American Journal of Epidemiology
Copyright O 1997 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol. 146, No. 7
Printed in U.S.A
Reduced Cholesterol Is Associated with Recent Minor Illness
The CARDIA Study
David R. Jacobs, Jr.,1 Brian Hebert,2 Pamela J. Schreiner,1 Stephen Sidney,3 Carlos Iribarren,1 and
Stephen Hulley4
Lower levels of plasma total cholesterol have been observed during severe infection, but it is not known
whether the minor illnesses encountered in the general population are also associated with reduced cholesterol. This paper examines the relation between minor illness and plasma lipids, using 7- and 10-year follow-up
data from more than 3,000 generally healthy participants in the Coronary Artery Risk Development in Young
Adults (CARDIA) Study. At both 7 and 10 years, approximately 8.5% of participants stated they had been "ill
with cold, flu, fever, or vomiting in the past 24 hours." In both cross-sectional and longitudinal analyses, the
plasma total cholesterol was about 5 mg/dl lower (p < 0.006) and high density lipoprotein cholesterol about
1.2 mg/dl lower (p < 0.12) in those who reported minor illness than in those who did not. Plasma triglycerides
did not vary with minor illness. The authors conclude that reductions in plasma total, low density, and high
density lipoprotein cholesterol mark an acute phase response even during minor illness. These reductions may
bias surveys over a limited geographic area during a short period because the proportion with minor illness
may vary locally. Because this effect should be stronger with more precise illness diagnosis, clinicians should
avoid making measurements for cholesterol management when illness may alter plasma lipid levels and the
resulting decisions. Am J Epidemiol 1997;146:558-64.
cold; influenza; lipids; nausea; smoking; vomiting
MATERIALS AND METHODS
Several studies have demonstrated modified blood
lipids and lipoproteins in severe infection (1-6), and a
similar phenomenon is seen in traumatic injury (7, 8)
and surgery (7, 9, 10). We asked whether lower lipid
levels would be associated with the minor illnesses
encountered in participants of population surveys. We
hypothesized that blood cholesterol would be slightly
lower in survey participants who reported they had
been ill on the day before examination. Data from
the year 7 and the year 10 examinations of a cohort
study of young adults were analyzed to test these
hypotheses.
Participants and measurements
The data used in this study were gathered as part of
the multicenter study, Coronary Artery Risk Development in Young Adults (CARDIA). The four field sites
were located at the University of Alabama at Birmingham, Birmingham, Alabama; Northwestern University, Chicago, Illinois; University of Minnesota, Minneapolis, Minnesota; and Kaiser Permanente Medical
Care Program, Oakland, California. Participants were
recruited by telephone at the Birmingham and Chicago
sites. The Minneapolis site used door-to-door recruitment in areas lacking telephones in addition to telephone recruitment. Oakland participants were selected
at random from among members of the Kaiser Permanente prepaid health plan. The baseline (year 0) recruitment was stratified by race (black or white), gender, age (18-24 vs. 25-30 years), and education (those
with less than a high school education and those with
at least a high school education). Approximately half
of those invited participated in the baseline examination. In total, 4,086 and 3,950 CARDIA participants
took part in the year 7 and 10 examinations, respectively (80.6 and 78.5 percent of those examined at
baseline, excluding deaths).
Received for publication April 1, 1996, and accepted for publication June 10, 1997.
Abbreviations: AIDS, acquired immunodeficiency syndrome;
CARDIA, Coronary Artery Risk Development in Young Adults; HIV-1,
human Immunodeficiency virus type 1; HDL, high density lipoprotein; LDL, low density lipoprotein.
1
Division of Epidemiology, School of Public Health, University of
Minnesota, Minneapolis, MN.
2
University of North Dakota, Grand Forks, ND.
3
Division of Research, Kaiser Permanente, Oakland, CA.
4
Department of Epidemiology and Biostatistics, University of
California at San Francisco, San Francisco, CA.
Reprint requests to Dr. David R. Jacobs, Jr., Division of Epidemiology, School of Public Health, University of Minnesota, 1300
South Second Street, Suite 300, Minneapolis, MN 55454.
