Introduction

Almost a third of American adults have high blood pressure (BP),¹ and only about half of those with high BP have it under control.² Sodium (Na) reduction is commonly recommended as a means to lower BP. Current clinical and public health recommendations for Na intake are expressed as milligrams of Na per day (mg/d),^2–4 whether these recommendations should be expressed in terms of Na to energy intake (Na density, mg/kcal) has not been explored.

Na density is the ratio of Na to energy in a diet. For example, a 2000 kcal diet that contains 2400 mg Na has an Na density of 1.2 mg/kcal. Because Na and energy intakes are so highly correlated, it is much harder for someone who is larger and more active to achieve the current Na recommendations of 2300 or 2400 mg/d than it is for someone who is smaller and less active simply because of their differing energy intakes.^2,3 Furthermore, smaller or less active people might not experience the benefits of Na reduction at an intake of 2300 mg/d because that absolute amount of Na results in a high Na density.

The original analysis of the DASH (Dietary Approaches to Stop Hypertension)–Na trial by Sacks et al⁵ is perhaps the most influential evidence underlying current Na recommendations.⁶ A recent policy statement from the American College of Cardiology and the American Heart Association concluded that “Reducing sodium intake to a mean of 2400 mg per day (d), relative to 3300 mg/d, lowers BP by 2/1 mm Hg, and reducing intake to a mean of 1500 mg/d lowers BP by 7/3 mm Hg”,² based largely on the results of the DASH–Na trial. However, the experimental variable in DASH–Na was not absolute Na (mg/d), but rather Na density (mg/kcal) at 3 levels.⁶

The main purpose of this secondary analysis of the DASH–Na trial was to determine whether the strength of the relationship between Na intake and BP varied with energy intakes. We also investigated whether this relationship varied with race and obesity status.

Methods

Data are available at the National Heart, Lung and Blood Institute Biologic Specimen and Data Repository and were accessed at (https://biolincc.nhlbi.nih.gov/home/). Age and race of the participants were provided by the DASH–Na trial principal investigator (L.J.A.) and were matched at the individual level by Biologic Specimen and Data Repository staff.

The DASH–Na study was a multicenter, randomized, crossover trial comparing the effects on BP of 3 levels of Na intake at the same energy intake, that is, Na density, in 2 types of diets, the DASH diet and a control diet.⁶ BP measurements (after screening) were taken by trained, certified observers using a random-zero sphygmomanometer and following the protocol that was used in the DASH and Trials of Hypertension Prevention studies.^7–9 Institutional Review Boards at each center approved the original study protocol and subsequent analyses. Written informed consent was obtained from each participant. This analysis was approved by the University of Utah, Johns Hopkins University, and Wake Forest University Institutional Review Board.

Participants were adults whose BP exceeded 120/80 mm Hg, including those with stage 1 hypertension (a systolic BP [SBP] of 140–159 mm Hg or a diastolic BP [DBP] of 90–95 mm Hg). The 3 Na density levels were defined as high (1.6 mg/kcal, representing a target of 3450 mg/d Na at an energy intake of 2100 kcal and reflecting typical consumption in the United States), intermediate (1.1 mg/kcal, representing a target of 2300 mg/d at 2100 kcal and reflecting Na recommendations),¹⁰ and low (0.5 mg/kcal, representing a target of 1150 mg/d at 2100 kcal and reflecting a level was hypothesized to further lower BP).

The DASH–Na trial was a tightly controlled feeding study. Food was provided to participants; in addition, up to 3 servings of allowed beverages including tea, coffee, some kinds of diet soda, and up to 14 alcoholic beverages per week. During a 2-week run-in period, participants (n=412) ate the high-Na control diet, which was designed to be similar to what many Americans eat. Participants were then randomly assigned to the DASH diet or the control diet. Participants ate their assigned diet, provided at 3 Na density levels, for 30 consecutive days in random order in a crossover design with up to 5 days between study periods. Other than the Na levels which varied by period, the assigned diets (DASH and control diets) remained similar in all other ways, including their potassium, magnesium, and calcium content.

