All estimates are based on population-weighted data from the National Health and Nutrition Examination Survey. The unweighted sample size ranges for each sex-age group and lipid are as follows. Among males aged 6 to 11 years, the minimum for total cholesterol in 2003-2004 was 402 and the maximum in 2013-2014 was 549; the minimum for high-density lipoprotein (HDL) and non-HDL cholesterol in 2015-2016 was 493 and the maximum in 2013-2014 was 549. Among females aged 6 to 11 years, the minimum for total cholesterol in 1999-2000 was 411 and the maximum in 2001-2002 and 2013-2014 was 500; the minimum for HDL and non-HDL cholesterol in 2009-2010 was 471 and the maximum in 2013-2014 was 500. Among males aged 12 to 19 years, the minimum for total cholesterol in 2011-2012 was 566 and the maximum in 2001-2002 was 1081; the minimum for HDL and non-HDL cholesterol in 2011-2012 was 566 and the maximum in 2009-2010 was 629; the minimum for triglycerides in 2007-2008 was 248 and the maximum in 2003-2004 was 498; the minimum for low-density lipoprotein (LDL) cholesterol in 2007-2008 was 248 and the maximum in 2003-2004 was 496; the minimum for apolipoprotein B in 2007-2008 was 248 and the maximum in 2005-2006 was 438. Among females aged 12 to 19 years, the minimum for total cholesterol in 2007-2008 was 495 and the maximum in 2001-2002 was 1056; the minimum for HDL and non-HDL cholesterol in 2007-2008 was 495 and the maximum in 2013-2014 was 615; the minimum for triglycerides in 2007-2008 was 191 and the maximum in 2001-2002 was 457; the minimum for LDL cholesterol in 2007-2008 was 191 and the maximum in 2001-2002 was 454; and the minimum for apolipoprotein B in 2007-2008 was 191 and the maximum in 2013-2014 was 276.
aAnalysis limited to 2007 and later due to HDL cholesterol bias prior to 2007; see Methods section for details.
bFor triglycerides, the data are expressed as geometric means and log-transformed units for the β.
eTable 1. Maximal Sample Sizes, by Survey Cycle: National Health and Nutrition Examination Survey, 1999-2016
eTable 2. Characteristics of Examined Youth with Complete Versus Any Missing Data for Lipids and Apolipoprotein B, by Survey Cycle: National Health and Nutrition Examination Survey, 1999-2016
eTable 3. Age- and Race/Ethnicity-Adjusted Changes in the Prevalence of Apolipoprotein B Concordance and Discordance with Non-HDL Cholesterol and LDL Cholesterol Over Time in US Youth Ages 12 to 19 Years, 2007-2014
eTable 4. Age-Adjusted Trends in Mean Levels of Lipids and Apolipoprotein B Over Time by Race/Ethnicity and BMI Category in US Youth Ages 6 to 19 Years, 1999-2016
eTable 5. Age-Adjusted Changes in the Prevalence of Ideal Levels of Lipids and Apolipoprotein B Over Time by Race/Ethnicity and BMI Category in US Youth Ages 6 to 19 Years, 1999-2016
eTable 6. Age-Adjusted Changes in the Prevalence of Adverse Levels of Lipids and Apolipoprotein B Over Time by Race/Ethnicity and BMI Category in US Youth Ages 6 to 19 Years, 1999-2016
Customize your JAMA Network experience by selecting one or more topics from the list below.
Perak AM, Ning H, Kit BK, et al. Trends in Levels of Lipids and Apolipoprotein B in US Youths Aged 6 to 19 Years, 1999-2016. JAMA. 2019;321(19):1895–1905. doi:10.1001/jama.2019.4984
What are the recent trends in serum levels of lipids and apolipoprotein B in US youths aged 6 to 19 years?
In this serial cross-sectional analysis of nationally representative data from 26 047 youths during the period 1999-2000 to 2015-2016, decreasing linear trends were observed in mean levels of total cholesterol (from 164 mg/dL to 155 mg/dL), non–high-density lipoprotein cholesterol (from 108 mg/dL to 100 mg/dL), low-density lipoprotein cholesterol (from 92 mg/dL to 86 mg/dL), triglycerides (from 78 mg/dL to 63 mg/dL), and apolipoprotein B (from 70 mg/dL to 67 mg/dL), and increasing linear trends were observed in mean levels of high-density lipoprotein cholesterol (from 52.5 mg/dL to 55.0 mg/dL).
