点击显示 收起
1 From the Graduate Program of Nutrition, National Taiwan Normal University, Taipei, Taiwan (L-CL, C-YH, and S-HH); the School of Public Health, Taipei Medical University, Taipei, Taiwan (C-YY); the School of Public Health, National Defense Medical College, Taipei, Taiwan (M-SL); and the Clinic of Gout, Taipei Municipal Ho-Ping Hospital, Taipei, Taiwan (C-LC).
2 Supported by the National Science Council, Taiwan, Republic of China (grant NSC 88-2314-B-003-001). 3 Address reprint requests to L-C Lyu, Department of Human Development and Family Studies, National Taiwan Normal University, #162 Section 1, Hoping East Road, Taipei, Taiwan 10610, Republic of China. E-mail: t10010{at}cc.ntnu.edu.tw.
ABSTRACT
Background: Gout has been a significant metabolic disorder for Chinese men in Taiwan; however, there is insufficient information on diet and lifestyle risk factors in this population.
Objective: The purpose of this case-control study was to explore potential dietary and lifestyle risk factors associated with gout in Chinese men.
Design: Between 1998 and 1999, we recruited and conducted face-to-face interviews with patients from outpatient clinics in Taipei who had incident gout (n = 92) and with their healthy coworkers (controls; n = 92).
Results: Systolic blood pressure, diastolic blood pressure, waist-to-hip ratio, waist-to-height ratio, and body mass index were significantly higher in cases than in controls. Family histories of gout and diabetes mellitus were strong risk factors for gout. Frequencies of vegetable and fruit consumption were significantly lower in cases than in controls. Logistic regression analyses showed that high alcohol intake and low intakes of fiber, folate, and vitamin C increased the risk of gout, but no association was found with purine intake. After covariates were controlled for, the adjusted odds ratios for the middle and highest tertiles of waist-to-height ratio (0.50–0.54 and =" BORDER="0"> 0.55, respectively) were 3.89 (95% CI: 1.32, 11.46) and 4.37 (1.18, 16.22), respectively, but no linear association was found for waist-to-hip ratio and waist circumference.
Conclusions: Consumption of alcohol, but not of purine, may be a significant dietary risk factor for gout. Food sources rich in dietary fiber, folate, and vitamin C, such as fruit and vegetables, protect against gout. Waist-to-height ratio, which indicates central obesity, has a significant linear effect on gout occurrence, independent of body mass index.
Key Words: Gout • case-control study • diet • obesity • waist-to-height ratio • Taiwan
INTRODUCTION
Gout is a metabolic disorder associated with altered uric acid metabolism and hyperuricemia. Various epidemiologic studies have reported effects of modernization and affluence on lifestyle, which have led to hyperuricemia and an increased prevalence of gouty arthritis in Asian populations (1–10). In Taiwan, a high prevalence of hyperuricemia and gout was observed. In a national survey from 1986 to 1989, Chou et al (2) reported a prevalence rate of gout of 0.5%. Moreover, the National Nutrition and Health Survey (1993–1996) showed that 22% of men (blood uric acid concentration > 7.7 mg/dL) and 23% of women (blood uric acid concentration > 6.6 mg/dL) in Taiwan who were > 45 y old had hyperuricemia (3).
Until now, most of these studies were conducted to identify risk factors for hyperuricemia, including ethnic (4–6), enzymatic (11–13), and environmental predispositions (14–17). Among acquired factors, reversible lifestyle factors contributed to increased blood uric acid concentrations. These factors were suggested to be a high-purine diet, alcohol consumption, and obesity (18, 19). However, no case-control studies have examined factors that are important to the etiology of gout. Thus, this age-matched case-control study was conducted to explore potential risk factors and protective lifestyle factors for gout in a Chinese population. Our primary objectives were to answer the following questions. First, are purine and alcohol consumption risk factors for gout? Second, is the pattern of body adiposity a risk factor for gout that is independent of general obesity?
SUBJECTS AND METHODS
Study subjects
From July 1988 to August 1989, we contacted 236 male patients who visited the outpatient clinic at the Taipei Municipal Ho-Ping Hospital mainly for bone and joint problems and were willing to participate in future in-person lifestyle interviews. Initial criteria for eligible cases and controls were as follows: Han ethnicity, age 20–70 y, generally good health with no regular use of medications such as diuretics and aspirin, no change in diet in the past year, and residency in Taipei city or county in a household with a full-time worker who had an available home address and a contact telephone number. We used the diagnostic criteria for gout of Wallace et al (20) and Hart and Fry (21), which follow the guidelines recommended by the American College of Rheumatology. A total of 95 episodic cases of diagnosed gout and 53 nongout controls (as hospital controls) were later identified and enrolled in the study by C-LC. We also contacted office coworkers of gout patients and recruited 46 healthy subjects without gout who were within a 5-y age range (as friend controls) (22). The 2 groups of controls did not differ from each other in age range and dietary intake distributions. Three cases and 7 controls did not complete the interviews, and thus the final analyses consisted of 184 subjects (92 cases and 92 controls). All subjects signed an informed consent form in accord with the Helsinki Declaration of 1975 as revised in 1983.
