Total cholesterol and low-density lipoprotein alterations in children and adolescents from Brazil: a prevalence meta-analysis

ABSTRACT Objective: The aim of the present study was to evaluate the prevalence of total cholesterol (TC) and low-density lipoprotein (LDL) alterations in children and adolescents in Brazil. Materials and methods: A systematic review and meta-analysis of prevalence. The search for articles was carried out in the databases: Medline (PubMed), Embase, Scientific Electronic Library Online (SciELO), Latin American and Caribbean Literature in Health Sciences (Lilacs). The meta-analysis was performed using the random effects model. The I² test was used to identify heterogeneity. Results: The present metanalysis revealed a significant prevalence of altered lipid profile in children and adolescents in Brazil. Regarding lipoprotein fractions, the prevalence of altered TC level was 27.47% (95% CI 24.36-30.82), and a smaller prevalence was observed for LDL cholesterol (19.29% – 95% CI 15.21-24.16). The models revealed high heterogeneity (I² = 99%; p < 0.01), however the precise source of it was not identified; although type of school, age group, year and the region of Brazil appeared to influence the estimations of altered lipid profiles. Conclusion: An important prevalence of lipid alterations was observed among Brazilian children and adolescents. Those results reinforce the importance of knowing the lipid profile of children and adolescents to perform early interventions for this public.

. International cutoff like the National Heart, Lung and Blood Institute (NHLBI) consider elevated TC, a cholesterol value > 200 mg/dL and a borderline value between 170 and 199 mg/dL; while the LDL value is pathological if it is > 130 mg/dL and borderline if > 110 mg/dL (2). However, in most Brazilian studies, the definition adopted is based on The Brazilian Society of Cardiology (BSC). The BSC classifies lipid profile for ages 2 to 19 years as follow: TC borderline 150-169 mg/dL; elevated ≥ 170 mg/dL and LDL borderline 100-129 mg/dL; elevated ≥ 130 mg/dL) (3)(4)(5).
Increases in LDL levels are the main predictor of CVD and LDL is the main component of TC (1). Therefore, both, TC and LD, are the focus of this study. In children and adolescents, lipid alterations can be risk factors to CVD; however, it usually occurs due to obesity (6). The association between lipid disorders and comorbidities, such as hypertension, obesity, and diabetes, are the main risk factors influencing the development of CVD (7). The Bogalusa Heart Study from the United States reported that atheromatous lesions in the aorta begin in childhood and it increases from 10 years of age until adulthood (8). The Bogalusa Heart Study revealed the presence of fatty streaks, which are precursors of atherosclerotic plaques, in the aorta and in the coronary bed of children and adolescents; those aortic injuries were correlated with elevated serum LDL levels (8,9).
Obesity, family medical history, physical inactivity, inadequate dietary patterns/habits are risk factors for lipid profile alterations in children and adolescents (10,11). Dietary habits and physical exercise are modifiable risk factors and, as such, can be subjected to intervention (12,13). However, to perform a better public health intervention it is important to understand the prevalence of lipid alteration in children and adolescents, as well as its geographical distribution, and other risk factors that may be related to the increase of the serum lipids.
Aiming to understand the impact of risk factors for CVD, the "Estudo de Riscos Cardiovasculares em Adolescentes" (ERICA) study (14) evaluated adolescents in public and private schools of Brazilian cities with populations >100,000. Results from that study, however, did not include children, and were limited to cities with large populations.
Several studies have reported the prevalence of TC and LDL alterations among children and adolescents in Brazil (15)(16)(17)(18). However, the diversity of cutoff points adopted for the classification of altered TC and LDL among Brazilian studies makes it difficult to compare results internationally. Moreover, the prevalence of TC and LDL alterations and in children and adolescents from Brazil remains unclear. The aim of the present study was to evaluate the prevalence of TC and LDL alterations in children and adolescents in Brazil through meta-analysis and identify aspects that influence these rates.

