Implications of Palliative Care Consultation Timing among a Cohort of Hospice Decedents

Abstract

Background: Earlier palliative care consultation is associated with less intensive medical care and improved quality outcomes for patients with cancer. However, there are limited data about how the timing of palliative care affects utilization among noncancer patients exposed to palliative care consultation.

Objective: Comparison of health care utilization for hospice decedents who received early versus late palliative care.

Design: A retrospective cohort study utilizing hospital and hospice administrative databases.

Setting/Subjects: Patients with cancer and noncancer diagnoses who received specialty palliative care consultation before dying at a local hospice.

Measurements: Comparing early (>90 days before death) versus late (<90 days before death) palliative care, outcome measures included intensive care unit (ICU) utilization and hospice length of stay (LOS).

Results: Of 233 hospice decedents in 2014 who had palliative care referrals, 36 (15.4%) had early and 197 (84.5%) had late referrals. Nearly half of the patients had a noncancer hospice diagnosis. Only 6% of the early group used the ICU in the last month of life, whereas 56% of the late group did. Patients receiving early palliative care had a longer median hospice LOS than those with late palliative care (138 days vs. 8 days).

Conclusions: Early palliative care appears to reduce intensive medical care and increase hospice LOS for patients with a variety of end-stage diseases.

Introduction

Many studies of cancer populations associate earlier palliative care referral with positive effects on a range of clinical outcomes, including less aggressive medical care in the last months of life, particularly reduced intensive care utilization.¹ This exploratory report addresses care patterns of patients enrolled in hospice after early and late specialty palliative care exposure. This gap is relevant as similar studies focus almost exclusively on cancer patients, when >70% of hospice decedents carry noncancer terminal diagnoses.²

Despite research supporting early palliative care, early specialty palliative care remains atypical and definitions of early vary widely.^3,4 Most referrals to palliative care occur late in the disease trajectory.^5–7 Furthermore, a majority of patients referred to hospice die within 30 days of initiating the benefit and ∼28% die within 1 week of enrollment.² For perspective, in 1990, <15% of hospice enrollees died within a week.⁸

The trend toward very short hospice length of stay (LOS) has been increasing, despite the proliferation of hospice and palliative care services over the past decade.⁹ End-of-life (EOL) care has grown arguably more aggressive among cancer and noncancer patients.^10–12 From the perspective of affected families, this aggressive EOL care is not associated with higher-quality care.¹³

Early palliative care's effects on EOL utilization patterns are inadequately described.¹⁴ Furthermore, sparse data exist defining the association between early specialty palliative care consultation and hospice LOS, particularly for noncancer patients.^15,16 Few studies address hospice LOS as an outcome of interest, beyond identifying patients not referred to hospice or referred within three days of death. We, therefore, undertook a retrospective cohort study to analyze the association between early palliative care referral, intensive care unit (ICU) use, and hospice LOS in patients with and without cancer diagnoses.

Materials and Methods

Study design and data sources

This was a retrospective cohort study of decedents who received palliative care at a large urban tertiary referral center and were referred to a single nonprofit hospice agency. A convenience sample of data from 2014 was used. Inclusion criteria were patients of age 18 years or older who died under the care of this hospice agency during 2014 and who had received an inpatient or outpatient palliative care consultation at Vanderbilt Medical Center at any time.

The palliative care team provided consultation services for both outpatient and inpatients diagnosed with end-stage illnesses. The team consisted of physicians, nurse practitioners, and a social worker. The team engaged in goals of care conversations, prescribed or recommended medications, and was involved in disposition planning to the local community hospice agency. The institutional review board of our institution approved this study.

Variable construction

The primary exposure of interest was the timing of palliative care. Patients exposed to palliative care >90 days before death were considered to have early palliative care and those patients exposed to palliative care <90 days before death were considered to have received late palliative care. The cutoff of three months was chosen based on recent studies' definition of early palliative care in the outpatient cancer setting.^17,18 The primary outcome of interest was ICU use within the last six months of life with hospice LOS and hospitalizations as secondary outcomes. Patient were considered to have ICU use within the last six months of life if they were admitted to the ICU at XXX Medical Center at any time during the last six months of their life.

