Na and calorie intake and mortality

[Guest guest]

Hi All,

Sodium ion and calorie intake and mortality are examined and discussed in the

below.

Calories matter?

Alderman MH, Cohen H, Madhavan S.

Dietary sodium intake and mortality: the National Health and Nutrition

Examination

Survey (NHANES I).

Lancet. 1998 Mar 14;351(9105):781-5.

Comment in:

Sodium intake and mortality

Lancet. 1998 May 16;351(9114):1508-9

Karl Engelman

Alderman and co-workers1 report an inverse relation between reported

sodium

intake and mortality. Although I have neither the statistical nor

epidemiological

expertise of these investigators, I am certain that their analyses and erroneous

conclusions are based on faulty data.

They purport to show increased mortality, especially among those in the lowest

quartile of sodium intake based on estimated (by dietary recall), but not

measured,

values. However, what about the validity of the sodium intake data on which

these

conclusions were based? The lowest quartile of men and women reported mean

sodium

intakes of only 1041 mg and 678 mg (45·3 and 29·5 mEq), respectively. To believe

the

validity of this reported very low sodium intake, one must assume that 25% of a

randomly picked adult population had spontaneously adopted mean sodium intakes

that

render food frankly unpalatable and are difficult to achieve, with the high salt

content of processed foods. Indeed, 85–90% of salt intake is obligatory (already

in

the food) rather than due to the discretionary use of the salt shaker. To attain

such low sodium intake, food sources would have to be restricted to unprocessed

food

components or specially prepared low-salt products and could not use regular

bread.

I find this hard to believe.

To base scientific conclusions on such dietary recall data is hazardous, since

reported and measured intakes can vary widely. Although calorie intake is not

amenable to simple immediate measurement and verification, under normal

circumstances there is an excellent correlation between actual sodium intake and

measured excretion in any 24 h period. The difficulty of correlating reported

salt

intake with measured output is shown in a study in which a select group of

educated

and motivated healthy volunteers on an ad lib diet were asked to keep

contemporaneous dietary diaries while also collecting daily 24-h urine samples.2

On

one occasion, average daily urinary sodium excretion increased by over 50%, but

the

source of the increased sodium intake could not be found in the food diaries

made

after each meal.

The same quartile also reported exceptionally low calorie intake even though

they

were quite overweight (body-mass index >25 kg/m2). These questionable data,

which

form the basis for Alderman's article, might be explained if they were very poor

estimators of their diet, or if, rather than their actual intake, they reported

what

they thought they should have been eating, given their much higher hypertension

and

heart disease rates.

Powerful computers and sophisticated statistical software packages enable

investigators to do massive number-crunching with ease, but the recipe for this

type

of research should be flavoured with a tincture of commonsense. In the end,

Alderman

and colleagues are correct when they stated that their study “does not justify

any

particular recommendations”.

Sodium intake and mortality

Lancet. 1998 May 16;351(9114):1508.

Hugh de Wardener and Graham A MacGregor

Alderman and colleagues (March 14, p 781)1 claim that sodium intake

significantly and inversely relates to cardiovascular mortality. This claim is

incorrect because the relation is not significant (table 3 p<0·086, 95% CI

0·77–1·02). Their second claim, that salt density (sodium/calories) is directly

related to cardiovascular mortality, is correct (p<0·006).

The contradictory nature of these findings accords with the unreliability of the

method used by the National Health and Nutrition Examination Survey (NHANES I).

It

consisted of one 24 h recall of the food consumed, with no assessment or

measurement

of salt added during the cooking and at the table.

In the 1970s when NHANES I was carried out, added salt accounted for around 50%

of

intake, so dietary calculation of sodium intake based only on the sodium content

of

food gave misleading results. It is about 25 years since those critical of the

relation of salt to blood pressure were dismissive of conclusions based on such

an

inaccurate and haphazard way of assessing salt intake.

Alderman and co-workers try to prop up their claim by stating that it agrees

with

previous findings of theirs in treated hypertensive patients,2 and that in “both

studies baseline sodium intake was assessed”. This statement is not true. It

suggests that the assessment of sodium intake in the two studies was the same.

In

Alderman's previous study, however, the usual (baseline) sodium intake of the

hypertensive individual was not measured. Instead, the investigators used the

sodium

content of a 24 h urine collected after 5 days of temporary salt restriction to

stimulate renin to construct a renin-sodium profile. Not to point this fact out,

and

to then claim that the hypertensive patients who consumed “the lowest salt diet

had

a fourfold greater likelihood of myocardial infarction than did those with a

higher

salt intake” was misleading. The term salt diet suggested that this was the

hypertensives' usual (baseline) intake of salt, when it was the salt intake on

the

fifth day of a temporary reduction. There is no evidence that urinary excretion

of

sodium after a temporary reduction in salt intake for 5 days bears any relation

to

the patients' (usual) baseline intake. It will depend on the patient's ability

to

follow instructions. This study in hypertensives is therefore irrelevant and

does

not contribute in any way to the relation between usual (baseline) salt intake

and

outcome.

