Veterans’ use of inpatient and outpatient palliative care: The national landscape

Abstract

Background:

Palliative care improves the quality of life for people with life-limiting conditions, which are common among older adults. Despite the Veterans Health Administration (VA) outpatient palliative care expansion, most research has focused on inpatient palliative care. This study aimed to compare veteran characteristics and hospice use for palliative care users across care settings (inpatient vs. outpatient) and dose (number of palliative care encounters).

Methods:

This national cohort included veterans with any VA palliative care encounters from 2014 through 2017. We used VA and Medicare administrative data (2010–2017) to describe veteran demographics, socioeconomic status, life-limiting conditions, frailty, and palliative care utilization. Specialty palliative care encounters were identified using clinic stop codes (353, 351) and current procedural terminology codes (99241–99245).

Results:

Of 120,249 unique veterans with specialty palliative care over 4 years, 67.8% had palliative care only in the inpatient setting (n = 81,523) and 32.2% had at least one palliative care encounter in the outpatient setting (n = 38,726), with or without an inpatient palliative care encounter. Outpatient versus inpatient palliative care users were more likely to have cancer and less likely to have high frailty, but sociodemographic factors including rurality and housing instability were similar. Duration of hospice use was similar between inpatient (median = 37 days; IQR = 11, 112) and outpatient (median = 44 days; IQR = 14, 118) palliative care users, and shorter among those with only one palliative care encounter (median = 18 days; IQR = 5, 64).

Conclusions:

This national evaluation provides novel insights into the care setting and dose of VA specialty palliative care for veterans. Among veterans with palliative care use, one-third received at least some palliative care in the outpatient care setting. Differences between veterans with inpatient and outpatient use motivate the need for further research to understand how care settings and number of palliative care encounters impact outcomes for veterans and older adults.

INTRODUCTION

Palliative care improves the quality of life for people with life-limiting conditions, which are common among older adults.^1–4 To meet the needs of veterans with life-limiting conditions, palliative and hospice care has been one of the fastest-growing health services in the Veterans Health Administration (VA) over the past decade.^5,6 Interdisciplinary palliative care teams were required in all VA Medical Centers starting in 2003, and the VA began expanding outpatient palliative care in 2013.^7–9 However, little is known about veterans’ use of palliative care across care settings, including veteran factors associated with care setting and number of encounters, or dose, of palliative care.

Palliative care can help support older adults with life-limiting conditions, including veterans, by addressing the physical, mental, spiritual, and social distress that often accompany illness.^10–14 Palliative care improves the quality of life and quality of end-of-life care; however, most research has been conducted in the inpatient setting.^3–7 Providers in and out of the VA perceive outpatient palliative care as an opportunity to improve the value of care given its patient-centeredness and potential to decrease healthcare use and costs.^9,15,16 Despite increasing evidence of the benefits of palliative care in addition to hospice, the effects of palliative care interventions vary as a result of heterogeneity in the intervention itself, the care setting, dose or number of encounters, and the clinical and psychosocial needs of the patient population.^17,18

Few studies have examined differences in patient characteristics and dose of palliative care by care setting. Among patients with metastatic non-small-cell lung cancer, one study found the use of outpatient versus inpatient-only specialty palliative care occurred earlier relative to death, but did not compare patient characteristics between care settings.¹⁹ Most palliative care research focuses on specialty palliative care, referring to care provided by a clinician with specialized training in palliative care or a member of their interdisciplinary care team; however, primary palliative care provided by nonspecialists may also be appropriate depending on the patient’s needs.^20–23 To support population health outcomes research across care settings, foundational knowledge about palliative care use by care setting is needed, including differences in patient characteristics and palliative care timing in the care trajectory.

