Hypertension is a key modifiable risk factor for cardiovascular disease (CVD), the leading cause of morbidity and mortality in the United States.¹ Studies indicate that excess dietary sodium is related to increased blood pressure,^2–4 while high potassium intake is related to lower blood pressure,⁵ and the sodium-to-potassium ratio may be more strongly associated with blood pressure than either sodium or potassium alone.⁶ However, most studies examining these associations have used self-reported dietary measures to estimate intake, including the National Health and Nutrition Examination Survey (NHANES), which has traditionally used 24-hour dietary recalls. Self-reported dietary intake is limited by participant ability to recall detailed information on foods and beverages, and portion sizes. In addition, food composition databases may not accurately reflect product reformulations, and dietary recalls do not capture sodium intake from salt added at the table.⁷ Dietary recalls may also be subject to differential bias across weight status categories, with more frequent under-reporting of intake among obese persons.⁸

Twenty-four hour urinary electrolyte excretion is not subject to the limitations of dietary measures. When collection is complete, 24-hour urine is recommended as the gold standard for assessing sodium intake.^{9, 10} Twenty-four hour dietary recalls may underestimate average sodium intake by 4%–34% compared to 24-hour urinary excretion.^{11, 12} Potassium intake estimated from dietary recalls may be higher than urinary excretion measures by up to 16%.^{11, 12} Accounting for day-to-day variation in urinary excretion of electrolytes is required for assessment of intake in observational studies of health outcomes.^{11, 13}

In 2014, for the first time, NHANES collected 24-hour urine samples and measured urinary electrolytes among a representative sample of U.S. adults ages 20–69 years. These data allow assessment of the associations of urinary sodium and potassium excretion with blood pressure in the U.S. population, unbiased by self-report of food intake. We hypothesized that higher excreted sodium and sodium-to-potassium ratio would be significantly associated with higher blood pressure and odds of hypertension, and that higher potassium excretion would be inversely associated with blood pressure and odds of hypertension.

Over half of U.S. adults age 20–69 were classified as hypertensive (weighted prevalence 28.2%, 95% confidence interval [CI] 21.6–34.8) or pre-hypertensive (23.1%, 19.5–26.6), combined. Adults with hypertension were older, and had higher BMI, than those with prehypertension or optimal blood pressure (both p<0.01 for trend, Table 1). An addition, a higher proportion of adults with hypertension had a history of CVD, diabetes, or chronic kidney disease (all p<0.01).

Sodium, potassium, and sodium-to-potassium ratio did not differ by hypertension status after adjustment for age, sex, race/ethnicity, and BMI (Table 2). In fully-adjusted models additionally including education, history of CVD, diabetes, chronic kidney disease, smoking status, and physical activity (Figure 1), average SBP was higher across increasing quartiles of sodium excretion (p<0.01 and sodium-to-potassium ratio (p<0.01). Across higher quartiles of potassium excretion, SBP was lower (p<0.01). Associations between estimated usual excretion of sodium, potassium, and their ratio with systolic blood pressure were approximately linear (Figure 2), and tests for non-linearity were not significant.

In fully adjusted linear regression models (Table 3), sodium excretion (per 1000 mg/day higher) was directly associated with SBP (4.58 mmHg, 95% CI 2.64, 6.51), DBP (2.25 mmHg, 95% CI 0.83, 3.67), and mid blood pressure (3.41 mmHg, 95% CI 1.97, 4.86). Potassium excretion (per 1000 mg/day higher) was inversely associated with SBP (−3.72 mmHg, 95% CI −6.01, −1.42) and mid blood pressure (−1.98 mmHg, 95% CI −3.37, −0.59). Molar sodium-to-potassium ratio (per 0.5 unit higher) was directly associated with SBP (1.72 mmHg, 95% CI 0.76, 2.68) and mid blood pressure (1.01 mmHg, 95% CI 0.31, 1.71). In the fully adjusted multivariable logistic model (Table 4), persons in the highest (Q4, 87.5^th percentile) compared to lowest quartile of sodium excretion (Q1, 12.5^th percentile) had 4 times greater odds of having hypertension (OR 4.22, 95% CI 1.36, 13.15).

Results remained robust in sensitivity analyses (eTable 1). Interactions between independent variables (sodium, potassium, or sodium-to-potassium ratio) and other model covariates were not statistically significant (data not shown) after correction for multiple comparisons. Although the interaction of sodium excretion with BMI was not significant, BMI-stratified models suggest the association of sodium and blood pressure was significant among obese but not overweight or healthy weight adults (see eTable 2).

