Lipid Screening in Childhood and Adolescence for Detection of Multifactorial Dyslipidemia: Evidence Report and Systematic Review for the U.S. Preventive Services Task Force

Abstract

Importance:

Multifactorial dyslipidemia, characterized by elevated total cholesterol (TC) or low-density lipoprotein cholesterol (LDL-C), is associated with dyslipidemia and markers of atherosclerosis in young adulthood. Screening for dyslipidemia in childhood could delay or reduce cardiovascular events in adulthood.

Objective:

Systematically review the evidence on benefits and harms of screening adolescents and children for multifactorial dyslipidemia for the U.S. Preventive Services Task Force (USPSTF).

Data sources:

MEDLINE, Cochrane Central Register of Controlled Trials, and PubMed between January 1, 2005 and June 2, 2015; studies included in a previous USPSTF evidence report; and reference lists of relevant studies and ongoing trials. Surveillance was conducted through April 9, 2016.

Study Selection:

Fair- and good- quality studies in English of participants age 0–20 years.

Data Extraction and Synthesis:

Two investigators independently reviewed abstracts and full-text articles and extracted data into evidence tables. Results were qualitatively summarized.

Main Outcomes and Measures:

Dyslipidemia (TC≥200mg/dL or LDL-C≥130mg/dL) and atherosclerosis in childhood; myocardial infarction and ischemic stroke in adulthood; diagnostic yield (number of confirmed cases per children screened); harms of screening or treatment.

Results:

Screening of children for multifactorial dyslipidemia has not been evaluated in randomized clinical trials. Based on one observational study (n=6,500) and nationally representative prevalence estimates, the simulated diagnostic yield (initial screening yield x positive predictive value [from a study with confirmatory testing] = simulated diagnostic yield) of screening for elevated TC varies between 4.8% and 12.3% (higher in obese children [12.3%], and at the ages when TC naturally peaks: 7.2% age 9 to 11, and 7.2% age 16 to 19). One good quality RCT (N=663) found a modest effect of intensive dietary counseling for a low-fat, low-cholesterol diet on lipid levels at one year in children age 8 to 10 with mild-to-moderate dyslipidemia; mean between-group difference in TC change from baseline was −6.1 mg/dL (95% CI, −9.1 to −3.2, p<.001). Between-group differences dissipated by year five. The intervention did not adversely affect nutritional status, growth, or development over 18 years. One observational study (N=9,245) found TC concentration at age 12–39 was not associated with death before age 55.

Conclusions and relevance:

The diagnostic yield of lipid screening varies by age and BMI. No direct evidence was identified for benefits or harms of childhood screening or treatment on outcomes in adulthood. Intensive dietary interventions may be safe with modest short-term benefit of uncertain clinical significance.

INTRODUCTION

Elevations in total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) are precursors to atherosclerosis and coronary heart disease (CHD). Identifying and treating dyslipidemia in adults over age 40 is common clinical practice in the United States.¹ TC and LDL-C concentrations in healthy children vary with age: they are low at birth, increase until age 2, peak before puberty, decrease during adolescence, and increase again during late adolescence and young adulthood.² TC and LDL-C concentrations are generally higher in girls and peak about 1 year earlier than in boys.²

Childhood dyslipidemia is commonly defined as TC≥200mg/dL or LDL-C≥130mg/dL using fixed cut points derived from population norms.³ Childhood dyslipidemia is typically multifactorial, with the exception of lipid disorders of genetic etiology characterized by very high lipid concentrations, such as familial hypercholesterolemia (FH). Multifactorial dyslipidemia may be associated with environmental and behavioral factors,^4,5 as well as obesity,^6–8 with or without inherited susceptibility.

Dyslipidemia in childhood and adolescence is not a disease, but is a risk factor for atherosclerosis and may contribute to CHD in adulthood.⁹ However, lipid measurement in youth imperfectly identifies adults with dyslipidemia; elevated LDL-C in adolescence (ages 12–18) has a positive predictive value of only 32.9% to 37.3% for dyslipidemia 15 to 20 years later¹⁰ and lower for children younger than age 12.¹¹

Screening youth for dyslipidemia may have the potential to identify affected youth, reduce long-term cholesterol burden through intervention, and prevent or delay cardiovascular events in adulthood. However, in 2007 the United States Preventive Services Task Force (USPSTF) found insufficient evidence (I recommendation) to recommend for or against routine selective or universal lipid screening of children or adolescents.¹² The purpose of this evidence report was to assist the USPSTF in updating its previous recommendations on screening children and adolescents for multifactorial dyslipidemia, defined as elevations in TC or LDL-C not due to FH.