Each participant’s energy intake was adjusted to ensure that his or her weight remained constant throughout the study, that is, energy intake was equal to energy requirement. Table S1 in the online-only Data Supplement shows the absolute Na targets at the 3 energy density levels by energy intake. Larger or more active people received more food and, therefore, more Na than smaller or less active people at each of the 3 Na levels.¹¹

Inclusion/Exclusion Criteria

The study participants were identified as black, white, or other race, and Hispanic or not; the number of participants of a race/ethnicity other than non-Hispanic black or white was not large enough to be analyzed separately. Therefore, from the original sample of 412 participants, 19 were excluded (Hispanics [n=12] and those of races other than black and white [n=7]). Data were also excluded from a study period when a measured BP during or at the end of the study period was not available (n=3 participants fully excluded; n=40 periods). The authors defined a valid study day as having a value for energy intake, not >2 meals missed, and not >4 servings of food that was not allowed according to daily diaries kept by the participants. Also excluded were study periods having <12 valid study days or <5 valid days in the last week of the period (n=11 participants and n=35 study periods) or having an absolute Na intake >6700 mg/d (n=2 study periods), resulting in 379 participants and 1102 study periods included in these analyses.

Energy and Na Intakes

Daily energy intake consumed was computed from the sum of energy from meals served based on the nutrient content of the menus, allowed beverages, and the supplemental nutrition bars that were provided to meet energy targets. The usual energy intake value for each period used in this data analysis was the mean daily energy intake, calculated using all valid diet days. Na intake was calculated in a similar manner. Na density was computed as the mean Na intake for a period divided by the mean energy intake for the period. The mean Na densities (mg/kcal) of the diets consumed by participants, calculated by the authors, for both the DASH and control diets were 0.6 for the low-Na density diets, 1.2 for the intermediate Na density diets, and 1.7 for the high-Na density diets (Table S2).

Statistical Analysis

Statistical analyses were conducted with SAS version 9.4 (SAS Institute, Inc, Cary, NC). Usual energy, Na intakes, and Na densities were estimated using means, SDs, and ranges stratified by Na density and diet arm. Baseline characteristics were described using frequencies for categorical variables and means and SDs for continuous variables. For illustration purposes, we present participant characteristics at roughly the lower, middle 2, and upper quartiles of energy intake. Differences in participant characteristics by energy intake were tested by χ² test, Monte Carlo estimate of Fisher exact test, or ANOVA.

A mixed-effects model was used to model SBP and DBP as a function of Na and energy (both as continuous variables) and their interaction, while accounting for age, sex, smoking status (current versus former or never), race, cohort, carryover (Na intake in the previous period), and clinical center in all analyses. Diet type (DASH or control) was included in all models; the effect of Na and energy intake on BP was allowed to vary by diet by including interactions of diet type with Na intake and diet type with energy intake. Estimates and corresponding standard errors at Na and energy levels were estimated using linear contrasts. Baseline BP was not included in the models because it is in the causal pathway for BP at the end of each feeding period.

The a priori plan was to determine whether the impact of Na and energy on BP was different in blacks and whites. Therefore, we evaluated interactions of race with Na and energy, diet type, and relevant interactions and conducted stratified analysis of the relationship of Na and intake (both as continuous variables) with BP with the same covariates for adjustments, except race. To test whether the slope of the association between Na and BP differed by obesity status, interaction terms for obesity status (body mass index [BMI] or ≥30 kg/m²) with Na intake, energy intake, and diet type and relevant interactions were modeled and then the models were fit stratified by obesity status.

Results

Participants with higher energy intakes were more likely to be male, white, and married with a higher BMI and a lower SBP than participants with lower energy intakes (Table; Table S3).

