1. INTRODUCTION

Uric acid is the final oxidation product of purine metabolism in human beings. High uric acid concentration results from increased production, decreased excretion of uric acid, or a combination of both processes. The prevalence of hyperuricemia was above 21% based on the US National Health and Nutrition Examination Survey 2007-2008 study.1 According to the statistics in 1993-1996, Taiwan was notably affected by a relatively high rate of hyperuricemia. There is 43.7% of hyperuricemia in men and 27.4% in women in Taiwan.2

At present, hyperuricemia is regarded as a serious public health problem. Hyperuricemia can result in gout3 and is related to cardiovascular disease,4,5 chronic kidney disease,6 and cancer.7–9 Specific dietary components are thought to affect concentrations of uric acid. Mainly, vegetarian diets may target multiple pathways in uric acid pathogenesis. Vegetarians consume more vegetables, whole grains, and nuts while away from purine-rich meat and seafood.10,11 Compared with omnivores, vegetarians are considered to have a lower concentration of uric acid. However, there are some vegetarians who are still in the status of hyperuricemia. The reason is that there is a pile of other factors, such as body mass index (BMI) and age, linked to uric acid pathogenesis, while the causal nature of these relationships needs to be clarified.8,12

The independent risk factors for the incidence of hyperuricemia have been investigated a lot by former studies.13–17 However, the previous research established an available prediction model of hyperuricemia,18 but not for people consuming vegetarian diets. Therefore, we created predictive models of hyperuricemia for vegetarians using a large population-based cohort in Taiwan. This study aims to help vegetarians discover other risk factors and avoid more adverse effects associated with hyperuricemia.

2. METHODS 2.1. Design and study participants

Fig. 1 illustrates the enrollment for the study. This retrospective cross-sectional study was conducted to create a model to predict hyperuricemia based on demographics, comorbidities, and commonly available biochemistry tests in vegetarians. We enrolled individuals aged above 40 years and underwent health exams and food questionnaires at the Health Examination Center of Taipei Tzu Chi Hospital (New Taipei City, Taiwan) from September 5, 2005, to December 31, 2016. Omnivores, subjects aged <40, and those with incomplete demographic information or biochemical data were excluded (Fig. 1). The study was conducted in accordance with the Declaration of Helsinki and was approved by the institutional review board at the Taipei Tzu Chi Hospital (approval number: 07-X-104). Informed consent was waived because of the retrospective study design.

Fig. 1:

Flowchart of patient selection. The database included 42 633 individuals who received physical checkups at the Health Examination Center of Taipei Tzu Chi Hospital. After excluding subjects younger than 40 y old (n = 3710), those with incomplete identification information (n = 1619) or missing biochemical data (n = 1293), and omnivores (n = 28,680). Finally, there were 7331 vegetarians entered the final analyses.

2.2. Clinical assessment

A validated food questionnaire was used for evaluating dietary patterns. Diet categories were divided into vegan, lacto-ovo vegetarian, and omnivore. Lacto-ovo vegetarians were defined as consuming eggs, dairy products, or both but without other animal products. Vegans were defined as consuming exclusively vegetables and fruits. Omnivores were defined as consuming both plant-based and animal-based foods. The structured questionnaire containing questions about age, gender, medical history, and functional health patterns was completed by a well-trained nurse at the entry of the study. Height and weight were measured using an automatic electronic meter (SECA GM-1000, Seoul, Korea) for calculating the BMI (kg/m2). An automatic blood pressure machine (Welch Allyn 53000, New York, NY, USA) was used to measure blood pressure. Venous blood was drawn after at least 8 hours of fasting. Measurements included serum creatinine, albumin, glycated hemoglobin (HbA1c), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and uric acid (Dimension RXL Max integrated chemistry system; Siemens, Erlangen, Germany). The estimated glomerular filtration rate (eGFR) was calculated by using the creatinine-based Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation.19 Urine protein was tested using a single dipstick analysis with an automated urine analyzer (Arkray 4030, Tokyo, Japan). These results were reported as a 6-grade scale: absent (<10 mg/dL), trace (+/−) (10-20 mg/dL), 1+ (30 mg/dL), 2+ (100 mg/dL), 3+ (300 mg/dL), or 4+ (>1000 mg/dL). Proteinuria was defined as trace, 1+, or above. CKD was defined as either the presence of proteinuria or an eGFR of ≤60 mL/min per 1.73 m2.20

2.3. Outcome measures

Hyperuricemia was defined as the concentration of serum uric acid level greater than 7.0 mg/dL based on the Taiwan Guideline for the Management of Gout and Hyperuricemia.21

