State of the Art:
Salt and Health Outcomes
Francesco P Cappuccio MD MSc DSc FRCP FFPH FBHS FAHA
Professor of Cardiovascular Medicine & Epidemiology
Head, WHO Collaborating Centre
University of Warwick, Coventry, UK
Disclosures: Technical Advisor to the World Health Organization, the Pan American Health Organization,
Member of C.A.S.H., W.A.S.H., UK Health Forum and Trustee of the Student Heart Health Trust
Vice-President, British Hypertension Society – all unpaid.
Salt – Blood Pressure – CVD Model
SALT
BP
CVD
Salt – Blood Pressure – CVD Model
• Hunter-gatherers societies have low salt consumption
(<1g per day)
• Salt entered the food chain as preservative for food
• Migrant populations to urbanized areas increase their salt
intake by several folds
• Salt intake is now high (population averages 7.5 to 12.5 g
per day, with higher extremes)
• In Westernized societies, salt in the food chain is not
needed to preserve food any longer (refrigeration)
SALT
BP
CVD
• 75% or more of salt in food derives from salt added in the
manufacturing process
• Individuals only control a small part of the salt they eat
• High salt in processed and manufactured food leads to
profit to industry through a ‘cycle of profit’ (e.g.
palatability; hygroscopic properties; taste addiction; thirst)
High salt consumption around the world
10.0 g/day
12.5 g/day
7.5 g/day
WHO target
5 g/day max.
Powles J et al. BMJ Open 2013;3:e003733
Average Global Consumption 10.2 g/day
Salt – Blood Pressure – CVD Model
SALT
BP
•
High salt intake is associated with
high BP and with the rise in BP
with age
•
The effect is seen at all ages, in
both genders, in all ethnic groups,
in rich and poor.
•
Migrant populations to urbanized
areas increase their salt intake by
several folds and their BP increases
in parallel.
•
Large primates fed high salt diets
show a rise in BP and when fed a
low salt diet their BP falls.
•
In humans, children and adults, a
reduction in salt intake in RCTs
causes a dose-dependent fall in BP.
•
The fall is greater as we get older,
as a function of baseline BP and in
people of black African ancestry.
CVD
Effects of reduced sodium intake on
systolic blood pressure
Mozaffarian D et al. NEJM 2014:371:624-34
A reduction in dietary salt intake reduces blood
pressure in adults …
Cappuccio FP & Capewell S. Functional Food Rev 2015; 7: 41-61
… and children
Aburto NJ et al. BMJ 2013; 346: f1326
The lower the salt, the lower the blood pressure
Change in Systolic Blood Pressure
(mmHg)
4
2
0
-2
-4
Normotensives
-6
-8
Hypertensives
-10
-12
-30
-50
-70
-90
-110
Change in Urinary Sodium
(mmol/24h)
He
9 FJ, MacGregor GA. J Hum Hypertens. 2002;16:761-70
-130
Salt – Blood Pressure – CVD Model
•
Large primates fed high salt diets show
a rise in BP and an increase in stroke
deaths. On a low salt diet the reverse is
seen.
•
High BP is a predictor of CVD in
populations (observational).
•
Countries that have reduced salt
consumption have seen a reduction in
CV occurrence over many years.
•
A reduction in BP with drugs reduces
CVD within 5 years or less (elderly)
(RCTs)
•
High salt intake is associated with high
mortality from stroke, CHD and CVD
(observational).
•
A reduction in salt intake reduces the
incidence of vascular events (small and
short RCTs).
•
Long-term surveillance confirms
benefits of reduced salt on CV
outcomes (observational).
SALT
BP
CVD
Stroke and Ischemic Heart Disease (CHD)
as a function of Systolic BP
Stroke
256
80–89 y
128
70–79 y
64
Mortality*
CHD
Age at risk
Age at risk
256
80–89 y
128
70–79 y
64
60–69 y
32
50–59 y
16
40–49 y
60–69 y
32
50–59 y
16
8
8
4
4
2
2
1
1
0
120
140
160
180
Usual SBP (mmHg)
Lewington et al. Lancet 2002;360:1903–13
0 120
140
160
180
Usual SBP (mmHg)
A small shift in mean BP reduces number of
hypertensives and burden of attributable disease
Reducing mean BP
Reducing %
hypertensives
Reducing events
attributable to BP
In 2010, global sodium consumption estimated at 3.95g per day (9.875g salt per day)
Globally, 1.65m annual CV deaths attributed to sodium intake >2g per day (>5g salt per day)
These deaths accounted for nearly 10% of CV deaths
85% of these deaths occurred in LMICs and 40% were premature (<70 years)
Mozaffarian D et al. NEJM 2014:371:624-34
Risk of stroke associated with salt intake in population
Strazzullo
P et al. BMJ 2009; 339: b4567
14
He FJ, MacGregor GA. Lancet 2011:378:380-2
New Scientist, 3 December 2011
Domain 1 - Errors with the greatest potential to alter the direction of association
Systematic error in sodium assessment

Lower risk: 24h urine collections not part of routine clinical practice, no quality assurance, not excluding incomplete collections.

