This Article Abstract Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Alderman, M. H. Articles by Laragh, J. H. Search for Related Content PubMed PubMed Citation Articles by Alderman, M. H. Articles by Laragh, J. H.

(Hypertension. 1995;25:1144-1152.)
© 1995 American Heart Association, Inc.

Articles

Low Urinary Sodium Is Associated With Greater Risk of Myocardial Infarction Among Treated Hypertensive Men

Presented in part at The American Society of Hypertension Seventh Meeting, New York, NY, May 6-10, 1992.

Michael H. Alderman; Shantha Madhavan; Hillel Cohen; Jean E. Sealey; John H. Laragh

From the Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine (M.H.A., S.M., H.C.), Bronx, and the Cardiovascular Center, Department of Medicine, Cornell University Medical College (J.E.S., J.H.L.), New York, NY.

	Abstract

Top Abstract Introduction Methods Results Discussion References

 
Abstract A sodium-reduced diet is frequently recommended for hypertensive individuals. To determine the relationship of sodium intake to subsequent cardiovascular disease, we assessed the experience of participants in a worksite-based cohort of hypertensive subjects. The 24-hour urinary excretion of sodium (U_NaV), potassium, creatinine, and plasma renin activity was measured in 2937 mildly and moderately hypertensive subjects who were unmedicated for at least 3-4 weeks. Morbidity and mortality in these systematically treated subjects were ascertained. Men and women were stratified according to sex-specific quartiles of U_NaV. Subjects in these strata were similar in race, cardiovascular status, and pretreatment and intreatment blood pressure. Subjects with lower U_NaV were thinner, excreted less potassium, and had higher plasma renin activity. During an average 3.8 years of follow-up, a total of 55 myocardial infarctions occurred. Myocardial infarction and U_NaV were inversely associated in the total population and in men but not in women, who sustained only nine events. In men, age- and race-adjusted myocardial infarction incidence in the lowest versus highest U_NaV quartile was 11.5 versus 2.5 (relative risk, 4.3, 95% confidence interval, 1.7-10.6). No association was observed between non–cardiovascular disease mortality (n=11) and U_NaV. There was a significant linear trend in proportions of myocardial infarction by U_NaV quartile, with a break point after the lowest U_NaV quartile. In the Cox multivariate analysis, log plasma renin activity, age, systolic pressure, and cholesterol as continuous variables as well as left ventricular hypertrophy and smoking had a direct association, and U_NaV (P=.036) had an inverse, independent association with the incidence of myocardial infarction among these treated hypertensive men.

Key Words: sodium • renin • myocardial infarction • hypertension • blood pressure

	Introduction

Top Abstract Introduction Methods Results Discussion References

 
Because sodium reduction can lower blood pressure (BP), at least in some hypertensive individuals,¹ and because the reduction of BP is beneficial, it has been assumed that a diet low in sodium would also reduce subsequent cardiovascular disease (CVD). As a result, a sodium-reduced diet is widely recommended, particularly for hypertensive patients, as both a primary and an adjunctive treatment.^{2 3 4}

Although much attention has been paid to the relation of sodium intake to fluid balance, blood volume, BP, renal function, plasma norepinephrine, plasma renin activity (PRA), cholesterol, and cardiovascular responsiveness,^{5 6 7 8 9} there is no information about the association of different levels of sodium intake with morbidity and mortality.

In a previous study,⁹ to determine the relation of the renin-sodium profile to subsequent cardiovascular and overall health outcomes in treated hypertensive patients, we measured PRA and 24-hour excretion of urinary sodium (U_NaV) in patients after at least 3 to 4 weeks without medication and before initiation or recommencement of antihypertensive drug therapy and then related these values to cardiovascular disease experience. Those with a high renin-sodium profile were most likely to sustain a myocardial infarction (MI) and had the lowest U_NaV.

In the present study, we examined the relation of U_NaV alone to subsequent morbidity and mortality in 2937 treated hypertensive subjects. We now report that in the group as a whole and particularly in men, baseline U_NaV was inversely related to subsequent MI, cardiovascular morbidity and mortality, and all-cause mortality but not to non-CVD mortality.

	Methods

Top Abstract Introduction Methods Results Discussion References

 
Subjects
The study design and methods are identical to those applied previously in the same cohort.⁹ Subjects were sequential entrants to a previously described union-sponsored systematic hypertension treatment program who entered care between January 1981 and June 1990.^{10 11} Entry BP criteria were an untreated systolic BP greater than or equal to 160 mm Hg or a diastolic BP greater than or equal to 90 mm Hg or under treatment with antihypertensive medication at the time of screening. Of these 3067 subjects, 1773 were previously treated and discontinued antihypertensive therapy for 3 to 4 weeks before providing a 24-hour urine collection for sodium determination.¹² Excluding 130 subjects whose follow-up was less than 2 months, the remaining 2937 (96%) formed the study cohort. Baseline information included demographic data, personal and parent or sibling (less than 60 years of age) history of MI and stroke, cigarette smoking habits, physical examination by a nurse and physician, electrocardiographic (ECG) findings (as recorded by program physicians), routine clinical chemistry, PRA, urine protein and electrolytes, and pretreatment BP. At each annual reexamination, an intervening history was obtained. All clinical data were obtained and treatment decisions made without knowledge of U_NaV or PRA values. The study was carried out according to a protocol approved by the institutional review committee.

