 |
|
Primary Prevention of High Blood Cholesterol Concentrations in the United States
David C. Goff, Jr, MD, PhD;
Darwin R. Labarthe, MD, PhD;
George Howard, DrPH;
Gregory B. Russell, MS
Arch Intern Med. 2002;162:913-919.
ABSTRACT
| |
Background Mean concentrations of total cholesterol (TC) among adults have declined
in the United States for decades. Whether the decline has been owing to prevention
of high TC levels or treatment of high TC levels once present is not known.
Objective To determine whether population-wide influences and/or the high-risk
approach have been operating to produce the well-known decline in mean TC
concentration in the US population.
Methods We examined changes in the distribution of TC levels across US birth
cohorts as sampled in the National Health Examination Survey and the National
Health and Nutrition Examination Surveys I, II, and III. We tested the hypotheses
that the age-adjusted 10th, 25th, 50th, 75th, and 90th percentiles of TC levels
were lower in more recent US birth cohorts than in earlier cohorts.
Results Data were analyzed for 49 536 participants born between 1887 and
1975 and examined at ages 18 through 74 years between 1959 and 1994. The 10th,
25th, 50th, 75th, and 90th percentiles of TC levels (adjusted for age, race,
and sex) were estimated to be lower by 3.4, 3.9, 4.7, 5.7, and 7.1 mg/dL (0.09,
0.10, 0.12, 0.15, and 0.18 mmol/L), respectively, for every successive 10
years in date of birth (P<.001 for each estimate).
Conclusions The declines in TC levels associated with successive birth cohorts were
greater at the upper aspect of the distribution, probably because of the combination
of population influences and treatment effects. The differences seen at the
lower percentiles support the contention that a strong prevention effect occurred
in the US population from 1959 through 1994. Greater understanding of this
dramatic shift in the distribution of TC levels could support future prevention
efforts.
INTRODUCTION
MEAN SERUM total cholesterol (TC) concentrations have declined in the
United States during the past several decades.1
It has not been apparent whether this shift in mean TC concentration has been
owing to reductions in blood TC concentrations through broadly occurring behavioral
or environmental changes in the population, treatment of clinically recognized
high blood cholesterol concentrations, or a combination of these influences.
Rose2 contrasted the expected changes in the
population distribution of a risk factor in response to either (1) population-wide
influences, such as effective prevention efforts, or (2) treatment of the
subgroup of the population with the risk factor present. According to this
paradigm, population-wide influences would shift the entire cholesterol distribution
to lower levels by reducing contemporaneous cholesterol concentrations and/or
by reducing the rate of increase in cholesterol concentrations with aging.
The approach of treating those at the highest risk, with selective attention
to persons with high blood cholesterol levels, could affect only the upper
aspect of the cholesterol distribution (by reducing the cholesterol concentrations
of only those persons selected for treatment).
The mean population level of cholesterol would be lower by either (or
both) a downward shift in the entire distribution or a decrease in the upper
extreme of the cholesterol distribution. However, a reduction in mean cholesterol
concentration because of a shift in the entire population distribution would
be expected to have much greater impact on population mortality than would
a reduction in mean cholesterol owing to treatment of persons with high blood
cholesterol levels alone.3 Therefore, to understand
the observed temporal changes in cardiovascular disease mortality, it is important
to determine the nature of shifts that have occurred in the cholesterol distribution
of the US population beyond an exclusive focus on either the arithmetic mean
or the upper extreme of the distribution. From a public health perspective,
reliance on cholesterol-lowering medications cannot be a long-term solution
for the control of epidemic high blood cholesterol levels that affect millions
of US adults. Population-wide preventive strategies promoting appropriate
behavior change are in principle much preferred, if they can be shown to be
effective.2 Therefore, understanding the forces
that determine the observed distributions of serum TC concentrations in human
populations is necessary for the soundest development of long-range public
health approaches to prevention of high blood cholesterol levels. The goal
of the present epidemiologic investigation is to determine whether population-wide
influences and/or the high-risk approach have been operating to produce the
well-known decline in mean TC concentration in the US population.