The
International Prospective Register of Systematic Reviews approved the research protocol (CRD42018103796). This systematic review followed the procedure suggested by the guidelines Preferred Reporting Items for Systematic Reviews and Metaanalysis (PRISMA) (19).

Eligibility criteria
Eligibility criteria included cross-sectional and baseline of cohort studies investigating prevalence rates of altered lipid profile among children and adolescents in Brazil.
To be included, studies were required to describe the prevalence of TC and/or altered LDL levels of children (2-10 years old) or adolescents (>10-19 years old), and report information collected in a community or in schools of Brazil. The studies must have used the international cutoff criteria for diagnosis (NHLBI) or the latest cutoff proposed by the BSC (3)(4)(5). There were no restrictions towards publication date, language, or publication status. Studies that evaluated children or adolescents with specific health conditions (e.g., diabetes, psychological and/or genetic diseases, populations with specific congenital problems, genetic syndromes, endocrine or immunological dysfunction, or primary hypertension), were excluded. Interventional studies were also excluded due to its inclusion criteria and sample size estimations that usually do not allow to estimate prevalence adequately.

Research strategy
Literature searches were performed in the Medline (PubMed), Embase, Scientific Electronic Library Online (SciELO), and Literatura Latino-Americana e do Caribe em Ciências da Saúde (Lilacs) databases using Medical Subject Headings (MeSH) terms and entries for PubMed and Embase, and DeCS (Health Sciences Descriptors) for the SciELO and Lilacs databases up to March 2022. Full-text versions of all potentially eligible articles were downloaded from the electronic databases or requested directly to the authors via e-mail. All searches were performed independently by two reviewers (VMJB and TPRS). Search strategies were tested using MESH and the related indexing terms for each database (Table S1) with the keywords ''dyslipidemia'', "total cholesterol", "low-density lipoprotein", ''prevalence'', ''children", "adolescents" and ''Brazil''. No data and language restrictions were applied. An independent manual search of the reference lists of the retrieved articles was also performed.
included in the review were identified, the study with the most complete data was considered. However, if the studies were identified from the same sample and, for example, one has a prevalence of TC and the other has a prevalence of LDL, both were included in the review. Observational studies were included if they provided cross-sectional data from baseline.
For data extraction, an electronic spreadsheet was created in which the following information was recorded: study name; authors; year of data collection; city; state; objective; age group; type and size of the sample; TC and LDL levels. TC and LDL levels were considered elevated according to the cutoff values determined by the authors in each study.

Risk of bias within studies
Quality assessment of the studies was performed using the Newcastle-Ottawa Scale, with an adapted version of the scale for cross-sectional studies (20), with a maximum of 10 points for the least risk of bias study. Two authors (CKD and TPRS) evaluated the risk of bias. Although differences in quality assessment scores between investigators were unusual, they were resolved by consensus. The risk of publication bias across studies was explored with funnel plot asymmetry and Egger's Test. Trim-to-fill correction was used in the presence of publication bias.

Data analysis
The primary endpoints were the prevalence of altered TC and LDL cholesterol levels with the corresponding 95% confidence interval (CI). Summary measures were estimated for the total population and for subgroups defined according to age group, type of school, year of publication, and region of the country (i.e., Brazil). Brazil is geopolitically divided into five regions with at least three states on each region. Heterogeneity was assessed using the chi-squared test with statistical significance set at p < 0.10, and its magnitude was determined using the I² statistic. Meta-analysis was performed using a random effect model and weighted according to the inverse of variance. The meta-analyses were performed with articles using the same cutoff for diagnosis (NHLBI or BSC). Therefore, the same outcome has two forest plots: one with studies using BSC criteria and other for studies using the NHLBI criteria. Analyses were performed using the command "Metaprop" in RStudio version 3.4.4, adopting statistical significance at p < 0.05.