Alderman and colleagues also state that there is “convincing evidence of adverse

effects of a low sodium diet on important physiological characteristics”

Detailed

reading of the papers does not support this assertion,3 indeed, one report they

cite4 is outstandingly irrelevant. The results were based on a questionnaire

answered by 11150 people. Daily salt intake from salt in the food was calculated

to

be 8 g daily and another 8·6 g was calculated to be added either during cooking

or

at the table. Therefore, average daily salt intake was 16·6 g—a much higher salt

intake than the average for the USA. A dietary assessment showed that a

substantial

number of participants consumed less than two-thirds of the recommended dietary

intake of vitamin B6, calcium, magnesium, and iron. For some bizarre reason, the

authors then proposed that, theoretically, these potential dietary deficiencies

could have been produced by a low salt intake.

It does not embellish Alderman and co-workers' crusade against the worldwide

recognition that the present consumption of salt should be reduced, to

misrepresent

previous work, make misleading statistical claims, and introduce other

irrelevant

material.

Sodium intake and mortality

Lancet. 1998 May 16;351(9114):1509

Heikki Karppanen and Eero Mervaala

Rather than shed new light on sodium intake and mortality, Alderman and

colleagues' report1 brings unnecessary confusion into the discussion on the

relation

between dietary sodium and mortality.

Alderman and co-workers used a single 24 hour dietary recall on nutrient intake

in

1971–75 to estimate an individual's long-term intake of sodium and calories.

Their

findings show that the investigators have failed in their attempt to find out

whether dietary sodium is associated with mortality in a general population.

They

argue that middle-aged people are likely to have a stable nutrient intake over

many

years. If so, how do they explain the fact that most people in their lowest

sodium

quartile survived the 20 year follow-up period at a calorie intake that should

have

resulted in death from starvation? Among women in the lowest sodium quartile,

the

calculated calorie intake was only 989 kcal, which is about 50% of the

recommended

daily allowance of 1800–2100 kcal for women in this age group. Among men in the

lowest sodium quartile, the calculated calorie intake was 1473 kcal, which is

50–60%

of the recommended dietary daily allowance of 2300–2800 kcal for men in this age

group. For the maintenance of basal metabolism only, middle-aged women require

at

least 1200 kcal daily, and middle-aged men at least 1500 kcal per day. Any

physical

activity increases the calorie requirement. We were surprised that the women on

the

low calorie intake were, on the average, 4 kg (6%) heavier than those in the

highest

sodium quartile, despite a two-fold calorie intake in the latter group. Hence,

the

calculations and conclusions have been based on invalid data.

The investigators overlook evidence that the dramatic decrease in all-cause and

cardiovascular mortality among the 5 million population of Finland during the

past

20 years reflects the decrease in sodium intake, and some increase in the intake

of

potassium.2, 3 and 4

Since the 1970s, changes in the intake of sodium have been carefully monitored

by

measurement of 24 hour urinary sodium excretion in representative population

samples. As a result of the continuous decrease, the average intake of sodium is

now

about 30% lower than it was 20 years ago. Although there has been an increase in

obesity, consumption of alcohol and smoking among women between 1972 and 1992,

there

has been a 10 mm Hg fall in the population average of diastolic blood pressure,

and

about 60% decrease in deaths from stroke and ischaemic heart disease among men

and

women aged 30-59 years.2, 3 and 4 Only 0·5–0·6 mm Hg of the fall of the

population

blood pressure could be attributed to antihypertensive drug treatment.4 The

decrease

in sodium intake and in the dietary sodium-to-potassium ratio in Finland is the

main

factor accounting for the 9·5 mm Hg fall in the population average of diastolic

blood pressure. Evidence that a general reduction of dietary sodium in the

population is beneficial is based on hard mortality endpoints. During the same

period that the average intake of sodium has decreased 30%, the life expectancy

of

Finns has increased by about 5 years.

Sodium intake and mortality

Lancet. 1998 May 16;351(9114):1509-10

H Alderman, Hillel Cohen and Shantha Madhavan

Authors' reply

Sir—Hugh de Wardener and Graham MacGregor ignore the significant inverse

association

of sodium to all-cause mortality (p=0·0069) reported from NHANES 1 and note the

borderline significance (p=0·086) of sodium to cardiovascular (CVD) mortality.

The

values we reported were for the full model. In the best-fit model, in which

variables with a p value greater than 0·10 were eliminated, sodium showed a

highly

significant inverse association with CVD mortality: the hazard ratio was 0·86

(95%

CI 0·81–0·93); p<0·0001. Although common practice is to report the best-fit

model,

we conservatively chose to present the full model.

The independent associations, in opposite directions, of sodium and

sodium/calorie

to mortality, probably reflect different physiological events. On the basis of

additional analyses, we can now report that the inverse sodium to mortality

associations persisted within each sodium/calorie quartile.

de Wardener and MacGregor also note the imprecision with which table salt is

estimated in NHANES 1. In multivariate analysis, the responses (always,

sometimes,

never) were not significantly associated with mortality, which probably reflects

the

small contribution of table salt compared with salt consumed within food. For

discretionary salt to change the results of our study, it would be necessary for

those reporting low table salt to have eaten a high sodium diet, which

contradicts

reports that individuals who have low-sodium diets also tend to use less table

salt

and vice versa.1

de Wardener and MacGregor repeat a previous, inaccurate, statement2 about

baseline

sodium in our study of hypertensives,3 which we have already answered at

length.4

Briefly, patients were uniformly advised to maintain usual diet while avoiding

high-sodium foods. Their sodium intake matched that seen in US Intersalt sites.