METHODS
POPULATION
Data for this report came from the National Health Examination Survey
(NHES) and the National Health and Nutrition Examination Surveys (NHANES)
I, II, and III. The designs of the NHES and NHANES series have been published
previously.4-7
In brief, these surveys represent repeated, independent cross-sectional surveys
of representative samples of the civilian, noninstitutionalized population
of the United States, 18 through 74 years of age.4-7
In NHANES III, there was no upper age cutoff point; however, since persons
older than 74 years were excluded from the earlier surveys, those participants
were excluded from the present analyses. All racial and ethnic groups and
both sexes were included. These data were collected to represent serial cross-sections
of the population of the United States. We constructed a series of birth cohorts
from approximately 1887 through 1975, the birth years that would meet age
eligibility criteria for at least one in the series of surveys.
VARIABLES
The NHES and NHANES series included data describing the participant's
date of birth, date of examination, age, sex, race, and TC concentration.
Lipoprotein cholesterol concentrations were available in only the 2 most recent
surveys; hence, these data were not examined in the present analyses. The
date of birth and age at examination were used as 2 primary factors that predict
TC concentration. Given these 2 variables, the distribution of cholesterol
levels at a fixed age could be described as a function of the year of birth
(or "birth cohort") of the participants. For the NHES and NHANES III, neither
year of birth nor exact date of examination were available. Thus, for those
2 surveys, we estimated year of birth by subtracting the participant's age
from the midyear of the examination period (1961 for NHES, 1989 for NHANES
III phase 1, and 1993 for NHANES III phase 2). Data regarding Hispanic ethnicity
were not collected before NHANES III; therefore, analyses to differentiate
persons of Hispanic ethnicity could not be performed.
In all surveys, serum TC concentration was scheduled to be measured
on all examined adults, regardless of fasting status. A description of the
procedures used for blood sample collection and measurement of TC has been
published previously.1
Cholesterol measurements from each of the 4 surveys were standardized
according to the criteria of the Centers for Disease Control and Prevention
(CDC) or the CDC–National Heart, Lung, and Blood Institute Lipid Standardization
Program.8 The NHES I measurements were performed
by the CDC Lipid Standardization Laboratory9
using a modified ferric chloride reference method, and the values were corrected
to the subsequently adopted CDC reference cholesterol method that is based
on the method of Abell et al.10 The rationale
and factors used to make the adjustments have been discussed previously.11 The NHANES I measurements were made in the CDC Lipid
Standardization Laboratory, but with a newer reference method.11
In NHANES II, serum samples were analyzed in the George Washington University
Lipid Research Clinic Laboratory using a Liebermann-Burchard reaction method.12-13 This method used serum calibration
pools to adjust measured values to equivalent CDC reference values.12, 14 In NHANES III, cholesterol levels
were measured enzymatically in The Johns Hopkins University Lipid Research
Clinic Laboratory using a commercially available reagent mixture (Cholesterol/HP,
catalog No. 816302; Boehringer Mannheim Diagnostics, Indianapolis, Ind) based
on the method of Allain et al.15
ANALYTIC PLAN
We examined age-related changes in the distribution of TC concentration
across birth cohorts as sampled in the series of surveys to determine whether
more recent birth cohorts were attaining lower blood cholesterol distributions
than earlier birth cohorts. We tested the hypotheses that the age-adjusted
10th, 25th, 50th, 75th, and 90th percentiles of TC concentration were lower
in successively more recent US birth cohorts than in earlier cohorts. We contend
that changes at the 10th, 25th, and 50th percentiles reflect population-wide
influences alone, whereas changes at the 75th and 90th percentiles could reflect
the combined effects of population-wide influences and high-risk approaches.
For use in graphical presentations, 8 birth cohorts were constructed (1887-1899,
1900-1909, 1910-1919, 1920-1929, 1930-1939, 1940-1949, 1950-1959, and 1960-1969).