METHODS

Scope of Review

The Agency for Healthcare Research and Quality (AHRQ) commissioned two systematic evidence reviews to support the USPSTF in updating its 2007 recommendation statement on screening for lipid disorders in children. This review focuses on benefits and harms of screening for and treatment of multifactorial dyslipidemia in children and youth aged 0 to 20 years. A separate systematic review to update the 2007 USPSTF recommendations on heterozygous FH.¹³

Using USPSTF methods,¹⁴ an analytic framework and eight key questions (KQs) were developed to assess evidence on the effect of screening and treatment on intermediate outcomes, health outcomes, and harms; the diagnostic yield of screening, and the association between intermediate outcomes in childhood and adult health outcomes (Figure 1). Adult health outcomes of interest were MI and ischemic stroke. Intermediate outcomes included lipid concentrations (TC, LDL-C, HDL-C, non-HDL-C and TG) and atherosclerosis markers (carotid intima-medial thickness [CIMT], calcium score, and autopsy findings).

Figure 1.

Analytic framework and key questions

^a Intermediate outcomes include lipid levels (total, LDL, HDL, and non-HDL cholesterol; triglycerides) and atherosclerosis markers (carotid intima-media thickness, calcium score, pathological findings).

Abbreviations: HDL = high-density lipoprotein; LDL = low-density lipoprotein; HDL = high-density lipoprotein; MI = myocardial infarction.

Evidence reviews for the USPSTF use an analytic framework to visually display the key questions (KQs) that the review will address to allow the Task Force to evaluate the effectiveness and safety of a preventive service. The questions are depicted by linkages that relate interventions and outcomes. Further details are available from the USPSTF procedure manual.⁴³

Screening Key Questions

• 1
  Does screening for multifactorial dyslipidemia in asymptomatic children and adolescents delay or reduce the incidence of MI or stroke in adulthood?
• 2
  Does screening for multifactorial dyslipidemia in asymptomatic children and adolescents improve intermediate outcomes (i.e., improve lipid concentrations or reverse or slow the progression of atherosclerosis) in childhood and adolescence?
• 3
  What is the diagnostic yield of screening for multifactorial dyslipidemia in children and adolescents?
• 4
  What are the harms of screening for multifactorial dyslipidemia in children and adolescents?

Treatment Key Questions

• 5
  Does treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents delay or reduce the incidence of adult MI and stroke events?
• 6
  Does treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents improve intermediate outcomes (i.e., reduce lipid concentrations or reverse or slow the progression of atherosclerosis) in childhood and adolescence?
• 7
  What are the harms of treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents?

Outcomes Key Question

• 8
  What is the association between intermediate outcomes in childhood and adolescence and future incidence of MI and stroke events in adults?

FH and other monogenic conditions were excluded from this review, as were renal, infectious, hepatic, inflammatory and storage disorders, types 1 and 2 diabetes, and several other syndromes that confer secondary risk for elevated LDL-C or TC. Detailed study methods and a list of excluded studies, including reasons, are listed in the full report.¹⁵

Data Sources and Searches

A literature search was conducted using several databases, including MEDLINE and PubMed, BMJ Clinical Evidence, Canadian Agency for Drugs and Technologies in Health, Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, Health Technology Assessment (Centre for Reviews and Dissemination), Institute for Clinical Systems Improvement, Institute of Medicine, and National Institute for Health and Clinical Excellence. The search included studies published January 1, 2005 or later. The original search was conducted on February 12, 2014, and updated on June 13, 2014, December 16, 2014 and June 2, 2015. Since June 2015, we continued to conduct ongoing surveillance through article alerts and targeted searches of high-impact journals to identify major studies published in the interim that may affect the conclusions or understanding of the evidence and therefore the related USPSTF recommendation. The last surveillance was conducted on April 9, 2016, and identified no new relevant studies. The search strategies are listed in the online supplement.

All studies included in the previous USPSTF evidence report⁹ were reviewed, along with the reference lists of several reports, including the 2011 NHLBI Expert Panel Report,¹⁶ publications from large cohort studies with longitudinal data, and studies included in other relevant systematic reviews and meta-analyses. Relevant articles were solicited from expert reviewers and Clinicaltrials.gov was searched to identify relevant ongoing trials.

Study Selection

All study selection procedures used dual, independent review. The title and abstracts and reviewed, followed by full text of all potentially relevant citations, against the inclusion and exclusion criteria for design, population, screening, intervention, outcomes, and setting. Discrepancies were resolved through discussion.

The screening population of interest was asymptomatic people age 0 to 20 years. Eligible screening interventions were defined as a lipid panel (fasting or non-fasting lipid measurement, TC or LDL-C alone or in combination with HDL) delivered in a universal or selective screening strategy. Although non-HDL-C was not among the included screening interventions, no studies were excluded that screened youth using non-HDL-C. Screening studies with modalities not relevant to primary care practice were excluded.

