Keys A
Am J Clin Nutr. 1984 Aug;40(2):351-9. doi: 10.1093/ajcn/40.2.351.
Serum cholesterol change, delta Chol mg/dl, from change only in dietary cholesterol, mg/1,000 kcal, x, was analyzed with data from 39 controlled experiments on 64 adults with the model delta Chol = b(x2k-x1k). Values of b were found by iteration for exponents k = 0.3 to k = i.0 yielding eight regression equations to check predicted against observed. The smallest root mean square deviation regression was 7 mg/dl given by delta Chol = 1.5(x2(1/2)-x1(1/2] which predicted an average serum cholesterol change of 18.7 versus 19.9 observed. The relationship pertains to calorie balance on a wide range of nutritionally acceptable diets. A prediction equation proposed from experiments in Massachusetts that took k = 1.0 overpredicted the effect of dietary cholesterol in these experiments by 300%.
利用来自对64名成年人进行的39项对照实验的数据,分析了仅因膳食胆固醇(毫克/1000千卡)变化x导致的血清胆固醇变化(Δ胆固醇,毫克/分升),模型为Δ胆固醇 = b(x₂ᵏ - x₁ᵏ)。通过对指数k从0.3到1.0进行迭代来确定b值,得到八个回归方程,用于将预测值与观测值进行对照检验。最小均方偏差回归为7毫克/分升,由Δ胆固醇 = 1.5(x₂⁽¹/²⁾ - x₁⁽¹/²⁾)给出,该方程预测的平均血清胆固醇变化为18.7,而观测值为19.9。这种关系适用于各种营养可接受饮食的热量平衡情况。马萨诸塞州实验提出的一个预测方程,其中k = 1.0,在这些实验中对膳食胆固醇的影响预测高估了300%。