At first, treatment generally began with either hydrochlorothiazide or propranolol.¹³ By the mid-1980s, - and/or ß-adrenergic blockers, calcium channel blockers, and angiotensin-converting enzyme (ACE) inhibitors were also used.

BP Measurements
All BP readings were taken by a registered nurse using a standard sphygmomanometer. An average of the second and third seated readings in a set of three was the BP for the day. The "final BP" was the average of all BP readings obtained during the 6 months before the last visit or preceding a morbid or mortal event.

Laboratory Methods
Subjects were advised to maintain their usual diet, while avoiding foods excessively high in salt, for the 4 or 5 days preceding their 24-hour urine collection. This same advice was standard to this program and therefore provided to all subjects throughout the course of treatment. Subjects were given plastic containers and instructed to discard the first void on the first day and include it on the second day. Sodium and potassium were measured by a flame photometer (Instrumentation Laboratories) until 1985 and thereafter by an ion-selective electrode (Beckman Astra). The two methods produced comparable results. For creatinine, the Jaffe rate calorimetry method (Beckman Astra) was used.

PRA was measured by an enzyme kinetic assay followed by radioimmunoassay of angiotensin I.¹⁴ Results are expressed as nanograms angiotensin I per liter per second. Blood samples were processed at room temperature at the worksite. Plasma was then frozen. PRA was later measured after rapid thawing of plasma to room temperature in samples that were thawed for the first time. In this way, inadvertent cryoactivation of plasma prorenin was avoided.

Assessment of Adequacy of Urine Collection
Observed creatinine clearance, which depends on urine collection, was compared with the estimated creatinine clearance, computed by applying the Cockcroft and Gault formula.¹⁵ The calculation of this formula is based on serum creatinine, age, and body weight. A previous comparison of this calculation with creatinine clearance actually measured in a metabolic ward revealed that 94% of the values estimated by this formula fell within ±35% of observed creatinine clearance.¹⁶ Therefore, as a means to validate the findings in the entire group, we applied this formula to isolate a subgroup whose urine collection met arbitrary standards.

Morbidity and Mortality
Morbid and mortal events were assessed through review of hospital charts and death certificates by blinded (without access to patient clinic charts) physicians. In cases without such confirmation, physicians outside the program, family members, friends, or union records were consulted.

Morbid and mortal events were classified according to the International Classification of Disease, Ninth Revision, Clinical Modification. CVD events included MI (code 410) and cerebrovascular disease (codes 430 to 434 and 436 to 438, henceforth referred to as strokes). All deaths were categorized as either CVD or non-CVD. For subjects with more than one event during follow-up, only the first CVD since initiation of therapy was included in the present analysis. Data were recorded on computer-compatible forms and processed for computer storage and analysis.

During 3.5 years of median follow-up (range, 0.2 to 9.5 years), 282 events (221 morbid and 61 mortal) occurred. Of these, there were 117 CVD events (55 MI, 23 stroke, 8 coronary [angioplasty/surgery] revascularization, 9 unstable angina, 6 congestive heart failure, and 16 other cardiovascular deaths) and 165 non-CVD events. Hospital records, death certificate, or both confirmed 46 of 55 (84%) MIs, 15 of 23 (65%) strokes, and 15 of 16 (94%) other CVD deaths.

Statistical Analysis
Baseline characteristics were assessed according to sex- specific quartiles of 24-hour U_NaV by univariate analysis. Median values were computed for those with skewed distributions. Cause-specific event rates (per 1000 person-years) were computed by sex-specific quartile of U_NaV level for the total population, and MI rates were adjusted for age and race calculated separately for men and women for each sodium stratum. Relative risk (RR) and 95% confidence interval (CI) between the lowest and the other sodium groups determined the relationship of U_NaV and the incidence of CVD. Similar analyses were done after the study population had been stratified by demographic characteristics, selected cardiovascular risk factors, and clinical chemistry measures at entry. Further analysis included testing for a linear trend in proportions of MI incidence by quartile of U_NaV.¹⁷ The Cox proportional hazards regression model^{18 19} was used to determine the effect of U_NaV on MI and CVD while simultaneously controlling for potential confounders and other covariates. Also, Cox models with and without interaction were used and compared by estimating the difference in their log likelihood (L₁ and L₂) and testing the significance of 2(L₂-L₁) after a ² distribution with 1 df. To further validate these results, analyses were also performed with data on the 2016 (69%) study subjects who met Cockcroft and Gault criteria for adequacy of 24-hour urine collection.¹⁵ All clinical chemistry measures are reported in SI (Système International) units, using conversion factors. All statistical analyses were performed with SPSS and BMDP software.

	Results

Top Abstract Introduction Methods Results Discussion References

 
Subject Characteristics
Men (N=1900) and women (N=1037) differed in most baseline characteristics, including U_NaV values as shown in Table 1. Median U_NaV was 115 mmol/d overall, 126 for men and 97 for women. As U_NaV levels for men and women were significantly different, sex-specific quartiles were determined. Among men and women (Tables 2 and 3), all four quartiles contained similar proportions of subjects who were white, smoked, and had prior CVD, positive immediate family history of CVD, prior treatment, and left ventricular hypertrophy (LVH) by ECG. Body weight, urine volume, and urinary potassium were positively related to U_NaV, whereas age was inversely related to U_NaV for both sexes. During the study, body weight increased similarly in each of the four groups, by 2 to 3 lb for men and 3 to 5 lb for women (initial ranges: men, 175 to 193 lb; women, 153 to 173 lb; final ranges: men, 177 to 195 lb; women, 157 to 178 lb). Serum cholesterol was inversely related to U_NaV in women. Blood urea nitrogen in men was similar in the first three quartiles and significantly (P<.005) higher in the highest quartile. Median PRA was inversely related to U_NaV in men and women.