The treatment population of interest was people with multifactorial dyslipidemia (ideally screen-detected) age 0 to 20 who were treated with lipid-lowering drugs or lifestyle interventions. All reported clinical and laboratory harms associated with interventions that had any evidence of treatment benefit were included.

Studies of efficacy or effectiveness were limited to fair or good-quality randomized clinical trials that were conducted in countries with a United Nations Human Development Index¹⁷ greater than 0.9. Studies conducted in very high HDI countries are more likely to be applicable to US settings. Trials, cohort studies, and observational studies that reported clinical or laboratory harms were included; case series and case reports were excluded.

Simulation of Diagnostic Yield

For the diagnostic yield question (KQ3), screening studies initially were required to include confirmatory testing, to allow calculation of proportion of persons screened who were confirmed cases. However, because only one Ohio-based study met all inclusion criteria for KQ3, studies were included if they involved large US populations and reported results of lipid screening (using a TC concentration of 200 mg/dL or greater, an LDL-C concentration of 130 mg/dL or greater, or both on a single occasion to define dyslipidemia. Combining the prevalence of elevated TC concentrations with the positive predictive value from the Ohio study allowed for computation of a simulated diagnostic yield (initial screening yield × positive predictive value (PPV) of the initial screen = diagnostic yield).

Data Extraction and Quality Assessment

Two reviewers independently critically appraised articles meeting inclusion criteria as good, fair, or poor in accordance with USPSTF guidance (eTable 1).¹⁴ Poor-quality studies were those with important limitations that could invalidate study findings and were excluded from this review.

For all included articles, data were abstracted into evidence tables, including study characteristics, study design elements, test characteristics for screening studies, intermediate and health outcomes and harms, including all relevant subgroups where available.

Data Synthesis and Analysis

Summary tables of study characteristics, population characteristics, intervention characteristics, and outcomes were created separately for each KQ. For treatment studies, lipid concentrations were expressed as percent change or difference from baseline. Data were not combined across treatment studies. No key questions had a sufficient number of included studies to permit meta-analysis.

RESULTS

A total of 7,137 unique abstracts and 537 full-text articles were reviewed (eFigure). Of these, 16 articles were included. These include four screening studies (eight articles),^7,8,18–23 2 treatment studies (five articles),^24–28 one study on treatment harms (5 articles),^26–30 and one study (one article) of association between intermediate and health outcomes.³¹ Three articles were included for both KQ6 and KQ7.

Screening and Health Outcomes

KQ1: Does screening for multifactorial dyslipidemia in asymptomatic children and adolescents delay or reduce the incidence of MI or stroke in adulthood?

No studies were identified.

KQ2: Does screening for multifactorial dyslipidemia in asymptomatic children and adolescents improve intermediate outcomes (i.e., improve lipid levels or reverse or slow the progression of atherosclerosis) in childhood and adolescence?

No studies were identified.

Diagnostic Yield of Screening

KQ3: What is the diagnostic yield of screening for multifactorial dyslipidemia in children and adolescents?

Three fair-quality studies,^7,18,22,23 and one good-quality study (NHANES)^8,19–21 met the inclusion criteria (Table 1). Only one study used confirmatory testing to establish dyslipidemia.¹⁸ In a large pediatric group practice setting, 6,500 people age 3 to 18 (mean 6.4) presenting for well child care between 1986 and 1988 provided a non-fasting blood sample. Those with non-fasting TC concentrations of 200 mg/dL or greater returned for a fasting lipid profile. The prevalence of elevated non-fasting TC concentrations was 8.5%. Of these, 88% returned for a second screen after 1 to 6 weeks, and 77% of these had a fasting LDL-C concentration of 130 mg/dL or more (PPV=77%). The diagnostic yield was 5.8%.

Table 1.

