High Density Lipoprotein Cholesterol, Total Cholesterol

Screening, and Myocardial Infarction

The Framingham Study
Robert D. Abbott, Peter W. F. Wilson, William B. Kannel, and William P. Castelli
The relation between high density lipoprotein cholesterol (HDL-C) and the development of myocardlal Infarction was examined In 2425 subjects, aged 50 to 79 years,
who were enrolled In the Framingham Study from 1969 to 1971. After 12 years of
follow-up, men in the bottom three quartiles of HDL-C (<52 mg/dl) experienced a 60%
to 70% excess of myocardlal Infarction as compared to men whose HDL-C levels were
higher (p<0.05). The effect of HDL-C was especially strong In women. In separate
comparisons to the 4th quartlle of HDL-C (> 67 mg/dl), the risk of myocardial Infarction
Increased from a fourfold excess In the adjacent 3rd quartlle (56 to 66 mg/dl, p<0.01)
to a nearly sixfold excess In the 1st quartile (<46 mg/dl, p < 0.001). These results
persisted after adjusting for age and other risk factors. In addition, a significant effect
of HDL-C remained In subjects who had the lowest concentrations of total cholesterol
(=s 192 mg/dl In men and 211 mg/dl In women) In which 29% had levels of HDL-C (s 36
mg/dl In men and 46 mg/dl In women) that were associated with a marked elevation In
the Incidence of myocardlal Infarction. We conclude that screening for total cholesterol alone In men and women aged 50 and older may not adequately Identify the coronary candidate. In addition, selective screening of HDL-C only for Individuals with
high concentrations of total cholesterol can leave the false Impression that low total
cholesterol Is uniformly associated with a healthy risk profile.
(Arteriosclerosis 8:207-211, May/June 1988)

everal epidemiologic studies have demonstrated that

high density lipoprotein cholesterol (HDL-C) is inversely related to the incidence of coronary heart disease
morbidity and mortality.1"7 Fewer studies have established
a relation between HDL-C and specific coronary manifestations, including angina pectoris, coronary insufficiency,
and myocardial infarction.2-3- * In addition, some researchers report an uncertain or questionable association,8'9i 10
while most fail to provide data for women.3"10 In this report,
a closer examination of the effect of HDL-C as a determinant of the specific manifestation of myocardial infarction
is presented for men and women aged 50 and older. We
describe the incidence of myocardial infarction as it occurred for various ranges of HDL-C, after adjusting for age
and other cardiovascular risk factors. We further examine
the effect of HDL-C within groups of subjects with low and
high concentrations of total cholesterol and the resulting
implications for screening for total cholesterol alone. Findings for this report are based on 12 years of follow-up of a
sample of subjects originally enrolled in the Framingham
Study.

Methods
Since 1948, the Framingham Study has biennially followed 5209 men and women for the development of cardiovascular disease. Sampling methods, response rates,
follow-up, and examination procedures have been described elsewhere.11-12
From 1969 to 1971, fasting levels of HDL-C were first
determined among study participants after heparin-manganese chloride precipitation of fresh plasma, following a
modified protocol adopted by the Lipid Research Clinics.13
For this report, subjects were followed for 12 years from
the time of HDL-C measurement for the development of
myocardial infarction. All subjects were free of coronary
heart disease (including myocardial infarction, angina pectoris, and coronary insufficiency) at the time the follow-up
began. Further details on the definition of coronary heart
disease are given elsewhere.14'15
In the course of follow-up, the diagnosis of myocardial
infarction was based on electrocardiogram (ECG) and enzyme criteria. The ECG criteria included ST segment elevation associated with terminal inversion of T waves and
the loss of initial QRS potential (pathologic Q waves of
0.04 second or greater), followed by serial changes indicating reversion toward normal. A stable ECG pattern,
including pathologic Q waves, or a loss of initial QRS potential in those leads in which this would not be expected,
was used as evidence for an old or remote myocardial
infarction. An interim myocardial infarction (between clinic
exams) was noted when changes from a previous tracing

Robert D. Abbott is at the Statistical Resource Section, National

Heart, Lung, and Blood Institute, Bethesda, Maryland. Peter W.F.
Wilson and William P. Castelli are at the Framingham Epidemiology Research Section, National Heart Lung, and Blood Institute,
Framingham, Massachusetts. William B. Kannel is at the Section
of Preventive Medicine and Epidemiology, Boston University
School of Medicine, Evans Research Foundation, Boston, Massachusetts.
Received August 7, 1987; revision accepted November 4,
1987.

