A, Sodium models were adjusted for race, height, activity, television/video time, percentage of calories from solid fat and added sugars, and dietary fiber. B, Potassium models were adjusted for race, height, activity, television/video time, and percentage of calories from solid fat and added sugars. C, Potassium to sodium ratio models were adjusted for race, height, activity, television/video time, percentage of calories from solid fat and added sugars, dietary fiber, and energy intake.
eTable. Dietary Variables for All Girls Stratified by Sodium and Potassium Categories.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Buendia JR, Bradlee ML, Daniels SR, Singer MR, Moore LL. Longitudinal Effects of Dietary Sodium and Potassium on Blood Pressure in Adolescent Girls. JAMA Pediatr. 2015;169(6):560–568. doi:10.1001/jamapediatrics.2015.0411
Identification of risk factors early in life for the development of high blood pressure is critical to the prevention of cardiovascular disease.
To study prospectively the effect of dietary sodium, potassium, and the potassium to sodium ratio on adolescent blood pressure.
Design, Setting, and Participants
The National Heart, Lung, and Blood Institute’s Growth and Health Study is a prospective cohort study with sites in Richmond, California; Cincinnati, Ohio; and Washington, DC. Participants included 2185 black and white girls initially aged 9 to 10 years with complete data for early-adolescent to midadolescent diet and blood pressure who were followed up for 10 years. The first examination visits were from March 1987 through February 1988 and follow-up continued until February 1999. Longitudinal mixed models and analysis of covariance models were used to assess the effect of dietary sodium, potassium, and the potassium to sodium ratio on systolic and diastolic blood pressures throughout adolescence and after 10 years of follow-up, adjusting for race, height, activity, television/video time, energy intake, and other dietary factors.
Mean dietary sodium and potassium intakes and the mean potassium to sodium ratio in individuals aged 9 to 17 years. To eliminate potential confounding by energy intake, energy-adjusted sodium and potassium residuals were estimated.
Main Outcomes and Measures
Mean systolic and diastolic blood pressures throughout adolescence and at the end of follow-up (individuals aged 17-21 years).
Sodium intakes were classified as less than 2500 mg/d (19.4% of participants), 2500 mg/d to less than 3000 mg/d (29.5%), 3000 mg/d to less than 4000 mg/d (41.4%), and 4000 mg/d or more (9.7%). Potassium intakes ranged from less than 1800 mg/d (36.0% of participants) to 1800 mg/d to less than 2100 mg/d (26.2%), 2100 mg/d to less than 2400 mg/d (18.8%), and 2400 mg/d or more (19.0%). There was no evidence that higher sodium intakes (3000 to <4000 mg/d and ≥4000 mg/d vs <2500 mg/d) had an adverse effect on adolescent blood pressure and longitudinal mixed models showed that those consuming 3500 mg/d or more had generally lower diastolic blood pressures compared with individuals consuming less than 2500 mg/d (P = .18). However, higher potassium intakes were inversely associated with blood pressure change throughout adolescence (P < .001 for systolic and diastolic) and at the end of follow-up (P = .02 and P = .05 for systolic and diastolic, respectively). While the potassium to sodium ratio was also inversely associated with systolic blood pressure (P = .04), these effects were generally weaker compared with effects for potassium alone.
Conclusions and Relevance
In this study of adolescent girls, consumption of 3500 mg/d of sodium or more had no adverse effect on blood pressure. The beneficial effects of dietary potassium on both systolic and diastolic blood pressures suggest that consuming more potassium-rich foods during childhood may help suppress the adolescent increase in blood pressure.
The scientific community has historically believed that most people in the United States consume excessive dietary sodium, resulting in an elevated risk of hypertension and cardiovascular disease.1 While in developed countries usual sodium intake is estimated to be approximately 3500 mg/d,2 the current Dietary Guidelines for Americans recommend limiting sodium intake to less than 2300 mg/d for healthy individuals between 2 and 50 years of age.3 Adults older than 50 years, black individuals, and individuals with hypertension, diabetes mellitus, or chronic kidney disease (a total of about 50% of the US population) are advised to limit their intake to 1500 mg/d. Dietary sodium restriction is considered a key strategy for reducing childhood and adult blood pressure (BP).4
Conflicting data, including evidence that intakes of sodium less than 1500 mg/d may do more harm than good for some individuals, led the Institute of Medicine to convene an expert panel on the topic. Their 2013 report2 concluded there was insufficient evidence to support the current sodium Dietary Guidelines for Americans. Furthermore, there was no evidence to support treating population subgroups (eg, black individuals) differently from the rest of the US population.