558
Cholesterol in Minor Illness
Before phlebotomy was performed at years 7 and
10, the participants were asked, "Have you been ill in
the past 24 hours (e.g., cold, flu, fever, vomiting)?"
(yes or no). No further differentiation was made between these symptoms of minor illness, nor were any
other illness details gathered.
Among participants who were excluded from year 7
analyses were 62 pregnant women, 325 who had
fasted less than 8 hours, and 106 who had missing
blood lipid values. An additional 82 were excluded
because they were missing covariate values; 3,473
participants were included in the analyses presented
here. Excluded from year 10 analyses were 28 due to
pregnancy, 203 due to fasting less than 8 hours, and 83
due to missing blood lipid values. An additional 58
were excluded because they were missing covariate
values; 3,528 participants were included in the analyses presented here. The 31 participants who reported
illness at both examinations were omitted from analyses of plasma lipid changes; 2,867 participants attended both examinations and were available for
change analyses.
Using a vacutainer containing ethylenediaminetetraacetic acid, blood samples were drawn the morning
after an overnight fast. The plasma was separated and
stored at — 70°C in airtight vials until it was shipped to
the Northwest Lipid Research Laboratories at the University of Washington in Seattle, Washington. Both
total cholesterol and triglycerides were measured enzymatically within 6 weeks of collection (11). High
density lipoprotein (HDL) cholesterol was determined
by precipitation with dextran sulfate/magnesium chloride (12). Low density lipoprotein (LDL) cholesterol
was calculated using the Friedewald equation (13),
omitting individuals with plasma triglycerides ^400
mg/dl. Questionnaires were used to obtain information
about the following potential covariates: smoking, oral
contraceptive use, self-reported history of diabetes,
total physical activity, diet, alcohol consumption, race,
sex, and education (14—16). Weight and height were
measured in light clothes (e.g., short-sleeved shirt,
shorts, and socks) without shoes. Measurements were
recorded to the nearest 0.5 pound for weight and 0.5
cm for height. Body mass index was computed as
weight/height2 (kg/m2). Values of saturated fat, polyunsaturated fat, total calories, and total cholesterol
intake (15) were used to compute Keys score (17). All
variables included in this investigation were measured
using identical protocols at the year 7 and 10 examinations, except diet, which was measured only at year
7. Quality control procedures were conducted at the
local and central level throughout all examination
cycles.
Am J Epidemiol
Vol. 146, No. 7, 1997
559
Statistical analysis
To assess the difference in plasma lipid levels between those reporting and not reporting illness in the
past 24 hours, we examined data from each examination cross-sectionally and also from those who reported illness only at year 10 ("became ill") or reported illness only at year 7 ("got better") to use each
person as his own control, defining change as year 10
minus year 7. Analogous to a crossover design without
randomization, an unbiased estimate of the lipid
change attributable to illness, removing time trend
between the two examinations, is as follows:
[(change in those who became ill) —
(change in those who got better)]/2.
This estimate assumes that each person's plasma lipid
level may be represented as the sum a + t + r,
where a is the individual lipid level (mg/dl) at year 7,
/is 0 at year 7 and the change in mg/dl by year 10, and
r is 0 or the offset associated with recent illness {t and
r are common to all participants). Analysis of covariance was used to adjust illness-related difference and
change for potential covariates.
RESULTS
Dlness was reported in the 24 hours before the clinic
visit in 8.4 percent (290/3,472) of participants at the
year 7 examination and in 8.8 percent (309/3,528) of
participants at the year 10 examination. Of those not ill
at year 7, 8.1 percent (214/2,639) became ill at year
10, and 7.5 percent (197/2,622) of those not ill in the
24 hours before year 10 had been ill at year 7. Black
men were slightly less likely than black women, white
men, or white women to report illness in the 24 hours
before either examination.