Lipids and apolipoprotein B changed favorably in US youths during recent years.
Favorable trends occurred in the lipid levels of US youths through 2010, but these trends may be altered by ongoing changes in the food supply, obesity prevalence, and other factors.
To analyze trends in levels of lipids and apolipoprotein B in US youths during 18 years from 1999 through 2016.
Design, Setting, and Participants
Serial cross-sectional analysis of US population–weighted data for youths aged 6 to 19 years from the National Health and Nutrition Examination Surveys for 1999 through 2016. Linear temporal trends were analyzed using multivariable regression models with regression coefficients (β) reported as change per 1 year.
Survey year; examined periods spanned 10 to 18 years based on data availability.
Main Outcomes and Measures
Age- and race/ethnicity-adjusted mean levels of high-density lipoprotein (HDL), non-HDL, and total cholesterol. Among fasting adolescents (aged 12-19 years), mean levels of low-density lipoprotein cholesterol, geometric mean levels of triglycerides, and mean levels of apolipoprotein B. Prevalence of ideal and adverse (vs borderline) levels of lipids and apolipoprotein B per pediatric lipid guidelines.
In total, 26 047 youths were included (weighted mean age, 12.4 years; female, 51%). Among all youths, the adjusted mean total cholesterol level declined from 164 mg/dL (95% CI, 161 to 167 mg/dL) in 1999-2000 to 155 mg/dL (95% CI, 154 to 157 mg/dL) in 2015-2016 (β for linear trend, −0.6 mg/dL [95% CI, −0.7 to −0.4 mg/dL] per year). Adjusted mean HDL cholesterol level increased from 52.5 mg/dL (95% CI, 51.7 to 53.3 mg/dL) in 2007-2008 to 55.0 mg/dL (95% CI, 53.8 to 56.3 mg/dL) in 2015-2016 (β, 0.2 mg/dL [95% CI, 0.1 to 0.4 mg/dL] per year) and non-HDL cholesterol decreased from 108 mg/dL (95% CI, 106 to 110 mg/dL) to 100 mg/dL (95% CI, 99 to 102 mg/dL) during the same years (β, −0.9 mg/dL [95% CI, −1.2 to −0.6 mg/dL] per year). Among fasting adolescents, geometric mean levels of triglycerides declined from 78 mg/dL (95% CI, 74 to 82 mg/dL) in 1999-2000 to 63 mg/dL (95% CI, 58 to 68 mg/dL) in 2013-2014 (log-transformed β, −0.015 [95% CI, −0.020 to −0.010] per year), mean levels of low-density lipoprotein cholesterol declined from 92 mg/dL (95% CI, 89 to 95 mg/dL) to 86 mg/dL (95% CI, 83 to 90 mg/dL) during the same years (β, −0.4 mg/dL [95% CI, −0.7 to −0.2 mg/dL] per year), and mean levels of apolipoprotein B declined from 70 mg/dL (95% CI, 68 to 72 mg/dL) in 2005-2006 to 67 mg/dL (95% CI, 65 to 70 mg/dL) in 2013-2014 (β, −0.4 mg/dL [95% CI, −0.7 to −0.04 mg/dL] per year). Favorable trends were generally also observed in the prevalence of ideal and adverse levels. By the end of the study period, 51.4% (95% CI, 48.5% to 54.2%) of all youths had ideal levels for HDL, non-HDL, and total cholesterol; among adolescents, 46.8% (95% CI, 40.9% to 52.6%) had ideal levels for all lipids and apolipoprotein B, whereas 15.2% (95% CI, 13.1% to 17.3%) of children aged 6 to 11 years and 25.2% (95% CI, 22.2% to 28.2%) of adolescents aged 12 to 19 years had at least 1 adverse level.
Conclusions and Relevance
Between 1999 and 2016, favorable trends were observed in levels of lipids and apolipoprotein B in US youths aged 6 to 19 years.
