You are on page 1of 11

1305

Association Between Blood Pressure and


Serum Lipids in a Population
The Troms0 Study
Kaare H. B0naa, MD, and Dag S. Thelle, MD

Background. High blood pressure has been associated with elevated atherogenic blood lipid
fractions, but epidemiological surveys often give inconsistent results across population sub-
groups. A better understanding of the relation between blood pressure and blood lipids may
provide insight into the mechanism(s) whereby hypertension is associated with increased risk
of coronary heart disease.
Methods and Results. We assessed the cross-sectional relations of serum total cholesterol, high
density lipoprotein (HDL) cholesterol, non-HDL cholesterol (total minus HDL cholesterol),
and triglyceride levels with blood pressure in a population of 8,081 men 20-54 years old and
7,663 women 20-49 years old. Stratified analyses and multivariable methods were used to
control for potential confounding anthropometric and lifestyle variables. Total and non-HDL
cholesterol levels increased significantly with increasing systolic or diastolic blood pressure in
both sexes. Men 20-29 years old had steeper regression slopes for blood pressure by total
cholesterol level than did women of similar age. In men, the association between blood pressure
and total cholesterol level decreased with age, whereas in women, it increased with age. Body
mass index modified the relation, whereas smoking, physical activity, and alcohol consumption
had little influence on the association. Triglyceride levels increased with blood pressure, but
this relation was weak in lean subjects. HDL cholesterol level correlated positively with blood
pressure in population subgroups having a high alcohol consumption.
Conclusion. These results support the hypothesis that there are biological interrelations
between blood pressure and blood lipids that may influence the mechanisms whereby blood
pressure is associated with risk of coronary heart disease. (Circulation 1991;83:1305-1314)

H ypertension and hypercholesterolemia each some investigators consider the association to be of


predispose to coronary heart disease, but little biological importance.7'11
the two acting in concert alter risk substan- Recent experiments indicate that impaired endo-
tially because their combined effects are considered thelium-dependent vascular relaxation in patients
to be multiplicative rather than additive.1-3 Persons with essential hypertension14 may be associated with
with a combination of risk factors are, therefore, at hypercholesterolemia,15,16 and it has been suggested
particularly high risk of coronary heart disease. that low density lipoprotein (LDL) cholesterol in
Hypertensive subjects frequently have higher choles- itself may be a modifiable risk factor for hyperten-
terol levels than do normotensive subjects.4-6 A sion.17 Also, heightened sympathetic activity or pe-
positive relation between serum cholesterol level and ripheral insulin resistance may be related to hyper-
blood pressure has been reported in many epidemi- lipidemia in hypertension.18'19
ological studies,37-13 but the results have often been Few attempts have been made to analyze the relation
inconsistent across population subgroups,10-13 and between blood pressure and serum lipid levels or to
assess the determinants within a population for this
From the Institute of Community Medicine, University of association. A better understanding of the interrelation
Troms0, Troms0, Norway. between blood pressure and blood lipid levels may be
Supported by the Norwegian Research Council for Science and
the Humanities and by the Norwegian Council on Cardiovascular of relevance for the understanding of how essential
Diseases. The screening was carried out in cooperation with the hypertension is related to the etiology and pathogenesis
National Health Screening Service, Oslo. of arteriosclerosis and, thereby, also for selecting the
Address for correspondence: Kaare H. B0naa, Institute of appropriate therapeutic approach. We, therefore, stud-
Community Medicine, University of Troms0, Postuttak, N-9000
Troms0, Norway. ied the cross-sectional relation of blood pressure to
Received August 14, 1990; revision accepted December 4, 1990. serum total cholesterol, high density lipoprotein (HDL)