Inaccurate assessments were likely to have been distributed throughout the

population, resulting in an underestimation of the true magnitude of the

observed

inverse relation of sodium to events.

Karl Engelman comments on the unreliable estimation of sodium intake by dietary

recall. We agree that NHANES 1 probably underestimated sodium. The real concern,

however, is not that across-the-board under reporting occurred, but whether

errors

in reporting of sodium biased the observed relation. We used stratified and

multivariate analyses to account for the possibility that sicker patients might

have

consumed less sodium and died sooner. The inverse relation of sodium to

mortality

held. We are now able to report two further analyses. Exclusion of 2161

participants

who reported recent change in appetite or dietary alteration produced sodium to

mortality results for the restricted group consistent with the whole. When we

excluded 1926 participants who reported less than a 1000 kcal intake daily, the

results were similar.

Heikki Karppanen and Eero Mervaala refer to ecological reports of communities

with

aggregate changes in several behaviours that have enjoyed a coincidental decline

in

mortality. Sodium intake was not measured in the studies cited, but referenced

from

different, serial cross-sectional studies. There were concurrent changes in

dietary

fat and other nutrients. Without direct linkage of various changes in the

characteristics of individuals to their outcomes, it is mere speculation to

attribute a decline in mortality to one of the many factors (sodium) that

changed.

NR Poulter and on behalf of The Faculty 31st International Society and

Federation of

Cardiology 10-day Teaching Seminar in Cardiovascular Disease, Epidemiology and

Prevention

The faculty consisted of: Barret-Connor, C , Vinjar

Fønnebø,

Kay-Tee Khaw, Darwin R Labarthe, Neil R Poulter, K Srinath Reddy, Grethe S Tell,

Dag

S Thelle

Dietary sodium intake and mortality: NHANES

Lancet. 1998 Sep 19;352(9132):987-8.

The analyses of data from the National Health and Nutrition Examination Survey

(NHANES I), reported by Alderman and colleagues (March 14, p 781)1 are

seriously flawed. The analyses as presented do not exclude individuals with

known

cardiovascular disease (CVD) or hypertension, who were disproportionately

represented in the lowest sodium quartile. Such individuals usually report lower

sodium intake and have increased risk of fatal coronary heart disease. Analyses

excluding participants with a history of CVD (but apparently not those with a

history of hypertension) were reportedly similar but should have been presented.

Participants reportedly in the lowest quartile of sodium intake not only had the

highest prevalence of known CVD and history of hypertension but were also the

oldest, had the highest proportion of black participants, and had the lowest

body

weights, all of which, except sodium intake, are associated with increased death

rates. It would therefore be expected that, irrespective of sodium intake, those

in

the lowest quartile would exhibit the highest mortality.

The inverse association between blood pressure and reported sodium intake

conflicts

with most previous publications.2 However, reported calorie intake was also

inversely associated with blood pressure and mortality. Most nutrients in the

diet

increase with increasing calorie intake. Thus, to avoid confounding by low

calorie

intake, the pertinent analyses were those in which sodium intake was corrected

for

dietary calories.

Data on smoking were collected, but not included in analyses. Because smokers

have

different diets,3 lower body weights and blood pressures,4 are from lower

socioeconomic strata, and have higher death rates than nonsmokers, this omission

makes serious confounding likely.

The inclusion in the multivariate analyses of variables such as sodium intake

and

blood pressure, which may well be related in the causal pathway result in

over-fitting the model, can critically change the size and even the direction of

the

effect under investigation. These errors notwithstanding, the investigators

misinterpret their findings. For example, the reported relative risk for

all-cause

mortality (RR) associated with a 1313 mg difference in sodium intake was 0·88.

However, using their data to estimate the effect of a 1313 mg difference in

sodium

intake in their model, one must assume some specific individual values of

calorie

and of sodium intake. If we compute the RR of two people with sodium intakes of

4000

mg and 2700 mg, respectively, and the same calorie intake of 2000 kcal (8360

kJ),

the estimated RR is actually 1·00. A similar computation with CVD mortality as

the

outcome, gives an estimated RR of 1·02 rather than 0·89, as Alderman and

colleagues

reported.

Despite all these major methodological flaws, higher sodium per calories intake

was

significantly associated with higher rates of cardiovascular (p=0·0056) and

all-cause mortality (p=0·0004). It is hard to understand why the investigators

discount these results as unsupportive of recommendations to use foods with

lower

salt content.