View this table: [in this window] [in a new window]   Table 1. Baseline Demographic and Clinical Characteristics of Subjects With Hypertension According to Sex

View this table: [in this window] [in a new window]   Table 2. Baseline Demographics and Clinical Characteristics of Men With Hypertension According to Quartile of 24-Hour Urinary Sodium Excretion

View this table: [in this window] [in a new window]   Table 3. Baseline Demographics and Clinical Characteristics of Women With Hypertension According to Quartile of 24-Hour Urinary Sodium Excretion

Hemoglobin, hematocrit, and serum potassium were similar across sodium groups and did not change during follow-up. Proportions of smokers and of subjects with LVH by ECG as well as mean change in cholesterol levels were not significantly different between the sodium quartiles at the end of follow-up. Initial mean systolic BP was similar in sodium quartiles; in the highest sodium group, mean diastolic BP values were 1 and 2 mm Hg higher than in the lowest sodium group in men (99 versus 98 mm Hg) and women (95 versus 93 mm Hg), respectively. Treatment brought nearly 80% of all subjects to the normal BP range (mean: 139.4/87.6 mm Hg) and eliminated interquartile differences except in men whose mean systolic BP in the lowest sodium quartile (139 mm Hg) was significantly higher (P<.05) than that in the highest (137 mm Hg).

U_NaV and Incidence of MI, Stroke, and CVD
The median length of follow-up of subjects in all four sodium strata was similar (3.43, 3.76, 3.34, and 3.45 years). Of the 117 CVD events in both sexes, 96 (82%) occurred in men (Table 4). Of these 96, 46 (48%) were MIs and 17 (18%) were strokes. In Table 4, unadjusted overall incidence rates of MI per 1000 person-years were inversely related to U_NaV for the total population, ranging from 8.1 (lowest quartile) to 2.9 (highest quartile) with an RR of 2.8 (95% CI=1.3-6.1). This was not true for stroke (RR=1.2, 95% CI=0.3-4.0), non-CVD morbidity (RR=1.2, 95% CI=0.8-1.8), or non-CVD deaths (RR=0.8, 95% CI=0.2-3.0). A similar but stronger relationship of MI and U_NaV was observed for men, with an RR of 5.2 (95% CI=2.0-13.5) for the lowest versus highest quartile. The observed MI rates of women (nine events) were in the opposite direction, with no statistical significance between quartiles and a power of 13%.

View this table: [in this window] [in a new window]   Table 4. Unadjusted Cause-Specific Incidence Rates per 1000 Person-Years in Men and Women by Quartile of 24-Hour Urinary Sodium Excretion

In view of these differences and the paucity of events in women, further analysis focused on men. For men, in the middle two U_NaV quartiles, which span the range of usual sodium intake (89 to 174 mmol/24 h), the incidences of MI, stroke, and total CVD were indistinguishable. The lowest quartile of U_NaV differed significantly in MI, CVD, and all-cause mortality from all others, as well as from the two middle groups. By contrast, the highest sodium quartile differed in MI or CVD from the lowest but not from the two middle quartiles. These results suggest a break point at the lowest sodium level.

In Table 5, age- and race-adjusted MI rates among men reflected a clustering of events in the lowest U_NaV quartile that differed significantly from the referent (fourth) quartile as well as the upper three quartiles combined.

View this table: [in this window] [in a new window]   Table 5. Incidence of Myocardial Infarction per 1000 Person-Years in Men With Hypertension According to Quartile of 24-Hour Urinary Sodium Excretion at Baseline

Stratified analysis revealed that even after adjustment for individual risk factors such as age, race, and LVH, the relationship of U_NaV to MI generally persisted but varied according to particular demographic or clinical characteristics (Fig 1). The relationship of U_NaV to MI expressed as the RR of the lowest quartile to the upper three combined (referent) was stronger in younger (4.5, P=.001) than older (1.2, P=.665) subjects. There were very few MIs in blacks, who in this population had an average U_NaV similar to that of whites. However, 5 of the 7 (71%) events that occurred in blacks were in the lowest sodium group compared with 13 of 28 (46%) in whites, although these distributions, reflecting small numbers, did not differ significantly. In the lowest U_NaV group, MI rate was four times higher among subjects with LVH initially than among those without LVH (35.6 versus 9.4 per 1000 person-years). Of note was the lack of evidence of protection against MI associated with increased urinary potassium. Moreover, the lowest sodium-to-potassium ratio was associated with more MIs than were those with the highest ratios.

View larger version (39K): [in this window] [in a new window]   Figure 1. Bar graphs show incidence of myocardial infarction according to quartile of urinary sodium excretion and age, race, and left ventricular hypertrophy (LVH) in men.

Subjects with a higher PRA were more likely to experience an event than those with lower levels. When subjects were divided according to tertile of PRA to represent high, normal, and low, those with U_NaV lower than the median had both higher PRA and a higher rate of MI within each PRA tertile. The inverse relationship of U_NaV to MI persisted regardless of urine volume.

