The type and amount of dietary fat are associated with serum lipid and lipoprotein concentrations in children as well as in adults (1-7). Dietary alterations in serum lipid levels are safe and relatively efficacious in young children (8-11). Moreover, evidence from several large studies reveals that serum total cholesterol (TC) and other adult cardiovascular disease risk factors seem to be relevant even in childhood and predict later coronary disease and mortality (12-14). Total dietary fat by itself, independent of the relative content of fatty acids, is not associated with serum cholesterol levels or coronary heart disease. However, a decrease in total fat consumption may facilitate reduction in saturated fatty acid intake (15). Also, fat is high in energy density, and a key strategy for preventing weight gain is to limit such foods. Given the increasing risk of overweight and obesity in children, fat therefore may be important in weight control (16). Therefore, several expert groups have recommended reduci ng fat levels to 30% of energy or less and saturated fat to less than 10% of energy in the diets of adults and children over the age of 2 years (10,11,15,17- 22). In contrast, some Canadian expert groups and individuals in the United States question the appropriateness of suggesting such modifications in child eating patterns. They argue that substituting carbohydrate for fat in child diets may be dangerous or inefficacious not only with respect to glycemic response but with respect to its effect on blood lipids (9,23-25). Also, they suggest that higher fat eating patterns may be more acceptable, particularly for preschool children (26). Monounsaturated fat--rich diets, somewhat higher in both total fat than the 30% of energy that is usually recommended but low in saturated fat, may reduce both total and low density lipoprotein cholesterol (LDL-C) while preserving high density lipoprotein cholesterol (HDL-C) levels and thus may have particular advantages (15,27,28). Finally, recent studies in adults suggest t hat individual dietary fatty acids differ markedly in their effects on serum lipids and lipoprotein levels (29,30). The effects of alterations in a specific fatty acid or in a group of fatty acids on serum lipids (while holding total fat and energy intakes constant) have rarely been studied in children's diets. In dietary counseling, parents are often asked to substitute foods of differing nutrient composition for each other. The real-life utility in modeling data, such as these, is to allow various types of theoretical dietary substitutions to be evaluated and compared.
In light of these questions, we examined the associations between diet and serum cholesterol levels in a large multiethnic cohort of American children who participated in the Child and Adolescent Trial for Cardiovascular Health (CATCH). CATCH was a randomized, controlled community intervention trial, in which school lunchroom staff were trained to reduce the fat content of school meals. The resulting variability in dietary fat and macronutrient intake, as reported by a sample of students who provided a 24-hour food recall and a blood sample, allowed us to explore by statistical modeling the potential impact of various dietary modifications on serum total and HDL-C. Such modifications might include a combination of low saturated fat and low total fat intake (<30% of energy); low saturated fat intake coupled with total fat intake of 30% of energy; and low saturated fat intake with higher intake of total fat (>30% of energy). Because total fat intake affects levels of other macronutrients, especially carbohydrat e, and may also affect total energy intake, these factors were also considered. Age, gender, and weight, which seem to mediate these associations, were also taken into account.
Overview of CATCH
The CATCH was a National Heart, Lung, and Blood Institute-- sponsored, multicenter, intervention study to promote healthful behaviors and to reduce risk factors for heart disease among elementary school children. CATCH involved 4 field centers (California, Louisiana, Texas, and Minnesota). Interventions consisted of school-based (school foodservice, physical education, classroom curricula) and family-based (home curricula, family fun nights) components. It was the first multicenter field trial to test the effectiveness of such interventions on decreasing daily intake of total fat, saturated fatty acids, and sodium in the diets of elementary school children. Several publications describe the CATCH study design, hypotheses, evaluations, and outcomes in detail (31-34). Briefly, fat intake was significantly reduced (P<.01) among children in intervention schools at follow-up (from 32.7% to 30.3% of energy consumed) compared with children in control schools (from 32.6% to 32.2% of energy). Much of this difference c ame from the decline in the intake of saturated fatty acids in the intervention schools (from 12.7% to 11.4% of energy) compared with minimal change among children in the control schools (from 12.5% to 12.1%). CATCH provided an opportunity to investigate the associations between fatty acid intakes (from 24-hour recalls) and cardiovascular risk factor variables in this multiethnic sample of children. A more complete description of the sample for dietary recalls is provided by Lytle et al (35).
From the CATCH baseline cohort of 5,106 children, 3,218 (63%) were randomly selected for the dietary substudy, of whom 1,920 (60%) participated at baseline (Grade 3). After loss to follow-up and exclusion of implausible values (46 at baseline, 47 at follow-up), a total of 1,182 children (37% of the random sample; 473 control and 709 intervention) provided usable dietary data at both baseline and follow-up (Grade 5). At each assessment period, research nutritionists who were trained in dietary assessment obtained a diet record--assisted 24-hour recall covering a weekday. All of the 1,182 children with paired dietary data provided blood samples for lipid analysis at baseline (by definition of the CATCH cohort), and all but 14 provided blood at follow-up, leaving a sample of 1,168 for follow-up blood lipids. Each analysis included 1,182 baseline values and 1,168 paired follow-up …