207

208

ARTERIOSCLEROSIS

VOL

showed development of an unexplained loss of R wave

potential or the appearance of pathologic Q waves.
The enzyme criteria for the diagnosis of myocardial infarction were based on hospital reports showing abnormal
concentrations of SGOT of at least 50 units/I, LDH of at
least 200 units/I, CPK greater than 100 ILJ/ml, or positive
CPK-MB isoenzyme, along with a history of prolonged
ischemic pain. When available, autopsy reports were also
used to provide evidence of myocardial infarction.
To illustrate the effects of HDL-C, 12-year, age-adjusted
incidence rates of myocardial infarction were calculated by
quartiles of HDL-C. For men, the 1st through 4th quartiles
consisted of the following ranges of HDL-C, respectively:
12 to 36, 37 to 44, 45 to 52, and 53 to 129 mg/dl. For
women, corresponding quartiles were as folbws: 23 to 46,
47 to 55, 56 to 66, and 67 to 139 mg/dl.
For purposes of detecting an independent effect of
HDL-C on myocardial infarction, control of potential confounding risk factors was necessary. Important risk factors
measured at the beginning of follow-up included total cholesterol, systolic blood pressure, body mass index [weight
divided by height squared (kg/m2)], diabetes, current ciga :
rette use, and the receipt of postmenopausal estrogen
replacement therapy in women.
To help illustrate the relation between the additional risk
factors and concentrations of HDL-C, mean levels of each
risk factor were compared across quartiles of HDL-C.
Adjustments were made for age by using analysis of
covariance and logistic regression models.16 Techniques
used to measure the risk factors have been reported
elsewhere.15
To estimate the independent effect of HDL-C, proportional hazards models17 were used to follow subjects over
12 years for the development of a myocardial infarction.
Relative risks of myocardial infarction were estimated
comparing the 1st through 3rd quartiles of HDL-C to the
4th quartile. Estimates of relative risk were based on the
corresponding regression coefficient associated with each
of the lower three quartiles compared separately to the 4th
quartile. Comparisons were adjusted for age alone and
also for age and the preceding covariates.
In addition, the effect of HDL-C within each quartile of
total cholesterol was examined. For men, the 1st through
4th quartiles consisted of the following ranges of total cholesterol concentrations, respectively: 116 to 192, 193 to
215, 216 to 244, and 245 to 376 mg/dl. For women, corresponding quartiles were as follows: 124 to 211,212 to 237,
238 to 266, and 267 to 412 mg/dl.
The fit of the proportional hazards model was examined
by observing changes in the proportionality assumption
with time and with risk factor level. The results indicated
that use of such models was appropriate. All tests of significance were two-sided.

Results
Of the 5209 participants in the original Framingham cohort, fasting levels of HDL-C were first determined in 2788
subjects during biennial examinations received from 1969
to 1971. Among this group, 1007 men and 1418 women
were free of coronary heart disease and constituted the