Higher risk: other 24h urine collections, all dietary assessments, spot and overnight urine collections.
Reverse causality

Lower risk: participants recruited from general population and pre-existing CVD excluded

Intermediate risk: sick populations not excluded or included despite stated otherwise; presence of CVD risk factors; specific sick
populations

Higher risk: specific sick populations (eg: heart failure, kidney disease, diabetes); removal of sick participants from analysis changes
direction of association
Domain 2 - Errors with some potential to alter the direction of association
Potential for residual confounding

Incomplete adjustment: not including 2 or more of age, sex, race, SES, cholesterol, BMI or weight, smoking, diabetes; if diet-based,
total calories; in urine-based weight, BMI or creatinine excretion

Imbalance across sodium intake levels: age difference across sodium groups >5 years; sex or race distribution across sodium groups
>20%

Inadequate follow-up: low level of follow-up (<80%) or of uncertain quality for outcome assessment
Domain 3 - Errors with the potential to lead to a false null result
Random error in sodium assessment

Lower risk: more than four 24h urine assessments on average; FFQs

Intermediate risk: between 22-4 24h urine collections, or corrections for regression dilution bias; dietary reports

Higher risk: urine collection <24h or single 24h urine collection; single dietary recall or 1-dat food record
Insufficient power

Less than 80% power to detect a 10% reduction in relative risk for every standard deviation in sodium intake
Domain 4 - Studies using same data with divergent results

NHANES I studies: same age group, same follow-up – inverse vs positive association