As shown in table 1, plasma total cholesterol was
6.3 mg/dl less in the 290 persons reporting illness in
the 24 hours preceding the year 7 examination than in
the 3,182 not reporting illness. The illness-related difference was reduced to 5.4 mg/dl after adjustment for
race, sex, age, education, smoking, body mass index,
oral contraceptive use (among women), self-reported
history of diabetes, total physical activity, Keys score,
and alcohol consumption at year 7. Similarly, using
data from the year 10 examination and adjusting for all
covariates, plasma total cholesterol was 7.3 mg/dl
lower in those who were ill in the past 24 hours than
in those not ill. Cholesterol decreased by 3.9 mg/dl in
the change analysis, comparing the examination in
which the person reported illness with the examination
in which no illness was reported and adjusting for
changes in covariates. Corresponding results for HDL
560
Jacobs et al.
TABLE 1. Mean level and change In plasma total cholesterol by recent illness status, unadjusted and
adjusted* at year 7 and 10 examinations, the CARDIA Study
Unadjusted
\fear 7 cross-sectional analysis
III in past 24 hours
Not ill in past 24 hours
No.
Mean± SEt
Mean±SE
290
3,182
170.1 ±1.8
176.4 ± 0.6
171.2 ±1.9
176.6 ±0.6
-6.3 ± 2.0
0.0019
-6.4 ± 2.0
0.0065
171.6 ±1.9
178.5 ± 0.6
171.5 ±1.9
178.7 ±0.6
-6.9 ± 2.0
0.0007
-7.3 ± 2.0
0.0003
6.3 ±1.7
-2.1 ± 1.4
1.3 ±0.5
5.8 ± 1.6
-2.1 ±1.5
1.2 ±0.5
-4.2 ±1.1
0.0001
-3.9 ±1.0
0.0002
Difference
p value
Year 10 cross-sectional analysis
ID in past 24 hours
Not 31 in past 24 hours
309
3,219
Difference
p value
Year 10-7 change analysis: mean
change in plasma total
cholesterol by recent illness
status
III year 7, not id year 10
Not ill year 7, ill year 10
Not ill, year 7 and year 10
Adjusted
197
214
2,425
Difference^
p value
* Adjusted for age, race, sax, education, smoking status, alcohol intake, body mass index, physical activity,
Keys score (see text), diabetes, and use of oral contraceptives,
t SE, standard error.
i Effect of illness on cholesterol, removing an estimate of time trend in cholesterol, is estimated as follows:
0.5 x (cholesterol change in persons ill at year 10 only - cholesterol change in persons ill at year 7 only).
cholesterol are shown in table 2. The adjusted illnessrelated differences were —1.1 at year 7, —1.7 at year
10, and —1.2 mg/dl in the change analysis. Illnessrelated LDL cholesterol differences were approximately the difference between the total cholesterol and
the HDL cholesterol differences (data not shown).
Estimated illness-related differences for plasma triglycerides, after adjustment for all covariates, were
- 0 . 7 {p = 0.8) at year 7, 0.8 (p = 0.8) at year 10, and
—2.7 mg/dl (p — 0.25) in the change analysis.
Inspection revealed that these changes in total cholesterol associated with minor illness were neither
dominated by a few very large values nor restricted to
people with very high or very low cholesterol. Although mean plasma total cholesterol was lower in
persons ill in the 24 hours before examination in all
race-sex and smoking groups, illness-related differences seemed to be greater in men than in women
(table 3) and in never smokers than in former or
current smokers (table 4). Findings for HDL cholesterol varied by race, sex, and smoking status in a
similar pattern (data not shown).
In table 5, levels of several nonlipid covariates according to illness status, each adjusted for age, race,
sex, and education, are shown. No variable reveals a
consistent pattern. Only smoking at year 7 and physical activity at year 10 are related to illness. Illness in
the 24 hours before both the year 7 and the year 10
examinations was twice as likely in smokers at year 7
(1.8 percent, 12/663) as in never and former smokers
(0.9 percent, 19/2,204). Duration of fasting varied
little by illness status. The 90th percentile of fasting
duration at year 7 was 15.5 versus 15.3 hours and at
year 10 was 15.8 versus 16.2 hours in those who were
ill versus not ill in the 24 hours before the examination.
