High blood pressure or hypertension is the leading preventable risk factor for cardiovascular diseases and is estimated to account for about 54% of deaths from stroke and 47% of deaths from coronary heart disease in adults worldwide [1]. Furthermore, the risk for cardiovascular disease increases progressively with blood pressure, beginning at pressures as low as 115/75 mmHg [2]. According to recent estimates, roughly one third of U.S. adults have high blood pressure [3], and in 2010, high blood pressure was estimated to be responsible for $156 billion in direct and indirect costs [4]. Results of randomized controlled trials have consistently shown reducing sodium intake reduces average blood pressure [5]–[7]. Conversely, observational studies indicate dietary potassium intake is negatively associated with blood pressure in some studies [8]–[10]; but not others [11]–[14].

Although the exact mechanisms by which sodium and potassium levels affect blood pressure are not well understood, evidence suggest that the altered sodium and potassium homeostasis play a key role in pathogenesis of hypertension [15]. However, population-based studies of the association between blood pressure or hypertension with the sodium-to-potassium ratio are limited by the use of intake estimates based on one 24-hour dietary recall [16], [17]. The lack of adjustment for day-to-day variability increases measurement error and may result in the estimates of the associations being biased toward the null. Furthermore, previous studies did not adjust for diabetes, heavy alcohol use, or physical activity, nor examined potential interactions between intake and age, sex, race/ethnicity or other covariates [16], [17]. In this study, we used 2005–2010 data from the National Health and Nutrition Examination Survey (NHANES) to estimate the associations of blood pressure and hypertension with sodium and potassium intake and their ratio among U.S. adults adjusting for within-person variability in intake. We additionally adjusted for potential confounding variables and tested for interactions between estimated intakes and age, sex, race/ethnicity, and other characteristics.

Compared to normotensive participants, hypertensive participants were older, more likely to be male, non-Hispanic black, former smokers, physically inactive, had higher BMI, and were more likely to have history of cardiovascular disease, diabetes and self-reported chronic kidney disease (Table 1).

Overall, the study participants' average estimated usual intakes of sodium, potassium and sodium-to-potassium ratio were 3,569 mg/d, 2,745 mg/d, and 1.41, respectively, and did not differ by hypertension status (Table 2). However, among certain population subgroups, mean intakes did differ by hypertension status. Mean sodium intake was higher in hypertensive participants vs. those who were not among those participants aged 20–50 years or who were physically active. Mean sodium intake was lower in hypertensive participants vs. those who were not among women, among participants with a BMI<25.0, with diabetes, or who were physically inactive. Mean potassium intake was lower in hypertensive participants vs. those who were not among those participants who were aged ≥51 years or had diabetes, but mean potassium intake was higher in hypertensive participants among those with a BMI≥30, or who were physically active. The mean sodium-to-potassium ratio was higher among hypertensive participants vs. those who were not in both age strata, 20–50 years and ≥51 years (Table 2).

After adjustment for potential confounders, intake of sodium, potassium and their ratio were significantly associated with systolic blood pressure, with an increase of 1.04 mmHg (95% confidence interval (CI), 0.27–1.82) and a decrease of 1.24 mmHg (95% CI, 0.31–2.70) for every 1,000 mg/d increase in sodium and potassium intake, respectively, and an increase of 1.05 mmHg (95% CI, 0.12–1.98) for every 0.5 unit increase in sodium-to-potassium ratio (Table 3). Overall, diastolic blood pressure was not associated with the estimated intakes, with one exception; every 1,000 mg/d increase of potassium intake was associated with 0.75 mmHg (95% CI, 0.22–1.28) decrease in diastolic blood pressure. The relationship between systolic blood pressure and potassium intake differed by age groups (p<0.001 for interaction). The effect of potassium intake on systolic blood pressure (comparing 90^th (3,751 mg/d) to 10^th percentile (1,831 mg/d)) appeared to be stronger among older age groups (Figure 1).

The adjusted average systolic blood pressure among study participants ranged from 118.2 mmHg (95% CI, 117.1–119.3) among those in the lowest quartile of sodium intake to 120.4 mmHg (95% CI, 119.5–121.3) (p = 0.010) among those in the highest (Figure 2A); from 120.3 mmHg (95% CI, 119.4–121.2) among those in the lowest quartile of potassium intake to 118.2 mmHg (95% CI, 117.2–119.6) (p = 0.010) among those in the highest (Figure 2B); and from 118.7 mmHg (95% CI, 117.8–119.6) among those in the lowest quartile of sodium-to-potassium ratio to 119.9 mmHg (95% CI, 119.2–120.6) among those in the highest (p = 0.028) (Figure 2C).