If following appropriate analyses of their NHANES data, the findings were to

remain,

it would then be important to attempt to replicate the findings in other

studies,

given that most previously published data have shown a direct association

between

sodium intake and blood pressure2 and that all low-bloodpressure populations

have

vanishingly low rates of CVD.5

Referred to by: Dietary sodium intake and mortality: NHANES,

The Lancet, Volume 352, Issue 9132, 19 September 1998, Page 988

H Aldermana, Hillel W Cohena and Shantha Madhavana

The analyses of data from the National Health and Nutrition Examination Survey

(NHANES I), reported by Alderman and colleagues (March 14, p 781)1 are

seriously flawed. The analyses as presented do not exclude individuals with

known

cardiovascular disease (CVD) or hypertension, who were disproportionately

represented in the lowest sodium quartile. Such individuals usually report lower

sodium intake and have increased risk of fatal coronary heart disease. Analyses

excluding participants with a history of CVD (but apparently not those with a

history of hypertension) were reportedly similar but should have been presented.

Participants reportedly in the lowest quartile of sodium intake not only had the

highest prevalence of known CVD and history of hypertension but were also the

oldest, had the highest proportion of black participants, and had the lowest

body

weights, all of which, except sodium intake, are associated with increased death

rates. It would therefore be expected that, irrespective of sodium intake, those

in

the lowest quartile would exhibit the highest mortality.

The inverse association between blood pressure and reported sodium intake

conflicts

with most previous publications.2 However, reported calorie intake was also

inversely associated with blood pressure and mortality. Most nutrients in the

diet

increase with increasing calorie intake. Thus, to avoid confounding by low

calorie

intake, the pertinent analyses were those in which sodium intake was corrected

for

dietary calories.

Data on smoking were collected, but not included in analyses. Because smokers

have

different diets,3 lower body weights and blood pressures,4 are from lower

socioeconomic strata, and have higher death rates than nonsmokers, this omission

makes serious confounding likely.

The inclusion in the multivariate analyses of variables such as sodium intake

and

blood pressure, which may well be related in the causal pathway result in

over-fitting the model, can critically change the size and even the direction of

the

effect under investigation. These errors notwithstanding, the investigators

misinterpret their findings. For example, the reported relative risk for

all-cause

mortality (RR) associated with a 1313 mg difference in sodium intake was 0·88.

However, using their data to estimate the effect of a 1313 mg difference in

sodium

intake in their model, one must assume some specific individual values of

calorie

and of sodium intake. If we compute the RR of two people with sodium intakes of

4000

mg and 2700 mg, respectively, and the same calorie intake of 2000 kcal (8360

kJ),

the estimated RR is actually 1·00. A similar computation with CVD mortality as

the

outcome, gives an estimated RR of 1·02 rather than 0·89, as Alderman and

colleagues

reported.

Despite all these major methodological flaws, higher sodium per calories intake

was

significantly associated with higher rates of cardiovascular (p=0·0056) and

all-cause mortality (p=0·0004). It is hard to understand why the investigators

discount these results as unsupportive of recommendations to use foods with

lower

salt content.

If following appropriate analyses of their NHANES data, the findings were to

remain,

it would then be important to attempt to replicate the findings in other

studies,

given that most previously published data have shown a direct association

between

sodium intake and blood pressure2 and that all low-bloodpressure populations

have

vanishingly low rates of CVD.5

PMID: 9519949

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve & db=pubmed & dopt=Abstra\

ct & list_uids=9519949 & query_hl=34

Introduction

Ecological, observational, and experimental data largely, but not invariably,

support the view that a restricted-sodium diet is associated with, and can

induce,

lower blood pressure.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 and 12 Based on these and

other associations13 with intermediate physiological variables, some groups

recommend restriction of daily sodium intake to 106 mmoles (approximately 2400

mg).14 This recommendation is justified by the expectation that the beneficial

changes in intermediate physiological variables outweigh any harmful effects, so

that the net effect would be lower cardiovascular morbidity and mortality, or an

overall improvement in the quality of life. We could find no empirical data to

support these expectations directly.

Although favourable effects on intermediate variables translate into health

benefits

in many circumstances, this is not always the case.15 and 16 In previous studies

in

hypertensive patients, a low-sodium diet was associated with adverse effects on

quality of life17 and an increase in cardiovascular morbidity and mortality.18

To

find out whether dietary sodium is associated with mortality in a general

population, we examined the relation of sodium intake, measured in 1971–75, to

all-cause and cardiovascular-disease (CVD) mortality, up to mid-1992, among

participants in the first National Health and Nutrition Examination Survey

(NHANES

I).

Methods

Participants

There were 20 729 individuals aged 25–75 years at the time of the NHANES I

survey

from 1971 to 1975. From the total sample, 14 407 (70%) underwent medical

examination, and 11 348 of these underwent a nutrition investigation based on a

24 h

recall. Data on sodium intake were missing for two participants, who were

therefore

excluded. Vital status was obtained for the remaining 11 346 participants

through

interview, tracing, and searches of the national death index. Deaths and causes

of

death were confirmed from death certificates. Details on the plan and operation

of

NHANES I have been published previously.19 and 20 Since information on mortality

was

incomplete after June, 1992, the cut-off date for this analysis was June 30,

1992.

Participants not reported as dead before this date were presumed to be alive.

The length of follow-up for each individual, expressed as the number of

person-years

contributed, was calculated from the baseline to the date of death or to the

cut-off

date.