Optimal lipid levels are 1 of 7 critical factors that define ideal cardiovascular health during childhood according to the American Heart Association.1,2 Conversely, nonoptimal childhood levels of lipids (including total cholesterol,3 high-density lipoprotein [HDL] cholesterol,4,5 non-HDL cholesterol,6,7 low-density lipoprotein [LDL] cholesterol,3,7 and triglycerides3,5) are associated with concurrent presence and extent of early atherosclerotic lesions at autopsy, and with later subclinical atherosclerosis by imaging. In addition, childhood apolipoprotein B levels, indicating total atherogenic lipoprotein particle concentration (independent of the cholesterol content within those particles, which can vary), may predict adult subclinical atherosclerosis even more strongly than cholesterol levels.8
Data on lipid levels for US youths can therefore provide insight into current and future cardiovascular health, as well as highlight pediatric dyslipidemia–related resource needs9 and inform public health efforts.10 In prior analyses of lipid levels in US youths, favorable trends were observed for mean levels of all lipid types (total cholesterol, HDL cholesterol, non-HDL cholesterol, triglycerides, and LDL cholesterol) from 1988-1994 to 2007-201011; nevertheless, 20% of youths had adverse levels for at least 1 lipid in 2011-2012.12 Trends in the prevalence of ideal levels of lipids and apolipoprotein B in US youths have not been described.
The objective of the present study was to analyze trends in serum lipids and apolipoprotein B during up to 18 years from 1999-2000 to 2015-2016 in US youths aged 6 to 19 years.
We used data from the National Health and Nutrition Examination Survey (NHANES), which uses a complex, multistage probability sampling design to select a representative sample of the civilian noninstitutionalized US population.13 NHANES combines in-home interviews with mobile examinations and laboratory tests, including HDL and total cholesterol in youths aged 6 to 19 years and fasting triglycerides and apolipoprotein B in a subset of adolescents aged 12 to 19 years. Written informed consent, assent, or both was obtained from all participants.
We included youths aged 6 to 19 years who attended an examination during any NHANES cycle from 1999-2000 to 2015-2016; the overall examination response rate was 81% (range, 65%-86% across cycles).14 We analyzed all available data, excluding individuals from particular analyses if relevant variables were missing. The analyses were designed to maintain sample sizes of 200 or greater whenever possible. The Northwestern University institutional review board waived the need for review because the research did not involve human participants.
We accessed data from 1999-2000 through 2015-2016 to include all 9 continuous NHANES data cycles (vs earlier intermittent cycles). We included all available NHANES cycles for total cholesterol (1999-2016), triglycerides and LDL cholesterol (1999-2014), and apolipoprotein B (2005-2014). For HDL and non-HDL cholesterol, we included data from 2007-2016 only because NHANES documentation indicates that differing assay methods and laboratories before 2007 caused bias within the HDL cholesterol values.15
We analyzed data for the following age categories: 6 to 8 years, 9 to 11 years, 12 to 15 years, and 16 to 19 years. These age categories were constructed because lipid levels change with maturation; however, pubertal staging was unavailable. To preserve sample size, in some of the analyses we categorized children as being aged 6 to 11 years, adolescents as aged 12 to 19 years, and the combined ages sample (aged 6-19 years) included all youths. Per the NHANES protocol, race and ethnicity were self- or parent-reported from provided categories and coded as Mexican, other Hispanic, non-Hispanic white, non-Hispanic black, or other race; and starting in 2011, non-Hispanic Asian became a separate category from other race. To maintain consistency across NHANES years, the race/ethnicity adjustment variable combined non-Hispanic Asian with other race.
Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared and used NHANES-collected standardized height and weight measurements. The 2000 US Centers for Disease Control and Prevention sex-specific BMI-for-age growth charts for youths aged 2 to 20 years were used to categorize BMI as underweight (<5th percentile), normal weight (≥5th to <85th percentile), overweight (≥85th to <95th percentile), or obese (≥95th percentile).
