Clinical Science (1979) 57.455s-4589
Salt and hypertension
LARS WILHELMSEN
Deparlment of Medicine, &Ira Hospital, G6teborg. Sweden
Introduction
The title of this panel discussion is ‘Salt and
hypertension’, but I would like to broaden the
scope of my contribution to comprise salt and
blood pressure. The reason is that the question of
salt is not only of importance for the management
of hypertension in the clinical context, but the
association between salt and blood pressure may
have major epidemiological implications for individuals with a wide range of blood pressures.
Most of those who may benefit from reduced salt
intake are not definitely hypertensive.
The following introductory lectures by Dr T.
Morgan and Dr F. 0. Simpson will deal with the
evidence for and against salt being of importance
for the determination of the blood pressure, and the
present paper will deal with some general epidemiological, statistical and methodological problems
which I think are of importance for the following
discussion.
Salt and the blood pressure
The association between salt and blood pressure
has been discussed for several decades, but a
review by Fries (1976) has broadened the interest
extensively.
Associations have been found between the mean
salt excretion for a certain population and the mean
blood pressure or the prevalence of hypertension.
However, differences in measurement of blood
pressure, selection bias in the study groups and the
presence of several confounding factors such as
diet, life style etc. make it dimcult to draw firm
conclusions from these findings.
Analyses of the association between salt excretion and blood pressure for single individuals
within populations have, with few exceptions, not
shown a correlation. However, the chances of
finding such correlations are poor for several
reasons.
Let us assume that the global relationship has
the form shown in Fig. 1, where the ellipse
symbolizes a hypothetical cluster of points in a
global sample. When the relationship within, for
example, a Swedish population sample is analysed
most points will lie within the narrow range
indicated, because the variation in salt intake within
that population is rather small. Thus it is difficult
to find a significant correlation because of the
small variation in the horizontal direction.
Each individual is also rather poorly characterized because usually salt excretion for only 2448 h is measured. However, the daily salt intake
varies considerably between days at least in
European, North American and Australian
populations. The mean of seven to nine 24 h
collections are often needed to characterize an
individual’s salt intake/excretion.
The blood pressure of the single individual is also
poorly defined by a single or a few measurements.
Thus there is reason to believe that both the values
on the horizontal and those on the vertical axis are
subject to bias, which tends to weaken any
correlations between individuals. When analysing
correlations between population means with great
variations in salt intake/excretion and blood pressure the above-mentioned bias is of less importance.
It is highly probable that not only the current
salt intake, but also the intake over several years
influences the blood pressure. It is virtually
impossible to obtain this history for salt intake over
a period of several years before the examination.
Finally, it is generally believed that the blood
pressure is strongly determined by genetic factors.
These factors tend to be of primary importance for
the individual blood pressure in a study of
individuals with fairly similar salt intake. When
455s
L. Wilhelmsen
456s
l 2 I
1000
C
800-
8
kx
600-
.0
5
1 0
15
20
25
30
Salt intake (g/day)
FIG. 1. Relationship between salt intake and blood
pressure in a hypothetical global sample. Effect of
restriction of one of the variables under study. From
Svardsudd (1978).
P
400-
200 -
1/ ,
-------.....
------..
~
I
115
,
,
135
, , , , , , , , ,
155
175
195
215
, I
235
Systolic blood pressure (mmHg)
population means for salt and blood pressure are
studied there will be a much wider distribution for
these variables, and the genetic factors will be
much less important when the mean pressures for
the populations are used. This does not exclude salt
intake as being of importance for the blood
pressure value in that particular population as a
whole.
The most reliable method of studying the effect
of salt is to study groups who have to change their
dietary salt intake, but keep other relevant habits
unchanged. However, in real life, particularly when
dealing with long-term assessment (over decades)
such changes are evidently very difficult to study
scientifically. Investigations performed so far indicate a definite decrease of blood pressure with
decreasing salt intake in the short term (Morgan,
Gillies, Morgan, Adam, Wilson & Carney, 1978).
Blood pressure and risk of disease
It is a well-known fact that the incidence of stroke,
coronary heart disease and the total mortality
increase with increasing blood pressure. The
relationships are not the same in all age groups,
and they are not the same for all the diagnoses in
question. Fig. 2, which is based upon the 5 year
follow-up of the Bergen Study (Holme & Waaler,
1976; Wilhelmsen, Berglund & Wedel, 1979),
shows that for all but the highest blood pressures,
coronary heart disease (CHD) is much more common as an end-point than is stroke among 50-59
year-old men. The risk increase for stroke comes at
higher blood pressures and is then much steeper
than for CHD. The difference between C H D and
stroke is less pronounced in the higher age groups.
FIG. 2. Relationship between 5 year mortality and
systolic blood pressure in the Bergen Study (Holme 8i
Waaler, 1976; Wilhelmsen, Berglund & Wedel, 1979).
Thus blood pressures far below those currently
used as cut-off points for antihypertensive drug
therapy are associated with an increased risk of
morbid events, at least in industrialized countries.
This has partly to do with the interplay between the
high blood pressure and the other major C H D risk
factors: high serum cholesterol and smoking.
Stroke, however, is less dependent on serum
cholesterol and smoking.
Fig. 3 shows the risk of CHD, non-fatal and
fatal events, in 855 men followed for 13 years in the
study of men born in 1913 in Goteborg
(Wilhelmsen et al., 1979). The risk associated with
any blood pressure is dependent on the smoking
habits. A non-smoker or an ex-smoker has only a
moderate risk increase with increasing blood
pressure, whereas the heavy smokers have a more
rapid risk increase with increasing pressure. A similar additional effect is seen for serum cholesterol.