The VA provides a use case for comparing inpatient and outpatient specialty palliative care use due to the national standards that apply to both settings.²⁴ According to VA standards, specialty palliative care consists of an interdisciplinary care team performing assessments and making a plan to address multiple aspects of care: pain and symptom management; goals of care and associated treatment decisions; advance care planning, psychosocial, spiritual, and other issues; family meetings; and referrals to hospice and other VA and community services. The increasing prevalence of outpatient palliative care use in the VA over the past 10 years mirrors the expansion of community-based palliative care nationally, and learnings from the VA may be useful in understanding care needs for older adults with life-limiting illness more broadly.^25,26

To inform the implementation and dissemination of outpatient palliative care, administrators and researchers need additional evidence comparing patient characteristics among people with different types of palliative care encounters. The primary aim of this study was to describe veteran characteristics for palliative care users by care setting (inpatient-only or outpatient) and dose (one encounter or more than one) among a national cohort of veterans. In addition, we describe the duration of palliative care and hospice use (Medicare and VA) by subgroups defined by setting and dose. Describing the landscape of palliative care use in the VA across care settings will help to inform policies to expand access to appropriate palliative care use in the setting preferred by older adults with serious illness and their families.

METHODS

Palliative care use for veterans was measured using linked national VA and Medicare administrative data sources. Medicare administrative data includes claims data for hospice benefits, and is essential for characterizing end-of-life care; however, we did not measure non-hospice palliative care in Medicare claims due to low sensitivity.^27,28 In comparison, methods for using VA administrative data to identify specialty palliative care use have been validated across care settings.^29,30

The sample (N = 120,249) includes veterans with at least one VA specialty palliative care encounter in fiscal year 2014 (beginning October 1, 2013) through 2017 (ending September 30, 2017). Specialty palliative care use was defined using VA Corporate Data Warehouse administrative data using previously validated algorithms, which have higher sensitivity compared to the documentation available in Medicare claims data.^28–30 The specialty palliative care definition aligns with VA guidance and uses combinations of stop codes indicating a consult provided by a palliative care team (351, 353) and outpatient current procedural terminology (CPT) codes (99241–99245 or 99201–99205) or inpatient CPT codes (99221–99223 or 99251–99255). Our definition of VA specialty palliative care required both stop codes and CPT codes, which were found to have a sensitivity of 0.89 and nearly perfect specificity.²⁹ Palliative care encounters that were not part of a hospice assessment were identified using stop code 353 and hospice palliative care encounters were identified as stop code 351.²⁹ Encounters conducted by telehealth were identified using the VA service category code for telehealth. We additionally identified “other” palliative care encounters using ICD-9 or 10 codes that indicate palliative care (ICD-9 code V66.7 or ICD-10 code Z51.5 in the primary or secondary ICD position) as would be used in Medicare billing data, even if these encounters did not meet the VA criteria for specialty palliative care.^27,30 Palliative care ICD codes demonstrate the prevalence of VA palliative services being provided outside of specialty clinics, and we were unable to determine whether the other palliative services provided were specialty care or not.

Outcomes

For each veteran with any specialty palliative care encounters from 2014 to 2017, we defined the number of specialty palliative care encounters overall and by care setting, inpatient or outpatient clinic, during the same period. We defined the first outcome as a binary indicator for one palliative care encounter compared to more than one encounter to understand factors associated with the level of exposure, as well as the potential clinical impact that multiple visits may have on the relationship and trust between the veteran and provider compared to only one visit. We defined the second outcome as a binary indicator for any specialty palliative care use in the outpatient clinic, with or without an inpatient palliative care encounter, compared to only in the inpatient setting. This comparison was important because the core motivation of the study was to describe the cohort of veterans receiving outpatient specialty palliative care and understand how the cohort might differ from veterans receiving inpatient specialty palliative care.

Covariates

Life-limiting conditions were defined using an algorithm validated among veterans using VA and Medicare claims data from 2010 to 2017 (Supplementary Table S1).³¹ Using VA and Medicare data, we identified veterans who met the criteria for each before their first palliative care encounter during the study period: cancer, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), dementia with complications, end stage renal disease (ESRD), and HIV/AIDS with complications. We defined binary indicators because veterans may have multiple conditions and categories are not mutually exclusive.

