Relationships Between Blood Pressure and 24‐Hour Urinary Excretion of Sodium and Potassium by Body Mass Index Status in Chinese Adults

Abstract

This study examined the impact of overweight/obesity on sodium, potassium, and blood pressure associations using the Shandong‐Ministry of Health Action on Salt Reduction and Hypertension (SMASH) project baseline survey data. Twenty‐four–hour urine samples were collected in 1948 Chinese adults aged 18 to 69 years. The observed associations of sodium, potassium, sodium‐potassium ratio, and systolic blood pressure (SBP) were stronger in the overweight/obese population than among those of normal weight. Among overweight/obese respondents, each additional standard deviation (SD) higher of urinary sodium excretion (SD=85 mmol) and potassium excretion (SD=19 mmol) was associated with a 1.31 mm Hg (95% confidence interval, 0.37–2.26) and −1.43 mm Hg (95% confidence interval, −2.23 to −0.63) difference in SBP, and each higher unit in sodium‐potassium ratio was associated with a 0.54 mm Hg (95% confidence interval, 0.34–0.75) increase in SBP. The association between sodium, potassium, sodium‐potassium ratio, and prevalence of hypertension among overweight/obese patients was similar to that of SBP. Our study indicated that the relationships between BP and both urinary sodium and potassium might be modified by BMI status in Chinese adults.

Cardiovascular disease (CVD), including heart disease and stroke, accounted for 41% of all deaths and 23% of national healthcare spending in China in 2005.1 Unfortunately, the prevalence of hypertension, the leading risk factor for CVD, has increased dramatically in China, from 5.1% in 1959 to 29.6% in 2009 among adults aged 18 years and older.2, 3 High sodium intake, low potassium intake, and obesity all increase the risk of hypertension.4, 5, 6 In China, sodium intake decreased over the past two decades, but studies suggest that average daily intake in 2009 (4700 mg/d) was almost double the Chinese Nutrition Association recommendation (2400 mg/d).7 The average daily potassium intake among Chinese adults is very low, resulting in a high sodium‐potassium ratio, indeed one of the highest in the world.5, 7, 8 In addition, overweight and obesity have increased remarkably in China over the past two decades; in 2010, 30.6% of Chinese adults were overweight and 12% were obese.9, 10

The US Institute of Medicine recommended that analyses on the effects of dietary sodium on health outcomes should also examine potassium intake and account for potential confounding factors.11 In addition, accumulating evidence has suggested that in humans the response of blood pressure (BP) to changes in sodium intake may vary by body weight/BMI or the presence of the metabolic syndrome.12, 13, 14, 15, 16 In China, several population‐based studies have linked sodium and potassium intake and their ratio with BP and risk of hypertension.17, 18, 19 However, few studies have examined the relationship between sodium and potassium intake and BP in analyses stratified by BMI status.20, 21 The INTERSALT study examined the Chinese subsample to assess the effect modification of BMI on the association of sodium and BP.20 However, the sample was relatively small and was based on just three Chinese neighborhoods. In addition, the sample was collected 20 to 30 years ago, when the prevalence of hypertension and obesity in China were relatively low. Therefore, we sought to understand how the association of sodium and BP might be modified by BMI status among Chinese adults in the contemporary setting.

We used baseline survey data from 2011 obtained as part of the Shandong and Ministry of Health Action on Salt and Hypertension (SMASH) project to examine the association between (1) urinary sodium, urinary potassium, and their ratio; and (2) BP and the prevalence of hypertension. We also explored whether the associations of sodium and potassium with BP and hypertension prevalence are modified by BMI.

Methods

Study Population

The SMASH project survey was a representative cross‐sectional survey of adults aged 18 to 69 years in Shandong Province, which had 96 million residents in 2010. Using a complex, multistratified cluster sampling method, we selected 2112 participants for a timed 24‐hour urine collection from 20 counties/districts across Shandong province. A detailed description of the study design and methods for the SMASH project survey is available elsewhere.22 Our study was conducted according to Declaration of Helsinki guidelines, and all procedures involving human subjects were approved by the ethics committee of the Shandong Center for Disease Control and Prevention. Participants provided written informed consent.

24‐Hour Urine Collection

We used the INTERSALT method to perform the 24‐hour urine collection.23 Each participant was instructed on this process by a trained health professional at the collection field sites. The participant was given a standard plastic container with boric acid (around 1 g) used as preservative and was told to discard the first void of the day and collect all the urine in the container during the following 24‐hour period including the first void from the next morning. The health professional recorded the beginning and ending time for each urine collection and the total hours between the first and last void collected. A standard interview by questionnaire was administered to each participant at the end of the 24‐hour period to assess the completeness of urine collection. The volume of urine was measured on a standard platform at the field site by a laboratory technician. The samples were kept in a freezer at −20°C and delivered to the certified laboratory (ADICON Clinical Laboratory Inc, Shanghai, China) in Jinan, Shandong. Urinary sodium and potassium were measured by the ion‐selective electrode method on an Olympus AU 680 Chemistry‐Immuno Analyzer (Shinjuku, Tokyo). The coefficient of variation was 1.5% for sodium and 2.5% for potassium. Urinary creatinine was measured by the picric acid method using the Olympus AU640 Chemistry‐Immuno Analyzer (here the coefficient of variation was 3.0%). Individual sodium and potassium excretion values were calculated from their concentration in the 24‐hour sample.