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


1306 Circulation Vol 83, No 4 April 1991

cholesterol, and triglyceride levels in a free-living pop- with the participants in a sitting position; a tourni-
ulation of more than 16,000 young and middle-aged quet was used but was released before sampling.
men and women. This large data set allowed an analysis After 30 minutes, the coagulated samples were cen-
of how these relations depend on sex, age, body weight, trifuged at 1,OOOg for 10 minutes, the serum was
and lifestyle factors. transferred to plastic tubes, kept between 1C and
10C, and analyzed within 48 hours (samples drawn
Methods before weekends were analyzed within 80 hours).
Subjects Total cholesterol level was measured directly by the
In 1979-1980, all men 20-54 years old and all enzymatic oxidase method, using a commercially
women 20-49 years old (21,329 subjects altogether) available kit (Boehringer Mannheim 148393, Mann-
living in the municipality of Troms0, Norway, were heim, FRG), and HDL cholesterol was assayed by
invited to participate in a health survey. Seventy- the same procedure after precipitation of lower
eight percent (n=16,621) of those invited attended. density lipoproteins with heparin and manganese
The attendance rate was higher in women than in chloride. Non-HDL cholesterol level was calculated
men, and it increased with age; from 72.4% in women as total cholesterol minus HDL cholesterol level.
20-29 years old to 91.2% in women 40-49 years old Triglycerides were enzymatically determined as glyc-
and from 61.6% in men 20-29 years old to 85.1% in erol (Boehringer 15725). Participants were examined
men 50-54 years old. The methods were similar to about time since last meal. The age- and sex-adjusted
those of the former Tromso study20 and the Norwe- meanSEM levels of serum triglycerides were
gian county studies.21 After exclusion of 288 subjects 1.470.01, 1.42+0.01, 1.330.01, 1.220.02, and
receiving current treatment for hypertension, 51 di- 1.100.04 mmol/l in subjects that had fasted less
abetics, 219 pregnant women, and 319 subjects with than 1, 1-2, 2-4, 4-8, and more than 8 hours
missing values for any measurement, 8,081 men and (p<0.0001 for linear trend). In 323 men and women
7,663 women were included in the present analysis. who had fasted for more than 8 hours, very low
density lipoprotein (VLDL) cholesterol (mmol/l)
Questionnaires level was calculated as 0.47 multiplied by triglyceride
level (mmol/l), and LDL cholesterol level was calcu-
The letter of invitation contained questions on lated as total cholesterol minus the sum of VLDL
medical history, current use of antihypertensive drugs and HDL cholesterol levels, according to the formula
(yes/no), smoking (yes/no), and physical activity in of Friedewald et al.27 All the laboratory measure-
leisure time (sedentary, moderate, intermediate, or ments were performed at the Institute of Medical
intensive). The validity of the responses to questions Biology, University of Troms0.
on use of drugs, smoking, and physical activity was
confirmed by checking medical records,22 measure- Statistical Analysis
ments of serum thiocyanate concentrations,23 and Two-way analysis of covariance was performed
bicycle ergometry,24 respectively. The questionnaire separately for each sex and systolic and diastolic
was checked for logistics at the examination, and blood pressure with the sPss-x program.28 The analy-
women were interviewed about the use of oral contra- ses were performed with the lipids as dependent
ceptives and menopausal status. A second question- variables, blood pressure, and age (5-year age
naire that also covered alcohol habits was filled in at groups) as grouping variables, and log1o body mass
home and returned by 88.2% of those who attended index as the covariate. Because analyses using log-
the survey. The reported frequency of inebriation transformed and nontransformed triglyceride values
(never or not last year, a few times per year, or once or gave similar results, the latter were used. Time since
more per month) is strongly associated with levels of last meal was included as a covariate only in the
serum y-glutamyltransferase levels25 and with the analyses of triglyceride levels. Linear trend was eval-
usual alcohol consumption26 in this population. uated by testing partial regression coefficients in
corresponding multiple linear regression analyses of
Measurements the lipids. The frequency distribution of total choles-
The participants were not requested to fast. Height terol level was compared in subjects with diastolic
and weight were measured in light clothing without blood pressure less than 70 mm Hg and greater than
shoes: body mass index was calculated as the weight 99 mm Hg in a pooled analysis of men and women. In
in kilograms divided by the square of the height in this analysis, age adjustment was done by the direct
meters. Blood pressure was measured by nurses method29 with 5-year age groups and the total study
trained according to tape recordings produced by the population (n = 15,744) as standard.
London School of Hygiene and Tropical Medicine, Next, the age-adjusted regression coefficients
with the first and fifth Korotkoff phases recorded as (slopes) for blood pressure on the lipids were calcu-
systolic and diastolic blood pressures, respectively. lated within strata divided by age, body mass index,
After 4 minutes of rest, two readings were taken with smoking habits, physical activity at leisure, and the
a 1-minute interval, using a standard stethoscope and frequency of inebriation by alcohol. Residual analysis
mercury sphygmomanometer. The lower value is within each stratum did not reveal extreme data
used in this report. Venipuncture was performed points influencing the results. The significance of