Veteran and health system level factors that are expected to be associated with palliative care patterns were identified using VA administrative data (demographics and health status) and linked Medicare claims data (health status). Race and ethnicity variables were combined to create categories non-Hispanic White, non-Hispanic Black, Hispanic ethnicity of any race, and other races and ethnicities or missing values. VA benefit priority status was used to identify high disability (priority==1,4) and low income (priority==5). Eligibility for VA benefits varies depending on the priority group assignment, and values 1 and 4 indicate assignment due to disability. The risk of unstable housing was assessed using a previously validated algorithm using VA data.³² Other veteran characteristics included rural status (rural-urban commuting code >4), most recent marital status (currently married vs. any other marital status), and age (defined at first palliative care encounter). The Area Deprivation Index from 2015 was linked with veteran zip codes.³³ High area deprivation was defined as being in the most deprived deciles relative to their state (top 30%). Health system site characteristics included high complexity and rural location. High complexity was defined using VA medical center complexity levels, which are based on levels of patient volume, patient risk, teaching and research, and intensive care unit levels.

Complex health needs included measures of frailty, depression, and the VA Care Assessment Needs score. Frailty was measured using the JEN Frailty Index as of the first palliative care encounter year (categories low frailty <4, medium frailty 4–7, high frailty 8+).³⁴ Depression was measured using indicators in the Medicare Beneficiary Summary File chronic conditions warehouse. Care Assessment Need scores rely on health services utilization data predicting either hospitalization or death within 1 year, and the VA data generates scores frequently as needs change over time.³⁵ We measured the Care Assessment Need score within 90 days of palliative care encounter, and 24,485 (13.0%) of included veterans were missing the Care Assessment Need score during that period, potentially due to low or no use of VA health services in the year before the palliative care encounter.

Palliative care characteristics and timing

The first, last, and count of palliative care encounters for each unique veteran was identified as the first and last encounter that met criteria for specialty palliative care or other (ICD codes only) palliative care from 2014 to 2017, including inpatient, outpatient, and telehealth settings. Time since life-limiting condition is reported among palliative care users meeting criteria for any life-limiting condition using Medicare claims and VA data.

Utilization measures were intended to measure health status by describing intensity of care in the period directly before and after the initial specialty palliative care encounter and were limited to 90 days around the index date. Healthcare utilization was measured using VA and Medicare claims data. We present counts for emergency department visits, inpatient days, intensive care unit (ICU) days, VA community living facility days, and Medicare skilled nursing facility (SNF) days. Counts in the period after palliative care encounter are presented as a rate per 90 days alive to account for censoring due to high mortality rates. Hospice enrollment was measured as the date of the first hospice claim in Medicare or VA. Days from hospice enrollment to death were evaluated among hospice users with a documented date of death during or after the study period (through 2022); while our health services data were limited to 2014–2017, the VA data provided dates of death that occurred through 2022 improving our measure of time to death.

This study protocol was approved by the Durham VA institutional review board.

Statistical analysis

Descriptive statistics were evaluated for three groups: veterans with inpatient-only specialty palliative care, veterans with any outpatient specialty palliative care encounter (with or without inpatient encounter), and veterans with ICD palliative care encounters. The group with ICD palliative care encounters were presented for illustrative purposes, but not included in the main multi-variable logistic regression analysis We conducted sensitivity analyses including all three groups (Supplementary Table S2) and comparing characteristics by cancer LLC versus other LLCs (Supplementary Tables S3 and S4).

We created a veteran-level cross-sectional data set pooled across 4 years, including covariates at both the veteran-level and site-level. Among the veterans who had at least one specialty palliative care encounter during the period (in any setting), we used logistic regression to evaluate conditional associations between veteran and site characteristics and the probability of two binary outcomes: at least one outpatient specialty palliative care encounter with or without an inpatient palliative care encounter (vs. inpatient only), and having only one specialty palliative care encounter in any care setting (vs. more than one). All available characteristics and prior 90-day utilization were included in the models, except for ICU and SNF days due to collinearity with hospital days. Models included Veterans Integrated Services Networks (VA health system) and study year as fixed effects.

RESULTS

Of 188,234 veterans with any palliative care encounters during the period 2014–2017, two-thirds (n = 120,249) had a specialty encounter documented in the inpatient or outpatient setting (Table 1). ICD-only palliative care encounters were more likely at low complexity facilities and among veterans using home-based primary care, with 6+ chronic conditions or depression, and over age 85. Most palliative care users with a specialty encounter received palliative care services only in the inpatient setting (n = 81,523), and about one-third of users had at least one palliative care encounter documented in the outpatient setting (n = 38,726).