Test for Linear Trend in the U_NaV-MI Relationship
Further analysis included fitting a linear regression between quartiles of U_NaV and proportions of MI events in each sodium stratum for men and testing for a linear trend (Fig 2). The slope (b=-0.0112) was roughly 3.5 times its standard error of 0.0031, showing that the linear trend was significant at P.01. Deviations from linearity were not significant (²=2.097, P>.05 at 2 df). The best break point for the relationship of U_NaV to MI was after the first sodium quartile, based on the proportions in quartile I compared with those in II through IV (²=10.6, P=.001) and among II within IV (²=3.02, P=.22).

View larger version (12K): [in this window] [in a new window]   Figure 2. Line graphs show linear trend in the relationship of urinary sodium excretion and myocardial infarction among men with hypertension. indicates observed; +, estimated.

Validation of Event Rates in Defined Subgroups of Study Subjects
Since estimation of sodium intake depended on a single 24-hour urine collection, various attempts were made to establish its accuracy. Assessment of only those 1298 men whose estimated urinary creatinine clearance values by the method of Cockcroft and Gault¹⁵ were within 35% of the measured values yielded a result similar to that of the whole group. An inverse relationship between MI and U_NaV (I: 11.7, II: 6.1, III: 8.4, and IV: 1.8/1000 person-years; I versus IV: RR=6.5, 95% CI=1.8-23.3) was observed. It should be noted that application of the Cockcroft and Gault formula resulted in the removal of subjects from all four U_NaV quartiles (35%, 34%, 30%, and 28% from I through IV, respectively).

Data on a different subgroup (n=1695) of subjects with a 24-hour U_NaV of 35 to 240 mmol produced a similar inverse sodium-to-MI association.

Drug Use
Physicians were unaware of U_NaV levels at the time of first prescription. The two principal initial drugs, diuretics and ß-blockers, were prescribed for 42% and 24% of all subjects, respectively. ACE inhibitors, calcium channel blockers, and -blockers were also prescribed as first-line drugs in recent years. Initial and final drug use were similar within each U_NaV stratum. Overall, MI and CVD events were also similar by drug use. Finally, when diuretic use was included in multivariate analyses, it did not remain in the model nor did it alter the significant associations of other factors identified in the original analysis.

Multivariate Analyses
The risk of MI was examined in men by multivariate analysis, with MI as the dependent variable and U_NaV along with other factors known to influence the incidence of MI as the independent variables. U_NaV, age, systolic BP, log PRA, urinary potassium, and cholesterol as continuous variables, along with smoking status and LVH as categorical variables, were each found to be independently associated with the incidence of MI in the best-fitting model (Table 6). Of these associations, a negative direction was observed for U_NaV only. Race, diastolic BP, body weight, fasting blood sugar, urine volume, serum creatinine, creatinine clearance, prior treatment status, and history of CVD did not remain in the model. The exclusion of diastolic BP from the model may be due to the collinearity of the two BP measures. In an alternate model in which PRA was not included, U_NaV was more significantly associated with MI than when log PRA was included in the model as in Table 6 (P=.026 vs .036). Because of the known relationship of PRA and U_NaV, the product of U_NaV and log PRA as an interaction term was also introduced in another model. Although associated with MI, its presence did not significantly improve the likelihood of the model without it (-270.08 versus -269.95). Similar results were observed in an analysis using total CVD in place of MI as the outcome variable.

View this table: [in this window] [in a new window]   Table 6. Variables Associated With Risk of Myocardial Infarction in Men With Hypertension as Estimated by Cox Proportional Hazards Regression: Best-Fitting Model

	Discussion

Top Abstract Introduction Methods Results Discussion References

 
The principal finding of this prospective cohort study is that U_NaV was inversely associated with MI and total CVD in treated hypertensive subjects. Perhaps as important was the finding that subjects consuming the highest amount of sodium did not experience more MI events than did those consuming the model sodium diet. This relationship existed whether U_NaV was treated as a categorical or continuous variable and was independent of preexisting disease or differences in conventional risk characteristics, PRA, and renal function. Although this association was demonstrable in men and in the population as a whole, an inverse tendency was observed among women. This was not significant and was based on a small number of events (nine), hence low power (13%), and may have been due to chance. However, it may also represent a real difference. Indeed, differences in the impact of CVD risk factors such as fasting blood glucose and serum cholesterol between men and women have been noted elsewhere.^{20 21 22}

We could find no previous reports relating cardiovascular morbidity or all-cause mortality to U_NaV. Short-term experimental studies that relate sodium intake to changes in BP have not been designed to address the ultimate health effect.²³ Epidemiological studies have revealed that in communities characterized by very low sodium diets, life spans tend to be short.^{24 25 26} Concern about low sodium diets was first raised by Meneely et al²⁷ and Dahl and Love,²⁸ who linked decreased salt intake to growth failure and increased mortality in rats. By contrast, a high sodium diet increased stroke but not MI among rats.²⁹

In an observational study, there must be a concern for undetected confounding, because it is impossible to entirely exclude the prospect that the observed associations are the result of such confounding. Presumably, analysis of other data will either confirm or refute the present findings. A strength of this particular study is that systematic determination of known relevant factors in all subjects was performed at entry. Indeed, this investigation is another in a series of studies with the common purpose of identifying, in hypertensive individuals, entry characteristics that are associated with subsequent cardiovascular events.^{9 30 31 32 33 34} The study cohort was stable and had long-term systematic treatment with virtually complete end point ascertainment.

