Epidemiologic Evidence on Salt and Blood Pressure

Table 1 shows average systolic blood pressure in two contrasting communities, US whites and Solomon Islanders.1,2 The data were collected in about 1970, when the Solomon Island tribes lived a hunter-gatherer lifestyle with little contact with Western civilization. The familiar rise in blood pressure with age did not occur in the hunter-gatherers, a fact that was confirmed by studies of several such communities about 30 years ago.3 Migration studies have demonstrated corresponding increases in blood pressure when economically undeveloped people moved to urban environments.4 Average blood pressure levels among older people in Western populations are high in comparison to hunter-gatherers, an observation which belies the term “normotensive.” Table 2 summarizes data on the differences between the two communities in the major determinants of blood pressure—dietary salt and potassium, body weight, habitual exercise, and alcohol consumption. All differ between the two communities, and all contribute to the rise in blood pressure with age in the Western community. Average Systolic Blood Pressure (mm Hg) in Typical American and Low Salt Hunter-Gatherer Communities According to Age Inland tribes. Average Systolic Blood Pressure (mm Hg) in Typical American and Low Salt Hunter-Gatherer Communities According to Age Inland tribes. Average Values of Determinants of Blood Pressure in Typical American and Hunter-Gatherer Communities Adjusted to US height.2 Average Values of Determinants of Blood Pressure in Typical American and Hunter-Gatherer Communities Adjusted to US height.2 All the factors listed in Table 2 should be tackled in efforts to lower the population average levels of blood pressure among older people in Western populations. But salt is particularly important for three reasons—its effect on blood pressure is large,3 the dietary intake in Western populations is particularly high, so a large reduction could be attained, and, with the cooperation with the manufactured food industry, lowering dietary salt could be easily and straightforwardly accomplished. Age-specific estimates of the size of the association between salt and blood pressure indicate that approximately 40% of the difference in blood pressure between the two communities in Table 1 is attributable to salt.3 This has been confirmed directly by the same study of Solomon Islanders: the blood pressure of coastal tribes whose salt intake was similar to that of Americans (as they used sea water in cooking), but they differed from Americans in the other determinants of blood pressure; their blood pressure lay approximately 40% of the way between that of the low-salt inland Solomon Islanders (Table 1) and the Americans.2 The relationship between salt and blood pressure cannot adequately be expressed as a simple average reduction in blood pressure that might be expected in any trial of salt reduction. Table 3 lists four important factors that influence the relationship. Characteristics of a Study TH Influence the Observed Relitionship Between Salt and Blood Pressure Characteristics of a Study TH Influence the Observed Relitionship Between Salt and Blood Pressure The importance of age is shown in Table 1. Salt, and the other determinants of blood pressure, have little impact at age 15 to 25. Their effect is strongly age-dependent; “sensitivity” to salt and other factors increases with age on average. The effect of a higher starting blood pressure is illustrated by a second important difference between the blood pressure of the two populations shown in Table 1. Not only is the average blood pressure greater in the Western than the hunter-gatherer community, but so is the standard deviation of blood pressure (a measure of the range of blood pressure across the population).3 The age-specific differences in blood pressure between individuals on the 90th percentile of the distribution of blood pressure in each community are therefore greater than the differences between the averages (or 50th percentiles) shown in Table 1, whereas the differences between individuals in the 10th percentile of blood pressure in each community are smaller. A change in salt or in other determinants of blood pressure would therefore be expected to have a greater effect in persons with comparatively high, rather than low, blood pressure. The change of blood pressure must increase with the magnitude of the change in dietary sodium; the association is continuous.3 In a trial, the duration of the sodium reduction is important; the reduction in blood pressure is not fully attained until about a month after the reduction in dietary sodium.5 Such a time effect is not surprising in view of the physiological changes that follow a reduction in dietary sodium. Total body sodium falls for example, and in the trials of salt reduction5 an average weight loss of about 1 kg occurred, which corresponds to 1 L water; its loss is unlikely to be complete in a few days. Similarly, diuretics lower blood pressure, but exert their maximum effect only after about a month. Simple metaanalyses of the trials of dietary salt reduction that do not take these factors into account may be misleading. One such metaanalysis concluded that in people with average blood pressure the effect of dietary salt reduction reduced blood pressure by only 1 mmHg.6 The average age of the subjects, however, was 22 years, and Table 1 indicates that only a small effect is to be expected at this age. The error was to generalize the result to older age groups. A second important error was made in failing to appreciate that the average trial duration of 2 weeks was too short to attain the maximum blood pressure reduction. The trials of shortest duration tend also to be those with the greatest salt reduction (since a large dietary salt reduction precludes eating almost any manufactured food, or eating in any restaurant or canteen, and can therefore be sustained for only a few days; only a more modest reduction in salt could be sustained for several weeks). In metaanalyses, therefore, the effect of duration tends to be cancelled by the effect of dose. Failure to take this interaction into account led to the incorrect conclusion that duration was of little consequence, and failure to omit short trials underestimated the blood pressure reduction per unit sodium reduction. The four factors that influence the relationship between salt and blood pressure listed in Table 3 are all important, but they give rise to too many permutations for metaanalyses of the trials of dietary salt reduction to accommodate. The trials cannot quantify the relationship in each 10 year age group and according to different levels of starting blood pressure: most of the subjects with average blood pressure were under the age of 30 and most of the subjects with higher blood pressure were aged 45 to 60.5,6 Also many of the trials were too short (1 or 2 weeks) to show the full effect, and the salt reduction attained in different trials varied greatly in magnitude. Estimates are best derived from observational studies comparing blood pressure between populations with differing dietary sodium intakes (so-called ecological studies), because they are sufficiently large to provide the extensive data necessary. An analysis with information from 47,000 persons has provided the most data, and therefore