Racial Differences in the Growth of Noncancer Diagnoses Among Hospice Enrollees

Article Outline

• Abstract
• Introduction
• Methods
  • Data Source
  • Covariates
  • Outcome
  • Analysis
• Results
• Discussion
• References
• Copyright

Abstract 

Patients with noncancer life-limiting illnesses now represent over half of all hospice enrollees, compared to only one-quarter of enrollees in 1992. Whether this growth in enrollees with noncancer diagnoses has been similar for Caucasians and African Americans, a group historically underrepresented in hospice, has not been described. The purpose of this study was to compare rates of noncancer diagnoses among African American and Caucasian hospice enrollees. We analyzed data from the administrative database of VITAS Healthcare Corporation, including all African Americans and Caucasians discharged from hospice between January 1, 1999 and December 31, 2003. Of the 166,390 eligible discharges, 14.6% were African American, and 85.4% were Caucasian. Over the five-year study period, there was a similar increase in the crude proportion of enrollees with noncancer diagnoses in both groups, from 42% to 49.7% among African Americans and 57.9% to 64.3% among Caucasians. However, in multivariate analysis (adjusted for age, gender, admission level of care, payment source, Health Maintenance Organization (HMO) use, discharge year, and hospice program characteristics—size, location, presence of an inpatient unit), African Americans had 32% lower odds of having a noncancer (vs. cancer) diagnosis than Caucasians (odds ratio [OR] 0.68 [0.66, 0.77]). While numerous studies document lower rates of hospice use among African Americans than Caucasians, these findings suggest disease-specific differences in patterns of hospice use, with greater disparities in hospice use among African Americans with noncancer diagnoses than those with cancer diagnoses. Targeted efforts to increase hospice use among African Americans with noncancer diagnoses may be important in reducing racial disparities in overall hospice use and improving the quality of care for dying African Americans.

Key Words: Hospice, African Americans, race, end-of-life care, noncancer diagnoses

Back to Article Outline

Introduction 

Over the last two decades, the number of hospice enrollees has increased by more than sixfold, from 158,000 enrollees in 1985 to over one million in 2004.¹ In addition to this dramatic growth in service utilization, hospices are now serving a more diverse population of patients with a broad spectrum of life-limiting illnesses. Although hospices have traditionally cared for cancer patients, increasing numbers of patients with noncancer diagnoses are accessing services. In 1992, 24% of hospice enrollees had noncancer diagnoses, compared to 54% in 2004.¹, ²

Along with this shift in diagnoses, the demographic profile of hospice enrollees has changed. Historically underrepresented in hospice, a growing number of African Americans are now accessing services. In the 1970s, hospice enrollees were almost entirely Caucasian.³ In 2004, minorities represented 22.7% of hospice patients, with African Americans making up the largest minority group at 8.1%.¹ From 1998 to 2002, the proportion of African American Medicare decedents who used hospice increased from 12% to 18%.⁴

Despite these positive changes in utilization, African Americans continue to use hospice at significantly lower rates than Caucasians.⁵, ⁶, ⁷, ⁸, ⁹ In the last National Mortality Follow-back Survey, the odds of hospice enrollment for blacks was 34% lower than that of whites (OR 0.66; 95% confidence interval [CI] 0.55, 0.78).⁵ Additionally, among all Medicare decedents in 2002, 26% of Caucasians used hospice services, compared to only 18% of African Americans.⁴

One way to explore this ongoing racial disparity in end-of-life care is to better understand how hospice use by different ethnic groups varies among those with different life-limiting illnesses. For example, no studies have examined whether racial differences in hospice use exist to the same degree across both cancer and noncancer diagnoses or whether the growth in noncancer diagnoses has been similar among African American and Caucasian hospice enrollees. Such analyses could help to delineate disease-specific racial disparities in hospice use and ultimately inform targeted efforts to increase hospice use among those subgroups of African Americans who are least likely to access services.

Therefore, the goal of this analysis was to explore racial differences in admission diagnoses among African-American and Caucasian hospice enrollees. Because of the extraordinary growth in noncancer diagnoses among hospice enrollees, we focus on racial differences in noncancer diagnoses in a large sample of hospice patients.

