点击显示 收起
1 From the Department of Public Health and Primary Care, Institute of Public Health, University of Cambridge School of Clinical Medicine, Cambridge, United Kingdom (K-TK, AW, RL, EO, NW, and ND), and the MRC Dunn Human Nutrition Unit, Cambridge, United Kingdom (SB)
2 EPIC-Norfolk is supported by program grants from the Medical Research Council UK and Cancer Research UK and through additional support from the European Union, Stroke Association, British Heart Foundation, Department of Health, and the Wellcome Trust. 3 Reprints not available. Address correspondence to Kay-Tee Khaw, Clinical Gerontology Unit, Box 251, University of Cambridge School of Clinical Medicine, Addenbrooke's Hospital, Cambridge CB2 2QQ United Kingdom. E-mail: kk101{at}medschl.cam.ac.uk.
ABSTRACT
Background: Abundant evidence indicates that a high sodium intake is causally related to high blood pressure, but debate over recommendations to reduce dietary sodium in the general population continues. A key issue is whether differences in usual sodium intake within the range feasible in free-living populations have clinical or public health relevance.
Objective: We examined the relation between blood pressure and urinary sodium as a marker of dietary intake.
Design: This was a study of 23104 community-living adults aged 4579 y.
Results: Mean systolic and diastolic blood pressure increased as the ratio of urinary sodium to creatinine increased (as estimated from a casual urine sample), with differences of 7.2 mm Hg for systolic blood pressure and 3.0 mm Hg for diastolic blood pressure (P < 0.0001) between the top and bottom quintiles. This trend was independent of age, body mass index, urinary potassium:creatinine, and smoking and was consistent by sex and history of hypertension. The prevalence of those with systolic blood pressure 160 mm Hg halved from 12% in the top quintile to 6% in the bottom quintile; the odds ratio for having systolic blood pressure 160 mm Hg was 2.48 (95% CI: 1.90, 3.22) for men and 2.67 (95% CI: 2.08, 3.43) for women in the top compared with the bottom quintile of urinary sodium. Estimated mean sodium intakes in the lowest and highest quintiles were 80 and 220 mmol/d, respectively.
Conclusions: Within the usual range found in a free-living population, differences in urinary sodium, an indicator of dietary sodium intake, are associated with blood pressure differences of clinical and public health relevance. Our findings reinforce recommendations to lower average sodium intakes in the general population.
Key Words: Sodium salt blood pressure population diet
INTRODUCTION
Abundant evidence indicates that high sodium intake is causally related to high blood pressure (1-8). Recent expert committees in the United States from the National High Blood Pressure Education Program and the American Heart Association have recommended limiting daily sodium intake in adults to not more than 100 mmol Na, or 6 g salt (9, 10). In the United Kingdom, the Scientific Advisory Committee on Nutrition in 2003 concluded that "Reducing the average salt intake of the population is likely to decrease the burden of high blood pressure and improve public health" and recommended a reduction in the population mean daily intake from the current average of 150 mmol Na (9 g salt) to 100 mmol Na (6 g salt) (11).
Nevertheless, debate over the relevance of dietary sodium intake to blood pressure in the general population continues. The issue now is not whether high dietary sodium may be etiologically related to high blood pressure, but whether differences in sodium intake within the range feasible and acceptable in the long term in free-living populations have clinical or public health significance. Hooper et al (7) argued that "intensive interventions, unsuited to primary care or population prevention programs, provide only small reductions in blood pressure and sodium excretion and effects on deaths and cardiovascular events are unclear." The 2003 European Guidelines on Cardiovascular Disease Prevention do not mention sodium (12).
There are 2 main lines of argument against sodium reduction to reduce blood pressure in the general population. First, changes in sodium intake achieved with highly motivated participants in trials in controlled circumstances over a relatively short term may not be easily generalizable to the free-living population. Second, only a small subset of the population may be sodium sensitive and such persons may be overrepresented in some trials. Thus, within the realistic feasible range of sodium intake in free-living industrialized populations, the magnitude of effect on blood pressure may be of trivial clinical or public health relevance (7). In the present study, we examined the relation between blood pressure and urinary sodium as a marker of dietary intake in men and women living in the general population.
SUBJECTS AND METHODS
EPIC-Norfolk (the European Prospective Investigation into Cancer in Norfolk) is a prospective population study of 25000 men and women aged 4579 y unselectively recruited from general practice age-sex registers in Norfolk, United Kingdom, who participated in a baseline survey (13). The participants were comparable with national population samples with respect to many characteristics, including anthropometric indexes, blood pressure, and lipids, but had a somewhat lower prevalence of current smokers. The EPIC-Norfolk Study was approved by the Norfolk Health District Ethics Committee.
At the baseline survey between 1993 and 1997, participants completed a detailed health and lifestyle questionnaire. Medical history was ascertained with the question, "Has a doctor ever told you that you have any of the following?", which was followed by a list of conditions including "high blood pressure (hypertension) requiring treatment with drugs." Smoking history was derived from responses to the questions, "Have you ever smoked as much as one cigarette a day for as long as a year?" and "Do you smoke cigarettes now?" Habitual physical activity assessed both work and leisure time activity during the past year, and individuals were allocated to 4 ordered categories of overall activity (14).
The participants attended a clinic where a health examination was carried out by trained nurses. Body mass index was estimated as weight in kilograms/(height in meters)2. Blood pressure was measured by using an Accutorr noninvasive oscillometric blood pressure monitor (Datascope Medical, Huntingdon, United Kingdom) after the participant had been seated for 5 min. The mean of 2 readings was used for analysis.
A casual urine specimen was requested from each participant. They were frozen without preservative at 20°C. In 19982002, the urine samples were thawed and assayed for sodium, potassium, and creatinine concentrations (mmol/L). Urinary electrolyte ratios (sodium:creatinine, potassium:creatinine, and sodium:potassium) were calculated.
To enable estimates of dietary sodium ranges in this population, we also assessed dietary sodium by the use of several independent methods. Participants were asked to complete a 7-d diet diary prospectively. The diaries are being coded and entered into a computer and analyzed for average daily nutrient intake with a specially developed program (15). This method was validated by using weighed-food records, 24-h urine collections checked for completeness by using para-aminobenzoic acid excretion, and blood biomarkers over 1 y in a subset of participants. The correlation between sodium intake estimated from the diaries and 24-h urinary sodium was r = 0.48 (16). Because of the labor-intensive requirements for analysis of the diet diaries, only a sample of 7000 diaries has been coded and analyzed to date for average daily nutrient and food intakes. We estimated the range of dietary sodium and potassium intakes in the population by using the subset for whom 7-d diet diaries had been analyzed. This is almost certainly an underestimate for sodium, however, because the diet analyses do not take into account salt added at the table, and the food-composition data are derived from samples measured without salt during food preparation. Thus, we also obtained an independent estimate by using 24-h urinary sodium and potassium excretion in a subsample of 340 men and women participating in continuing validation and calibration studies who had up to six 24-h urine collections over 1 y (16, 17). We used data from 23104 participants aged 4579 y who attended the health examination and had available results from a spot urine specimen analyzed for sodium, potassium, and creatinine concentrations.
Because we wished to examine the independent relation of sodium alone, we used the ratio of sodium to creatinine in the analyses but adjusted for potassium where appropriate. We categorized the participants by quintiles of urinary sodium:creatinine. We calculated the mean systolic and diastolic blood pressure of men and women by quintile of urinary sodium:creatinine after adjustment for age, body mass index, cigarette smoking, and urinary potassium:creatinine. Differences were tested by using analysis of variance; significance values are shown for the trend test. In persons with no history of hypertension, we examined the percentage of those with newly diagnosed hypertension, arbitrarily defined as systolic blood pressure > 140 mm Hg or 160 mm Hg or diastolic blood pressure > 90 mm Hg or 95 mm Hg by quintile of urinary sodium:creatinine. We also examined the percentage of those with optimal blood pressure, which was defined as systolic blood pressure < 120 mm Hg or diastolic blood pressure < 70 mm Hg.
The regression of blood pressure on urinary sodium:creatinine as a continuous variable, adjusted for age, body mass index, smoking, and urinary potassium:creatinine, was estimated. All data were analyzed by using SPSS for WINDOWS, version 11.5 (SPSS Inc, Chicago).
RESULTS
The characteristics of the population are shown in Table 1. Fourteen percent of this older population reported a history of hypertension.
View this table:
TABLE 1. Characteristics of men and women aged 4579 y in the European Prospective Investigation into CancerNorfolk Cohort, 199319971
The characteristics of the men and women by category of urinary sodium:creatinine are shown in Table 2. Mean body mass index did not differ significantly. However, mean systolic and diastolic blood pressures differed significantly, with a trend for increasing blood pressure with increasing category of sodium:creatinine. Whereas urinary potassium:creatinine was significantly positively related to urinary sodium:creatinine (r = 0.40, P < 0.001), urinary potassium:creatinine was weakly negatively correlated with blood pressure (for SBP: r = 0.06, P < 0.001; for DBP: r = 0.04, P < 0.001, adjusted for sex, age, BMI, and sodium:creatinine).
View this table:
TABLE 2. Characteristics of 10 812 men and 12 922 women aged 4579 y in the European Prospective Investigation into CancerNorfolk Cohort by category of urinary sodium:creatinine, 19931997
Mean systolic and diastolic blood pressure, adjusted for age, smoking, body mass index, and urinary potassium:creatinine, are shown in Table 3. There was a significant trend of increasing mean systolic and diastolic blood pressure with increasing quintile of urinary sodium:creatinine. This trend was consistent by sex, by history of hypertension, and by age group (< and >60 y). The magnitude of the relation was an 7-mm Hg difference in mean systolic blood pressure and an 3.5-mm Hg difference in mean diastolic blood pressure between those in the top and bottom quintiles of urinary sodium:creatinine. The magnitude was greater in those older than 60 y, with a difference of 10 mm Hg in systolic blood pressure between those in the top and bottom quintiles; this difference was similar to the difference in systolic blood pressure observed between th top and bottom quintiles of body mass index (not shown).
View this table:
TABLE 3. Adjusted mean systolic (SBP) and diastolic (DBP) blood pressure and distribution of blood pressure categories by urinary sodium:creatinine for men and women aged 4579 y in the European Prospective Investigation into CancerNorfolk Cohort, 199319971
The odds ratio for individuals with newly diagnosed hypertension, defined by using arbitrary systolic or diastolic threshold criteria, increased with increasing quintile of urinary sodium:creatinine, with a significant 2.5-fold likelihood of systolic blood pressure >160 mm Hg in the top compared with the bottom quintile. The prevalence of those with newly diagnosed hypertension defined by using arbitrary systolic or diastolic threshold criteria also increased with increasing quintile of urinary sodium:creatinine: there was nearly a 2-fold difference between the top and bottom quintiles. Conversely, the percentage of persons with desirable blood pressure levels decreased with increasing urinary sodium:creatinine. The whole distribution of blood pressure shifted upward with increasing urinary sodium:creatinine.
In the multivariate regression analysis, the regression slope of blood pressure on urinary sodium:creatinine (Table 4) was not significantly different in men and women. The regression slope did not differ by history of hypertension but was significantly greater in those aged 60 y than in those aged <60 y. There was no significant difference in the regression slope by body mass index category.
View this table:
TABLE 4. Regression of systolic (SBP) and diastolic (DBP) blood pressure on urinary sodium:creatinine adjusted for age, BMI, smoking, and urinary potassium:creatinine in 10 812 men and 12 922 women aged 4579 y in the European Prospective Investigation into CancerNorfolk Cohort, 19931997
Shown in Table 5 are the estimated mean daily dietary sodium and potassium intakes in the population based on the subset of 7256 individuals who had available 7-d diary data coded for nutrient intake and on the random sample of 340 individuals in the calibration study who had six 24-h urine collections over 1 y. The population estimates made by using 2 independent methods (7-d diaries and 24-h urinary collections) were broadly comparable for dietary potassium. Correlation coefficients for urinary sodium:creatinine and dietary sodium intake estimated from the 7-d diary were r = 0.14 (P < 0.0001) for men and r = 0.11 (P < 0.0001) for women. Those for urinary potassium:creatinine and dietary potassium estimated from 7-d diaries were r = 0.21 (P < 0.0001) for men and r = 0.27 (P < 0.0001) for women. As expected, because added table salt was not included in the 7-d diet diary estimates, dietary sodium intake estimated from diet records was an underestimate compared with 24-h urine collections, particularly in the upper quintiles, whereas there was good agreement across all quintiles for potassium.
View this table:
TABLE 5. Daily sodium and potassium intakes by sex-specific quintiles of intake estimated from 7-d food records and from six 24-h urine collections in subsamples of men and women aged 4579 y in the European Prospective Investigation into CancerNorfolk Cohort, 199319971
Estimated mean sodium excretion was 161 mmol/d for men and 125 mmol/d for women on the basis of 24-h urine collections. We also used a third independent estimate of dietary sodium intake based on the mean sodium concentration in the casual urine samples, which was 96 mmol/L for the 10812 men and 73 mmol/L for the 12922 women. Assuming a mean daily urine volume of 1.6 L (18), the average 24-h intake of 154 mmol for men and 118 mmol for women was close to the estimate derived from the 24-h urine collections.
DISCUSSION
Blood pressure relates to urinary sodium:creatinine
Systolic and diastolic blood pressure related significantly to urinary sodium:creatinine as estimated from a casual urine sample in this population, independently of age, body mass index, cigarette smoking, and urinary potassium:creatinine and consistently in men and women. That such a relation was observed was surprising given the huge measurement error in assessing individual sodium status based on a single casual urine sample. Random errors are likely to attenuate any relation rather than produce significant associations.
Although trials have shown that changing sodium intake changes blood pressure, correlations within a population between sodium intake and blood pressure have been less consistent (4, 19). Possible explanations for the inconsistency are that sodium is not an important determinant of differences in blood pressure between individuals within a population; alternatively, studies may have lacked power to detect an association due to small sample size or large measurement error in estimating dietary sodium intake. Sodium is a particular problem for dietary assessment because of the discretionary salt added in cooking and at the table that is not adequately captured by dietary instruments. The difficulties in assessing sodium intake with dietary instruments, coupled with large intraindividual variation, have led to recommendations that multiple, timed urine collections are required to characterize an individual's intake (20, 21).
Nevertheless, several population studies have reported associations between blood pressure and urinary sodium and potassium estimated from causal urine samples (22-24). These empirical observations may have several explanations. The measurement error incurred by using electrolyte ratios from a casual urine sample and assuming constant creatinine excretion to characterize an individual, although substantial, may be outweighed by volume errors in collecting several timed urine collections, particularly in general populations who may not be highly motivated. It is possible that persons who are hypertensive may be taking medication such as diuretics that may increase sodium excretion. However, the relation of blood pressure with urinary sodium was apparent after excluding those with known hypertension who might be taking such medication. Blood pressure per se may influence electrolyte excretion, although in the steady state it is likely that net electrolyte output would reflect input. Indeed, in this cohort, urinary sodium:creatinine and potassium:creatinine estimated from the single casual urine sample correlated significantly with dietary sodium and potassium intakes estimated from 7-d diaries despite the underestimation of dietary sodium in the dietary assessment. Other work comparing spot urine samples with 24-h urine collections supports the utility of spot urinary electrolyte ratios (25, 26).
It is therefore plausible that the relation between blood pressure and urinary sodium:creatinine in this population reflects a causal relation, although the magnitude of the effect is likely to be attenuated by the large random measurement error. The results in Table 3 suggest that even this attenuated relation shows differences of potential clinical and public health significance. Although other factors, such as obesity or nutrients, most notably potassium, are also important influences on blood pressure (27, 28), we wished to address the issue of sodium independently of potassium. Thus, the analyses were confined to urinary sodium:creatinine rather than urinary sodium:potassium but were adjusted for potassium as appropriate.
Relation apparent in the whole population
The difference in mean systolic blood pressure between the top and bottom quintiles of sodium was 7 mm Hg. This difference is comparable with the differences in blood pressure in this population between the top and bottom quintiles of body mass index, which is well accepted as an important determinant of blood pressure. There was no evidence that this was due to a subset of the population who were sodium sensitive. The whole distribution of blood pressure in the population was shifted upward with increasing urinary sodium such that the prevalence of those who might be defined as hypertensive according to accepted criteria increased, and, conversely, the prevalence of those with the lowest blood pressure decreased. The prevalence of hypertension doubled between the top and bottom quintiles of urinary sodium excretion, and the prevalence of those with optimal or the lowest blood pressure was cut in half. The regression slopes were not significantly different between persons who had a history of hypertension and those who did not. However, sodium sensitivity did appear to increase with age, with a steeper blood pressure rise with increasing sodium in older than in younger persons, an observation that has been noted previously (29). These observational data have effect sizes comparable with those observed in the Dietary Approaches to Stop Hypertension sodium trial (8).
Health implications
Because the risk of adverse health events rises continuously with blood pressure across the whole blood pressure distribution (30, 31), a substantial proportion of the events attributable to higher-than-optimal blood pressure does not occur in the small percentage of persons with very high blood pressure and at very high risk but in the large proportion of persons in the middle of the population distribution who are at moderately increased risk. Conversely, a small reduction in risk of a large number of persons may result in a large reduction in the risk of an entire population (32). Since Geoffrey Rose first expounded these concepts, the population-based approach to risk factor reduction as a complement to treatment of high-risk individuals has gained general currency. A reduction in the population mean systolic blood pressure of 5 mm Hg is estimated to reduce stroke, coronary heart disease, and all-cause mortality by 14%, 9%, and 7%, respectively (9). On an individual basis, our study indicates that free-living individuals have one-half the risk of high blood pressure when sodium intake is 80 compared with 220 mmol/d. Nevertheless, policies and recommendations aimed at reducing average salt intake in the population to reduce overall blood pressure have been debated vigorously (7, 33-35). One recent review concluded that interventions to reduce dietary sodium have trivial effects and do not support the need for public health action (7).
The current data indicate that this criticism appears misplaced. The estimated mean sodium intake in this population based on a random sample who had up to six 24-h urine collections was 120 mmol/d, with a mean in the bottom quintile of 80 mmol and a mean in the top quintile of 220 mmol. The UK Scientific Advisory Committee on Nutrition (11) recommended reducing average salt intake in Britain from the current British average of 9 to 6 g/d, that is, from 150 to 100 mmol Na; recommendations in the United States suggest 100 mmol Na as an upper limit. A daily intake of 100 mmol Na is clearly feasible and achievable, because 40% of the current, free-living population cohort already appeared to have dietary sodium intake within this range. The difference between this group and those in the top quintile for sodium was 7 mm Hg in mean systolic blood pressure with about a doubling of hypertension prevalence.
In the current analysis, we did not address the issue of the relation between dietary sodium and cardiovascular disease endpoints. Nevertheless, other studies have indicated a positive relation of dietary sodium with cardiovascular disease. Tuomilehto et al (36) reported that persons in the top compared with the lowest quintile of 24-h urinary sodium excretion in a Finnish cohort had significantly higher relative risks of 1.51, 1.45, and 1.26 for coronary heart, cardiovascular, and total mortality, respectively. In the follow-up to the first National Health and Nutrition Examination Survey, He et al (37) reported a 61% increase in cardiovascular disease mortality in overweight persons and 39% increase in all-cause mortality associated with 100-mmol higher sodium intake.
Conclusion
Within the range found in a free-living population, even modest and entirely feasible differences in sodium intake are associated with blood pressure differences of clinical and public health relevance. Our findings reinforce current recommendations to lower sodium intake in the general population.
ACKNOWLEDGMENTS
SB, ND, K-TK, and NW are principal investigators in the EPIC-Norfolk population study. RL is responsible for data management, record linkage, and overall computing. AW is responsible for nutritional analyses. EO worked on retrieval and analysis of the urine samples. K-TK conducted the data analyses and wrote the paper with co-authors and is guarantor for this paper. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. None of the study sponsors had any role in study design, collection, analysis and interpretation of data, writing of the report, or the decision to submit the paper for publication. None of the authors had a conflict of interest.
REFERENCES