DESIGN: Cross-sectional study.

PARTICIPANTS: A random sample of aged Medicare beneficiaries who died 1996 to 1999, N=241,047.

MEASURES: We estimate differences in mean Medicare expenditures by year before death and by age, gender, race, and area income, adjusting for comorbidities and Medicaid enrollment.

RESULTS: Expenditures for blacks are lower in the second and third years before death and are not significantly different from whites in the last year of life (LYOL) (y3=70%, P<.0001; y2=82%, P<.0001; LYOL=119%, P=.098). Differences in expenditures between decedents with area incomes over 35,000 compared to under35,000 compared to under 20,000 attenuate by the LYOL (y3=116%, P<.0001; y2=107%, P<.0001; LYOL=96%, P<.0001). Expenditure patterns for women versus men vary by age. Among the younger cohorts (68 to 74 and 75 to 79), expenditures are higher for women in all 3 years before death. This difference attenuates among older cohorts; in the oldest cohort (90+), expenditures for men exceed those for women by 11% in the LYOL (P<.0001). Older beneficiaries have higher expenditures in the second and third years before death but lower expenditures in the LYOL. On average, the youngest cohort expended $8,017 more in the LYOL relative to the oldest cohort, whereas in the third years before death, the oldest cohort’s expenditures were $8,017 more in the LYOL relative to the oldest cohort, whereas in the third years before death, the oldest cohort’s expenditures were 5,270 more than those for the youngest cohort (P<.0001).

CONCLUSIONS: Age-associated differences in aggregate Medicare payments for end-of-life care are more substantial than other differences. The fact that other differences attenuate in the LYOL may reflect having overcome barriers to health care, or reflect an effective ceiling on the opportunities to provide services for persons with overwhelming illness.