TABLE 1.

Characteristics of veterans with documented palliative care encounters.

Veteran demographics	Veterans with inpatient-only specialty PC N = 81,523	Veterans with outpatient, specialty PC N = 38,726	Veterans with Other PC encounters N = 67,985	Total N = 188,234
	%	%	%	%
Age at first PC encounter	75.0 (11.8)	73.5 (11.5)	77.0 (12.1)	75.4 (11.9)
Mean (SD)
Under age 65	20.20%	23.40%	16.60%	19.50%
Age 65 to <75	32.20%	34.80%	27.60%	31.10%
Age 75 to <85	23.40%	22.20%	24.20%	23.40%
Age 85+	24.30%	19.60%	31.60%	26.00%
Female	2.60%	3.10%	3.00%	2.90%
Race and ethnicity
Non-Hispanic White	72.30%	75.50%	79.20%	75.40%
Non-Hispanic Black	17.80%	15.30%	13.40%	15.70%
Hispanic	6.80%	5.80%	4.30%	5.70%
Missing or other	3.20%	3.40%	3.20%	3.20%
Married	43.00%	48.50%	51.10%	47.10%
Rural status	29.00%	30.20%	30.30%	29.70%
High area deprivation	25.20%	21.20%	21.10%	22.90%
Priority status for high disability	39.80%	37.00%	38.90%	38.90%
Priority status for low-income	29.00%	30.60%	22.90%	27.10%
High risk of unstable housing	21.70%	22.00%	20.90%	21.50%
LLC prior to PC encounter
Cancer	30.90%	39.80%	25.10%	30.60%
CHF	35.60%	19.70%	19.30%	26.40%
COPD	36.50%	25.40%	24.90%	30.00%
Dementia	27.30%	12.60%	17.30%	20.70%
ESRD	7.70%	6.10%	4.70%	6.30%
HIV/AIDS	1.10%	0.80%	0.60%	0.80%
No prior LLC	9.90%	21.40%	33.00%	20.60%
Complex health needs
JEN frailty category
Low	3.20%	7.50%	9.40%	6.30%
Med	34.30%	39.00%	37.20%	36.30%
High	62.60%	53.50%	53.40%	57.40%
Depression, ever diagnosed	24.30%	23.00%	29.60%	25.90%
CAN score
<80	9.60%	12.40%	15.50%	12.30%
80 to <90	9.20%	12.70%	13.70%	11.50%
90 to 100	64.80%	68.30%	58.30%	63.20%
Missing CAN	16.50%	6.60%	12.50%	13.00%
Charlson >1	47.00%	55.20%	48.20%	49.10%
Chronic conditions 6+	30.10%	28.10%	40.20%	33.40%
Home-based primary care	11.40%	11.80%	23.20%	15.70%
Site characteristics
High complexity	83.30%	80.40%	75.30%	79.80%
Rural status	6.10%	4.40%	5.50%	5.50%

Note: The inpatient specialty palliative care category excludes veterans with one or more outpatient specialty palliative care encounters to describe mutually exclusive categories. All comparisons are statistically significant at p < 0.001 due to the sample size.

Abbreviations: CAN, care assessment needs; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; ESRD, end stage renal disease; HIV/AIDS, human immunodeficiency virus/acquired immunodeficiency syndrome; LLC, life-limiting condition; PC, palliative care.

Characteristics of palliative care users (Table 1)

Veteran demographics were similar between the palliative care users in the inpatient and outpatient settings, including age, race, marital status, and rurality. Socioeconomic factors such as priority status and housing instability were similar, though inpatient-only palliative care users were more likely to live in an area with high deprivation. Differences in life-limiting conditions category, number of conditions, Care Assessment Need scores, and frailty were observed between groups. Veterans accessing specialty palliative care in the outpatient versus inpatient setting were more likely to have cancer (40% vs. 31%) and less likely to have congestive heart failure (20% vs. 36%), COPD (25% vs. 37%), or dementia (13% vs. 27%). Outpatient versus inpatient-only palliative care users were less likely to have more than one life-limiting condition (31.9% vs. 50.2%, respectively). Outpatient palliative care users were also more likely to not meet criteria for any life-limiting condition (21% vs. 10%) and have low frailty scores (8% vs. 3%). The VA identifies veterans with high Care Assessment Needs scores (98%–100) as eligible for PC; however, palliative care users in our sample often had scores lower than 98 or no score. High scores were more likely among inpatient-only palliative care users (42.7%) than outpatient palliative care users (29.9%).