Additional analyses were done on a restricted sample of 9962 participants who,

at

baseline, reported no pre-existing cardiovascular disorders. Those excluded had

a

history of circulatory diseases (International Classification of Diseases, 9th

Revision, codes 390–450), rheumatic heart disease, or heart operation (ICD-9

procedure codes 29–30).

Measurements

Data on nutrient intake were available from a single 24 h dietary recall. Sodium

intake was expressed as mg per day, and total calorie intake as calories per

day.

Qualitative data on use of table salt were also available from the dietary

frequency

questionnaire (responses never, sometimes, or always).

As part of the physical examination in NHANES I, one blood-pressure value was

recorded. For 33% of participants this value was the mean of three readings, two

with the participant seated and once with him or her supine. In the remaining

67% of

participants, one reading was done with the participant seated. Of the total 11

346

participants, 16% had baseline blood-pressure readings in the hypertensive range

(systolic & #8805;160 mm Hg, diastolic & #8805;95 mm Hg, or both). Both the actual

systolic blood pressure and history of hypertension were included in

multivariate

analyses as independent variables.

Statistical analysis

Baseline characteristics, including blood pressure, were defined for male and

female

participants. All statistical tests for continuous (Student's t test) and

categorical variables ( & #967;2) were two-tailed. Since distributions of sodium

intake for men and women differed significantly, baseline characteristics were

assessed according to sex-specific quartiles of 24 h dietary sodium and calorie

intake. The differences between quartiles were tested for statistical

significance

by ANOVA. All-cause and CVD death rates (per 1000 person-years), adjusted for

age

and sex, were calculated by specific dietary quartiles separately for men and

women,

standardised by the direct method, with age and sex distribution of the whole

cohort

as reference. The differences in mortality rates between the lowest and highest

quartiles were tested for statistical significance.

Because sodium intake and total calorie intake were highly correlated (r=0·65),

a

sodium/calorie ratio (expressed as sodium per calorie intake per day) was

calculated

for each participant to reflect the concentration of salt in the diet.

Participants

were stratified according to sex-specific quartiles, and all analyses described

for

sodium and total calorie intake were repeated for sodium/calorie ratio.

proportional-hazards regression models were constructed individually as well as

in

combination, to assess the association of the three dietary measures (sodium,

total

calories, and sodium per calorie) with all-cause and total CVD mortality, with

simultaneous control for such available relevant variables as age at baseline,

sex,

race, body-mass index, history of CVD and hypertension, and systolic blood

pressure.

Stepwise backward regression analysis was done to estimate hazard ratios and 95%

CI

from a full model. To assess whether models with inclusion of other dietary

measures

were better than that with sodium intake as an individual measure, the change in

the

log-likelihood value was tested for significance as a & #967;2 statistic with 1

degree of freedom.

All analyses were repeated on the restricted study sample of participants with

no

reported history of pre-existing CVD at baseline. All statistical analyses were

done

by means of SPSSWIN (version 7.0).

Results

The 4478 (39·5%) men differed significantly (p<0·05) from the 6868 women in mean

age

(52·9 vs 46·7 years), weight (76·6 vs 66·1 kg), and blood pressure (138/86 vs

134/82

mm Hg) and frequency of reported history of CVD (15 vs 11%) and hypertension (18

vs

15%). Mean daily sodium intake (2515 vs 1701 mg) and calorie intake (2159 vs

1471

kcal [1 kcal=0·0042 MJ]) were also significantly higher for men than for women.

In

response to the question on use of table salt more men than women replied

“always”

(39 vs 24%) and fewer replied “never” (38 vs 53%).

Baseline characteristics of participants by quartile of sodium intake, total

calorie

intake, and sodium/calorie ratio were analysed for men and women separately. For

men

and for women, there were significant differences (ANOVA, p<0·05) across the

four

quartiles of sodium intake in mean age, blood pressure, bodyweight, and

body-mass

index (women only), use of table salt, and the proportions of black people and

individuals with a history of CVD or hypertension (table 1). Mean calorie intake

increased from the lowest to the highest sodium-intake quartile, but

surprisingly,

body-mass index was similar across the quartiles of sodium intake for men.

Sodium

and calorie intakes were closely correlated (r=0·65). The characteristics of the

four quartiles of total calorie intake (not shown) were very similar to those of

the

quartiles of sodium intake.

Table 1. Baseline characteristics by quartile (Q) of sodium intake from 24 h

recall

....................................

Characteristic* Men Women

--------Q1 (n=1120) Q2 (n=1118) Q3 (n=1120) Q4 (n=1120) Q1 (n=1717) Q2 (n=1721)

Q3

(n=1713) Q4 (n=1717)

.....................................

