Differences in Medicare Utilization and Expenditures in the Last Six Months of Life among Patients with and without Alzheimer's Disease and Related Disorders

Abstract

Background:

Previous research has not focused on differences at the end of life among Medicare beneficiaries with, and without, a diagnosis of Alzheimer's disease and related disorders (ADRDs).

Objective:

The purpose of this study was to examine differences in utilization of inpatient services and Medicare expenditures (overall and by category) in the last six months of life for patients with, versus those without, a diagnosis of ADRD.

Design:

The study used 2013 Medicare Research Identifiable Files (5% sample) to study utilization and expenditures for a full six months before death (n = 7895 for ADRD; n = 30,639 for non-ADRD). A generalized linear model with a gamma distribution was used to examine the association of ADRD with end-of-life service utilization and expenditures.

Results:

ADRD patients were overall less expensive than their non-ADRD peers through reduced use of high-cost services, and urban patients were more likely than rural patients to use hospice and other services among both the ADRD and non-ADRD groups. After controlling for age, gender, race, dual eligibility, residence, region, chronic conditions, and type of service utilization, ADRD beneficiaries cost Medicare 11% less than non-ADRD beneficiaries (p < 0.01).

Conclusions:

Future research should examine the informal caregiving costs of caring, which is a significant part of care for an ADRD patient, as the residential setting of the beneficiary highly influences costs.

Introduction

Alzheimer's disease and related disorders (ADRDs) are a category of degenerative brain diseases that often lead to dementia.¹ In 2013, the ADRD death rate was 26.8 per 100,000, making it the sixth leading cause of death in the United States.² This is expected to continue as the baby boomer generation ages.³ Because an individual with ADRD can live for many years with increasing disability and declining health,^4–6 the reported cause of death may be an illness stemming from complications of ADRDs such as secondary infections, sepsis, and frailty.^7–10

Health care utilization for ADRD patients is costly, with patients often having multiple disease conditions such as a stroke or depression accompanying ADRD.¹¹ The health care costs of dementia increase significantly throughout the duration of the illness due to the cost of caregiving in a formal setting such as a nursing home.¹² The health care costs of ADRD have been estimated to be similar to those of heart disease and higher than expenditures for cancer.¹³ Estimates include $33,329 annually per case, adjusted for other comorbid conditions.¹⁴

For individuals aged ≥65, Medicare covers much of the costs for inpatient, outpatient, home health, hospice, skilled nursing facility services (skilled medical care), and prescription drugs. Medicare does not cover long-term custodial care (nonmedical care) unless it is part of an acute care episode, leaving much of ADRD costs to be paid by the individual. Medicaid becomes the primary payer for this custodial care, but only after the individual's assets have been reduced to approximately $2,000, depending on the state where the beneficiary resides.^15,16 Medicare expenditures contribute to spending variation at the end of life among ADRD patients.¹⁷

Little research has been conducted on the end of life among Medicare beneficiaries with ADRD. ADRD has been associated with increased hospital utilization with health care utilization differences among those with ADRD and those without ADRD narrowing at the end of life.^18,19 Previous research has focused on the last year of life, with no studies examining the differences in health care utilization and expenditures for those with ADRD and those without ADRD during the last six months of life.¹⁹ The last six months of life is a common metric for examining intensity of end-of-life services.^20,21 This is also the most costly time period for Medicare beneficiaries, with >25% of Medicare expenditures incurred during this time period.^22,23

Therefore, an examination of differences in health care utilization and expenditures for those with ADRD and those without ADRD during the last six months of life is needed. The purpose of this study was to examine differences in utilization of services and Medicare expenditures (overall and by category) in the last six months of life for patients with, versus those without, ADRD. Given previous research focusing on the last year of life, we expect that expenditures among Medicare beneficiaries with ADRD versus those without ADRD may differ during the last six months of life.