Timing of palliative care encounters (Table 2)

TABLE 2.

Characteristics of palliative care use among veterans by care setting.

	Veterans with inpatient-only specialty PC N = 81,523	Veterans with outpatient, specialty PC N = 38,726		Other PC encounters (ICD-only) N = 67,985
	% or Mean (SD)	% or Mean (SD)	StDiff	% or Mean (SD)
Index PC encounter characteristics
Within 2 years of LLC, %	42.80	37.90	0.10	29.60
Inpatient, %	100.00	9.90	4.26	36.50
PC episode characteristics
Only one PC encounter, %	68.60	24.60	0.98	61.80
Duration, days first to last PC	58.53 (163.08)	159.66 (248.53)	−0.63	78.92 (206.14)
Days between visits	34.85 (49.82)	28.80 (40.10)	−0.48	35.81 (59.75)
PC encounter types, count
Specialty inpatient	1.14 (0.43)	0.32 (0.67)	1.45	n/a
Specialty outpatient	n/a	1.19 (0.67)		n/a
Other inpatient (ICD-only)	1.23 (0.62)	0.56 (0.88)	0.88	0.47 (0.58)
Other outpatient (ICD-only)	0.71 (3.3)	3.87 (6.16)	−0.64	2.68 (6.94)
Total PC encounters (all types)	3.08 (3.6)	5.94 (6.66)	−0.53	3.15 (6.86)
Telehealth encounters	2.39 (3.88)	3.13 (4.71)	0.17	2.24 (3.50)
Mortality, %
90 days of first PC	63.00	30.80	0.68	53.90
1 year of first PC	76.60	56.30	0.44	68.20
Hospice Use, %
Prior hospice, total	18.10	8.60	0.28	25.90
Subsequent hospice, total	66.40	66.50	−0.00	66.00
Subsequent hospice, VA	53.70	48.40	0.11	41.20
Subsequent hospice, Medicare	27.40	40.70	−0.28	44.30
Hospice days, total	81.99 (165.49)	106.57 (176.47)	−0.14	112.51 (198.97)
Utilization before index PC
Emergency department visits	1.07 (1.64)	0.65 (1.36)	0.28	0.50(1.12)
Inpatient days	1.68 (1.21)	0.67 (1.03)	0.90	0.68 (1.04)
ICU days	0.51 (0.96)	0.64 (1.04)	−0.14	0.40 (0.78)
SNF days	0.10 (0.39)	0.04 (0.23)	0.19	0.04 (0.24)
Utilization after index PC
Emergency department visits	1.07 (0.22)	0.69 (0.13)	−0.04	0.80 (0.20)
Inpatient days	1.40 (0.65)	0.25 (0.38)	0.58	0.59 (0.62)
ICU days	0.34 (0.90)	0.20 (0.05)	0.02	0.24 (0.08)
SNF days	0.16 (0.04)	0.10 (0.02)	0.04	0.12 (0.03)

Note: Group means were significantly different (p < .05) for all characteristics due to the sample size; thus, standardized differences are presented, and differences greater than 0.1 standard deviation are in bold to indicate a meaningful difference.

Abbreviations: ADI, Area Deprivation Index; ICU, intensive care unit; LLC, life-limiting condition; PC, palliative care; SNF, skilled nursing facility; VA, Veterans Health Administration.

Compared to inpatient palliative care users, outpatient palliative care users were less likely to have had hospice use before palliative care (18% vs. 9%), and more likely to have only one palliative care encounter (25% vs. 69%). The duration between first and last palliative care encounter and number of encounters was greater for outpatient (158 days, six encounters) versus inpatient-only (58 days, three encounters) groups, primarily due to higher 90-day (63% vs. 30%) and 1 year (77% vs. 56%) mortality in the inpatient group. Outpatient versus inpatient palliative care users had lower rates of healthcare utilization before and after the first palliative care encounter.