The strength of the relationships between absolute Na and both SBP and DBP varied with energy intake such that BP rose more steeply with increasing Na at lower energy intake than at higher energy intake (interaction of Na and energy, P<0.001; Figure 1). There were no significant differences between men and women in this model. We chose to show comparisons at the midpoint of the low energy group (2100 kcal, higher Na density) and high energy group (3200 kcal, lower Na density) at the current recommended upper limit of Na (2300 mg)³ and to present these estimates by diet type (Figure 1A and 1C) to illustrate the interaction between Na and energy. At 2300 mg of Na on the control diet, those with usual energy intake of 2100 kcal had an average SBP that was 3.0 mm Hg (95% CI, 0.2–5.8) higher than those with usual energy intake of 3200 kcal; DBP was 2.7 mm Hg (95% CI, 1.0–4.5) higher. On the DASH diet (Figure 1B and 1D), the differences between 2100 kcal and 3200 kcal/d intake at 2300 mg were attenuated and not significantly different from zero (SBP, 1.3 mm Hg; 95% CI, −1.4 to 4.0; DBP, 1.3 mm Hg; 95% CI, −0.4 to 3.1). However, on the DASH diet, the differences between 2100 kcal and 3200 kcal/d intake at 3600 mg were significantly different from zero (SBP, 3.4 mm Hg; 95% CI, 0.7–6.1; DBP, 2.1 mm Hg; 95% CI, 0.4–3.9).

• Download figure
• Download PowerPoint

The interaction of Na intake and race in the full model was significant for DBP (P=0.03), but not for SBP (P=0.07). There was no significant interaction between race and energy, or among race, Na intake, and energy for SBP or DBP. In stratified models, the association of Na with SBP was stronger at lower energy intake (higher Na density) than higher energy intake (lower Na density) in both blacks and whites (both P<0.001), whereas the association of Na and DBP varied with energy intake among blacks (P=0.001) but not whites (P=0.288). On the control diet in blacks at 2300 mg, the difference in SBP between individuals whose energy intake was 2100 kcal when compared with 3200 kcal was 3.8 mm Hg (95% CI, 0.0–7.7), whereas in whites, it was not significantly different than zero (2.4 mm Hg; 95% CI, −1.6 to 6.4; Figure 2A and 2B). On the control diet in blacks at 2300 mg Na intake, DBP was 4.0 mm Hg (95% CI, 1.6–6.3) higher at 2100 kcal compared with 3200 kcal, whereas in whites, it was not significantly higher (1.8 mm Hg; 95% CI, −0.8 to 4.5; Figure 2C and 2D). On the DASH diet at 2300 mg Na intake, SBP did not differ at between 2100 and 3200 kcal in blacks (SBP, 2.4 mm Hg; 95% CI, −1.3 to 6.0) and whites (0.3 mm Hg; 95% CI, −4.3 to 3.7); differences were also not significant for DBP in blacks (1.6 mm Hg; 95% CI, −0.6 to 3.9) or whites (1.1 mm Hg; 95% CI, −1.5 to 3.7) on the DASH diet.

• Download figure
• Download PowerPoint

There was a significant 3-way interaction between Na intake, energy, and obesity status for SBP (P=0.033); for DBP, the 3-way interaction was not significant, but there was a significant interaction between energy intake and obesity (P=0.005). We, therefore, used stratified models to describe the relationship between energy and Na by obesity status (Figure 3). In stratified models, the association of Na with SBP was stronger at lower energy intake (higher Na density) than at higher energy intake (lower Na density) in both nonobese (interaction of Na with energy; P<0.001) and obese (interaction of Na with energy; P=0.028); the association of Na and DBP by energy intakes was similar (interaction of Na with energy, P=0.005 nonobese; P=0.049 obese). On the control diet at 2300 mg Na intake in the nonobese, SBP was 4.7 mm Hg (95% CI, 1.0, 8.5) higher at 2100 kcal compared with 3200 kcal (Figure 3A), and DBP was 3.1 mm Hg (95% CI, 0.7, 5.5) higher (Figure 3C). In obese at 2300 mg Na intake on the control diet, there were no differences in SBP (0.6 mm Hg; 95% CI, −3.8 to 4.9; Figure 3B) or DBP (1.8 mm Hg; 95% CI, −1.0 to 4.5; Figure 3D) between 2100 and 3200 kcal. On the DASH diet at 2300 mg Na, there were no significant differences in SBP at 2100 versus 3200 kcal regardless of obesity status (2.5 mm Hg; 95% CI, −1.0 to 6.0 for nonobese and −1.3 mm Hg; 95% CI, −6.0 to 3.4 for obese). On the DASH diet at 2300 mg Na, DBP was higher at 2100 versus 3200 kcal in the nonobese (2.1 mm Hg; 95% CI, 0.1–4.4), but not in the obese (−0.3 mm Hg (95% CI, −3.2 to 2.6).