Demographic characteristics

Age (years) 56·9 (14·3) 54·4 (15·5) 51·7 (15·5) 48·6 (15·1) 49·8 (16·0) 49·2

(16·0)

47·8 (15·9) 43·9 (14·9)

Black race 24·4% 17·0% 13·3% 8·8% 26·0% 18·3% 15·4% 11·5%

History

CVD 17·6% 15·4% 14·5% 11·3% 11·5% 12·0% 9·4% 9·6%

Hypertension 21·6% 18·1% 15·5% 14·9% 16·8% 15·0% 14·2% 12·6%

Anthropometric characteristics

Body-mass index (kg/m2) 25·7 (4·3) 25·4 (4·0) 25·2 (4·2) 25·5 (4·1) 26·6 (6·0)

25·6

(5·7) 25·3 (5·5) 24·6 (5·5)

Bodyweight (kg) 76·0 (14·5) 76·4 (13·4) 76·4 (14·2) 77·7 (13·7) 68·4 (16·3) 66·3

(15·2) 65·6 (14·3) 64·3 (14·9)

Blood pressure (mm Hg)

Systolic 142·4 (24·9) 138·8 (33·2) 136·0 (22·3) 134·4 (20·6) 136·7 (26·8) 134·9

(26·1) 133·7 (26·2) 129·5 (24·5)

Diastolic 87·3 (14·0) 85·8 (13·0) 84·5 (12·1) 84·6 (11·8) 83·5 (13·7) 82·6

(13·8)

81·8 (13·1) 80·2 (12·8)

24 h dietary recall

Sodium intake (mg) 1041 (322) 1832 (195) 2647 (282) 4538 (1489) 678 (229) 1232

(138)

1791 (196) 3105 (1002)

Calorie intake (kcal) 1473 (638) 1930 (708) 2297 (732) 2937 (1050) 989 (408)

1331

(434) 1589 (518) 1976 (682)

Sodium/calorie (mg/kcal) 0·80 (0·35) 1·07 (0·42) 1·27 (0·44) 1·67 (0·61) 0·76

(0·35)

1·02 (0·35) 1·25 (0·48) 1·70 (0·66)

Use of table salt

Always 33·9% 39·2% 37·2% 45·9% 20·4% 21·5% 24·5% 30·0%

Never 46·2% 35·5% 36·4% 31·9% 57·4% 57·0% 52·5% 46·9%

....................................

* Data presented as mean (SD) or as % of participants.

Baseline characteristics generally differed (ANOVA, p<0·05), for men and for

women,

across quartiles of sodium/calorie ratio, with the exceptions of systolic blood

pressure in men; weight, body-mass index, and use of table salt always in women;

and

diastolic blood pressure, history of hypertension, history of CVD, and use of

table

salt “never” in both sexes (table 2).

Table 2. Baseline characteristics by quartile (Q) of sodium per calorie intake

from

24 h recall

.....................................

Characteristic* Men Women

--------Q1 (n=1118) Q2 (n=1120) Q3 (n=1120) Q4 (n=1119) Q1 (n=1716) Q2 (n=1717)

Q3

(n=1717) Q4 (n=1717)

.....................................

Demographic characteristics

Age (years) 50·3 (15·9) 52·9 (15·3) 54·0 (15·2) 54·5 (15·0) 46·1 (15·5) 48·1

(16·0)

48·7 (16·0) 47·9 (15·9)

Black race 20·1% 14·8% 17·0% 11·6% 23·2% 17·4% 16·8% 13·8%

History

CVD 12·5% 14·6% 15·4% 16·0% 9·3% 10·4% 11·7% 11·1%

Hypertension 17·1% 15·6% 18·4% 19·0% 13·9% 14·2% 14·6% 16·0%

Anthropometric characteristics

Body-mass index (kg/m2) 25·7 (4·2) 25·4 (4·1) 25·2 (4·2) 25·4 (4·1) 25·6 (5·9)

25·5

(5·5) 25·4 (5·9) 25·6 (5·6)

Bodyweight (kg) 77·7 (14·3) 76·4 (13·7) 76·0 (14·1) 76·2 (13·7) 66·4 (15·9) 66·2

(14·9) 65·7 (15·5) 66·1 (14·7)

Blood pressure (mm Hg)

Systolic 137·0 (22·8) 137·0 (22·9) 138·8 (22·6) 138·8 (23·7) 131·6 (25·5) 133·7

(25·8) 134·9 (26·4) 134·5 (26·4)

Diastolic 85·7 (12·8) 85·1 (12·8) 85·9 (12·8) 85·6 (12·9) 81·7 (13·7) 82·1

(13·3)

82·3 (13·1) 82·0 (13·4)

24 h dietary recall

Sodium intake (mg) 1436 (730) 2134 (904) 2663 (1177) 3827 (1845) 920 (483) 1408

(579) 1815 (767) 2662 (1254)

Calorie intake (kcal) 2330 (1069) 2208 (916) 2098 (922) 2003 (898) 1520 (706)

1506

(609) 1467 (604) 1391 (604)

Sodium/calorie (mg/kcal) 0·62 (0·16) 0·97 (0·82) 1·27 (0·10) 1·95 (0·52) 0·60

(0·16)

0·94 (0·08) 1·24 (0·11) 1·96 (0·58)

Use of table salt

Always 45·5% 37·8% 37·8% 35·3% 23·7% 25·4% 23·9% 23·5%

Never 35·5% 36·8% 37·8% 39·9% 53·9% 52·2% 53·5% 54·0%

..........................................

* Data presented as mean (SD) or as % of participants.

