The Role of Race and Ethnicity in Predicting Length of Hospice Care among Older Adults

Abstract

Background:

The purpose of the study was to examine both direct and interactive roles of race/ethnicity with patients' characteristics (age, gender, relationship with caregiver, diagnosis, referral source, and payment type) in predicting length of hospice care.

Method:

This study included a total of 16,323 patients 65 years of age and older (M_age=81.4, SD=8.3) who were served by a hospice in central Florida during a four-year period, 2002-2006. Survival analyses were conducted using the Cox proportional hazards model to predict the length of hospice care and test the interaction effects of race/ethnicity.

Results:

The majority of subjects (83.5%) were white, 7.6% were African-American, and 8.9% were Hispanic. During the study period, 58.5% died. All patient characteristics were significantly associated with the length of hospice care (p < .05). Overall, Hispanics had the longest hospice stay (M=98.84 days), followed by African-Americans (M=90.29) and whites (M=88.20). With the exception of African-American women who were no more likely to stay longer under hospice care than African-American men, the women in this study stayed longer under hospice care than men did. Patients referred from long-term care (LTC) settings had shorter stays in hospice care compared to those referred by physicians in other settings. Additionally, African-Americans and Hispanics referred from LTC had significantly shorter hospice stays than those referred by primary physicians.

Conclusion:

In this limited sample of hospice patients, length of stay was longer for minority patients than white patients.

Introduction

Hospice care provides essential services including medical care, pain management, as well as emotional and spiritual support for patients at the end of life (EOL).¹ Although hospice use among racial/ethnic minorities has grown during the past several years,² African-Americans and Hispanics were found to use hospice services at significantly lower rates compared to European-Americans, even after controlling for other sociodemographic characteristics, medical history, prognosis, and access to health care.^1,3–8

The optimal length of hospice care for each patient varies based on several factors such as type of disease and its course, the patient's performance, and access to care.^9,10 Each of these factors make it difficult for health care professionals to agree on the optimal length of a patient's hospice stay.¹⁰ Although it is difficult to decide when it is the “right” time, late referrals could be associated with poor EOL care outcomes, higher concerns with the care, and lower patient satisfaction with the quality of EOL care.¹¹

Despite some inconsistent findings across racial/ethnic groups regarding length of hospice care, a number of patients' characteristics have been consistently associated with the length of hospice care. Shorter length of hospice care has been associated with: younger age, male gender, referrals from an in-patient facility (e.g., nursing homes or assisted living settings) or hospital settings, cancer diagnosis, and having Medicare-only health insurance (versus those with Medicaid coverage or Medicare and private health insurance).²

Of the studies that have examined utilization of hospice care and the length of hospice care across different racial/ethnic groups,^{1,4–8,12,13} most have used race/ethnicity as predictors or examined racial/ethnic groups separately, and none addresses the interactive role of race/ethnicity in the relationship between patients' characteristics and length of hospice care. This is an important gap in prior research, because evidence suggests that patients' characteristics (age, gender, relationship with caregiver, diagnosis, referral source, and payment type) are differentially associated with race/ethnicity.⁷ For example, a recent study by Johnson et al.¹⁴ highlights the importance of considering racial differences in the relationship between referral location before hospice enrollment and length of hospice care. The referral source/location (e.g., acute care setting vs. non-acute care setting) could be linked to the timing of referrals and the length of hospice care ultimately impacting hospice use. The omission of the interactive role of race/ethnicity may obscure variations within racial/ethnic groups and lead to an overgeneralization of the association between patients' characteristics and length of hospice care. Thus, the purpose of the study was to examine both direct and interactive roles of race/ethnicity with other patient characteristics in predicting length of hospice care. Explaining group-specific effects could help generate hospice services that accommodate the special circumstances of minority populations.⁵

Method

Sample

The data were provided by a not-for-profit independent hospice provider via its electronic medical records system in central Florida. The sample included a total of 16,323 patients age 65 and older who were enrolled at the hospice over a four-year period, August 2002 to August 2006. Of the participants, 83.5% were white, 7.6% were African-American, and 8.9% were Hispanic. The study was approved by the University of South Florida Institutional Review Board (IRB) and by the IRB of the hospice organization from which the data were obtained.