8, No 3,

MAY/JUNE

1988

subjects for this study. At the beginning of follow-up, subjects were aged 50 to 79 years.
Table 1 gives the age-adjusted, 12-year incidence of
myocardial infarction by quartile of HDL-C. For both sexes,
the highest rate of myocardial infarction occurred among
subjects whose HDL-C was in the lowest quartile
(15.3/100 for men and 10.6/100 for women). These rates
were significantly greater (p<0.05) than the rates experienced by those whose HDL-C fell in the highest quartile
(9.4/100 for men and 1.4/100 for women).
For men, the rates of myocardial infarction in the 2nd
and 3rd quartiles of HDL-C were similar (14.8/100 and
14.9/100, respectively). For women, the rate of myocardial
infarction declined significantly with each increase in
HDL-C quartile (p<0.01).
Table 2 provides the age-adjusted mean levels of the
selected risk factors. For women, total cholesterol increased with each increase in HDL-C quartile. In each sex
group, subjects in the lowest quartile of HDL-C had significantly lower levels of total cholesterol as compared to
those in the highest quartile (p<0.05).
For women, systolic blood pressure decreased with increased HDL-C. Women in the lowest quartile of HDL-C
had significantly higher blood pressures than those in the
top quartile (p<0.05). No such association was seen in
men.
Body mass index was the risk factor most strongly related to HDL-C. Compared to subjects in each of the bottom
three quartiles of HDL-C, those in the highest quartile were
significantly leaner (p<0.01).
In both sexes, diabetes was also more common in those
in the lowest quartile of HDL-C as compared to the other
quartiles. Compared to women in the top quartile, it was
significantly more frequent in the lowest quartile, as was
the use of cigarettes (p<0.05). Past or current use of
postmenopausal estrogens by women was also significantly less frequent in the 1 st and 2nd HDL-C quartiles as
compared to the top quartile (p<0.01).
Table 3 provides estimates of the relative risk of myocardial infarction comparing each of the 1st through 3rd quarTable 1. Age-adjusted 12-Year Incidence (Rate/100)
of Myocardial Infarction by Quartile of High Density
Llpoproteln Cholesterol
HDL-C
quartile (mg/dl)
Men
1st (12 to 36)
2nd (37 to 44)
3rd (45 to 52)
4th (53 to 129)
Total
Women
1st (23 to 46)
2nd (47 to 55)
3rd (56 to 66)
4th (67 to 139)
Total

Number
at risk

Number
of events

Age-adjusted
rate

239
275
231
262
1007

36
41
34
25
136

15.3*
14.8
14.9
9.4
13.5

356
354
353
355
1418

38
26
23
5
92

10.6*
7.4*

6.5f
1.4
6.5

Significant excess of events compared to the 4th quartile:

*p<0.05, fP<0.01, $p<0.001.

HDL CHOLESTEROL AND MYOCARDIAL INFARCTION

Table 2. Age-adjusted Mean Levels of Selected Risk
Factors by Quart Me of High Density Lipoproteln Cholesterol
HDL-C quartile
Risk factor

4th

221.5
(42.9)
138.5
(20.3)
27.1t
(3.5)

217.6
(37.6)
139.4
(20.3)

26.2f
(3.1)

140.7
(21.0)
25.3
(3.4)

9.3
(29.3)
32.5
(46.7)

7.9
(26.9)
34.6
(47.8)

8.7
(28.4)
35.6
(47.9)

19.5t
(39.2)

19.4f
(39.6)

242.7
(39.4)
139.4
(22.7)
25.6*
(4.0)
7.5
(26.6)
32.1
(46.5)
23.6
(42.3)

243.9
(38.9)
138.8
(21.5)
24.3
(3.8)

(46.8)

238.9
(41.1)
139.6
(22.1)
26.1*
(4.0)
6.4
(24.7)
29.7
(45.7)

213.3t
(41.6)

140.6
(22.4)
27.9$
(3.6)
11.9
(32.2)
35.1
(48.0)

236.4*
(45.8)

Diabetes (%)
Cigarette use (%)

142.4*
(22.4)
27.5*
(5.0)
11.5*
(32.3)
33.7*

Past or current estrogen use (%)

223.8
(38.8)

6.7
(24.1)
26.6
(45.5)
29.2
(46.0)

The numerals in parentheses are standard deviations.

Significantly different from the 4th quartile: *p < 0.05, t p < 0.01,
* p < 0.001.

Table 3. Estimated Relative Risk of Myocardlal

Infarction for the 1 st through 3rd Quartiles of HDL-C as
Compared to the 4th Quartile
HDL-C quartile
comparison
1st versus 4th

Men

Women

Risk factor
Ageadjusted adjusted

Risk factor
Ageadjusted adjusted

1.7*

0-0,2.8)
2nd versus 4th
3rd versus 4th

209

Table 4. Age-adjusted 12-Year Incidence (Rate/100)

of Myocardlal Infarction by Quartile of HDL-C within
Each Quartile of Total Cholesterol
HDL-Cquartile

Total cholesterol quartile (mg/dl)

1st

2nd

3rd

4th

116 to 192*

193 to 215
17.0(10/59)
18.0(11/61)
12.5 (8/64)
13.1 (10/76)

216 to 244
12.9 (8/67)
18.3(11/59)
20.2(12/60)
8.7 (6/64)

245 to 376
14.3 (6/42)
14.2(12/84)
19.7 (0/51)
10.9 (8/73)

Men

Cigarette use (%)

Women
Total cholesterol
(mg/dl)
Systolic blood pressure (mm Hg)
Body mass index
(kg/m2)

3rd

2nd

1st

Men
Total cholesterol
(mg/dl)
Systolic blood pressure (mm Hg)
Body mass index
(kg/m2)
Diabetes (%)

Abbott et al.