Included studies by key question

Author, year	Study family	Qualityb	Country	N	Study design	Mean age, years (range)	Population and setting Intervention (for trials)	Outcomes assessed	Dates of data collection
Screening studies (KQ3)
Garcia 198918	--	Fair	U.S. (Ohio)	6,500	Cross-sectional	6.4 (3–18)	White, middle-class children presenting for well-child care at a large group practice. 6,500 (100%) of population was White.	Non-fasting blood sample followed by fasting lipid profile if TC ≥200 mg/dL. LDL-C ≥130 mg/dL was considered elevated.	1986–1988
Kit 20158	NHANES	Good	U.S.	13,172	Cross-sectional	(8–17)	Nationally representative U.S. samplec; Female participants ranged from 741/1558 (47.6%) in 2009–2010 to 1173/2333 (50.3%) in 2001–2002	Single lipid assessment. Non-fasting TC ≥200 mg/dL or fasting LDL-C ≥130 mg/dL were considered elevated.	1999–2012
Kit 201219	NHANES	Good	U.S.	16,116	Cross-sectional	(6–19)	Nationally representative U.S. samplec; Female participants ranged from 2009/4205 (47.8%) in 2007–2010 to 2954/5759 (51.3%)	Single lipid assessment. Non-fasting TC ≥200 mg/dL or fasting LDL-C ≥130 mg/dL were considered elevated.	1988–2010
CDC 201020	NHANES	Good	U.S.	3,125	Cross-sectional	(12–19)	Nationally representative U.S. samplec, Female participants=1491 (47.7%)	Single lipid assessment. Non-fasting TC ≥200 mg/dL or fasting LDL-C ≥130 mg/dL were considered elevated.	1996–2006
Ford 200921	NHANES	Good	U.S.	9,868	Cross-sectional	(6–17)	Nationally representative U.S. samplec, Female participants=4907 (49.7%)	Single lipid assessment. Non-fasting TC ≥200 mg/dL or fasting LDL-C ≥130 mg/dL were considered elevated.	1999–2006
Margolis 201422	--	Fair	U.S.	29,360	Retrospective analysis of automated medical records	NRd (3–19)	Youth with ≥1 visit during the study period in 3 large US health systems. Female participants=15404 (52.5%). Race and ethnicity for female participants included White:5974 (38.8%); Asian: 1008 (6.5%); Black: 1521 (9.9%); Hispanic: 3044 (19.8%); Other: 237 (1.5%); and Missing: 3620 (23.5%). Race and ethnicity distribution for male study participants is comparable	First TC measurement on record. TC ≥200 mg/dL was considered elevated.	2007–2010
Ice 2011;7 Ritchie 201023	CARDIAC	Fair	U.S. (West Virginia)	23,263	Cross-sectional	10.84 (NR)	Fifth grade students in all 55 counties of West Virginia. Population was 53.2% female. Race and ethnicity included White (93%), Bi-racial (2%), Black (3%), and Asian, Hispanic, and Other (<1% each).	Cardiac risk factor assessment with one-time fasting lipid panel (TC, LDL-C, HDL-C, and TG). Family history premature coronary heart disease before age 55 (assessed by parent survey). TC ≥200 mg/dL or LDL-C≥130 mg/dL were considered elevated.	2003–2008
Treatment studies (KQ6, KQ7)
Wong 201324e	--	Fair	Canada	32 IG: 16 CG: 16	Randomized clinical trial	13 (8–18)	Youth receiving care at a specialty lipid clinic, with fasting serum LDL-C concentration 135–193mg/dL, first-degree family history of hypercholesterolemia or premature atherosclerotic cardiovascular disease, and compliance with the NCEP Step II diet. Female participants=15 (46.9%). Intervention group received 30 g/day of ground flaxseed via specially prepared muffins and bread. Control group received placebo muffins and bread.	LDL-C, TC, and TG assessed at 4 weeks.	2009–2010
DISC 1993;25 DISC 1995;26 Obarzanek 1997;29 Lavigne 1999;30 Obarzanek 2001;27 Dorgan 201128	DISC	Good	U.S.	663 IG: 334 CG: 329	Randomized clinical trial	Boys: 9.7 (8–10); Girls 9.0 (8–10)	Pre-pubertal children with LDL-C between ≥ 80th and < 98th percentiles for age and sex recruited from public and private elementary schools. Female participants=301 (45.4%). Race and ethnicity included White=574 (86.6%); Black=56 (8.4%); Other=33 (5.0%). Intervention group received modified NCEP Step II diet delivered using a family-based counseling approach (years 0–3) followed by lower intensity dietary counseling (years 4–8). Control group received feedback to parent about child’s baseline cholesterol concentration, written heart-healthy diet materials.	LDL-C and TC, and potential harms evaluated at 1, 3, 5, 7–8, and 18 years after randomization.	1987– 2008
Outcomes study (KQ8)
Saydah 201331	--	Good	U.S.	9,245	Prospective study of NHANES participants with 12–18 year follow up	26.1 (12–39)	Population was 50.4% female. Race/ethnicity included Non-Hispanic white (77.2%), Non-Hispanic black (14.7%) Mexican American (8.1%). Baseline mean lipid values (SE) were total cholesterol: 182.0 (0.82); HDL cholesterol: 50.4 (0.37); Non-HDL cholesterol: 131.6 (0.90).	Mortality from endogenous and all-causes before age 55	1988–1994 to 2006

