Serum Cholesterol Response to Changes in Dietary Lipids

Abstract

An analysis is presented of the data from ninety-nine sets of controlled experiments, each providing average serum cholesterol response (Δ cholesterol) in a group of men, in calorie balance, to a known change in the diet. The recently reported data of Hegsted et al. are in good agreement with predictions based on a formulation published earlier, the correlation between observed and predicted values being r = 0.92, and the root mean square error 14.9 mg. cholesterol per 100 ml. of serum. In that formulation, saturated fatty acids with 12 through 17 carbon atoms (S’), and dietary cholesterol, expressed as the square root of the mg. cholesterol per 1,000 kcal. of diet, raise the serum level and the two effects are linearly additive: monoenes and stearic acid are without significant effect; and polyenes depress the serum level so that about two parts of polyene will offset the effect of about one part of S’. The formulation of Hegsted et al. gives predicted values that are well correlated with observed values, r = 0.87, but the root mean square error is 28.0. This large value for the root mean square error reflects the fact that, as shown independently, serum cholesterol is not, as assumed by Hegsted et al., a simple linear function of dietary cholesterol.