Body Weight, Sodium, Potassium, and Blood Pressure

The global pattern of illness has become increasingly similar, with a progressive shift to morbidity and mortality from chronic diseases. Cardiovascular disease (CVD) is the leading cause of death in all of the world's most populous nations, including China, India, the United States, Indonesia, and Brazil, and accounts for almost 40% of all deaths worldwide.1 While the role of genetics remains unclear, poor diet, physical inactivity, and cigarette smoking are central to CVD risk in most individuals and populations.2 Unhealthy diet and physical inactivity manifest as overweight and obesity, high blood pressure (BP), lipid abnormalities, diabetes mellitus, and the metabolic syndrome. The 2010 Global Burden of Illness Study identified high BP as the most important of 67 risk factors studied for worldwide risk of mortality (9.4 million deaths per year; 95% confidence interval [CI], 8.6–10.1) and disability‐adjusted life years (7.0%; 95% CI, 6.2–7.7%).3 High BP was responsible for more deaths than the second (tobacco smoking) and third (acute respiratory infections) leading risk factors combined. Prevention and treatment of high BP are essential elements in any meaningful effort to reduce the population burden of illness from CVD and other BP‐related disorders.2

A wide array of dietary factors have been related to high BP, including high sodium (Na) and calorie intake and low potassium (K) intake.4, 5 These variables are closely interrelated, with correlation coefficients of about 0.7 to 0.8.6, 7, 8 There is general acceptance of a strong direct association between both body weight and dietary Na with BP and a strong inverse association between dietary K and BP, with no apparent threshold for any of these relationships.9, 10, 11 An important question is whether the effect of dietary Na on BP is modified by body weight or intake of dietary K. Clinical trials have the potential to provide the most pertinent information treatment effect modification. Observational studies can provide only indirect estimates of intervention effect and the potential influence of an intervention effect modifier. However, compared with clinical trials, observational investigations often provide an opportunity to examine a much larger and more representative study sample, and, in the case of cohort analyses, a much longer period of observation is often feasible.

Some observational studies have suggested a greater “effect” of dietary Na on BP at higher compared with lower levels of body weight.12, 13, 14, 15 In the cross‐sectional Yi Migrant Study, an analysis was conducted among 831 men in whom body weight was measured in a standardized fashion, BP was estimated using an average of nine recordings obtained over a 3‐day period and Na intake was derived from an average of three consecutive 24‐hour urine specimens.14 Both age and body mass index (BMI) modified the association of urinary Na with BP throughout the entire range of urinary Na, with the modifying effect of higher BMI on systolic BP being about 5 mm Hg greater at the highest compared with the lowest level of urinary Na for patients 38 years and older. In the INTERSALT study, adjustment for BMI reduced the effect of a 100 mmol higher level of urinary Na (corrected for regression dilution bias) on systolic BP from 6.0 mm Hg to 3.1 mm Hg.16 However, there was no evidence that BMI modified the effect of Na on BP. In a prospective cohort analysis of the first National Health and Nutrition Examination Survey Follow‐up Study, with up to 10 years of observation, dietary Na estimated from a 24‐hour dietary recall was associated with CVD morbidity and mortality in overweight but not in nonoverweight adults.17 Clinical trials, especially those that employ a factorial design, allow for more direct exploration of the interrelationship between the effects of body weight and dietary Na on BP. The two largest and most prolonged, 30 to 36 months, randomized controlled trials of weight loss and Na reduction effects on BP utilized a factorial design.18, 19 The Trial of Nonpharmacologic Interventions in the Elderly (TONE) was conducted in 975 seniors with hypertension who were well controlled on a single antihypertensive medication.18 Phase II of the Trial of Hypertension Prevention (TOHP) was conducted in 2382 adults aged 30 to 54 years with untreated high normal BP.19 Both trials confirmed weight loss and dietary Na reduction as effective non‐drug interventions for lowering BP. Consistent with experience in the Yi People Study, TONE demonstrated an additive effect of the two interventions. For example, after 3 months of intervention, while continuing on a single antihypertensive medication, systolic BP was reduced by an average of 5.3 mm Hg, 4.0 mm Hg, 3.4 mm Hg, and 0.8 mm Hg in the combined weight loss and Na reduction, weight loss, Na reduction, and usual care groups, respectively. Following subsequent withdrawal of antihypertensive medication, 50% of those assigned to the combination intervention remained “event‐free” (mostly, need to recommence antihypertensive medication) 24 months after drug withdrawal compared with 39.6%, 37.9%, and 23.9% for the Na reduction, weight loss, and usual care groups, respectively. There was little evidence for such additivity in the TOHP phase II study. This may have been a reflection of less effective reduction in Na intake for the combined Na reduction and weight loss group compared with Na reduction on its own.

