Correspondence

3 Rakova N, Jüttner K, Dahlmann A, et al. worldwide. 3,4 The first (1990 and We created a multidimensional
Long-term space flight simulation reveals
infradian rhythmicity in human Na(+) balance.
2010) assessed intake on the basis view of the published data on how
Cell Metab 2013; 17: 125–31. of 24 h urinary sodium excretion much sodium individuals consume
4 Lerchl K, Rakova N, Dahlmann A, et al. (UNaV), measured in 142 surveys of globally and how this consumption
Agreement between 24-hour salt ingestion
and sodium excretion in a controlled 24 h UNaV and 103 dietary surveys relates to cardiovascular disease and
environment. Hypertension 2015; 66: 850–57. in 187 countries.3 These data were all-cause mortality, and longevity
5 Olde Engberink RHG, van den Hoek TC, representative of 74% of the global (figure).5,6 The worldwide range of
van Noordenne ND, van den Born BH,
Peters-Sengers H, Vogt L. Use of a single adult population. Mean sodium sodium intake is virtually identical
baseline versus multiyear 24-hour urine intake was 3·95 g/day with a range to the range associated with both
collection for estimation of long-term sodium
intake and associated cardiovascular and renal of 2·3–5·5 g/day. For Asia and eastern the lowest risk of cardio­ v ascular
risk. Circulation 2017; 136: 917–26. Europe, mean intake was 4·2 g/day, disease events, including deaths,
while western Europe, North America, and greatest longevity, suggesting
The US recommendation for sodium Australia, and New Zealand had mean an optimal intake of 3·5–4·5 g/day.5,6
intake is less than 2·3 g/day for healthy intakes of 3·4–3·8 g/day. This concordance of data is consis­
individuals and less than 1·5 g/day The second analysis relied on peer- tent with a physiologically-regulated
for those at risk of cardiovascular reviewed publications of 190 data­ process7 that maintains sodium intake
disease.1 In 2013, the US National sets of 24 h UNaV surveys involving within an optimal range. Breaching
Academy of Medicine concluded that nearly 7000 people in 45 countries, either the lower or upper limits of this
insufficient evidence was available to primarily from western Europe, range appears to be associated with
support such a recommendation. The North America, Japan, Australia, increased all-cause mortality risk and
question remains of whether existing and New Zealand.4 Mean 24 h UNaV shortened life expectancy.
data provide compelling evidence over five decades was 3·7 g/day with In summary, worldwide sodium
that these guidelines are feasible and a range of 2·6–4·8 g/day and no intake of free-living individuals
will confer health benefit. The recent significant variation over time. These appears impervious to changes in
report by Mente and colleagues 2 two datasets, involving hundreds of dietary intake mandated by public
adds to the literature challenging the thousands of individuals, documented policies or guidelines, and an intake
long-standing presumption of health virtually identical global means and of 3–4·5 g/day seems to predict
benefits. ranges, seemingly unaffected by healthier longer lifespan than intakes
Two analyses of sodium intake demographics, culinary preferences, outside of this range.6 Thus, achieving
revealed remarkable consistency food supply, and time. the recommended sodium intake in
the general population appears neither
Comparison of risk of death vs life expectancy based on sodium intake feasible nor likely to be beneficial.
2·2 80 DAM reports personal fees from the Scientific
Advisor to ConAgra Foods, personal fees from the
Composite risk–death and CVD event hazard ratio (95% CI)

Scientific Advisor to Grocery Manufactures

USA
Association, grants from the Academy of Nutrition
70
China and Dietetics, non-financial support from the Board
Healthy life expectancy at birth (years)

1·8
of Directors of the American Society of Nurtrition,
and non-financial support from the Board of
60 Directors of the National Board of Physician
Nutrition Specialists, outside of the submitted work.
1·4 MHA and JCG declare no competing interests.