For most youths, blood was drawn in the nonfasting state. The subset (45%-50%) of adolescents aged 12 to 19 years who attended a morning examination were requested to fast; self-reported fasting times of 8.5 to 24 hours were required for the measurement of triglycerides and apolipoprotein B (otherwise, values were set as missing).
Venous blood samples were frozen and shipped to standardized laboratories. There were changes made in the use of laboratory equipment during the NHANES cycles; however, NHANES uses the US Centers for Disease Control and Prevention’s Lipid Standardization Program to ensure accuracy and precision of measurements between laboratories and over time. Across included NHANES cycles, levels of total cholesterol and triglycerides were directly measured using enzymatic assays, and levels of HDL cholesterol and apolipoprotein B were directly measured using immunoassays.
A correction factor was applied to apolipoprotein B levels from 2005-2006 that was recommended by NHANES to account for cycle-specific laboratory and equipment differences. Level of non-HDL cholesterol was calculated as the difference between HDL and total cholesterol. Level of LDL cholesterol was calculated using the Friedewald equation16 for serum samples with triglyceride values of 400 mg/dL or less and set as missing for samples with triglyceride values greater than 400 mg/dL.
Levels of lipids and apolipoprotein B were classified as ideal, borderline, or adverse according to the most recent pediatric lipid guidelines.17 Cut points for ideal and adverse, respectively, were as follows: less than 170 mg/dL and 200 mg/dL or greater for total cholesterol, greater than 45 mg/dL and less than 40 mg/dL for HDL cholesterol, less than 120 mg/dL and 145 mg/dL or greater for non-HDL cholesterol, less than 90 mg/dL and 130 mg/dL or greater for triglycerides, less than 110 mg/dL and 130 mg/dL or greater for LDL cholesterol, and less than 90 mg/dL and 110 mg/dL or greater for apolipoprotein B.
Apolipoprotein B levels also were classified as concordant or discordant with cholesterol levels using a published method.18 Linear regression of apolipoprotein B on non-HDL cholesterol (or LDL cholesterol, secondarily) was performed for the entire sample (2007-2014), and the difference between each individual’s observed and regression-predicted apolipoprotein B levels (ie, [observed apolipoprotein B] − [regression-predicted apolipoprotein B]) was calculated as the individual’s residual value.
Residual values of −5 to 5 mg/dL identified concordance. More negative residual values identified discordance with low apolipoprotein B level (favorable profile). More positive residual values identified discordance with high apolipoprotein B level (unfavorable profile), which is a pattern that has been associated with the presence of small, dense LDL particles and elevated cardiovascular disease risk.8
Survey procedures in SAS version 9.3 (SAS Institute Inc) were used to account for the complex NHANES design. We applied survey weights to generate US population–level estimates. We calculated means (geometric means for triglycerides, given its skewed distribution) and proportions of youths with ideal and adverse levels and apolipoprotein B and cholesterol concordance or discordance for each survey period, using multivariable linear regression models to adjust for changes in population structure over time.
For the primary analyses, the means and proportions were calculated by sex and age strata, adjusting for continuous age and race/ethnicity. In the secondary analyses, the means and proportions were calculated by race/ethnicity strata adjusting for age and by BMI strata adjusting for age and race/ethnicity. Data for strata with inadequate sample sizes13 (other Hispanic, other race, underweight) were not analyzed. The decision to stratify was made a priori based on previously reported differences in lipid levels among these subgroups11,12 (ie, interaction analyses were not performed).
For the analyses of temporal trends, each 2-year NHANES cycle constituted a period; however, when the sample sizes were inadequate,13 the NHANES cycles were combined into 2 equal periods and simple change was analyzed. We tested for linear temporal trends using age- and sex-specific multivariable regression models adjusting for age and race/ethnicity, and fitting each NHANES cycle as an ordinal variable scaled as a 2-year interval so that regression coefficients (β) could be reported per 1 year. Trends for triglyceride levels were tested using log-transformed values because of the skewed distribution for triglycerides.