Diet history and lifestyle interview
The diet history questionnaire consisted of questions on dietary habits, one 24-h recall, and a total-diet Chinese food-frequency questionnaire consisting of questions arranged by meal sequence on the intake of 493 items during the previous year. A test of relative validity showed strong correlations of macro- and micronutrient intakes with 7-d records (r = 0.38, P < 0.05), and reproducibility was consistently high for most nutrients, with Spearman correlation coefficients between 0.42 for vitamin A and 0.79 for vitamin B-12. When we designed this Chinese food-frequency questionnaire, we took particular care to include all foods high in purines so that our primary research hypotheses could be adequately tested. To collect more specific information on dietary patterns, this questionnaire was supplemented with questions on family and individual eating habits. Lifestyle questions, including questions on drinking, smoking, and betel nut chewing, and questions on the use of dietary supplements were also asked during the interview. Abstention from drinking was defined as having < 1 alcoholic drink/mo. Visual aids for showing 3 portion sizes were developed to facilitate accuracy in reporting long-term dietary intakes on the food-frequency questionnaire and actual intakes on the 24-h recall. A similar methodology was used by Lyu et al (23). Intakes of total fat; animal and plant protein; dietary purine (24–26); cholesterol; fat-soluble vitamins such as vitamins A, D, and E; dietary fiber; and water-soluble vitamins including thiamin, riboflavin, niacin, vitamin C, and folate were assessed in this study. The completeness rates of selective nutrient databases were 75% for folate, 85% for purine, 94% for vitamin C, 97% for dietary fiber, and 98% for alcohol.
Physical activity level was assessed with the use of structured questionnaires on the basis of 3 categories (work, leisure time, and house work). The duration and intensity of physical activity were considered in evaluating overall physical activity levels (27). Three types of anthropometric measurements were included in this study: body size (height and weight), body girth (waist and hip), and body skinfold thicknesses (triceps). In addition, information on personal medical history, medication use, and medical history of first-degree relatives (parents and siblings) were collected in the in-person interviews. Most of the subjects who were interviewed took 1.5–2.5 h to complete the interview.
Statistical analysis
We performed all univariate and multivariate analyses by using SAS, version 6.12 (SAS Institute Inc, Cary, NC). For univariate analysis, comparisons between means were made with the use of Student’s t test, and comparisons between values for categorical variables were made with the use of the chi-square test. Average daily nutrient intakes were calculated from the total-diet food-frequency questionnaire and were expressed as nutrient density [amount per 1000 calories (4.185 MJ)] in the analyses. Moreover, nutrient intakes from 24-h recalls were also used for verification of possible dietary associations; however, alcohol intake was not used because of the limited amounts consumed in the short time period. Dietary variables in the analyses included protein, animal protein, plant protein, fat, animal fat, plant fat, carbohydrates, dietary fiber, soluble fiber, insoluble fiber, sodium, phosphorus, calcium, iron, -carotene, ß-carotene, retinal, total vitamin A, vitamin E, thiamin, riboflavin, niacin, vitamin B-6, vitamin B-12, vitamin C, folate, saturated fat, monounsaturated fat, polyunsaturated fat, cholesterol, and purine. Nutrient intakes were categorized into upper, middle, and lower thirds of the range for all subjects including cases and controls. Spearman correlation coefficients were calculated between anthropometric measures and dietary variables. Potential confounding factors were identified and adjusted for in multivariate analyses. Adjustment variables included age and educational level as continuous variables. The unconditional logistic regression model produced odds ratios and 95% CIs, and all results, including trends, were considered significant if P < 0.05.