Abbreviations: KQ = key question, SD = standard deviation, NR = not reported, TC = total cholesterol, mg/dL = milligram(s) per deciliter, LDL-C = low density lipoprotein cholesterol, HDL-C = high density lipoprotein cholesterol, TG = triglycerides, IG = intervention group, CG = control group, NCEP = National Cholesterol Education Program, g = grams, NHANES = National Health and Nutrition Examination Survey

^aArticles in the same “study family” use the same study population or same data source to describe multiple results

^bQuality assessed using criteria developed by the U.S. Preventive Services Task Force¹⁴ and the Newcastle-Ottawa Scale⁴⁴

^cSample weights used to obtain prevalence estimates representative of the civilian non-institutionalized US population

^dThis study does not report mean age. The age distribution for the cohort is: 3–5 y: 2.7%; 6–8 y: 6.8%; 9–11 y: 15.4%; 12–16 y: 41.6%; 17–19 y: 33.4%.²²

^eWong 2013 study was evaluated for KQ6 only

The PPV of 77% for screening for elevated TC concentrations from the confirmatory testing study was combined with data from large recent studies to simulate population-based diagnostic yields. Three large population-based studies that used only a single lipid test (without confirmatory testing) and had screening yields of 8%–11% were used for this purpose.^7,8,19

Three studies reported the prevalence of elevated lipid concentrations by age and body mass index (BMI) subgroups. These studies defined “healthy weight” as BMI <85^th percentile for age and sex, “overweight” as BMI between 85^th and 95^th percentile for age and sex, and “obese” as BMI ≥95^th percentile for age and sex.

NHANES is a nationally representative sample survey.³² All four of the NHANES articles^8,19–21 included adolescents; two included children as young as 6–8 years. All used the National Cholesterol Education Program (NCEP) cut points of 200 mg/dL for TC and 130 mg/dL for LDL-C.³ Participants provided non-fasting samples for TC testing; 12-to-19 year olds also provided fasting samples for LDL-C testing. The highest rates of elevated TC were in ages 9 to 11 years (9.4%; 95% CI 7.3–11.6) and 16 to 19 years (9.4%; 95% CI 7.0–11.8);¹⁹ there were no significant differences in age-specific prevalence.^8,19,21 Prevalence of elevated LDL-C was higher in children with obesity (14.2%, 95% CI 10.2–19.6) than in healthy weight children (prevalence ratio 2.5; 95% CI, 1.6 to 3.8).²⁰

The CARDIAC study screened 23,263 West Virginia 5^th-grade students statewide between 2003 and 2008^7,23 for elevated lipid concentration and family history of early CHD or hypercholesterolemia. After a nonfasting screening test, the prevalence of elevated TC (200 mg/dL or greater) was 10.7%:⁷ 7.5 % for children with healthy weight, 11.5 % for overweight children, and 16.0 % for obese children. Prevalence of elevated LDL-C (greater than 130 mg/dL) was 8.7 % overall,⁷ 8.3 % in those with a family history of early CHD, and 9.5% in those without.²³

In a study of 3-to-19-year-olds enrolled in three large health systems in California, Colorado, and Minnesota,²² elevated TC concentrations using NCEP criteria were increased significantly with BMI: 10.7 (95% CI 10.2–11.3%) in the obese group, 8.6 (95% CI 7.8–9.4%) in the overweight group, and 6.7 (95% CI 6.2–7.1%) in healthy weight group. A similar association was found for LDL-C measurements.

Simulated diagnostic yields ranged between 4.8% and 12.3% for different age and BMI subgroups. The highest diagnostic yields appear to be in obese children (12.3%), and at the ages when TC naturally peaks: children age 9 to 11 (7.2%), and adolescents age 16 to 19 (7.2%).¹⁹ The subgroup with the highest estimated diagnostic yield was obese children age 10 to 11 (12.3%).⁷

Harms of Screening

KQ4: What are harms of screening for multifactorial dyslipidemia in children and adolescents?

No studies were identified.

Benefits of Treatment of Multifactorial Dyslipidemia

KQ5: Does treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents delay or reduce the incidence of adult MI and stroke events?

No studies were identified.

KQ6: Does treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents improve intermediate outcomes (i.e., improve lipid levels or reverse or slow the progression of atherosclerosis) in childhood and adolescence?