There was variation in timing and duration of palliative care use by care setting (Figure 1). Across all care settings, most palliative care users had only one specialty encounter prior to death or hospice enrollment (69.3%, 84,324). About one-third of veterans using palliative care initiated palliative care within 2 years of a documented life-limiting condition in either the inpatient or outpatient setting (median = 1.8 and 1.7 years, respectively.) The duration between the first and last documented palliative care use (excluding hospice) was typically longer among outpatient palliative care users (median = 106 - days, IQR = 27, 306) compared to inpatient palliative care users (median = 80 days, IQR =25, 245). Duration of hospice use was similar between inpatient (median = 37 days; IQR = 11, 112) and outpatient (median = 44 days; IQR = 14, 118) palliative care users. Veterans with only one palliative care encounter typically enrolled in hospice only 18 days (median = 18 days; IQR = 5, 64 days) prior to death, and follow-up palliative care encounters may not be needed due to hospice enrollment or death.

FIGURE 1.

Timing and duration of veterans’ palliative care and hospice use by care setting and dose (2014–2017). LLC, life-limiting conditions; PC, palliative care of unknown type; SPC=specialty palliative care. Among veterans with LLC who had at least one palliative care encounter, we defined four groups: (1) veterans with no encounters with the specialty palliative care clinic, but who did have palliative services outside the specialty clinic (indicated by ICD-9/10 PC codes); (2) veterans with only one palliative care encounter in an inpatient or outpatient specialty clinic (indicated by stop codes); (3) veterans with more than one encounter for palliative care in an inpatient specialty clinic, but no outpatient encounters; (4) veterans with more than one encounter for palliative care in an outpatient specialty clinic. Time since life-limiting condition (LLC) is reported among those with any LLC documented.

Factors associated with characteristics of palliative care use (Table 3)

TABLE 3.

Veteran and site characteristics associated with palliative care setting and dose among veterans with specialty palliative care.

	Outpatient versus inpatient-only PC encounter		Only one versus more than one, PC encounter
	n = 120,249		n = 120,249
	Adjusted OR	95% CI	Adjusted OR	95% CI
Veteran demographics
Age at first PC encounter
Under age 65	Ref		Ref
Age 65 to <75	0.93***	(0.89–0.96)	1.12***	(0.89–0.96)
Age 75 to <85	0.81***	(0.77–0.85)	1.31***	(0.77–0.85)
Age 85+	0.64***	(0.61–0.67)	1.57***	(0.61–0.67)
Female	1.10*	(1.01-1.20)	0.96	(1.01-1.20)
Race and Ethnicity
Non-Hispanic White	Ref		Ref
Non-Hispanic Black	0.87***	(0.84–0.91)	0.95***	(0.84–0.91)
Hispanic	0.78***	(0.73–0.83)	0.99***	(0.73–0.83)
Other	0.99	(0.91–1.07)	1.02	(0.91–1.07)
Married	1.20***	(1.17–1.24)	0.91***	(1.17–1.24)
Socioeconomic status
Rural status	0.95*	(0.92–0.99)	1.10***	(0.92–0.99)
Priority status for high disability	0.90***	(1.17–1.24)	1.02***	(1.17–1.24)
Priority status for low-income	1.00	(0.96–1.04)	0.99	(0.96–1.04)
High risk of unstable housing	1.12***	(1.08–1.16)	0.71***	(1.08–1.16)
LLC before PC encounter
Cancer	1.25***	(1.21–1.30)	0.69***	(1.21–1.30)
CHF	0.67***	(0.65–0.70)	1.05**	(0.65–0.70)
COPD	0.74***	(0.71–0.76)	1.05***	(0.71–0.76)
Dementia	0.52***	(0.50–0.55)	1.42***	(0.50–0.55)
ESRD	0.82***	(0.77–0.87)	1.25***	(0.77–0.87)
HIV/AIDS	0.91	(0.78–1.05)	0.93	(0.78–1.05)
Complex health needs
JEN frailty category
Low	Ref		Ref
Med	0.66***	(0.62–0.71)	1.02***	(0.62–0.71)
High	0.63*	(0.59–0.68)	0.96***	(0.59–0.68)
Depression, ever diagnosed	1.05*	(1.01–1.09)	0.96*	(1.01–1.09)
CAN Score
<80	Ref		Ref
80 to <90	1.21***	(1.14–1.28)	0.88***	(1.14–1.28)
90 to 100	1.74***	(1.66–1.82)	0.70***	(1.66–1.82)
Missing CAN	0.39***	(0.36–0.41)	1.81***	(0.36–0.41)
Charlson >1	0.81***	(0.77–0.84)	1.10***	(0.77–0.84)
Chronic conditions 6+	1.28***	(1.23–1.33)	0.94***	(1.23–1.33)
Home-based primary care	1.16***	(1.11–1.22)	0.58***	(1.11–1.22)
Site characteristics
High complexity	0.82***	(0.78–0.85)	1.17***	(0.78–0.85)
Rural status	0.60***	(0.56–0.65)	2.00***	(0.56–0.65)
Prior utilization
Inpatient days	0.36***	(0.35–0.37)	1.33***	(0.35–0.37)
Emergency department visits	1.03***	(1.02–1.04)	0 97***	(1.02–1.04)
VA community living center	0.26***	(0.24–0.29)	2.70***	(0.24–0.29)