Anthropometric and BP Data

Weight, height, and BP were measured by trained health professionals using standardized methods.22 According to Chinese guidelines for overweight and obesity,24 normal weight is defined as a BMI ≥18.5 kg/m² to <24 kg/m², while overweight is defined as a BMI of 24 kg/m² to <28 kg/m². Patients with a BMI ≥28 kg/m² are considered obese. We combined overweight and obese participants in a single group (overweight/obese) to produce more stable estimates.

BP was measured with the patient in the sitting position three times 5 minutes apart on one occasion by trained health professionals with an electronic sphygmomanometer (HEM‐7071; Omron Healthcare, Inc, Lake Forest, IL). The average of the three measurements was used. Hypertension was defined by the Chinese guidelines for the management of hypertension and the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) as a systolic BP (SBP) ≥140 mm Hg or a diastolic BP (DBP) ≥90 mm Hg, or a self‐report of taking antihypertensive medication(s) in the past 2 week.25, 26

Data Analysis

An incomplete urine collection was defined as either a 24‐hour urinary volume <500 mL or a 24‐hour urinary creatinine volume beyond the sex‐specific mean±2 standard deviations (SDs). We identified 88 participants with incomplete 24‐hour urine collection and excluded them from our analytic sample. An additional 76 participants who were underweight (BMI <18.5 kg/m²) were also excluded, leaving a total of 1948 adults for the data analysis.

The covariates in this analysis included age, sex, educational attainment, smoking status, frequency of drinking alcohol, physical activity, and use of antihypertensive medication(s). Data on these variables were collected by in‐person interview using a standardized questionnaire. Frequency of drinking alcohol was categorized in four groups: nondrinker, <1 d/wk, 1 to 2 d/wk, or ≥3 d/wk. Physical activity was defined by the frequency of leisure‐time exercise: none, <3 d/wk, or ≥3 d/wk. Tobacco smoking status was categorized as none, former smoker, or current smoker. The missing values were coded as a missing group within the covariates. Ninety participants had missing values for covariates: 18 for smoking, 14 for physical activity, and 60 for drinking (two participants had two missing values and 88 had one missing value).

Statistical Analysis

We calculated the weighted mean of the 24‐hour urinary sodium output, potassium output, and the sodium‐potassium ratio, by the selected covariates and BMI. The sample weights were determined by the design weight and a post‐stratification weight adjustment to correct for oversampling or undersampling.22, 27 The Wald‐F test was used to assess the differences across the categories of the covariates. We used multivariable linear regression to examine the associations of urinary sodium output, potassium output, and the sodium‐potassium ratio with SBP and DBP. We estimated the adjusted β‐coefficients associated with having a 1‐SD higher urinary sodium (SD=85 mmol), a 1‐SD higher potassium (SD=19 mmol), and a 1‐unit difference in the sodium‐potassium ratio among all adults and by BMI status.

The dose‐response relationship between urinary sodium excretion and BP while adjusting for covariates was tested for linearity through the fitting of a restricted cubic spline function.28 There is no evidence of significant departure from linearity (P=.07–.11). We calculated the middle value of each quintile (ie, 10th, 30th, 50th, 70th, and 90th percentiles) of 24‐hour urinary sodium excretion in the total population and by BMI percentile, and we used the coefficients from the linear regression models to estimate the adjusted mean of the SBP/DBP associated with each of these values by BMI status.29 Similarly, for the SBP/DBP models with potassium or the sodium‐potassium ratio, restricted cubic splines were also not significant (P=.25–.84); thus, we assumed a linear dose‐response relationship for all three urinary measurements in relation to BP.

We used logistic regression to estimate the adjusted odds ratio (OR) for the associations of 24‐hour urinary sodium, potassium, and the sodium‐potassium ratio with hypertension. Again, restricted cubic splines were not significant (P=.47–.63), suggesting linear dose‐response relationships of 24‐hour urinary sodium, potassium, and the sodium‐potassium ratio for prevalence of hypertension. Thus, we estimated the ORs comparing the mid‐value of each quintile with the lowest quintile (Q5, Q4, Q3, Q2 vs Q1) while adjusting for covariates.

In both linear and logistic regression, we used three models to adjust for covariates. In model 1, we adjusted for age, sex, rural/urban residency, and region; in model 2, in addition to the covariates in model 1, we adjusted for educational attainment, smoking status, drinking of alcohol, physical activity, and use of antihypertensive medication; in model 3, we further adjusted for BMI as a continuous variable. In models 1 to 3 for sodium excretion, we included potassium as a covariate, and for potassium excretion, we included sodium as a covariate.

We tested interactions between urinary sodium and potassium excretion, their ratio, and the broad BMI classification (normal vs overweight/obese) by including the interaction terms in the linear or logistic regression models. We found significant interactions by BMI group for the associations between urinary sodium excretion and SBP (P=.04) and DBP (P=.03) and a marginally significant interaction for the association of the sodium‐potassium ratio and SBP (P=.07). We found a significant interaction by BMI classification for the association between the sodium‐potassium ratio and hypertension (P=.001) but not for sodium (P=.10) or potassium (P=.61) alone.