No studies meeting the inclusion criteria evaluated the effect of medications on intermediate outcomes in children or adolescents with multifactorial dyslipidemia. For lifestyle modification, one good quality trial (4 articles)^25–28 and 1 fair quality treatment trial (1 article)²⁴ were included (Table 1). A four-week trial tested the effect of flaxseed supplementation of 30 g per day delivered through specially prepared muffins and bread. The control group received placebo muffins and bread.²⁴ Compared to placebo, flaxseed was associated with lower, but not statistically significant, TC concentrations (relative decrease of −4%, 95% CI −10% to 2%, p=.20) and LDL-C concentrations (relative decrease of −5%, 95% CI −12% to 2%, p=.15). Flaxseed also was associated with lower HDL-C concentrations relative to placebo (mean percent change −15%, 95% CI −24% to −6%, p=0.001).

The Dietary Intervention Study in Children (DISC) was an RCT of a modified NCEP Step II (low fat, low cholesterol) diet³ with a multi-year intervention and long-term followup of children with LDL-C between 80^th and 98^th percentiles for age and sex (age 8–10 at baseline, n=663). During the first 6 months, participants had group visits and individual family visits, followed by a 2.5-year maintenance phase to support adoption of the low fat diet. Parents of control group participants received feedback about the child’s cholesterol concentration and educational materials. Children in the intervention group received dietary counseling at a lower intensity until about 8 years post-randomization.²⁷ A subset of 230 female participants was assessed again as adults, about 18 years post-randomization.²⁸

Adherence to the diet during the intervention period was good,²⁶ and the children in the intervention group had improved dietary quality.³³ Small, statistically significant decreases in LDL-C and TC relative to the control group were seen in participants randomized to the dietary intervention at years 1 and 3. Adjusted mean between group differences in change from baseline at one year were −6.1 mg/dL (95% CI, −9.1 to −3.2, p<.001) for mean TC and −4.8 mg/dL (95% CI, −7.4 to −2.2, p<.001) for mean LDL-C. The groups did not differ significantly at Year 5 (LDL-C and TC),²⁹ year 7 (LDL-C and TC),²⁷ or Year 18 (LDL-C)²⁸ (Figure 2).

Figure 2.

Effect of dietary intervention (DISC) on mean total cholesterol in children (aged 8–10 at baseline)^a

^a Mean age of study participants was 9.5 years at Baseline and 17.0 years at Year 7. Mean age was not reported at other time points.

Notes:

Source: Obarzanek et al., 2001²⁷

Error bars represent 95% CIs

Median followup was 7.2 years (range: 6.5 to 9.3 years)

The numbers following IG (intervention group) and CG (control group) represent the number of participants in each study arm contributing data at each time point

In this same study, LDL-C showed a similar pattern to TC over time for both study groups.

Abbreviations: DISC = Dietary Intervention Study in Children, mg/dL = milligram(s) per deciliter, IG = intervention group, CG = control group

Harms of Treatment of Multifactorial Dyslipidemia

KQ7: What are the harms of treatment of multifactorial dyslipidemia with lifestyle modifications and/or lipid-lowering medications in children and adolescents?

Only the DISC trial was included for treatment harms (Table 1). Weight, height, and BMI were assessed at baseline and annually through the final visit in the original study and (for the subset of 230 women) at year 18.^27,28 Skin fold thickness and of mid-arm, waist, and hip circumferences were taken at several points. Sexual maturation was evaluated at every visit through year 5 or until the participant reached Tanner Stage 5. Age at menarche was recorded for females at year 7 and at year 18. Blood pressure was measured at baseline, year 1, year 3, and year 18. Laboratory measures, macronutrient and micronutrient intake were measured at several time points. Psychosocial assessments³⁰ were done at year 3 to assess social emotional development, depression/anxiety, academic development, and eating disorders. The DISC trial identified no harms of a modified NCEP Step II dietary intervention with behavioral counseling in children with multifactorial dyslipidemia.

Relationship Between Intermediate Outcome and Adult Health Outcomes

KQ8: What is the association between intermediate outcomes in childhood and adolescence and future incidence of MI and stroke events in adults?

One good-quality study met the inclusion criteria (Table 1).³¹ This study identified a cohort of 9,245 NHANES participants age 12–39 years and measured deaths before age 55 through linkage with the National Death Index. TC concentrations were greater than 200 mg/dL in 28.5% of the cohort and greater than 240 mg/dL in 7.6%. At the end of the study, 283 participants (3.1%) had died. Moderately elevated TC concentrations (200 to 239 mg/dL) were not significantly associated with a relative hazard (RH) for death before age 55 for males (RH for all causes: 0.75, 95% CI 0.42–1.37; RH for endogenous causes: 0.71, 95% CI 0.34–1.51) or for females (RH for all causes: 1.39, 95% CI 0.71–2.70; RH for endogenous causes 0.77, 95% CI 0.36–1.62). A very high TC concentration (≥240 mg/dL) was associated with a greater risk of death from endogenous causes (RH 2.58; 95% CI, 1.31 to 5.08) in females only.