The adjusted ORs comparing prevalence of hypertension among adults in the highest quartile vs. those in the lowest quartiles were 1.40 (95% CI, 1.07–1.83) for sodium intake, 0.72 (95% CI, 0.53–0.97) for potassium intake, and 1.30 (95% CI, 1.05–1.61) for sodium-to-potassium ratio (Table 4). Adjusted ORs were 1.16 (95% CI, 1.02–1.32), 0.81 (95% CI, 0.67–0.98) and 1.26 (95% CI, 1.05–1.51), respectively, for every 1,000 mg/d increase in sodium and potassium intake and every 0.5 unit increase in the sodium-to-potassium ratio. The associations did not differ significantly by age, gender, race/ethnicity, BMI, or education (P>0.06 for all interactions).

In a sensitivity analysis, we used intake data from the 1^st 24-hour dietary recall to examine their association with blood pressure and hypertension, and found that compared with the results using the usual intake, the mean intake of sodium, potassium and their ratio from the 1^st 24-hour dietary recall did not change significantly (Table S1 in File S1), but substantial attenuation occurred in the association between all three intake variables and blood pressure and hypertension (Tables S2 and S3 in File S1).

We repeated analyses including 3,780 NHANES participants who took antihypertensive medications. The results of this analysis generally showed a greater correlation between the three intake variables and blood pressure than did the results of our main analyses in which we excluded participants who took antihypertensive medications (Tables S4 and S5 in File S1).

In this large nationally representative sample of U.S. adults who were not taking anti-hypertensive medications, we found that usual sodium intake and sodium-to-potassium ratio were positively associated and usual potassium intake was negatively associated with blood pressure and hypertension. After adjustment for demographic and other characteristics, the pattern of association was largely consistent across gender, race/ethnicity, education, and BMI categories. However, the effect of potassium intake on the systolic blood pressure appeared to be stronger among the older than among younger participants (p<0.001 for interaction).

Our findings of positive association between sodium intake and blood pressure were consistent with results from animal experiments, epidemiological studies, and clinical trials [8], [29]–[35]. A recent meta-analysis of 34 trials reported that a 75 mmol per 24 hours reduction in urinary sodium excretion (equivalent to sodium reduction of 1760 mg/d) reduced on average −4.18 mmHg (95% CI, −5.18, −3.18) for systolic and −2.06 mmHg (95% CI, −2.67, −1.45) for diastolic blood pressure [36].

Although we found potassium intake to be negatively associated with blood pressure and hypertension, results from previous studies of this relationship have been inconsistent. Our results are difficult to compare with these previous studies because of differences in the characteristics of study participants [8]–[14], or in methods used to assess potassium intake [26]. Of three studies whose results did not show an association between potassium intake and systolic blood pressure [12]–[14], one assessed participants' potassium intake on the basis of a single 24-hour dietary recall [12]; the lack of association may thus possibly be related to measurement error caused by day-to-day variability in potassium intake as suggested by the attenuation of our results when we used only one 24-hour diet recall. The three other studies adjusted for energy intake among other variables [11], [13], [14]. However, had we adjusted for energy intake in our study using the residual method, the estimated decrease in systolic blood pressure associated with a 1,000 mg/d increase in potassium intake would still have been significant (1.86 mmHg (95% CI, 0.92–2.80)) for every 1,000 mg/d increase in potassium intake. The InterSalt study used one 24-hour potassium excretion and found that potassium excretion was negatively correlated with both systolic and diastolic blood pressure in individual subjects after adjustments for confounding variables such as age, gender, BMI, alcohol intake, and sodium excretion [35]. A recent meta-analysis of 21 randomized controlled trials reported that increased 24 hour urinary potassium excretion reduced systolic blood pressure by 5.93 mmHg (95% CI, 1.70, 10.15) and diastolic blood pressure by 3.78 mmHg (95% CI, 1.43, 6.13) based on the mean differences between intervention and control groups [37]. In addition, we found a significant interaction between potassium intake and age on the systolic blood pressure. The stratified analysis suggested that the effect of potassium intake on systolic blood pressure appeared to be stronger among older age groups. However, the possible biologic mechanism of differential effects of potassium intake on blood pressure by age is unknown, and we cannot rule out the possibility that the significance of these findings are by chance. Further study is needed to clarify this association.