Sensitivity Analysis

We conducted several sensitivity analyses. First, we repeated our analysis by excluding the known hypertensive participants who were taking an antihypertensive medication. Awareness of their hypertension status might have changed their intake of sodium (Table S1). Second, we also repeated our analysis by excluding participants who had self‐reported CVD or diabetes (n=71) at baseline (Table S2). Third, we examined the distribution of 24‐hour urinary volume and urinary creatinine by the selected characteristics to assess the possible effects on the completeness of the urine collection (Table S3). Fourth, we repeated the analysis using alternative measures of incompleteness of urine collection based on excretion of urinary creatinine and other variables (Tables S4–S8). Finally, we repeated the analysis by stratifying participants into three groups: normal weight, overweight, and obese (Tables S9–S11).

Statistical analyses were performed with SAS 9.3 (SAS Institute Inc, Cary, NC). All tests were two‐sided and a P value <.05 was considered significant.

Results

The mean age of the 1948 participants was 41.4 years (SD=13.9). Of them, 907 were classified as normal weight, 684 were overweight, and 357 were obese. More than half of the population (53.4%) was overweight or obese, and 22.9% had hypertension.

In the population, the mean 24‐hour urinary sodium excretion was 235.7 (mmol) (95% confidence interval [CI], 231.6–239.9), and the mean potassium excretion was 40.5 (mmol) (95% CI, 39.6–41.5). The mean sodium‐potassium ratio was 6.8 (95% CI, 6.6–7.0). Mean 24‐hour urinary sodium and potassium excretion were both significantly higher among overweight/obese adults than among those of normal weight, although mean sodium‐potassium ratio did not significantly differ (Table 1).

Table 1.

The Weighted Mean of 24‐Hour Sodium Excretion, Potassium Excretion, and Sodium‐Potassium Ratio by BMI Status

Characteristic	All Participants (N=1948)			Normal Weighta (n=907)			Overweight/Obesea (n=1041)
	Na, mmol	K, mmol	Na/K	Na, mmol	K, mmol	Na/K	Na, mmol	K, mmol	Na/K
All	235.7	40.5	6.8	223.6	38.9	6.7	246.4	42.0	6.9
Age, y
<50	235.4	40.4	6.8	222	38.7	6.7	247.9	41.9	6.9
≥50	236.4	41.0	6.8	227.9	39.4	6.6	243.1	42.2	6.9
Sex
Male	244.3	39.8	7.1	230	38.7	6.8	257	40.7	7.4
Female	226.3b	41.4	6.5b	216.6b	39.1	6.6	234.8b	43.4b	6.4b
Hypertension
Yes	245.3	40.1	7.3	223.8	39.1	6.7	243.5	42.5	6.6
No	232.9c	40.7	6.7c	222.7	37.3	6.7	252.7	41	7.5c

Abbreviations: K, potassium; Na, sodium. ^aAccording to the Chinese guidelines for overweight and obesity, normal weight is defined as a body mass index (BMI) of ≥18.5 kg/m² to <24 kg/m², while overweight is defined as a BMI of 24 kg/m² to <28 kg/m². Patients with a BMI ≥28 kg/m² are considered obese. ^bMale patients significantly different than female patients within the total, normal weight, and overweight populations. ^cHypertensive patients significantly different than nonhypertensive patients within the total, normal‐weight, and overweight populations.

Among all adults, each SD increase in 24‐hour urinary sodium excretion value (SD=85 mmol) was associated with a significant increase of 1.09 mm Hg (95% CI, 0.22–1.96) in SBP after adjusting for age, sex, urban/rural status, region, educational attainment, smoking status, drinking, physical activity, use of antihypertensive medication, and urinary excretion of potassium. After additional adjustment for BMI, the association between sodium excretion and SBP was not significant (Table 2).

Table 2.

Adjusted Association of Sodium Excretion, Potassium Excretion, and Sodium‐Potassium Ratio With Systolic Blood Pressure by BMI Status

	Na		K		Na/K
	β‐Coefficient (95% CI)a	P Value	β‐Coefficient (95% CI)a	P Value	β‐Coefficient (95% CI)a	P Value
All population
Model 1b	1.34 (0.42–2.25)	.007	−0.96 (−1.75 to −0.17)	.021	0.49 (0.33–0.64)	<.001
Model 2c	1.09 (0.22–1.96)	.018	−0.85 (−1.53 to −0.16)	.019	0.43 (0.30–0.57)	<.001
Model 3d	0.46 (−0.33 to 1.26)	.231	−0.88 (−1.58 to −0.19)	.016	0.37 (0.26–0.49)	<.001
Normal‐weight populatione
Model 1b	−0.39 (−1.81 to 1.03)	.566	−0.27 (−1.58 to 1.05)	.67	0.15 (−0.15 to 0.44)	.307
Model 2c	−0.55 (−2.00 to 0.89)	.425	−0.09 (−1.27 to 1.08)	.865	0.12 (−0.17 to 0.41)	.395
Model 3d	−0.66 (−2.01 to 0.69)	.313	−0.21 (−1.45 to 1.02)	.717	0.12 (−0.18 to 0.43)	.403
Overweight/obese populatione
Model 1b	1.90 (0.85–2.94)	.002	−1.74 (−2.55 to −0.93)	<.001	0.67 (0.42–0.91)	<.001
Model 2c	1.65 (0.73–2.58)	.002	−1.60 (−2.39 to −0.81)	.001	0.61 (0.38–0.83)	<.001
Model 3d	1.31 (0.37–2.26)	.01	−1.43 (−2.23 to −0.63)	.002	0.54 (0.34–0.75)	<.001

