Hospice terminal care is now used by 10% to 15% of elderly Americans at variable points before their deaths.
By examining the duration of patient survival after enrollment in hospice care, we sought to identify individual and market factors associated with the timing of hospice use.
We linked Medicare claims, census information, and Area Resource File data to form a national cohort of 151,410 hospice patients admitted in 1993 and followed up until late 1996. We examined this cohort with Cox regression and other means.
The primary outcome measure was survival after hospice enrollment.
The patients had a mean±SD age of 79.0 ± 7.4 years; 10.2% were nonwhite; 51.4% were female; and 71.3% had cancer. Median survival after hospice enrollment was 30 days (interquartile range, 10-86 days). After adjustment for measured patient, provider, and market factors, several variables were associated with relatively earlier hospice enrollment, farther from death. Compared with complementary groups, nonwhites were enrolled in hospice 4 days earlier; women, 5 days earlier; older people, 1 day earlier; and those with substance abuse, psychiatric disease, or dementia, each 3 days earlier. After adjustment, income and education were not associated with the timing of enrollment. Patients residing in markets with more hospital beds, greater hospice capacity, or a higher proportion of generalists were enrolled earlier.
Even after adjustment for certain clinical attributes, individual social factors and local market factors were associated with survival after hospice enrollment. Certain socially disadvantaged groups were enrolled earlier, as were those residing in areas with more medical institutions. The decision to enroll patients in hospice may depend on both non-clinical and clinical factors.
*From the Departments of Medicine and Sociology, and the Center on Aging and Population Research Center, University of Chicago, Chicago, Illinois.
†From the School of Medicine, Harris Graduate School of Public Policy Studies, and the Center on Aging and Population Research Center, University of Chicago, Chicago, Illinois.
Supported by a grant from the Alzheimer's Association (TRG-95-033) (N.A.C.); by the Soros Faculty Scholars Program of the Project on Death in America (N.A.C.); by a Geriatric Academic Program Award, National Institute on Aging, National Institutes of Health (5 K12 AG00488-06) (N.A.C.); and by a Medical Scientist National Research Service Award from the National Institutes of Health (5 T32 GM07281) (T.J.I.).
Address correspondence to: Nicholas A. Christakis, MD, Section of General Internal Medicine, University of Chicago Medical Center, 5841 S Maryland Ave, MC 2007, Chicago, IL 60637.
Received August 23, 1999; initial review completed October 19, 1999; accepted December 30, 1999.