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


B0naa and Thelle Association of Blood Pressure With Serum Lipids 1307

differences between the regression coefficients was pressure relations were similar within strata divided
tested according to the formula according to levels of physical activity in leisure time,
alcohol consumption, or (in women) the use of oral
-B2)
z = (B1 2 contraceptives (data not shown).
where B1 and B2 are the regression coefficients and Figure 2 shows that the age-adjusted frequency dis-
SE1 and SE2 are the respective standard errors. The tribution of total cholesterol level is shifted to the right
mean values for cholesterol by diastolic blood pres- for persons having diastolic blood pressure greater than
sure and age were computed. The mean values for 99 mm Hg compared with those having diastolic blood
triglyceride levels were computed by diastolic blood pressure less than 70 mm Hg. The cholesterol distribu-
pressure and body mass index (adjusted for age and tion curve for subjects with diastolic blood pressure of
time since last meal by one-way analyses of covari- 70-99 mm Hg had an intermediate position. Among
ance). Last, we evaluated the independent associa- persons with diastolic blood pressure greater than 99
tion between the lipids and blood pressure by testing mm Hg, 61.5%, 29.7%, 11.1%, and 4.5% had total
partial regression coefficients in multiple linear re- cholesterol levels greater than 6, 7, 8, and 9 mmol/l,
gression analyses. The tests are two tailed through- respectively. The corresponding values for subjects with
out. A correlation matrix of the variables included in diastolic blood pressure less than 70 mm Hg were
the study is provided in the "Appendix." 33.3%, 13.2%, 3.2%, and 0.5%. Cholesterol levels less
than 5 mmol/1 were found in 14% of those with
Results diastolic blood pressure greater than 99 mmHg in
Total Cholesterol Level contrast to 32.1% of those with diastolic blood pressure
The age-adjusted Pearson correlation coefficient less than 70 mm Hg.
of serum total cholesterol level with diastolic blood
pressure was r=0.14 in men and r=0.10 in women; HDL Cholesterol and Triglyceride Levels
the coefficients for systolic blood pressure were
slightly lower. Table 1 shows the progressive increase In men, the age-adjusted levels of HDL cholesterol
in the age-adjusted levels of serum total cholesterol showed a slight increase with systolic but not with
with increasing levels of diastolic or systolic blood diastolic blood pressure, whereas in women, there
pressure in men and women. Controlling for body was no significant linear trend (Table 1). Controlling
mass index modifies the relation, but the association for body mass index strengthens these associations.
remains highly significant. Large contrasts were found when participants were
The age-adjusted linear regression slopes for sys- stratified by alcohol consumption: Low consumers
tolic and diastolic blood pressures on total choles- tended to have an inverse relation between HDL
terol level were significant for men and women in all cholesterol and blood pressure, whereas a positive
sex- and age-specific strata (Table 2). Whereas the relation was found in high consumers (Table 3). The
slopes for women increased with age, the slopes for associations remained nonsignificant in the lower
men decreased. The slopes were steeper for men consumption category after statistical adjustments
20-29 years old than for women in the same age for body mass index, whereas the positive relation
group (p<0.01). The linear regression slopes under- was strengthened in the other categories (not
estimate the age and sex contrasts, however, because shown).
there is a nonlinear relation between blood pressure The positive association of serum triglyceride levels
and cholesterol in men 20-39 years old and in with blood pressure was significant also after linear
women 40-49 years old (Figure 1). In these groups, adjustment for body mass index (Table 1), but the
the increase in cholesterol level at diastolic blood increase with blood pressure is greater in subjects with
pressures greater than 99 mm Hg is particularly
great. In men 20-29 years old, the cholesterol level body mass index higher than the median value than in
increase between extreme diastolic blood pressure leaner subjects. This contrast is particularly strong for
categories was 1.03 mmol/l (20.2%). Postmenopausal diastolic blood pressure (Table 3 and Figure 3).
women 40-49 years old had steeper slopes than did
menstruating women of similar age (not significant) Multivaniable Analysis
(Table 2). Postmenopausal women had significantly Serum total cholesterol, HDL cholesterol, and
steeper slopes than did women 20-39 years old triglyceride levels each have an independent posi-
(p=0.0002 for systolic and p=0.001 for diastolic tive relation with diastolic and with systolic blood
blood pressures).
The blood pressure-cholesterol association was pressures (data not shown) when accounting simul-
significant within strata divided according to body taneously for the confounding effects of age, body
mass index, but the slopes tended to be steeper for mass index, and lifestyle variables (Table 4). HDL
those with body mass index above the sex-specific cholesterol was not significantly associated with
median value than in leaner subjects. Nonsmokers diastolic blood pressure when body mass index was
had somewhat steeper slopes than did smokers (Ta- removed as a covariate from the multiple linear
ble 2). For both sexes, the total cholesterol-blood regression models.

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


1308 Circulation Vol 83, No 4 April 1991
TABLE 1. Mean Levels of Serum Total Cholesterol, High Density Lipoprotein Cholesterol, and Triglycerides in 8,081 Men and 7,663 Women
According to Blood Pressure: The Troms0 Study
Total cholesterol (mmol/l), HDL cholesterol (mmol/1), Triglycerides (mmol!l),
adjusted adjusted adjusted
For age For age For For age
Blood pressure (mm Hg) n For age and BMI For age and BMI age and BMI
Men
Systolic
<110 224 5.78 5.87 1.43 1.39 1.38 1.47
110-124 2,824 5.88 5.92 1.43 1.42 1.52 1.56
125-139 3,143 5.97 5.96 1.45 1.45 1.65 1.64
140-154 1,515 6.14 6.10 1.47 1.48 1.78 1.74
.155 375 6.30 6.20 1.54 1.57 1.87 1.76
p values
Equality 0.000 0.000 0.000 0.000 0.000 0.000
Linear trend 0.000 0.000 0.000 0.000 0.000 0.000
Diastolic
<70 705 5.73 5.80 1.45 1.42 1.53 1.61
70-79 2,291 5.85 5.90 1.45 1.43 1.53 1.58
80-89 3,118 5.98 5.98 1.45 1.45 1.64 1.62
90-99 1,537 6.16 6.10 1.46 1.48 1.74 1.67
.100 430 6.42 6.29 1.47 1.52 2.02 1.86
p values
Equality 0.000 0.000 0.84 0.000 0.000 0.000
Linear trend 0.000 0.000 0.38 0.000 0.000 0.000
Total or mean-+SD 8,081 5.98+1.15 1.450.45 1.630.94
Women
Systolic
<110 1,139 5.54 5.58 1.74 1.73 1.04 1.07
110-124 4,197 5.60 5.61 1.73 1.73 1.05 1.06
125-139 1,674 5.73 5.71 1.74 1.75 1.11 1.06
140-154 511 5.89 5.82 1.76 1.79 1.21 1.17
.155 142 6.08 5.97 1.74 1.78 1.31 1.24
p values
Equality 0.000 0.000 0.61 0.005 0.000 0.000
Linear trend 0.000 0.000 0.56 0.001 0.000 0.000
Diastolic
<70 1,218 5.53 5.56 1.72 1.70 1.03 1.05
70-79 3,001 5.57 5.59 1.74 1.73 1.05 1.06
80-89 2,533 5.62 5.71 1.76 1.76 1.10 1.09
90-99 766 5.83 5.76 1.73 1.76 1.18 1.13
>100 145 6.17 6.08 1.78 1.82 1.25 1.19
p values
Equality 0.000 0.000 0.049 0.000 0.000 0.002
Linear trend 0.000 0.000 0.071 0.000 0.000 0.000
Total or mean+SD 7,663 5.65-+1.06 1.740.40 1.080.57
Total cholesterol, HDL cholesterol, and triglyceride levels were adjusted for age or age and BMI analysis of covariance. Triglycerides were
also adjusted for time since last meal. Equality by analysis of covariance; linear trend by multiple regression analysis.
HDL, high density lipoprotein; BMI, body mass index.