Back to Article Outline

Methods 

Data Source 

Data for this analysis were obtained from VITAS, the largest hospice provider in the United States. After obtaining the required approval from the hospice provider and the Duke University Health System Institutional Review Board, patient-level files were used to create the study database. The data were abstracted from the computerized central administrative and clinical database of the hospice provider. Preprogrammed parameters, control processes for changing data entry, and routine monitoring of data entry for accuracy and timeliness by staff at each program provide ongoing system-wide data quality checks.

This analysis included all Caucasian and African American hospice enrollees who were discharged from VITAS hospice programs in eight states—California, Florida, Illinois, New Jersey, Ohio, Pennsylvania, Texas, and Wisconsin—between January 1, 1999 and December 31, 2003. A total of 166,390 enrollees were eligible for inclusion in this analysis.

Covariates 

Variables for the final model were theoretically derived based on their relevance as potential confounders of the relationship between race and hospice admission diagnoses. Selection was based on the potential relationship of the variable to the prevalence of cancer or noncancer diagnoses (i.e., gender, age, residence in a nursing home, hospice program location), or in the case of hospice program characteristics, based on program resources that may differentially affect enrollment of those with cancer or noncancer diagnoses (i.e., presence of inpatient hospice unit, program size—large programs may offer services and resources not offered by smaller programs).

The final model included demographic variables, hospice use variables, and hospice program variables. In addition to race, demographic variables included age (≤65, >65) and gender. Hospice use variables included admission level of care (routine home care in a private residence, routine home care in a nursing home, continuous care, and inpatient care in a hospital-based hospice unit operated by the hospice program or in a contract bed in an area hospital); payment source (Medicare, Medicaid, charity, other); and health maintenance organization (HMO) (yes, no). Hospice program variables included hospice location (California, Florida, Illinois, New Jersey, Ohio, Pennsylvania, Texas, Wisconsin); inpatient unit (yes, no); program size; and year of discharge (1999, 2000, 2001, 2002, 2003). The inpatient unit variable referred to whether the subject was enrolled in a hospice program that operated its own hospital-based hospice unit in addition to providing services in private residences and nursing homes. The program size variable was based on the proportion of the total sample of enrollees served by a given program. The sample included enrollees served by 27 programs. Programs were arranged from largest to smallest based on the proportion of the total sample of enrollees served by the program. We dichotomized the programs at the point that constituted approximately 50% of the total sample used for this analysis. Using this method, the six largest programs that together served a total of 51.5% of enrollees in the sample were classified as large programs, whereas the other 21 programs were classified as small programs.

Outcome 

The outcome variable for this analysis was the enrollees' primary admission diagnosis. All primary admission diagnoses were included and categorized as cancer or noncancer.

Analysis 

Chi-squared analyses were used to compare African American and Caucasian hospice enrollees across categorical demographic, hospice use, and hospice program variables, and a nonparametric median test was used to examine racial differences in length of stay. The Cochran Armitage Trend Test was used to examine trends in the growth of noncancer diagnoses during the five-year study period for both African American and Caucasian hospice enrollees. Differences were considered statistically significant at P<0.05.

To determine if race (African American vs. Caucasian) was an independent predictor of admission diagnosis (noncancer vs. cancer), the analysis included a logistic regression model. As noted above, the final model was theoretically derived and included those variables considered important predictors of hospice admission diagnoses (race, age, gender, payment source, HMO, admission level of care, hospice location, inpatient unit, hospice size, and year of discharge). Additionally, we tested for race×year interaction. Because the interaction was not statistically significant (P>0.05), we excluded it from our final model. The final model included only main effects. All analyses were performed using SAS Version 9.3.

Back to Article Outline

Results 

Between 1999 and 2003, a total of 166,390 African-American and Caucasian enrollees were discharged from the VITAS hospice programs included in our analysis. African Americans represented 14.6% of the total sample. Fig. 1 displays the breakdown of admission diagnoses for the total sample and for each racial group. A significantly smaller proportion of African American than Caucasian enrollees had noncancer admission diagnoses (46.8% vs. 61.6%).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  Admission diagnoses by race for total sample (P<0.001: Chi-squared test of difference in proportions by race).