Persons born in 1970 through 1975 were excluded from the graphical presentations
because this birth cohort would have contributed a single point. The 5 specified
percentile values of the distribution of TC were determined for the 8 birth
cohorts across 6 age groups (18-24, 25-34, 35-44, 45-54, 55-64, and 65-74
years). Plotting these percentile values within each stratum defined by age
and birth cohort displays the unadjusted birth cohort patterns of association
between age and the percentile for TC.
The primary goal of these analyses was to determine whether there were
differences in the selected percentiles of TC across birth cohorts at a fixed
age. An analysis of covariance approach was used, wherein the expected value
for a percentile of the TC distribution was modeled as a function of the birth
year, age, and nonlinear and interaction terms, specifically:
where TCx is the x-th percentile (10th, 25th, 50th, 75th,
or 90th) of the TC distribution, Age is the age of the participant at the
time of the survey, BY is the birth year of the participant, and  k are the regression parameters. The goal of this analysis was to assess
whether there were differences in the age-related pattern of TC across birth
years. These differences were assessed in the full model described herein
and in a simplified model excluding age–by–birth year interaction
terms. The Age2 term was included in the model to account for the
curvilinear nature of the age-related pattern of TC concentration. The Age
x BY and Age2 x BY interaction terms were included
in the models to test whether the age-related pattern of TC differed across
birth cohorts, that is, whether the slope of the age-related "change" in TC
was more or less steep across birth cohorts. These models were fit using the
asymmetric square error loss approach used by Efron.16
In ordinary regression (or least squares), the relationship between predictor
variables and the mean value for the outcome variable is estimated by providing
equal weight to residuals above and below the estimated regression line. Efron16 suggested that the relationship between predictor
variables and percentiles of the distribution can be estimated by "shifting"
the regression line by assigning differential weight to residuals above the
regression line relative to those below the regression line. By more heavily
weighting residuals above (relative to below) the regression line, the line
that minimizes the weighted sum of squares will shift the regression line
upward. For any specific weight, the slope and intercept defining a unique
regression line can be found by Newton-Raphson methods.17
The percentile associated with the regression line can be determined by tabulating
the number of observations above and below the estimated line. Specific percentiles
of interest for these analyses (10th, 25th, 50th, 75th, and 90th) were found
by an additional Newton-Raphson search. The variances of the estimated parameters
were provided by bootstrap methods with 100 replications.18-19
Analyses were adjusted for sex and ethnicity. Additional percentile regression
analyses were performed to estimate the 1st, 5th, 15th, 20th, 30th, 35th,
40th, 45th, 55th, 60th, 65th, 70th, 80th, 85th, 95th, and 99th percentiles.
These percentiles were used in developing a graphical display of the estimated
TC distributions of 50-year-old persons born in 1910 (measured in 1960) and
1940 (measured in 1990).
Complex sampling strategies were used in the individual surveys to enable
the estimation of population characteristics such as prevalence of high blood
cholesterol levels that were applicable to the noninstitutionalized adult
population of the United States. The incorporation of these sampling weights
in the current analyses was not feasible because appropriate statistical techniques
have not been developed for weighted percentile regression. These sampling
weights would have a major impact on the estimation of the prevalence of high
blood cholesterol levels in the US population, but in general have a lesser
impact on the estimation of associations between TC and other variables, such
as year of birth. All analyses were conducted using STATA 5.0 statistical
software (Stata Corp, College Station, Tex).