Cholesterol Fractions and Blood Pressure in a pooled analysis adjusting for sex and age are shown.
Fasting Subjects The values indicate that both VLDL and LDL cho-
Table 5 presents the association between diastolic lesterol levels are increased at high blood pressures,
blood pressure and total, VLDL, LDL, and HDL but the trends remained significant only for total
cholesterol levels in 323 fasting subjects. Because the cholesterol level (p=0.03) after statistical adjust-
trends were similar in men and women, the results of ments for body mass index.

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


B0naa and Thelle Association of Blood Pressure With Serum Lipids 1309

TABLE 2. Stratified Linear Regression of Blood Pressure on Men Women


Serum Total Cholesterol: The Troms0 Study 7.0 -

Dependent variable/ Blood pressure 0


stratum n Systolic Diastolic E 6.5 -
40-49
Men -5
CD
6.0 -
Age (yr) a
2
0
20-29 2,433 1.5+0.2 1.0-+-0.2 0
30-39
- 5.5 -
30-39 3,045 1.20.2 1.40.1 0 2-2
I-
40-49 1,785 1.20.3 0.9+0.2
5.0 -
50-54 818 1.10.5 0.90.3 .41

Body mass index (kg/m2) A A 5 A .0Y, 1 9 1


70 80 90 100 70 80 90 100
<24.0 4,040 0.90.2 0.60.1 Diastolic blood pressure (mmHg)
>24.0 4,041 1.10.2 1.2+0.1
Cigarette smoking
FIGURE 1. Plot of mean concentrations of serum total
cholesterol levels (mmol/l) by diastolic blood pressure in men
Yes 4,072 1.10.2 1.00.1 20-29, 30-39, 40-49, and 50-54 years old and in women
No 4,009 1.30.2 1.50.1 20-29, 30-39, and 40-49 years old. Cells with less than 20
All men 8,081 1.20.1 1.20.1 observations were pooled with adjacent category. T bars are
Women SEM.
Age (yr)
20-29 2,892 0.60.2 0.80.2 but the results have been varied. Although no relation
30-39 3,029 0.90.2 0.90.2 was found in some studies,30,31 findings were inconsis-
40-49 1,742 1.90.3 1.10.2 tent across population strata in other studies.10-13 Sev-
Body mass index (kg/m2) eral factors may explain these discrepancies. First, the
<22.0 3,829 0.70.2 0.50.1 age and sex distribution of the population under study
.22.0 3,834 1.2+0.2 1.10.1 may influence the results. The present study consisted
Cigarette smoking of a predominantly young adult population, and the
Yes 3,666 1.00.2 0.70.1 results indicate that the strength of the association is
No 3,997 1.50.2 1.30.1 greater in younger than in older men. Next, biological
Menopausal status
interactions between risk factors may be more easily
detected in populations with a high prevalence of
Postmenopausal 293 2.90.7 1.60.4 elevated risk factors32; this may explain why the clus-
Premenopausal 1,263 1.60.4 0.90.2 tering of cholesterol level with blood pressure seems to
All women 7,663 1.20.1 0.90.1 be stronger in a high-risk population, such as in Eastern
Plus and minus values are age-adjusted regression coefficientsSE Finland,8 than in a low-risk population, such as in
for blood pressure (mm Hg) regressed on serum total cholesterol Southern California.3 Last, many studies relied on
level (mmol/l). Body mass index is grouped by the sex-specific median linear adjustments, and the results were often reported
value. Menopausal status includes women 40-49 years old not using as a partial correlation or regression coefficient: they
oral contraceptives (n=32) and not uncertain about their meno-
pausal status (n=154).

Discussion 30
The present study consisted of a large population a
of men and women with a broad age range, which U
-R
allowed a detailed analysis of the association be-
tween total cholesterol level and blood pressure. The 0
' 20

consistent positive relation between blood pressure c

and cholesterol level within population strata sug- a0


gests that there is a biological interrelation between 10
the two major coronary heart disease risk factors.
The results show that differences in total cholesterol
levels are not confined to comparisons of "hyperten-
sive" versus "normotensive" subjects but that there
are graded and continuous interrelations throughout 4 5 6 7 8 9
the usual range of blood pressure in this population; Total cholesterol (mmol/L)
the higher prevalence of extreme lipid abnormalities FIGURE 2. Frequency distribution of serum total cholesterol
in hypertension is secondary to a shift of the whole level in 1,923 men and women with diastolic blood pressure
distribution curve to the right. less than 70 mm Hg (solid line) and in 575 men and women
Previous research has also shown a positive associa- with diastolic pressure greater than 99 mm Hg (dashed line).
tion of blood pressure with total cholesterol level,3,7-13 Age adjusted.