Abbreviations: CI, confidence interval; K, potassium; Na, sodium. ^aβ‐coefficients for 24‐hour urine sodium and potassium are presented as per one standard deviation; the estimated population standard deviation for 24‐hour sodium and 24‐hour potassium was 85.0 mmol and 19.0 mmol, respectively. P values for interactions between 24‐hour sodium, potassium excretion, and body mass index (BMI) status were 0.04 and 0.14 for systolic blood pressure and 0.03 and 0.49 for diastolic blood pressure, respectively. ^bModel 1 adjusted for age, sex, urban‐rural, and regions. We also included potassium excretion in the regression model for sodium and sodium in the regression model for potassium. ^cModel 2 adjusted for all factors in model 1 plus educational attainment, smoking status, alcohol intake, physical activity, and antihypertensive medication use. ^dModel 3 adjusted for all factors in model 2 plus BMI as a continuous variable. ^eAccording to the Chinese guidelines for overweight and obesity, normal weight is defined as a BMI of ≥18.5 kg/m² to <24 kg/m², while overweight is defined as a BMI of 24 kg/m² to <28 kg/m². Patients with a BMI ≥28 kg/m² are considered obese.

Each rise of 1 SD in potassium excretion (SD=19 mmol) was associated with a significantly lower SBP (−0.88 mm Hg; 95% CI, −1.58 to −0.19) when sodium excretion and all the covariates were included in the analysis (Table 2). The association between DBP and 24‐hour urinary sodium excretion was not significant, as was the case for the association of DBP and potassium excretion (Table 3). The sodium‐potassium ratio was positively and significantly associated with increases in SBP and DBP, with or without adjustment for BMI (Table 2 and Table 3). We observed stronger and more consistent associations between BP and urinary sodium excretion, potassium excretion, and their ratio in the overweight/obese population than among those of normal weight (Table 2 and Table 3).

Table 3.

Adjusted Association of Sodium Excretion, Potassium Excretion, and Sodium‐Potassium Ratio With Diastolic Blood Pressure by BMI Status

	Na		K		Na/K
	β‐Coefficient (95% CI)a	P Value	β‐Coefficient (95% CI)a	P Value	β‐Coefficient (95% CI)a	P Value
All population
Model 1b	0.89 (0.10–1.67)	.03	−0.53 (−1.30 to 0.24)	.16	0.26 (0.09–0.42)	.005
Model 2c	0.69 (−0.07 to 1.44)	.071	−0.46 (−1.14 to 0.22)	.172	0.22 (0.07–0.37)	.006
Model 3d	0.05 (−0.61 to 0.71)	.864	−0.50 (−1.11 to 0.11)	.103	0.16 (0.03–0.29)	.02
Normal‐weight populatione
Model 1b	−0.39 (−1.55 to 0.77)	.478	−0.39 (−1.53 to 0.74)	.471	0.07 (−0.20 to 0.34)	.598
Model 2c	−0.60 (−1.68 to 0.49)	.257	−0.29 (−1.29 to 0.71)	.545	0.04 (−0.21 to 0.29)	.719
Model 3d	−0.69 (−1.69 to 0.32)	.166	−0.39 (−1.42 to 0.64)	.428	0.05 (−0.21 to 0.30)	.707
Overweight/obese populatione
Model 1b	1.06 (0.36–1.75)	.006	−0.86 (−1.54 to −0.18)	.017	0.33 (0.17–0.48)	<.001
Model 2c	0.91 (0.27–1.54)	.008	−0.77 (‐1.42 to −0.13)	.022	0.30 (0.15–0.44)	.001
Model 3d	0.59 (−0.04 to 1.22)	.064	−0.62 (−1.26 to 0.03)	.062	0.24 (0.10–0.38)	.002

Abbreviations: CI, confidence interval; K, potassium; Na, sodium. ^aβ‐coefficient for the sodium‐potassium ratio is presented as per 1 unit change. P values for interactions between sodium‐potassium ratio and body mass index (BMI) status were 0.07 for systolic blood pressure and 0.12 for diastolic blood pressure, respectively. ^bModel 1 adjusted for age, sex, urban‐rural, and regions. ^cModel 2 adjusted for all factors in model 1 plus educational attainment, smoking status, alcohol intake, physical activity, and antihypertensive medication use. ^dModel 3 adjusted for all factors in model 2 plus BMI as a continuous variable. ^eAccording to the Chinese guidelines for overweight and obesity, normal weight is defined as a BMI of ≥18.5 kg/m² to <24 kg/m², while overweight is defined as a BMI of 24 kg/m² to <28 kg/m². Patients with a BMI ≥28 kg/m² are considered obese.

By quintile of urinary sodium, the adjusted SBP increased from 120.9 mm Hg (95% CI, 119.6–122.1) in quintile 1 to 121.6 mm Hg (95% CI, 119.1–124.1) in quintile 5 (P=.313). Among overweight/obese men and women, it increased from 124.1 mm Hg (95% CI, 122.2–126.0) in quintile 1 to 127.1 mm Hg (95% CI, 125.2–128.9) in quintile 5 (P=.010) (Figure).

Figure 1.