RESULTS
POPULATION CHARACTERISTICS
Across the 4 surveys, 52 646 men and women met year of birth and
age eligibility criteria to be included in this analysis, including 6530 participants
from NHES, 16 704 from NHANES I, 12 504 from NHANES II, and 16 908
from NHANES III. This composite data file included 29 145 women and 23 501
men, 41 430 white participants, 10 053 black participants, and 1163
persons of other or unknown race or ethnicity. Data regarding TC concentration
were missing for 3110 persons (5.9%), leaving 49 536 participants for
analysis. The distribution of these participants by year of birth (as reported
or estimated) and age is shown in the Table
1. Because those born in the earliest birth cohorts were old at
the time of the first examination, the data for the earliest birth cohorts
were limited necessarily to the older age groups. Likewise, those participants
in the most recent birth cohorts could not have achieved an advanced age by
the time of the final examination, and as such the data for the most recent
birth cohorts were limited necessarily to the younger age groups.
|
|
|
|
Distribution of Participants With Serum Total Cholesterol Concentrations
in the National Health Examination Survey and the National Health and Nutrition
Examination Surveys I, II, and III by Year of Birth and Age at Examination
|
|
|
TC CONCENTRATION AND AGE
Observed and estimated percentile curves for TC concentrations of persons
aged 18 through 74 years are shown by birth cohort in Figure 1. At each percentile shown, TC concentration was greater
at successively older ages except for a plateau or decrease for persons older
than 60 years. In the regression analyses of TC percentiles that included
age, age2, race, sex, and year of birth, the coefficients for the
linear age terms were positive (higher TC levels at older ages) and significant
(P<.001). Conversely, the coefficients for the
age2 (quadratic) term were negative (because of a declining rate
of increase across the whole age range and a decrease in TC concentration
at the oldest ages) and significant (P<.001).
|
|
|
|
Figure 1. Observed (left panels) and estimated
(right panels) 90th, 75th, 50th, 25th, and 10th percentile curves for serum
total cholesterol concentration for the age range 18 through 74 years by birth
cohort. In the panels depicting estimated patterns, the solid lines reflect
the ranges for which data were observed and the dotted lines reflect the ranges
for which the values were extrapolated from the observed data. The models
used to derive these estimates included the following independent variables:
age, age2 (quadratic), birth year, age by birth year, and age2 by birth year. Scale intervals are equal in all panels, whereas the
ranges differ as appropriate to each percentile shown. To convert cholesterol
from milligrams per deciliter to millimoles per liter, multiply by 0.02586.
|
|
|
TC CONCENTRATION AND BIRTH YEAR
At each percentile shown, more recent birth cohorts attained lower TC
concentrations than did earlier birth cohorts (Figure 1). The differences between the birth cohorts were larger
the higher the percentile considered; nevertheless, the observed declining
pattern with more recent birth cohorts was significant (P<.001) for all percentiles considered. Adjusted for age, sex, and
ethnicity, the 10th, 25th, 50th, 75th, and 90th percentiles of TC concentrations
were estimated to be 3.4, 3.9, 4.7, 5.7, and 7.1 mg/dL (0.09, 0.10, 0.12,
0.15, and 0.18 mmol/L) lower, respectively, for every successive 10 years
later in date of birth. Additional models (results not shown) were examined
to assess differences in this birth cohort effect between sex and ethnic groups.
Decreasing TC concentrations were observed at all 5 percentiles in both men
and women, with greater decreases observed in women than in men (for sex–by–birth
year interaction term, P<.001 in all 5 models).
Likewise, decreasing TC concentrations were observed at all 5 percentiles
for white and black participants, with greater decreases for white than black
participants at the lower 3 percentiles (for ethnicity–by–birth
year interaction term, P<.01 in all 3 models).
No ethnic difference was observed at the 75th and 90th percentiles.
The rate of the estimated decrease in the percentiles of TC also varied
by age as shown in Figure 1. Not
only were TC concentrations lower for more recent birth cohorts than for earlier
birth cohorts but also the apparent increase in TC concentration with age
was less rapid (for interaction terms, P<.001
in all 5 models). That is, the apparent rate of increase in TC concentrations
attributable to aging was slower in more recent cohorts than in earlier cohorts.
This finding is indicated by the divergence of the estimated curves shown
in Figure 1, at least through the
end of middle age. (These patterns of change in TC concentration by age are
based on observations made in independent samples of persons belonging to
any particular birth cohort at successive surveys.)
