Objective
This study examined the impact of hospice enrollment on the probabilities of hospital and nursing home admissions among a sample of frail dual-eligible assisted living (AL) residents.
Design
The study used a retrospective cohort design. We estimated bivariate probit models with 2 binary outcome variables: any hospital admissions and any nursing home admissions after assisted living enrollment.
Setting
A total of 328 licensed AL communities accepting Medicaid waivers in Florida.
Participants
We identified all newly admitted dual-eligible AL residents in Florida between January and June of 2003 who had complete state assessment data (n?= 658) and followed them for 6 to 12 months.
Measurements
Using the Andersen behavioral model, predisposing (age, gender, race), enabling (marital status, available caregiver, hospice use), and need (ADL/IADL, comorbidity conditions, and incontinence) characteristics were included as predictors of 2 binary outcomes (hospital and nursing home admission). Demographics, functional status, and caregiver availability were obtained from the state client assessment database. Data on diagnosis and hospital, nursing home, and hospice use were obtained from Medicare and Medicaid claims. Death dates were obtained from the state vital statistics death certificate data.
Results
The mean age of the study sample was 81.5 years. Three-fourths were female and 63 % were White. The average resident had a combined ADL/IADL dependency score of 11.49. Fifty-eight percent of the sample had dementia. During the average 8.9-month follow-up period, 6.8 % were enrolled in hospice and 10.2 % died. Approximately 33 % of the sample had been admitted into a hospital and 20 % had been admitted into a nursing home. Bivariate probit models simultaneously predicting the likelihood of hospital and nursing home admissions showed that hospice enrollment was associated with lower likelihood of hospital (OR?= 0.24, P < .01) and nursing home admissions (OR?= 0.56, P < .05). Significant predictors of hospital admissions included higher Charlson Comorbidity Index score and incontinence. Predictors of nursing home admissions included higher Charlson Comorbidity Index score, the absence of available informal caregiver, and incontinence.
Conclusions
Hospice enrollment was associated with a lower likelihood of hospital and nursing home admissions, and, thus, may have allowed AL residents in need of palliative care to remain in the AL community. AL providers should support and facilitate hospice care among older frail dual-eligible AL residents. More research is needed to examine the impact of hospice care on resident quality of life and total health care expenditures among AL residents.