Since the Institute of Medicine report was released, more studies have found U- or J-shaped relationships between sodium and cardiovascular risk or all-cause mortality, with individuals who consume less than 3.0 g or more than 6.0 g of sodium per day having higher risks.5,6 Studies of the direct effects of dietary sodium on BP have also yielded conflicting results, although current evidence suggests that most BP increases are associated with sodium intakes of more than 5000 mg/d.7 The relationship between dietary sodium and BP in children and adolescents is largely unexamined in prospective studies.8
Several meta-analyses of clinical trials indicate that increasing potassium intake (generally through supplement use) reduces BP in hypertensive adults9,10; however, population-based studies have failed to show such an association.11 Data related to potassium effects on BP in younger populations are limited and inconsistent.12 The Institute of Medicine has set the current Adequate Intake for potassium to be 4500 mg/d for children aged 9 to 13 years and 4700 mg/d for older children and adults. In the 2010 Dietary Guidelines for Americans,13 adolescent girls consumed an average of 1976 mg/d, significantly less than recommendations. It is unknown whether beneficial effects of dietary potassium would be evident at more modest levels of intake as has been found in adults.6,7
Several previous studies have suggested that when potassium intake is relatively high and sodium intake is low (leading to a higher potassium to sodium ratio), there will be many more benefits to BP than those associated with either nutrient alone.14 Few data exist in younger populations; however, at least 1 study has demonstrated an inverse association between the sodium to potassium ratio and BP in this age group.8
The primary objective of the current study was to evaluate the long-term effects of dietary sodium, potassium, and the potassium to sodium ratio on BP during and at the end of adolescence.
Our goal was to estimate long-term effects of dietary sodium, potassium, and the potassium to sodium ratio on blood pressure throughout adolescence among white and black girls. Higher sodium consumption (3000 to <4000 mg/d and ≥4000 mg/d vs <2500 mg/d) during adolescence had no effect on either systolic or diastolic blood pressure across 10 years of follow-up.
Compared with girls having the lowest potassium intakes (<1800 mg/d), girls with higher intakes (≥2400 mg/d) had lower systolic and diastolic blood pressures at the end of adolescence.
Girls with a higher potassium to sodium ratio (≥0.8) had systolic blood pressure levels that were 1.3 mm Hg lower than girls whose potassium to sodium ratios were below 0.6.
Potassium intake was more strongly and consistently associated with lower blood pressures from later childhood throughout adolescence compared with either the potassium to sodium ratio or sodium intake alone.
These analyses were approved by the Boston University Institutional Review Board and are based on previously collected data from the National Heart, Lung, and Blood Institute’s Growth and Health Study (NGHS), a 10-year longitudinal study to evaluate the development of obesity and other cardiometabolic outcomes in girls initially aged 9 to 10 years. Written informed consent was attained from the parents or guardians of study participants. The NGHS enrolled 2379 white and black girls during 1987 and 1988 from 3 diverse clinical centers. Follow-up continued until about 2000. Details of the original study design and methods are previously published.15 For the current analyses, girls with missing data on diet, BP, or potential confounding variables included in the final models (n = 194) were excluded, yielding a final sample of 2185 participants.
Diet was assessed using 3-day diet records (2 weekdays and 1 weekend day) during 8 of the 10 study years (examination years 1-5, 7, 8, and 10). Girls were instructed by a trained nutritionist in standard household measures to estimate portion sizes. When necessary, parents provided information on recipes, brands, and other details. Standardized in-person debriefing was completed and data were entered in the Nutrition Data System16 of the University of Minnesota to calculate nutrient intakes.
Blood pressure was measured annually following a standardized protocol with a V-Lok Cuff mercury sphygmomanometer (Baum Desktop Model). Three measurements were taken with a 30-second rest in between. The first 2 consecutive beats indicated the first Korotkoff sound. The fourth Korotkoff sound was the point of sound muffling and the fifth Korotkoff sound was the disappearance of sound. Mean systolic BP (SBP) and diastolic BP (DBP) were calculated as the mean of the second and third BP readings. The BP outcome in this study was derived from the means of all available measurements from ages 17 to 21 years.
Data on potential confounders, such as body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), physical activity, and television/video viewing were collected systematically across 10 years. Height was measured yearly in duplicate using a portable stadiometer and weight was measured on a digital scale. Physical activity was assessed at visits 1, 3, 5, and 7 to 10 using a validated habitual activity questionnaire.17 The habitual activity questionnaire score was calculated by multiplying an estimate of the metabolic equivalent level for each recorded activity by weekly frequency of participation and the portion of the year it was performed. Time spent watching television (hours per day) was estimated annually by questionnaire.