50 *David A McCarron, Joel C Geerling,

Michael H Alderman
1·0 dmccarron45@gmail.com
40 Private practice, Portland, OR 97209, USA (DAM);
ESC and NHS

Department of Neurology, Carver College of

Medicine, University of Iowa, Iowa City, IA, USA
WHO
AHA

0·6 30
(JCG); and Department of Epidemiology and
2000 4000 6000 Population Health, Albert Einstein College of
Sodium intake (mg/d) Medicine, Bronx, NY, USA (MHA)

Figure: Optimum intake of dietary sodium 1 US Department of Health and Human

The red line is the reported J-shaped curve for cardiovascular disease (CVD) and all-cause mortality related to Services and US Department of Agriculture.
2015–2020 dietary guidelines for Americans.
sodium intake,5 which is superimposed on the association of intake with longevity (black line).6 Overlaid on
Eighth Edition. December, 2015.
the mortality and longevity curves are the two reported ranges (95% CI) of worldwide sodium intake
https://health.gov/dietaryguidelines/2015/
(blue: 2·3–5·5 g/day³; green: 2·6–4·8 g/day⁴). The range of current worldwide sodium intake is in a so-called guidelines/chapter-1/key-recommendations/
sweet spot that aligns with the lowest all-cause mortality and greatest longevity. AHA=American Heart (accessed Feb 19, 2019).
Association. ESC=European Society of Cardiology. NHS=National Health Service.