RESULTS
The basic characteristics of the 92 cases and the 92 controls are shown in Table 1. There was no significant difference in age distribution between the cases and the controls. The educational level of the controls was significantly higher than that of the cases. Even though triceps skinfold thicknesses were not significantly different between the cases and the controls, the cases had significantly higher body mass index (BMI), waist circumference, hip circumference, waist-to-hip ratio, and waist-to-height ratio than did the controls. Relative to the respective lowest tertile, the odds ratio for the highest tertile was 2.62 for BMI (95% CI: 1.26, 5.42), 3.21 for waist circumference (95% CI: 1.53, 6.73), and 2.54 for hip circumference (95% CI: 1.23, 5.28); all had significant trends (P < 0.05). However, the trend for waist-to-hip ratio was not linear: relative to the lowest tertile, the odds ratio for the middle tertile was 3.10 (95% CI: 1.47, 6.58), whereas the odds ratio for the upper tertile was 2.13 (95% CI: 1.05, 4.31). The trend for waist-to-height ratio was linear, with odds ratios of 2.87 (95% CI: 1.37, 6.00) and 3.65 (95% CI: 1.73, 7.73) for the middle and upper tertiles relative to the lower tertile. Both systolic and diastolic blood pressures were significant risk factors for gout. Lifestyle factors, such as smoking history, abstention from drinking, and vegetarian status, were not significantly different between the cases and the controls, although the number of vegetarians was too small to draw a firm conclusion.
View this table:
TABLE 1 . Basic characteristics of the cases and the controls1
The medical histories of the subjects and their first-degree family members (parents and siblings) are shown in Table 2. There were no significant differences between the cases and the controls in the numbers having diabetes, hypertension, hyperlipidemia, liver disease, renal disease, lung disease, and other arthritis. However, the number of subjects who had a family history of diabetes or gout was significantly higher among the cases than among the controls (P < 0.05). Therefore, family histories of diabetes and gout are suggestive potential risk factors for gout. Regarding energy expenditure assessments, overall physical activity levels were not significantly different between the cases and the controls (data not shown).
View this table:
TABLE 2 . Personal medical histories of the cases and the controls and medical histories of their first-degree family members1
Associations between the risk of gout and nutrient intakes from a Chinese food-frequency questionnaire and a 24-h recall are shown in Tables 3 and 4, respectively. As shown in Table 3, dietary fiber, soluble fiber, vitamin C, and folate were protective dietary factors against gout, and alcohol was a risk factor. Relative to the respective lowest intake tertile, the odds ratios for the highest tertile were 3.27 for alcohol, 0.38 for dietary fiber, 0.44 for soluble fiber, 0.31 for vitamin C, and 0.33 for folate. Moreover, the intakes of dietary fiber, iron, vitamin A, riboflavin, and folate calculated from the 24-h recall showed protective effects against gout (Table 4). From both dietary assessment methods, consumption of dietary fiber and folate seemed to be consistent protective factors against gout.
View this table:
TABLE 3 . Associations of nutrient densities from a Chinese food-frequency questionnaire with the risk of gout1
View this table:
TABLE 4 . Associations of nutrient intakes from a 24-h recall with the risk of gout1
Average nutrient intakes calculated by using the 24-h recall method did not differ significantly between the cases and the controls, except for the intakes of dietary fiber, folate, calcium, and iron, which were significantly (P < 0.05) higher in the controls than in the cases (Table 4). In the controls, the average daily energy intake was 2305 kcal (9.653 MJ), and the average macronutrient energy distribution was as follows: 14% of total energy from protein, 33% from fat, and 51% from carbohydrates. Average daily purine intakes did not differ significantly between the 2 groups, with a mean (± SD) of 374 ± 208 mg for the cases and of 378 ± 221 mg for the controls. Monthly frequencies of selected dietary habits pertaining to gout among the cases and the controls are shown in Table 5. Vegetable and fruit consumption was significantly higher in the controls than in the cases, although the frequency of social dining was significantly lower in the controls (P < 0.05). The frequency of consumption of foods with a high purine content, such as seafood and organ meat, did not differ significantly between the cases and the controls.
View this table:
TABLE 5 . Frequencies of selected dietary habits pertaining to gout among the cases and the controls
Six multivariable models of gout occurrence consisting of variables for family history, obesity, and dietary factors after adjustment for age and education level are shown in Table 6. Consistently, the risk of gout among the subjects with family histories of diabetes or gout was > 3-fold (odds ratio: 3.55–4.17) and > 6-fold (odds ratio: 6.95–9.54), respectively, that of the subjects without such family histories. Obesity indexes including BMI, triceps skinfold thickness, waist circumference, hip circumference, waist-to-hip ratio, and waist-to-height ratio were highly correlated, and the central obesity index, the waist-to-height ratio (model 4), seemed to have more stable effects on gout than did general obesity (trend P < 0.05). After adjustment for the general obesity index (BMI) and alcohol intake, we found that the adjusted odds ratios for the middle (0.50–0.54) and highest (=" BORDER="0"> 0.55) tertiles of waist-to-height ratio were 3.89 (95% CI: 1.32, 11.46) and 4.37 (95% CI: 1.18, 16.22), respectively, but no linear association was found for waist-to-hip ratio and waist circumference. Model 4 explained 25.8% of variability with 80.8% concordance.