One possible explanation for these findings is that sicker subjects, or those with a poor outlook based on history, chose to eat less sodium. Data are available to examine this possibility. In fact, those subjects with preexisting disease (prior CVD, renal insufficiency, abnormal ECG, elevated creatinine, and LVH) did not concentrate in the lowest U_NaV group, nor did that quartile include more subjects who had a family history of premature CVD or had themselves received previous antihypertensive treatment. Although the risk of heart attack was indeed greatest among subjects with LVH, the prevalence of LVH was no greater in low than in high sodium subjects, and the association of low U_NaV and MI persisted whether this marker of disease severity was present or absent. A relationship between U_NaV and LVH measured by echocardiography has been previously reported,^{4 35} but correlations between LVH determinations by ECG and echocardiography are poor.³⁶ The absence of difference in LVH between sodium quartiles in the present study may reflect the lack of sensitivity of ECG as a measure of LVH. Weight change and BP response during the study were similar in the four groups. Non-CVD mortality was similar across the quartiles. Finally, multivariate analysis confirmed that the relationship of U_NaV to MI was independent of these potentially confounding factors.

It is also possible that those subjects destined to sustain an MI were excessively responsive to widespread public messages to limit sodium intake. This unlikely outcome would be in sharp contrast to the usual case, in which compliant individuals tend to have the best outcomes.³⁷ Again, no evidence exists to suggest that these subjects were at higher risk based on personal status or family history.

The association observed here depends on the measure of sodium in a single baseline 24-hour urine collection. The initial advice to avoid high sodium foods before urine collection may have influenced diet. Neither subjects nor their nurses or doctors were apprised of urine sodium findings. Throughout the study, all subjects were advised to maintain that same diet. A major alteration in diet was neither sought nor desired. It is reassuring to note that the mean U_NaV of the study subjects (126 mmol/24 h) was in the middle of the range (104 to 151 mmol/24 h) of US participants in INTERSALT, an epidemiological study in which a single 24-hour urine sodium measurement was linked to BP.³⁸

In the present study, as was the case when the relation of BP to stroke and MI or the relation of potassium intake to stroke was first examined, the relation of U_NaV to MI has been explored in a cohort study. A single baseline measure of the item of interest was made and linked to outcomes occurring many years later. In these circumstances, it is not possible to know the stability of the measure. However, some data support the contention that dietary intake, including electrolytes, tends to remain constant in middle-aged people over many years.³⁹ Moreover, in the present study, a contemporaneous normotensive control group drawn from the same population and receiving the same dietary counsel delivered a remarkably consistent U_NaV. In fact, annual 24-hour urine specimens of these normotensive subjects revealed a change (fall) in mean U_NaV of about 1 mmol/y over a 10-year period.⁴⁰

Because it has been reported previously that increased potassium intake measured once at baseline was associated with a protection against stroke that persisted over an entire 8-year follow-up period,⁴¹ we explored the role of potassium in this cohort. The earlier study, based on a single 24-hour dietary recall at baseline, did not report sodium intake or MI incidence. Although stroke incidence was too low to reliably test for potassium effect in the present study, a trend of reduced stroke with increasing potassium excretion was observed.

Since urine collections are notoriously inaccurate, unbiased subsets of individuals whose urinary excretion met some arbitrary objective criteria for accuracy were identified for further analysis.⁴² One approach was to select subjects whose urine volume or U_NaV were within specific ranges. In these subgroups, similar to the entire group in all relevant characteristics, the U_NaV-to-MI relationship held. More convincing, we believe, was the application of the formula developed by Cockcroft and Gault, and later validated empirically, to test for completeness of collection.^{15 16} Creatinine clearance was estimated from serum creatinine, age, and weight. When only subjects whose predicted urinary creatinine clearance was within 35% of that measured were assessed, the U_NaV-to-MI relationship was still demonstrable. In short, no strategy of subject inclusion or exclusion altered the strong association between U_NaV and MI.

It should also be noted that inaccuracy in urine collection was not simply caused by a tendency for subjects to deliver an incomplete collection. Many specimens were excluded because of excessive collection. More significantly, in terms of the reliability of the exposure data, the percent excluded in each U_NaV quartile was roughly similar and not concentrated in the lowest quartile.

Finally, random variation in eating patterns almost certainly generated specimens not reflective of the subject's usual pattern. The inevitable nondirectional misclassification of subjects in terms of their usual intake produces what has been described as regression dilution bias.⁴³ The tendency of this bias would be to blunt any relationship of exposure to outcome.⁴⁴ Thus, the magnitude of the association between U_NaV and MI found here most likely underestimated the true relationship.

This study was not designed to explain how a low sodium diet might influence MI incidence. Perhaps a diet low in sodium was also low in other nutrients. Calcium and iron, as well as potassium and virtually every other measured nutrient, correlate well with sodium intake and are therefore liable to be reduced in people on low sodium diets.⁴⁵ It is also possible that a low sodium intake, particularly in treated hypertensive individuals, decreased blood volume, increased blood viscosity, and in the face of some atherosclerosis, reduced coronary blood flow.^{46 47}

The hypothesis that stimulated this analysis was based on the previously reported association of PRA and MI in hypertensive patients.⁹ PRA is inversely related to sodium intake. We have here extended previous observations to show that PRA, as a continuous variable, is directly related to MI in hypertensive subjects. At the same time U_NaV, as a continuous variable, is inversely related to MI. When PRA was excluded from the Cox multiple regression model, the already significant sodium-to-MI link became substantially stronger. With the inclusion of PRA, both remained in the model, reflecting their independent association with MI. Taken together, these data suggest that much but not all of the association between reduced sodium intake and increased MI may be mediated through effects on the renin-angiotensin system.

