Statistical issues in service evaluation – a case of intermediate care