Beginning in the 1970s, cross‐sectional observational analyses began to show the relationship between dietary Na to K ratio and BP20, 21 and subsequently numerous studies identified a robust relationship between dietary Na/K and BP.10, 22, 23, 24 However, many of these reports were based on a relatively small sample size or imperfect assessment of dietary electrolyte values. In a Yi People Study of 419 men, a standardized protocol was used to measure BP, and dietary Na and K were estimated by averaging results from three consecutive 24‐hour samples and three 24‐hour dietary recalls.25 The two measures of dietary intake identified strong, significant relationships between BP and both Na, K, and Na/K. In a fully adjusted multivariate model, urinary Na/K was a better explanatory variable than either Na or K alone. These findings are consistent with clinical trials experience, where Na intake has been an important modifier for the effect of K on BP.26 The joint effects of Na and K on subsequent CVD was studied during extended follow‐up of 2974 TOHP participants who had not been assigned to an active Na reduction intervention.27 The participants were aged 30 to 54 years with untreated high normal BP at baseline, making reverse causality an unlikely possibility. Urinary Na, K, and Na/K were estimated using up to seven (median=5) carefully collected 24‐hour urine specimens. The cohort was followed for 10 to 15 years using active and passive surveillance. In a fully adjusted model, the relative risks (95% CI) of CVD for a 100 mmol higher level of urinary Na, 50 mmol higher level of urinary K, and 1.00 unit higher level of urinary Na/K were 1.25 (0.91–2.72), 0.83 (0.53–1.29), and 1.24 (1.05–1.46), respectively. A spline plot suggested that the relationship between urinary Na/K and CVD was linear, with no evidence of a threshold for risk.

In this issue of the Journal, Yan and colleagues28 make an important contribution to our understanding of the apparent influence that body weight has on the relationship between dietary Na, K, Na/K, and BP. Their study was cross‐sectional and conducted in China, where dietary Na and Na/K is very high compared with Western countries. However, the sample size was relatively large, helping to mitigate the challenge of random error associated with estimation of BP and dietary Na. Benefits of a large sample size are often overwhelmed by the potential for systematic error resulting from lack of attention to quality in the measurement of key variables. Yan and colleagues used rigorous methods, including careful measurements of 24‐hour urine for estimation of dietary Na. Their findings provide additional support for independent, complementary associations between dietary Na and K and BP. In addition, their data support the role of body weight as an effect modifier for the association between dietary Na and BP. They may be the first to provide detailed information on the role of body weight as an effect modifier for the relationships between dietary K and Na/K with BP.

In combination with the preponderance of information from prior reports, the new study raises an important question. Should we be more focused on Na/K in clinical practice? To date, practice guidelines have tended to focus on recommendations for modifying dietary Na rather than K or Na/K.29, 30, 31, 32 Perhaps this reflects the fact that trials of K supplementation have generally been of short duration and primarily based on pill supplementation rather than dietary change.26 Another reason may be that many guidelines recommend the Dietary Approaches to Stop Hypertension (DASH) diet33 or a Mediterranean diet. The DASH diet is rich in potassium (approximately 4400 mg/d), exceeding the World Health Organization recommendation for dietary K (3510 mg/d)34 and coming close to meeting the adequate intake level (4700 mg/d) recommended by the Institute of Medicine.35 Whatever the reason, the accumulated evidence suggests that special attention should be paid to Na reduction and K supplementation, in addition to weight loss counseling, in persons with hypertension who are overweight or obese. These are high‐risk individuals for whom attention to diet is particularly likely to yield a reduction in BP and enhance the effect of antihypertensive medication. In addition, the role of body weight as an effect modifier for the relationships between BP and dietary Na, K, and Na/K should be further explored in observational epidemiology and clinical trials.