Inferences from this study must be limited by its circumstances. Participants were healthy and successfully treated hypertensive subjects in whom any potential hypotensive effect of a low sodium intake was masked by more potent drug therapy. It is certainly possible that in untreated subjects, a low sodium diet might, by lowering BP, be beneficial. In this regard, it should be noted that Meade et al⁴⁸ failed to detect an association between PRA and MI or sudden death from coronary causes among mostly normotensive men. This further cautions against extrapolation of the findings in these treated hypertensive subjects to normotensive individuals.

To our knowledge, this is the first clinical study directly linking U_NaV to subsequent morbidity and mortality. These findings conflict with widely held popular belief. However, they are consistent with the biologically plausible hypothesis that, because sodium and renin are inversely related, a low sodium diet would be associated with increased MI. Certainly, these results do not justify any therapeutic recommendations. No single observational study can establish causality. The next step must be to examine other appropriate data.

	Acknowledgments

 
This work was supported in part by Specialized Center of Research (SCOR) grant HL-18323 from the National Heart, Lung, and Blood Institute, US Public Health Service.

	Footnotes

 
Reprint requests to Dr Michael H. Alderman, Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine, 1300 Morris Park Ave, Room 1311 Belfer, Bronx, NY 10461.

Received July 7, 1994; first decision August 9, 1994; accepted January 19, 1995.

	References

Top Abstract Introduction Methods Results Discussion References

 

