Biostatistical pitfalls in studies of atherosclerotic heart disease.