DISCUSSION
These results support the hypothesis that plasma
total, LDL, and HDL cholesterol decline in minor
illness, although to a lesser degree than has been
observed in more severe illnesses and trauma. In our
study, plasma triglycerides were unaffected by minor
illness. Variation in the level of cholesterol reduction
with minor illness was seen with race and sex and with
smoking status. This variation may have resulted from
differing perceptions of what constitutes minor illness
(specifics about the illness were not queried) and from
a reduced ability of former or current smokers comAm J Epidemiol
Vol. 146, No. 7, 1997
Cholesterol in Minor Illness
561
TABLE 2. Moan level and change in plasma high density Dpoproteln cholesterol by recent illness
statue, unadjusted and adjusted* at year 7 and 10 examinations, the CARDIA Study
Unadjusted
Vbar 7 cross-sectional analysis
III in past 24 hours
Not ill in past 24 hours
Mean± SEt
Maan±SE
290
3,182
51.1 ±0.9
51.9±0.2
50.5 ± 0.7
51.7 ±0.2
-0.7 ± 0.8
0.40
-1.1 ±0.7
0.12
48.3 ± 0.9
50.4 ± 0.2
48.2 ± 0.7
49.9 ± 0.2
-2.0 ±0.8
0.0136
-1.7 ±0.7
0.0152
-0.1 ±0.6
-2.1 ±0.5
-1.7 ±0.2
0.2 ± 0.6
-2.2 ± 0.5
-1.8 ±0.2
-1.0 ± 0.4
0.0131
-1.2 ±0.4
0.0022
Difference
p value
•fear 10 cross-sectional analysis
ID in past 24 hours
Not ill in past 24 hours
309
3,219
Difference
p value
\fear 10-7 change analysis: mean
change in plasma high
density lipoprotein cholesterol
by recent illness status
ID year 7, not ill year 10
Not ill year 7, ill year 10
Not ill, year 7 and year 10
Adjusted
Mo.
197
214
2,425
Difference^
p value
* Adjusted for age, race, sex, education, smoking status, alcohol intake, body mass index, physical activity,
Kays score (see text), diabetes, and use of oral contraceptives,
t SE, standard error.
i Effect of illness on cholesterol, removing an estimate of time trend in cholesterol, is estimated as follows:
0.5 x (cholesterol change in persons ill at year 10 only - cholesterol change in persons ill at year 7 only).
pared with never smokers to differentiate chronic
cough from new, minor illness. These distinctions
could not be investigated in our study because only a
single questionnaire item was used to elicit recent
minor illness.
In support of the notion that smokers suffer more
minor illness, a slightly higher prevalence of recent
illness was observed in smokers than in nonsmokers at
the CARDIA year 7 examination. Although this finding was not replicated at the year 10 examination, it
may reflect increased frequency of minor illnesses
among smokers, for example, by increasing susceptibility to bronchitis (18). Furthermore, such illnesses
may last longer in smokers, who cannot clear their
lungs as readily as nonsmokers. This is plausible given
that smokers have been observed to have a high rate of
absenteeism from work (19).
In a classic study, Groen et al. (1) observed reductions in and subsequent rebound of serum total cholesterol associated with illness during a 12-week controlled feeding, crossover experiment. Of the 20 men
and women who consumed a strict vegetarian diet, 12
contracted gastroenteritis in week 4 of one study period. Except that these participants ate less during
Am J Epidemiol
Vof. 146, No. 7, T997
acute illness, the study diet continued unchanged during and after the illness. The mean total cholesterol
was about 200 mg/dl at the beginning of the study
period and had reached 170 mg/dl by the onset of
illness. During the illness, mean cholesterol declined
to 108 mg/dl, rebounded to 209 mg/dl in weeks 7 and
8, and then returned by weeks 10-12 to the level
typical of the vegetarian diet for these individuals,
about 170 mg/dl. Concurrent higher fat diets, eaten by
40 other people, displayed no such pattern. A similar
pattern of reduction and rebound in cholesterol was
documented for four other people who contracted cold
or flu symptoms in other feeding periods of the study.
In recent studies of cholesterol levels during infection, a similar pattern of decreased cholesterol and
increased triglycerides was observed (4, 5).
Sammalkorpi et al. (6) found that during acute viral or
bacterial infections, low density lipoprotein becomes
triglyceride rich and cholesterol poor and that lipoprotein changes are greater in more severe infections.
