Dr. Hyon K. Choi served on the advisory boards (less than $10,000 each) for TAP and Savient Pharmaceuticals.
Sugar-sweetened soft drinks, diet soft drinks, and serum uric acid level: The third national health and nutrition examination survey
Version of Record online: 28 DEC 2007
Copyright © 2008 by the American College of Rheumatology
Arthritis Care & Research
Volume 59, Issue 1, pages 109–116, January 2008
How to Cite
Choi, J. W. J., Ford, E. S., Gao, X. and Choi, H. K. (2008), Sugar-sweetened soft drinks, diet soft drinks, and serum uric acid level: The third national health and nutrition examination survey. Arthritis & Rheumatism, 59: 109–116. doi: 10.1002/art.23245
- Issue online: 28 DEC 2007
- Version of Record online: 28 DEC 2007
- Manuscript Accepted: 15 JUN 2007
- Manuscript Received: 5 MAR 2007
Sugar-sweetened soft drinks contain large amounts of fructose, which may significantly increase serum uric acid levels and the risk of gout. Our objective was to evaluate the relationship between sugar-sweetened soft drink intake, diet soft drink intake, and serum uric acid levels in a nationally representative sample of men and women.
Using data from 14,761 participants age ≥20 years from the Third National Health and Nutrition Examination Survey (1988–1994), we examined the relationship between soft drink consumption and serum uric acid levels using linear regression. Additionally, we examined the relationship between soft drink consumption and hyperuricemia (serum uric acid level >7.0 mg/dl for men and >5.7 mg/dl for women) using logistic regression. Intake was assessed by a food-frequency questionnaire.
Serum uric acid levels increased with increasing sugar-sweetened soft drink intake. After adjusting for covariates, serum uric acid levels associated with sugar-sweetened soft drink consumption categories (<0.5, 0.5–0.9, 1–3.9, and ≥4 servings/day) were greater than those associated with no intake by 0.08, 0.15, 0.33, and 0.42 mg/dl, respectively (95% confidence interval 0.11, 0.73; P < 0.001 for trend). The multivariate odds ratios for hyperuricemia according to the corresponding sweetened soft drink consumption levels were 1.01, 1.34, 1.51, and 1.82, respectively (P = 0.003 for trend). Diet soft drink consumption was not associated with serum uric acid levels or hyperuricemia (multivariate P > 0.13 for trend).
These findings from a nationally representative sample of US adults suggest that sugar-sweetened soft drink consumption is associated with serum uric acid levels and frequency of hyperuricemia, but diet soft drink consumption is not.
Hyperuricemia is considered the precursor of gout, which is the most common inflammatory arthritis for adult men (1). The doubling of the prevalence (2) and incidence (3) of gout in the US over the last few decades (4) coincided with a substantial increase in soft drink and fructose consumption (5). For example, soft drink consumption in the US increased by 61% in adults from 1977–1997 (5), and sugar-sweetened soft drinks represent the largest single food source of calories in the US diet (5, 6). Conventional dietary recommendations for gout have focused on restriction of purine and alcohol intake, but not on sugar-sweetened soft drink intake (7, 8). Although these soft drinks contain low purine levels, they contain large amounts of fructose, which is the only carbohydrate known to increase uric acid levels (9–13). This led us to hypothesize that sugar-sweetened soft drink consumption is positively associated with serum uric acid levels, but that diet soft drink intake is not. To evaluate this hypothesis, we conducted a cross-sectional study based on the US Third National Health and Nutrition Examination Survey (NHANES-III) (14, 15). In addition, we examined whether the intake of orange juice, a common source of naturally occurring fructose, was associated with serum uric acid levels in this study.
SUBJECTS AND METHODS
Conducted between 1988 and 1994, the NHANES-III included a representative sample of the noninstitutionalized civilian US population, which was selected by using a multistage, stratified sampling design (14). After a home interview, participants were invited to attend examination sessions where blood and urine specimens were obtained. For participants unable to attend the examination for health reasons, a blood sample was obtained during the home interview. Our analysis was limited to participants age ≥20 years who attended the medical examination, and only included the 14,761 participants (6,906 men and 7,855 women) with complete information. We repeated our analyses among 14,317 participants after excluding those who self-reported gout or were taking allopurinol or uricosuric agents (n = 444).