Dietary intake and mortality

Of the 11 346 participants, 7423 were presumed alive and 3923 had died (1970

from

CVD) by June 30, 1992. All-cause mortality rates adjusted for age and sex

(figure)

were inversely and significantly related to sodium intake per day (lowest to

highest

quartile 23·18 to 19·01 per 1000 person-years; p<0·0001) and total calorie

intake

per day (25·03 to 18·40 per 1000 person-years; p<0·0001). A similar inverse

relation

was seen for CVD mortality rates from lowest to highest quartile of sodium

intake

(11·80 to 9·60 per 1000 person-years; p<0·0019) and calorie intake (12·80 to

8·94

per 1000 person-years; p<0·0002). The results were similar for the analysis

limited

to participants with no reported history of CVD at baseline (not shown).

For sodium/calorie ratio, there was a weak direct relation to all-cause

mortality

(lowest to highest quartile 20·27 to 21·71 per 1000 person-years; p=0·14) and a

significant direct relation to CVD mortality (9·73 to 11·35 per 1000

person-years;

p=0·017). When the analysis was restricted to participants with no reported

history

of CVD at baseline (not shown), both all-cause and CVD mortality were directly

related to sodium/calorie ratio, but the relation did not achieve statistical

significance.

Multivariate analysis

Because sodium intake, total calorie intake, and sodium/calorie ratio were all

associated with all-cause and CVD mortality, we undertook stepwise

proportional-hazards regression analysis with all three dietary measures in the

model (table 3). Models were constructed separately for all-cause and CVD

mortality

as dependent variables. These models had significantly better likelihood values

(p<0·001) than models limited to one dietary measure. In these single-measure

analyses, sodium intake had a significant inverse association with both

all-cause

and CVD mortality. Sodium/calorie ratio showed a significant positive relation

to

both all-cause and CVD mortality. However, in the model with both measures of

sodium

intake, total calorie intake no longer had an independent relation to mortality.

The

relation of sodium intake to all-cause mortality, although significant

(p=0·0069),

was small. For example, an increase in dietary sodium of 1000 mg was associated

with

a 10% reduction in mortality. When potassium intake, alcohol use, family income,

and

education of head of household were each added to the models, the results for

sodium

intake and sodium/calorie ratio did not change substantially. Both potassium

intake

and alcohol use were removed from the all-cause and CVD models during backwards

stepwise elimination. The results were similar when these analyses were

restricted

to the participants with no history of CVD at baseline.

Table 3. Variables associated with risk of all-cause and CVD mortality in

proportional-hazards regression (full model)

.......................................

Variable* All-cause mortality CVD mortality

--------ß p Hazard ratio (95% CI)† ß p Hazard ratio (95% CI)†

.......................................

Male 0·6286 <0·0001 1·88 (1·75–2·01) 0·6361 <0·0001 1·89 (1·71–2·09)

Black race 0·1585 0·0001 1·17 (1·08–1·27) 0·0465 0·4347 1·05 (0·93–1·18)

History of CVD 0·4033 <0·0001 1·50 (1·39–1·62) 0·4859 <0·0001 1·63 (1·46–1·80)

History of hypertension 0·1000 0·0241 1·11 (1·01–1·21) 0·0834 0·1668 1·09

(0·97–1·22)

Age (years) 0·0810 <0·0001 3·62 (3·44–3·82) 0·0922 <0·0001 4·33 (3·98–4·71)

Body-mass index (kg/m2) & #8722;0·0047 0·1932 0·98 (0·94–1·01) 0·0081 0·1000 1·04

(0·99–1·10)

Systolic blood pressure (mm Hg) 0·0057 <0·0001 1·15 (1·11–1·20) 0·0103 <0·0001

1·29

(1·23–1·36)

Sodium (mg) & #8722;0·0001 0·0069 0·88 (0·80–0·96) & #8722;0·00009 0·0864 0·89

(0·77–1·02)

Calories (kcal) & #8722;0·00001 0·8562 0·99 (0·91–1·08) & #8722;0·00002 0·7394

0·98

(0·87–1·11)

Sodium/calories (mg/kcal) 0·1955 0·0004 1·12 (1·05–1·19) 0·2159 0·0056 1·13

(1·04–1·24)

Table salt use (always) 0·0741 0·1201 1·08 (0·98–1·18) & #8722;0·0130 0·8510 0·99

(0·86–1·13)

Table salt use (never) 0·0057 0·8889 1·01 (0·93–1·09) & #8722;0·0012 0·9825 1·00

(0·89–1·12)

...........................................

* For categorical variables, yes=1. For table salt use variables,

reference=sometimes.

† For continuous variables, hazard ratios are for 1 SD change. SDs: age=15·9

years,

body-mass index=5·15 kg/m2, systolic blood pressure=24·98 mm Hg, sodium=1313 mg,

calories=849 kcal, sodium/calorie=0·5787 mg/kcal.

All analyses were repeated with stratification for age (<65 and & #8805;65

years).

For the older age-group, the relation of sodium intake, sodium/calorie ratio,

and

calorie intake to all-cause and CVD mortality were similar to those observed for

the

whole study population and were statistically significant. Trends for the

younger

group were similar, but did not achieve significance.

Discussion

Our main findings are that dietary sodium intake is inversely associated with

all-cause and CVD mortality, and that dietary sodium/calorie ratio is directly

associated with both mortality rates. These associations, although small, are

significant and independent, both of each other and of other factors known to

influence mortality.