Methods

Population

The study population was drawn from the 5% sample of 2013 Medicare Research Identifiable Files (n = 2,972,192). The data files used included the beneficiary master summary (enrollment information, chronic conditions), carrier claims (physician), MedPar (inpatient and skilled nursing facility), home health, hospice, and outpatient files. Analysis was restricted to beneficiaries enrolled in Medicare Part A and Part B throughout the entire year of 2013, before death (n = 120,068). To study utilization for a full six months before death, beneficiaries who died before July 1, 2013 were excluded (sample n = 57,111). Only fee-for-service beneficiaries (sample n = 41,860) with complete demographic information were included (sample n = 40,574). The sample was further limited to beneficiaries who had utilized at least one type of service (i.e., inpatient, outpatient, hospice, home health, skilled nursing facility, ambulance, or physician) during the year and thus had nonzero expenditures (n = 38,534). ADRD patients were defined as individuals with at least one inpatient, skilled nursing facility, home health, outpatient, hospice, or carrier claim with an ICD-9/CPT4/HCPCS code of DX 331.0–331.7, 290.0–290.43, 294.0–294.8, or 797.xx during the calendar year 2013 (n = 7895 for ADRD; n = 30,639 for non-ADRD). The study was approved by the University of South Carolina Institutional Review Board.

Variable definitions

Total end-of-life cost was the sum of the total Medicare expenditures for inpatient, outpatient, physician, hospice, home health, and skilled nursing facility use during the last six months of life for ADRD patients. Expenditures for prescription drug claims (i.e., Part D data) were not available and thus not included.

Within the 2013 Medicare Research Identifiable Files, we are not able to determine the geographical rates paid to providers in rural and urban areas. However, we are able to identify whether the beneficiary resided in a rural area, based on the county of residence. We will compare expenditures, ADRD versus without ADRD, within rural and within urban counties, as providers are paid different rates based on geography.²⁴ County of residence was used to classify beneficiaries as rural or urban, using Urban Influence Codes (UICs). Metropolitan counties (UICs 1,2) were categorized as urban, while rural counties included Micropolitan (UICs 3,5,8), small adjacent (UICs 4,6,7), and remote rural (UICs 9,10, 11, 12) counties.

Using the Andersen behavioral model of health services use,²⁵ predictor variables of predisposing characteristics (age, sex, race/ethnicity, and dual eligibility), enabling resources (region), and need (count of chronic conditions) were chosen to adjust for differences in health care expenditures during the last six months of life. We adjusted for age (<65, 65–74, 75–84, 85–94, and ≥95), sex, race/ethnicity (non-Hispanic African American, non-Hispanic white, other, and Hispanic), region (Northeast, Midwest, South, and West) and dual eligibility. Dual eligibility was used as a proxy for low socioeconomic status.²⁶ A beneficiary was designated as dually eligible if he or she had been eligible for Medicaid enrollment for any time between 1 and 12 months. The count of chronic conditions, collected from the Master Beneficiary Summary File, was used as a proxy for evaluated need. These conditions were identified by using previously validated claims-based algorithms.²⁷ Chronic conditions were collapsed into the following: asthma, heart disease, heart failure, chronic kidney disease, cancer, chronic obstructive pulmonary disease, hyperlipidemia, hypertension, osteoporosis, and stroke. Conditions were then categorized by counts: no conditions except ADRD, one–two, three–four, or five or more conditions. Finally, to control for community enabling resources that might increase or decrease overall utilization, we included access to hospice, measured by the ratio of hospice organizations to population, with no hospice in the county set as the lowest level, inpatient beds (in quartiles) and home health availability, measured by the ratio of agencies to population. None of these three variables were significant in multivariable analysis, thus all were dropped from the final analysis. All county-level data were obtained from the Area Health Resource File.

