Independent associations of sodium intake with measures of body size and predictive body fatness
Disclosure: The authors declared no conflict of interest.
Author Contributions: SY was responsible for the study concept, preliminary literature review, writing the manuscript, and overseeing the data analysis. MF updated the literature, conducted the data analysis, and provided text/edits to the manuscript. JB provided conceptual oversight and commentary and edits to the manuscript. All authors had final approval of the submitted version.
Stella S. Yi is currently at Department of Population Health, New York University School of Medicine, New York, USA
Abstract
Objective
Observational studies highlight a possible relationship between sodium intake and obesity. This investigation explores the cross-sectional relationships between sodium intake and measures of body size and fatness (body mass index [BMI], weight, waist circumference, predictive body fatness).
Methods
Analyses were performed using data from participants in the National Health and Nutrition Examination Survey (NHANES) 2009-10 with two 24-h dietary recalls and measures of body size and fatness (n = 4,613). Regression analyses assessed the relationships of sodium (1,000 mg/day) with outcomes, adjusting for caloric intake. Analyses are presented overall and by sex; data were weighted to be representative of the non-institutionalized US adult population.
Results
Positive associations between sodium intake and measures of body size and predictive body fatness were observed, and the magnitude of association was larger in women than in men. For each 1,000 mg/day higher sodium intake, BMI was 1.03 kg/m2 higher; weight was 2.75 kg higher; waist circumference was 2.15 cm higher; and predictive body fatness was 1.18% higher after adjustment for energy intake.
Conclusions
Longitudinal analyses examining associations between sodium intake and measures of body size and body fatness are needed.
Introduction
Recent analyses in the National Health and Nutrition Examination Survey (NHANES) reported a 20% (HR, 1.2, 95% CI, 1.03-1.41 per 1,000 mg/day) increased risk of all-cause mortality with higher sodium intake (1). Historically, studies have focused on adverse effects of sodium on blood pressure and cardiovascular disease (2). Few studies have examined the direct relationship between sodium and measures of body size and fatness in the general population. High levels of sodium increase adipocyte mass and insulin sensitivity in rats (3), and are associated with insulin resistance, metabolic syndrome, and inflammation in humans (4, 5). Two studies demonstrated a positive relationship between higher sodium intake, as measured by 24-h dietary recalls, and risk of being overweight (6) or obese in Korean adults (7). The Monitoring Trends in Cardiovascular Disease (MONICA) study found that over a six-year period, increased 24-h urinary sodium excretion was associated with increased body fat, decreased fat free mass, but not with waist circumference or weight among Danish adults (8).
Sodium intake may also influence body size and fatness through higher caloric intake since sodium intake is higher as more calories are consumed. Increased sodium intake is related to increased fluid consumption in adults. While some studies (6-8) have controlled for caloric intake, others have not. The objective of the current analysis was to assess the cross-sectional associations of sodium intake with measures of body size (body mass index (BMI), waist circumference, weight) and predictive body fatness after accounting for total energy intake in a population-based sample of US adults.
Methods
The NHANES is a nationally representative survey designed to assess the health and nutritional status of non-institutionalized civilians in the United States. Sampling methodology has been described previously (9). Briefly, NHANES participants completed a household interview to collect demographic and general health information followed by a visit to a Mobile Examination Center for anthropometric measures and a dietary interview. Trained interviewers administered the first dietary recall using the USDA multiple-pass method in person with a computer-assisted interview to standardize procedures, and a second dietary interview was conducted by telephone 3-10 days later.
The 2009-2010 NHANES had a sample size of 10,537. Participants were excluded from the analysis if they were <20 years of age (n = 4,319), were missing body measures (n = 523), were pregnant (n = 58), did not complete two 24-h dietary recalls (n = 859), or had an average energy intake outside the range of 500-3,500 kcal/day for women and 800-4,000 kcal/day for men (10) (n = 165). The final analytic sample size was n = 4,613.
Average sodium intake (mg/day) and energy consumption (kcal/day) from the two recalls were computed. BMI was calculated from weight and height (kg/m2). Predictive body fatness was calculated using sex-specific formulas that were recently evaluated in the NHANES population and were shown to perform best for assessing body fat in general population samples (males: Kagawa equation; females: Gomez-Ambrosi equation; see below) (11).