The proportion of enrollees with noncancer diagnoses by race for each year of the study period is shown in Fig. 2. Throughout the study, African American enrollees were less likely than Caucasian enrollees to have noncancer admission diagnoses. However, with each subsequent year of the study, the proportion of patients with noncancer diagnoses increased significantly in each racial group (trend for each racial group: P<0.0001). This growth was similar among African-American and Caucasian enrollees. As a result, the percentage point difference between the two groups in the proportion of enrollees with noncancer diagnoses was relatively stable during the five-year study period (trend for difference: P=0.09).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2 

  Noncancer diagnoses by race, 1999–2003 (P<0.0001: Cochran Armitage Trend Test for African Americans; P<0.0001: Cochran Armitage Trend Test for Caucasians; P = 0.09: Cochran Armitage Trend Test for difference between racial groups).

Sample characteristics by race and diagnosis are listed in Table 1. Compared to Caucasians, a greater proportion of African Americans in both diagnostic categories were aged ≤65, used charity or Medicaid as a payment source, and were admitted at an inpatient level of care. African American enrollees with both cancer and noncancer diagnoses had significantly longer median lengths of stay than Caucasian enrollees.

Table 1. Sample Characteristics by Race and Diagnoses

	Noncancer		Cancer
	African Americans	Caucasians	African Americans	Caucasians
	(n=11,367) 11.49%	(n=87,541) 88.51%	(n=12,917) 19.14%	(n=54,565) 80.86%
Agea
Age ≤65	22.8%	5.7%	35.1%	23.3%
Age >65	77.2%	94.3%	64.9%	76.7%
Genderb
Female	61.3%	63.6%	51.4%	52.3%
Marital statusa
Married	22.2%	29.8%	32.9%	47.2%
Payment sourcea
Medicare	80.3%	94.4%	66.4%	77.0%
Medicaid	13.5%	1.8%	16.1%	4.2%
Charity	1.6%	0.6%	3.0%	1.5%
Other	4.6%	3.3%	14.6%	17.2%
HMOa
Yes	34.0%	31.0%	42.0%	44.8%
Admissiona
Level of care
Routine home care in private residence	26.7%	26.7%	57.4%	57.5%
Routine home care in a nursing home	29.9%	35.6%	11.4%	12.5%
Continuous care	3.0%	4.7%	3.0%	4.3%
Inpatient care	40.4%	33.1%	28.2%	25.7%
Median length of stay in daysc (range)	12 (1–1,930)	10 (1–3,662)	16 (1–1,932)	14 (1–1,911)
Admitted to program with inpatient unita
Yes	86.3%	71.7%	82.3%	68.8%
Hospice program sized, e
Large programs	54.5%	54.2%	40.2%	49.3%
Location of programa
Florida	36.6%	34.0%	25.4%	30.3%
Chicago	18.9%	12.3%	26.9%	13.2%
Ohio	7.6%	7.5%	4.4%	5.1%
California	6.2%	18.9%	9.0%	21.4%
Texas	19.9%	20.9%	20.2%	21.8%
New Jersey	0.1%	0.2%	0.1%	0.2%
Philadelphia	9.9%	3.5%	12.5%	5.5%
Wisconsin	1.0%	2.6%	1.6%	2.6%
Year of dischargea
1999	14.7%	16.7%	17.9%	19.4%
2000	18.1%	19.3%	19.1%	20.3%
2001	20.4%	20.6%	20.2%	20.4%
2002	22.6%	21.1%	21.4%	20.0%
2003	24.2%	22.3%	21.5%	19.9%

Percentages may not add up to 100% because of rounding.

The results of the multivariate analysis are in Table 2. Compared to Caucasians, African Americans had 32% lower odds of having a noncancer (vs. cancer) diagnosis (OR 0.68 [0.66, 0.77]). Other variables associated with significantly lower odds of noncancer (vs. cancer) diagnoses included a payment source of “other coverage” (vs. Medicare); enrollment in an HMO; admission level of care of routine home care in a private residence or continuous care (vs. inpatient care); admission to a hospice program with an inpatient unit; admission to a “large program.” Variables associated with significantly higher odds of noncancer (vs. cancer) diagnoses included age >65, female gender, Medicaid (vs. Medicare) as payment source, and admission level of care of routine home care in a nursing home (vs. inpatient care). Additionally, the odds of noncancer diagnoses increased in each subsequent year of the study period.