View this table:
TABLE 6 . Odds ratios and 95% CIs for gout of multivariable models consisting of variables for family history, obesity, and dietary factors after adjustment for age and education level
Because of the high correlations between the intakes of dietary fiber, vitamin C, and folate (r = 0.67 for dietary fiber and vitamin C, r = 0.57 for dietary fiber and folate, r = 0.64 for vitamin C and folate, P < 0.01), we decided to put dietary fiber in model 5 and folate in model 6. From model 5, the protective effect from dietary fiber was borderline (P for trend = 0.09) after adjustment for covariates. Because intakes of dietary fiber and alcohol were correlated (r = -0.21, P < 0.01), when we removed alcohol from the model, dietary fiber showed a protective effect (P for trend = 0.05), with odds ratios of 0.43 (95% CI: 0.17, 1.08) and 0.41 (95% CI: 0.17, 1.00) for the middle and upper tertiles, respectively (data not shown). In model 6, folate intake also showed a protective effect against gout, with adjusted odds ratios of 0.30 (95% CI: 0.12, 0.77) and 0.37 (95% CI: 0.15, 0.92) for the middle and upper tertiles, respectively. This model explained 28.8% of variability with 82.8% concordance. The nonsignificant results for alcohol from multivariate analyses could be explained on the basis that alcohol intake was significantly correlated with education level in this population (r = -0.20, P < 0.01). When we removed education level from the models, alcohol intake showed significant odds ratios in the upper tertiles (95% CI: 1.13, 7.62) for models 1–4 (P for trend = 0.02–0.03); the attenuation of effects by adjustment for education level was also shown for dietary fiber in model 5 (P for trend = 0.03) and weight-to-height ratio (P for trend = 0.02) and folate (P for trend = 0.01) in model 6 when education level was removed from the models (data not shown). The consistently higher odds ratios in a dose-response fashion and the significant trends without adjustment for education level in the models suggested that alcohol was a risk factor for gout and that dietary fiber and folate were protective factors against gout.
DISCUSSION
The objective of this case-control study was to determine whether dietary intakes and related lifestyle factors are associated with the development of gout. Although the medical community generally agrees that gout prevention can be achieved through lifestyle changes including weight loss, restricting protein and calorie intake, limiting alcohol consumption, and avoiding the use of diuretics (18, 19, 28), supporting quantitative data are not available from any dietary case-control studies. Our study was the first to identify potential dietary factors affecting gout occurrence in a Chinese sample, and surprisingly, we did not find that the intakes of fat, total protein, animal protein, and purine are related to gout occurrences, as we had previously assumed.
Besides animal sources, plant foods including whole grains, beans, peas, lentils, asparagus, cauliflower, sprouts, spinach, and mushrooms are high in purines (18, 24, 26). Soy products such as soy milk, tofu, and pressed tofu are predominant plant sources of protein in Taiwan, and the intake of soy products contributed a fair amount to the purine intake in our study population. We estimated that 20% of purine intake in this population was from soy products. A misleading nutrition claim that soy products cause gout is a popular health belief among Asian populations with high soy consumption (14, 29). Yamakita et al (29) conducted a study in Japan to examine the effect of tofu on uric acid metabolism in healthy subjects and subjects with gout and found no significant increase in plasma or urine uric acid concentrations or in uric acid clearance in gout patients with normal uric acid clearance. In Taiwan, one study compared the blood uric acid concentrations in vegetarians who usually consumed a fairly large amount of soy products as protein sources with those in nonvegetarians and found lower blood uric acid concentrations in the vegetarians (30). Our data support the observation that increased consumption of foods from plant sources, especially fruit and vegetables, reduce the risk of gout development, but probably not in the way suggested by the "purine content theory." The complexity of human uric acid metabolism and the difficulties of human diet assessment may account for the uncertain causal relation of dietary purine and nucleoprotein intakes, as well as amino acid and protein intakes, with gout (31, 32).
Our findings agree with many other findings that alcohol intake is strongly associated with gout occurrence. The ethanol and purine content in alcoholic beverages may account for the association with hyperuricemia and gout. For example, beer is reported to have a high guanosine content from yeast and barley fermentation (33). A possible mechanism for the association of alcohol intake with gout includes the overproduction of lactic acid and fatty acids, which affect the pH values of body fluids and alter the renal excretion of uric acid (18, 32, 33). In addition, nucleotide overproduction occurs after injection of ethanol, and one study in Japan showed that this effect occurred via a disturbance of ATP metabolism (34).