DISCUSSION

Screening strategies proposed for dyslipidemia have included both selective and universal screening. Several expert groups recommend screening based on family history, primarily for detecting FH.^3,16,34,35 A 2011 guideline from the National Heart, Lung, and Blood Institute (NHLBI) recommended universal screening at 9 to 11 years and again at 17 to 21 years, with selective screening at other ages.¹⁶ Elevated LDL-C and TC concentrations are associated with higher BMI^6–8,36 and waist circumference,^6,36 suggesting a rationale for selective screening. However, BMI is most strongly associated with elevated triglyceride (TG) concentration, which, not being atherogenic, is not included in the definition of multifactorial dyslipidemia. Current rates of dyslipidemia screening in children and adolescents are low, estimated between 2.5 % and 3.2%.³⁷ Current guidelines do not recommend lipid-lowering medications to treat multifactorial dyslipidemia in youth¹⁶ and pharmacologic treatment of 8-to-20-year-olds with lipid-lowering agents is rare.³⁸

This evidence report found that in population-based studies, 8%–11% of children and adolescents screened positive at the fixed NCEP cut point of ≥200 mg/dL with a single TC test, with higher rates in obese children and to a lesser extent in overweight children. After confirmatory testing, screening is estimated to result in diagnostic yields ranging from 4.8% to 12.3%, with higher yields in higher BMI subgroups, if screening were widely adopted (Table 2). However, no direct evidence was found for the effect of screening for multifactorial dyslipidemia on dyslipidemia and atherosclerosis in childhood or on myocardial infarction and ischemic stroke in adulthood.

Table 2.

Summary of evidence table

Key Question	Studies (k)	Observations (n)	Design	Major limitations	Consistency	Applicability	Overall quality	Summary of findings
KQ1 (screening & health outcomes)	0	0	N/A	N/A	N/A	N/A	N/A	N/A
KQ2 (screening & intermediate outcomes)	0	0	N/A	N/A	N/A	N/A	N/A	N/A
KQ3 (screening diagnostic yield)	k=4 (8 articles)	n=78,792a	Cross-sectional	In non-NHANES studies, populations were almost exclusively white. Only one study included confirmatory testing (n=6,500); all others used a single test so true diagnostic yield is not available	Taken together the results are internally consistent.	All studies were conducted in U.S. settings. Two were in U.S. primary care settings and are most directly applicable. One study was conducted in a school-based setting and is likely relevant to primary care. NHANES data may not have direct relevance to primary care.	Fair	The one study that included confirmatory testing found a diagnostic yield of 5.8%. Data from studies using a single test found simulated diagnostic yields of 4.8% to 12.3%.
KQ4 (screening harms)	0	0	N/A	N/A	N/A	N/A	N/A	N/A
KQ5 (treatment in childhood & health outcomes)	0	0	N/A	N/A	N/A	N/A	N/A	N/A
KQ6 (treatment in childhood & intermediate outcomes )	k=2 (5 articles)	n=695	RCT Longitudinal followup of one trial cohort to 18 years post-randomization	In the larger study (diet) the cutoffs used to define dyslipidemia were lower than the evidence report criteria. The smaller study (flaxseed) had a small sample size (n=32) and was limited to a high-risk population in a tertiary care setting.	N/A – different interventions	The diet study is likely applicable to a U.S. setting if the relatively high intensity of dietary counseling could be replicated in primary care. The flaxseed study has limited applicability to a primary care setting.	Fair	The diet study found lower LDL-C and TC levels at 1 and 3 year followup in the intervention group. On longitudinal followup of the trial cohort the treatment effects were attenuated. The study of flaxseed supplementation found no effect at 4 weeks on lipid levels.
KQ7 (treatment harms)	k=1 (5 articles)	n=663	RCT; Longitudinal followup of trial cohort to 18 years post-randomization	Cutoffs used to define dyslipidemia were lower than the evidence report criteria.	N/A – one study	The DISC study is likely applicable to a U.S. setting if the relatively high intensity of family counseling could be replicated in primary care.	Fair	No harms reported at any time point during the trial or on long-term followup for anthropometric, laboratory, psychosocial or maturation measures.
KQ8 (association: childhood intermediate outcomes & adult health outcomes)	k=1	n=9,245	Longitudinal analysis of NHANES data	Outcomes on children within the total study population (age 12–39) were not reported separately.	N/A – one study	NHANES data may not have direct relevance to primary care.	Fair	At followup 283 people were deceased before age 55. Leading causes of death for people age 12–19 at baseline were accident, self-inflicted injury, circulatory causes, and cancer. In multivariate models neither very high cholesterol (TC ≥240 mg/dL) nor moderately increased TC (200–239 mg/dL) was independently associated with death, when both sexes were combined. For females only a very high cholesterol was associated with a greater risk of death before 55 (RH 2.58; 95% CI 1.31, 5.08).