Systolic blood pressure values by body mass index (BMI) status and 24‐hour sodium excretion, 24‐hour potassium excretion, and sodium‐potassium ratio. Data source: Shandong and Ministry of Health Action on Salt and Hypertension (SMASH) project 2011 survey, among adults (N=1948) in Shandong Province, China. According to the Chinese guidelines for overweight and obesity, overweight is defined as a BMI of 24 to <28. Those with a BMI ≥28 are considered obese. Footnote: We updated Figure footnote as ‘We used 10th, 30th, 50th, 70th, and 90th percentiles as the mid‐value of each quintile for sodium and potassium excretion and sodium‐potassium ratio. The value for P10, P30, P50, P70, and P90 for sodium was 137.0, 188.9, 230.3, 255.4, and 341.3 mmol for the entire population, 129.9, 178.6, 217.9, 249.5, and 312.5 mmol among the normal‐weight population, and 144.4, 201.3, 239.8, 257.9, and 365.4 mmol among the overweight/obese population. The value for P10, P30, P50, P70, and P90 for potassium was 19.7, 29.2, 37.7, 47.6, and 65.0 mmol for the entire population, 19.2, 28.3, 36.5, 45.5, and 61.9 mmol among the normal‐weight population, and 20.1, 30.3, 39.0, 49.2, and 68.0 mmol among the overweight/obese population. The value for P10, P30, P50, P70, and P90 for sodium‐potassium ratio was 3.4, 4.7, 5.8, 7.7, and 11.0 for the entire population, 3.3, 4.6, 5.7, 6.4, and 10.9 among the normal‐weight population, and 3.5, 4.7, 5.9, 7.8, and 11.2 among the overweight/obese population.’

In a comparison of the highest and the lowest quintiles of sodium excretion, the OR for prevalence of hypertension in the population as a whole, after adjustment for all covariates except for BMI, was 1.62 (95% CI, 1.13–2.31). After additional adjustment for BMI, it was 1.30 and no longer significant (95% CI, 0.91–1.85). In the same comparison of quintiles, the fully adjusted OR (model 3) for potassium excretion was 0.62 (95% CI, 0.42–0.91), and for the sodium‐potassium ratio was 1.54 (95% CI, 1.23–1.92) (Table 4). The associations of sodium, potassium, and the sodium‐potassium ratio with hypertension appeared to be stronger in the overweight/obese population than among those of normal weight (Table 4).

Table 4.

Adjusted Odds for Risk of Hypertension by BMI Status and Quintile of Sodium Excretion, Potassium Excretion, and Sodium‐Potassium Ratio