Measures

The outcome variable, length of hospice care, was measured by the number of days between enrollment in hospice and the date of death. Race/ethnicity (white, African American, and Hispanic) was the major predictor variable. Other covariates included age; gender (0=male, 1=female); primary caregiver relationships (spouse, daughter, son, and other); diagnosis (cancer, Alzheimer's disease/dementia, COPD/CHF/heart disease, and other); primary payer (private health insurance/private pay, Medicaid, Medicare); and referral sources (primary physician, hospital, LTC settings including nursing home and assisted living, and other).

Data analysis

Descriptive analyses–chi-square tests and one-way analysis of variance (ANOVA)–were used to examine the relationship of racial/ethnic groups to patients' characteristics and to the length of hospice care. A survival analysis using the Cox proportional hazards model was used to estimate the relative risk/hazard of the event (i.e., death in this study).^15,16 Since a “higher risk of death” could mean “shorter length of hospice care,” the two terms were interchangeably used in this study.² A hazard ratio greater than 1.00 indicates a higher risk of death (i.e., a shorter length of hospice care), relative to the reference group.

To test the modifying effects of race/ethnicity, each interaction term was added into the main effects model step by step. A modifying or interaction effect occurs when the modifying variable (race/ethnicity) influences the direction and strength of the relationship between an independent variable and a dependent variable, and the interaction between the modifying variable and the independent variable is significant. Post hoc analyses further examined the significant interaction effects to explain the specific influences of race/ethnicity on a certain variable. To compare the main effects model and the interactions model, a likelihood ratio chi-square test was used to assess the model fit.¹⁷

Results

Sample characteristics

Of 16,323 patients, 58.5% (n=9,543) died. The characteristics of all participants across the three racial/ethnic groups are presented in Table 1. Hispanic patients were more likely to stay longer under hospice care (M=98.84 days, SD=145.54) than African-Americans (M=90.29, SD=139.85) or whites (M=88.20, SD=139.05).

Table 1.

Characteristics of Hospice Patients by Racial/Ethnic Groups (N=16,323)

	White (n=13,628)	African-American (n=1,245)	Hispanic (n=1,450)	p value
Age, M (SD)	81.64 (8.22)	79.45 (8.58)	81.29 (8.29)	<.01
Gender, n (%)				.02
Male	6,011 (44.1)	589 (47.3)	677 (46.7)
Female	7,617 (55.9)	656 (52.7)	773 (53.3)
Caregiver, n (%)				<.01
Spouse	4,459 (33.0)	270 (21.9)	401 (27.8)
Daughter	4,149 (30.7)	442 (35.9)	560 (38.8)
Son	2,267 (16.8)	158 (12.8)	194 (13.5)
Other	2,618 (19.4)	362 (29.4)	287 (19.9)
Diagnosis, n (%)				<.01
Cancer	5,125 (37.9)	524 (42.3)	511 (35.3)
Alzheimer's/dementia	2,217 (16.4)	178 (14.4)	282 (19.5)
COPD/CHF/heart disease	3,035 (22.6)	221 (17.8)	306 (21.2)
Other^a	3,127 (23.1)	317 (25.6)	347 (24.0)
Referral Source, n (%)				<.01
Primary physician	4,650 (35.8)	410 (34.2)	535 (38.6)
Hospital	5,074 (39.1)	586 (48.9)	674 (48.6)
Long-term care	2,504 (19.3)	150 (12.5)	157 (11.3)
Other^b	745 (5.7)	52 (4.3)	21 (1.5)
Primary Payer, n (%)				<.01
Private pay/PHI	844 (6.2)	85 (6.8)	136 (9.4)
Medicaid	24 (.2)	11 (.9)	66 (4.6)
Medicare	12,653 (93.6)	1,144 (92.3)	1,244 (86.0)
Length of Service (days), n (%)	88.20 (139.05)	90.29 (139.85)	98.84 (145.54)	.02
Median (days)	30	34	40
0-7 days, n (%)	3521 (25.8)	276 (22.2)	314 (21.7)	<.01
8-30 days	3376 (24.8)	318 (25.5)	331 (22.8)
31-180 days	4665 (34.2)	471 (37.8)	559 (38.6)
181 or more days	2066 (15.2)	180 (14.5)	246 (17.0)