In the secondary analyses, we tested for trends in race/ethnicity and BMI strata. In a sensitivity analysis, we repeated our primary trend analyses, treating 2-year NHANES cycle as a categorical variable with the earliest cycle as the referent, and using orthogonal-contrast matrices to test for linear or quadratic trends. The results from the analyses using this method were similar overall to our primary analyses and are not presented. For all analyses, hypotheses were tested using an α level of .05 based on a 2-tailed test, with no adjustment for multiple comparisons. Because this increased the potential for type I error, we considered the secondary (subgroup) analyses to be exploratory.
To address missing data, we calculated the number and proportion of examined youths in each cycle with any missing data for levels of lipids or apolipoprotein B. We then compared demographics and BMI between youths with and without missing values. To further assess the effects of missing data, we used the SAS SURVEYIMPUTE procedure to impute missing levels for total cholesterol, which was the lipid with the highest proportion of missing data (up to 23% for some age groups and NHANES cycles). The resulting estimates were essentially identical and the conclusions were unchanged; therefore, reported estimates are from the analyses excluding missing data, consistent with prior reports of lipid levels in US youths and adults.11,12,19
A total of 26 047 youths aged 6 to 19 years participated in an NHANES examination from 1999 to 2016 and were eligible for inclusion (weighted mean age, 12.4 years; female, 51%). Sample sizes and characteristics varied between cycles according to sampling strategy (eg, intentional oversampling of adolescents in 1999-2006 and of non-Hispanic Asians in 2011-2016; eTable 1 in the Supplement).
Across NHANES cycles, 17% to 25% of children (aged 6-11 years) and 10% to 12% of adolescents (aged 12-19 years) were missing data on HDL or total cholesterol, and 15% to 19% of adolescents eligible for the analyses of triglycerides, LDL cholesterol, and apolipoprotein B were missing data for at least 1 of these measurements. Compared with youths with complete data, youths with missing data tended to be younger, were less often Mexican, and had lower BMI percentiles for some cycles, but sex distribution was similar (eTable 2 in the Supplement).
Table 1 and the Figure show age- and race/ethnicity-adjusted mean levels and trends for levels of lipids and apolipoprotein B. In all youths (aged 6-19 years), total cholesterol level decreased linearly during the 18-year period from 164 mg/dL (95% CI, 161 to 167 mg/dL) in 1999-2000 to 155 mg/dL (95% CI, 154 to 157 mg/dL) in 2015-2016 (β, −0.6 mg/dL [95% CI, −0.7 to −0.4 mg/dL] per year). The contributions to total cholesterol made by HDL and non-HDL cholesterol were in opposite directions, but both were favorable from the standpoint of risk for atherosclerosis. The adjusted mean HDL cholesterol level increased from 52.5 mg/dL (95% CI, 51.7 to 53.3 mg/dL) in 2007-2008 to 55.0 mg/dL (95% CI, 53.8 to 56.3 mg/dL) in 2015-2016 (β, 0.2 mg/dL [95% CI, 0.1 to 0.4 mg/dL] per year). The adjusted mean non-HDL cholesterol level decreased from 108 mg/dL (95% CI, 106 to 110 mg/dL) in 2007-2008 to 100 mg/dL (95% CI, 99 to 102 mg/dL) in 2015-2016 (β, −0.9 mg/dL [95% CI, −1.2 to −0.6 mg/dL] per year).
Among adolescents, the decline in non-HDL cholesterol level was accounted for by declines in both levels of triglycerides and LDL cholesterol. The adjusted geometric mean for triglycerides declined from 78 mg/dL (95% CI, 74 to 82 mg/dL) in 1999-2000 to 63 mg/dL (95% CI, 58 to 68 mg/dL) in 2013-2014, a mean decrease of 1.5% per year (log-transformed β, −0.015 [95% CI, −0.020 to −0.010] per year). The adjusted mean LDL cholesterol level declined from 92 mg/dL (95% CI, 89 to 95 mg/dL) in 1999-2000 to 86 mg/dL (95% CI, 83 to 90 mg/dL) in 2013-2014 (β, −0.4 mg/dL [95% CI, −0.7 to −0.2 mg/dL] per year). The adjusted mean apolipoprotein B level declined from 70 mg/dL (95% CI, 68 to 72 mg/dL) in 2005-2006 to 67 mg/dL (95% CI, 65 to 70 mg/dL) in 2013-2014 (β, −0.4 mg/dL [95% CI, −0.7 to −0.04 mg/dL] per year).
