associated with medically intensive services, with bias-corrected 95% CIs generated through bootstrapping with 1000 replications.We adjusted for multiple hypothesis testing by constraining the familywise error rate to no more than 0.05 (2-tailed level of significance) across the 5 outcomes studied. 44

Results
The study cohort consisted of 48 937 veterans who received care through the VA and Medicare.Our cohort was majority white (90.8%), male (98.9%), and had a mean (SD) of 8.8 (3.9) comorbidities (Table 1).Cohort members were most likely to die of lung and bronchus cancer (31.1%) or prostate cancer (20.8%).Cohort members had a cancer diagnosis for a median (interquartile range) of 34.6 (16.0-54.8)months before death.In unadjusted analyses, patients who received a medically intensive service had a longer time from cancer diagnosis to death than patients who did not (P < .001using a Wilcoxon rank sum test) (Table 1).
More than half of the cohort (58.9%) received at least 1 medically intensive service in the last month of life (Table 2).Patients who received medically intensive services were significantly more likely to have comorbidities, although the difference was small (Table 1).The most frequently occurring intensive service was insufficient hospice exposure (36.6%), followed by a hospital stay without an ICU admission (30.3%); the least frequently occurring intensive service was chemotherapy (11.0%) (Table 2).Patients were most likely to receive 1 (28.6%) or 2 (22.1%) intensive services in the last month of life; few patients had 3 or more intensive services.
Costs varied substantially by type of intensive service provided, with ICU stays associated with highest additional financial consequence and chemotherapy associated with lowest additional financial consequence for the health system (Figure 1 and Figure 2).Patients with an ICU stay in the last month of life had $21 093 (95% CI, $20 364-$21 689) higher health system costs and $1222 (95% CI, $1178-$1238) higher expected beneficiary costs than those who did not have an ICU stay.Patients with a non-ICU hospital stay had $8590 (95% CI, $8224-$8772) higher health system costs and $771 (95% CI, $749-$778) higher expected beneficiary costs than those who did not have a non-ICU hospital stay.Patients with 2 or more emergency department visits in the last month of life had $11 140 (95% CI, $10 623-$11 495) higher health system costs and $879 (95% CI, $853-$901) higher expected beneficiary costs.Patients who received 3 or fewer days of hospice had $13 134 (95% CI, $12 713-$13 501) higher health system costs and $811 (95% CI, $778-$825) higher expected beneficiary costs than those who received hospice for the greater (recommended) number of days.
Patients who received chemotherapy had $3460 (95% CI, $2927-$3880) higher health system costs and $942 (95% CI, $888-$969) higher expected beneficiary costs than those who did not receive any chemotherapy in the last month of life.
The total estimated mean (SD) health care system costs for patients with intensive services were as follows: $23 612 ($5528) for any intensive service; $35 235 ($10 067) for hospital admission with ICU stay; $27 007 ($8629) for 2 or more emergency department visits; $25 438 ($6729) for  a All differences between no medically intensive service and receipt of such service are statistically significant after adjusting for multiple hypothesis testing (all P < .001).

Discussion
Despite recommendations more than half of cancer decedents receive medically intensive services in the last month of life.We found the costs associated with nonrecommended intensive services added a mean of almost $16 000 to health care system costs and more than $1100 to expected beneficiary costs in the last month of life alone, bringing total spending in the last month of life to a mean of $24 000 for the health system and $1300 for the beneficiary.Although it is not

JAMA Network Open | Geriatrics
Health System and Beneficiary Costs for Intensive End-of-Life Medical Services surprising that intensive medical services cost more, quantifying the magnitude of these costs can help spur efforts to reduce it.
Prior studies have estimated overall health system costs in the last 6 months of life for Medicare beneficiaries to range from means of $41 712 to $74 212 45,46 and medians around $22 000. 47,48To our knowledge, no other study has quantified the additional costs associated with intensive medical services at the end of life, and there is no other published evaluation of beneficiary expectations of payment or patient costs associated with intensive medical services at the end of life.In this work, we found ICU stays were associated with the highest excess financial consequence to the health system, of more than $21 000.Chemotherapy was associated with the lowest excess financial consequence to the health system of approximately $3500.The low excess financial consequence of chemotherapy is partially because patients not receiving chemotherapy were still receiving other intensive services in the last month of life, contributing to their total costs.
Many factors the patient-clinician decision to pursue medically intensive services.
These may include patient desire for such care, clinician belief that it will provide significant medical or palliative benefit, patient denial of limited prognosis, or miscommunication about prognosis.
0][51] Evidence also indicates that patients and clinicians may not always share the same understanding about prognosis; in 1 large sample study, more than two-thirds of patients with metastatic solid tumor were unaware that their chemotherapy had no curative intent. 52Thus, miscommunication or misunderstanding of prognosis is of particular concern.
Ideally, patient-clinician decisions to pursue medically intensive services should involve discussions of the likelihood of benefit, risks, and side effects, including potential financial consequences, of these interventions.The present study provides reference cost estimates that may help inform those discussions.The present analysis indicates that patients experience approximately $1250 out-of-pocket health costs in the last month of life due to medically intensive services.To place this number in context, the median annual household income of a Medicare beneficiary in 2014, the last year of this analysis, was $24 150, 53 or $2013 a month.Using these figures, expected beneficiary spending on medical services that have a low likelihood of helping them and could harm them may represent 62% of the household income of the typical Medicare enrollee in the last month of his life.
Indeed, analyses of Medicare-only beneficiaries find that beneficiaries with a new cancer diagnosis have out-of-pocket costs that are a mean of 24% of their household income. 54The present study indicates that as cancer progresses, expected beneficiary responsibility for intensive medical services represents a higher proportion of household income, rising to almost two-thirds of household income in the last month of life.This study also identified costs in the last month of life only, and almost 90% of our cohort received no chemotherapy in this time frame.Thus, analyses looking farther back from death would yield much higher estimates of drug costs.This, coupled with the high cost of chemotherapies and immunotherapies that have come to market after fiscal year 2014 (the last year of our data), 55,56 indicate that the financial consequences due to cancer care will only grow.
The purchasing of medical care, including chemotherapy and end-of-life care, is largely guided by physicians; recent work finds that physician beliefs, rather than patient beliefs, explain much of oft-noted geographic variation in end-of-life spending. 57Thus, physicians have a strong opportunity to limit medically intensive interventions at the end of life in ways that can avoid patient financial consequences.For example, early goals-of-care conversations with patients may help to ensure that patients receive the care they want and need.The present study also highlighted that given the low likelihood of benefit and the potential for financial consequences, as well as patient concerns about the cost of cancer care, 21,22 it may be worthwhile for physicians to discuss this openly with patients before proceeding with medically intensive interventions.

JAMA Network Open | Geriatrics
Health System and Beneficiary Costs for Intensive End-of-Life Medical Services Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

Disclaimer:
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US government.

Figure 1 .
Figure 1.Health System Costs in the Last Month of Life due to Medically Intensive Services 40 000 35 000 30 000 25 000 20 000

Figure 2 .
Figure 2. Beneficiary Costs in the Last Month of Life Due to Medically Intensive Services All analyses were conducted in Stata/MP, version 15.1 (StataCorp), from February to August 2019.

Table 1 .
Demographic Characteristics and Comorbidities

Table 2 .
Receipt of Medically Intensive Services in the Last 30 Days of Life

Table 3 .
Estimated Costs of Medically Intensive Care for the Health System and the Beneficiary in the Last Month of Life