Sodium, Blood Pressure, and Cardiovascular Disease

In this issue of Hypertension, Naser et al. report values for 24-hour urinary sodium excretion when measured directly and when the Kawasaki, Tanaka, or INTERSALT formulas were used to derive estimates from spot urine sodium measurements¹. Their analysis was performed using pooled data from three longitudinal studies conducted in Bangladesh. Careful attention was paid to collection of the exposure (urinary sodium) and outcome (blood pressure) variables. All the formula-based estimates for 24-hour excretion of sodium provided biased results compared to directly measured 24-hour urinary sodium excretion, with a mean bias (difference between formula estimates and direct collection) of −230, −1035, and −1196 mg/24-hours for the Kawasaki, Tanaka, and INTERSALT formulas, respectively. Compounding these errors, each formula overestimated 24-hour urinary sodium at lower levels and underestimated 24-hour urinary sodium at higher levels of sodium excretion. There was a linear association between directly collected 24-hour urinary sodium and systolic BP, whereas all three formula-derived estimates of 24-hour urinary sodium resulted in a J-shaped association with systolic BP. In an earlier report, based on a median follow-up of 24 years in TOHP I and TOHP II participants, a linear association was noted between directly collected 24-hour urinary sodium and total mortality whereas the association between 24-hour urinary sodium estimated using the Kawasaki formula and total mortality was J-shaped².

The relationship between dietary sodium and blood pressure (BP) as well as cardiovascular disease (CVD) has been studied extensively in observational and experimental studies. The most definitive evidence for a beneficial effect of sodium reduction on CVD would be confirmation in event-based randomized controlled trials. The closest to this “gold standard” test for benefit in a pure sodium reduction intervention is the randomized comparison of long-term follow-up in Trials of Hypertension Prevention (TOHP) participants^{3, 4}. After 10–15 years of follow-up, the adjusted hazard ratio [HR] (95% confidence interval [CI]) for all CVD events was 0.70 (0.53–0.94) albeit follow-up was incomplete for the non-fatal events³. A subsequent intention-to-treat analysis of all participants randomized to sodium reduction compared to control, conducted after a median follow-up of 25.7 (TOPH I) and 22.4 (TOHP II) years, yielded a HR (95% CI) of 0.85 (0.66 to 1.09) for all-cause mortality⁴. A recent meta-analysis that included results from five trials, including TOHP I and II, reported a treatment-related CVD relative risk (95% CI) of 0.72 (0.59–0.89)⁵. The Salt Substitute and Stroke Study (SSaSS), a large cluster-randomized trial being conducted in participants with a history of stroke or a high risk for stroke, is testing the efficacy of a sodium reduction intervention, based on partial replacement of sodium by potassium, for prevention of CVD⁶. The trial is well powered to test the primary outcome (fatal and non-fatal stroke) and it will provide results for a practical means of sodium reduction, albeit not a “pure” sodium reduction intervention.

Given the challenges of conducting definitive event-based randomized controlled trials to test the efficacy of sodium reduction for prevention of CVD, the association between sodium intake and both BP and CVD in observational studies has been studied extensively. Judgements regarding causality for associations in observational studies of sodium and CVD are challenging because residual confounding is hard to eliminate and because reverse causality is a common concern, especially in the studies conducted in participants with comorbid conditions⁷. In addition, validity of the methods used to estimate urinary sodium excretion has been a serious concern. The gold standard for measurement of sodium excretion is assessment using 24-hour urinary collections. Collection of a 24-hour urine is burdensome for participants and accuracy of a 24-hour urinary sodium measurement requires careful attention to detail by investigators. Many investigators have chosen to measure sodium concentration in single “spot” urine collections and then employ one of several formulae to estimate 24-hour urinary sodium excretion. Unfortunately, validation studies have demonstrated that spot urines provide biased estimates of sodium excretion and that values derived from spot urines are typically too high at lower and too low at higher levels of urinary sodium excretion⁸.

In randomized controlled trials, reductions in sodium intake have consistently resulted in BP lowering⁹. In a recent report, Filippini et al. studied the dose-response relationship between sodium excretion and BP in 85 trials with 24-hour urinary sodium measurements and a trial duration of at least four weeks¹⁰. Overall, there was an approximately linear relationship between achieved sodium intake (or change in sodium intake) and both systolic and diastolic BP, with no suggestion of a flattening of the curve either at the highest or lowest levels of sodium excretion. The relationship pattern was similar in those with or without hypertension, albeit the slope was, as expected, steeper for those with hypertension. There was also a suggestion of greater BP lowering in those consuming higher levels of sodium at baseline. A previous meta-analysis conducted in 133 randomized controlled trials identified a larger BP lowering effect of sodium reduction in older and non-white trial participants⁹. BP is generally accepted as one of the best surrogates for CVD, especially stroke and heart failure. A reasonable inference from the effects of dietary sodium on BP is that CVD benefits should accrue. Although some observers have claimed adverse effects of sodium reduction due to activation of the renin-angiotensin-aldosterone (RAAS) system, such effects are modest with typical sodium reduction recommendations and activation of the RAAS is common with drugs such as diuretics and RAAS inhibitors that have been well proven as effective for prevention of CVD.

The elegant report by Naser et al. adds to the evidence that formulae for estimation of 24-hour sodium excretion using spot urine values provide erroneous, biased results that yield misleading associations with health outcomes. They should not be used to report on health associations and the time has come to stop publishing misleading manuscripts based on formulae that employ spot urines to estimate 24-hour sodium excretion. Likewise, calls for public policy decisions based on such misleading reports should be recognized as counterproductive to implementation of the science-based recommendations to reduce dietary sodium consumption provided by the American Heart Association, the National Academies of Sciences, Engineering, and Medicine, the World Health Organization, and many other governmental and non-governmental organizations. What is needed is implementation of sodium reduction recommendations, especially a progressive decrement in the addition of sodium during food processing and food preparation (which typically accounts for 80–90% of dietary sodium). This represents one of the most achievable approaches for prevention of hypertension and reduction of the enormous health and financial consequences of excessive dietary sodium intake in countries all around the world.