Fruit and vegetables, which are rich in micronutrients including folate, vitamin C, and dietary fiber, were found to have a significant protective role in our study. Previous research has shown that approximately two-thirds of the uric acid produced each day is excreted in urine and that one-third is eliminated directly in intestinal secretions and saliva (35, 36). This may explain the possible protective effect against gout of the intake of dietary fiber and soluble fiber that was shown in the estimation of usual diets with the Chinese food-frequency questionnaire. Fiber has been recognized as being beneficial for intestinal motility and as having a potential role in binding uric acid in the gut for excretion, and thus the intake of fiber has been suggested to result in a lower risk of gout, as was observed in the present study. The hypouricemic effect of folate was suggested by Oster in 1977 (37), and the mechanism was presumed to be mediated through the inhibition of xanthine oxidase. However, one intervention study conducted by Boss et al (38) using folate supplementation failed to lower blood uric acid concentrations. The results of the present study show high correlations (r = 0.57–0.67) of the consumption of dietary fiber, folate, and vitamin C in this population with fruit and vegetable intake, which may account for the protective effects.
We found that hypertension and obesity are strong risk factors for gout. In 1997 Li et al (7) reported a population study from Beijing that showed strong associations of serum uric acid concentrations with triacylglycerol and glucose concentrations and BMI (7). A cross-sectional population study of 910 men and 603 women in Hong Kong showed positive associations of serum uric acid with BMI, waist-to-hip ratio, systolic and diastolic blood pressure, fasting glucose, triacylglycerol, and apolipoprotein B, as well as a negative association with HDL cholesterol (39). The authors suggested that serum uric acid may be a marker for the presence of an adverse cardiovascular disease risk, which is strongly related to hypertension, hyperlipidemia, and diabetes mellitus. The inclusion of hyperuricemia as a part of syndrome X, which is associated with insulin resistance and coronary artery disease risk factors, was also suggested by some clinicians (19, 40).
Obese people often have hyperuricemia, which frequently leads to painful attacks of gout. It has been suggested that increased serum uric acid (41) and gout occurrence (42) are closely associated with an increase in visceral fat accumulation. The significance of visceral fat obesity has been noticed for many interrelated chronic metabolic disorders (43). Many studies have compared the associations of these proxy measures of abdominal obesity, including waist circumference, waist-to-hip ratio, and waist-to-height ratio, with coronary artery disease risk factors (44–46). As our multivariable models suggest, indexes of central obesity, especially waist-to-height ratio, are better linear indexes for identifying the risk of gout in this population than are waist circumference and waist-to-hip ratio. In a Japanese population study, waist-to-height ratio was also proposed to be a better predictor of multiple coronary risk levels than was waist-to-hip ratio (47).
Many physiologic studies suggested a strong association between hyperuricemia and gout (19, 48). However, as with other blood biochemical disorders, only 5% of hyperuricemic subjects developed clinical symptoms of gout (19, 48). Gout has been recognized as a disorder of altered uric acid metabolism (either overproduction or underexcretion of uric acid) that results in high blood uric acid concentrations (18, 19, 48). Besides diet, the other factors that are related to reductions in uric acid excretion include renal dysfunction, lead toxicity, and medication side effects (19, 21). Also, various genetic metabolic factors affecting uric acid production and excretion complicate the scenario for gout occurrence. In the present study, the average uric acid concentration in the 92 patients with gout was 9.22 mg/dL, with 62 subjects having the underexcretion type (24-h urinary excretion of uric acid < 600 mg) and 30 subjects having the overproduction type (> 600 mg uric acid/d). Potential lifestyle risk factors were not found to differ between these subgroups in the present study.
The strength of this study was that we recruited incident cases of gout rather than patients with long-term gout; therefore, selection bias was limited. The weakness of this study was that only male patients were included in the analysis. Sex differences have often been observed, and the differences are reported to result from differences in sex hormones (49). In conclusion, we confirmed that alcohol intake, but not purine intake, is a strong risk factor for developing gout. Fruit and vegetables, which are rich in dietary fiber, folate, and vitamin C, are protective against gout.
ACKNOWLEDGMENTS
We thank Feng-Hsin Chang for typing assistance and Jeng-Long Feng for valuable advice on energy expenditure and exercise physiology.
C-YH and S-HH contributed to data collection and database management, C-YY and M-SL assisted in the statistical analyses and contributed to data analyses, and C-LC was responsible for the diagnosis and recruitment of participants. L-CL designed and oversaw this project and contributed to the writing of the manuscript. None of the authors of this manuscript had any conflicts of interest.
REFERENCES