This nationally representative cross-sectional study, using the first-ever collection of 24-hour urine in NHANES to estimate usual excretion, demonstrated a direct association between sodium excretion and multiple measures of blood pressure among US adults. Sodium-to-potassium ratio was also directly associated with SBP, while potassium excretion was inversely associated with SBP. These results are consistent with prior findings of urinary electrolyte excretion and blood pressure in other studies.^25–28

The magnitude of the association between SBP and 24-hour urinary sodium excretion in the present study (4.58 mmHg higher per 1000 mg/d difference in estimated usual excretion), was much greater than previously reported using 24-hour dietary intake data (1.04 mmHg higher per 1000mg/d difference in estimated usual intake) from NHANES 2005–2010.²⁹ Compared to self-reported dietary intake data, urinary sodium excretion does not depend on the accuracy of self-report or food composition tables, and 90–95% of ingested sodium is excreted through urine.² In addition, dietary methods, which can be affected by self-report and nutrient coding errors, often fail to capture salt added in preparation (cooking) or at the table and some may not capture non-food sources of sodium such as supplements and antacids, medications, or tap water.³⁰ Such sources of measurement error in dietary studies may weaken the observed associations.

Similarly, the magnitude of the association between SBP and potassium excretion also appears larger in the present study (−3.72 mmHg lower per 1000mg/d difference in estimated usual excretion) compared to a prior investigation using dietary intake data from NHANES 2005–2010 (−1.24 mmHg lower per 1000mg/d difference in estimated usual intake).²⁹ However, the present study found that sodium excretion was significantly associated with DBP but potassium excretion was not. This may be due to greater within-individual variability in urinary potassium vs. sodium excretion as a measure of intake. Approximately 77–90% of potassium consumed is excreted in urine, compared to 90–95% of sodium,² and the fraction of potassium excreted in urine varies between individuals and may be affected by race and medications.^{2, 31}

Our findings are subject to the following limitations. First, the results are subject to potential selection bias, given that the response rate for the 24-hour urine collection was 50% (75% of 66%). Although sample weights for the 24-hour urine collection were adjusted for non-response, differences between respondents and non-respondents could reduce generalizability. Second, while 24-hour urinary excretion is considered the gold standard for estimated sodium intake, imperfect collection can under- or over-estimate excretion. For example, missed voids may result in under-estimates, or collection beyond 24-hour may result in over-estimates. To minimize this error, NHANES did not analyze samples that were collected for less than 22 hours and the longest collection time analyzed was 27.1 hours; adjusted 24-hour urine volume was used to estimate usual sodium and potassium excretion. In addition, we performed sensitivity analyses restricted to persons with complete collections based on creatinine criteria for completeness,²⁴ and results remained robust. Third, due to the cross-sectional design of NHANES and lack of temporality, causality cannot be inferred from these data. Fourth, 24-hour urine was collected on only up to two days, 3–10 days apart, which could have diminished the observed associations with blood pressure due to random measurement error.¹³ Fifth, although a broad set of covariates were included in regression models, unmeasured confounding by additional factors could also play a role. Sixth, antihypertensive medications containing diuretics can impact sodium and potassium excretion, which could confound results. Among participants who reported taking antihypertensive medications, information was not available regarding medication type, so diuretic use could not be included specifically in multivariable models. However, results remained consistent in sensitivity analyses restricted to persons who reported not taking any antihypertensive medications. Finally, only 32 participants had estimated usual sodium excretion ≤2300 mg/day, so we were unable to examine this group separately to study the sodium and blood pressure relationship among persons meeting the 2015–2020 Dietary Guidelines for Americans recommendation of consuming <2300 mg sodium/day.

In conclusion, based on this cross-sectional analysis, we observed a linear, dose-response association between 24-hour sodium excretion and hypertension among US adults, and an inverse association between potassium excretion and hypertension. These associations were independent of numerous covariates in a large, diverse sample of adults. These cross-sectional findings using gold-standard assessment methods, when added to the evidence base from longitudinal and interventional studies, support dietary advice to lower sodium intake and increase consumption of potassium-containing foods as part of a healthy dietary pattern as recommended in the 2015–2020 Dietary Guidelines for Americans. A dietary pattern high in potassium, such as the Dietary Approaches to Stop Hypertension (DASH) diet which focuses on vegetables, fruits, whole grains, low fat dairy and lean protein, in combination with a low-sodium diet, had greater blood pressure lowering effects than either dietary approach alone.³² Advice from doctors and other health care professionals about sodium reduction is associated with action to reduce sodium intake.³³ The health impact of individual and population-level strategies to increase potassium intake merits further research. Recommended public health strategies for sodium reduction focus on reducing the primary source of sodium in the US diet, salt added to commercially processed and prepared foods.^34–36 Continued monitoring of the related health impact of such strategies is essential given that even small reductions in population sodium intake are projected to prevent thousands of deaths due to heart disease and stroke and save billions of dollars in health care costs annually.³⁷