Analytic approach

Bivariate analysis was used to examine differences in key variables of interest by rurality using chi-square tests, α = 0.05. Wilcoxon rank-sum tests were used to detect statistically significant differences in total median and mean expenditures, as well as by service category. Medicare expenditures were log-transformed to normalize and account for the skewed data. A generalized linear model with a gamma distribution was used to examine the diagnosis of ADRD on end-of-life Medicare expenditures. Coefficients were transformed into cost ratios for ease of interpretation. Regression results on log-transformed end-of-life Medicare expenditures demonstrate the difference in log-transformed expenditures between the indicated group and the referent group. Cost ratios, exponentiated coefficient estimates, were calculated to examine the ratio between the expenditures for the indicated group versus the referent group. These will be presented as percentage differences. Types of service utilization were included in the regression model to control for the intensity of services provided to patients at the end of life. All data management and analyses were conducted using SAS Version 9.3 (SAS Institute, Inc.).

Results

The majority of the ADRD decedents were urban (80.3%), female (69.2%), aged ≥85 (61.9%), and non-Hispanic white (85.2%; see Table 1). Less than half of the ADRD decedents were dual-eligible beneficiaries (42.3%). Over 40% (42.5%) of the ADRD decedents resided in the southern region of the United States. A large proportion of the ADRD sample had three or more chronic conditions (83.7%), not including ADRD. On average, the total cost for Medicare expenditures in the last six months of life for Medicare beneficiaries with ADRD was $22,387 (±$29,786; median = $12,544).

Table 1.

Total Medicare Expenditures, Demographics, and Comorbidity for Beneficiaries in Their Last Six Months of Life, by Alzheimer's Disease and Related Disorders Status, 2013 (N = 38,534)

	ADRD		Non-ADRD
Variable	N = 7895	%	N = 30,639	%	p Value
Gender					<0.0001
Female	5448	69.0	17,199	56.1
Male	2447	31.0	13,440	43.9
Age (years)					<0.0001
<65	79	1.0	3400	11.1
65–74	536	6.8	6811	22.2
75–84	2406	30.5	9321	30.4
85–94	4108	52.0	9352	30.5
≥95	766	9.7	1755	5.7
Dual eligibility					<0.0001
Yes	3324	42.1	9137	29.8
No	4571	57.9	21,502	70.2
Race					<0.0001
Non-Hispanic African American	657	8.3	2870	9.4
Hispanic	337	4.3	1298	4.2
Non-Hispanic other	171	2.2	884	2.9
Non-Hispanic white	6730	85.2	25,587	83.5
Region					<0.0001
Northeast	1429	18.1	5818	19.0
Midwest	2013	25.5	7696	25.1
South	3361	42.6	11,954	39.0
West	1092	13.8	5171	16.9
Rurality					<0.0001
Rural	1556	19.7	7127	23.3
Urban	6339	80.3	23,512	76.7
Chronic conditions category					<0.0001
0	240	3.0	2053	6.7
1–2	1013	12.8	3401	11.1
3–4	1777	22.5	6145	20.1
≥5	4865	61.6	19,040	62.1

ADRD, Alzheimer's disease and related disorders.

Demographic differences between ADRD and non-ADRD decedents were significant across all categories. Non-ADRD decedents were younger (p < 0.001) and less likely to be dually eligible than ADRD decedents (70.2% vs. 57.9%, p < 0.0001). As well, non-ADRD decedents were more likely to have zero chronic conditions, and less likely to have between one and two and three and four chronic conditions, compared with non-ADRD decedents (p < 0.0001). On average, the cost of medical care in the last six months of life for Medicare beneficiaries without ADRD was $31,753 (±$40,989; median = $18,680).

Within both rural and urban decedents, ADRD patients incurred lower costs at the end of life than did their non-ADRD peers, largely through a pattern of reduced use of high-cost services (Table 2). We will compare expenditures, ADRD versus without ADRD, within rural and within urban, as providers are paid different rates based on geography.²⁴ Thus, within rural residents, ADRD beneficiaries had lower total median expenditures than rural beneficiaries without ADRD ($10,550 vs. $16,570, p < 0.001), as well as lower mean expenditures ($18,187 vs. $26,973, p < 0.0001). Similarly, urban ADRD beneficiaries were less costly in the last six months of life than their urban non-ADRD counterparts, with lower median ($13,039 vs. $19,354, p < 0.001) and mean expenditures ($23,417 vs. $33,201, p < 0.0001).

