Get access

Hospice Usage by Minorities in the Last Year of Life: Results from the National Mortality Followback Survey


Address correspondence to K. Allen Greiner, MD, MPH, Assistant Professor of Family Medicine and Preventive Medicine, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City, KS 66160. E-mail:


OBJECTIVES: To examine racial/ethnic variations in rates of hospice use in a national cohort and to identify individual characteristics associated with hospice use.

DESIGN: Secondary analysis of the 1993 National Mortality Followback Survey (NMFS), a nationally obtained sample using death certificates and interviews with relatives (proxy respondents) to provide mortality, social, and economic data and information about healthcare utilization in the last year of life for 23,000 deceased individuals.

SETTING: Hospice care.

PARTICIPANTS: Individuals aged 15 and older who died in 1993. Subjects were included in this analysis if they died of nontraumatic causes (N = 11,291).

MEASUREMENTS: Hospice use was dichotomized by proxy responses indicating use or nonuse of home or inpatient hospice services. The percentage of individuals using hospice services in the last year of life was calculated.

RESULTS: Unadjusted bivariate results found that African Americans were less likely to use hospice than whites (odds ratio (OR) = 0.59; P < .001) and that those without a living will (LW) (OR = 0.23; P < .001) and without a cancer diagnosis (OR = 0.28; P < .001) were less likely to use hospice. The negative relationship between African Americans and hospice use was unaffected when controlled for sex, education, marital status, existence of a LW, income, and access to health care. Logistic models revealed that presence of a LW diminished the negative relationship between African Americans and hospice use, but the latter remained significant (OR = 0.83; P = .033). A subanalysis of subjects aged 55 and older showed a significant interaction between access to care and race/ethnicity with respect to hospice use (P = .044). Inclusion of income in this multivariable logistic model attenuated the relationship between African-American race/ethnicity and hospice use (OR = 0.77), and the difference between whites and African Americans became only marginally statistically significant (P = .060).

CONCLUSION: In the 1993 NMFS, hospice use was negatively associated with African-American race/ethnicity independent of income and access to healthcare. The relationship is not independent of age, insurance type, or history of stroke. For subjects aged 55 and older, access to healthcare may be an important confounder of the negative relationship between African-American race/ethnicity and hospice use. Consistent with previous studies, this analysis found that African Americans were less likely to use LWs than whites. The reduced importance of African-American race/ethnicity on hospice use with the inclusion of presence of a LW in logistic models suggests that similar cultural processes may shape differences between African Americans and whites in advance care planning and hospice use.