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


1310 Circulation Vol 83, No 4 April 1991

TABLE 3. Stratified Linear Regression of Blood Pressure on Serum High Density Lipoprotein Cholesterol and on
Serum Triglycerides: The Troms0 Study
Dependent variable/ Blood pressure
stratum n Systolic Diastolic
Regression coefficient+SE on serum HDL cholesterol
Men
Inebriated by alcohol
Never or not last year 1,484 0.10.8 -1.20.6
A few times per year 3,459 0.90.5 -0.3+0.3
Once or more per month 2,045 2.50.6 0.7+0.5
Women
Inebriated by alcohol
Never or not last year 3,307 -0.90.6 -0.20.4
A few times per year 2,779 0.70.6 0.70.4
Once or more per month 577 3.81.2 2.60.9
Regression coefficientSE on serum triglycerides
Men
Body mass index
<24.0 4,040 1.30.3 0.00.2
.24.0 4,041 1.50.2 1.20.1
Women
Body mass index
<22.0 3,829 1.10.4 0.10.3
.22.0 3,834 2.20.3 1.7+0.2
Plus and minus values are age-adjusted regression coefficientsSE for blood pressure (mm Hg) regressed on serum
HDL cholesterol (mmol/l) or serum triglycerides (mmol/l). Information about alcohol consumption was unavailable for
1,093 men and 1,000 women.
Body mass index values are grouped by the sex-specific median value.

may underestimate nonlinear associations, however, tions. One reason for this is that most studies,
such as the relation between diastolic blood pressure including this one, rely on measurements of the
and total cholesterol level among younger men in the variables on a single occasion. Such measurements
present study (Table 2 and Figure 1). are subject to random fluctuations due to the mea-
Correlation coefficients less than 0.20 from cross- surement process and to temporary deviations from
sectional studies may not seem impressive, but they an individual's usual value.33 Because of the diluting
can, nevertheless, indicate important biological rela- effects of such fluctuations, estimates based on single
measurements will underestimate the real associa-
BMI tion between the level of usual blood cholesterol and
2.2 -
24+ the level of usual blood pressure, implying that the
actual clustering of the risk factors may be stronger
Men than observed. In the Seven Countries study, the
8
_i 1.8-
correlation of serum cholesterol level with blood
pressure was r=0.13 and that of body mass index with
24
I2
0
A
blood pressure was r=0.20.7 Compared with blood
> 1.4 -
cholesterol level, the random fluctuation of body
22+ mass index is negligible, implying that the real asso-
- Women ciation of cholesterol level with blood pressure may
.r.22
not be much less than that of body mass index.
1.0 -
The present study suggests that clustering of
. _1 1 cholesterol level with blood pressure is more pro-
70 80 90 100
nounced in men 20-39 years old than in women of
Diastolic blood pressure (mmHg) similar age and that clustering decreases with age in
FIGURE 3. Plot of mean concentrations of serum triglyceride men and increases with age in women. Similar
levels (mmol/l) by diastolic blood pressure in men and women results have been reported for men,8,9,11 and a
greater and less than the sex-specific median value for body recent Italian study found similar contrasts in age
mass index (BMI). Adjusted for age and time since last meal. trends between the sexes.13 Selection bias may
T bars are SEM. partially explain such age and sex differences if the

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


B0naa and Thelle Association of Blood Pressure With Serum Lipids 1311

TABLE 4. z Values for Variables Included in Multiple Regression disease is higher in men than in women. This kind
Analyses of Diastolic Blood Pressure: The Troms0 Study of bias is not likely to be large, however, given that
Independent variable z the incidence of coronary heart disease is still low in
Men (n=6,979) the older age groups of the present study popula-
Total cholesterol 7.4 tion and because the nonresponse was rather small
HDL cholesterol ... 2.7 ... in these age groups.
Triglycerides ... ... 3.9 A final possibility is that these contrasts are the
Age 17.7 21.2 21.8 results of physiological mechanisms that operate dif-
Log body mass index 17.3 19.4 17.7 ferently depending on age and sex. One such factor
may be sympathetic nervous overactivity. Heightened
Smoke cigarettes -4.7 -3.7 -4.4
Physical activity -2.5 -3.1 -2.6 sympathetic drive is a characteristic of young people
Inebriated by alcohol 4.2 3.7 4.2
with hypertension,34,35 maybe particularly in young
Time since last meal ... ... 10.9
men.5 Augmented sympathetic nerve activity, associ-
ated with increases in blood pressure, may also
R2 0.15 0.15 0.16 produce unfavorable metabolic effects, such as in-
Women (n=6,658) creases in atherogenic blood lipid concentra-
Total cholesterol 7.1 ... ... tions.18,36-38 Enhanced sympathetic drive can, there-
HDL cholesterol ... 4.1 ... fore, be one underlying factor for the greater
Triglycerides ... ... 4.2 association of cholesterol level and blood pressure in
Age 14.8 17.0 17.9 younger than in older men. In a recent population-
Log body mass index 15.2 17.1 15.4 based study of young adults,5 borderline hyperten-
Smoke cigarettes -4.0 -2.3 -3.7 sion, associated with sympathetic hyperactivity34 and
Physical activity -0.5 -0.8 -0.3 lipid abnormalities,39 was three times more frequent
Inebriated by alcohol 0.9 0.1 0.5
in men than in women. Generally, among young
Time since last meal ... ... 8.2
adults, hypertension is more prevalent in men than in
women.10 However, we are not aware of studies that
R2 0.13 0.13 0.13 have specifically examined the hypothesis that young
z values > 1.96 correspond to p<0.05. men are more susceptible to the pressor and meta-
Information about physical activity or frequency of inebriation bolic effects of sympathetic overactivity than are
was unavailable in 1,102 men and 1,005 women. young women.
The nonfasting status of the great majority of
attendance rate varies and if nonattenders have risk participants precluded the calculation of VLDL
factor levels different from those of attenders. and LDL cholesterol levels in these subjects. The
However, nonattendance in the younger age groups subgroup analyses in fasting subjects suggest that
in this study was mostly due to work situations or the association of blood pressure with cholesterol is
other social commitments and was due to a much due to both VLDL and LDL cholesterol levels, but
lesser extent to health or risk factor status. Because these results should be interpretated with care
we found a steeper slope in men of this age, a because these subjects may not represent the total
selection bias is an unlikely explanation for the study population. In the total study group, the
difference between the two sexes in the younger age age-adjusted correlation of non-HDL cholesterol
group. In the older age groups, selective survival level with diastolic blood pressure was r=0.13 for
and low attendance by those already affected by the men and r=0.10 for women. The correlation of
disease could explain part of the difference between LDL cholesterol level with non-HDL cholesterol
the sexes because the prevalence of coronary heart was r=0.98 in the fasting group. These results suggest