Table 2.

Service Use and Mean/Median Medicare Expenditures, by Service Category and Alzheimer's Disease and Related Disorders Status, among Beneficiaries with any Expenditures in Their Last Six Months of Life, 2013 (N = 38,534)

	Rural ADRD	Rural non-ADRD	Within rural, ADRD vs. non-ADRD p values	Urban ADRD	Urban non-ADRD	Within urban, ADRD vs. non-ADRD p values
Total Medicare expenditures among those with any expenditure
Median	$10,550	$16,570	<0.0001	$13,039	$19,354	<0.0001
Mean	$18,187	$26,973	<0.0001	$23,417	$33,201	<0.0001
Inpatient
% Service use	60.6	66.9	<0.0001	61.0	68.3	<0.0001
Median	$12,459	$16,701	<0.0001	$14,668	$19,729	<0.0001
Mean	$18,402	$26,582	<0.0001	$24,373	$33,663	<0.0001
Number of stays
Mean	1.8	2.0	<0.0001	1.8	2.1	<0.0001
Median	1.0	2.0	0.0002	1.0	2.0	<0.0001
Outpatient
% Service use	86.1	88.1	<0.0001	74.3	78.3	<0.0001
Median	$77	$283	<0.0001	$108	$436	<0.0001
Mean	$297	$724	<0.0001	$431	$1,091	<0.0001
Number of visits
Mean	5.7	7.4	<0.0001	4.5	6.2	<0.0001
Median	4.0	5.0	<0.0001	3.0	4.0	<0.0001
Physicians
% Service use	56.1	76.6	<0.0001	57.3	78.6	<0.0001
Median	$222	$385	<0.0001	$291	$502	<0.0001
Mean	$490	$869	<0.0001	$647	$1,185	<0.0001
Number of visits
Mean	6.0	8.5	<0.0001	7.2	10.4	<0.0001
Median	3.0	5.0	<0.0001	4.0	6.0	<0.0001
Hospice
% Service use	53.0	44.0	<0.0001	63.5	48.4	<0.0001
Median	$1,838	$1,557	0.001	$2,358	$2,089	<0.0001
Mean	$2,155	$2,057	0.013	$2,935	$2,719	<0.0001
Number of episodes
Mean	1.1	1.0	0.02	1.1	1.0	0.05
Median	1.0	1.0	0.02	1.0	1.0	0.05
Home health
% Service use	21.7	27.8	<0.0001	29.5	33.7	<0.0001
Median	$2,178	$2,185	0.028	$2,906	$2,675	<0.0001
Mean	$2,450	$2,416	0.025	$3,325	$3,032	<0.0001
Number of episodes
Mean	1.2	1.2	0.5	1.3	1.3	0.05
Median	1.0	1.0	0.50	1.0	1.0	0.04
SNF
% Service use	27.5	24.1	<0.0001	25.8	23.8	<0.0001
Median	$10,116	$9,043	0.34	$12,232	$11,179	0.002
Mean	$12,576	$11,886	0.23	$14,615	$13,787	0.0009
Number of stays
Mean	1.4	1.5	0.07	1.5	1.5	0.10
Median	1.0	1.0	0.09	1.0	1.0	0.12

SNF, skilled nursing facility.