Decreased cholesterol is seen in asymptomatic chronic
infection with the human immunodeficiency virus type
1 (HIV-1) (20-22), and in the acquired immunodeficiency syndrome (AIDS) (22-24), whereas triglycer-
562
Jacobs et al.
TABLE 3. Mean difference (cross-sectional) and change (longitudinal) in plasma total cholesterol by recent Illness status,
unadjusted and adjusted* at year 7 and 10 examinations, stratified by race and sex, the CARDIA Study
Cross-secdonal dtfference
Year 7
Longitudinal change
from year 10 to year 7
Vfear 10
Unadjusted
Adjusted
Unadjusted
Adjusted
Unadjusted
Adjusted
Black male
Mean plasma total cholesterol difference
p value for difference
(42/837)t
-2.8
0.63
-1.9
0.73
(56/648)t
-14.1
0.0057
-16.7
0.0008
(26/40/450)+
-5.3
0.082
-5.0
0.076
Black female
Mean plasma total cholesterol difference
p value for difference
(79/850)t
-2.0
0.58
-1.0
0.78
(90/881 ) t
-2.0
0.59
-2.0
0.58
(52/53/625)+
-2.8
0.14
-1.6
0.39
White male
Mean plasma total cholesterol difference
p value for difference
(83/841 ) t
-13.1
0.0013
-13.2
0.0008
(78/B08)t
-14.2
0.0007
-13.4
0.0011
(58/63/672)+
-7.6
0.0002
-7.6
0.0002
White female
Mean plasma total cholesterol difference
p value for difference
(86/B54)t
-4.3
0.21
(85/B82)f
-0.1
0.98
-1.1
0.76
(61/58/678)+
-2.1
0.31
-2.0
0.34
-5.1
0.14
* Adjusted forage, race, sex, education, smoking status, alcohol intake, body mass index, physical activity, Keys score (see text), diabetes,
and use of oral contraceptives.
t Given in parentheses for each analysis are number ill in the past 24 hours/number not ill in the past 24 hours.
+ Given in parentheses for change analysts are number ill in the past 24 hours at year 7 but not at year 10/number not ill in the past 24
hours at year 7 but ill at year 10/number not ill in the past 24 hours at either year 7 or year 10.
TABLE 4. Mean difference (cross-sectionai) and change (longitudinal) In plasma total cholesterol by recent Mine i status,
unadjusted and adjusted* at year 7 and 10 examinations, stratified by smoking status, the CARDIA Study
Cross-secUonaJ dWerence
>fear7
Longitudinal change
from year 10 to year 7t
VbaMO
Unadjusted
Adjusted
Unadjusted
Ad)usted
Unadjusted
Adjusted
Never smokedf
Mean plasma total cholesterol difference
p value for difference
(153/1,887)+
-9.4
0,0007
-8.3
0.002
(186/1,928)+
-7.0
0.007
-7.1
0.0049
(115/135/1,490)§
-5.4
0.0001
-5.4
0.0001
Former smoker
Mean plasma total cholesterol difference
p value for difference
(51/509)+
-4.6
0.34
-4.1
0.39
(50/517)+
-4.6
0.34
-4.7
0.33
(41/34/401)§
-2.1
0.47
-0.5
0.83
Current smoker
Mean plasma total cholesterol difference
p value for difference
(86/788)+
-1.8
0.65
-0.9
0.82
(73/774)+
-8.1
0.06
-8.5
0.05
(41/45/534)§
-2.8
028
-32
0.21
* Adjusted for age, race, sex, education, smoking status, alcohol intake, body mass index, physical activity, Keys score (see text), diabetes,
and use of oral contraceptives.
t In the change analysis, smoking status is as reported at year 10. Omission of 311 individuals who reported a change in smoking
status between year 7 and year 10 yielded virtually identical results (data not shown).
+ Given in parentheses for each analysis are number ill in the past 24 hours/number not ill in the past 24 hours.
§ Given in parentheses for change analysis are number ill in the past 24 hours at year 7 but not at year 10/number not ill in the past 24
hours at year 7 but ill at year 10/number not ill in the past 24 hours at either year 7 or year 10.
ides are increased in symptomatic HTV infection and
AIDS. Beisel and Fiser (3) cautioned in 1970 regarding the need to account for the complicating presence
of an infectious agent in studies of lipid metabolism.