TABLE 5. Adjusted Mean Levels of Total, Very Low, Low, and High Density Lipoprotein Cholesterol in 323 Fasting
Men and Women, According to Diastolic Blood Pressure: The Troms0 Study
Cholesterol (mmol/l)
n Total VLDL LDL HDL
Diastolic blood pressure (mm Hg)
<80 111 5.94+0.11 0.50+0.02 3.910.11 1.54+0.04
80-89 133 6.100.10 0.49+0.02 4.070.10 1.540.04
.90 79 6.560.13 0.60+0.03 4.380.13 1.580.05
p values
Equality 0.003 0.007 0.032 0.75
Linear trend 0.001 0.015 0.011 0.50
Values are meanSE, adjusted for age and gender.
VLDL, very low density lipoprotein; LDL, low density lipoprotein; HDL, high density lipoprotein.
Equality by analysis of covariance. Linear trend by multiple regression analysis.

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


1312 Circulation Vol 83, No 4 April 1991

Appendix
Pearson Correlation Coefficients Between the Variables Included in the Analyses of Blood Pressure and Serum Lipids: The Troms0 Study
Age SBP DBP TC HDL TG BMI Cig Act Alc
Men (n=8,081)
SBP 0.12
DBP 0.31 0.51
TC 0.44 0.14 0.25
HDL 0.08 0.06 0.03 0.09
TG 0.08 0.13 0.11 0.33 -0.04
BMI 0.25 0.20 0.30 0.29 -0.13 0.27
Cig 0.00 0.00 -0.04 0.09 -0.05 0.07 -0.08
Act -0.08 -0.01 -0.06 -0.11 0.03 -0.07 -0.06 -0.16
Alc -0.31 0.01 -0.06 -0.12 0.10 -0.03 -0.07 0.19 -0.03
Time since last meal 0.01 -0.03 0.12 0.04 0.03 -0.12 -0.00 0.02 -0.03 0.00
Women (n=7,663)
SBP 0.25
DBP 0.29 0.60
TC 0.41 0.19 0.22
HDL 0.18 0.05 0.07 0.19
TG 0.07 0.11 0.09 0.29 -0.19
BMI 0.29 0.27 0.27 0.26 -0.10 0.19
Cig -0.13 -0.06 -0.09 0.03 -0.16 0.09 -0.12
Act 0.02 -0.03 0.00 -0.02 0.06 -0.07 0.00 -0.10
Alc -0.35 -0.11 -0.11 -0.18 -0.01 -0.03 -0.12 0.26 -0.03
Time since last meal 0.05 0.01 0.11 0.05 0.03 -0.09 0.04 -0.02 -0.01 -0.01
Coefficients greater than 0.02 are significant atp<0.01, and coefficients greater than 0.03 are significant atp<0.001. Eleven men and five
women had missing values for physical activity, and 1,093 men and 1,000 women had missing values for frequency of inebriation. Smoking
was coded as no=0 and yes= 1, physical activity and inebriation in three categories, and time since last meal in four categories.
SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; HDL, high density lipoprotein; TG, triglyceride; BMI,
body mass index; Cig, cigarette smoking; Act, physical activity; Alc, frequency of inebriation.

that blood pressure is associated with the athero- normalities, it is also possible that functions of
genic blood lipid fractions in both the LDL and biological membranes influencing blood pressure
VLDL regions of the lipoprotein distribution.40 The may be altered, depending on plasma fatty acid
age-adjusted relation of HDL cholesterol level with composition or lipoprotein concentrations. We re-
blood pressure was weak, but a significant positive cently showed that w-3 polyunsaturated fatty acids,
association was found in the highest alcohol con- which are found at high levels in seafoods, may
sumption category. Alcohol has a pressor effect41 lower both blood pressure and triglyceride levels in
and may increase HDL cholesterol level by increas- hypertension.43 Other research suggests that the
ing the levels of lipoprotein lipase.42 Body mass lowering of LDL cholesterol level is associated with
index is positively related to blood pressure and lowered incidence of hypertension17 and that LDL
inversely related to HDL cholesterol level (see cholesterol may blunt endothelium-dependent vas-
"Appendix") and, thus, strongly modifies the asso- cular relaxation.1',16
ciations between HDL cholesterol level and blood In the present study, men with diastolic blood
pressure. pressure greater than 99 mm Hg had on average a
The increases in total cholesterol or triglyceride total cholesterol level 0.69 mmol/l higher than men
level with blood pressure were greater in over- with diastolic blood pressure less than 70 mm Hg; in
weight than in lean subjects. This suggests that body younger men, the difference was greater. Such a
mass in itself or factors associated with body mass cholesterol difference alone imparts a 30% incre-
are related to concomitant elevations of blood ment in 8-year risk of myocardial infarction in this
pressure and blood lipids. Peripheral insulin resis- population. The synergistic relation between risk
tance and hyperinsulinemia have been suggested factors implies, however, that the increment may be
as mechanisms linking hyperlipidemia with hyper- larger at high levels of blood pressure. These results
tension, and decreased insulin sensitivity may be provide population-based support for the hypothe-
found more frequently in obese than in nonobese ses that there are biological interrelations between
persons.5.19 blood pressure and atherogenic blood lipid frac-
Although sympathetic nervous activity and insu- tions and that pathophysiological factors underlying
lin resistance may link hypertension and lipid ab- these interrelations may influence the mechanisms

