INTRODUCTION: Acute admissions to hospital are rising. As a part of a service evaluation we examined pathways of patients following hospital discharge depending on data available on admission to hospital.

METHODS: We merged data available on admission to the Wrexham Maelor hospital from an existing data-base in the Acute Medical Unit with follow up data from local social services as part of a data sharing agreement. Patients requiring support by social services post-discharge were matched with patients not requiring social services from the same post-code.

RESULTS: Stepwise logistic regression analysis identified candidate variables predicting likely support need. Decision tree analysis identified sub-groups of patients with higher likelihood to require support by social services after discharge from hospital. We found patients with normal physiology on admission as evidenced by a value of zero for the National Early Warning Score who were frail or older than 85years were most likely to require support after discharge.

CONCLUSIONS: Information available on admission to hospital might inform long term care needs. Prospective testing is needed. The algorithms are prone to be dependent on availability of local services but our methodology is expected to be transferable to other organizations.