Mean sodium and potassium intakes were calculated from diet records at each examination between ages 9 and 17 years and were examined as both continuous and categorical variables. The mean potassium to sodium ratio was derived by dividing each girl’s age-specific mean potassium intake by her age-specific mean sodium intake and then estimating the mean potassium to sodium ratio across all years from ages 9 to 17 years. Cutoff values were selected through sensitivity analyses to reflect the trends observed in the data while optimizing analytic power. Sodium intake levels were classified as less than 2500 mg, 2500 mg to less than 3000 mg, 3000 mg to less than 4000 mg, and 4000 mg or more. Potassium was classified as less than 1800 mg, 1800 mg to less than 2100 mg, 2100 mg to less than 2400 mg, and 2400 mg or more. The potassium to sodium ratio was classified as less than 0.6, 0.6 to less than 0.7, 0.7 to less than 0.8, and 0.8 or more.
Because both sodium and potassium were highly correlated with total energy (r = 0.81 for sodium and r = 0.70 for potassium), mean sodium and potassium intakes were also expressed as energy-adjusted residuals (to prevent collinearity associated with adding energy to the multivariable models). The residual method reduced variation in dietary sodium and potassium intake resulting from differences in body size.18 Energy-adjusted sodium and potassium residuals were classified as less than −300 mg, −300 mg to less than −100 mg, −100 mg to less than 100 mg, and 100 mg or more.
Analysis of covariance models were used to examine the relationship between categories of sodium and potassium intake (as well as their energy-adjusted residuals) and the potassium to sodium ratio on adolescent SBP and DBP at the end of follow-up (individuals aged 17-21 years). The following potential confounding or effect-modifying variables were explored for inclusion in the multivariable models: age, race, socioeconomic status, physical activity, television/video time, height, weight, baseline and follow-up BMI, percentage of lean body mass, age at menarche, baseline SBP and DBP, and a wide range of dietary-intake variables including energy; low-fat dairy servings; fiber; whole grains; total fruits and vegetables; percentage of calories from fat, saturated fat, carbohydrates, and protein; intakes of red and processed meats; solid fat and added sugars; Healthy Eating Index score; and other minerals (eg, calcium and magnesium). Variables that changed the effect estimates (mean differences) by 10% or more were retained in the final models. These included race, height, activity, television/video time, percentage of calories from solid fat and added sugars, energy (potassium to sodium ratio model only), and mean fiber intake (sodium and potassium to sodium ratio models only).
Linear mixed-effects regression models for data with repeated measures were used to estimate the effects of sodium, potassium, and the potassium to sodium ratio on adjusted mean SBP and DBP throughout adolescence. A sodium (or potassium) intake group-by-age interaction term was used to estimate BP values across time for different levels of (categorical) dietary intake. The F test for overall group differences was completed in addition to individual tests of all 2-way group comparisons (eg, high vs low, high vs moderate, and moderate vs low intakes of dietary sodium). Participant-specific random intercepts were used to account for the correlation owing to repeated BP measures. The unstructured covariance matrix was used in the mixed models to reflect the correlation among random intercepts. For these longitudinal mixed models, further sensitivity analyses led to 3 categories of intake. Sodium was classified as less than 2500 mg (low), 2500 mg to less than 3500 mg (medium), and 3500 mg or more (high). Potassium was classified as less than 1800 mg (low), 1800 mg to less than 2400 mg (medium), and 2400 mg or more (high). The potassium to sodium ratio was classified as less than 0.6 (low), 0.6 to less than 0.75 (medium), and 0.75 or more (high). The same covariates were retained in the final longitudinal models. All analyses were performed using Statistical Analysis Systems software version 9.1 (SAS Institute Inc).
Table 1 shows baseline characteristics of NGHS girls according to sodium and potassium intakes. Girls in the highest sodium intake category (≥4000 mg/d) were predominantly black, from lower socioeconomic status families, and watched more television than those consuming less sodium (P < .001 for all). In contrast, girls with higher potassium intakes (≥2400 mg/d) were more often white, more active, had lower BMIs, and watched less television (P < .001 for all).
The eTable in the Supplement adds information on dietary intakes associated with sodium and potassium intakes. Individuals with the highest sodium and potassium intakes consumed the most calories as well as the most dairy, fruits and vegetables, and fiber (P < .001 for all). Higher sodium intakes were also associated with higher fat intakes, while consuming more potassium was linked with higher protein intakes.