Note. COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; PHI, private health insurance; LTC, long-term care.

Diagnoses were based on ICD-9 codes, and “other” diagnoses include HIV/AIDS, amyotrophic lateral sclerosis (ALS), and end stage multiple sclerosis.

“Other” referral sources include hospice and home health care.

Length of hospice care

The results from the Cox proportional hazard analyses adjusted for all study variables are presented in Table 2. In Model 1 (main effects model), higher risk of having a shorter length of hospice care was associated with being of the white race, being male, having a spouse caregiver, having a cancer diagnosis, being referred from hospital or LTC settings, and having Medicare or Medicaid as a primary source of payment. Model 2 included the interaction terms and was a better fit than Model 1 with a statistically significant change in −2 log likelihood (df=3, χ²=23.78, p<.01).

Table 2.

Adjusted Cox Proportional Hazard Models of Patient Characteristics and Length of Hospice Care (N=16,323)^a

	Model 1			Model 2
	HR	CI	p value	HR	CI	p value
Race/Ethnicity (ref=white)
African-American	0.80	0.74-0.87	<.01	0.63	0.52-0.75	<.01
Hispanic	0.79	0.74-0.86	<.01	0.63	0.63-0.85	<.01
Age	1.00	0.99-1.00	0.61	1.00	0.99-1.00	0.56
Female (ref=male)	0.82	0.74-0.86	<.01	0.80	0.77-0.84	<.01
Caregiver (ref=spouse)
Daughter	0.93	0.88-0.98	<.01	0.93	0.88-0.98	<.01
Son	0.93	0.87-0.99	0.03	0.93	0.87-0.10	0.04
Other	0.91	0.85-0.97	<.01	0.91	0.86-0.97	<.01
Diagnosis (ref=cancer)
Alzheimer's/dementia	0.69	0.64-0.73	<.01	0.68	0.64-0.73	<.01
COPD/CHF/heart disease	0.62	0.58-0.65	<.01	0.62	0.58-0.65	<.01
Other	0.91	0.87-0.97	<.01	0.91	0.86-0.96	<.01
Referral Source (ref=primary physician)
Hospital	2.40	2.29-2.51	<.01	2.38	2.26-2.50	<.01
Long-term care	1.32	1.24-1.41	<.01	1.28	1.20-1.38	<.01
Other	0.95	0.84-1.07	0.39	0.91	0.80-1.03	0.15
Primary Payer (ref=Private pay/PHI)
Medicaid	1.26	0.95-1.67	0.12	1.28	0.96-1.70	0.10
Medicare	1.60	1.44-1.79	<.01	1.60	1.44-1.78	<.01
Interactions
African-American×Female				1.21	1.03-1.42	0.02
African-American×LTC referral				1.34	1.02-1.77	0.04
Hispanic×LTC referral				1.38	1.07-1.78	0.01
-2 Log likelihood	162128.27			162104.49

Note.

A hazard ratio greater than 1.0 means a higher risk of having a shorter length of hospice care.

HR, hazard ratio; CI, confidence interval; Ref, reference group; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; PHI, private health insurance; LTC, long-term care.