1294 www.thelancet.com Vol 393 March 30, 2019


2 Mente A, O’Donnell M, Rangarajan S, et al.
Urinary sodium excretion, blood pressure,
cardiovascular disease, and mortality:
a community-level prospective epidemiological
cohort study. Lancet 2018; 392: 496–506.
3 Powles J, Fahimi S, Micha R, et al. Global,
regional and national sodium intakes in 1990
and 2010: a systematic analysis of 24 h urinary
sodium excretion and dietary surveys
worldwide. BMJ Open 2013; 3: e003733.
4 McCarron DA, Kazaks AG, Geerling JC, Stern JS,
Graudal NA. Normal range of human dietary
sodium intake: a perspective based on 24-hour
urinary sodium excretion worldwide.
Am J Hypertens 2013; 26: 1218–23.
5 Mente A, O’Donnell M, Rangarajan S, et al.
Associations of urinary sodium excretion with
cardiovascular events in individuals with and
without hypertension: a pooled analysis of data
from four studies. Lancet 2016; 388: 465–75.
6 Messerli FH, Hofstetter L, Bangalore S. Salt and
heart disease: a second round of “bad science”?
Lancet 2018; 392: 456–58.
7 Lowell, BB. New neuroscience of homeostasis
and drives for food, water, and salt. N Eng J Med
2019; 380: 459–71.
Franz Messerli and colleagues1 discuss
the Article by Andrew Mente and
colleagues2 about salt, blood pressure,
and cardiovascular outcomes, asking
the rhetorical question of whether
it is “bad science”. We posit that the
50-year-old controversy on salt and
cardiovascular morbidity (uniform
benefit of salt reduction versus harmful
effects for some at very low salt
intake—ie, the J curve association) is an
issue of no accurate data on salt intake
instead.
Rakova and colleagues3 have shown
that under a condition of completely
controlled salt intake for long periods,
salt excretion exhibits a circaseptan
rhythm with large variability, implying
that under these ideal conditions
(irreproducible in normal life), a week
of urine collections would be required
to estimate intake. Compare this ideal
situation with the daily variation of
salt intake over years, estimated by
one random urine sample in most
studies, and it becomes obvious that
one of the three datasets needed to
provide a scientific answer (accurate
salt intake, accurate outcomes, and
absence of confounders) is simply
unavailable in all studies to date.
Regarding confounders, the effects
of salt on intermediate cardiovascular
risk factors (eg, blood pressure, insulin
www.thelancet.com Vol 393 March 30, 2019 1295
resistance, and oxidative stress) are
influenced in a major way by the
salt sensitivity of blood pressure
phenotype, a fact only taken into
consideration (despite no accurate
data on salt intake either) in the
Genetic Epidemiology Network of Salt
Sensitivity (also known as GenSalt).
The study proposed by Jones and
colleagues 4 in the federal prison
population has the potential to
control diet and measure excretion
for prolonged periods. We would
like to suggest that analogous to
mendelian randomisation, such a
study should include stratification of
subjects by obtaining data on putative
predictive markers of salt sensitivity,5
given that actual phenotyping would
probably be costly and unfeasible.
We declare no competing interests.
*Fernando Elijovich, Cheryl L Laffer
fernando.elijovich@vanderbilt.edu
Division of Clinical Pharmacology, Department of
Medicine, Vanderbilt University School of Medicine,
Nashville, TN 37232, USA
1 Messerli FH, Hofstetter L, Bangalore S. Salt and
heart disease: a second round of “bad science”?
Lancet 2018; 392: 456–58.
2 Mente A, O’Donnell M, Rangarajan S, et al.
Urinary sodium excretion, blood pressure,
cardiovascular disease, and mortality:
a community-level prospective epidemiological
cohort study. Lancet 2018; 392: 496–506.
3 Rakova N, Jüttner K, Dahlmann A, et al.
Long-term space flight simulation reveals
infradian rhythmicity in human Na+ balance.
Cell Metab 2013; 17: 125–31.
4 Jones DW, Luft FC, Whelton PK, et al. Can we
end the salt wars with a randomized clinical
trial in a controlled environment? Hypertension
2018; 72: 10–11.
5 Elijovich F, Weinberger MH, Anderson CAM,
et al. Salt sensitivity of blood pressure. a
scientific statement from the American Heart
Association. Hypertension 2016; 68: e7–46.
Authors’ reply
24 h urine collection is a reference
method, rather than the gold stand­
ard for measuring sodium intake.
A key test of whether different
measures of sodium intake are valid
is to compare each measure with an
independent biomarker, such as blood
pressure (BP)—an association upon
which current recommendations for
reducing sodium intake are based.
Large epidemiological studies with
Correspondence
24 h urine collections have shown a
weak1 or no2,3 association between
24 h sodium excretion and BP, but
several studies that use early-morning
fasting urine to estimate sodium
intake (with adequate methods—
namely the appropriate formula for
fasting urine and correct timing of
fasting urine collection immediately
after a proper 24 h urine collection)
show stronger associations with BP
than 24 h urine collections,4,5 similar
to the 2·42 mm Hg reduction in
systolic BP per 1 g lowering of sodium
consumption observed in randomised
trials.6
A morning fasting urine collection
reflects overnight basal excretion, and if
used to estimate sodium intake, is not
only sufficiently robust for stratifying
groups of people according to intake
to assess associations with BP (and
more importantly with cardiovascular
disease and mortality), but might
even be a better indicator of long-term
sodium stored in tissue compartments
Philippe TURPIN/Getty Images
in the body (such as the vasculature
or bone) than a single 24 h collection
(which is subject to the additional
variability from short-term changes
in dietary intakes during the day).
Furthermore, first morning urine is
less subject to the limitations of a 24 h
urine (ie, high amounts of incomplete
collection—as much as 50%—or biases
from excluding participants who
are unwilling to provide a 24 h urine
sample), which can distort the results
of studies that use 24 h urine or limit
their generalisability. Additionally,
any method that estimates sodium
intake needs to be applicable to large
populations if the association between
sodium intake and clinical events is to This online publication has been
be ascertained reliably. Such large-scale corrected. The corrected version
first appeared at thelancet.com
applicability would require studies in on April 8, 2019
which several thousand events accrue,
which in turn would require studies of
some tens of thousands of participants,
in whom sodium intake and outcomes
are measured without biases. These
studies would be more feasible and less
biased with simpler methods, such as
collection of overnight fasting urine