Table 2 shows the effect of sodium intake on late adolescent BP. For both absolute intakes and energy-adjusted sodium residuals, there was no association with late-adolescent SBP or DBP among black or white girls.
Table 3 shows the beneficial effect of increased potassium intake on late-adolescent BP. Overall, girls in the highest category of potassium intake had statistically significantly lower late-adolescent SBP and DBP levels than those in lower categories (P = .02 for SBP and P = .05 for DBP). Effects were confounded by energy intake in black girls. After adjusting for total energy via the residual method, black girls in the highest potassium category had SBP levels that were about 1.5 mm Hg lower than those in the lowest potassium category (P = .07). Overall, both absolute and energy-adjusted potassium intakes were associated with lower SBP and DBP levels.
Table 4 shows a statistically significant decrease in SBP with an increasing potassium to sodium ratio (P = .04 for all girls). In race-stratified analyses (with less statistical power), results were similar but not statistically significant. There was no statistically significant effect on DBP.
The Figure shows BP changes throughout adolescence associated with sodium and potassium intake categories and the potassium to sodium ratio. There was no clear longitudinal effect of dietary sodium on either SBP or DBP. Those with higher sodium intakes (≥3500 mg/d) tended to have lower mean DBP levels. Girls with higher potassium intakes (≥2400 mg/d) had lower SBP and DBP levels throughout adolescence compared with girls with lower potassium intakes (P < .001). Effects of the potassium to sodium ratio were weaker and marginally statistically significant for SBP only.
Sodium was unrelated to SBP or DBP changes throughout adolescence or BP levels at the end of adolescence, even among participants with intakes more than 4000 mg/d. Further, there was no evidence that lower sodium intakes (<2500 mg/d) benefitted BP in black or white girls. In contrast, higher potassium consumption tended to be inversely associated with BP, with the beneficial effect of energy-adjusted potassium intake in black girls being stronger for SBP. The potassium to sodium ratio was inversely associated with SBP levels only.
Several cross-sectional studies in children and adolescents have examined dietary sodium and BP, including an earlier NGHS analysis that found no clear association between sodium or potassium and BP in 9- to 10-year-old girls.19 Results from different survey periods in the National Health and Nutrition Examination Surveys have been variable.20,21 However, a 2013 reanalysis of National Health and Nutrition Examination Surveys data across multiple survey periods found that higher sodium intakes (≥3754 mg per 2000 kcal/d) were associated with a higher prevalence of elevated BP but only among nonblack children and adolescents.22 In the cross-sectional UK National Diet and Nutrition Survey, SBP increased by 0.4 mm Hg for every gram-per-day increase in salt intake.23 Diastolic BP results were not shown. The well-known methodologic limitations of cross-sectional studies limit their usefulness. In particular, such studies are prone to confounding, as well as reporting bias, and the design makes it difficult to determine the direction of the etiologic relationship.
Prospective studies and clinical trials of sodium intake and BP in children and adolescents also yield inconsistent results. One of the few prospective studies, an early report of Dutch adolescents, found that sodium (measured by 24-hour excretion) was unrelated to BP.8 Clinical trials of sodium restriction in children have been largely characterized by short duration, small sample sizes, and high attrition rates. Null results were found in several short-term studies (3-4 weeks’ duration) of younger adolescents.24-27 In boys, small inverse effects on SBP and DBP were found in one 12-week study28 while another found no effect.29 A crossover trial carried out in 2 boarding schools reduced sodium intake through changes in school food service; this estimated 15% to 20% reduction in sodium intake led to a 1.7-mm Hg and 1.5-mm Hg decrease in SBP and DBP, respectively.30
To our knowledge, few data exist on the effects of potassium or the potassium to sodium ratio on childhood BP. In the earlier cited prospective study of Dutch adolescents, potassium excretion was inversely associated with SBP while a lower potassium to sodium ratio led to SBP increases.8 A review of 12 early observational studies concluded that the relationship between potassium intake/excretion and childhood BP is unclear.12 Similarly, smaller trials of potassium supplementation also found little effect on BP, although methodologic issues limited the interpretability of these results.29 In 1 well-controlled randomized trial in China, a potassium-calcium supplement suppressed the normal age-related increase in childhood BP,31 with greater beneficial effect evident among salt-sensitive children. A randomized crossover study of late adolescents/young adults with mild elevated BP found that potassium supplementation added to sodium restriction lowered BP while sodium restriction alone had no effect.32
Separating the effects of dietary sodium or potassium from associated diet patterns in observational studies is difficult. For example, dietary sodium may be closely linked with a less-healthy lifestyle, especially given the high sodium concentrations in processed foods and snacks. An adverse effect of dietary sodium could actually result from confounding by other dietary or behavioral risk factors. In this study, we controlled for the US Department of Agriculture food category of solid fats and added sugars as a marker of a less-healthy pattern and explored possible confounding by diet quality using Healthy Eating Index scores (no confounding was observed). We also controlled for other lifestyle and sociodemographic factors that may have affected diet and BP indirectly. Body mass index could be a potential causal intermediate between sodium or potassium and BP. Studies of adults33,34 and 1 trial in obese adolescents35 suggest that body weight modifies the effects of sodium on BP but we did not observe that in our analyses. After baseline and follow-up BMI were added to the multivariable models separately, both changed effect estimates by less than 5%, leading us to remove BMI from the final models. We carefully explored the effect of various individual and composite dietary factors to determine the independent effects of sodium and potassium on BP in these girls.