Table 2. Multivariate Analysis of Association of Demographic, Hospice Use, and Hospice Program Variables with Admission Diagnoses

	Noncancer vs. Cancer Diagnoses
	Odds Ratio	95% Confidence Interval
Race
Caucasians (reference)
African Americans	0.68	[0.66, 0.70]
Age
≤65 (reference)
>65	1.80	[1.72, 1.88]
Gender
Male (reference)
Female	1.35	[1.32, 1.38]
Payment source
Medicare (reference)
Medicaid	1.53	[1.43, 1.63]
Charity	0.96	[0.86, 1.07]
Other coverage	0.51	[0.48, 0.55]
HMO
No (reference)
Yes	0.92	[0.90, 0.94]
Admission level of care
Inpatient (reference)
Routine home care (private residence)	0.37	[0.36, 0.38]
Routine home care (nursing home)	1.97	[1.90, 2.03]
Continuous care	0.76	[0.72, 0.81]
Admitted to program with inpatient hospice unit
No (reference)
Yes	0.92	[0.97, 0.98]
Hospice program sizea
Small (reference)
Large	0.91	[0.87, 0.96]
Location
Florida (reference)
Chicago	0.58	[0.54, 0.61]
Ohio	0.82	[0.78, 0.86]
California	0.66	[0.60, 0.73]
Texas	0.68	[0.66, 0.71]
New Jersey	0.69	[0.52, 0.92]
Philadelphia	0.48	[0.45, 0.52]
Wisconsin	0.53	[0.48, 0.60]
Year of discharge
1999 (reference)
2000	1.12	[1.08, 1.16]
2001	1.22	[1.18, 1.27]
2002	1.29	[1.25, 1.34]
2003	1.40	[1.35, 1.45]

Back to Article Outline

Discussion 

This study examined racial differences in hospice admission diagnoses. In this analysis of 166,390 hospice patients, the proportion of enrollees with noncancer diagnoses increased in both racial groups over the five-year study period. However, the proportion of African Americans with noncancer diagnoses remained lower than the proportion of Caucasians with noncancer diagnoses in each year of the study. Not only were African Americans unable to “catch up” to Caucasians, but the percentage point difference between the groups in the proportion of enrollees with noncancer diagnoses was relatively stable throughout the five-year period. Even after controlling for demographic, hospice use, and hospice program variables that may affect admission diagnoses, African Americans were still significantly less likely than Caucasians to have noncancer admission diagnoses.

A number of studies document lower rates of hospice use among African Americans compared to Caucasians.¹, ⁴, ⁵, ⁶, ⁷, ⁸, ⁹ The findings from this analysis indicate disease-specific differences in patterns of hospice use and suggest greater disparities in hospice use among African Americans with noncancer diagnoses than those with cancer diagnoses. Targeted efforts to close the “noncancer gap” by increasing hospice use among African Americans with noncancer diagnoses may reduce racial disparities in overall hospice use. For example, in our sample, 61.6% of Caucasians had noncancer diagnoses compared to only 46.8% of African Americans. If the number of African Americans with cancer diagnoses remained constant, an additional 9,357 African Americans with noncancer diagnoses would need to enroll in hospice to increase the proportion of African Americans with noncancer diagnoses to 61.6%. This represents an 82.3% increase in the number of African Americans with noncancer diagnoses and a 38.5% increase in the total number of African Americans in the sample.

Many of the factors thought to explain the lower use of hospice services by African Americans in general may have an even greater impact on hospice use among those with noncancer diagnoses. For example, a number of studies cite lack of knowledge of hospice services as a barrier to hospice use by African Americans.¹⁰, ¹¹, ¹², ¹³ One myth about hospice care is that it is a service for cancer patients only. Therefore, even African Americans with some knowledge of hospice may not know about the availability of services for both cancer and noncancer diagnoses.