In sex- and age-specific analyses, favorable total cholesterol trends were statistically significant in all subgroups (Table 1 and the Figure). The temporal trends for non-HDL cholesterol level reached statistical significance in both sexes, in children (aged 6-8 years and aged 9-11 years), and in some adolescent subgroups. The temporal trends for HDL cholesterol level reached statistical significance in both sexes and in both age groups of children. The temporal trends for triglyceride level were statistically significant in all subgroups of adolescents, whereas the temporal trends for LDL cholesterol level reached statistical significance in both age subgroups and in males. The temporal trends for apolipoprotein B level did not reach statistical significance in any subgroup.
The age- and race/ethnicity-adjusted trends in the prevalence of ideal levels of lipids and apolipoprotein B appear in Table 2 and Table 3. Among youths overall, the prevalence of ideal levels of HDL, non-HDL, and total cholesterol each increased linearly over time. Among adolescents, the prevalence of ideal levels of triglycerides and LDL cholesterol also increased linearly, whereas the prevalence of ideal levels of apolipoprotein B remained high without statistically significant change. In age- and sex-specific analyses, temporal trends varied in statistical significance for HDL cholesterol level (significant only in females aged 6-11 years), non-HDL cholesterol level (no clear pattern), and LDL cholesterol level (significant only in 12- to 15-year-olds and males). Otherwise, temporal trends were statistically significant among all subgroups for levels of total cholesterol and triglycerides and among no subgroups for levels of apolipoprotein B.
Among all youths, the proportions with all ideal levels of HDL, non-HDL, and total cholesterol increased by a mean of 1.1% (95% CI, 0.6% to 1.5%) per year up to 51.4% (95% CI, 48.5% to 54.2%) in 2015-2016. Among adolescents, the proportion with ideal levels of all lipids and apolipoprotein B was 39.6% (95% CI, 33.7% to 45.4%) in 2007-2008 and 46.8% (95% CI, 40.9% to 52.6%) in 2013-2014, but the linear trend did not reach statistical significance (β, 1.0% [95% CI, −0.2% to 2.3%] per year).
The age- and race/ethnicity-adjusted changes in the prevalence of adverse levels of lipids and apolipoprotein B from earlier to later periods appear in Table 4. Among all youths, the prevalence of adverse levels decreased significantly for non-HDL and total cholesterol, whereas the difference did not reach statistical significance for HDL cholesterol. Among adolescents, the prevalence of adverse levels decreased significantly for triglycerides but nonsignificantly for LDL cholesterol and apolipoprotein B.
The proportion of youths with at least 1 adverse level of HDL, non-HDL, or total cholesterol decreased significantly for all youths from 23.1% to 19.2% (P = .002) and for children from 22.2% to 15.2% (P < .001), but not for adolescents (from 23.7% to 21.8%; P = .27). The proportion of adolescents with at least 1 adverse level of total cholesterol, HDL cholesterol, non-HDL cholesterol, triglycerides, LDL cholesterol, or apolipoprotein B did not significantly change from 25.4% to 25.2% (P = .92).
The age- and race/ethnicity-adjusted changes in the prevalence of apolipoprotein B concordance and discordance with non-HDL and LDL cholesterol from earlier to later periods appear in eTable 3 in the Supplement. Among all adolescents, no significant changes were detected. The decreases in discordance with a high level of apolipoprotein B for a given level of non-HDL or LDL cholesterol (unfavorable profile) reached statistical significance in a few subgroups (eg, vs non-HDL cholesterol in 16- to 19-year-olds; P = .04), but sample sizes were limited.
The adjusted trends for levels of lipids and apolipoprotein B by race/ethnicity and BMI category appear in eTables 4-6 in the Supplement. In these exploratory analyses, temporal trends in mean levels and in the prevalence of ideal and adverse levels were generally directionally consistent across racial/ethnic groups and BMI categories, but variable in magnitude and in whether statistical significance was reached.