The inverse association of salt intake with mortality is consistent with the

findings of a similar observational study of 3000 participants in a systematic

programme to control hypertension.18 In both studies, baseline sodium intake was

assessed and related (with control for known confounders) to subsequent

morbidity or

mortality. Different ways of estimating sodium intake—dietary recall and 24 h

urinary measurement—produced similar results in the two studies. NHANES I

baseline

data, unlike the study in hypertensive patients, included information on other

dietary factors. Since there was a strong correlation of sodium intake with

total

energy consumption, we explored the relation of calorie intake to mortality,

both as

an isolated variable and in relation to its sodium content. We found that both

sodium alone and the sodium concentration in the diet helped to explain, in

opposite

directions, variations in all-cause and cardiovascular mortality.

These findings suggest that, although there may be a specific relation of sodium

to

survival, it is not likely to be simple; at the very least, it must be

considered in

the total dietary context. Moreover, given the genetic, behavioural,

environmental,

and dietary heterogeneity in most industrialised societies, different

individuals

may have different optimum sodium intakes. Other research supports the concept

that

the diet as a whole may be a more important determinant of health outcome than

an

individual component.21 The high degree of correlation between most dietary

components and total calorie consumption suggests that there may be important

interaction with other individual dietary components.

The available data do not provide an explanation for the observations. A

favourable

effect of sodium restriction on various intermediate physiological variables has

been shown, particularly, but not exclusively, for blood pressure and other

haemodynamic characteristics. In addition to the single outcome study suggesting

increased morbidity in treated patients on a low-sodium diet,18 there is

convincing

evidence of adverse effects of a low-sodium diet on important physiological

characteristics, including the sympathetic system and the renin-angiotensin

system

in particular.22, 23, 24, 25 and 26 The ultimate health outcome of any

intervention

is the sum of all its biological effects. For a low-sodium diet, harm may

outweigh

benefit.

An important limitation of this study is the reliability of the dietary

measures.

Assessment of the exposure (dietary sodium and calories) was made only once, at

baseline, and by recall rather than objective measurement. The dietary recall of

sodium and calorie intake is probably an underestimation, as NHANES III, using

new

methods, suggests.27 Memory can be faulty, estimates of portion size can be

mistaken, and diet can change from day to day. However, to the extent that such

variation was random, it would tend to mute the relation of exposure (sodium and

sodium/calories) to outcome (death). Although a bias is possible, there is no

inherent reason to suspect one. Also, we have no information about dietary

patterns

after the baseline examination. However, other studies have shown that

middle-aged

people are likely to have a stable nutrient intake over many years.28 and 29

Nevertheless, no valid inferences can be made about the effect of usual,

long-term

dietary patterns. The substantial limitations of the dietary measures should be

seen

in the context that these data are the best available for a large, long-term

population study in which mortality outcomes are also available. These data have

been used for other studies linking nutrients to outcome.30, 31 and 32

The outcome data are more reliable, since all-cause mortality is an unequivocal

endpoint; ascertainment was unbiased and complete. Confidence in less precise

disease-specific mortality must be guarded.

All research, especially that with non-experimental designs, is prey to

confounding.

Ecological studies of sodium in populations are the most susceptible to

confounding,

but observational studies of individuals, such as ours, are not exempt. There

can be

control for some known confounders, but not for all. The problem arises when an

unrecognised factor, or a factor for which insufficient data are available,

confounds the observed associations by being associated with both exposure and

outcome. Smokers, among whom we expect the mortality rate to be higher, might

also

have different dietary intake, but we did not have data to address this issue

here.

However, in a previous study, data on smoking were available, and smoking did

not

affect the inverse sodium to mortality relation.18

Another difficulty is that people with CVD, who are therefore at higher risk of

mortality, might be more likely to adopt a low-sodium diet. To address this

issue,

we undertook stratified analysis that excluded participants with pre-existing

CVD,

as well as multiple regression analysis. In both cases, the relation of sodium

to

mortality persisted. Nevertheless, confounding by indication, despite efforts to

control for it, cannot be entirely excluded as a possible contributor to these

results.

Because of the controversy that surrounds the issue of appropriate sodium

intake,

these new data are valuable. Much research links dietary sodium and its

variation to

intermediate biological characteristics, but this study adds to the smaller body

of

information relating sodium intake to ultimate morbidity and mortality. The

information derives from a methodologically sound study carried out in a

probability

sample of the US population. During follow-up of more than two decades, 3923

deaths

occurred.

Even if the association is valid, and confirmed elsewhere, it does not sustain

any

claim of causality. Even if the association is significant, dose-related, and

independent, it may be a marker for some other dietary or non-dietary factor and

therefore not causally related. In addition, the observational data here provide

no

insight into what might occur if consumption were artificially altered to modify

absolute sodium intake or its dietary concentration. A manipulated sodium intake

would not inevitably yield the outcomes that occur in individuals naturally

consuming that amount of salt. In other words, these data provide no support for

a

recommendation to increase (or decrease) intake of salt or to decrease its

concentration in the diet.

Al Pater, PhD; email: old542000@...

__________________________________________________