Note: Models included veterans integrated services networks (VISN) and year fixed effects.

Abbreviations: CAN, care assessment needs; CHF, congestive heart failure; CI, 95% confidence interval; COPD, chronic obstructive pulmonary disease; ESRD, end stage renal disease; HIV/AIDs, human immunodeficiency virus/acquired immunodeficiency syndrome; LLC, life-limiting condition; OR, odds ratio; PC, palliative care.

^*p-value < 0.05;

^**p-value < 0.01;

^***p-value< 0.001.

Conditional on other factors including demographics and prior utilization, odds of outpatient palliative care encounter are increased for veterans with cancer (aOR = 1.25; CI = 1.21, 1.30) and reduced for veterans with other types of conditions compared to not meeting criteria for life-limiting condition before palliative care use. Veterans receiving care in low complexity or rural facilities had reduced odds of any outpatient palliative care encounter (aOR = 0.82; CI = 0.78, 0.85; and aOR = 0.60; CI = 0.56, 0.65) compared to inpatient only.

Conditional on other factors including demographics and prior utilization (Table 3), veterans with cancer had lower odds (adjusted odds ratio (aOR) = 0.69; 95% confidence interval (CI) = 0.67, 0.71) of only one palliative care encounter, meaning cancer patients were more likely to have a second palliative care encounter or more. Non-cancer conditions were associated with increased odds of only one palliative care encounter. The risk of housing instability was associated with reduced odds of having only one palliative care encounter (aOR = 0.71; CI = 0.69, 0.74). Veterans receiving care in low-complexity or rural facilities had increased odds (aOR = 1.17; CI = 1.12, 1.22; and aOR = 2.00; CI = 1.87, 2.14) of experiencing only one palliative care encounter.

DISCUSSION

This national evaluation provides novel insights into the care setting and timing of VA specialty palliative care use for veterans, as well as hospice use before and after PC. Among the cohort, one-third received at least some of their palliative care in the outpatient setting. Veterans with any outpatient specialty palliative care were more likely to have cancer and less likely to have heart failure or COPD. Across care settings, hospice enrollment and duration are similar across inpatient and outpatient groups. Differences between veterans with inpatient and outpatient palliative care use motivate the need for further research to understand how care setting and timing of palliative care impacts outcomes for veterans and older adults with life-limiting conditions in and out of the VA.