Blood cholesterol and its constituents are also decreased immediately after injury or surgery and remain
low for approximately 2 weeks (9, 10, 25). The reduction in cholesterol is greater in major surgery than in
more minor surgery such as laparotomy (9, 25). Blood
triglycerides initially decrease after major surgery and
then increase to levels above baseline within a few
days (9, 25).
A limitation of this study is that it was not a randomized controlled experiment. As a result, we cannot
completely rule out other explanations besides an
acute phase response for the illness-related changes in
plasma cholesterol. Although fasting times were equal
for those who were ill and not ill at each examination,
Am J Epidemiol
Vol. 146, No. 7, 1997
Cholesterol in Minor Illness
563
TABLE 5. Mean* of various characteristics by recent illness, adjusted for age, race, sex, education (age and education adjusted
only for raee and sex), the CARDIA Study
\feax 7 cross-section
11 In
past 24
hours
(n = 290)
Ago (years)
Education (years)
Body mass index (kg/m2)
Physical activity (exercise units)
Kays score (mg/dl)
AJcohd intake (ml/day)
Current smoker (%)
Ever had diabetes (%)
Current use of oral contraceptives
(women only) (%)
31.8
14.6
Not m In
past 24
houre
(n = 3.182)
3Z0
0.4
14.6
26.8
347.3
40.0
11.6
24.8
0.9
10.0
10.9
26.5
358.7
40.8
11.7
30.9
Difference
In
-O.3
0.0
-0.3
Vtoar 10 cross-section
P
value
0.19
HI In
past 24
hours
(n = 309)
35.2
14.6
27.8
-0.4
0.74
0.35
0.47
0.19
0.94
0.02
0.42
309.9
ND*
10.7
23.5
1.3
-O.9
0.63
0.2
11.5
0.8
0.1
6.1
No) II In
past 24
hours
(n = 3,219)
35.0
14.6
Difference
P
value
0.17
-0.09
0.30
-31.50
0.42
0.55
0.42
0.04
1.6
-0.63
-0.81
-0.33
0.61
0.74
0.65
1.1
-0.89
0.15
27.5
341.4
ND
11.4
24.3
* ND, diet history not done at year 10.
we have no information about respective amounts
eaten at the last meal. In accordance with the study of
Groen et al. (1), we would expect the plasma cholesterol to recover once the illness abated; however, the
interval between examinations was 3 years, and we
have no information about recovery time.
Three conclusions may be drawn from this observation of reduced plasma cholesterol in minor illness.
The first may be relevant to the epidemiologic observation of increased nonatherosclerotic disease rates
when plasma total cholesterol is low (26). Because it is
known that blood cholesterol is reduced in an acute
phase response (27), likely mediated by release of
cytokines (28), the decline in plasma total, LDL, and
HDL cholesterol observed in CARDIA is likely a
reflection of an acute phase response to minor illness.
These acute phase modifications in blood lipids and
lipoproteins may serve a functional purpose in the
short term. Groen et al. (1) speculated that cholesterol
may be reduced because it is needed to synthesize
cortical hormones in response to stress. Harris et al.
(29, 30) have shown experimentally that chylomicrons
protect rats from toxic effects of Escherichia coli
endotoxin, and Feingold and Grunfeld (31) recently
extended this finding. Endotoxin-induced mortality
was 80-100 percent in rats whose blood cholesterol
and triglycendes had been drastically reduced by either of two drugs. Mortality was 0-30 percent in
untreated control rats and in either group of drugtreated rats whose blood cholesterol and triglycerides
had been restored to the normal range by lipoprotein
infusion. Whether changes in blood lipids during major infection or trauma or in chronic infections or
inflammation have any pathogenetic significance
should be investigated.
