Smith W R, Kellerman A, Brown J S
Division of General Internal Medicine, University of Tennessee, Memphis, USA.
J Am Geriatr Soc. 1995 Sep;43(9):1052-7. doi: 10.1111/j.1532-5415.1995.tb05573.x.
To determine whether nursing homes transferred their terminal care patients to one public hospital, and if so, the impact of these transfers on nursing home and hospital mortality rates.
A retrospective cohort study using both Medicare claims and hospital data to construct a mortality prediction model, as well as a cross-sectional descriptive comparison of state nursing home mortality rates.
A public hospital, one of 32 hospitals designated as overall mortality outliers for each of the first three Medicare mortality reports.
Patients (n = 1235) included in the index hospital's 1988 Medicare mortality report; nursing homes (n = 289) included in state data on 1988 nursing home mortality rates.
Patients transferred from nursing homes to the index hospital were nearly twice as likely to die as the hospital's other Medicare patients in univariate (P = .0001) as well as multivariate analysis (OR = 1.68, 95% CI = 1.17 - 2.40). Terminal care admissions accounted for 33.9% of deaths from nursing homes but only 16.2% of other deaths (P = .009). The 1988 death/discharge rate at one of two nursing homes responsible for nearly all transfers was 15.2%, less than half the state average. Yet it contributed 59.3% of nursing home deaths and 22.7% of terminal care patients to the hospital. The hospital's 20.0% 1988 Medicare mortality rate was just above the expected range.
Two nursing homes transferred terminal care to a single hospital. This resulted in a lower-than-average mortality rate for one of the nursing homes and a higher-than-expected Medicare mortality for the hospital. Without these end-of-life transfers, the hospital's Medicare mortality rate would have been within the expected range.