Greater preferences for the use of life-sustaining therapies and denial of death at the end-of-life are other reasons that may explain the lower use of hospice services by African Americans.¹⁰, ¹⁴, ¹⁵, ¹⁶, ¹⁷, ¹⁸ These factors may be even more important in the face of noncancer life-limiting illnesses. For example, heart disease, dementia, chronic obstructive pulmonary disease (COPD), and other noncancer diagnoses may not be viewed as terminal illnesses. Unlike cancer, these diagnoses have a less predictable dying trajectory with a longer period of functional impairment prior to death, making prognostication more difficult.¹⁹, ²⁰, ²¹, ²² Additionally, with many noncancer diagnoses, acute exacerbations of illness occur intermittently with periods of recovery after the use of life-sustaining therapies. In the absence of a “recognized” life-limiting illness, African Americans with noncancer diagnoses may be even less likely to access hospice services.

This analysis has several limitations. First, the study sample is drawn from a single national hospice provider and includes enrollees admitted to 27 hospice programs in eight states. Although these data are from the largest hospice provider in the United States, our results may not be generalizable to other hospices in other states throughout the United States. For example, in this sample, both the proportion of total enrollees with noncancer diagnoses (59.4% vs. 54%) and the proportion of African Americans (14.6% vs. 8.1%) exceeded the data reported for hospice enrollees nationwide. This may, in part, reflect targeted marketing efforts by the hospice provider and an increased market-share among African Americans and those with noncancer diagnoses. The greater representation of both African American enrollees and enrollees with noncancer diagnoses in our sample may have resulted in a smaller difference in the proportion of noncancer diagnoses between the two races than the difference that exists in the general population of hospice patients. Future studies should include a nationally representative sample of hospice enrollees and hospice providers.

Our sample also differs from national data in the distribution of noncancer diagnoses. Nationally, heart disease is the most common noncancer diagnosis, followed by dementia.¹ In our sample, dementia was the most common noncancer diagnosis, followed by heart disease. Dementia represented 23.3% of noncancer diagnoses in our sample, compared to 16.5% nationally, and heart disease represented 20.5% of noncancer diagnoses, compared to 22.6% of noncancer diagnoses nationally.

Another limitation of this analysis is the absence of population-based data. We have no information on the proportion of African American decedents with noncancer diagnoses in the hospice service areas. Because African Americans have higher mortality rates than Caucasians from heart disease and cerebrovascular disease, the lower proportion of African American than Caucasian enrollees with noncancer diagnoses in our sample is unlikely due to a smaller proportion of African American decedents with noncancer diagnoses in the hospice service areas.

The findings of this analysis suggest both positive changes in hospice use among African Americans and an opportunity for further efforts to decrease racial disparities in end-of-life care. First, there was a similar growth in the crude proportion of African Americans and Caucasians with noncancer diagnoses in our sample over the five-year study period. This suggests an increase in access to and acceptance of hospice among those with noncancer diagnoses in both racial groups. The persistent difference between the two groups in the proportion of enrollees with noncancer diagnoses suggests a disease-specific focus for ongoing efforts to increase hospice use among African Americans. Future research identifying barriers to hospice use by African Americans with noncancer diagnoses could inform the development of targeted interventions to increase hospice use among African Americans with noncancer diagnoses. Such interventions, if successful, could significantly reduce overall disparities in hospice use and serve as a model for improving end-of-life care for all racial and ethnic groups.