	OR (95% CI)a					P_trend b	Overall OR(95% CI)a
	Q1	Q2	Q3	Q4	Q5
All population
Na
Mid‐value of quintile/range	137.0	188.9	230.3	255.4	341.3		22.1, 750.8
Model 1c	1.00	1.14 (1.05–1.23)	1.26 (1.09–1.46)	1.34 (1.12–1.62)	1.66 (1.21–2.29)	.001	1.24 (1.08–1.41)
Model 2d	1.00	1.13 (1.03–1.24)	1.25 (1.06–1.47)	1.32 (1.08–1.63)	1.62 (1.13–2.31)	.004	1.22 (1.05–1.42)
Model 3e	1.00	1.07 (0.98–1.17)	1.13 (0.96–1.32)	1.16 (0.95–1.43)	1.30 (0.91–1.85)	.110	1.11 (0.96–1.29)
K
Mid‐value of quintile/range	19.7	29.2	37.7	47.6	65.0		25.4, 227.0
Model 1c	1.00	0.92 (0.86–0.99)	0.86 (0.75–0.99)	0.79 (0.64–0.98)	0.69 (0.49–0.97)	.018	0.85 (0.74–0.99)
Model 2d	1.00	0.91 (0.85–0.99)	0.84 (0.73–0.98)	0.77 (0.61–0.97)	0.65 (0.45–0.94)	.013	0.84 (0.72–0.98)
Model 3e	1.00	0.90 (0.83–0.98)	0.83 (0.71–0.96)	0.75 (0.59–0.95)	0.62 (0.42–0.91)	.008	0.82 (0.70–0.96)
Na/K
Mid‐value of quintile/range	3.4	4.7	5.8	7.7	11.0		0.3, 37.3
Model 1c	1.00	1.08 (1.04–1.11)	1.15 (1.07–1.23)	1.28 (1.14–1.45)	1.56 (1.26–1.94)	<.001	1.06 (1.03–1.09)
Model 2d	1.00	1.08 (1.04–1.12)	1.16 (1.08–1.24)	1.30 (1.15–1.48)	1.61 (1.29–2.01)	<.001	1.06 (1.03–1.10)
Model 3e	1.00	1.07 (1.03–1.11)	1.14 (1.07–1.23)	1.27 (1.13–1.44)	1.54 (1.23–1.92)	<.001	1.06 (1.03–1.09)
Normal‐weight populationf
Na
Mid‐value of quintile/range	129.9	178.6	217.9	249.5	312.5		38.9, 749.6
Model 1c	1.00	1.01 (0.87–1.17)	1.01 (0.77–1.32)	1.01 (0.72–1.42)	1.02 (0.57–1.83)	.931	1.01 (0.79–1.29)
Model 2d	1.00	0.96 (0.79–1.19)	0.94 (0.65–1.36)	0.92 (0.58–1.47)	0.87 (0.39–1.95)	.709	0.94 (0.67–1.32)
Model 3e	1.00	0.95 (0.78–1.17)	0.92 (0.64–1.33)	0.90 (0.56–1.43)	0.83 (0.37–1.86)	.623	0.93 (0.66–1.29)
K
Mid‐value of quintile/range	19.2	28.3	36.5	45.5	61.9		25.1, 227.0
Model 1c	1.00	0.92 (0.84–1.01)	0.86 (0.72–1.02)	0.79 (0.60–1.04)	0.68 (0.44–1.06)	.063	0.85 (0.71–1.03)
Model 2d	1.00	0.93 (0.82–1.05)	0.87 (0.69–1.09)	0.80 (0.56–1.14)	0.70 (0.40–1.24)	.184	0.86 (0.68–1.10)
Model 3e	1.00	0.93 (0.82–1.05)	0.86 (0.68–1.10)	0.80 (0.55–1.15)	0.69 (0.38–1.26)	.186	0.86 (0.67–1.10)
Na/K
Mid‐value of quintile/range	3.3	4.6	5.7	7.4	10.9		0.7, 28.7
Model 1c	1.00	1.01 (0.96–1.05)	1.01 (0.93–1.10)	1.02 (0.87–1.19)	1.03 (0.79–1.36)	.796	1.00 (0.97–1.04)
Model 2d	1.00	1.00 (0.93– 1.07)	1.00 (0.88–1.14)	1.00 (0.79–1.27)	1.00 (0.66–1.53)	.992	1.00 (0.95–1.06)
Model 3e	1.00	1.00 (0.93–1.07)	1.00 (0.88–1.14)	1.00 (0.79–1.26)	1.00 (0.66–1.52)	.996	1.00 (0.95–1.06)
Overweight/obese populationf
Na
Mid‐value of quintile/range	144.4	201.3	239.8	257.9	365.4		22.1, 750.8
Model 1c	1.00	1.14 (1.03–1.26)	1.26 (1.06–1.51)	1.35 (1.07–1.69)	1.67 (1.13–2.48)	.005	1.24 (1.05–1.46)
Model 2d	1.00	1.15 (1.03–1.28)	1.28 (1.06–1.55)	1.37 (1.08–1.74)	1.72 (1.14–2.60)	.005	1.25 (1.06–1.49)
Model 3e	1.00	1.10 (0.99–1.23)	1.20 (0.99–1.45)	1.26 (0.98–1.60)	1.48 (0.97–2.26)	.046	1.18 (0.99–1.40)
K
Mid‐value of quintile/range	20.1	30.3	39.0	49.2	68.0		4.69, 155.17
Model 1c	1.00	0.90 (0.83–0.98)	0.82 (0.70–0.96)	0.74 (0.58–0.95)	0.62 (0.41–0.91)	.009	0.82 (0.69–0.96)
Model 2d	1.00	0.89 (0.81–0.98)	0.80 (0.67–0.96)	0.71 (0.54–0.94)	0.58 (0.37–0.90)	.008	0.79 (0.66–0.96)
Model 3e	1.00	0.90 (0.82–0.99)	0.82 (0.68–0.98)	0.73 (0.56–0.96)	0.60 (0.39–0.94)	.013	0.81 (0.67–0.97)
Na/K
Mid‐value of quintile/range	3.5	4.7	5.9	7.8	11.2		0.34, 37.25
Model 1c	1.00	1.10 (1.05–1.14)	1.19 (1.10–1.29)	1.38 (1.20–1.59)	1.77 (1.37–2.28)	<.001	1.08 (1.04–1.12)
Model 2d	1.00	1.10 (1.06–1.14)	1.20 (1.12–1.29)	1.39 (1.22–1.59)	1.81 (1.44–2.28)	<.001	1.08 (1.05–1.11)
Model 3e	1.00	1.09 (1.04–1.13)	1.17 (1.08–1.27)	1.33 (1.15–1.53)	1.66 (1.29–2.14)	<.001	1.07 (1.03–1.11)

^aFor urinary sodium (Na) and potassium (K) excretion, odds ratios (ORs) are for per one standard deviation (SD) increase in excretion. For sodium‐potassium ratio, ORs is per unit change. ^b P value for trend across percentiles of urinary excretion of sodium, potassium or sodium‐potassium ratio based on F test; all tests were two‐tailed. P values for interactions between 24‐hour sodium, potassium, and sodium‐potassium ratio and body mass index (BMI) status on risk for hypertension were .10, .61, and .001, respectively. ^cModel 1 adjusted for age, sex, urban‐rural, and regions. ^dModel 2 adjusted for all factors in model 1 plus educational attainment, smoking status, alcohol intake, physical activity, and antihypertensive medication use. We also included potassium in the regression model for sodium and sodium in the regression model for potassium. ^eModel 3 adjusted for all factors in model 2 plus BMI as a continuous variable. ^fAccording to the Chinese guidelines for overweight and obesity, normal weight is defined as a BMI of ≥18.5 kg/m² to <24 kg/m², while overweight is defined as a BMI of 24 kg/m² to <28 kg/m². Patients with a BMI ≥28 kg/m² are considered obese.