• Download figure
• Download PowerPoint

Discussion

This analysis was conducted to determine whether the relationship between Na and BP varies with energy intake. Several key findings emerged. First, the slope of the relationship BP with Na intake varied by energy intake, suggesting that Na density may reflect the relationship with BP better than absolute Na intake does. Second, as previously reported, the effect of Na reduction was attenuated in the setting of the DASH diet compared with the control diet, suggesting that aspects of diet also influence the BP response to changes in Na intake.^5,12,13 In addition to attenuating the BP response to Na, the DASH diet also lowered BP, consistent with evidence that aspects of diet, including an increased intake of potassium, independently lower BP.¹⁴ Third, the association of Na with energy intake on BP persisted when the analyses were stratified by race, except for DBP among whites. Fourth, with stratification by obesity status, larger differences in BP by energy intake were observed among those who were nonobese compared with obese.

Although the DASH–Na study controlled Na density, and not absolute Na intake, the original results of this trial were interpreted in terms of Na intake rather than Na density. The lack of reporting results by Na density may reflect the absence of data on energy intake in many studies. Investigations of the relationship of Na density with mortality in observational studies have provided mixed results. The investigation of Na density with mortality in the NHANES II follow-up study (National Health and Nutrition Examination Survey) and from NHANES III suggested an inverse association of Na and cardiovascular disease mortality, even when considering Na with respect to energy intake.^15,16 In these studies, however, Na and energy intake were estimated using one 24-hour recall, which is not adequate to estimate an individual’s usual dietary intake because of large day-to-day variation in both Na intake and energy intake and their ratio. Further, energy intake was implausibly low in both studies. It is possible that the observed relationship between Na intake and cardiovascular disease mortality may be explained by under-reporting bias, that is, if obese people report lower Na density and are more likely to die of cardiovascular disease, the observed relationship may reflect the impact of obesity rather than diet. The most recent evidence from a prospective cohort study (n=716, 19-year follow-up) aligns with the results of the current analysis, with reported incidence of cardiovascular events twice as high among those in the highest versus lowest quartile of Na density (22% versus 11%, respectively; P=0.005) according to 7-day food records collected at baseline.¹⁷

Current Na recommendations are expressed in absolute amounts (ie, 2300–2400 mg/d),^2–4,18 and this reflects how study findings have been reported^2,3,19–21; however, when Na recommendations are applied in practice, they are often, but not intentionally, energy adjusted. For example, in the Healthy US-style dietary pattern, these amounts range from 921 mg Na/d at 1000 calories to 2392 mg Na/d at 3200 calories.²² This is because everyone is encouraged to choose nutrient-dense, low-Na foods. The Healthy Eating Index, an index that represents adherence of dietary patterns with the Dietary Guidelines for Americans, assesses food and nutrient intakes, including Na, on an energy-adjusted basis, that is, the scoring standards are expressed as densities.^23–25 In the Healthy Eating Index-2010 and Healthy Eating Index-2015, an Na density of 1.1 mg/kcal or lower receives the highest (best) score for the Na component of the index. In addition, the Dietary Reference Intakes for Na for children and older adults are lower than for young and middle-aged adults because they, too, were set by energy adjustment.⁴ Finally, a recent evidence-based guideline for dietitians for the management of hypertension recommends an Na restriction of 1500 to 2400 mg/d based on several factors, including energy intake.²⁰