In addition, we analyzed the NHANES-III Supplemental Nutrition Survey (SNS) of Older Americans (16), because total fructose intake was estimated only in this subpopulation. The NHANES-III SNS was conducted to estimate usual dietary intake in a sample of older subjects (age ≥50 years), and to identify the characteristics that influence dietary survey collection. Our analysis of these data was limited to the 2,570 participants with complete information.
Uric acid measurement.
Serum uric acid levels were measured by oxidization with the specific enzyme uricase to form allantoin and H2O2 (Hitachi Model 737 Multichannel Analyzer; Boehringer Mannheim Diagnostics, Indianapolis, IN). Details about quality-control procedures have been published elsewhere (15). Values are reported in milligrams per deciliter; to convert to micromoles per liter, multiply by 59.48.
Assessment of sweetened soft drink and diet soft drink intake.
During the home interview, intakes of sugar-sweetened soft drinks, diet soft drinks, and orange juice were determined from responses to the food-frequency questionnaire that was administered to participants to assess their usual consumption over the prior month. Food-frequency questionnaire assessment of dietary intake has been shown to be a valid and reliable method for assessing average dietary consumption (17, 18).
Assessment of covariates.
The average daily intakes of total meat, seafood, and dairy foods were derived from responses to the food-frequency questionnaire (19, 20). Intake of total energy and fructose was calculated from a 24-hour dietary recall. Total caffeine intake was calculated using US Department of Agriculture food composition sources (21). We estimated that the caffeine content of the individual beverages was 137 mg/cup of coffee, 47 mg/cup of tea, and 46 mg/bottle or can of cola beverage (21, 22). The NHANES-III collected information on body measurements (including height and weight), medication use (including diuretics, antihypertensives, allopurinol, and uricosuric agents), medical conditions (including self-reported hypertension and gout), and serum creatinine levels. Glomerular filtration rate (GFR) was estimated by using the simplified Modification of Diet in Renal Disease study equation (23–25):
Body mass index (BMI) was calculated by dividing a participant's weight in kilograms by the square of their height in meters.
All statistical analyses were computed using survey commands of Stata statistical software (e.g., svymean and svyreg; Stata Corporation, College Station, TX) to incorporate sample weights and adjust for clusters and strata of the complex sampling design.
We used linear regression modeling to evaluate the relationship between beverage and fructose intake and serum uric acid level. Sweetened and diet soft drink consumption were each categorized into 5 groups of average daily servings: 0, <0.5, 0.5–0.9, 1–3, and ≥4 servings/day. Fructose intake in the NHANES-III SNS was categorized into 4 groups: <10, 10–49.9, 50–74.9, and ≥75 gm/day. These cut-points were chosen, considering the distribution of the data, to ensure sufficient sample size for each category according to the Analytic and Reporting Guidelines of the NHANES-III (26). Multivariate models for sugar-sweetened soft drink consumption were adjusted for age; sex; smoking status; total energy intake; BMI; use of diuretics, beta-blockers, allopurinol, and uricosuric agents; self-reported hypertension; GFR; and intake of meat, seafood, dairy foods, coffee, tea, total caffeine, diet soft drinks, and orange juice. Multivariate models for fructose intake were adjusted for the same variables, except for intake of diet soft drinks and orange juice. Trends in serum uric acid levels across categories of intake were assessed in linear regression models by using the median values of each category to minimize the influence of outliers. We also performed logistic regression with a dichotomous outcome of hyperuricemia (serum uric acid level >7.0 mg/dl in men and >5.7 mg/dl in women) (15), adjusting for the same covariates. We examined the potential impact of an alternative definition of hyperuricemia (serum uric acid level >6.0 mg/dl regardless of sex) in these regression models.
We explored potential interactions by sex, age group (<60 years versus ≥60 years), BMI (<25 kg/m2 versus ≥25 kg/m2), and alcohol use (abstainer versus drinker) by testing the significance of interaction terms added to our final multivariate models. For all difference estimates and odds ratios (ORs), we calculated 95% confidence intervals (95% CIs). All P values were 2-sided.