The pattern of association remained largely unchanged by excluding the participants currently using antihypertension medications or with self‐reported CVD or diabetes (Tables S1 and Table S2). We examined the distribution of urine volume and creatinine by participant characteristics and did not find differential levels of incompleteness (Table S3). In addition, our results remained largely unchanged in sensitivity analyses conducted on subsets of the study population using five different creatinine criteria to exclude participants with potentially incomplete urine collections (Tables S4–S8). When analyzing BMI as three groups, ie, normal, overweight and obese, the association between sodium excretion and SBP and prevalence of hypertension were significant among the overweight participants, but not among the obese participants. (Tables S9–S11).

Discussion

Using a representative sample of the Shandong adult population aged 18 to 69 years, we estimated a high average sodium intake (235.7 mmol, or 5400 mg/d), a low average intake of potassium (40.5 mmol, or 1500 mg/d), and a very high sodium‐potassium ratio (6.8). Higher sodium excretion and the sodium‐potassium ratio were associated with higher BP and prevalence of hypertension, and higher potassium excretion was inversely associated with BP and hypertension risk, while the observed associations appeared to be stronger in the overweight/obese population than in adults of normal weight.

Several earlier studies have examined the association between sodium intake and BP in various Chinese populations,5, 17, 18, 19, 20, 21, 30, 31, 32, 33 including populations in northern and southern China (Table S12). The majority of the studies showed a positive association between sodium intake and BP,5, 19, 21, 30, 31, 32, 33 and the association in some of these reports was attenuated after adjusting for covariates, including BMI.5, 21, 33 Among these studies, three examined the effect of BMI on the sodium and BP relationship.20, 21, 33 In the multinational INTERSALT and INTERMAP studies, no significant interactions were found on the association between urinary sodium, potassium intake, and BP by BMI among Chinese participants.20, 21 In the Yi migration study, a greater association between sodium and BP was found among persons with higher levels of BMI.33 However, in the above three studies, the samples of Chinese participants were limited (600–900 persons), and prevalence of hypertension and obesity was low.

The role played by body weight in the association between sodium intake and BP remains open to debate.34, 35, 36 Two studies suggested that obese adolescents had greater BP response/association to sodium intake than did their lean counterparts.15, 16 In the Trial of Nonpharmacologic Interventions in the Elderly (TONE) study, among older obese people with hypertension, there appeared to be an interaction between sodium reduction and weight loss.13 That study found that although the combined intervention (weight loss plus reduced sodium intake) was more beneficial than either of the interventions alone in reducing BP, the effect of the two interventions combined was not purely additive.13 Elsewhere, the Genetic Epidemiology Network of Salt Sensitivity (GenSalt) study suggested that the metabolic syndrome, of which obesity is a major component, was associated with differential effects of salt reduction on BP, with the number of risk factors for metabolic syndrome increased, the risk of BP changes of high/low sodium dietary intervention increased.14 Moreover, an independent association of sodium intake with the risk of cardiovascular disease was found only in the overweight participants in the first National Health and Nutrition Examination Survey Epidemiologic Follow‐up Study.37

On the other hand, some studies did not find an effect of BMI on the sodium‐BP relationship.20, 21, 38 In the INTERSALT study, the pooled regression coefficients for the effect of sodium intake on BP did not differ between participating centers or individuals in the upper (higher) BMI group and those in the lower BMI group.20 It was suggested that controlling for BMI might lead to overadjustment. Because sodium intake was positively associated with BMI, and BMI is better measured than sodium intake, it may dominate combined regression models, and thus the greater sodium intake among the higher BMI participants might have contributed to the association of BMI with BP.

In our study, BMI was moderately correlated with 24‐hour sodium excretion (r=0.15, P<.001). Adjusting for BMI attenuated the association for the population as a whole, which was consistent with the INTERSALT study findings, although we still observed a significant statistical interaction on the multiplicative scale of sodium and the sodium‐potassium ratio with BP and prevalence of hypertension. Stratified analysis by BMI category in our study showed a significant sodium‐SBP association among overweight/obese participants but not among those of normal weight, suggesting the possible modification of the observed sodium‐BP associations by BMI.

Our findings of effect modification by BMI status on the association between sodium intake and BP, if true, might have significant public health implications for the prevention and control of hypertension in the Shandong Province and China as a whole. In the past 30 years in China, the trends for dietary sodium intake, prevalence of overweight/obesity, and hypertension have changed significantly. In the 1980s, while dietary sodium intake was already high, the prevalence of hypertension and overweight/obesity were low.9, 10 Although the consumption of sodium has decreased slightly over time,7 the prevalence of hypertension and overweight/obesity increased dramatically.2, 3, 10 In China in 2010, of every 10 adults, three were hypertensive and four were overweight or obese.3, 9 Given the rapid increase in the prevalence of overweight and obesity in Shandong Province as well as in China as a whole and the observed effect modification by BMI status, the high sodium intake and high sodium‐potassium ratio might play an increasingly important role in the increased prevalence of hypertension and subsequent risk of CVD. In Shandong Province, the SMASH project, led by the local government and agencies from multiple sectors, is taking action to reduce salt intake to 10 g/d in 2015 as well as to improve hypertension control.22 Our findings in this examination of survey data from the SMASH project suggest that sodium reduction should occur alongside interventions to reduce obesity such as increasing physical activity and promotion of healthy diets to achieve better hypertension control and consequent reduced risk of CVD in Shandong.

