Predicting place of discharge from hospital for patients with a stroke or hip fracture on admission.

OBJECTIVES: To determine the predictive power of patient and service characteristics on place of discharge following hospital admission for an acute stroke and for hip fracture.
METHODS: Prospective cohort of 440 acute stroke and 572 hip fracture patients aged 65 years or over admitted from home to six district general hospitals and associated community hospitals, three in the North and three in the South of England.
RESULTS: Age, marital status, living arrangements, mental health status at admission, pre-admission self-rated disability, pre-admission use of home-care services, post-admission staff assessments of functional dependency as measured by Barthel Index and of confusion as measured by the modified Crichton Royal Behavioural Rating Scale and nursing staffs' expectation of place of discharge are all significantly related to place of discharge for stroke and hip fracture participants. Logistic regression correctly predicted discharge destination for 87% of stroke patients from data available at time of admission and 83% of hip fracture patients. Of the 30% of stroke patients discharged to an institution, the model correctly predicted 73%. However, of the 19% of hip fracture patients discharged to an institution, only 28% were correctly predicted.
CONCLUSIONS: Data about older patients admitted to hospital with an acute condition should be routinely collected by hospital staff to inform clinical management and to permit risk-adjusted audit.