Two studies examined the relationship between blood pressure and sodium-to-potassium ratio from dietary recall data [16], [17]. Both studies used intake estimated derived from a single 24-hour recall, and neither adjusted for chronic kidney disease, diabetes, alcohol use, or physical activity. However, the results of both showed a positive associations between subjects' blood pressure and their sodium-to-potassium ratio, but the association was found only among people who consumed <400 mg of dietary calcium intake per day in one study [33]. In these two studies, the magnitudes of the increase in systolic blood pressure per unit change in sodium-to-potassium ratio (2.2 and 2.8 mm Hg) was similar to our finding of a 2.1 mm Hg increase. In the sensitivity analyses, we found no significant difference in the magnitude of association by levels of usual dietary calcium intake (<400 mg/d, 400–800 mg/d and >800 mg/d) (results not shown). The InterSalt study found that 24-hour urinary excretion of sodium-to-potassium ratio was also positively related to blood pressure after adjustments for age, gender, BMI, and alcohol intake [35]. Similar results were found from the Dallas Heart Study using urinary excretion of sodium-to-potassium ratio after adjustments for age, gender, race, diabetes, BMI, total cholesterol, estimated glomerular filtration rate, and urine albumin/creatinine ratio [38].

According to the World Health Organization, the sodium-to-potassium ratio should be ≤1 [39] which is also consistent with the Dietary Guidelines for Americans [40]. In the Dietary Approaches to Stop Hypertension (DASH) diet, which is rich in potassium-containing fruits, vegetables, and low-fat dairy foods, the target for sodium and potassium consumption were 3,000 mg and 4,700 mg (0.6 sodium-to-potassium ratio) for 2,100 kcal diet per day, respectively [41]. This DASH diet was shown to be effective in decreasing blood pressure levels in adults without hypertension and adults with stage I isolated systolic hypertension. Further decreases in average sodium intake (to 2,300 and 1,500 mg/d) and sodium-to-potassium ratio (to 0.5-0.3) provided additional benefits. Results from other studies have shown that a low-sodium/high-potassium diet also decreases blood pressure in adults on antihypertensive medications [42], [43]. In one of these studies, participants' sodium intake was lowered by providing them with sodium-free bread [42]. Given that grain products are a major source of sodium in the American diet [44], the increased availability of low-sodium products coupled with increased availability of low-cost fruits and vegetables may make it easier for U.S. adults to adopt low-sodium/high-potassium diets.

Our study has major strengths. This is, to our knowledge, the first study to assess the association between usual intake, particularly sodium-to-potassium ratio, and blood pressure and hypertension status in large nationally representative sample of U.S. adults. We used a measurement error model to estimate usual sodium, potassium intake and their ratio from two 24-hour dietary recalls accounting for within individual variation in intake. We also examined interactions with important sociodemographic and clinical characteristics.

A number of issues must be taken into account when interpreting our results. First, our reliance was upon results of 24-hour dietary recall to estimate participants' usual sodium and potassium intake. Compared with 24-hour sodium excretion, sodium intake from repeated 24-hour dietary recall interviews in NHANES may under or overestimate actual intake. Sodium intake is highly correlated with energy intake and a previous validation study using the 24-hour dietary recall used in this study suggests energy intake from 24-hour dietary recalls may underestimate energy intake by 11% [45]. However, both sodium and potassium intake estimated from replicate 24-hour dietary recalls correlates significantly with 24-hour excretion and studies suggest data from recalls can provide a valid estimate of associations [46], [47]. Second, our estimation of usual sodium and potassium intakes excluded sodium and potassium from table salt, supplements and antacids, resulting in an underestimate of overall sodium and potassium intake. However, the contributions of these additional components to overall sodium and potassium intake are small (<6% all together). Third, the clinical classification of hypertension among individuals is based on the average of two seated blood pressure measurements, properly measured with well-maintained equipment, at each of two visits to the office or clinic (21). As blood pressure in this study was measured during a single mobile examination visit, some individuals without hypertension may be included among those with hypertension, likely diminishing the associations. Fourth, questions for physical activity differed between the NHANES 2005–2006 and 2007–2010 cycles. To minimize the influence of this inconsistency in physical activity measure, we defined less than 10 minutes of moderate and/or vigorous activity per week as physically inactive. Furthermore, our study was cross-sectional, thus the associations between sodium, potassium and their ratio and blood pressure and hypertension should be interpreted with caution. Finally, we cannot rule out unmeasured confounding factors. For example, there might be unmeasured medical conditions that might explain the association between sodium, potassium intake and their ratio with blood pressure and the hypertension.

Our results indicated that sodium intake was positively associated with systolic blood pressure and hypertension, that potassium intake was negatively associated with both, and that the sodium-to-potassium ratio was positively associated with both. These results support those from randomized controlled trials showing reduced sodium consumption and increased potassium consumption can help prevent hypertension, and hence, cardiovascular disease.