Statistical analysis
Analyses were stratified by sex, due to significant differences in sodium and calorie intake and anthropometry in men and in women. Mean (± standard error) sodium, calories, BMI, weight, waist circumference, and predictive percent body fat were calculated. Pearson correlations were calculated to examine associations between sodium, calories, and body measures. Multivariable linear regression models were constructed for each of the four continuous body measure outcome variables. The crude model examined the relationship between sodium and body measures without adjustment. Model 1 was adjusted for age, sex (in overall models only), race/ethnicity, and education, and Model 2 included Model 1 covariates plus energy intake using the residuals method. In the residuals method (12), sodium is regressed on calories, and the calculated regression residuals for each observation are added to expected sodium values. The standard multivariable method of adding calories as a continuous covariate was also conducted. Analyses incorporated sampling weights (SUDAAN, version 10.0; Research Triangle Institute, Research Triangle Park, North Carolina).
Results
Among non-pregnant US adults aged 20+ years, mean sodium intake was 3,543.8 ± 35.1 mg/day and mean calorie intake was 2,088.1 ± 20.1 kcal/day (Table 1). Daily sodium intake and calorie consumption were significantly higher among males than females (P-value <0.001). Weight and waist circumference were higher among males compared to females (P-value <0.001). There was no significant difference in BMI by sex. Sodium intake was highly correlated with energy intake overall (r = 0.80, P < 0.01) and when stratified by sex (males: r = 0.77, P < 0.01; females: r = 0.75, P < 0.01). Sodium intake and energy intake were not correlated with BMI (sodium: r = 0.03, P = 0.06; energy intake: r = −0.02, P = 0.17).
| Overall | Males | Females | Male vs. female | ||||
|---|---|---|---|---|---|---|---|
| Mean | SE | Mean | SE | Mean | SE | P-value | |
| Sodium intake (mg/day) | 3543.8 | 35.1 | 4104.8 | 50.7 | 3016.1 | 35.6 | <0.001 |
| Calories (kcal/day) | 2088.1 | 20.1 | 2412.8 | 31.7 | 1782.6 | 16.3 | <0.001 |
| Body mass index (kg/m2) | 28.8 | 0.1 | 28.9 | 0.3 | 28.7 | 0.2 | 0.51 |
| Weight (kg) | 82.1 | 0.5 | 89.6 | 0.8 | 75.3 | 0.5 | <0.001 |
| Waist circumference (cm) | 98.4 | 0.5 | 101.5 | 0.7 | 95.6 | 0.5 | <0.001 |
| Predictive body fatness (%) | 35.5 | 0.2 | 30.5 | 0.4 | 40.1 | 0.2 | <0.001 |
- Sodium and calories are calculated as the mean from two 24-h dietary recalls.
In adjusted analyses, a difference of 1,000 mg/day of sodium was associated with higher body measure values for all outcomes (BMI: 1.03 kg/m2; weight: 2.75 kg; waist circumference: 2.15 cm; predictive body fatness: 1.18%) even after accounting for energy intake (Table 2). The differences per 1,000 mg/day of sodium were larger in females than in males. For example, a difference of 1,000 mg/day of sodium was associated with a 0.56 kg/m2 higher BMI in females, but a 0.37 kg/m2 higher BMI in males. Results did not differ when models were adjusted for energy intake using the multivariable adjustment compared to the residual method adjustment.