1.6
(1.0,2.7)
1.6
(1.0,2.8)

1.7
(1.0,2.9)
1.7*
(1.0,2.9)
1.6
(1.0,2.8)

7.6*
5.8*
(3.0,19.5) (2.2,15.2)
5.3*
4.9*
(2.0,13.7) (1.9,12.8)

4.6f

4.0t

(1.8,12.2) (1.5,10.6)

The risk factor adjusted relative risks are adjusted for age, total
cholesterol, systolic blood pressure, body mass index, diabetes,
cigarette use, and the receipt of estrogen therapy by women. The
numerals in parentheses are the 95% confidence intervals.
Significant relative risk: *p<0.05, t P < 0 . 0 1 , * p < 0.001.

tiles of HDL-C to trie 4th quartile. Relative risks are adjusted for age alone, and for age and the selected risk factors
described in Table 2.
Among men in the 1st through 3rd quartiles of HDL-C,
the risk of myocardial infarction was 60% to 70% greater
than in the top quartile. Only the excess risk in the 1st
quartile of HDL-C was significantly greater than the risk in
the 4th quartile after adjustment for age (p<0.05). When
adjusted for the other risk factors, this excess was no long-

1st 17.1 (12/71)

2nd 9.5 (7/71)
3rd
7.0 (4/56)
4th
2.2 (1/49)
Women
124 to 211*
9.4(10/104)
1st
2nd 5.4 (5/91)
6.0 (5/84)
3rd
1.3 (1/80)
4th

212 to 237* 238 to 266* 267 to 412f

7.6 (7/91) 12.6(10/77) 13.1 (11/84)
7.8 (7/89) 5.6 (5/90) 10.7 (9/84)
8.9 (8/90)
5.9 (7/83) 5.3 (5/96)
2.0 (2/99)
0.0 (0/85) 2.3 (2/91)

The numerals In parentheses represent the number of events/

number at risk.
Significant HDL-C effect In the given range of total cholesterol:
*p<0.05, t P < 0 . 0 1 .

er significant, although a significant excess in the 2nd

quartile emerged. When comparing the incidence of myocardial infarction among men in the combined bottom three
quartiles of HDL-C ( s 5 2 mg/dl), men in the top quartile
were at significantly lower risk (p<0.05).
For women, the inverse association between HDL-C
and myocardial infarction was much clearer. Without exception, lower concentrations of HDL-C were significantly
associated with an increased relative risk of myocardial
infarction. After adjusting for age and the other risk factors,
those in the bottom 25% of HDL-C values experienced
nearly six times the rate of coronary events as compared to
those in the top 25% (p < 0.001). Women in the middle two
quartiles of HDL-C also experienced an excess of coronary events as compared to the highest quartile. For those
in the 2nd quartile of HDL-C, there was a nearly fivefold
excess (p<0.01), and in the adjacent 3rd quartile of moderately high levels of HDL-C, a fourfold excess was experienced (p<0.01).
Table 4 gives the 12-year, age-adjusted incidence of
myocardial infarction by quartile of HDL-C within each
quartile of total cholesterol. In both sexes, the effect of low
HDL-C levels persisted in subjects who had the lowest
concentrations of total cholesterol. For men whose total
cholesterol fell below the 25th percentile, the 12-year incidence of myocardial infarction declined significantly
(p<0.01) from 17.1/100 in the 1st quartile of HDL-C to
2.2/100 in the 4th quartile. For women, a similar and significant (p< 0.05) decline from 9.4/100 in the 1 st HDL-C quartile to 1.3/100 in the top quartile was observed.
The effect of HDL-C appeared to diminish in men with
total cholesterol concentrations exceeding 192 mg/dl. Although not significant, those in the top quartile of HDL-C
usually experienced the lowest rates of myocardial infarction. In contrast, the effect of HDL-C in women remained
graded and strong for the entire distribution of total cholesterol values (p<0.05). In addition, the effect of HDL-C was