Modifying effects

To interpret the significant interaction effects, each interaction term was compared to possible counterparts, which highlights the racial variations in certain predictors (Tables 2 and 3). Being an African-American female was associated with approximately the same risk of shorter length of hospice care as being an African-American male, thus cancelling effects (longer length of hospice care) of being female or African-American. African-Americans or Hispanics who were referred from LTC increased the risk of shorter length of hospice care considerably, that is, they were much less likely to stay for a longer length of time under hospice care compared to their respective counterparts referred by primary physicians.

Table 3.

Significant Interaction Effects of Race/Ethnicity with Patient Characteristics (N=16,323)^a

Interaction variable	HR	Reference
African-American female	0.76	White female
	0.97	African-American male
	0.61	White male
African-American referred from LTC	0.84	White referred from LTC
	1.72	African-American referred by primary physician
	1.08	White referred by primary physician
Hispanic referred from LTC	1.01	White referred from LTC
	1.77	Hispanic referred by primary physician
	1.29	White referred by primary physician

Note.

A hazard ratio greater than 1.0 means a higher risk of having a shorter length of hospice care.

HR, hazard ratio; LTC, long-term care.

Discussion

This study explored factors affecting the length of hospice care until death among older adults by considering how race/ethnicity moderates the relationships between patients' characteristics and the length of hospice care. The results of this study are consistent with prior studies that examine the relationship between patients' personal characteristics (gender, relationships with caregiver, diagnosis, referral source, and payment type.⁵ For example, patients with chronic degenerative diseases such as Alzheimer's disease or dementia may stay longer under hospice care than cancer patients with acute disease progressions.¹⁰

In the context of lower utilization of hospice care by minority patients, our finding that minority patients who enroll in hospice tend to have longer lengths of stay suggests that minorities who enroll may be prepared or referred earlier in the course of their illness than white patients. On the other hand, the finding may be interpreted to indicate fewer racial/ethnic minorities with the need for hospice services are accessing these services due to their lack of knowledge, mistrust of hospice care, and their cultural values^1,4–8 whereas whites may seek more aggressive care and exhaust other options until they are admitted to hospice care at a later stage in their illness, shortening their hospice stay. Further research should focus on testing if longer lengths of stay by minority patients are present nationally, and the reasons behind these differences.

This study has limitations. First, the data were limited to central Florida and to one hospice agency. However, the study included a large sample (N=16,323) and the proportions of race/ethnic groups were comparable to the national population.⁹ Second, administrative data lack detailed clinical information (e.g., patients' physical and mental/psychological conditions, and fulfillment of emotional/spiritual needs). Finally, although we examined three racial/ethnic groups, we could not explore the cultural differences within and between the groups. It is also important to note that race/ethnicity is socially constructed and reflects different cultural beliefs and meaning, and different access to valued resources in society.

Implications for practice

It is critical for health care professionals to understand that race/ethnicity is associated with hospice utilization and the length of hospice care. It is encouraging that rates of hospice utilizations among racial/ethnic minorities have increased over the past decade.⁹ Studies on the topic claim that when overall hospice utilization rates are higher, differences between whites and other groups are less obvious.¹⁸ Thus, concerted efforts should be made to increase hospice use for the general population and to reduce the racial/ethnic disparity in hospice use. Also, future studies should explore the structural and process-based barriers that prevent racial/ethnic minorities from utilizing hospice care systems and how these barriers could influence the length of hospice care and the quality of EOL care.

It is important for health care professionals to be aware of populations who are at risk of receiving late hospice referrals. There might be good reasons for patient to stay under hospice care only for a short amount of time. For example, if a patient is referred from a hospital after a sudden acute condition, then that patient might be referred for hospice care and die in a short amount of time. Also, health care professionals may provide nonhospice palliative care options, such as doctor consultations for pain and symptom management in LTC or hospital settings using interdisciplinary team-based palliative care approaches for patients hospitalized after an acute episode of illness.¹⁹ However, if health care professionals, especially in LTC settings, delay referrals, which may deprive a patient of necessary hospice care, then this problem should be addressed. Thus, collaborative efforts should be developed between hospice care and LTC providers to expedite referrals in order to provide timely hospice care.²⁰

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Cohen

. Racial/ethnic disparities in hospice care: A systematic review. Journal of Palliative Medicine June, 2008; 11,5:763–768.

