Population versus clinical views in coronary disease: can epidemiological data be useful to clinicians?