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


B0naa and Thelle Association of Blood Pressure With Serum Lipids 1313

whereby hypertension is associated with increased 18. Bravo EL: Metabolic factors and the sympathetic nervous
risk of coronary heart disease. system. Am J Hypertens 1989;2:339S-344S
19. Reaven GM, Hoffman BB: Hypertension as a disease of
References carbohydrate and lipoprotein metabolism. Am J Med 1989;
87(suppl 6A):2S-6S
1. Report of Inter-Society Commission for Heart Disease 20. Thelle DS, F0rde OH, Try K, Lehmann EH: The Troms0
Resources: Primary prevention of the atherosclerotic diseases. heart study: Methods and main results of the cross-sectional
Circulation 1970;42:A55-A95 study. Acta Med Scand 1976;200:107-118
2. Brand RJ: An Examination of the Association Between A-B 21. Bjartveit K, Foss OP, Gjervig T, Lund-Larsen PG: The
Behavior and Coronary Heart Disease Incidence: Proceedings of cardiovascular disease study in Norwegian counties: Back-
the Forum on Coronary Prone Behavior, St. Petersburg, Florida, ground and organization. Acta Med Scand Suppl 1979;634:1-70
June 1-3, 1977. Washington, DC; 1978. US Dept of Health, 22. Kristiansen T: Hypertonikere i Troms0 3 (Hypertension in the
Education, and Welfare publication No. NIH 78-1451 Third Troms0 Survey). Troms0, Norway, University of Troms0,
3. Criqui MH, Barrett-Connor E, Holdbrook MJ, Austin M, 1990
Turner JD: Clustering of cardiovascular disease risk factors. 23. F0nneb0 Knutsen S: The Troms0 survey: The family interven-
Prev Med 1980;9:525-533 tion study (thesis). Troms0, Norway, University of Troms0,
4. Williams RR, Hunt SC, Hopkins P, Stults BM, Wu LL, 1990
Hasstedt SJ, Barlow GK, Stephenson SH, Lalouel J-M, Kuida 24. Lochen M-L, Rasmussen K: The Troms0 study: Physical
H: Familial dyslipidemic hypertension: Evidence from 58 Utah fitness, self-reported physical activity, and their relationship to
families for a syndrome present in approximately 12% of other coronary risk factors. Submitted.
patients with essential hypertension. JAMA 1988;259: 25. Nilssen 0, F0rde OH, Brenn T: The Troms0 study: Distribu-
3579-3586 tion and population determinants of gamma-glutamyltrans-
5. Julius S, Jamerson K, Mejia A, Krause L, Schork N, Jones K: ferase. Am J Epidemiol 1990;132:318-326
The association of borderline hypertension with target organ 26. Nilssen 0, F0rde OH: The Tromso study: The positive pre-
changes and higher coronary risk: Tecumseh Blood Pressure dictive value of gamma-glutamyltransferase and an alcohol
study. JAMA 1990;264:354-358 questionnaire in the detection of early stage risk drinkers. J
6. MacMahon SW, Macdonald GJ, Blacket RB: Plasma lipopro- Intern Med 1991 (In press)
tein levels in treated and untreated hypertensive men and 27. Friedewald WT, Levy RI, Fredrickson DS: Estimation of the
women: The National Heart Foundation of Australia Risk concentration of low-density lipoprotein cholesterol in plasma,
Factor Prevalence Study. Arteriosclerosis 1985;5:391-396 without use of the preparative ultracentrifuge. Clin Chem
7. Keys A, Aravanis C, Blackburn H, Buzina R, Djordjevic BS, 1972;18:499-502
Dontas AS, Fidanza F, Karvonen MJ, Kimura N, Menotti A, 28. SPSS-X User's Guide, ed 3. Chicago, SPSS Inc., 1988
Mohacek I, Nedeljkovic S, Puddu V, Punsar S, Taylor HL, van 29. Armitage P, Berry G: Statistical Methods in Medical Research,
Buchem FSP: Seven Countries: A Multivariate Analysis of Death ed 2. Oxford, Blackwell Scientific Publications, 1987
and Coronary Heart Disease. Cambridge, Mass/London, Har- 30. Ostrander LD: Coronary risk factors in a community: Findings
vard University Press, 1980 in Tecumseh, Michigan. Circulation 1976;53:152-156
8. Salonen JT, Puska P, Kottke TE, Heinonen OP: Coronary risk 31. Natvig H, Borchgrevink CF, Dedichen J, Owren PA, Schi0tz
factor clustering patterns in eastern Finland. Int J Epidemiol EH, Westlund K: A controlled trial of the effect of linolenic
1981;10:203-210 acid on incidence of coronary heart disease: The Norwegian
9. Hjermann I, Helgeland A, Holme I, Lund-Larsen PG, Leren vegetable oil experiment of 1965-66. Scand J Clin Lab Invest
P: The association between blood pressure and serum choles- Suppl 1968;22:1-20
terol in healthy men: The Oslo study. J Epidemiol Community 32. Grundy SM, Wilhelmsen L, Rose G, Campbell RWF, Assman
Health 1978;32:117-123 G: Coronary heart disease in high-risk populations: Lessons
10. Castelli WP, Anderson K: A population at risk: Prevalence of from Finland. Eur Heart J 1990;11:462-471
high cholesterol levels in hypertensive patients in the Framing- 33. Thompson SG, Pocock SJ: The variability of serum cholesterol
ham study. Am J Med 1986;80:23-32 measurements: Implications for screening and monitoring. J
11. Stamler J, Rhomberg P, Schoenberger JA, Shekelle RB, Dyer Clin Epidemiol 1990;43:783-789
A, Shekelle S, Stamler R, Wannamaker J: Multivariate analy- 34. Julius S, Schork N, Schork A: Sympathetic hyperactivity in
sis of the relationship of seven variables to blood pressure: early stages of hypertension: The Ann Arbor data set. J
Findings of the Chicago Heart Association Detection Project Cardiovasc Pharmacol 1988;12(suppl 3):S121-S129
in Industry, 1967-1972. J Chronic Dis 1975;28:527-548 35. Mark AL: Cyclosporine, sympathetic activity, and hyperten-
12. Criqui MH, Cowan LD, Heiss G, Haskell WL, Laskarzewski sion. N Engl J Med 1990;323:748-750
PM, Chambless LE: Frequency and clustering of nonlipid 36. Sacks FM, Dzau VJ: Adrenergic effects on plasma lipoprotein
coronary risk factors in dyslipoproteinemia: The Lipid metabolism: Speculation on mechanisms of action. Am J Med
Research Clinics program prevalence study. Circulation 1986; 1986;80(suppl 2A):71-81
73(suppl I):I-40-I-50 37. Day JL, Metcalfe J, Simpson CN: Adrenergic mechanisms in
13. Laurenzi M, Mancini M, Menotti A, Stamler J, Stamler R, control of plasma lipid concentrations. Br Med J 1982;284:
Trevisan M, Zanchetti A: Multiple risk factors in hyperten- 1145-1148
sion: Results from the Gubbio study. J Hypertens 1990;8(suppl 38. Maziere C, Mazi&re JC, Mora L, Gardette J, Polonovski J:
l):S7-S12 Epinephrine decreases low density lipoprotein processing and
14. Panza JA, Quyyumi AA, Brush JE, Epstein SE: Abnormal lipid synthesis in cultured human fibroblasts. Biochem Biophys
endothelium-dependent vascular relaxation in patients with Res Commun 1985;133:958-963
essential hypertension. N Engl J Med 1990;323:22-27 39. Julius S, Mejia A, Jones K, Krause L, Schork N, van de Ven C,
15. Ltischer TF: Functional abnormalities of the vascular endo- Johnson E, Petrin J, Sekkarie MA, Kjeldsen SE, Schmouder
thelium in hypertension and atherosclerosis. Scand J Clin Lab R, Gupta R, Ferraro J, Nazzaro P, Weissfeld J: "White coat"
Invest Suppl 1990;50:28-32 versus "sustained" borderline hypertension in Tecumseh,
16. Henry PD: Hyperlipidemic arterial dysfunction. Circulation Michigan. Hypertension 1990;16:617-623
1990;81:697-699 40. Frick MH, Elo 0, Haapa K, Heinonen OP, Heinsalmi P, Helo
17. Ekelund L-G, MacMahon RP, Whaley FS, Corder EH, P, Huttunen JK, Kaitaniemi P, Koskinen P, Manninen V,
Rubenstein CL: Does lowering of cholesterol with cholestyr- Maenpia H, Milkdnen M, Manttiri M, Norola S, Pasternack
amine decrease the incidence of hypertension in hypercholes- A, Pikkarainen J, Romo M, Sjoblom T, Nikkila EA: Helsinki
terolemic men? The LRC coronary primary prevention trial Heart Study: Primary-prevention trial with gemfibrozil in
(abstract). CVD Epidemiol Newslett 1988;43:13 middle-aged men with dyslipidemia: Safety of treatment,

