OBJECTIVE: To identify characteristics of nursing homes and residents associated with particularly long or short hospice stays. DESIGN/ SETTING: Observational study using administrative data on resident characteristics and hospice utilization from a large regional hospice linked with publicly available data on nursing home characteristics. PARTICIPANTS: A total of 13,479 residents who enrolled in hospice during 2001-2008. MEASUREMENTS: Logistic regression models of the probability of a long (>180 days) or very short (<or=3 days) stay, adjusting for nursing home characteristics, a measure of nursing home quality developed using Minimum Data Set Quality Indicator/Quality Measures data, and resident characteristics. RESULTS: Nursing home characteristics were not statistically significant predictors of long stays. The probability of a short stay increased with the facility's nurse staffing ratio and decreased with the share of residents covered by Medicaid. Men (relative to women) and blacks (relative to whites) were less likely to have a long stay and more likely to have a short stay, while those 70 years or younger (relative to those 81-90) and residents with Alzheimer's disease/dementia were more likely to have long stays and less likely to have short stays. Fourteen percent of hospice users were discharged before death because they failed to meet Medicare hospice eligibility criteria, and these residents had longer lengths of stay, on average. CONCLUSION: Few facility characteristics were associated with very long or very short hospice stays. However, high rates of discharge before death that may reflect a less predictable life trajectory of nursing home residents suggests that further evaluation of the hospice benefit for nursing home residents may be needed.
OBJECTIVE: To identify characteristics of nursing homes and residents associated with particularly long or short hospice stays. DESIGN/ SETTING: Observational study using administrative data on resident characteristics and hospice utilization from a large regional hospice linked with publicly available data on nursing home characteristics. PARTICIPANTS: A total of 13,479 residents who enrolled in hospice during 2001-2008. MEASUREMENTS: Logistic regression models of the probability of a long (>180 days) or very short (<or=3 days) stay, adjusting for nursing home characteristics, a measure of nursing home quality developed using Minimum Data Set Quality Indicator/Quality Measures data, and resident characteristics. RESULTS: Nursing home characteristics were not statistically significant predictors of long stays. The probability of a short stay increased with the facility's nurse staffing ratio and decreased with the share of residents covered by Medicaid. Men (relative to women) and blacks (relative to whites) were less likely to have a long stay and more likely to have a short stay, while those 70 years or younger (relative to those 81-90) and residents with Alzheimer's disease/dementia were more likely to have long stays and less likely to have short stays. Fourteen percent of hospice users were discharged before death because they failed to meet Medicare hospice eligibility criteria, and these residents had longer lengths of stay, on average. CONCLUSION: Few facility characteristics were associated with very long or very short hospice stays. However, high rates of discharge before death that may reflect a less predictable life trajectory of nursing home residents suggests that further evaluation of the hospice benefit for nursing home residents may be needed.
Authors: Katherine Freund; Michelle T Weckmann; David J Casarett; Kristi Swanson; Mary Kay Brooks; Ann Broderick Journal: J Hosp Med Date: 2011-11-15 Impact factor: 2.960