| Body mass index (kg/m2) | Weight (kg) | Waist circumference (cm) | Predictive body fatness (%) | |||||
|---|---|---|---|---|---|---|---|---|
| Difference | 95% CI | Difference | 95% CI | Difference | 95% CI | Difference | 95% CI | |
| Crude | ||||||||
| Overall | 0.29* | (0.04, 0.53) | 3.45*** | (2.57, 4.34) | 1.38*** | (0.75, 2.01) | −1.35*** | (−1.59, −1.12) |
| Male | 0.28* | (0.02, 0.55) | 1.51** | (0.56, 2.46) | 0.48 | (−0.24, 1.20) | 0.17 | (−0.17, 0.51) |
| Female | 0.33 | (−0.03, 0.70) | 1.66** | (0.66, 2.67) | 0.78 | (−0.01, 1.58) | 0.00 | (−0.53, 0.53) |
| Model 1a | ||||||||
| Overall | 0.47** | (0.24, 0.71) | 1.72*** | (1.00, 2.44) | 1.12*** | (0.61, 1.63) | 0.44* | (0.16, 0.73) |
| Male | 0.37** | (0.14, 0.60) | 1.47** | (0.66, 2.28) | 0.87* | (0.24, 1.50) | 0.36* | (0.05, 0.67) |
| Female | 0.56** | (0.21, 0.91) | 1.92** | (0.85, 2.99) | 1.40** | (0.68, 2.13) | 0.43 | (−0.01, 0.87) |
| Model 2b | ||||||||
| Overall | 1.03*** | (0.70, 1.35) | 2.75*** | (1.90, 3.60) | 2.15*** | (1.41, 2.90) | 1.18*** | (0.73, 1.64) |
| Male | 0.81** | (0.38, 1.24) | 2.51** | (1.12, 3.90) | 1.85** | (0.70, 3.00) | 0.97** | (0.36, 1.59) |
| Female | 1.32*** | (0.74, 1.90) | 3.02** | (1.36, 4.68) | 2.48** | (1.04, 3.93) | 1.48*** | (0.78, 2.17) |
- a Adjusted for age, race/ethnicity, sex (in overall models), education.
- b Adjusted for age, race/ethnicity, sex (in overall models), education, and after accounting for the residuals of sodium regressed on calories.
- *P < 0.05, **P < 0.01, ***P < 0.001
Discussion
Sodium intake was associated with measures of body size and fatness after adjustment for energy intake in this nationally representative sample. For each 1,000 mg/day higher sodium intake, BMI was 1.03 kg/m2 higher; weight was 2.75 kg higher; waist circumference was 2.15 cm higher; and predictive body fatness was 1.18% higher. Each of these differences was larger in women than in men.
Studies investigating the association between sodium and body size in adults are few (4, 6, 7, 13). A prospective analysis from the Dallas Heart Study examined associations between the sodium–potassium ratio derived from first morning spot urine samples with total-body percentage fat measured by dual-energy X-ray absorptiometry in a multiethnic cohort (14). The authors found a positive association between the sodium–potassium ratio and total-body percentage fat (0.43 percent units, 95% CI: 0.15, 0.72 per 3-unit change in ratio). Similarly, a longitudinal analysis using data from the MONICA project demonstrated positive relationships between 24-h urinary sodium excretion (per 100 mmol) and subsequent increases in body fat (0.24 kg, 95% CI: 0.05 to 0.43, P = 0.02) and decreases in fat free mass (−0.21 kg, 95% CI: −0.40 to −0.01, P = 0.04), but only after adjustment for demographics, change in body weight, and calories (8). No associations were observed in crude models of sodium excretion and body fat or fat free mass, or between sodium excretion and changes in body weight or waist circumference. Similar to the results observed in the current study, coefficients became larger after adjustment for caloric intake, indicating a negative confounding effect of calories on the relationship between sodium and measures of body size.
Strengths of this study include a sample representative of the US non-institutionalized population and the inclusion of body size measures beyond BMI. This analysis also accounted for differences by sex, a topic not covered in prior analyses. A limitation of this cross-sectional investigation, however, is that reverse causality (i.e., the impact of obesity on dietary reporting of sodium or energy intakes) (15, 16) cannot be ruled out. However the corroboration of results across different measures of body size, and the weak to nonexistent correlations between BMI and energy intake (and sodium intake) lends credibility to these results.
To close, sodium intake was associated with measures of body size and body fat in this cross-sectional analysis. These results and those of previous studies point to either a potentially larger impact of sodium intake on the body measures of women, or differences in the associations due to differential misreporting of calorie intake by sex. Future research should consider sex as an effect modifier, and should include accurate assessments of sodium and caloric intake to account for differential reporting by obesity status. Additional longitudinal studies to establish temporality of sodium on measures of body size and body fat are warranted.
