RESPONSE TO A CHOLESTEROL-LOWERING DIET - EFFICACY IS GREATER IN HYPERCHOLESTEROLEMIC SUBJECTS EVEN AFTER ADJUSTMENT FOR REGRESSION TO THE MEAN

PURPOSE: To define how much regression to the mean confounds apparent responsiveness in subgroup analyses, and to test, using techniques tha t remove regression to the mean, whether hypercholesterolemic subjects are more likely to respond to diet. PATIENTS AND METHODS: Data collec ted on 812 men and women participating in the Minnesota Coronary Surve y Dietary Trial who had at least 2 total cholesterol measurements on a high-saturated-fat diet and 1 cholesterol measurement on a low-satura ted-fat diet were analyzed for the effects of initial serum cholestero l and regression toward the mean on measurement of diet responsiveness . RESULTS: If regression towards the mean is not taken into account, d ietary responsiveness in patients with mean cholesterol levels of 280 mg/dL was -25%, whereas dietary responsiveness in subjects with mean s erum cholesterol levels of 156 mg/dL was -5%. After regression toward the mean was taken into account, subjects with high initial serum chol esterol levels had an 18% reduction in serum cholesterol levels wherea s subjects with lower levels had an 11% reduction. Even after regressi on toward the mean is accounted for, subjects with high serum choleste rol levels were significantly more diet-responsive (p <0.005). CONCLUS ION: The efficacy of a cholesterol-lowering diet for individuals can b e overestimated or underestimated if only single measurements are used to determine response. Subjects with hypercholesterolemia even after adjustment for regression towards the mean, are more diet-responsive t han subjects with lower cholesterol levels. Dietary therapy should rem ain the first step in the treatment of hypercholesterolemia, and shoul d also be effective in reducing cholesterol levels in the population a t large.