The birth cohort changes shown in Figure
1 have an impact on the estimated distribution of TC concentration
at any given age for 2 or more contrasting birth cohorts. Thus, the estimated
distributions of TC concentration for 50-year-old persons born in 1910 and
1940 are shown in Figure 2. The
entire distribution of TC concentrations was shifted to lower levels in the
1940 birth cohort relative to the 1910 birth cohort, with a greater shift
in the upper range of the TC distribution. This view of the changing distribution
illustrates clearly the effects described by Rose2
and addressed herein.
|
|
|
|
Figure 2. The estimated distributions of
serum total cholesterol concentration for 50-year-old persons born in 1910
and 1940. The models used to derive these estimates included the following
independent variables: age, age2 (quadratic), birth year, age by
birth year, and age2 by birth year. The prevalence is the estimated
proportion of people with an exact cholesterol concentration. To convert cholesterol
from milligrams per deciliter to millimoles per liter, multiply by 0.02586.
|
|
|
COMMENT
These analyses demonstrate that the entire distribution of TC concentrations
has shifted to lower levels in the United States. This downward shift, previously
described primarily for the mean,1, 20-21
is present at even the lower percentiles (10th and 25th) of blood pressure,
where pharmacologic management can be assumed to have had virtually no impact.
Thus, this shift in the overall distribution cannot be attributed solely to
treatment effects but must have resulted to an important degree from population-wide
behavioral and environmental influences on TC concentrations. The finding
that the shift was observed among women and men and among black and white
populations supports the contention that population-wide behavioral and environmental
influences are operating to cause this birth cohort effect.
Previously, populations have been shown to differ in the slope of cholesterol
increase with age.22 In this report, the slope
of cholesterol increase with increasing age was less steep for more recent
birth cohorts than for earlier cohorts. This finding indicates that the forces
that influence the slope of the cholesterol increase with age may be dynamic
and may therefore be modifiable through planned prevention strategies. If
this pattern were to continue, we could expect more recent birth cohorts to
develop clinically defined high blood cholesterol concentrations less commonly
in the future than have earlier birth cohorts. Such a finding would be strongly
supportive of an effect of primary prevention of high blood cholesterol levels
in the US population.
The population influences responsible for this substantial change in
cholesterol development are not yet known. Adverse trends in physical activity
and obesity have been reported; however, important beneficial dietary changes
may have occurred, for example, increased consumption of fruits and vegetables
and decreased consumption of foods containing saturated fatty acids. A recent
report from the CARDIA Study20 documented birth
cohort–related reductions in TC and low-density lipoprotein cholesterol
concentrations in association with reductions in dietary saturated fatty acid
and cholesterol intake. It is also instructive to note that pellagra was endemic
in much of the United States in the early 1900s, an observation that underscores
the magnitude of the dietary changes that have occurred during the past century.
The shift to lower blood cholesterol concentrations was more pronounced
in the upper range of the distribution. The larger decrease in the upper range
could in theory reflect the combined effects of prevention and treatment,
with the prevention effect seen at the lower percentiles complemented by the
specific treatment effect on those with recognized and treated high blood
cholesterol levels. However, use of effective cholesterol-lowering medications
was not widespread during the period covered by these surveys.21
Alternatively, the greater decline at the upper aspect of the cholesterol
distribution could reflect a greater cholesterol-lowering effect of societal
and behavioral changes among persons with high cholesterol concentrations
whether due to genetic differences in response to behavior change or more
extreme initial behavior patterns with correspondingly greater latitude for
improvement.