Covariates measured included the patient's age at death, gender, race, hospice diagnosis, and whether the initial exposure to palliative care was inpatient or outpatient. Payers were classified as Medicaid, Medicare, private, or uninsured. The hospice diagnosis was separated into nine categories (solid tumor, hematological, cirrhosis, congestive heart failure, neurodegenerative, stroke, pulmonary, renal, and other) and then subcategorized to cancer hospice diagnosis (solid tumor or hematological) and noncancer (all other diagnoses).

Analysis

We explored the association between timing of palliative care consultation and ICU use in the last six months of life using a logistic regression model adjusting for observed patient characteristics. Secondary analyses included a subgroup analysis by type of diagnosis and unadjusted comparisons of secondary outcomes. Secondary EOL care outcomes include ICU use in the last month of life, number of days in the ICU in the last six months of life, number of hospitalizations in the last six months of life, number of patients with a hospice LOS less than three days, and hospice LOS. For the unadjusted analyses, hypothesis testing was conducted using Fischer's Exact Test for categorical variables and the Mann–Whitney U test for continuous variables. We provided 95% Wilson confidence intervals (CIs) for the categorical variables. Analyses were conducted using R (version 3.2.2).

Results

Of 233 hospice decedents in 2014 who had palliative care referrals, 36 had early and 197 had late referrals. The distribution of patient characteristics, payment type, and type of initial consultation is given in Table 1. The majority of patients had their initial exposure to palliative care in the inpatient setting (94.8% of cohort, 77.8% early, and 98% late). Nearly half of the patients in this cohort had a hospice diagnosis other than cancer (48.5% of cohort, 55.6% early, and 47.2% late).

Table 1.

Patient Characteristics

	Early palliative care (n = 36)	Late palliative care (n = 197)	pa
Age, median [IQR]	64.00 [54, 79]	68.0 [56, 77]	0.74
Gender, count (%)
Male	16 (44.4)	100 (50.8)	0.59
Female	20 (55.6)	97 (49.2)
Race, count (%)
White	28 (77.8)	143 (72.6)	0.38
Black	6 (16.7)	48 (24.4)
Other	2 (5.6)	6 (3.0)
Cancer, count (%)
No	20 (55.6)	93 (47.2)	0.37
Yes	16 (44.4)	104 (52.8)
Insurance type, count (%)
Medicare	19 (52.8)	114 (57.9)	0.14
Medicaid	12 (33.3)	38 (19.3)
Nonpublic/none	5 (13.9)	45 (22.8)
Initial exposure to palliative care, count (%)
Inpatient	28 (77.8)	193 (98.0)	<0.001
Outpatient	8 (22.2)	4 (2.0)
Number of palliative care consults, median [IQR]
Inpatient	3 [2, 7]	3 [2, 5]	0.99
Outpatient	0 [0, 1]	0 [0, 0]	<0.001

^aThe p value is obtained from a Mann–Whitney U test for the continuous variables and Fisher's exact test for the categorical variables.

IQR, interquartile range.

The percentage of those who utilized the ICU in the last six months of life was greater for those entering palliative care late (56%, 95% Wilson CI 49%–63%) than for those enrolling in palliative care early (19%, 95% Wilson CI 10%–35%). The adjusted odds ratio for ICU use in the last six months of life showed a strong protective effect of early palliative care compared with late palliative care (0.15, 95% CI 0.05–0.41). The logistic regression model adjusted for age, gender, a hospice diagnosis of cancer, payment type, and whether the initial exposure to palliative care was in an outpatient setting. Age was modeled nonlinearly using restricted cubic splines with three degrees of freedom. There was a stronger protective effect when focusing only on those without a cancer diagnosis (0.03, 95% CI 0.00–0.24). The adjusted odds ratio for ICU use was 0.36 (95% CI 0.09–1.33) when focusing on those with a cancer diagnosis.

