Choice of spread by a random sample of the British population. Association with socio-economic status and risk factors for cardiovascular disease.