The US mortality from coronary heart disease increased during this century
until the mid-1960s and has been declining since.23
One may speculate that since high blood cholesterol levels stand as one of
the most important risk factors for coronary heart disease,23
the downward shift in the distribution of cholesterol since the middle of
the 20th century may be playing a central role in the more than 50% decline
in coronary heart disease mortality during the same period. A previously demonstrated
downward shift in blood pressure may also have contributed to the decline
in coronary heart disease mortality.24 Coronary
heart disease has a complex risk factor structure that prominently includes
other risk factors, including cigarette smoking, diabetes mellitus, dietary
imbalance, and physical inactivity.23 The pattern
of increasing coronary heart disease mortality until the mid-1960s may reflect
changes in these other risk factors, most notably the increase in smoking
rates during the early part of this century until the mid-1960s. As levels
of these other risk factors, including smoking rates, have stabilized or declined,
it is likely that the declining trend in the entire cholesterol distribution
has played an important role in the decrease in coronary heart disease mortality
observed in the United States during the past 40 years.
A limitation inherent in the data available through these surveys is
the lack of repeated measures of cholesterol concentrations with increasing
age for specific individuals. Rather, these data represent repeated independent
samples from these birth cohorts. Thus, the use of these data to describe
age-related changes in cholesterol concentrations is similar to the use of
cross-sectional data to construct growth charts for children. In the analogous
scenario involving the construction of growth charts, the tempo of growth
in a typical child is markedly blunted in cross-sectional data. Thus, growth
charts constructed from cross-sectional data do not adequately reveal the
typical growth pattern of an individual. Nevertheless, if growth charts constructed
based on cross-sectional data from different birth cohorts of children differed
substantially, one would be able to identify the fact that some birth cohorts
of children were growing faster (or slower) or taller (or shorter) than other
birth cohorts. Likewise, although the tempo of cholesterol change with age
for a typical individual cannot be accurately described using these data,
substantial differences in the change of cholesterol with age between birth
cohorts are equally compelling as representing a cohort effect as the analogous
example regarding growth in children.
The alternative (and in principle stronger) study design, a series of
population-based cohort studies following persons born between 1887 and 1974
from ages 18 through 74 years, can no longer be performed. Establishing and
following current birth cohorts would be of great interest but would address
the question of future changes in the distributions of cholesterol rather
than previous changes. Thus, the present epidemiologic approach was the only
means available for the stated purposes of gaining greater insight relevant
to population-level changes in cholesterol distributions during the past several
decades.
Techniques for cholesterol concentration measurement differ across surveys.1 The most likely effect of this change on the analyses
reported herein would be to introduce random error or "noise" and thereby
to decrease the likelihood of observing a consistent change across birth cohorts.
Furthermore, if the quality of cholesterol measurements improved with successive
surveys, the expected effect of this change in method would be a reduction
in the number of extreme measurements caused by measurement error. As the
proportion of extreme measurements due to error decreased, there would be
an associated increase in the lower percentiles (10th and 25th) of cholesterol
in the absence of other influences. Therefore, if temporal changes in measurement
error were important influences, we would at worst have underestimated the
true decline in these lower percentiles.
The nature of the shift in the cholesterol distribution demonstrated
herein supports the contention that changes in population-wide behavior and
environmental conditions have contributed to the decline in mean cholesterol
concentrations observed in the United States. As a result, these findings
support the potential utility of planned population approaches to risk factor
reduction and chronic disease prevention.
Efforts aimed at preventing the development of high blood cholesterol
levels (ie, primary prevention of hyperlipidemia) through changes in health
behaviors at the population level should go forward simultaneously with continued
and intensified efforts to improve the control of high blood cholesterol levels
through pharmacologic means. Further effort should be devoted to developing
an understanding of the population-wide changes that have contributed to this
decline in cholesterol concentrations. This knowledge could enhance greatly
our prospects for the prevention of cardiovascular diseases.
AUTHOR INFORMATION
Accepted for publication August 29, 2001.
This work was supported by grant 1 RO3 HL58697 from the National Heart,
Lung, and Blood Institute of the National Institutes of Health, Bethesda,
Md.
Corresponding author and reprints: David C. Goff, Jr, MD, PhD, Public
Health Sciences and Internal Medicine, Wake Forest University School of Medicine,
Medical Center Boulevard, Winston-Salem, NC 27157-1063.