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


1314 Circulation Vol 83, No 4 April 1991

changes in risk factors, and incidence of coronary heart 43. B0naa KH, Bjerve KS, Straume B, Gram IT, Thelle D: Effect
disease. N Engl J Med 1987;317:1237-1245 of eicosapentaenoic and docosahexaenoic acids on blood
41. Puddey IP, Beilin LJ, Vandongen R, Rouse IL, Rogers P: pressure in hypertension: A population-based intervention
Evidence for a direct effect of alcohol consumption on blood trial from the Troms0 study. N Engl J Med 1990;322:795-801
pressure in normotensive men: A randomized controlled trial.
Hypertension 1985;7:707-713
42. Eckel RH: Lipoprotein lipase: A multifunctional enzyme
relevant to common metabolic diseases. N Engl J Med 1989; KEY WORDS * risk factors * triglycerides * epidemiology
320:1060-1068 cholesterol * hypertension * coronary heart disease

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015


Association between blood pressure and serum lipids in a population. The Troms Study.
K H Bnaa and D S Thelle

Circulation. 1991;83:1305-1314
doi: 10.1161/01.CIR.83.4.1305
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 1991 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/83/4/1305

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:


http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:


http://circ.ahajournals.org//subscriptions/

Downloaded from http://circ.ahajournals.org/ by guest on December 1, 2015

You might also like