The population's mean age was 45 years. The mean serum uric acid level was 5.32 mg/dl (6.05 mg/dl in men and 4.63 mg/dl in women) and 18% were hyperuricemic (19% of men and 17% of women). The relevant characteristics according to sweetened soft drink, diet soft drink, and orange juice intake are shown in Table 1. With increasing sweetened soft drink intake, meat intake and the proportion of men tended to increase, but age and frequency of hypertension and diuretic use tended to decrease. With increasing diet soft drink intake, BMI and frequency of hypertension tended to increase. With increasing orange juice intake, dairy intake tended to increase.
|Intake, servings/day||Participants, no.||Age, years||Men, %||BMI, kg/m2||Diuretic use, %||History of hypertension, %||Alcohol, servings/day||Total meat, servings/day||Seafood, servings/day||Dairy foods, servings/day||Uric acid medication use, %†||Creatinine, mg/dl|
|Sweetened soft drinks|
|Diet soft drinks|
Serum uric acid levels and sweetened soft drink, diet soft drink, and orange juice intake.
Serum uric acid level increased with increasing sweetened soft drink intake, as shown in Table 2 and Figure 1. After adjusting for age, BMI, sex, and nondietary variables, serum uric acid levels associated with sugar-sweetened soft drink consumption categories (<0.5, 0.5–0.9, 1–3, and ≥4 servings/day) were greater than the levels associated with no intake by 0.12, 0.20, 0.38, and 0.45 mg/dl, respectively (95% CI 0.18, 0.71; P < 0.001 for trend). After further adjusting for dietary variables, the differences were slightly attenuated but remained significant (P < 0.001 for trend). In contrast, diet soft drink intake was inversely associated with serum uric acid levels after adjusting for age, BMI, sex, and nondietary variables (P = 0.001 for trend) (Table 2). After additionally adjusting for dietary variables, the inverse association was attenuated and became insignificant (P = 0.13 for trend). There was a modest association between increasing orange juice intake and serum uric acid levels (Table 2). After adjusting for covariates, serum uric acid levels in individuals with orange juice intake ≥1 serving/day were greater than those with no intake by 0.17 mg/dl (95% CI 0.01, 0.34; P = 0.009 for trend) (Table 2).
|Intake, servings/day||Participants, no.||Unadjusted difference (95% CI)||Multivariate difference (95% CI)†||Multivariate difference (95% CI)‡|
|Sweetened soft drinks|
|0||5,655||0 (referent)||0 (referent)||0 (referent)|
|<0.5||4,844||0.10 (0.02, 0.19)||0.12 (0.05, 0.19)||0.08 (0.01, 0.15)|
|0.5–0.9||3,236||0.29 (0.19, 0.38)||0.20 (0.13, 0.27)||0.15 (0.06, 0.24)|
|1–3||857||0.46 (0.32, 0.61)||0.38 (0.25, 0.51)||0.33 (0.21, 0.46)|
|≥4||186||0.49 (0.20, 0.77)||0.45 (0.18, 0.71)||0.42 (0.11, 0.73)|
|P for trend||–||< 0.001||< 0.001||< 0.001|
|Diet soft drinks|
|0||10,063||0 (referent)||0 (referent)||0 (referent)|
|<0.5||2,368||−0.09 (−0.19, 0.02)||−0.08 (−0.15, −0.01)||−0.03 (−0.11, 0.05)|
|0.5–0.9||1,677||−0.19 (−0.29, −0.09)||−0.22 (−0.31, −0.14)||−0.14 (−0.24, −0.03)|
|1–3||525||−0.11 (−0.28, 0.06)||−0.17 (−0.31, −0.03)||−0.07 (−0.21, 0.08)|
|≥4||145||−0.08 (−0.44, 0.29)||−0.28 (−0.61, 0.05)||−0.12 (−0.43, 0.19)|
|P for trend||–||0.046||0.001||0.131|
|0||3,304||0 (referent)||0 (referent)||0 (referent)|
|<0.5||5,837||0.02 (−0.06, 0.10)||0.01 (−0.07, 0.09)||0.02 (−0.06, 0.09)|
|0.5–0.9||5,287||0.09 (−0.01, 0.18)||0.07 (0.00, 0.14)||0.08 (0.01, 0.15)|
|≥1||344||0.13 (−0.15, 0.40)||0.15 (−0.03, 0.34)||0.17 (0.01, 0.34)|
|P for trend||–||0.042||0.015||0.009|
The results of logistic regression with hyperuricemia as a dichotomous outcome were similar. The multivariate ORs for hyperuricemia according to sugar-sweetened soft drink consumption categories (<0.5, 0.5–0.9, 1–3, and ≥4 servings/day) as compared with no intake were 1.01, 1.34, 1.51, and 1.82, respectively (P = 0.003 for trend). In contrast, diet soft drink consumption was not associated with hyperuricemia (P = 0.46 for trend). The multivariate OR for hyperuricemia according to orange juice consumption categories (<0.5, 0.5–0.9, and ≥1 serving/day) as compared with those with no use were 1.12, 1.30, and 1.39, respectively (P = 0.005 for trend). An alternative definition of hyperuricemia (serum uric acid level >6.0 mg/dl, regardless of sex) did not materially alter these results.