Timing and use of palliative care differs for veterans using outpatient compared to only inpatient palliative care, which may be a factor in the veteran and family experience of care. Earlier inpatient or outpatient palliative care is associated with greater family satisfaction with care,³⁶ intensity of care,^19,37 and even longer life expectancy.^19,38 Similarly, late enrollment in hospice has been associated with the intensity of care at the end of life and family satisfaction with care.^39–41 We found that more than two-thirds (69%) had only one specialty palliative care encounter and with a short interval (median of 18 days) for hospice use. These veterans may have had only a single palliative care encounter because of the end-stage nature of their disease, resulting in shorter times to hospice enrollment or death. Mortality rates were lower among veterans accessing specialty palliative care in outpatient setting relative to those in the inpatient settings, which is expected due to their lower prevalence of life-limiting conditions; however, quantifying the duration of survival following initiation of outpatient palliative care is novel. Among veterans with a second palliative care encounter, most veterans had 80–100 days (about 3 months) between their first and last specialty palliative care encounter in both inpatient and outpatient settings. In some cases, one palliative care encounter may be the most beneficial and appropriate experience for older adults who experience a rapid decline in their final weeks of life; however, early initiation of palliative care, when appropriate, is expected to improve quality of life for patients and caregivers.

Compared to non-cancer conditions, veterans with life-limiting cancer were more likely to use outpatient specialty palliative care. This difference may be because of the more robust evidence for the benefits of early, outpatient palliative care in oncology compared to other conditions. In cancer patients, a meta-analysis found outpatient palliative care integrated with oncology interventions had positive effects on short-term quality of life, symptom burden, and survival.⁷ Preliminary evidence suggests outpatient palliative care consults may be effective in patients with certain conditions such as advanced heart failure and COPD.^4,42–48 Systematic reviews specifically evaluating specialist outpatient palliative care among patients with heart failure have reported improvement in patient-reported outcomes; however, outpatient palliative care interventions were heterogeneous, and most of the evidence was limited to small trials and observational studies.^44–46

There has been a historical partnership between palliative care and cancer care nationally, and we observe a similar pattern in the VA. The connection between the two disciplines has been further supported by institutional mechanisms, particularly within Medicare.⁴⁹ For example, Medicare value-based payment models such as the Oncology Care Model have supported infrastructure for palliative care services to be embedded within oncology clinics.⁷ The model requires quality measures that align with the goals of specialty palliative care. Similarly, the new Medicare model Guiding an Improved Dementia Experience aims to improve patient and caregiver quality of life and requires services common in palliative care such as longitudinal prognosis education, caregiver education and supports, and connections to social services. Over the next decade, new payment models may create potential reimbursement to support a similar increase in the palliative care infrastructure and workforce embedded within neurology clinics.

This national, cross-sectional cohort study sought to describe current patterns of specialty palliative care use in the VA across settings; however, there are many limitations to these data. First, we are unable to discern the veteran experience in these administrative data. Second, these data do not reflect veterans’ use of non-hospice specialty palliative care outside of the VA, and within the VA, care setting may be misclassified for up to 10% of encounters.²⁹ However, this misclassification decreased over time. Third, due to the cross-sectional study design, our results are impacted by censoring at the beginning and end of the study period. Duration of palliative care may be underestimated for veterans accessing palliative care at the end of the study period with limited follow-up time. Some veterans may have accessed palliative care before the study period, which would lead to overestimation of the time from life-limiting condition documentation to first palliative care use. Overall, our period of observation (4 years) and mean follow-up time (3 years) was much longer than the average duration of palliative care use, so we expect that the results reflect the typical experience for the population of palliative care users in the VA during this period. Our veteran cohort is comprised of mostly male older adults and is not generalizable to non-veterans.

This national evaluation fills a gap in understanding of veteran characteristics, dose and timing of VA palliative care, and Medicare or VA hospice use among veterans receiving palliative care in the inpatient or outpatient setting. Future work should seek to understand the causes and outcomes of differences in palliative care patterns for veterans with non-cancer life-limiting conditions. More evidence is needed to understand whether the use of specialty palliative care in outpatient clinics contributes to improved outcomes for veterans and older adults with life-limiting illness and their caregivers.

Key points

• This national cohort study compared veteran characteristics and hospice use for palliative care users by care setting (inpatient vs. outpatient) and dose (number of palliative care encounters).

• Outpatient versus inpatient palliative care users were more likely to have cancer and less likely to have high frailty, but sociodemographic factors and hospice use were similar.

• Differences between veterans with inpatient and outpatient use motivate the need for further research to understand how care setting and number of palliative care encounters impact outcomes for veterans and older adults.

Why does this paper matter?

This national evaluation provides novel insights into the care setting and dose of Veterans Health Administration specialty palliative care use for US veterans.