Utilization rates differed between ADRD and non-ADRD decedents. ADRD beneficiaries were less likely to have utilized inpatient, outpatient clinic, physician, or home health services than their peers without the disorder. On average, rural ADRD beneficiaries were less likely to have utilized inpatient hospitalizations than rural non-ADRD beneficiaries (60.6% vs. 66.9% utilization; $12,459 vs. $16,701 median expenditures, p < 0.0001). There were stark differences between ADRD and non-ADRD decedents in the utilization of physician services during the last six months of life. Over 40% of ADRD decedents did not have any physician expenditures versus just >20% of non-ADRD (p < 0.0001). ADRD beneficiaries, compared with non-ADRD beneficiaries, were more likely to use hospice and skilled nursing facility services. Expenditure differences in these two services between ADRD and non-ADRD decedents were modest, however, in comparison with differences associated with hospital use. For example, rural decedents with ADRD incurred a median of $1838 in hospice expenditures during the last six months of life, compared with $1557 for non-ADRD decedents (p < 0.001; Table 2).

After controlling for age, gender, race, dual eligibility, residence, region, chronic conditions, and type of service utilization, ADRD beneficiaries cost Medicare 11% less than non-ADRD beneficiaries (p < 0.01; Table 3). Individuals with five or more chronic conditions were 33% more costly during the last six months of life than individuals who had no conditions besides ADRD (p < 0.001). Individuals utilizing inpatient services were 557% more costly during the last six months of life than individuals who had not utilized inpatient services (p < 0.001). Individuals who utilized skilled nursing facilities (SNF) were 90% more costly during the last six months of life than individuals who did not (p < 0.01). Individuals who utilized hospice were 17% more costly during the last six months of life than individuals who did not. Individuals who utilized outpatient, physician, and home health services were also more costly among individuals who utilized these services than individuals who had not utilized these services.

Table 3.

Regression Results on Log-Transformed End-of-Life Medicare Expenditures for Beneficiaries with and without Alzheimer's Disease and Related Disorders in the Last Six Months of Life (N = 38,534)

	Estimate	Cost ratio ^a
Intercept	8.35^*
ADRD	−0.11^*	0.89
Chronic conditions
No of conditions	Referent
1–2	−0.05	0.96
3–4	0.07^*	1.07
≥5	0.29^*	1.33
Inpatient	1.72^*	5.6
Outpatient	0.25^*	1.28
Physician	0.13^*	1.14
SNF	0.64^*	1.90
Home health	0.32^*	1.38
Hospice	0.16^*	1.17

Also adjusting for age, gender, race, dual eligibility, residence, and region.

The cost ratio is calculated as the exponent of the estimated coefficient. This is the dollar amount between the particular group and the referent.

Significant at p = 0.01.

Discussion

This article offers the first examination of differences in service utilization and Medicare expenditures (overall and by category) in the last six months of life for patients with, versus those without, ADRD. The findings from this study demonstrate that Medicare expenditures for ADRD patients in the last six months of life are, on the whole, less costly than non-ADRD patients, with differences during the last six months of life existing after adjusting for age, gender, race, dual eligibility, region, chronic conditions, and type of service utilization. Our findings are important as we examine numerous types of Medicare expenditures among ADRD and non-ADRD patients at the end of life; going beyond just inpatient expenditures, including outpatient, physician, hospice, home health, and skilled nursing facility expenditures.

Our findings may be important for policymakers in light of programs such as the National Alzheimer's Project Act, an act with the goal of tracking the costs associated with this disease for Medicaid and Medicare, and providing quality and efficient care for patients with ADRD.²⁸ Underdiagnosis of ADRD is a known issue in this field, and there is some evidence that undiagnosed ADRD patients have much higher utilization and cost.²⁸ Our findings of higher diagnosis, utilization, and expenditures among urban ADRD patients may signify that expensive undiagnosed patients are “moving into the system” at higher rates in urban areas than in rural areas, thus leading to the observed higher utilization and expenditures.