Serum uric acid level according to sex, age group, BMI, and alcohol use.
We also conducted stratified analyses to evaluate whether the association between sweetened soft drink consumption and serum uric acid level varied according to sex, age group, BMI, and alcohol use (Figure 2). There was no significant interaction with these variables (P for interaction >0.1) except for sex (P for interaction = 0.001) (Figure 2). The multivariate differences for soft drink consumption categories (<0.5, 0.5–0.9, 1–3, and ≥4 servings/day) were 0.15, 0.21, 0.40, and 0.52 mg/dl, respectively, for men (P for interaction <0.001) and 0.04, 0.10, 0.19, and 0.19 mg/dl, respectively, for women (P = 0.02 for trend) (Figure 1).
Fructose intake and serum uric acid levels in the NHANES-III SNS.
Among the participants of the NHANES-III SNS (age ≥50 years; n = 2,570), increasing fructose intake was associated with increasing serum uric acid levels. After adjusting for all covariates, serum uric acid levels associated with fructose consumption categories (10–49.9, 50–74.9, and >75 gm/day [the latter being equivalent to ∼5 cans of cola/day]) were greater than those associated with <10 gm/day by 0.05, 0.43, and 0.88 mg/dl, respectively (95% CI 0.24, 1.53; P = 0.003 for trend). The multivariate ORs for hyperuricemia according to the corresponding fructose consumption levels were 1.03, 2.05, and 4.11, respectively (P = 0.003 for trend).
In this nationally representative sample of US men and women, we found that serum uric acid levels significantly increased with increasing sugar-sweetened soft drink intake. The association was independent of dietary and other risk factors for hyperuricemia such as age, sex, BMI, alcohol use, renal function, hypertension, and diuretic use. The association persisted across subgroups stratified by sex, age, BMI, and alcohol use, and it tended to be greater in men. We also found a modest association with orange juice intake. In contrast, there was no significant association between diet soft drink consumption and serum uric acid level.
The difference in serum acid level between the extreme categories of sugar-sweetened soft drink consumption was 0.4 mg/dl. This magnitude of a population mean difference in serum uric acid level (20, 27) can be translated into a clinically relevant difference in the risk for incident gout, as demonstrated in our previous studies (19, 28). For example, an increase of 1 daily serving in beer intake was associated with a mean serum uric acid level increase of 0.4 mg/dl in a cross-sectional analysis of the NHANES-III participants (19), and with a 50% increased risk of incident gout in our prospective analysis of the Health Professionals Follow-Up Study (27). This potentially significant impact on the eventual risk of gout is also supported by our results using hyperuricemia as a dichotomous outcome, and when using various definitions for hyperuricemia.
Sugar-sweetened soft drinks contain large amounts of fructose from added sugars, whereas orange juice and other sweet fruit juices contain naturally occurring fructose. Fructose induces uric acid production by increasing ATP degradation to AMP, a uric acid precursor (4, 9, 29–31). Fructose phosphorylation in the liver uses ATP, and the accompanying phosphate depletion limits regeneration of ATP from ADP, which in turn serves as a substrate for the catabolic pathway to uric acid formation (32). Thus, within minutes after fructose infusion, plasma (and later urinary) uric acid concentrations increase (9). In conjunction with purine nucleotide depletion, rates of purine synthesis de novo are accelerated, thus potentiating uric acid production (33). In contrast, glucose and other simple sugars do not have this effect (13). Furthermore, fructose could indirectly increase serum uric acid levels and the risk of gout by increasing insulin resistance and circulating insulin levels (34). Both experimental studies in animal models and short-term feeding trials among humans suggest that higher fructose intake contributes to insulin resistance, impaired glucose tolerance, and hyperinsulinemia (35–38).