210

ARTERIOSCLEROSIS

VOL

not significantly altered in the presence of high or low concentrations of total cholesterol.
Discussion
The focus of this report is on individuals aged 50 and
older, where findings demonstrate that HDL-C is inversely
related to the development of myocardial infarction, particularly in women. In addition, the data provide further
evidence of an effect of low levels of HDL-C on the development of hard or definite coronary events, comprising
53% of all coronary heart disease observed in the Framingham sample.
Inclusion of coronary deaths without a documented
myocardial infarction (occurring in 37 of 1007 men and 23
of 1418 women) failed to alter considerably the magnitude
of these findings. Including such events increased the relative risk calculations in Table 3 only slightly for men. Nevertheless, the excess risk of either myocardial infarction or
death from coronary heart disease in each of the bottom
HDL-C quartiles significantly exceeded the rate in the top
quartile in separate comparisons ( p < 0.05). In women, the
corresponding relative risks remained significant, but were
lower than those displayed in Table 3, increasing from a
threefold excess of events in the 3rd quartile of HDL-C
(p<0.01) to a fivefold excess in the 1st quartile
( p < 0.001). In both men and women, however, most of the
coronary deaths that occurred without a documented myocardial infarction occurred suddenly (45 of 60 within 1 hour
of symptom onset) and may have involved other risk factors unique to sudden death.
The effect of including other coronary endpoints, such
as angina pectoris and coronary insufficiency, has been
described elsewhere in a recent publication.1 In that report,
the effect of HDL-C as a continuous variable was examined. In the current report, while choosing to consider a
more objective coronary manifestation, we examined disease incidence across ranges of HDL-C. Doing so eliminates the need to assume a strict linear relationship involving HDL-C and enables detection of an excess of event
rates in categories of HDL-C that may not be obvious from
constrained parametric models. The additional stratification of total cholesterol into quartiles also permits the detection of an interaction effect with HDL-C, which appears
to be absent.
Unfortunately, the data and analyses remain limited, as
it is difficult to determine how much myocardial disease
could be prevented by increasing HDL-C levels in a population of men and women aged 50 and older. It seems
possible that some disease could be prevented, based on
the theory that HDL-C can remove cholesterol from arterial
walls. 18 ' 19 Other mechanisms or metabolic derangements
may also exist, however. Clearly, such mechanisms are
complicated, since low concentrations of HDL-C often
appear with additional atherosclerotic traits, especially
hypertriglyceridemia and obesity. Additional traits may
also include hyperinsulinemia, clotting abnormalities, sex
hormone imbalances,20'21 and other risk factors not routinely measured at all Framingham examinations.22'23'24
It seems apparent, however, that elevated levels of
HDL-C are protective against coronary heart disease.