the most precise estimates.3 Regression analyses of average blood pressure on sodium intake, and of the standard deviation of blood pressure on sodium intake, in 10-year age groups, separately across economically developed and undeveloped communities were done. Linear models fitted the data well, and the association was of similar magnitude in men and women.3Table 4 shows the estimates of the differences in systolic and diastolic blood pressure per 100 mmol/24 h difference in sodium intake according to age and percentile of the blood pressure distribution. In older persons the association is large: a 100 mmol/24 h change in sodium intake is associated with a difference of about 10/5 mm Hg in blood pressure, and in persons on the 95th percentile of the distribution (“hypertensives”), the estimated blood pressure difference is 15/7 mm Hg. Estimated Change in Systolic and Diastolic Blood Pressure (mm Hg) PER 100 mmol/24 H Change in Sodium Intake by Percentiles of the Blood Pressure Distribution3 Estimated Change in Systolic and Diastolic Blood Pressure (mm Hg) PER 100 mmol/24 H Change in Sodium Intake by Percentiles of the Blood Pressure Distribution3 Validation in six different ways has confirmed that these are accurate estimates of the relationship between sodium and blood pressure according to age. Regression analyses of blood pressure differences across communities with different sodium intakes were conducted separately in economically developed communities (such as the US whites in Table 1) and economically undeveloped communities (such as the Solomon Islanders), and the separate analyses in the two very different types of community yielded similar results.3 The data in Table 4 are a weighted average. The separate analyses in economically developed and undeveloped communities avoided material confounding the estimates by other determinants of blood pressure. Potassium intake, body mass index, and alcohol intake differ on average between developed and undeveloped communities (Table 2), but these determinants are not associated either across developed communities or across undeveloped communities with sodium intake (that is, there is no systematic tendency for communities with higher potassium intake, for example, to also have either higher or lower sodium intake on average).3 The estimates of the relationship between sodium and blood pressure should be unbiased. The estimates of the relationship are very close to those from the large Intersalt study (which were adjusted for differences in dietary potassium, alcohol, and body mass index).7,8 The estimates of the relationship were similar to those from 14 published cross-sectional studies of blood pressure and 24 h urinary sodium conducted in American, European, and Asian communities.9 The relationship in such cross-sectional studies is greatly attenuated by the large degree of within-person day-to-day variation in sodium intake; in this analysis the cross-sectional data were adjusted to allow for this. In an analysis of 33 trials of 5 or more weeks' duration of the effect, predictions (based on the data summarized in Table 4) of the reduction in blood pressure in each trial according to the age and starting blood pressure of the subjects and the size of the sodium reduction closely matched a wide range of observed reductions (between 1 and 36 mm Hg).5 This renders bias implausible since it would be too great a coincidence for a source of error to reproduce such close matching between observed and predicted blood pressure reductions; the effect of bias would rather be to destroy such matching. In a second analysis restricted to those trials that avoided possible bias because the high and low salt diets were otherwise identical, the order of the diet avoided regression to the mean, or the trials were double-blind with otherwise identical diets taken in random order, the matching between observed and predicted blood pressure reductions was equally close.10 Animal experiments have confirmed a relationship. The early experiments in rats by Dahl and his colleagues showed a dose-response relationship between increasing dietary salt and blood pressure with variations in responses between individuals comparable to the effect according to blood pressure percentiles shown in Table 4.11 The variation between individuals was approximately normally distributed, and largely genetically determined. Recent experiments in chimpanzees have shown an important effect of dietary salt reduction on blood pressure, which again varied in extent between individuals.12 The risk of stroke and ischemic heart disease increases continuously with blood pressure.13 The high average blood pressure in older people in Western populations (Table 1) is important, and a modest reduction in this average would produce an important decrease in the incidence of circulatory diseases.13 If dietary salt was reduced on average by 3 g (50 mmol sodium)/day, the average blood pressure would be expected to fall by about 5 mm Hg (systolic) in people over 50 (Table 4). This reduction in the large proportion of the population who are close to the average would be more important than the greater reduction in blood pressure (8 mm Hg, systolic) in the minority who are hypertensive. It has been estimated that the reduction in dietary salt of 3 g/day would reduce age-specific stroke mortality by about 22% and ischemic heart disease mortality by about 16%—an important public health achievement. There is an analogy with global warming: a small change in the average carries important long-term consequences. Dietary salt reduction is important for the entire community.14 The average dietary salt consumption is about 10 g/day in Western societies, but <1% of this in some communities.3 A reduction in average daily salt consumption from 10 g to below 1 g/day is physiologically possible, a reduction from 10 g to 7 g is a more realistic target. It could only be attained, however, with the cooperation of the manufactured food industry, since 75% of dietary salt is added by manufacturers.15 A concerted program to gradually reduce the added salt by 3 g/day would greatly benefit the public health. The sodium content of manufactured foods has fallen by 10% to 15% over 12 years16; an acceleration of the existing process is all that is required. Important in this respect are randomized trials showing that such a gradual reduction need have no deleterious effect on the perceived taste of manufactured food14; indeed, it would improve taste. Randomized trials have been conducted in which subjects were asked to choose blindly the most pleasant from a range of soups, crackers, or other foods that were identical in all respects except their salt content.17–19 Individuals who had been randomized to take a reduced salt diet for the previous few months chose less heavily salted foods as most pleasant, while those who remained on their usual diet chose the more heavily salted foods. Salt taste is a matter of custom, there is no innate biological tendency for people to prefer heavily salted food. The public is subjected to an unnecessarily large amount of dietary salt, mostly with no choice. A gradual reduction in this added salt, together with food labeling and the availability of low salt alternatives, would be a welcome and important contribution to the public health by the manufactured food industry.