Between 1999 and 2016, favorable temporal trends were observed in mean levels and in the distribution of ideal and adverse levels of HDL, non-HDL, and total cholesterol among youths aged 6 to 19 years. Among adolescents, favorable trends were observed for levels of triglycerides, LDL cholesterol, and apolipoprotein B. Despite these favorable trends, it was estimated at the end of the study period that 47% to 51% of US youths had all lipids at ideal levels and 19% to 25% had at least 1 adverse level. In the exploratory subgroup analyses, trends for levels of lipids and apolipoprotein B were generally directionally consistent across racial/ethnic groups and BMI categories.
In the context of published analyses,11,12 these results demonstrate a continuation of the favorable trends in mean lipid levels and prevalence of adverse lipid levels in US youths from 1988-1994 to 2007-2010. The findings also align with those from a report showing favorable lipid trends in US adults from 1999 to 2014.19 Given evidence that lipid levels track from youth to adulthood,20,21 these findings warrant some optimism about the future of atherosclerotic cardiovascular disease related to dyslipidemia. For example, the 13-mg/dL reduction in mean total cholesterol level among US adults from 1980 to 2000 was estimated to have resulted in the prevention or postponement of 82 830 deaths from coronary heart disease.22 In the present analysis, mean total cholesterol level among US youths declined by 9 mg/dL from 1999 to 2016.
Nevertheless, at the end of the study period, the proportion of youths with adverse levels of lipids or apolipoprotein B remained substantial (19%-25%), and the proportion with all ideal levels of lipids and apolipoprotein B was relatively low (47%-51%). An ideal cholesterol level constitutes 1 of 7 factors that define “ideal cardiovascular health,” a construct introduced by the American Heart Association with its 2020 strategic impact goal.1,2 This positive construct emphasizes primordial prevention (ie, prevention of the development of risk factors proximal to disease) as a populationwide strategy of particular relevance to youths. Ideal cardiovascular health is associated with lower prevalence and extent of subclinical atherosclerosis in youths23,24 and lower incidence of cardiovascular disease, cancer, and all-cause mortality in adults.25,26 The current findings provide a baseline for efforts to increase the prevalence of ideal lipid levels in youths.
To our knowledge, this study is also the first to analyze population-level apolipoprotein B trends in US youths. During the atherosclerotic disease process, retention of apolipoprotein B–containing particles within the arterial wall is a fundamental step that is driven in large part by apolipoprotein B–containing particle concentration.8 Particle concentration can be measured by serum apolipoprotein B level because apolipoprotein B has a 1:1 relationship with its lipoprotein particles. Because lipoprotein particles may carry average, high, or low amounts of cholesterol, apolipoprotein B levels may be discordant with non-HDL or LDL cholesterol levels (measured by weight, not particle number) in a given individual.
Although the optimal clinical use of apolipoprotein B is not yet clear, apolipoprotein B levels have been shown to be superior to traditional lipid levels for predicting atherosclerosis and cardiovascular disease events,18,27 particularly in younger adults with discordance between levels of apolipoprotein B (or particle number) and LDL cholesterol. Discordance is most common among individuals with obesity, the obesity-associated dyslipidemia of high triglycerides and low HDL cholesterol, and dysglycemia.28 In this context, the favorable trend in mean apolipoprotein B levels and the absence of an unfavorable trend in discordance with high apolipoprotein B levels observed in this study were unexpected, given known increases in childhood obesity29 and type 2 diabetes30 prevalence over a similar period. Similarly, it was unexpected that levels of triglycerides improved so substantially in light of the obesity trends.
It will be important to understand the reasons for the favorable lipid trends observed in this study to both acknowledge public health successes and plan future efforts. A wide variety of factors has been associated with lipid and apolipoprotein B levels, such as diet,31 physical activity,32 smoking,33 body size,34 pubertal stage,35 medications,36 and glycemia.37 Among US youths, some of these factors may be improving (eg, decreased trans fats in the food supply10), others are worsening (eg, BMI38), and others may be changing in unknown ways.
The strengths of this study include generalizability because of its use of nationally representative data and the comprehensiveness of the lipid and apolipoprotein B assessment using both continuous and categorical measures.