Abbreviations: KQ = key question, N/A = not applicable, NHANES = National Health and Nutrition Examination Survey, RCT = randomized clinical trial, LDL-C = low density lipoprotein cholesterol, TC = total cholesterol, mg/dL = milligram(s) per deciliter, DISC = Dietary Intervention Study in Children, RH = relative hazard, CI = confidence interval

^aN=78,792 is an estimate, as the four NHANES articles included for KQ3 have overlapping populations

No evidence was found on the harms of screening for multifactorial dyslipidemia in childhood. Since the majority of dyslipidemias identified in childhood do not develop clinically important lipid elevations or cardiovascular disease,^10,39 such “non-disease” could result in harms, such as labelling a child as “sick,” parent or child anxiety, and unnecessary or harmful testing and treatment. Further, the NCEP cut points are derived from population distributions,³ not from associations between lipid concentrations and clinical outcomes as they are for adults. Although widely accepted, these fixed cut points may over-identify children in the age groups currently targeted for screening, who naturally experience peaks in TC and LDL-C concentrations.

Evidence from one trial suggests that moderately dyslipidemic 8–10 year olds following a low-fat, low-cholesterol diet will experience a small decrease in lipid levels without evidence of harms (Table 2). The study diet closely resembled current macronutrient recommendations for children with elevated LDL-C≥130 mg/dL.¹⁶ The small beneficial effect of low fat, low cholesterol diet did not persist beyond three years and was consistent with the natural decrease in LDL-C concentration seen in early adolescence. The relatively high intensity of the counseling intervention limits its generalizability to primary care settings, where trained nutritional counselors may not be part of the health care team. In addition, the clinical importance of the small effect on cholesterol concentrations in a 3-year period is unclear. An RCT of flaxseed supplementation found no effect on TC concentrations; this trial included children with a family history of dyslipidemia so it is possible that some of the participants could have met criteria for FH.

Despite the clear link between lipid concentrations and CHD in adults, the association between elevated TC or LDL-C concentrations in youth and cardiovascular disease in adulthood is poorly understood. A single longitudinal study of adolescents and young adults from the NHANES database found no association between TC concentrations and death before age 55 years when men and women were combined, but did find that women who had an extremely high TC concentration (≥240 mg/dL) at age 12 to 39 years had a greater risk of death before age 55. This subgroup is likely dominated by women with familial hypercholesterolemia (FH), for whom premature CHD deaths are expected. The meaning of this finding is unclear because of the small number of deaths in this subgroup and since it represents both adolescence and young adulthood.

Research Needs

Large RCTs of lifestyle interventions with long-term followup would provide evidence to inform recommendations for management of children and youth with multifactorial dyslipidemia. Rigorous trials of dietary supplements and medications to reduce concentrations of atherogenic lipids in children and adolescents would also improve the body of evidence.

Evidence for the effect of screening in childhood on health outcomes in adulthood is notably lacking, and could be provided by large trials of both universal and selective screening strategies that use accepted cut points defining dyslipidemia, perform confirmatory testing, and include racially and ethnically diverse populations. Pediatric studies that screen for abnormal non-HDL or ApoB, which have emerged as strong markers of atherogenic risk in adults,^40–42 could also be useful. Long-term followup studies of pediatric cohorts into adulthood would provide evidence that would aid the understanding of the association between cholesterol concentrations in childhood with multifactorial dyslipidemia and adult health outcomes.

Though not a primary focus of this review, its findings suggest that the current commonly-accepted fixed cut points for defining dyslipidemia in childhood are imprecise given the normal, predictable, and well-documented fluctuations in lipid levels that occur during childhood and adolescence. Re-examination of cut-points that more precisely reflect the fluctuations of lipids during childhood, and their implications for screening, may be warranted.

Limitations

This review identified no screening studies with followup to adulthood, and only 2 RCTs of different dietary interventions. The larger trial targeted 8–10 year olds, so the effect of a dietary intervention on adolescents with dyslipidemia remains unknown. There was no evidence to support any other type of treatment for multifactorial dyslipidemia. This evidence report excluded FH, which is addressed in a companion review.¹³ However, it is worth noting that the two conditions are detected with the same screening test.

Conclusions

The diagnostic yield of lipid screening varies by age and BMI. No direct evidence was identified for benefits or harms of childhood screening or treatment on outcomes in adulthood. Intensive dietary interventions may be safe with modest short-term benefit of uncertain clinical significance.