Although salt has historically been believed to be the major determinant of elevated BP, many questions remain. It has been shown that the BP response to dietary sodium is variable across individuals, leading some to conclude that sodium may only matter in salt-sensitive individuals. Because most people are not believed to be salt sensitive, the effects of sodium on BP may not be readily apparent in the general population. This could explain the null effect of sodium in the current study.
The interaction of sodium and potassium could alter BP through a number of mechanisms that involve effects on the kidney, fluid volume, mediators of vascular resistance, vasoconstriction, the renin-angiotensin-aldosterone system, and the sympathetic nervous system.36 Potassium has been shown to affect response to dietary sodium in both normotensive and hypertensive individuals.37 Further, the interaction between potassium and sodium in some studies has been shown to differ in black and white individuals.38 Potassium supplementation enhances the capacity of the kidneys to excrete sodium, thereby reducing fluid volume and lowering BP.39 Studies in animal models have shown that increases in extracellular sodium may promote endothelial cell stiffness while increases in potassium may reduce stiffness via formation of nitric oxide.40
The current study had several strengths, including its prospective design, with approximately 10 years of follow-up, extensive 3-day diet records, and repeated annual BP measurements. These analyses also examined a number of dietary (nutrients, foods, and diet patterns) and lifestyle factors as potential confounders. Sodium and potassium intake were also explored as energy-adjusted residuals to reduce potential confounding by energy intake.
The study was limited by the lack of direct measures of sodium or potassium via 24-hour urine collections. Dietary report is subject to error in that younger individuals have difficulty reporting portion sizes and details of dietary intake. There is also the possibility of reporting bias, as has been shown among those consuming the most snack foods (which are high in dietary sodium).41 Because the NGHS included only adolescent girls aged 9 to 21 years, we cannot extrapolate these findings to adolescent boys or younger children.
This prospective study showed that black and white adolescent girls who consumed more dietary potassium had lower BPs in later adolescence. In contrast, the data indicated no overall effect of sodium intake alone on BP, and, thus, do not support the call for a global reduction in sodium intake among children and adolescents. This study emphasizes the need to develop methods for estimating salt sensitivity to be used in future studies of high-risk populations and points to the potential health risks associated with the existing low dietary potassium intakes among US children and adolescents.
Corresponding Author: Lynn L. Moore, DSc, MPH, Department of Medicine, Boston University School of Medicine, 801 Massachusetts Ave, Ste 470, Boston, MA 02118 (email@example.com).
Accepted for Publication: February 3, 2015.
Published Online: April 27, 2015. doi:10.1001/jamapediatrics.2015.0411.
Author Contributions: Dr Moore had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Buendia, Daniels, Moore.
Acquisition, analysis, or interpretation of data: Buendia, Bradlee, Singer, Daniels, Moore.
Drafting of the manuscript: Buendia, Daniels, Moore.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Buendia, Singer, Moore.
Obtained funding: Moore.
Administrative, technical, or material support: Buendia, Bradlee, Moore.
Study supervision: Daniels, Moore.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by grant R21DK075068 from the National Institute of Diabetes and Digestive and Kidney Diseases and grant DRI1110 from the National Dairy Council and Dairy Council of California.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content of this study is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases or the National Institutes of Health.
Additional Information: This article was prepared using National Heart, Lung, and Blood Institute Growth and Health Study research data obtained from the National Heart, Lung, and Blood Institute.