This study also has several limitations. First, blood was drawn in the nonfasting state for adolescents (aged 12-19 years) who attended afternoon or evening examinations as well as all children (aged 6-11 years). This would not have affected the measurement accuracy of HDL, non-HDL, or total cholesterol17,39; conversely, fasting was required for the measurement of triglycerides, LDL cholesterol (calculated using triglycerides), and apolipoprotein B; therefore, the sample size was limited. Second, the study periods for HDL and non-HDL cholesterol and apolipoprotein B were truncated due to potential bias or data unavailability during certain NHANES cycles. For HDL and non-HDL cholesterol, prior analyses of lipid levels from 1988-1994 to 2007-201011 provide some continuity with the present analyses of lipids from 2007-2008 to 2015-2016.
Third, because of the positive bias in HDL cholesterol for 1999-2006,15 LDL cholesterol levels for 1999-2006 may have been reciprocally negatively biased through the use of the Friedewald equation; however, such an effect would have biased the favorable trend results toward the null. Fourth, because LDL cholesterol level was considered missing for samples with triglyceride levels greater than 400 mg/dL, population LDL cholesterol levels may have been underestimated if samples with triglyceride levels greater than 400 mg/dL also had elevated LDL cholesterol levels; however, exclusion of LDL cholesterol data for this reason occurred for less than 1% of eligible samples. Fifth, youths using lipid-lowering medications were not excluded; however, this was rare (self-reported by 1 participant aged 16-19 years in 2013-2014 and 1 in 2015-2016).
Sixth, subgroup analyses must be interpreted with caution due to limitations in power related to sample sizes and risk of false-positive results related to multiple testing. Formal interaction testing was not performed due to concerns about type II errors, and any comparisons are exploratory and qualitative. Seventh, despite its importance for lipid profiles in youths, pubertal staging was unavailable, and even short age intervals are not perfect proxies for development. Eighth, pregnant youths were not excluded because pregnancy status was unavailable in the public use data set for individuals younger than 20 years; however, the proportion of girls younger than 20 years who were pregnant was less than 1% per NHANES documentation.40
Between 1999 and 2016, favorable trends were observed in levels of lipids and apolipoprotein B in US youths aged 6 to 19 years.
Corresponding Author: Amanda M. Perak, MD, MS, Department of Preventive Medicine, Northwestern University, 680 N Lake Shore Dr, Ste 1400, Chicago, IL 60611 (firstname.lastname@example.org).
Accepted for Publication: April 3, 2019.
Author Contributions: Drs Perak and Ning had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors approved the final manuscript as submitted and agreed to be accountable for all aspects of the work.
Concept and design: Perak, de Ferranti, Van Horn, Lloyd-Jones.
Acquisition, analysis, or interpretation of data: Perak, Ning, Kit, Van Horn, Wilkins, Lloyd-Jones.
Drafting of the manuscript: Perak, Lloyd-Jones.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Ning.
Obtained funding: Perak.
Administrative, technical, or material support: Van Horn.
Supervision: Van Horn, Wilkins, Lloyd-Jones.
Conflict of Interest Disclosures: Dr Wilkins reported receiving consulting fees from NGM Biopharmaceuticals. No other disclosures were reported.
Funding/Support: Dr Perak’s work was supported in part by training grant T32 HL069771 from the National Institutes of Health, a pediatric physician-scientist research award from the Department of Pediatrics, Northwestern University Feinberg School of Medicine, and career development award K23 HL145101 from the National Heart, Lung, and Blood Institute. Dr Wilkins’s work was supported in part by career development award K23 HL33601 from the National Heart, Lung, and Blood Institute. Under intra-agency agreement A-HL-17-001, the National Heart, Lung, and Blood Institute and the National Institutes of Health provided funding for select measures described in this article.
Role of the Funder/Sponsor: The National Institutes of Health had a role in the review and approval of the manuscript; however, the National Institutes of Health and other funding organizations had no role in design and conduct of the analysis; collection, management, analysis, and interpretation of the data; preparation of the manuscript; or the decision to submit the manuscript for publication.
Disclaimer: The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute, the National Institutes of Health, or the US Department of Health and Human Services.
Create a personal account or sign in to: