Curr Hypertens Rep (2015)7:4 DOI 10.1007/s11906-015-0559-8
PREVENTION OF HYPERTENSION: PUBLIC HEALTH CHALLENGES (P MUNTNER, SECTION EDITOR)
Dietary Sodium and Cardiovascular Disease Andrew Smyth 1,2 & Martin O’Donnell 1,2 & Andrew Mente 1 & Salim Yusuf 1
# Springer Science+Business Media New York 2015
Abstract Although an essential nutrient, higher sodium intake is associated with increasing blood pressure (BP), forming the basis for current population-wide sodium restriction guidelines. While short-term clinical trials have achieved low intake (6 months). Guidelines assume that low sodium intake will reduce BP and reduce cardiovascular disease (CVD), compared to moderate intake. However, current observational evidence suggests a J-shaped association between sodium intake and CVD; the lowest risks observed with 3–5 g/day but higher risk with 5 g/day) and increased risk of CVD. Although lower intake may reduce BP, this may be offset by marked increases in neurohormones and other adverse effects which may paradoxically be adverse. Large randomised clinical trials with sufficient follow-up are required to provide robust data on the long-term effects of sodium reduction on CVD incidence. Until such trials are completed, current This article is part of the Topical Collection on Prevention of Hypertension: Public Health Challenges * Andrew Smyth [email protected] Martin O’Donnell [email protected] Andrew Mente [email protected] Salim Yusuf [email protected] 1
Population Health Research Institute, Hamilton Health Sciences, McMaster University, Hamilton, ON, Canada
2
HRB Clinical Research Facility Galway, NUI, Galway, Ireland
evidence suggests that moderate sodium intake for the general population (3–5 g/day) is likely the optimum range for CVD prevention. Keywords Sodium . Salt intake . Cardiovascular disease . Hypertension . Guidelines
Introduction Sodium is required for normal physiological function, and total body sodium is tightly regulated (via multiple mechanisms) to maintain extracellular sodium concentrations within a narrow range [1]. The main dietary source of sodium is salt (sodium chloride), which accounts for approximately 95 % of daily intake. The majority of the world’s population (∼95 %) currently consume between 3 and 6 g/day of sodium [2, 3•]. Recently, there has been much debate about optimal sodium intake and the evidence linking sodium intake and cardiovascular disease. Current guidelines are based on the following assumptions: (i) any elevation in systolic blood pressure (BP) above 115 mmHg is associated with increasing risk of cardiovascular disease (CVD); (ii) measures of sodium intake are positively associated with elevated BP; (iii) reducing sodium intake will reduce BP irrespective of the level of sodium intake or BP level [4]; (iv) reducing sodium must therefore reduce CVD. However, this chain of events has never been fully demonstrated, and there are substantial data to question these assumptions [5]. Small and short-term clinical trial data show that sodium reduction reduces blood pressure, seen predominantly in those with elevated BP but not so in nonhypertensives [6]. These data have prompted guidelines to recommend that sodium intake should dramatically be reduced in the entire population [7] to less than 2 g/day (i.e. less
47
Page 2 of 8
than half current intake). These recommendations mean that most people will require major dietary change to achieve guideline targets. In this review, we discuss global sodium intake (including measurement issues) and review the evidence linking sodium intake to cardiovascular disease (including blood pressure).
Global Sodium Intake One of the most highly cited estimates of sodium intake comes from the INTERSALT Study, which estimated sodium intake from 24-h urine collections in 52 population samples in 32 countries (n = 10,079) [8]. In order to measure withinindividual variability, 8 % of the sample completed a second 24-h urine collection (3–6 weeks later). INTERSALT showed wide global variations in 24-h sodium excretion ranging from 0.46 g/day (Yanomamo Indians, Brazil) to 6.0 g/day in men and 5.4 g/day in women in Tianjin, northern China [9]. Subsequently, the INTERMAP study used two consecutive 24-h dietary recalls and one 24-h urine collection to estimate sodium intake in Japan, China, UK and USA [10]. They confirmed that the highest mean sodium excretion was found in northern China, followed by Japan, USA and the UK [11, 12]. A recent meta-analysis of cross-sectional studies (187 countries) estimated mean global intake at 3.95 g/day, with significant regional variations [2]. The Prospective Urban Rural Epidemiological Study (PURE Study) is the largest study (n>100,000) to report global variations in sodium intake (628 communities in 18 countries) with a mean intake of 4.9 g/day [3•]. As sodium is a nutrient rather than a food type, intake is embedded within the overall diet pattern; sources include discretionary (added during cooking or at the table) and non-discretionary (processed or pre-prepared foods) [13]. The proportion of discretionary to non-discretionary use varies between regions [14]. Importantly, the regions of the world with the highest non-discretionary use are not the regions with the highest overall intake.
Measurement The method used to measure sodium intake is a key methodological issue, as differences between methods may affect the absolute estimates of sodium intake and make comparisons between studies challenging. Two main approaches can be used to measure sodium intake—estimating sodium intake from dietary questionnaires or measuring urinary sodium excretion. The current reference standard is repeated 24-h urine collection for sodium, as 90–95 % of sodium ingested is thought to be excreted in the urine. Multiple collections are thought to be required to account for day-to-day variations in sodium intake. However, recent data suggest that non-renal mechanisms generate weekly and monthly infradian
Curr Hypertens Rep74: 1 )5 02(
rhythmicity of sodium storage (in the interstitium, especially the skin), independent of daily sodium intake [15, 16•]. These findings also suggest that variations in excretion, transfer of sodium between different compartments in the body, accumulation of sodium in the body and sodium intake itself may influence the association between sodium and cardiovascular disease. A key limitation of 24-h urine collection is the high frequency of incomplete sample collection (due to under collection) that may bias estimates (underestimate intake) or result in the exclusion of participants (e.g. those engaged in manual labour or those who have to travel for work), limiting generalizability. Based on these limitations, even single 24-h urine collections are associated with high rates of incomplete collection (as much as 30 %), and so repeat 24-h collection is impractical in large international epidemiologic studies. In order to address the limitations of 24-h urine collections (single or multiple), various formula-derived estimates, including the Kawasaki formula [17], Tanaka formula [18], INTERSALT formula [19] and Mege formula, were developed and validated against 24-h urine collections [20]. The timing and fasting status of urine samples and the population included are important considerations when choosing a formula. In North America, the Tanaka and INTERSALT formulae applied to non-fasting, random urine samples are associated with the least biased estimates [20]. An international (11 countries) study validated Kawasaki formula-derived estimates from fasting urine samples (ICC 0.71) [21], indicating that this approach is a robust and reliable method of estimating sodium intake in population-based studies. While multiple urine collections (random, fasting or 24-h) in the same individual is likely the best measure of usual (habitual) sodium intake, this too is impractical in large studies. Instead, the approach is to measure the degree of variability in the same individual at several time points in a subset of individuals (e.g. a few percent) in larger studies, and using the degree of correlation between two or measures in the same group of individuals, the estimate of sodium intake is statistically adjusted. This method has been used by several investigators including INTERSALT and PURE with regard to sodium estimates and also for other risk factors such as cholesterol or BP and is a widely accepted method to analyse epidemiologic associations [3•]. A key point is that if any estimate has greater imprecision, the use of this measure tends to underestimate the slope of the association between sodium intake and BP or CVD. However, it cannot change the shape of the association, e.g., a direct association to an inverse association. Further, if the variability between two measures is similar across the entire range of exposures (e.g. sodium or cholesterol), then there is no reason to expect that a linear relationship will be changed to one that is nonlinear or J-shaped. Sodium consumption from the diet may also be estimated using dietary questionnaires, such as food frequency questionnaires or 24-h dietary recall (ideally repeated multiple times).
Curr Hypertens Rep (2015)7:4
The main advantages of dietary methods over urinary methods include convenience, the ability to carry out repeated measurements easily and the ability to identify key dietary sources of excess sodium. Dietary methods are limited by recall bias, imprecision in estimating portion sizes, variations in the sodium content of food items (e.g. sodium content of a slice of bread may vary widely) and lack of information on sodium added during cooking or at the table. In addition, dietary methods need to be validated between regions due to significant regional variations in common foods and in the sodium content of food items. Because of these factors, the use of dietary methods tends to add imprecision which in turn underestimates the association between sodium and BP or CVD [22]. Therefore, any method of sodium estimation can underestimate the slope of the association with BP or with CVD, but these methods are not expected to change the direction or shape of the associations. The purpose and design of a study will influence the choice of method of measuring sodium. For example, for small studies that measure within-individual differences over time, prolonged (24 h) urine collections on different days/weeks will maximise individual-level precision. However, such an approach is not necessary for improving the precision of population estimates of sodium consumption, as the largest sources of variation in these are the marked interindividual differences in sodium consumption. This variance can only be reduced by including a large number of individuals and little improvement in precision of group means can be achieved by using prolonged collections (i.e. 24 h collections). Studies exploring the association between sodium intake and clinical outcomes (e.g. cardiovascular events) should focus on enhancing group-level estimates of sodium intake (which is mainly achieved by including a large number of individuals). Further, large numbers of events are required to precisely relate an exposure (e.g. sodium intake) to an outcome (e.g. cardiovascular disease); large studies which accrue several thousands of events are required to be reliable. The methods used in such studies need to be unbiased and practical. A single fasting or non-fasting urine sample is ideally suited for such studies. Moreover, by repeatedly obtaining such measures in a subsample of the study population, statistical adjustments can be made to calculate ‘usual or habitual’ levels of sodium consumption. Finally, since there are fewer refusals and missing data, the simple single measure of urine leads to less bias compared to 24-h collections (as approximately 30 % of approached individuals refuse to participate in research studies that require 24-h urine collections). These considerations suggest that estimates of sodium derived from single fasting urines (like BP or cholesterol) are reasonably reliable and perhaps even superior to more involved measures such as 24-h urine collections for epidemiological studies. It is important to note that to evaluate the impact of an intervention on sodium intake, different approaches may be preferred.
Page 3 of 8 47
Sodium Intake and Cardiovascular Disease Sodium is essential to produce osmotic pressure and maintain water in the extracellular space [1]. This requires a balance between sodium intake and excretion, maintained by the renal response to renin and aldosterone, the sympathetic nervous system and atrial natriuretic peptide production. Taken together, these mechanisms adapt to variations in sodium intake without significant increases in blood pressure. However, some individuals have a significant blood pressure response to moderate changes in sodium intake, known as salt sensitivity [23]. Low sodium intake may activate the reninangiotensin-aldosterone and sympathetic nervous systems and has adverse effects on the lipid profile [24]. However, most mechanistic studies of sodium reduction are small (n< 50) and had short durations of follow-up (median 28 days). The long-term effects of low sodium intake on biomarkers (such as CRP, IL-6, troponin, BNP/ProBNP and uromodulin [25–28]) have not been adequately studied. Current guidelines are primarily based on the association between sodium intake and blood pressure [29•] with the assumption that changing sodium has no other physiologic or clinical effects. INTERSALT reported a weak but significant positive association between mean sodium intake and blood pressure (p=0.0446), but there were four outlier communities (Brazilian tribes, Papua New Guinea and Kenya), which when excluded, resulted in loss of statistical significance (p=0.33) [8]. Subsequently, other studies reported a positive association between sodium intake and blood pressure. Recently, the Prospective Urban Rural Epidemiology (PURE) study, including >100,000 individuals from 18 countries, reported a positive but nonlinear association between sodium intake and blood pressure, with modest effects in non-hypertensive and younger individuals [3•]. Several short-term clinical trials (
PREVENTION OF HYPERTENSION: PUBLIC HEALTH CHALLENGES (P MUNTNER, SECTION EDITOR)
Dietary Sodium and Cardiovascular Disease Andrew Smyth 1,2 & Martin O’Donnell 1,2 & Andrew Mente 1 & Salim Yusuf 1
# Springer Science+Business Media New York 2015
Abstract Although an essential nutrient, higher sodium intake is associated with increasing blood pressure (BP), forming the basis for current population-wide sodium restriction guidelines. While short-term clinical trials have achieved low intake (6 months). Guidelines assume that low sodium intake will reduce BP and reduce cardiovascular disease (CVD), compared to moderate intake. However, current observational evidence suggests a J-shaped association between sodium intake and CVD; the lowest risks observed with 3–5 g/day but higher risk with 5 g/day) and increased risk of CVD. Although lower intake may reduce BP, this may be offset by marked increases in neurohormones and other adverse effects which may paradoxically be adverse. Large randomised clinical trials with sufficient follow-up are required to provide robust data on the long-term effects of sodium reduction on CVD incidence. Until such trials are completed, current This article is part of the Topical Collection on Prevention of Hypertension: Public Health Challenges * Andrew Smyth [email protected] Martin O’Donnell [email protected] Andrew Mente [email protected] Salim Yusuf [email protected] 1
Population Health Research Institute, Hamilton Health Sciences, McMaster University, Hamilton, ON, Canada
2
HRB Clinical Research Facility Galway, NUI, Galway, Ireland
evidence suggests that moderate sodium intake for the general population (3–5 g/day) is likely the optimum range for CVD prevention. Keywords Sodium . Salt intake . Cardiovascular disease . Hypertension . Guidelines
Introduction Sodium is required for normal physiological function, and total body sodium is tightly regulated (via multiple mechanisms) to maintain extracellular sodium concentrations within a narrow range [1]. The main dietary source of sodium is salt (sodium chloride), which accounts for approximately 95 % of daily intake. The majority of the world’s population (∼95 %) currently consume between 3 and 6 g/day of sodium [2, 3•]. Recently, there has been much debate about optimal sodium intake and the evidence linking sodium intake and cardiovascular disease. Current guidelines are based on the following assumptions: (i) any elevation in systolic blood pressure (BP) above 115 mmHg is associated with increasing risk of cardiovascular disease (CVD); (ii) measures of sodium intake are positively associated with elevated BP; (iii) reducing sodium intake will reduce BP irrespective of the level of sodium intake or BP level [4]; (iv) reducing sodium must therefore reduce CVD. However, this chain of events has never been fully demonstrated, and there are substantial data to question these assumptions [5]. Small and short-term clinical trial data show that sodium reduction reduces blood pressure, seen predominantly in those with elevated BP but not so in nonhypertensives [6]. These data have prompted guidelines to recommend that sodium intake should dramatically be reduced in the entire population [7] to less than 2 g/day (i.e. less
47
Page 2 of 8
than half current intake). These recommendations mean that most people will require major dietary change to achieve guideline targets. In this review, we discuss global sodium intake (including measurement issues) and review the evidence linking sodium intake to cardiovascular disease (including blood pressure).
Global Sodium Intake One of the most highly cited estimates of sodium intake comes from the INTERSALT Study, which estimated sodium intake from 24-h urine collections in 52 population samples in 32 countries (n = 10,079) [8]. In order to measure withinindividual variability, 8 % of the sample completed a second 24-h urine collection (3–6 weeks later). INTERSALT showed wide global variations in 24-h sodium excretion ranging from 0.46 g/day (Yanomamo Indians, Brazil) to 6.0 g/day in men and 5.4 g/day in women in Tianjin, northern China [9]. Subsequently, the INTERMAP study used two consecutive 24-h dietary recalls and one 24-h urine collection to estimate sodium intake in Japan, China, UK and USA [10]. They confirmed that the highest mean sodium excretion was found in northern China, followed by Japan, USA and the UK [11, 12]. A recent meta-analysis of cross-sectional studies (187 countries) estimated mean global intake at 3.95 g/day, with significant regional variations [2]. The Prospective Urban Rural Epidemiological Study (PURE Study) is the largest study (n>100,000) to report global variations in sodium intake (628 communities in 18 countries) with a mean intake of 4.9 g/day [3•]. As sodium is a nutrient rather than a food type, intake is embedded within the overall diet pattern; sources include discretionary (added during cooking or at the table) and non-discretionary (processed or pre-prepared foods) [13]. The proportion of discretionary to non-discretionary use varies between regions [14]. Importantly, the regions of the world with the highest non-discretionary use are not the regions with the highest overall intake.
Measurement The method used to measure sodium intake is a key methodological issue, as differences between methods may affect the absolute estimates of sodium intake and make comparisons between studies challenging. Two main approaches can be used to measure sodium intake—estimating sodium intake from dietary questionnaires or measuring urinary sodium excretion. The current reference standard is repeated 24-h urine collection for sodium, as 90–95 % of sodium ingested is thought to be excreted in the urine. Multiple collections are thought to be required to account for day-to-day variations in sodium intake. However, recent data suggest that non-renal mechanisms generate weekly and monthly infradian
Curr Hypertens Rep74: 1 )5 02(
rhythmicity of sodium storage (in the interstitium, especially the skin), independent of daily sodium intake [15, 16•]. These findings also suggest that variations in excretion, transfer of sodium between different compartments in the body, accumulation of sodium in the body and sodium intake itself may influence the association between sodium and cardiovascular disease. A key limitation of 24-h urine collection is the high frequency of incomplete sample collection (due to under collection) that may bias estimates (underestimate intake) or result in the exclusion of participants (e.g. those engaged in manual labour or those who have to travel for work), limiting generalizability. Based on these limitations, even single 24-h urine collections are associated with high rates of incomplete collection (as much as 30 %), and so repeat 24-h collection is impractical in large international epidemiologic studies. In order to address the limitations of 24-h urine collections (single or multiple), various formula-derived estimates, including the Kawasaki formula [17], Tanaka formula [18], INTERSALT formula [19] and Mege formula, were developed and validated against 24-h urine collections [20]. The timing and fasting status of urine samples and the population included are important considerations when choosing a formula. In North America, the Tanaka and INTERSALT formulae applied to non-fasting, random urine samples are associated with the least biased estimates [20]. An international (11 countries) study validated Kawasaki formula-derived estimates from fasting urine samples (ICC 0.71) [21], indicating that this approach is a robust and reliable method of estimating sodium intake in population-based studies. While multiple urine collections (random, fasting or 24-h) in the same individual is likely the best measure of usual (habitual) sodium intake, this too is impractical in large studies. Instead, the approach is to measure the degree of variability in the same individual at several time points in a subset of individuals (e.g. a few percent) in larger studies, and using the degree of correlation between two or measures in the same group of individuals, the estimate of sodium intake is statistically adjusted. This method has been used by several investigators including INTERSALT and PURE with regard to sodium estimates and also for other risk factors such as cholesterol or BP and is a widely accepted method to analyse epidemiologic associations [3•]. A key point is that if any estimate has greater imprecision, the use of this measure tends to underestimate the slope of the association between sodium intake and BP or CVD. However, it cannot change the shape of the association, e.g., a direct association to an inverse association. Further, if the variability between two measures is similar across the entire range of exposures (e.g. sodium or cholesterol), then there is no reason to expect that a linear relationship will be changed to one that is nonlinear or J-shaped. Sodium consumption from the diet may also be estimated using dietary questionnaires, such as food frequency questionnaires or 24-h dietary recall (ideally repeated multiple times).
Curr Hypertens Rep (2015)7:4
The main advantages of dietary methods over urinary methods include convenience, the ability to carry out repeated measurements easily and the ability to identify key dietary sources of excess sodium. Dietary methods are limited by recall bias, imprecision in estimating portion sizes, variations in the sodium content of food items (e.g. sodium content of a slice of bread may vary widely) and lack of information on sodium added during cooking or at the table. In addition, dietary methods need to be validated between regions due to significant regional variations in common foods and in the sodium content of food items. Because of these factors, the use of dietary methods tends to add imprecision which in turn underestimates the association between sodium and BP or CVD [22]. Therefore, any method of sodium estimation can underestimate the slope of the association with BP or with CVD, but these methods are not expected to change the direction or shape of the associations. The purpose and design of a study will influence the choice of method of measuring sodium. For example, for small studies that measure within-individual differences over time, prolonged (24 h) urine collections on different days/weeks will maximise individual-level precision. However, such an approach is not necessary for improving the precision of population estimates of sodium consumption, as the largest sources of variation in these are the marked interindividual differences in sodium consumption. This variance can only be reduced by including a large number of individuals and little improvement in precision of group means can be achieved by using prolonged collections (i.e. 24 h collections). Studies exploring the association between sodium intake and clinical outcomes (e.g. cardiovascular events) should focus on enhancing group-level estimates of sodium intake (which is mainly achieved by including a large number of individuals). Further, large numbers of events are required to precisely relate an exposure (e.g. sodium intake) to an outcome (e.g. cardiovascular disease); large studies which accrue several thousands of events are required to be reliable. The methods used in such studies need to be unbiased and practical. A single fasting or non-fasting urine sample is ideally suited for such studies. Moreover, by repeatedly obtaining such measures in a subsample of the study population, statistical adjustments can be made to calculate ‘usual or habitual’ levels of sodium consumption. Finally, since there are fewer refusals and missing data, the simple single measure of urine leads to less bias compared to 24-h collections (as approximately 30 % of approached individuals refuse to participate in research studies that require 24-h urine collections). These considerations suggest that estimates of sodium derived from single fasting urines (like BP or cholesterol) are reasonably reliable and perhaps even superior to more involved measures such as 24-h urine collections for epidemiological studies. It is important to note that to evaluate the impact of an intervention on sodium intake, different approaches may be preferred.
Page 3 of 8 47
Sodium Intake and Cardiovascular Disease Sodium is essential to produce osmotic pressure and maintain water in the extracellular space [1]. This requires a balance between sodium intake and excretion, maintained by the renal response to renin and aldosterone, the sympathetic nervous system and atrial natriuretic peptide production. Taken together, these mechanisms adapt to variations in sodium intake without significant increases in blood pressure. However, some individuals have a significant blood pressure response to moderate changes in sodium intake, known as salt sensitivity [23]. Low sodium intake may activate the reninangiotensin-aldosterone and sympathetic nervous systems and has adverse effects on the lipid profile [24]. However, most mechanistic studies of sodium reduction are small (n< 50) and had short durations of follow-up (median 28 days). The long-term effects of low sodium intake on biomarkers (such as CRP, IL-6, troponin, BNP/ProBNP and uromodulin [25–28]) have not been adequately studied. Current guidelines are primarily based on the association between sodium intake and blood pressure [29•] with the assumption that changing sodium has no other physiologic or clinical effects. INTERSALT reported a weak but significant positive association between mean sodium intake and blood pressure (p=0.0446), but there were four outlier communities (Brazilian tribes, Papua New Guinea and Kenya), which when excluded, resulted in loss of statistical significance (p=0.33) [8]. Subsequently, other studies reported a positive association between sodium intake and blood pressure. Recently, the Prospective Urban Rural Epidemiology (PURE) study, including >100,000 individuals from 18 countries, reported a positive but nonlinear association between sodium intake and blood pressure, with modest effects in non-hypertensive and younger individuals [3•]. Several short-term clinical trials (