Han

, Remsburg

, McAuley

, Keay

, Travis

. Length of hospice care among U.S. adults: 1992–2000. Inquiry Spring, 2007; 44,1:104–113.

Han

, Remsburg

, Iwashyna

. Differences in hospice use between black and white patients during the period 1992 through 2000. Medical Care, 2006; 44,8:731–737.

Greiner

, Perera

, Ahluwalia

. Hospice usage by minorities in the last year of life: results from the National Mortality Followback Survey. Journal of American Geriatrics Society July, 2003; 51,7:970–978.

Colón

, Lyke

. Comparison of hospice use and demographics among European Americans, African Americans, and Latinos. American Journal of Hospice and Palliative Medicine May/June, 2003; 20,3:182–190.

Haas

, Earle

, Orav

et al. Lower use of hospice by cancer patients who live in minority versus white areas. Journal of General Internal Medicine March, 2007; 22,3:396–399.

Carrion

. When do Latinos use hospice services? Studying the utilization of hospice services by Hispanics/Latinos. Social Work in Health Care March, 2010; 49,3:197–210.

Lepore

, Miller

, Gozalo

. Hospice use among urban black and white U.S. nursing home decedents in 2006. The Gerontologist, April 1, 2011; 51,2:251–260.

Organization NHaPC: Facts and figures: Hospice care in America. 2010. http://www.nhpco.org/files/public/Statistics_Research/Hospice_Facts_Figures_Oct-2010.pdf. 2011 January 13.

10.

Quill

. Is length of stay on hospice a critical quality of care indicator? Journal of Palliative Medicine, April 2007; 10,2:290–292.

11.

Teno

, Shu

, Casarett

, Spence

, Rhodes

, Connor

. Timing of referral to hospice and quality of care: Length of stay and bereaved family members' perceptions of the timing of hospice referral. Journal of Pain and Symptom Management, 2007; 34,2:120–125.

12.

McCarthy

, Burns

, Ngo-Metzger

, Davis

, Phillips

. Hospice use among Medicare managed care and fee-for-service patients dying with cancer. JAMA, May 7, 2003; 289,17:2238–2245.

13.

Lackan

, Ostir

, Freeman

, Mahnken

, Goodwin

. Decreasing variation in the use of hospice among older adults with breast, colorectal, lung, and prostate cancer. Medical Care, 2004; 42,2:116–122.

14.

Johnson

, Kuchibhatla

, Tulsky

. Racial differences in location before hospice enrollment and association with hospice length of stay. Journal of the American Geriatrics Society, 2011; 59,4:732–737.

15.

>SAS version 9.2[computer program]Cary, NC: SAS Institute, 2008.

16.

Allison

. Survival analysis using SAS: A practical guide. Cary, NC: SAS Institute, 2007.

17.

Long

. Regression models for categorical, limited dependent variables. Thousand Oaks, CA: Sage, 1997.

18.

Connor

, Elwert

, Spence

, Christakis

. Racial disparity in hospice use in the United States in 2002. Palliative Medicine, April 1, 2008; 22,3:205–213.

19.

Meier

, Isaacs

, Hughes

. Palliative care: Transforming the care of serious illness. San Francisco, CA: Jossey-Bass, 2010.

20.

Dobbs

, Hanson

, Zimmerman

, Williams

, Munn

. Hospice attitudes among assisted living and nursing home administrators, and the long-term care hospice attitudes scale. Journal of Palliative Medicine December, 2006; 9,6:1388–1400.