1. Cutler JA, Follmann D, Elliott P, Suh I. An overview of randomized trials of sodium reduction and blood pressure. Hypertension. 1991;17(suppl I):I-27-I-33.
2. Stamler R. Implications of the INTERSALT study. Hypertension. 1991;17(suppl I):I-16-I-20.
3. Weinberger MH, Cohen SJ, Miller JZ, Luft FC, Grim CE, Fineberg NS. Dietary sodium restriction as adjunctive treatment of hypertension. JAMA. 1988;259:2561-2565. [Abstract]
4. Schneider RE, Messerli FH, Garavaglia GE, Nunez BD. Dietary salt intake: a determinant of cardiac involvement in essential hypertension. Circulation. 1988;78:951-956. [Abstract/Free Full Text]
5. McCarron DA, Morris CD, Henry HJ, Stanton JL. Blood pressure and nutrient intake in the United States. Science. 1984;224:1392-1398. [Abstract/Free Full Text]
6. Folkow B, Ely DL. Cardiovascular and sympathetic effects of 240-fold salt intake variations: studies in rats, with comparisons to man. Acta Physiol Scand. 1989;136(suppl 583):89-96.
7. Swales JD. Studies of salt intake in hypertension: what can epidemiology teach us? Am J Hypertens. 1990;3:645-649. [Medline] [Order article via Infotrieve]
8. The Surgeon General's Report on Nutrition and Health. Washington, DC: US Government Printing Office; 1988:1-78. US Dept of Health and Human Services, Public Health Service publication 88-50211.
9. Alderman MH, Madhavan S, Ooi WL, Cohen H, Sealey JE, Laragh JH. Association of the renin-sodium profile with the risk of myocardial infarction in patients with hypertension. N Engl J Med. 1991;324:1098-1104. [Abstract]
10. Alderman MH, Schoenbaum EE. Detection and treatment of hypertension at the work site. N Engl J Med. 1975;293:65-68. [Abstract]
11. Alderman MH, Davis TK. Hypertension control at the work site. J Occup Med. 1976;18:793-796.[Medline] [Order article via Infotrieve]
12. Sealey JE, Campbell GA, Prebisz JJ. Hormone assays. In: Laragh JH, Brenner BM, eds. Hypertension: Pathophysiology, Diagnosis, and Management. New York, NY: Raven Press Publishers; 1990:1443-1460.
13. Joint National Committee. The 1980 Report of the National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med. 1980;140:1280-1285. [Medline] [Order article via Infotrieve]
14. Sealey JE. Plasma renin activity and plasma prorenin assays. Clin Chem. 1991;37:1811-1819. [Abstract/Free Full Text]
15. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31-41. [Medline] [Order article via Infotrieve]
16. Robertshaw M, Lai KN, Swaminathan R. Prediction of creatinine clearance from plasma creatinine: comparison of five formulae. Br J Clin Pharmacol. 1989;28:275-280. [Medline] [Order article via Infotrieve]
17. McDonough JR, Hames CG, Stulb SC, Garrison GE. Coronary heart disease among Negroes and whites in Evans county, Georgia. J Chronic Dis. 1965;18:463-468.
18. Cox DR. Regression models and life-tables. J R Stat Soc B. 1972;34:187-220.
19. Hopkins A. BMDP 2L: survival analysis with covariates—Cox models. In: Dixon WJ, ed. BMDP Statistical Software. Berkeley, Calif: University of California Press; 1985:576-594.
20. Seeman T, Mendes de Leon C, Berkman L, Ostfeld A. Risk factors for coronary heart disease among older men and women: a prospective study of community-dwelling elderly. Am J Epidemiol. 1993;138:1037-1049. [Abstract/Free Full Text]
21. Scheidt-Nave C, Barrett-Connor E, Wingard DL. Sex difference in fasting glycemia as a risk factor for ischemic heart disease death. Am J Epidemiol. 1991;133:565-576. [Abstract/Free Full Text]
22. Weijenberg MP, Feskens EJ, Bowles CH, Kromhout D. Serum total cholesterol and systolic blood pressure as risk factors for mortality from ischemic heart disease among elderly men and women. J Clin Epidemiol. 1994;47:197-205. [Medline] [Order article via Infotrieve]
23. The Trials of Hypertension Prevention Collaborative Research Group. The effects of nonpharmacologic interventions on blood pressure of persons with high normal levels: results of the trials of hypertension prevention, phase I. JAMA. 1992;267:1213-1220. [Abstract]
24. Oliver WJ, Cohen EL, Neel JV. Blood pressure, sodium intake and sodium-related hormones in the Yanomamo Indians, a `no-salt' culture. Circulation. 1975;52:146-151. [Abstract/Free Full Text]
25. Lowenstein FW. Blood pressure in relation to age and sex in the tropics and subtropics. Lancet. 1961;1:389-391.
26. Morse WR, Beh YT. Blood pressure amongst aboriginal ethnic groups of Szechwan province, West China. Lancet. 1937;1:966-967.
27. Meneely GR, Ball COT, Youmans JB. Chronic sodium chloride toxicity: the protective effect of added potassium chloride. Ann Intern Med. 1957;47:263-273.
28. Dahl LK, Love RM. Evidence for relationship between sodium (chloride) intake and human hypertension. Arch Intern Med. 1954;94:525-531.
29. Tobian L. Salt and hypertension: lessons from animal models that relate to human hypertension. Hypertension. 1991;17(suppl I):I-52-I-58.
30. Alderman MH, Melcher L, Drayer DE, Reidenberg MM. Increased excretion of urinary N-acetyl-B-glucosaminidase in essential hypertension and its decline with antihypertensive therapy. N Engl J Med. 1983;309:1213-1217. [Abstract]
31. Alderman MH, Madhavan S, Davis T. Reduction of cardiovascular disease events by worksite hypertension treatment. Hypertension. 1983;5(suppl V):V-138-V-143.
32. Ooi WL, Budner NS, Cohen H, Madhavan S, Alderman MH. Impact of race on treatment response and cardiovascular disease among hypertensives. Hypertension. 1989;14:227-234. [Abstract/Free Full Text]
33. Alderman MH, Ooi WL, Madhavan S, Cohen H. Treatment-induced blood pressure reduction and the risk of myocardial infarction. JAMA. 1989;262:920-924. [Abstract]
34. Madhavan S, Ooi WL, Cohen H, Alderman MH. Relation of pulse pressure and blood pressure reduction to the incidence of myocardial infarction. Hypertension. 1994;23:395-401. [Abstract/Free Full Text]
35. Liebson PR, Grandits G, Prineas R, Dianzumba S, Flack JM, Cutler JA, Grimm R, Stamler J. Echocardiographic correlates of left ventricular structure among 844 mildly hypertensive men and women in the Treatment of Mild Hypertension Study (TOMHS). Circulation. 1993;872:476-486.
36. Lee D, Marantz PR, Devereux RB, Kligfield P, Alderman MH. Left ventricular hypertrophy in black and white hypertensives: standard electrocardiographic criteria overestimate racial differences in prevalence. JAMA. 1992;267:3294-3299. [Abstract]
37. Haynes RB, Taylor DW, Sackett DL. Compliance in Health Care. Baltimore, Md: Johns Hopkins University Press; 1979.
38. INTERSALT Cooperative Research Group. INTERSALT: an international study of electrolyte excretion and blood pressure: results for 24 hour urinary sodium and potassium excretion. Br Med J. 1988;297:319-328.
39. Jensen OM, Wahrendorf J, Rosenquist A, Geser A. The reliability of questionnaire-derived historical dietary information and temporal stability of food habits in individuals. Am J Epidemiol. 1984;120:281-290. [Abstract/Free Full Text]
40. James GD, Sealey JE, Alderman MH, Laragh JH. Year to year stability of urine sodium, potassium, aldosterone and PRA in normotensive men and women. Am J Hypertens. 1993;6(part 2):86A-90A.
41. Khaw KT, Barrett-Connor E. Dietary potassium and stroke-associated mortality. N Engl J Med. 1987;316:235-240. [Abstract]
42. Liu K, Cooper K, McKeever J, McKeever P, Byington R, Soltero R, Stamler R, Gosch F, Stevens E, Stamler J. Assessment of the association between habitual salt intake and high blood pressure: methodological problems. Am J Epidemiol. 1979;110:219-226. [Abstract/Free Full Text]
43. Selmer R. Blood pressure and twenty-year mortality in the city of Bergen, Norway. Am J Epidemiol. 1992;136:428-440. [Abstract/Free Full Text]
44. MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J. Blood pressure, stroke, and coronary heart disease, part 1: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990;335:765-774. [Medline] [Order article via Infotrieve]
45. Engstrom AM, Tobelmann RC. Nutritional consequences of reducing sodium intake. Ann Intern Med. 1983;98(part 2):870-872.
46. Luft FC, Rankin LI, Bloch R, Weyman AE, Willis LR, Murray RH, Grim CE, Weinberger MH. Cardiovascular and humoral responses to extremes of sodium intake in normal black and white men. Circulation. 1979;60:697-706. [Free Full Text]
47. De Simone G, Devereux RB, Chien S, Alderman MH, Atlas SA, Laragh JH. Relation of blood viscosity to demographic and physiologic variables and to cardiovascular risk factors in apparently normal adults. Circulation. 1990;81:107-117. [Abstract/Free Full Text]
48. Meade TW, Cooper JA, Peart WS. Plasma renin activity and ischemic heart disease. N Engl J Med. 1983;329:616-619.[Abstract/Free Full Text]