RESULTS
In March 2022, the literature search identified 831 studies in the databases. After screening of titles and abstracts and, subsequently, full-text reading of the articles, 47 (14, studies were included in the present systematic review ( Figure 1). The characteristics of these studies are summarized in Table 1. Sample sizes ranged from 95 to 38,069 across the studies. All regions of Brazil were represented; however, few studies from the Northern and Midwest states were found. Most studies were conducted in cities in the Southeast region, with emphasis on the states of São Paulo and Minas Gerais ( Table 1).
Seven studies (21-27) exclusively evaluated lipid profile in 2,591 children, most of whom were from the Southeast region (Table 1), ranging in age from 6 to 10 years. Twenty-two studies (14,28-48) examined exclusively adolescents (10 to 19 years old), covering 45,331 individuals, and were performed predominantly in the Northeast and Southeast regions of Brazil (Table 1). In addition, 18 studies (49-66) evaluated simultaneously children and adolescents, totaling 24,400 individuals 2 to 19 years of age (Table 1) Regarding the studies methodological quality evaluation, 44 studies were graded ≥ 7 of 10 stars. The articles were mostly downgraded due to lack of assessment of non-respondents (Table 1).
A subgroup meta-analysis according to the cutoff criteria used to diagnose altered TC was performed. Pooled analysis for TC prevalence according to the NHLBI criteria indicated a tendency to a lower prevalence estimate ( Figure S1A). Therefore, the following analysis were performed for each diagnostic criterion, BSC and NHLBI.     (Figures S2A and S2B). A second subgroup meta-analysis examining the prevalence of elevated TC levels was performed according to the age group (Figures 2 A e B). Children exhibited a higher prevalence of altered TC levels (11.56% [95% CI 7.31-17.82]) than adolescents (4.78% [95% CI 3.45-6.59]) (p = 0.002) with the NHLBI criteria (Figure 2A), with no difference when using the BSC criteria ( Figure 2B). A third subgroup analysis for altered TC was performed with the studies in which the samples came from school. The analysis were divided by type of school: private or public schools ( Figure 2C). Children and adolescents in public schools presented a higher prevalence of altered TC levels ( Figure 2C) with the BSC criteria, and no differences was observed when using NHLBI criteria. A fourth subgroup meta-analysis examining the year when studies were published revealed no difference (p = 0.391) in the prevalence of altered TC levels with the BSC criteria nor the NHLBI criteria ( Figures S2A and S2B).
To conclude the TC analysis, a subgroup metanalysis was performed by gender. No difference in the prevalence of altered TC level between girls and boys (p = 0.3439) was observed.
Similar to TC, a subgroup meta-analysis according to the criteria adopted for altered LDL was performed. The estimated prevalence of elevated LDL levels in children and adolescents classified according to Figure 2A. Prevalence of high total cholesterol for age group according NHLBI criteria. Note: p-value = 0.021. Figure 2B. Prevalence of high total cholesterol for age group according BSC criteria Note: p-value = 0.990.

Lipid profile alterations in young Brazilians
Arch Endocrinol Metab. 2023;67/1 Figure 2C. Prevalence of high total cholesterol for school administrative dependency according BSC criteria. Note: p-value = 0.034.  Figure S1B) and the following analysis were performed for each criteria adopted. Regarding the regions of Brazil, the meta-analysis revealed that the South region had the highest prevalence of elevated LDL levels compared with the other regions (p < 0.001) using the NHLBI criteria ( Figure 3A). When the regions of Brazil were analyzed with BSC criteria, the Southeast region had the highest prevalence of elevated LDL levels compared with other regions (p < 0.01) ( Figure 3B). There was a tendency toward an elevated prevalence estimation of LDL in most recent studies (p < 0.05) with the NHLBI and BSC criteria (Figures 4A and 4B). There was no difference in the prevalence of altered LDL levels between girls and boys with the NHLBI criteria (p = 0.974), neither between age groups (children versus adolescents; p = 0.613) (Figures S3A and S3B).

NHLBI
Subgroup analysis by gender were not performed with BSC criteria because only Ribas and Silva (63) provided data separated by gender and no difference between genders was observed in this study. Children exhibited a higher prevalence of altered LDL levels ( Figure 5). No subgroup analysis was performed according to the type of school since all studies using BSC criteria were from public schools and all using NHLBI criteria were from private schools.
The visual inspection of the funnel plot for TC and for LDL (Figures S4A and S4B) indicated publication bias. Therefore, the trim and fill correction was performed for TC and LDL ( Figures S4C, S4D, S4E and S4F). The correction evidenced a very similar elevated TC prevalence estimation. On the other hand, the analysis for LDL evidenced that probably the prevalence of elevated LDL is underestimated in the previous analysis.

DISCUSSION
The present meta-analysis revealed a significant prevalence of altered of TC and LDL alteration in children and adolescents in Brazil. Type of school, age group, year and the region of Brazil appeared to influence estimations of altered lipid profiles. In addition, the cutoff points used for the diagnosis of altered TC and LDL levels varied across the studies and influenced the prevalence estimation. We must recognize that the results on subgroup analysis were dependent of the diagnoses criteria adopted and the associations were tenuous. Although some associations were just seen when using one of the diagnoses criteria, a clinical tendency was seen on the same direction with the other criteria.
There was significant variability in the prevalence estimates for elevated TC and LDL levels among the studies. Some authors reported a very high prevalence, such as Faria and cols., who reported a rate of 57% in 100 adolescents for TC and 50% for LDL cholesterol (34), while other authors reported low values, 3.3% for altered TC and 1.7% for altered LDL in 180 adolescents (31). This discrepancies in prevalence estimation may be due to selection bias of small sample studies. However, sensitive analysis with studies with sample sizes greater than 300 subjects did not evidenced over or under estimations of lipid profile alterations prevalence (data not shown). The Brazilian Society of Pediatrics (67) recommends the use of the same cutoff values that does the BSC (3)(4)(5). However, international references (2) consider high a cholesterol value > 200 mg/dL and border line a value between 170 and 199 mg/dL, while the LDL value is pathological if > 130 mg/dL and border line if > 110 mg/dL. Burgos and cols. (51), who analyzed 1,743 children and adolescents in the South of Brazil and adopted the borderline criteria of the NHLBI (TC > 170 mg/dL and LDL > 110 mg/dL), reported prevalence of 60.75% and 43.14%, respectively (51). Although several differences were observed on the prevalence estimates in this metaanalysis, the prevalence of altered lipids in the Brazilian pediatric population was very high, even using the most conservative limits for the diagnosis, the NHLBI parameters.
Cutoff criteria is an especially important issue because it defines treatment strategies; therefore, validity studies of diagnostic tests are needed to identify the optimal cutoff points for TC and its fractions in the child and adolescent population. In 2011, in the city of Londrina, 1000 adolescents between 11 and 16 years old were subjected to lipid profile evaluation and classified according to 3 diagnostic criteria (68). Different prevalence estimates were found according to the distinct criteria: TC (BSC 38.3%; National Cholesterol Education Program [NCEP] 11.2%; National Health and Nutrition Examination Survey [NHANES] 4.8%); and LDL (BSC/NCEP, 10.8% and NHANES, 5.9%). Overall, the prevalence of dyslipidemia according to each criterion was 61% (SBC), 28.6% (NCEP), and 24.2% (NHANES) (68).
Comparing the prevalence of elevated TC and LDL in public and private schools, the prevalence of altered TC and LDL in public schools were higher Prevalence n