From the Department of Public Health Sciences, Wake Forest University
School of Medicine, Winston-Salem, NC (Dr Goff and Mr Russell); Division of
Adult and Community Health, National Center for Chronic Disease Prevention
and Health Promotion, Centers for Disease Control and Prevention, Atlanta,
Ga (Dr Labarthe); and Department of Biostatistics, University of Alabama at
Birmingham School of Public Health (Dr Howard).
REFERENCES
| |
1. Johnson CL, Rifkind BM, Sempos CT, et al. Declining serum total cholesterol levels among US adults: the National
Health and Nutrition Examination Surveys. JAMA. 1993;269:3002-3008.
ABSTRACT
2. Rose G. Strategy of prevention: lessons from cardiovascular disease. Br Med J (Clin Res Ed). 1981;282:1847-1851.
3. Kottke TE, Gatewood LC, Wu SC, Park HA. Preventing heart disease: is treating the high risk sufficient? J Clin Epidemiol. 1988;41:1083-1093.
FULL TEXT
|
ISI
| PUBMED
4. National Center for Health Statistics. Plan and Initial Program of the Health Examination
Survey. Washington, DC: US Dept of Health, Education, and Welfare, National
Center for Health Statistics; 1965. Advance Data From Vital and Health Statistics,
No. 4.
5. National Center for Health Statistics. Plan and Operation of the Health and Nutrition Examination
Survey, United States, 1971-73. Washington, DC: US Dept of Health, Education, and Welfare, National
Center for Health Statistics; 1973. Advance Data From Vital and Health Statistics,
No. 10.
6. McDowell A, Engle A, Massey J, et al. Plan and Operation of the second National Health
and Nutrition Examination Survey, 1976-80. Hyattsville, Md: US Dept of Health and Human Services, National Center
for Health Statistics; 1981. Advance Data From Vital and Health Statistics,
No. 15.
7. Ezzati TM, Massey JT, Waksberg J, Chu A, Maurer KR. Sample Design: Third National Health and Nutrition
Examination Survey. Hyattsville, Md: US Dept of Health and Human Services, National Center
for Health Statistics; 1992. Advance Data From Vital and Health Statistics,
No. 113.
8. Myers GL, Cooper GR, Winn CL, Smith SJ. The Centers for Disease Control-National Heart, Lung, and Blood Institute
Lipid Standardization Program. Clin Lab Med. 1989;9:105-135.
ISI
| PUBMED
9. Moore Fe, Gordon T. Serum Cholesterol Levels of Adults, United States,
1960-62. Washington, DC: US Dept of Health, Education, and Welfare, National
Center for Health Statistics; 1967. Advance Data From Vital and Health Statistics,
No. 22.
10. Abell L, Levy GB, Brodie BB, Kendall FE. A simplified method for the estimation of total cholesterol in serum
and demonstration of its specificity. J Biol Chem. 1952;195:357-366.
FREE FULL TEXT
11. Abraham S, Johnson CL, Carroll MD. Total Serum Cholesterol Levels of Adults 18-74 Years: United States, 1971-74. Washington, DC: US Dept of Health, Education, and Welfare, National Center for Health Statistics; 1978. Advance Data From Vital and Health Statistics,
No. 205.
12. Lipid Research Clinics Program. Manual of Laboratory Operations: Lipid and Lipoprotein
Analysis. Bethesda, Md: National Institutes of Health; 1974:1. Revised 1982.
DHEW publication NIH 75-628.
13. Carroll M, Sempos C, Briefel R, Gray S, Johnson C. Serum Lipids of Adults 20-74 Years, United States,
1976-80. Hyattsville, Md: US Dept of Health and Human Services, National Center
for Health Statistics; 1993. Advance Data From Vital and Health Statistics,
No. 242.
14. Bachorik PS, Wood PD, Williams J, Kuchmak M, Ahmed S, Lippel K, Albers J. Automated determination of total plasma cholesterol: a serum calibration
technique. Clin Chim Acta. 1979;96:145-153.