Lower expenditures among rural ADRD beneficiaries during the last six months of life may be an indicator for disparities in rural–urban access to end-of-life care, which can help policymakers target programs to reach these people. We found that rural beneficiaries were less likely to utilize physician, hospice, and home health services; this could possibly be due to greater availability of home caregiving and support in rural areas. These findings were not entirely unsurprising due to lower rates of providers and hospice services available in rural areas.²⁹

Our study was restricted to use of fee-for-service Medicare claims data, and may not be generalizable to privately insured or Medicare advantage populations. We examined expenditures and utilization, keeping in mind that Medicare pays rural and urban providers different service rates.²⁴ Further limitations include use of the Center for Medicare and Medicaid Services' year-end flag, which does not indicate illness stage. This definition includes ADRD without information on stage of illness. We were not able to include out-of-pocket costs, costs for prescription drugs, durable medical equipment, or any costs covered by Medicaid for those patients who were dually eligible. Additional costs not included were adult day programs, in-home care, respite, custodial care in skilled facilities, and assisted living. Costs and stays were restricted to limitations of billing data. For example, hospice benefits are truncated to reflect billing cycles of 30 days in the claims data. Furthermore, this study focused on only the last six months of life; patients with ADRD likely could have been incurring costs for years.

Our findings of lower physician use among ADRD patients in the last six months of life may be a substitution effect for hospice. This is an area for future research. Further research should also examine the informal caregiving costs of caring, which is a significant part of the care for an ADRD patient.¹⁹ The burden of informal caregiving on the physical, emotional, and financial health of the caregiver may be extensive.^30,31

Footnotes

Acknowledgment

This work was funded by a grant from the Federal Office of Rural Health Policy.

Author Disclosure Statement

No competing financial interests exist.

References

American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 4th ed, Text Revision (DSMIV-TR). Arlington, VA: American Psychiatric Association, 2000.

Kochanek

, Murphy

, Xu

, Tejada-Vera

: Deaths: Final data for 2014. Natl Vital Stat Rep, 2016; 65:1–122.

Ortman

, Velkoff

, Hogan

: An aging nation: The older population in the United States. Population estimates and projections [Internet]. Suitland MD: US Department of Commerce, Economics and Statistics Administration US Census Bureau, 2014. https://census.gov/content/dam/Census/library/publications/2014/demo/p25–p1140.pdf (last accessed June 20, 2018)

Burns

, Jacoby

, Luthert

, Levy

: Cause of death in Alzheimer's disease. Age Ageing, 1990; 19:341–344.

Brunnstrom

, Englund

: Cause of death in patients with dementia disorders. Eur J Neurol, 2009; 16:488–492.

Lin

, Fillit

, Cohen

, Neumann

: Potentially avoidable hospitalizations among Medicare beneficiaries with Alzheimer's disease and related disorders. Alzheimers Dement, 2013; 9:30–38.

Ives

, Samuel

, Psaty

, Kuller

: Agreement between nosologist and cardiovascular health study review of deaths: Implications of coding differences. J Am Geriatr Soc, 2009; 57:133–139.

Forma

, Rissanen

, Aaltonen

, et al.: Dementia as a determinant of social and health service use in the last two years of life 1996–2003. BMC Geriatr, 2011; 11:14.

Romero

, Benito-Leon

, Mitchell

, et al.: Under reporting of dementia deaths on death certificates using data from a population-based study (NEDICES). J Alzheimers Dis, 2014; 39:741–748.

10.

Romero

, Benito-Leon

, Louis

, Bermejo-Pareja

: Under reporting of dementia deaths on death certificates: A systematic review of population-based cohort studies. J Alzheimers Dis, 2014; 41:213–221.

11.

Barnes

, Yaffe

: The projected effect of risk factor reduction on Alzheimer's disease prevalence. Lancet Neurol, 2011; 10:819–828.

12.

Zhu

, Scarmeas

, Torgan

, et al.: Longitudinal study of effects of patient characteristics on direct costs in Alzheimer disease. Neurology, 2006; 67:998–1005.

13.

