the Planella formula based on apolipoprotein B to calculate LDL- cholesterol. However, the reviewers recommended that we use apoli- poprotein A-I instead.
CORRESPONDENCE
formula gave the same LDL-cholesterol values as the Friedewald formula. When we submitted our first draft of the study for publication, we used the Planella formula based on apolipoprotein B to calculate LDLcholesterol. However, the reviewers recommended that we use apolipoprotein A-I instead. Their argument was the same as that of Wägner and Ordóñez-Llanos—ie, the formula includes apolipoprotein A-I instead of HDL-cholesterol. In all our statistical analyses we adjusted for fasting, and calculated all results separately according to nutritional state (which were seen at peer review). The results and the comparisons between apolipoprotein B and LDL-cholesterol were similar, again supporting the fact that apolipoprotein B and apolipoprotein AI are insensitive to the nutritional state, which is also supported by Wägner and Ordóñez-Llanos. In multivariate analyses comparing LDL-cholesterol and apolipoprotein B, we found that apolipoprotein B was much better than LDL-cholesterol irrespective of which method we used to calculate LDL-cholesterol. The risk ratios (for 1 SD change) in favour of apolipoprotein B were 1·33 for men and 1·53 for women (both p