Study Strengths and Limitations

A major strength of our study was the collection of timed 24‐hour urine specimens from representative population samples in Shandong Province under strict quality control. Potential limitations also exist. First, having a single 24‐hour urine sample from the participants might be insufficient to account for within‐ and between‐person variation in usual sodium intake,39, 40 variation that might attenuate the association with BP (“regression‐dilution”) as suggested by other studies.5, 18, 19 Second, in sensitivity analyses, we used urinary creatinine, urine volume, and body weight to assess completeness of urine collection, as we did not have an objective biomarker of completeness such as para‐aminobenzoic acid.41 Our approach might lead to the inclusion of some incomplete urine samples or to the exclusion of some complete urine samples and thus introduce an additional source of variability. Third, in our study, covariates were based on self‐report and thus subject to reporting error. For example, we did not obtain information on the intensity of physical activity or the frequency of alcohol consumption. Forth, calorie intake was not available in SMASH 2011, therefore we could not adjust for the total calorie intake in our analyses. Fifth, we observed an insignificant association between 24‐hour sodium and BP among the normal‐weight participants, but could not rule out the potential for reverse causality, when compared with the participants with certain medical conditions who might be less likely to be overweight and reduce their sodium intake.42 Furthermore, we observed a nonsignificant association of sodium and BP and prevalence of hypertension among obese participants. This nonsignificant association might be, at least partly, attributed to the limited sample size in the obese group; however, it deserves further investigation with a larger sample size. Finally, SMASH 2011 was a cross‐sectional study, and observed associations should be interpreted with caution. Follow‐up with repeated 24‐hour urine collection is desirable to evaluate the effects of sodium intake on BP and risk of hypertension and possible interactions with BMI.

Conclusions

In our study, the positive association of urinary sodium excretion and the inverse association of potassium excretion with SBP, DBP, and prevalence for hypertension were stronger within the overweight/obese participants than among those of normal weight. The sodium‐potassium ratio was significantly associated with BP and prevalence of hypertension regardless of BMI status and appeared to be a stronger predictor for prevalence of hypertension than sodium or potassium alone. The dietary pattern of high‐sodium and low‐potassium intakes contributes to the high burden of CVD, mainly stroke, in China, and along with the emerging obesity epidemic represents a major threat to public health. It is important to move toward lowered sodium intake and higher potassium intake in the Chinese population to help tackle the growing CVD epidemic, which has already affected the health and welfare of millions of people in China.

Sources of Funding

The survey on which this study was based was funded by the Chinese Center for Disease Control and Prevention, National Center for Noncommunicable and Chronic Disease Control and Prevention, the Technical Development Plan in Shandong (implemented by Shandong Center for Disease Control and Prevention, grant number 2012GSF11828). Paul Elliott is supported by the National Institute for Health Research (NIHR), Imperial College Healthcare NHS Trust (ICHNT), and Imperial College Biomedical Research Centre (BRC) (grant number P38084), the NIHR Health Protection Research Unit on Health Impact of Environmental Hazards, and the Medical Research Council – Public Health England (MRC‐PHE) Centre for Environment and Health.

Disclosures

None of the authors has any conflicts of interest.

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US Centers for Disease Control and Prevention, the National Institute for Health Research, or Department of Health in England.

Author Contributions

Z.Q.B. and L.H.W. helped in study design; J.L.T., X.L.G., X.F.Z., and J.Y.Z. helped in data collection and data cleaning; L.X.Y., Q.H.Y., M.E.C., and J.X.M. helped in data analysis and the manuscript writing; and Y.L.H., M.E. P.E., and S.Y.A. provided significant advice and contributed to the writing and editing of the manuscript.

Supporting information

Table S1. Association of urinary sodium, potassium, and sodium‐potassium ratio with blood pressure excluding participants taking antihypertensive medications (n=1823, with 877 normal weight and 946 overweight/obese).

Table S2. Adjusted association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio excluding participants with self‐reported chronic disease (n=1877, with 888 normal weight and 989 overweight/obese).

Table S3. Distribution of 24‐hour urinary volume and urinary creatinine by selected demographics––Shandong and Ministry of Health Action on Salt and Hypertension (SMASH) baseline survey, 2011.

Table S4. Association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio by Reinivuo's criteria (n=1415, with 604 normal weight and 811 overweight/obese).

Table S5. Association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio by World Health Organization criteria (n=1546, with 717 normal weight and 829 overweight/obese).

Table S6. Association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio by Malekshah's criteria (n=1289, with 577 normal weight and 712 overweight/obese).

Table S7. Adjusted association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio by Joossens and Geboers's criteria (n=1304, with 659 normal weight and 645 overweight/obese).

Table S8. Adjusted association of 24‐hour urinary sodium, potassium, and sodium‐potassium ratio by Knuiman's criteria (n=917, with 499 normal weight and 418 overweight/obese).

Table S9. Adjusted association of sodium excretion, potassium excretion, and sodium‐potassium ratio with systolic blood pressure by body mass index status.

Table S10. Adjusted association of sodium excretion, potassium excretion, and sodium‐potassium ratio with diastolic blood pressure by body mass index status.

Table S11. Adjusted odds for risk of hypertension by body mass index status and quintile of sodium excretion, potassium excretion, and sodium‐potassium ratio.

Table S12. The observational studies on the association between sodium intake and blood pressure in a Chinese population (from urine collection samples).

J Clin Hypertens (Greenwich). 2015;17:916–925. DOI: 10.1111/jch.12658. © 2015 Wiley Periodicals, Inc.