In populations with low levels of smoking and
low serum concentrations of cholesterol, stroke
may be as prevalent as or more prevalent than
CHD, and the risk increase with increasing blood
pressure may come at higher pressures but be
steeper. This finding has, for example, been
reported by Ashcroft & Desai (1978) from
Jamaica.
‘Population-attributable
measures
risk’
and
preventive
A factor that tends to be overlooked is the
importance not only of the risk but also of the
Round Table 2: Salt intake and hypertension
457s
Probability (PI of MI
for a 50 -year-old
man during
13 years follow-up
t
FIG. 3. Probability (P)of non-fatalor fatal myocardial infarction (MI) or sudden coronary death for a 50
year-old man during 13 years’ follow-up in relation to systolic blood pressure (BP, mmHg) and smoking
habits.
number of persons in the population who are
exposed to the risk. When both these variables are
taken into account we talk about the ‘populationattributable risk’. Follow-up data from the previously mentioned study of men born in 1913
(Wilhelmsen et al., 1979) illustrates the importance
of this concept (K. Sviirdsudd & G. Tibblin,
unpublished work). Fig. 4 shows the increasing risk
for stroke and C H D in this population sample in
relation to blood pressure, the distribution of blood
pressure and the number of new events which have
occurred at various pressures during 13.5 years’
follow-up. Most of the morbid events did not occur
in the group with the highest risk, but in the groups
with moderate risk, which contain a greater proportion of the men under study. In other words,
most people who are going to suffer disease due to
an elevation of the blood pressure have moderate
elevations which would not usually lead to drug
therapy. Any measure which would prevent blood
pressure increasing even in the moderate range
might be valuable in these individuals. Several
studies have indicated that there are relatively few
such modifiable factors (Sviirdsudd, 1978). Reduction of salt intake and weight reduction are,
however, two such measures.
From population data it can easily be calculated
that a general reduction of the blood pressure in all
individuals, excepting those with the lowest pressures, would have a much greater impact on morbidity than a reduction to what is considered
normal in those with definite ‘hypertension’. Thus
in the population sample of men born in 19 13 a
reduction of all blood pressures by 10 mmHg
would theoretically prevent as many morbid events
as reducing all systolic pressures to 135 mmHg or
lower. According to several studies, a reduction by
10 mmHg could be achieved by relatively moderate
salt restriction in many countries.
It is conceivable that salt intake is of much
greater importance in certain individuals, namely
those who are more sensitive to salt load possibly
L. Wilhelmsen
458s
Prevalence
%
Risk
Morbid events
%
40
n
60
60
40
40
20
20
30
20
10
0
‘La.
0,
0
100
0
120
0
140
0
160
0
180
.
.............
.
200
0
Systolic BP
FIG.4. Blood pressure (BP) distribution, risk for coronary heart disease or stroke, and number of such
morbid events in relation t o blood pressure during 13.5 years’ follow-up of 855 men aged 50 years at
entry.
because of a genetically determined relative renal
deficiency in salt handling or some other mechanism. Then, one would wish to know which
individuals have this deficiency and restrict their
salt intake, whereas other people could continue
eating as today. But we know far too little at
present to be able to identify these individuals in the
healthy population. It might be argued that the
evidence for salt being important as a risk factor is
so strong that a reduced salt intake should be
recommended for most people. This view parallels
the discussion on general reduction of dietary fat,
and reduction of smoking, for which exactly the
same arguments on ‘population-attributable risk’
are valid. This epidemiological approach to the
problems points to intervention measures other than
those most of us as clinicians are accustomed to.
The changes that would be required to meet the
salt-reduction goals would necessitate substantial
changes in the diet and cooking habits of most
people. Recommendations for such changes have
to be based on solid scientific evidence, and the
public should be assured of substantial benefit from
their changes. Do we have such evidence today?
Perhaps the following debate will help to clarify our
thoughts.
References
ASHCROFT,
M.I. & DESAI,P. (1978) Blood pressure in a rural
Jamaican community. Lancet, i, 1167-1 170.
FRIES, E.D. (1976) Salt, volume and the prevention of
hypertension. Circulation, 53,589-595.
HOLME,1. & WAALER,
H.T.H. (1976) Five-year mortality in the
city of Bergen, Norway, according to age, sex and blood
pressure. Acta Medica Scandinavica, 200,229-239.
MORGAN,T., GILLIES,A., MORGAN,G., ADAM,W., WILSON,
M. & CARNEY,S. (1978) Hypertension treated by salt
restriction. Lancet, i, 227-230.
K. (1978) High blood pressure. A longitudinal
SVARDSUDD,
population study of men born in 1913, with special reference
to development and consequences for health, Ph.D. Thesis.
University of Goteborg, Goteborg, Sweden.
L., BERGLUND,
G. & WEDEL,H. (1979) Benefits
WILHELMSEN,
of blood pressure treatment in the general middle-aged male
population. In: Mild Hypertension-Natural
History and
Trearrnent, pp. 47-55. Ed. Gross, F. & Strasser, T. Pitman
Medical. London.
DISCUSSION
Roberfson: Thank you very much Dr Wilhelmsen.
We wish to have the general discussion later, but if
anyone feels that they would like to emphasize or
criticize some points of fact that Dr Wilhelmsen
made, they can raise these now.
Hornych: In your large epidemiological study in
which you examined the relationship between
sodium intake or sodium excretion and hypertension, did you also look at the relationship of plasma
sodium and plasma chloride with blood pressure?
Wilhelmsen: These variables were not correlated
with the blood pressure level.
Robertson: If there are no other questions to be
raised immediately, I shall now invite Dr Morgan
to present his work with Dr Myers and Dr Carney,
indicating that sodium intake is important.