Beaton et al (Am J Clin Nutr 1979;32:2546-59) reported on the partitioning of variance in 1-day dietary data for the intake of energy, protein, total carbohydrate, total fat, classes of fatty acids, cholesterol, and alcohol. Using the same food intake data and the expanded National Heart, Lung and Blood Institute food composition data base, these analyses of sources of variance have been expanded to include classes of carbohydrate, vitamin A, vitamin C, thiamin, riboflavin, niacin, calcium, iron, total ash, caffeine, and crude fiber. The analyses relate to observed intakes (replicated six times) of 30 adult males and 30 adult females obtained under a paired Graeco-Latin square design with sequence of interview, interviewer, and day of the week as determinants. Neither sequence nor interviewer made consistent contribution to variance. In females, day of the week had a significant effect for several nutrients. The major partitioning of variance was between interindividual variation (between subjects) and intraindividual variation (within subjects) which included both true day-to-day variation in intake and methodological variation. For all except caffeine, the intraindividual variability of 1-day data was larger than the interindividual variability. For vitamin A, almost all of the variance was associated with day-to-day variability. One day data provide a very inadequate estimate of usual intake of individuals. In the design of nutrition studies it is critical that the intended use of dietary data be a major consideration in deciding on methodology. There is no "ideal" dietary method. There may be preferred methods for particular purposes.

As part of a population survey and a follow-up study of plasma lipid profiles by high temperature gas-liquid chromatography, we have determined the quantities and relative proportions of all major chemical classes and molecular species of llpids of plasma from 1200 subjects at Visit 2 of the Toronto-McMaster Lipid Research Clinic Prevalence Study. We have compared these values between our 24 subjects with ischemic vascular disease and 73 control subjects matched for age, sex, and plasma total cholesterol and triacylglycerols. The phosphatidylcholine/free cholesterol ratio showed the highest association with ischemic vascular disease of any of over 10 other lipid parameters and all the common risk indicators except high density lipoprotein cholesterol. The phosphatidylcholine/free cholesterol ratio had a reiatlve risk ratio of 20/4 (95% confidence limits, 15/9,23/1) and high density llpoproteln cholesterol, a risk ratio of 23/1 (95% confidence limits, 24/0,19/5) for ischemic vascular disease. The average ratio of phosphatidylcholine/free cholesterol for the ischemic vascular disease group was 1.36 and for the controls 1.51, the population average being 1.50. Plasma high density lipoproteln cholesterol had a significant correlation (R = 0.15) with the phosphatidytcholine/ffee cholesterol ratio in the total population sample. The Increased risk for ischemic vascular disease from a tower phosphatidylchollne/free cholesterol ratio may possibly be explained on the basis of decreased fluidity and stability of the lipoproteins due to a relative oversaturatfon with free cholesterol. (Arteriosclerosis 2: 296-302, July/August 1982) A 1976 review' has concluded that the cause and pathogenesis of atherosclerosis remains unsoived. A number of atherogenic processes, inciuding injury to the artery wall, plasma lipoprotein infiltration, thrombosis, and the cellular and tissue responses appear to complement each other. Studies over an extended period of time have established that elevated plasma cholesterol and triglyceride levels are related to the risk of developing ischemic vascular disease (IVD). 2 -3 Of the commonly measured lipoproteins, the low density lipoproteins (LDL), and, to a lesser extent, trie very low density lipopro- teins {VLDL) are positively associated with increased risk, while high density lipoprotein (HDL) is associated with reduced risk and possible protection against the disease. 4 " 6 Due to the many factors that affect biood lipid levels, individual variations in response to environmental conditions, and the fact that they have continuous unimodal frequency distributions in the population, it is impossible to define clearly the normal and abnormal ranges for plasma lipid and lipoprotein fractions. Their ideal concentrations, therefore, remain undecided but in general it is considered that the lower the level of total LDL and VLDL cholesterol and the higher the level of HDL cholesterol the less the risk of premature IVD. 7B However, the metabolic or biochemical and physiochemicai basis for their role in ...

No abstract

No abstract