The second conclusion is of practical importance in
survey research, given that minor illnesses such as
Am J Epidemiol
Vol. 146, No. 7, 1997
cold and flu are highly infective in local areas over
relatively brief periods. Therefore some particular surveys conducted in a narrow time frame over limited
geographic areas may include larger proportions suffering minor illness than other, parallel surveys,
whether in different localities or in different years,
resulting in biased assessment of plasma lipids. We
speculate that this phenomenon accounts for part of
the community by year component of variance in
encountered plasma cholesterol in the Minnesota
Heart Health Program (32) through which annual surveys were conducted in six localities, each within 2
months, during the 1980s. This speculation cannot be
checked because minor illness was not queried in these
surveys.
The third conclusion is of direct clinical importance.
Minor as well as major illness appears to explain part
of the variation in blood cholesterol measurements.
This variation is substantial —the unexplained withinindividual variation in blood cholesterol has a standard
deviation of 15-20 mg/dl (33). Additional research
should investigate the timing and extent of lipid
changes with respect to type and severity of major or
minor illness, as well as to other factors such as age,
gender, and smoking status. It appears that lipid assessment for cholesterol management is best done in
patients who are in their normal state of health. Future
studies may find a more substantial decrease in cholesterol when misclassification is diminished by measuring illness more precisely or in older patients. If
this is true, as suggested by the large effects observed
in the Groen et al. experiment (1), the decrease in
cholesterol associated with minor illness could be an
important source of within-individual variation that
should be factored into clinical decisions pertaining to
cholesterol management.
564
Jacobs et al.
ACKNOWLEDGMENTS
This research was supported by contracts N01-HC48047-48050 and N01-HC-95095 from the National Heart,
Lung, and Blood Institute, National Institutes of Health, and
grant R01 AG12264 from the National Institute on Aging.
REFERENCES
1. Grocn J, Tjiong BK, Kamminga CE, et al. The influence of
nutrition, individuality and some other factors, including various forms of stress, on the serum cholesterol: an experiment
of nine months duration in 60 normal human volunteers.
Voeding 1952;13:556-87.
2. Beisel WR. Impact of infectious disease upon fat metabolism
and immune functions. Cancer Res 1981;41:3797-8.
3. Beisel WR, Fiser RH. Lipid metabolism during infectious
illness. Am J Clin Nutr 1970;23:1069-79.
4. Alvarez C, Ramos A. Lipids, lipoproteins, and apoproteins in
serum during infection. Clin Chem 1986;32:142-5.
5. Lees RS, Fiser RH, Beisel WR, et al. Effects of an experimental viral infection on plasma lipid and lipoprotein metabolism. Metabolism 1972;21:825-33.
6. Sammalkorpi K, Valtonen V, Kerttula Y, et al. Changes in
serum lipoprotein pattern induced by acute infections. Metabolism 1988;37:859-65.
7. Keller H. Postoperative Laboratoriumsuntersuchungen. (In
German). (Postoperative laboratory investigations). Langenbecks Arch Chir 1973;334:631-9.
8. Wolfram G, Eckhart J, Zollner N. Storungen des Lipoproteinund Fettesaurenstoffwechsels bei Schwerverletzten. (In German). (Disturbances of lipoprotein and fatty acid metabolism
in the severely injured). Klinische Wochenschrift 1980;58:
1327-37.
9. Aufenanger J, Walter H, Katterman R. Untersuchungen zum
Lipid-und Lipoproteinstoffwechsel beim Menschen nach operativen Eingriffen. (In German). (Investigations of lipid and
lipoprotein metabolism in man after operative intervention).
Langenbecks Arch Chir 1993;378:41-8.
10. Kesteloot H, Cobbaert C, Meyns B, et al. Time course of
serum lipid and lipoprotein levels after coronary bypass
surgery: modification by pravastatin. Acta Cardiol 1992;47:
519-28.
11. Wamick GR. Enzymatic methods for quantification of lipoprotein lipids. Methods Enzymol 1986;129:101-23.
12. Wamick GR, Benderson J, Albers JJ. Dextran sulphate-Mg+
precipitation procedure for quantification of high density lipoprotein cholesterol in plasma. Clin Chem 1982;28:
1379-88.
13. Friedewald WT, Levy RI, Frederickson DS. Estimation of the
concentration of low density lipoprotein cholesterol in plasma.
Clin Chem 1972;18:499-501.
