[Food consumption scores and serum lipids levels in the population of São Paulo, Brazil].

OBJECTIVE To study food patterns assessed using scores of consumption and their relationship with serum total cholesterol (TOTAL-C), low density lipoprotein-cholesterol (LDL-C) and high density lipoproteins (HDL-C) concentration in the population of the metropolitan area of São Paulo, Brazil. METHODS Data on food frequency consumption (FFC), serum lipids, and other covariates were available for a representative sample of 1,045 adults. A 12-month retrospective food frequency questionnaire was used. FFC was assessed using scores of consumption, which was obtained by grouping food according to their composition into two large groups: score I (known as risk food for cardiovascular diseases) and score II (known as healthy food). The association between the scores and serum blood lipoprotein levels among the study population was analyzed through multiple linear regression analyses. Modeling step-wise techniques were used to enter the covariates into the linear models. RESULTS Increasing mean levels of TOTAL-C and LDL-C were seen from the lowest through the highest quintile of score I when compared to the score II, where decreasing mean levels of TOTAL-C and LDL-C were observed from the lowest to the highest quintile. The results of the linear regression analyses between serum TOTAL-C and LDL-C levels for both FFC score I and score II, after multivariate adjusting, showed a significant positive relationship with score I and a significant and inverse relationship with score II. CONCLUSIONS In population studies, FFC analyses through scores can be the choice method to evaluate the quality of diet and its potential effect on serum levels of TOTAL-C and LDL-C.