8, No 3,

MAY/JUNE

1988

Based on the Framingham data, optimal HDL-C levels

exceeded 52 mg/dl for men and 66 mg/dl in women.
Whether these levels apply to individuals younger than 50
needs further study.
Unfortunately, there are no data that indicate whether
therapies for specifically raising HDL-C will reduce the risk
of coronary heart disease. Additional studies should focus
on the benefits of weight reduction,25 dieting, 1926 and
physical activity27 to increase HDL-C per se and reduce
coronary risk.
Successful attempts at increasing HDL-C may have
their most noticeable effects in women aged 50 and older,
where the association between HDL-C and myocardial infarction was particularly strong in the Framingham data.
Among those in the Framingham sample, 75% had HDL-C
levels of less than 67 mg/dl, placing them at four to six
times the risk of developing a myocardial infarction compared to women with greater HDL-C levels. In addition, the
relative health advantage of women, as compared to men,
in terms of tower rates of coronary heart disease,28 begins
to disappear when levels of HDL-C are at 46 mg/dl or less.
The association between HDL-C and myocardial infarction also persisted for low levels of total cholesterol in both
men and women. This might be unexpected given the
hypothesis that HDL-C Is protective due to reverse cholesterol transport. Nevertheless, defects in catabolism of very
low density lipoprotein (VLDL) cholesterol may result in
reduced fasting HDL-C and low density lipoprotein (LDL)
cholesterol through the precursor-product relationship
among lipoproteins. As a result, levels of VLDL cholesterol
(and remnant particles) could become elevated and possibly contribute to atherosclerosis. This may be a reasonable explanation in the postprandial state, since control of
fasting LDL and VLDL cholesterol in the Framingham data
failed to explain or diminish the independent association
between HDL-C and myocardial infarction. Furthermore,
within low and high ranges of the less dense lipoprotein
particles, in addition to triglyceride, the effect of low HDL-C
levels remained, as it did across levels of total cholesterol.
Additionally, based on the predictive strength of these various lipid fractions, HDL-C appears to be the most important measure of coronary risk in the Framingham sample.
Subjects who had low levels of HDL-C also comprised a
considerable proportion of those who had low concentrations of total cholesterol (=s192 mg/dl in men and 211
mg/dl in women). Of those whose total cholesterol fell in
the 1st quartile, approximately 29% (71/247 men and
104/359 women) had levels of HDL-C (=s36 mg/dl in men
and 46 mg/dl in women) that were associated with a
marked elevation in the incidence of myocardial infarction.
In addition, among the latter group who went on to have
a myocardial infarction (12 men and 10 women), most
were free of other cardiovascular risk factors. For both men
and women, 80% were normotensive and 90% were nonobese (< 30 kg/m2). None of the subjects had levels of LDL
or VLDL cholesterol that exceeded the sex-specific 90th
percentile for the entire Framingham sample. Only 9% had
triglyceride concentrations above the 90th percentile.
Among the ten women with a myocardial infarction, 80%
were nondiabetic and 56% were nonsmokers. The only
possible risk factors in men that could be used in place of

HDL CHOLESTEROL AND MYOCARDIAL INFARCTION

HDL-C as markers of high coronary risk included diabetes
and cigarette smoking. Nevertheless, nondiabetics and
nonsmokers comprised a large proportion of the 12 men
who had a myocardial infarction (50% and 33%, respectively). As a result, there appears to be no clear way to
predict for men or women which subjects with low levels of
total cholesterol are at high risk of coronary heart disease
and should have HDL-C concentrations measured.
We conclude, therefore, that if cholesterol screening is
to occur in men and women aged 50 and older, a screening
for total cholesterol alone may not adequately identify the
coronary candidate. Such attempts at screening for total
cholesterol, while potentially beneficial for some, can incorrectly classify a substantial proportion of the general
population, unless HDL-C determinations are also made.
In addition, selective screening of HDL-C only for individuals with elevated concentrations of total cholesterol can
leave the false impression that low total cholesterol is uniformly associated with a healthy risk profile. Whether these
conclusions apply to younger individuals needs further
study.

11.
12.

13.
14.

15.

16.
17.
18.

Index Terms:

HDL cholesterol

19.
20.
21.

22.
23.
24.
25.

26.

27.

28.

