Methods

Demographic and geographic data
The city of Cuiabá, capital of Mato Grosso, is in the central part of South America, and it has a large, diverse population living in three different ecosystems (Figure 1). The ethnic composition of the city is the result of two distinct historical phases, the first occurring during the gold rush in the 18th century, with miscegenation between whites and Native Indians and later with slaves from Africa. The second phase of miscegenation occurred in the 1970s with the migration of people from southern and southeastern Brazil, consisting of people from many ethnic backgrounds, mainly of European descent. The migration was so extensive that the number of inhabitants in the capital quintupled in three decades. Therefore, the residents of this city come from various regions, which makes this city a representative place for an RI study from an ethnographic point of view.
Figure 1  Map of South America: Cuiabá, Mato Grosso, is located in central Brazil; the three ecosystems of the state are shown. 
The sample size calculation was based on CLSI recommendations, which require at least 120 reference individuals per age group [10]. This cross-sectional study evaluated 1,994 healthy children and adolescents who were randomly sampled from 20 schools and 25 daycare centers. Interviewers completed a written questionnaire assessing the individual’s background, his or her relatives and demographic and anthropometric data. The following inclusion criteria were used for this study: children and adolescents from 1 to 12 years, 11 months and 29 days of age without any underlying disease or diagnosed clinical complaints at the time of blood collection. Children and adolescents who had a chronic disease or who were taking medications that were not reported in the questionnaire but were discovered during clinical evaluation were excluded. Children and adolescents were included after parents or guardians signed the informed consent. All data were entered into Epidata 3.1 with a double entry. Of the 1,994 individuals who participated, 128 were excluded for the regular use of medication (n = 14, 12 children and 2 adolescents) or acute clinical symptoms present at the time of blood collection (n = 114, 100 children and 14 adolescents). Complaints referred to at the time of collection, which resulted in the exclusion of children, were fever (40 children and 5 adolescents), sore throat (28 children and adolescents 6), otalgia (27 children and 2 adolescents), dysuria (5 children) and dengue (1 teenager). Children taking the following medications were also excluded: Carbamazepine (4 children), Phenobarbital (3 children), Ritalin (1 child and 1 teenager), Seretide (2 children), Sabril (1 child), prophylactic Benzetacil (1 teenager) and tuberculosis prophylaxis (1 child).
The blood samples were collected after a fasting period of 3 hours for children from 1 to 2 years old, 6 hours for children from 2 to 5 years old and 12-14 hours for older children and adolescents. The samples were processed in a Cobas® 6000 analyzer (Roche Hitachi Cobas 6000 Analyzer, Hitachi High Technologies Corp., Tokyo, Japan) using enzymatic colorimetric and enzymatic homogeneous colorimetric methods. The nHDL-c and LDL-c levels were calculated (nHDL-c = TC - HDL-c and LDL-c = TC – HDL-c – TG/5), and the latter parameter was calculated using the Friedwald formula.

Ethics
The study was approved by the Ethics Committees at the Julio Müller University Hospital (# 947/2010) and the Faculty of Medicine of the Universidade de São Paulo (# 318/2011). The Municipal Department of Education and Health of Cuiabá city also reviewed and approved this study.

Statistics
We tested the homogeneity of variances using Bartlett’s test for each parameter by age, and depending on the result, ANOVA or a Kruskal-Wallis test was used to examine differences among the age groups. The Bonferroni post-hoc test was applied to verify pairwise differences between the means by ages if the ANOVA or Kruskal-Wallis test yielded a p-value less than 0.05, and age intervals with similar means were combined. Bartlett’s test was applied to the new age groups, followed by ANOVA or a Kruskal-Wallis test to check whether the new age groups maintained the age categorization. We excluded outliers (±3 standard deviations from the mean), and the RI was established as the resulting mean ±2 standard deviations. In addition, we calculated the percentile distribution to establish the RIs and adopted the criteria of the NHLBI Expert Panel on Integrated Guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents [5] as a proposed decision limit for this population. For TC, nHDL-c, LDL-c and TG, values below the 75th percentile were considered desirable or acceptable. Values from the 75th to approximately the 95th percentile were considered borderline, and values greater than or equal to the 95th percentile were considered high. For HDL-c, the 10th percentile was used as the lower limit. Therefore, the participants below the 10th percentile were considered to have a low concentration. A concentration above the 50th percentile was considered desirable. The significance level was 5% for all tests. The statistical analyses were performed using Minitab software, version 15 (Minitab, PA, USA) and SPSS, version 16 (Chicago, Illinois, USA).