NHANES III studies: different age groups, different follow-up – inverse vs positive association
Cobb LK et al. Circulation 2014; 129: 1173-86
SODIUM INTAKE AND CV EVENTS:
J-SHAPE RELATIONSHIP?
EPOGH Study: mortality rates and CV events by thirds of
24h urinary sodium excretion
CARDIOVASCULAR EVENTS
MORTALITY
All CV
Total
non-CV
CHD events
CV
Stroke
3
6
9
12 15
18
Population salt intake (g/day)
Stolarz-Skrzypek et al, JAMA 2011;305:1777–85
3
6
9
12 15
18
Population salt intake (g/day)
EPOGH Study: characteristics of male participants
Variables
Sodium intake tertile
Lowest
Highest
p
Urinary Sodium (mmol/24h)
120
290
<0.05
Urinary Volume (L/24h)
1.3
1.8
<0.05
Urinary Creatinine (mmol/24h)
12
16
<0.05
Urinary Potassium (mmol/24h)
62
85
<0.05
≤ Elementary school education
35%
20%
EXCLUSIONS: only if 24h urine volume < 300 ml
Stolarz-Skrzypek et al, JAMA 2011; 305: 1777–85
He FJ et al. Kidney Int 2011; 80: 696-8
All CVD events by “estimated” daily salt intake in
ONTARGET/TRANSCEND PATIENTS
Model was adjusted for age, sex, race/ethnicity (white vs nonwhite); prior history of stroke or myocardial infarction; creatinine, body mass
index; comorbid vascular risk factors (hypertension, diabetes mellitus, atrial fibrillation, smoking, low- and high-density lipoprotein); treatment
allocation (ramipril, telmisartan, neither, or both); treatment with statins, β-blockers, diuretic therapy, calcium antagonist, and antithrombotic
therapy; fruit and vegetable consumption, level of exercise; baseline blood pressure and change in systolic blood pressure from baseline to last
follow-up; and urinary potassium.
O’Donnell M et al. JAMA 2011; 306: 2229-38
Cause-Specific CVD events by “estimated” daily salt intake in
ONTARGET/TRANSCEND PATIENTS
Model was adjusted for age, sex, race/ethnicity (white vs nonwhite); prior history of stroke or myocardial infarction; creatinine, body mass
index; comorbid vascular risk factors (hypertension, diabetes mellitus, atrial fibrillation, smoking, low- and high-density lipoprotein); treatment
allocation (ramipril, telmisartan, neither, or both); treatment with statins, β-blockers, diuretic therapy, calcium antagonist, and antithrombotic
therapy; fruit and vegetable consumption, level of exercise; baseline blood pressure and change in systolic blood pressure from baseline to last
follow-up; and urinary potassium.
O’Donnell M et al. JAMA 2011; 306: 2229-38
All CVD events by “estimated” daily salt intake in
ONTARGET/TRANSCEND PATIENTS
• Old and sick patients in secondary prevention
70% hypertensives
37% diabetics
48% hypertensives with previous MI
21% previous stroke / TIA
3% AF (9% at follow-up)
• Multiple medications (overall 29% BUT ….41% in
the lowest sodium category)
• Inaccurate estimate of sodium intake, based on
‘spot urine samples’ using Kawasaki equation
O’Donnell M et al. JAMA 2011; 306: 2229-38
5g salt
Finnish Cohort (1998-2002)
Multicentre
2,807 adults with T1D (no ESRD)
Median f-up 10 years
217 deaths (7.7%)
Baseline 24h urinary sodium
Thomas FC et al. Diabetes Care 2011; 34: 861-6
424 microalbuminuric patients
122 ESRD events
Very few events stratified in 5 groups
Not graded relationship
‘At risk’ group?
Thomas FC et al. Diabetes Care 2011; 34: 861-6
Sodium estimated by single fasting morning urine – unreliable and biased
Sodium Study (n~100K) not comparable to Overall Study (n~160K) – selection bias
Fewer from India (5 v 18%) and more from China (42% v 30%)
More participants with ill-health (hypertension, BP medication, CHD, CVD)
Lower sodium excretion group (see Table 1, p. 603):
<3g per day (<7.5g of salt per day) – unable to discriminate on low sodium intake
Small sample size - wide confidence intervals
Mente A et al. NEJM 2014:371:601-11
Outcome
Outcome, sodium dose and statistical power
7
6
5
4
3
2
1
0
Overall P=.03
They really
must think
everyone is
real stupid
P=.01
P=NS
1000 mg 2000 mg 3000 mg 4000 mg 5000 mg 6000 mg 7000 mg
Dietary Sodium
Courtesy of Norm Campbell
<7.5 g salt
10 g salt
O’Donnell M et al. et al. NEJM 2014:371:612-23
Sodium estimated by single fasting morning urine – unreliable and biased
Sodium Study (n~100K) not comparable to Overall Study (n~160K) – selection bias
Fewer from India (5 v 18%) and more from China (42% v 30%)
More participants with ill-health (hypertension, BP medication, CHD, CVD)
Lower sodium excretion group (see Table 1, p.616):
<3g per day (<7.5g of salt per day) – unable to discriminate on low sodium intake
compared to higher sodium: 3y older; fewer men, Asians, smokers; more Africans and
non-Asians, urban; lower blood pressure; higher LDL-cholesterol, history of CVD and
diabetes, F&V intake, medication use – biased towards lower sodium excretion due to
age and gender, and presence of ill-health (reverse causality)
O’Donnell M et al. et al. NEJM 2014:371:612-23
Cappuccio FP et al. Eur Heart J 2013; May 8: on-line
‘Graded’ and ‘linear’ association between salt
intake and cardiovascular disease
5g salt
Cook NR et al. Circulation 2014; 129: 981-9
REVERSE ASSOCIATIONS OR J CURVE
Studies in people with disease
Reverse/J
Dong 2010
Y
Ekinci 2011
Y
O’Donnell 2014
Y
Tikellis
Y
Studies2013*
in people without disease
Lennie 2011
Y
Reverse/J
Cohen 2006
Y
Cohen 2008
Y
Geleijinse 2007
Y
Larsson 2008
Y
Stolarz-Skrzypek 2011
Y
from IOM 2013 analysis
Linear relationship of CVD outcomes
Studies in people
with dietary sodium
that support a
Sodium mg/day
Fresh foods with meat 500-< 1000
Fresh food vegans <500
Kempner rice diet ~150
Hunter gather populations <1000
Yaminano ~ 100
Long term physiological studies~
100
linear association
(from IOM report)
Cook 2007
Cook 2009*
Costa 2012
Gardener 2012
Heerspink 2012
Jafar 2006
14 studies, 9 free of
methodology flaws
Kono 2011
Arcand 2011
McCausland 2012
Nagata 2004
Takachi 2010
Thomas 2011
1250 mg sodium/day
from IOM 2013 analysis
Umesawa 2008
Yang 2011
Salt – Blood Pressure – CVD Model
SALT
BP
HOW FAR SHOULD WE GO
WITH REDUCTION OF SALT INTAKE?
THE LOWER THE BETTER ?
CVD
Role of salt in hypertension and CVD
overestimated?
A RANDOMIZED CONTROLLED TRIAL
OF SALT INTAKE REDUCTION
AND CVD MORTALITY ?
IS IT FEASIBLE?
Randomized controlled trial of salt reduction and
CVD mortality
• Need to recruit >28,000 participants
• >2,500 cardiovascular events
• Duration of at least five years
• Design challenges:
– Multi-centre – multi-national – multi-cultural
– Maintain two groups at different sodium intakes in an unfavorable
environment (declining salt consumption)
– Appropriate methodology for salt assessment in individuals
• Ethical problems
• Budget unlikely
Strazzullo P. J Hypertens 2011; 29: 829-31
Changes in diastolic blood pressure,
salt intake and stroke deaths in Finland
5600 mg
3360 mg
DBP
Salt
Karppanen
H et al. Progress, Cardiovascular Disease. 2006;49:59-75.
37
Stroke
Changes in salt intake, blood
pressure, stroke and IHD
mortality in England from
2003 to 2011
-40%
-42%
Health Survey for England
aged ≥16 years
2003 N=9183
2006 N=8762
2008 N=8974
2011 N=4753
-3.0 mmHg
-1.4 mmHg
-1.4 g/d
He FJ et al. BMJ Open 2014; 4: e004549
“What
is truly meaningful in life is
inversely related to what is measurable”