Guenther et al²⁶ demonstrated that it is possible to have a nutritionally adequate diet with an Na intake of 1500 mg/d (Na density 0.4–0.5 mg/kcal) by developing food choice scenarios that encompassed the Na density levels in the DASH–Na trial. Making only the lowest Na choices, such as unsalted potato chips, unsalted butter or margarine, and bread with a lower Na content, resulted in an Na density similar to the lowest Na density in the DASH–Na trial. Typical food choices resulted in Na density that was similar to the higher Na density in the DASH–Na trial (1.7 mg/kcal) and to the estimated average Na density of diets in the United States of 1.7 mg/kcal.²²

The authors were interested in understanding whether the association of Na with BP varied in blacks versus whites because of the higher prevalence of hypertension among blacks in the United States.²⁷ The finding of a slightly larger difference in the change in BP in response to Na among Blacks when compared with whites conflicts with a recent meta-analysis that found no difference in the blacks and whites in response to Na reduction²⁸ but is consistent with original findings from the DASH–Na trial.¹² Many mechanisms may contribute to black/white differences in hypertension. Chronic stress may be a contributor among blacks.²⁹ Differences in kidney function and Na retention are hypothesized reasons, as are differences in potassium homeostasis potentially related to differences in aldosterone levels and aldosterone sensitivity.³⁰

The finding of greater differences in BP by energy intake at the same level of Na intake among participants with normal weight status compared with those who were obese on the control diet is novel, though contrary to what might be expected based on the known increase in hypertension with obesity. Clinical trials show a sustained reduction in BP with Na reduction and weight loss³¹ and synergistic influence of Na reduction and weight loss on BP in the elderly.^32,33 Increases in energy needs are related to body surface area and exercise, and greater Na losses will occur with both. Furthermore, salt sensitivity of BP may be inversely related to physical activity. In this study, we do not have data to support whether any of these mechanisms were involved. We can only speculate on whether these findings would apply to the refractory nature of hypertension among individuals with low BMI common in certain ethnic groups. Further studies or secondary analyses of other trials are warranted to explore whether these results hold up among other racial/ethnic groups and body size differences.

Confounding of sex and energy intake is evident in the proportion of women and men in the highest and lowest levels of energy intake. In analyses controlling for energy intake, there we no significant sex differences in the relationship of Na and BP. Nonetheless, examination of the differences in Na and energy intake in men and women with respect to BP are warranted in light of the putative mechanisms that might account for differential BP response to diet by sex. Sympathetic nerve activity related to hypertension is different in young women when compared with men and older men and women.^34–36

Several strengths of the current analyses stem from the design of the DASH–Na trial.⁵ The study documents the effects of the DASH diet and Na reduction, alone and combined, on BP. Because it was a feeding study, Na and energy intakes were controlled and not prone to reporting bias. The crossover study design allowed for control of within-person factors that may impact BP. The study included sufficient numbers of non-Hispanic whites and blacks to examine relationships within these races separately. Our analyses also have limitations. An analysis of Hispanics and people of other races was not possible. Second, in this trial, 3 levels of Na density were tested, but the optimal Na density may or may not have been observed. Third, these analyses extend beyond the randomized design and are, therefore, subject to confounding by factors associated with Na homeostasis such as sex and energy intake such as physical activity.

Perspectives

Making Na recommendations based on energy intake instead of making recommendations based on absolute amounts of Na has practical advantages. Stating Na recommendations on a per calorie basis would reflect the way that dietary guidance for Na is currently applied in practice and would not change the overall message to consumers to reduce Na intake and to the food industry to reduce the Na content of foods. If Na recommendations were expressed as Na density (mg/kcal), it would be much easier for patients and dietitians to plan reduced Na diets because people with varying energy needs could eat the same foods but in varying amounts. Future work is needed to establish the optimal Na density. It would be helpful to establish whether the relationship between Na and BP varies when energy needs vary because of differences in physical activity rather than by differences in body size.

Footnotes