14. Friedman GD, Cutter GR, Donahue RP, et al. CARDIA: study
design, recruitment, and some characteristics of the examined
subjects. J Clin Epidemiol 1988;41:1105-16.
15. Liu K, Slattery M, Jacobs DR, et al. A study of the reliability
and comparative validity of the CARDIA dietary history. Ethn
Dis 1994 ;4:15-27.
16. Jacobs DR, Hahn LP, Haskell WL, et al. Validity and reliability of a short physical activity history: CARDIA and the
Minnesota Heart Health Program. J Cardiopulm Rehabil 1989;
9:448-59.
17. Anderson JT, Jacobs DR, Foster N, et al. Scoring systems for
evaluating dietary pattern effects on serum cholesterol. Prev
Med 1979;8:525-37.
18. Griffith DE, Perkins RC. Changing strategies for treatment of
chronic bronchitis. Semin Respir Infect 1994;9:49-57.
19. Green MS, Luz J, Gofer D. Absence from work among smokers and nonsmokers in Israeli industries—the CORDIS study.
Isr J Med Sci 1992;28:645-9.
20. Shor-Posner G, Basit A, Ying L, et al. Hypocholesterolemia is
associated with immune dysfunction in early human immunodeficiency virus-1 infection. Am J Med 1993;94:515-19.
21. Muga R, Tor J, Rey-Joly C, et al. Dislipemia e infecci6n por
VIH-1 en adictos a drogas intravenosas. (In Spanish). (Dyslipidemia and HIV-1 infection in intravenous drug addicts).
Med Clin (Bare) 1993;100:161-3.
22. Grunfeld C, Pang M, Doerrler W, et al. Lipids, lipoproteins,
triglyceride clearance, and cytokines in human immunodeficiency virus infection and the acquired immunodeficiency
syndrome. J Clin Endocrinol Metab 1992;74:1045-52.
23. Feingold KR, Krauss RM, Pang M, et al. The hypertriglyceridemia of acquired immunodeficiency syndrome is associated
with an increased prevalence of low density lipoprotein subclass pattern B. J Clin Endocrinol Metab 1993;76:1423-7.
24. Grunfeld C, Kotler DP, Shigenaga JK, et al. Circulating
interferon-alpha levels and hypertriglyceridemia in the acquired immunodeficiency syndrome. Am J Med 1991 ;90:
154-62.
25. Malmendier CL, Amerijckx JP, Bihain BE, et al. Changes in
apolipoprotein and lipids in patients after surgery. Biomed
Pharmacother 1985;39:192-5.
26. Jacobs DR Jr, Muldoon MF, Rastam L. Low blood cholesterol, nonillness mortality, and other nonatherosclerotic disease mortality: a search for causes and confounders. Am J
Epidemiol 1995;141:518-22.
27. Feingold KR, Funk JL, Moser AH, et al. Role for circulating
lipoproteins in protection from endotoxin toxicity. Infect Immun 1995;63:2041-6.
28. Ettinger WH, Harris T, Verdery RB, et al. Evidence for
inflammation as a cause of hypocholesterolemia in older people. J Am Geriatr Soc 1995;43:264-6.
29. Harris HW, Grunfeld C, Feingold KR, et al. Human very low
density lipoproteins and chylomicrons can protect against
endotoxin-induced death in mice. J Clin Invest 1990;86:
696-702.
30. Harris HW, Grunfeld C, Feingold KR, et al. Chylomicrons
alter the fate of endotoxins, decreasing tumor necrosis factor
release and preventing death. J Clin Invest 1993 ;91:1028-34.
31. Feingold KR, Grunfeld C. Role of cytokines in inducing
hyperlipidemia. Diabetes 1992;41(suppl 2):97-101.
32. Luepker RV, Murray DM, Jacobs DR, et al. Community
education for cardiovascular disease prevention: risk factor
changes in the Minnesota Heart Health Program. Am J Public
Health 1994;84:1383-93.
33. Jacobs DR, Anderson JT, Hannan P, et al. Variability in
individual serum cholesterol response to change in diet. Arteriosclerosis 1983;3:349-56.
Am J Epidemiol
Vol. 146, No. 7, 1997