This article has been cited by other articles:

				D. A. McCarron Dietary sodium and cardiovascular and renal disease risk factors: dark horse or phantom entry? Nephrol. Dial. Transplant., July 1, 2008; 23(7): 2133 - 2137. [Full Text] [PDF]

				N. Tzemos, P. O. Lim, S. Wong, A. D. Struthers, and T. M. MacDonald Adverse Cardiovascular Effects of Acute Salt Loading in Young Normotensive Individuals Hypertension, June 1, 2008; 51(6): 1525 - 1530. [Abstract] [Full Text] [PDF]

				B. D. Dickinson, S. Havas, and for the Council on Science and Public Health, Amer Reducing the Population Burden of Cardiovascular Disease by Reducing Sodium Intake: A Report of the Council on Science and Public Health Arch Intern Med, July 23, 2007; 167(14): 1460 - 1468. [Abstract] [Full Text] [PDF]

				N. R Cook, J. A Cutler, E. Obarzanek, J. E Buring, K. M Rexrode, S. K Kumanyika, L. J Appel, and P. K Whelton Long term effects of dietary sodium reduction on cardiovascular disease outcomes: observational follow-up of the trials of hypertension prevention (TOHP) BMJ, April 28, 2007; 334(7599): 885 - 885. [Abstract] [Full Text] [PDF]

				M. H. Alderman Evidence relating dietary sodium to cardiovascular disease. J. Am. Coll. Nutr., June 1, 2006; 25(3 Suppl): 256S - 261S. [Abstract] [Full Text] [PDF]

				C. L. Laffer, R. J. Bolterman, J. C. Romero, and F. Elijovich Effect of Salt on Isoprostanes in Salt-Sensitive Essential Hypertension Hypertension, March 1, 2006; 47(3): 434 - 440. [Abstract] [Full Text] [PDF]

				P. Meneton, X. Jeunemaitre, H. E. de Wardener, and G. A. Macgregor Links Between Dietary Salt Intake, Renal Salt Handling, Blood Pressure, and Cardiovascular Diseases Physiol Rev, April 1, 2005; 85(2): 679 - 715. [Abstract] [Full Text] [PDF]

				C. Nagata, N. Takatsuka, N. Shimizu, and H. Shimizu Sodium Intake and Risk of Death From Stroke in Japanese Men and Women Stroke, July 1, 2004; 35(7): 1543 - 1547. [Abstract] [Full Text] [PDF]

				V. Franco, S. Oparil, and O. A. Carretero Hypertensive Therapy: Part II Circulation, June 29, 2004; 109(25): 3081 - 3088. [Full Text] [PDF]

				M. H. Alderman Dietary Sodium and Cardiovascular Health in Hypertensive Patients: The Case against Universal Sodium Restriction J. Am. Soc. Nephrol., January 1, 2004; 15(90010): S47 - 50. [Abstract] [Full Text]

				P. Sawathiparnich, L. J. Murphey, S. Kumar, D. E. Vaughan, and N. J. Brown Effect of Combined AT1 Receptor and Aldosterone Receptor Antagonism on Plasminogen Activator Inhibitor-1 J. Clin. Endocrinol. Metab., August 1, 2003; 88(8): 3867 - 3873. [Abstract] [Full Text] [PDF]

				P. August Initial Treatment of Hypertension N. Engl. J. Med., February 13, 2003; 348(7): 610 - 617. [Full Text] [PDF]

				G. A MacGregor, F. J He, I. J Perry, M R Law, N J Wald, L. Hooper, C. Bartlett, G. Davey Smith, and S. Ebrahim Long term effects of advice to reduce dietary salt BMJ, January 25, 2003; 326(7382): 222 - 222. [Full Text]

				L. Hooper, C. Bartlett, G. Davey Smith, and S. Ebrahim Systematic review of long term effects of advice to reduce dietary salt in adults BMJ, September 21, 2002; 325(7365): 628 - 628. [Abstract] [Full Text] [PDF]

				M. H Alderman Salt, blood pressure and health: a cautionary tale Int. J. Epidemiol., April 1, 2002; 31(2): 311 - 316. [Full Text] [PDF]

				P. Elliott and J. Stamler Commentary: Evidence on salt and blood pressure is consistent and persuasive Int. J. Epidemiol., April 1, 2002; 31(2): 316 - 319. [Full Text] [PDF]

				G MacGregor and H. de Wardener Commentary: Salt, blood pressure and health Int. J. Epidemiol., April 1, 2002; 31(2): 320 - 327. [Full Text] [PDF]

				J. He and P. K Whelton Commentary: Salt intake, hypertension and risk of cardiovascular disease: an important public health challenge Int. J. Epidemiol., April 1, 2002; 31(2): 327 - 331. [Full Text] [PDF]

				W. M. Vollmer, F. M. Sacks, J. Ard, L. J. Appel, G. A. Bray, D. G. Simons-Morton, P. R. Conlin, L. P. Svetkey, T. P. Erlinger, T. J. Moore, et al. Effects of Diet and Sodium Intake on Blood Pressure: Subgroup Analysis of the DASH-Sodium Trial Ann Intern Med, December 18, 2001; 135(12): 1019 - 1028. [Abstract] [Full Text] [PDF]