There are important practical implications of our results. It was over 100 years ago that Osler prescribed diets low in fructose as a means to prevent gout (13). In his 1893 text (39) he wrote, “The sugar should be reduced to a minimum. The sweeter fruits should not be taken” (13). However, conventional dietary recommendations for gout have focused on restriction of purine intake, although low-purine diets are often high in carbohydrates, including fructose-rich foods (7). Our data provide evidence that fructose poses a substantial risk for hyperuricemia, thus supporting the validity and importance of Osler's approach. Furthermore, because fructose intake is associated with increased serum insulin levels, insulin resistance, and adiposity (35–38, 40, 41), the overall negative health impact from fructose is expected to be larger in gout patients, who often have the metabolic syndrome (63%) (42) and are overweight (71%) (43). Conversely, the conventional low-purine diet approach that allows fructose consumption may have contributed to the high prevalence of the metabolic syndrome observed in cross-sectional studies (42, 44, 45). Our findings support the importance of reducing fructose intake in dietary recommendations for hyperuricemia in order to reduce the patient's serum uric acid levels, as well as to improve long-term outcomes.
We found that the increase in serum uric acid level associated with sugar-sweetened soft drink intake tended to be larger among men than women. This could be due to differences in sex hormones. Studies in rats have shown that female sex hormones protect against the development of hyperinsulinemia associated with high fructose intake (46–48). Because hyperinsulinemia results in decreased renal excretion of urate and correlates with higher serum uric acid levels (49–51), the protective effect of estrogen may lead to an attenuated effect of fructose on serum uric acid levels. Confirmation of these findings by future studies would allow more refined dietary recommendations for both men and women with hyperuricemia or gout.
The strengths and limitations of our study deserve comment. This study was performed in a nationally representative sample of US women and men; thus, the findings are likely to be generalizable to the US adult population. As opposed to prospective studies, a cross-sectional study design tends to leave uncertainty regarding the temporal sequence of exposure–outcome relations, and is also vulnerable to recall bias. For example, if some participants switched their alcohol consumption to nonalcoholic beverages based on previously identified gout or hyperuricemia before the NHANES study, the previous alcohol effect might theoretically lead to a positive association with the nonalcoholic beverages. However, exclusion of individuals with self-reported, lifetime history of physician-diagnosed gout, or of those who were taking medications to treat hyperuricemia, did not materially alter our results. Furthermore, the fact that the positive association existed with sugar-sweetened soft drinks but not with diet soft drinks argues against this potential scenario. Given the absence of existing conventional recommendations on soft drink consumption for hyperuricemia and gout, it is unlikely that some participants changed their sweetened or diet soft drink intake based on previously identified hyperuricemia or gout. In the NHANES-III, the health examination component including serum uric acid measurement (outcome) was performed after the household interview that inquired about intake of these beverages during the past month (exposure). Thus, it appears implausible that the serum uric acid levels measured in this study would somehow systematically influence the reported intake of these beverages. Similarly, differential effects between the beverages would not be explained by these potential methodologic limitations.
In conclusion, our results suggest that sugar-sweetened soft drink intake is associated with a higher level of uric acid and frequency of hyperuricemia, but that diet soft drink intake is not. Furthermore, orange juice intake may also be associated with a higher level of serum uric acid. These data support strategies to reduce fructose consumption in the dietary recommendations for individuals with hyperuricemia and gout.
Dr. Hyon K. Choi 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 design. Jee Woong J. Choi, Ford, Gao, Hyon K. Choi.
Acquisition of data. Jee Woong J. Choi, Hyon K. Choi.
Analysis and interpretation of data. Jee Woong J. Choi, Ford, Gao, Hyon K. Choi.
Manuscript preparation. Jee Woong J. Choi, Ford, Gao, Hyon K. Choi.
Statistical analysis. Jee Woong J. Choi, Hyon K. Choi.
- 14National Center for Health Statistics. Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-94. Vital Health Stat 1 1994; 32: 1–407.
- 15Centers for Disease Control and Prevention. NHANES-III 1988–94 reference manuals and reports. Hyattsville (MD): National Center for Health Statistics; 1996.
- 25A simplified equation to predict glomerular filtration rate from serum creatinine [abstract]. J Am Soc Nephrol 2000; 11: 155A., , , .
- 26National Center for Health Statistics. Analytic and reporting guidelines: the Third National Health and Nutrition Examination Survey, NHANES-III (1988–94). Hyattsville (MD): CDC; 1996.
- 39Gout: the principles and practice of medicine. 2nd ed. New York: Appleton; 1893. p. 287–95..