Lipid profile alterations in young Brazilians
Arch Endocrinol Metab. 2023;67/1 than in private schools. A Brazilian study about price and availability of food products with and without trans fatty acids in food stores near elementary schools located in low-and medium-income neighborhoods observed that cheaper products containing trans fats were more readily available than products without trans fats, promoting the consumption of less nutritious food by underprivileged children and adolescents (69). Reducing trans fatty acids intake by children may result in improvements on the lipid profile (70). Although some studies have shown a predominance of obesogenic environments with high ultra-processed food intake mainly in private schools (71), this type of food is consumed by all social classes (72) and in consequence by all children, from public to private schools. Our TC estimate in private schools came from just two studies, therefore we cannot assume that it happens the same in other private schools. Furthermore, in our metaanalysis, the Southeast and South, the richest regions of Brazil, exhibited higher rates of lipid profile alterations. Through systematic search, several studies were found, contributing to a representative sample size of the population of Brazilian children and adolescents. However, despite the large number of articles, the Northern and Central-Western regions were underrepresented, which may have influenced the results according to region.
Another finding of the present study was the increase in the prevalence of elevated TC and LDL levels between 2000 and 2019. A tendency towards increase in prevalence rates was seen mainly after 2010. The pattern of food consumption in Brazilian population has been changing over the past few decades, with an increase in the consumption of ultra-processed foods (73). The early and abusive consumption of ultraprocessed foods is one of the most important factors associated with the increased prevalence of obesity and the risk for metabolic complications (3,7), such as changes in lipid profile (4,5,67,(74)(75)(76)(77)(78).
Another important factor in childhood and adolescence is the low level of physical activity and the increase of a sedentary lifestyle over the years (79). When it is associated with changes in eating habits, this behavior may contribute to increases in obesity and metabolic changes, including dyslipidemia (12,18). Data on Brazilian adolescents from all capitals of Brazil, indicates a prevalence of leisure-time physical inactivity around 54.3% (79). The Brazilian Society of Sports Medicine recommends regular physical activity and highlights the benefit of improved lipid profiles and believes there is an association between physical inactivity, obesity and dyslipidemia, and that obese children may become obese adults prone to illnesses related to weight gain (80,81). All this changes probably had some influence on the findings, although a selection bias could be present, especially in light of the profound heterogeneity reported on the analysis.
Higher rates of TC were detected in children when compared to adolescents, but no differences between boys and girls was observed. Lipoprotein concentrations changes considerably with normal growth and maturation, and varied according to sex (82)(83)(84)(85)(86). During pubertal growth, cholesterol is included in growing cells, leading to decreases in serum lipid values (87) but it increases on the later adolescence, approaching adult concentrations (84). Our sample of adolescents are mainly from individuals on the late phase (mean age 14.7 years old; data not shown). Therefore, screening recommendations should consider fluctuations in serum lipid levels during growth and sexual maturation. It is recommended that every child undergo a determination of TC level at 10 years of age by means of an examination of digital pulp capillary blood (4). Because adolescence is a critical period of life for the onset or persistence of obesity and its complications (86), knowing the nutritional status of this segment of the population is important.

Strengths and limitations
Finally, it should be noted that the present metaanalysis was designed according to the standards recommended by the Cochrane Collaboration (87) and adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement (19) but it still presents some limitations. One of the main limitations of the present study was the high heterogeneity, the origin of which could not be explained. Although several subgroup analyses were performed to identify the source of heterogeneity, it remained elusive. Another limitation is the fact that information regarding nutritional status, dietary intake and physical activity was not reported in the articles and, consequently, was not analyzed in this study. This may also have contributed to the high heterogeneity. Most studies estimated the LDL with Friedewald's formula, but no information was provided regarding In conclusions, the present study indicates a high prevalence of altered -if not abnormal -lipid levels among children and adolescents in Brazil. These results reinforce the importance of knowing the lipid profile of children and adolescents to perform early interventions for treatment as well as to promote healthy habits that lead to prevention of lipid profile alterations and its consequences.   Figure S4E. Funnel plot with the trim-to-fill correction for TC. Figure S4F. Funnel plot with the trim-to-fill correction for LDL.