cholesterol screening

211

lipoprotein cholesterol is not a major risk factor for Ischemic

heart disease in British men. Br Med J 1986:292:515-519
Dawber TR, Kannel WB, Lyell LP. An approach to longitudinal studies in a community: The Framingham Study. Ann NY
Acad Sd 1963;107:539-556
Gordon T, Moore FE, Shurtleff D, et al. Some methodologic
problems in the long term study of cardiovascular disease:
Observations from the Framingham Study. J Chronic Dis
1959;10:186-206
Upld Research Clinics Program. Manual of Laboratory Operations, vol 1. Upid and lipoprotein analysis. DHEW publ no
(NIH) 75-628. Bethesda, MD: NIH 1974
Sortie P. Cardiovascular diseases and death following myocardial infarction and angina pectoris: Framingham Study, 20
year follow-up, publication (NIH) 77-1247. Washington DC:
Dept of Health, Education, and Welfare 1977
Shurttcff D. Some characteristics related to the incidence of
cardiovascular disease and death: The Framingham Study,
18-year follow-up, publication (NIH) 74-559. Washington DC:
Dept of Health, Education, and Welfare 1974
Lane PW, Nelder JA. Analysis of covariance and standardization as instances of prediction. Biometrics 1982;38:613621
Cox DR. Regression models and life tables (with discussion).
J R Stat Soc 1972;34:series B:187-220
Levy Rl, Rtfklnd BM. The structure, function and metabolism
of high-density lipoproteins: A status report. Circulation 1980;
62:IV4-8
Grundy SM. Cholesterol and coronary heart disease: A new
era. JAMA 1986256:2849-2858
Gutal J, LaPorte R, Kuller L, et al. Plasma testosterone,
high density lipoprotein cholesterol and other lipoprotein fractions. Am J Cardiol 1981 ;48:897-902
Evans DJ, Hoffman RG, Kalkoff RK, Klssebah AH. Relationship of adrogenic activity to body fat topography, fat cell
morphology, and metabolic aberrations in premenopausal
women. J Clin Endocrinol Metab 1983;57:304-310
Klssebah AH, Nadarajen V, Murray R, et al. Relation of
body fat distribution to metabolic complications of obesity. J
Clln Endocrinol Metab 1982;54:254-259
Orchard TJ, Becker DJ, Bates M, Kuller LH, Drash AL.
Plasma insulin and lipoprotein concentrations: An atherogenlc association? Am J Epidemiol 1983;118:325-337
Wilson PWF, Garrison RJ, Abbott RD, Castelli WP. Factors associated with lipoprotein cholesterol levels: The Framingham Study. Arteriosclerosis 1983;3:273-281
Lasser NL, Grandrts G, Gagglula AW, et al. Effects of antihypertensive therapy on plasma llplds and lipoproteins in the
Multiple Risk Factor Intervention Trial. Am J Med 1984;
76:52-66
Ernst N, Fisher M, Bowen P, et al. Changes in plasma lipids
and lipoproteins after a modified fat diet. Lancet 1980;
1:111-112
Huttunen JK, Lanslmles E, Voutllalnen E, et al. Effect of
moderate physical exercise on serum lipoproteins: A controlled clinical trial with special reference to serum high-density lipoproteins. Circulation 1979;60:1220-1229
Lerner DJ, Kannel WB. Patterns of coronary heart disease
morbidity and mortality in the sexes: A 26-year follow-up of
the Framingham population. Am Heart J 1986;111:383-390

References
1. Castelli WP, Garrison RJ, Wilson PWF, et al. Incidence of
coronary heart disease and lipoprotein cholesterol levels.
JAMA 1986;256:2835-2838
2. Gordon T, Casts) 11 WP, HJortland MC, et al. High density
lipoprotein as a protective factor against coronary heart disease: The Framingham Study. Am J Med 1977;62:707-714
3. Enger SC, Hjermann I, F O M OP, et al. High density lipoprotein cholesterol and myocardial infarction or sudden death: A
prospective case-control study in middle aged men of the
Oslo Study. Artery 1979;5:170-181
4. Watklna LO, Neaton JD, Kuller LH (forthe MRFIT Research
Group). High density lipoprotein cholesterol and coronary
heart disease Incidence in black and white MRFIT usual care
men. Circulation 1985;71:417A
5. Goldbourt U, Hottzman E, Neufeld HN. Total and high density lipoprotein cholesterol In the serum and risk of mortality:
evidence of a threshold effect. Br Med J 1985;290:12391243
6. Gordon DJ (for the Upid Research Clinics Coronary Primary
Prevention Trial Investigators). Plasma high-density lipoprotein cholesterol and coronary heart disease in hypercholesterolemic men. Circulation 1985;72:lll-185
7. Miller NE, Thelle DS, Forde OH, et al. The Tromso Heart
Study: High-density lipoprotein and coronary heart disease: A
prospective case-control study. Lancet 1977;2:965-968
8. Keys A. Alpha lipoprotein (HDL) cholesterol in the serum and
risk of coronary heart disease and death. Lancet 1980;2:
603-606
9. Keys A, Karvonen MJ, Punsar S, et al. HDL serum cholesterol and 24-year mortality of men in Finland. Int J Epidemiol
1984;13:428-435
10. Pocock SJ, Shaper AQ, Phillips AN, et al. High density

Abbott et al.

myocardial infarction

Framingham Study