FULL TEXT
|
ISI
| PUBMED
15. Allain CC, Poon LS, Chan CSG, Richmond W, Fu PC. Enzymatic determination of total serum cholesterol. Clin Chem. 1974;20:470-475.
ABSTRACT
16. Efron B. Regression percentiles using asymmetric squared error loss. Stat Sinica. 1991;1:93-125.
17. Reddien GW. Newton-Raphson methods. In: Encyclopedia of Statistical Sciences.
Vol 6. New York, NY: John Wiley & Sons; 1985.
18. Efron B, Tibshirani R. An Introduction to the Bootstrap. New York, NY: Chapman & Hall; 1993.
19. Smith M, Howard G, Evans GW. Bootstrap estimation of the variance of percentile regression parameters. In: Proceedings of the 17th Annual SAS Users Group
International Conference. Cary, NC: SAS Institute; 1992:1283-1991.
20. Bild DE, Jacobs DR Jr, Liu K, et al. Seven-year trends in plasma low-density-lipoprotein-cholesterol in
young adults: the CARDIA Study. Ann Epidemiol. 1996;6:235-245.
FULL TEXT
|
ISI
| PUBMED
21. Burke GL, Sprafka JM, Folsom AR, Hahn LP, Luepker RV, Blackburn H. Trends in serum cholesterol levels from 1980 to 1987: the Minnesota
Heart Survey. N Engl J Med. 1991;324:941-946.
ABSTRACT
22. Keys A. Serum cholesterol and the question of "normal." In: Benson ES, Strandjord PE, eds. Multiple Laboratory Screening. New York, NY: Academic Press; 1969:147-170.
23. Labarthe DR. Coronary heart disease. In: Labarthe DR, ed. Epidemiology and Prevention
of Cardiovascular Diseases: A Global Challenge. Gaithersburg, Md: Aspen
Publishers Inc; 1998:43-72.
24. Goff DC Jr, Howard G, Russell GB, Labarthe DR. Birth cohort evidence of population influences on blood pressure in
the United States, 1887–1994. Ann Epidemiol. 2001;11:271-279.
FULL TEXT
|
ISI
| PUBMED
RELATED ARTICLE
Archives of Internal Medicine Reader's Choice: Continuing Medical Education
Arch Intern Med. 2002;162(8):951-952.
FULL TEXT
THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES
|
Lipid Screening and Cardiovascular Health in Childhood
Daniels et al.
Pediatrics 2008;122:198-208.
ABSTRACT
| FULL TEXT
Primary Prevention of Cardiovascular Disease in Nursing Practice: Focus on Children and Youth: A Scientific Statement From the American Heart Association Committee on Atherosclerosis, Hypertension, and Obesity in Youth of the Council on Cardiovascular Disease in the Young, Council on Cardiovascular Nursing, Council on Epidemiology and Prevention, and Council on Nutrition, Physical Activity, and Metabolism
Hayman et al.
Circulation 2007;116:344-357.
FULL TEXT
Over-The-Counter Statins Are Worth Considering in Primary Prevention of Cardiovascular Disease
Gotto
Circulation 2006;114:1310-1314.
FULL TEXT
Managing Abnormal Blood Lipids: A Collaborative Approach
Fletcher et al.
Circulation 2005;112:3184-3209.
ABSTRACT
| FULL TEXT
Intensive Lipid-Lowering Therapy in Patients with Coronary Heart Disease
Lauderdale and Sheehan
The Annals of Pharmacotherapy 2005;39:329-334.
ABSTRACT
| FULL TEXT
Prevention Conference VII: Obesity, a Worldwide Epidemic Related to Heart Disease and Stroke: Group II: Age-Dependent Risk Factors for Obesity and Comorbidities
St. Jeor et al.
Circulation 2004;110:e471-e475.
FULL TEXT
Trends in Self-reported Multiple Cardiovascular Disease Risk Factors Among Adults in the United States, 1991-1999
Greenlund et al.
Arch Intern Med 2004;164:181-188.
ABSTRACT
| FULL TEXT
|