Back to Article Outline

References 

1. National Hospice and Palliative Care Organization . NHPCO's 2004 facts and figures. Available from http://www.nhpco.org/files/public/Facts_Figures_for2004data.pdfAccessed February, 2006
   • View In Article
2. Medicare Payment Advisory Commission . Medicare beneficiaries' access to hospice. Figure 1-0: new hospice patients by diagnosis 1992–2000. May 2002; p. 4. Available from http://www.medpac.gov/publications/congressional_reports/may2002_HospiceAccess.pdfAccessed February, 2006
   • View In Article
3. Schumacher JD. Hospice care in the United States. Business briefing: long-term healthcare. 2004:14–15. Available from http://www.touchbriefings.com/pdf/886/lth041_schumacher.pdfAccessed February 20, 2006
   • View In Article
4. Medicare Payment Advisory Committee . New approaches in Medicare. Chapter 6: hospice care in Medicare: recent trends and a review of the issues. Figure 6-3: hospice use has increased among all races. June 2004: p. 143. Available from http://www.medpac.gov/publications/congressional_reports/June04_ch6.pdfAccessed February 20, 2006
   • View In Article
5. Greiner KA, Perera S, Ahluwalia JS. Hospice usage by minorities in the last year of life: results from the National Mortality Followback Survey. J Am Geriatr Soc. 2003;51:970–978
   • View In Article
   • MEDLINE
   • CrossRef
6. Virnig BA, Kind S, McBean M, Fisher E. Geographic variation in hospice use prior to death. J Am Geriatr Soc. 2000;48:1117–1125
   • View In Article
   • MEDLINE
7. Virnig BA, McBean AM, Kind S, Dholakia R. Hospice use before death. Variability across cancer diagnoses. Med Care. 2002;40:73–78
   • View In Article
   • MEDLINE
   • CrossRef
8. Colon M, Lyke J. Comparison of hospice use and demographics among European Americans, African Americans, and Latinos. Am J Hosp Palliat Care. 2003;20:182–190
   • View In Article
   • MEDLINE
9. Lackan NA, Ostir GV, Freeman JL, Mahnken JD, Goodwin JS. Decreasing variation in the use of hospice among older adults with breast, colorectal, lung, and prostate cancer. Med Care. 2004;42:116–122
   • View In Article
   • MEDLINE
   • CrossRef
10. Reese DJ, Ahern RE, Nair S, Warren C. Hospice access and use by African Americans: addressing cultural and institutional barriers through participatory action research. Soc Work. 1999;44(6):549–559
    • View In Article
    • MEDLINE
    • CrossRef
11. Born W, Greiner KA, Eldonna S, Butler J, Ahluwalia JS. Knowledge, attitudes, and beliefs about end-of-life care among inner-city African Americans and Latinos. J Palliat Med. 2004;7:247–256
    • View In Article
    • MEDLINE
    • CrossRef
12. Burrs FA. The African American experience: Breaking the barriers to hospices. Hosp J. 1995;10:15–18
    • View In Article
    • MEDLINE
    • CrossRef
13. Gordon AK. Hospice and minorities: a national study of organizational access and practice. Hosp J. 1996;11:49–70
    • View In Article
    • MEDLINE
    • CrossRef
14. Crawley LM, Payne R, Bolden J, et al. Palliative and end-of-life care in the African American community. JAMA. 2000;284(19):2518–2521
    • View In Article
    • MEDLINE
    • CrossRef
15. Neubauer BJ, Hamilton CL. Racial differences in attitudes toward hospice care. Hosp J. 1990;6(1):37–48
    • View In Article
    • MEDLINE
    • CrossRef
16. Hopp FP, Duffy SA. Racial variations in end-of-life care. J Am Geriatr Soc. 2000;48:658–663
    • View In Article
    • MEDLINE
17. Caralis PV, Davis B, Wright K, Marcial E. The influence of ethnicity and race on attitudes toward advance directives, life-prolonging treatments, and euthanasia. J Clin Ethics. 1993;4:155–165
    • View In Article
    • MEDLINE
18. Blackhall LJ, Frank G, Murphy ST, et al. Ethnicity and attitudes towards life-sustaining technology. Soc Sci Med. 1999;48:1779–1789
    • View In Article
    • MEDLINE
    • CrossRef
19. Teno JM, Weitzen S, Fennell ML, Mor V. Dying trajectory in the last year of life: does cancer trajectory fit other diseases?. J Am Geriatr Soc. 2001;4(4):457–464
    • View In Article
20. Fox E, Landrum-McNiff K, Zhong Z, et al. Evaluation of prognostic criteria for determining hospice eligibility in patients with advanced lung, heart, or liver disease. SUPPORT Investigators. Study to understand prognoses and preferences for outcomes and risks of treatment. JAMA. 1999;282(17):1638–1645
    • View In Article
    • MEDLINE
    • CrossRef
21. Luchins DJ, Hanrahan P, Murphy K. Criteria for enrolling dementia patients in hospice. J Am Geriatr Soc. 1997;49(9):1054–1059
    • View In Article
22. In:  Emanuel LL,  von Gunten CF,  Ferris FD editor. The education in palliative and end-of-life care curriculum. The EPEC Project, 1999. 2003;Curriculum available from www.epec.netAccessed February 24, 2006
    • View In Article