Summary measures for the secondary outcomes are given in Table 2 along with the p values for comparisons between the early versus late palliative care groups. There was a greater use of the ICU in last month of life for those exposed to palliative care late (41%, 95% Wilson CI 35%–48%) than those enrolling in palliative care early (6%, 95% Wilson CI 2%–18%). The median number of days in the ICU (Table 2) was also greater for those encountering palliative care late (3 days) than those entering palliative care early (0 days). The median LOS in hospice (Table 2) was different between early and late palliative care patients with a median of 138 days for those exposed early and only 8 days for those exposed late.

Table 2.

End-of-Life Health Care Utilization

	Early palliative care	Late palliative care
n	36	197
Number of ICU days in last six months, median [IQR]	0.0 [0.0, 0.0]	3.0 [0.0, 7.0]**
Number of hospitalizations in last six months, median [IQR]	1.0 [0.0, 2.0]	1.0 [1.0, 3.0]**
Hospice length of stay,a median [IQR]	138.0 [70.0, 266.5]	8.0 [3.8, 20.0]**
Hospice length of stay less than three days,a % (95% Wilson CI)	5.6 (1.5–18.1)	25.0 (19.5–31.5)*
ICU use in month prior, % (95% Wilson CI)	5.6 (1.5–18.1)	41.1 (34.2–47.8)**

^aOne patient in the late palliative care group was missing hospice length of stay, so was discarded from this analysis.

^*p value <0.01 and ^**p value <0.001, where the p value is obtained using the nonparametric Mann–Whitney U test for continuous outcomes and Fisher's exact test for the binary outcomes.

CI, confidence interval; ICU, intensive care unit.

Discussion

This retrospective study suggests that early palliative care is associated with a reduction in intensive EOL medical care and an increase in hospice LOS for patients with a variety of terminal diseases. Adjusting for where palliative care intervened (outpatient or inpatient) and several patient characteristics, earlier referrals were strongly associated with decreased ICU utilization in the last six months of life. As other studies of cancer patients suggest, early palliative care intervention, defined as occurring at least three months before death, reduces the intensity of EOL care to a degree that late palliative care does not. Moreover, this study suggests similar benefits of early palliative care for an array of noncancer diagnoses. Future research might prospectively investigate the association between early palliative care referrals and health care utilization—both within hospitals and hospices.

This study has several limitations that should be kept in mind when interpreting its results. An observational study of a single institution and single hospice agency may not be generalizable to other settings. It is possible that patients received ICU care at other hospitals. The 90 day cutoff definition of early palliative care is somewhat arbitrary and other cutoffs might find similar correlations. The observational nature of the study does not allow us to directly evaluate the effect of early palliative care on EOL care, but the adjusted logistic regression analysis does suggest a strong association between early palliative care and decreased ICU utilization in the last six months of life. Multivariable adjusted analyses of the other measures of EOL utilization were not feasible because of the small sample size. Moreover, it is probable that some uncontrolled confounding still exists in the results due to the limited number of covariates we could include in our adjusted analysis.

Nevertheless, this study adds to the evidence for the value of earlier palliative care as a way of reducing EOL ICU utilization and increasing hospice LOS. Although more robust studies are needed to more usefully define early palliative care, this brief report supports efforts to involve palliative care months earlier than is typical in the care of patients with both malignant and nonmalignant serious illness.

Author Disclosure Statement

Dr. Martin serves as medical director of outpatient palliative care at Vanderbilt Medical Center. Drs. Robbins, Martin, and Shinall practice both inpatient and outpatient palliative medicine at Vanderbilt Medical Center. Dr. Shinall's work was funded by the NIH/NCI (KIZCAO90625). The funder had no roll in the design, analysis, or publication of this study. Dr. Hackstadt is an assistant professor at the Department of Biostatistics at Vanderbilt Medical Center. None of the other authors has any relevant financial relationships to disclose.