Agency for Healthcare Research and Quality. Total expenses and percent distribution for selected conditions by type of service: United States, 2008. Medical Expenditure Panel Survey Household Component data [Internet]. Rockville MD: Agency for Healthcare Research and Quality, 2017. https://meps.ahrq.gov/data_stats/tables_compendia_hh_interactive.jsp?_SERVICE=MEPSSocket0&_PROGRAM=MEPSPGM.TC.SAS&File=HCFY2008&Table=HCFY2008_CNDXP_C&_Debug= (last accessed June 20, 2018)

14.

Hurd

, Martorell

, Delavande

, et al.: Monetary costs of dementia in the United States. N Engl J Med, 2013; 368:1326–1334.

15.

Bynum

, Rabins

, Weller

, et al.: The relationship between a dementia diagnosis, chronic illness, medicare expenditures, and hospital use. J Am Geriatr Soc, 2004; 52:187–194.

16.

Medicare.gov: How can I pay for nursing home care? [Internet]. Baltimore MD: Centers for Medicare and Medicaid Services, 2017. https://medicare.gov/what-medicare-covers/part-a/paying-for-nursing-home-care.html (last accessed June 20, 2018)

17.

Miller

, Intrator

, Gozalo

, et al.: Government expenditures at the end of life for short- and long-stay nursing home residents: Differences by hospice enrollment status. J Am Geriatr Soc, 2004; 52:1284–1292.

18.

Tolppanen

, Taipale

, Purmonen

, et al.: Hospital admissions, outpatient visits and healthcare costs of community-dwellers with Alzheimer's disease. Alzheimers Dement, 2015; 11:955–963.

19.

Feng

, Coots

, Kaganova

, Wiener

: Hospital and ED use among Medicare beneficiaries with dementia varies by setting and proximity to death. Health Aff (Millwood), 2014; 33:683–690.

20.

Skinner

, Wennberg

: How Much Is Enough? Efficiency and Medicare Spending in the Last Six Months of Life. National Bureau of Economic Research, Cambridge, MA, 1998, No. w6513.

21.

Wennberg

, Fisher

, Stukel

, et al.: Use of hospitals, physician visits, and hospice care during last six months of life among cohorts loyal to highly respected hospitals in the United States. BMJ, 2004; 328:607.

22.

Hogan

, Lunney

, Gabel

, Lynn

: Medicare beneficiaries' costs of care in the last year of life. Health Aff, 2001; 20:188–195.

23.

Riley

, Lubitz

: Long-term trends in medicare payments in the last year of life. Health Serv Res, 2010; 45:565–576.

24.

MedPAC: Payment basics [Internet]. Washington

: MedPAC, 2014. http://medpac.gov/-documents-/payment-basics (last accessed June 25, 2018)

25.

Andersen

: A Behavioral Model of Families' Use of Health Services. Chicago, IL: Center for Health Administration Studies, 1968.

26.

Komisar

, Feder

, Kasper

: Unmet long-term care needs: An analysis of Medicare-Medicaid dual eligibles. Inquiry, 2005; 42:171–182.

27.

Chronic Condition Data Warehouse: CCW Medicare administrative data user guide [Internet]. Chronic Condition Data Warehouse, 2017. https://ccwdata.org/web/guest/user-documentation (last accessed June 25, 2018)

28.

Alzheimer's Association: Draft national plan to address Alzheimer's Disease [Internet]. Chicago IL: Alzheimer's Association, 2012. http://alz.org/documents_custom/National_Plan_to_Address_Alzheimers.pdf (last accessed June 25, 2018)

29.

Chan

, Hart

, Goodman

: Geographic access to health care for rural Medicare beneficiaries. J Rural Health, 2006; 22:140–146.

30.

Langa

, Chernew

, Kabeto

, et al.: National estimates of the quantity and cost of informal caregiving for the elderly with dementia. J Gen Intern Med, 2001; 16:770–778.

31.

Crouch

, Probst

, Bennett

: Rural-urban differences in unpaid caregivers of adults. Rural Remote Health, 2017; 17:4351.