2.4. Statistical analyses

Data are expressed as number (percentage) or mean (SD). Student’s t-test and Chi-square test compared the means and proportions between the two groups. Receiver operating characteristic (ROC) curves were used to illustrate the diagnostic performance of three different prediction models. The models were built up stepwise based on traditional risk factors of hyperuricemia or gout and other clinical characteristics of the participants.19 Logistical regression was used to determine each variable’s odds ratio (OR) with hyperuricemia in all models. The covariates in multivariable logistical regressions were age and sex (model 1); age, sex, diabetes, hypertension, hyperlipidemia, and proteinuria (model 2); age, sex, diabetes, hypertension, hyperlipidemia, and proteinuria, eGFR, systolic blood pressure, HbA1c, HDL-C, LDL-C, BMI, serum albumin, alcohol drinking, and dietary habits (model 3, fully adjusted model). A two-tailed p value of less than 0.05 was considered statistically significant. All statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC) and STATA (version 15.1; Stata Corp, College Station, TX).

3. RESULTS 3.1. Patient characteristics

A total of 7331 vegetarians entered the final analyses. Table 1 shows the demographic and clinical information in patients with (n = 593) or without (n = 6738) hyperuricemia. Compared with patients without hyperuricemia, those with hyperuricemia were older (62.2 ± 10.3 vs 60.9 ± 9.4 years old, p = 0.001); predominantly male (74.9% vs 29.2%, p < 0.001); more had alcohol drinking (8.9% vs 5.6%, p = 0.001), hypertension (33.7% vs 17.6%, p < 0.001), diabetes mellitus (9.4% vs 5.8%, p < 0.001), CKD (27% vs 18.2%, p < 0.001) and history of gout (21.9% vs 2%, p < 0.001); and had higher BMI (25.3 ± 3.8 vs 22.9 ± 3.2, p < 0.001), higher aspartate aminotransferase (26.5 ± 13.9 vs 22.5 ± 20, p < 0.001), higher alanine aminotransferase (33.5 ± 23.3 vs 25.5 ± 20.1, p < 0.001), lower HDL-C (42.4 ± 11.6 vs 51.4 ± 14.8 mg/dL, p < 0.001), higher albumin (4.3 ± 0.4 vs 4.2 ± 0.3 mg/dL), higher creatinine (1.1 ± 0.6 vs 0.8 ± 0.3 mg/dL, p < 0.001), and lower CKD-EPI eGFR (76 ± 17 vs 87 ± 13 mL/min/1.73 m2, p < 0.001) values.

Table 1 - The demographics, comorbidities, and biochemical examinations in vegetarians with or without hyperuricemia Characteristics With hyperuricemia (n = 593) Without hyperuricemia (n = 6738) p Age group, y, n (%)  40-49 69 (11.6) 797 (11.8) 0.89a  50-69 181 (30.5) 2245 (33.3) 0.17a  60-69 209 (35.2) 2546 (37.8) 0.22a  ≥70 134 (22.6) 1150 (17.1) 0.001a  Age, y, mean (SD) 62.2 (10.3) 60.9 (9.4) 0.001b Gender, n (%)  Male 444 (74.9) 1965 (29.2) <0.001a  Female 149 (25.1) 4773 (70.8) <0.001a Current smoking, n (%) 7 (1.2) 38 (0.6) 0.07a Alcohol drinking, n (%) 53 (8.9) 374 (5.6) 0.001a BMI, kg/m2, mean (SD) 25.3 (3.8) 22.9 (3.2) <0.001b  >27, n (%) 151 (25.5) 707 (10.5) <0.001a Hypertension, n (%) 200 (33.7) 1187 (17.6) <0.001a SBP, mmHg, mean (SD) 126 (15) 119 (15) <0.001b Diabetes, n (%) 56 (9.4) 393 (5.8) <0.001a HbA1c, %, mean (SD) 5.7 (0.8) 5.6 (0.8) 0.08b AST, IU/L, mean (SD) 26.5 (13.9) 22.5 (20.0) <0.001b ALT, IU/L, mean (SD) 33.5 (23.3) 25.5 (20.1) <0.001b HDL-C, mg/dL, mean (SD) 42.4 (11.6) 51.4 (14.8) <0.001b LDL-C, mg/dL, mean (SD) 119.3 (30.8) 116.1 (30.4) 0.07b Serum albumin, g/dL, mean (SD) 4.3 (0.4) 4.2 (0.3) <0.001b Gout history, n (%) 130 (21.9) 132 (2.0) <0.001a Proteinuria, n (%) 92 (15.5) 1054 (15.6) 0.93a Serum creatinine, mg/dL, mean (SD) 1.1 (0.6) 0.8 (0.3) <0.001b eGFRc, mL/min/1.73 m2, mean (SD) 76 (17) 87 (13) <0.001b CKD, n (%) 160 (27.0) 1223 (18.2) <0.001a CKD stage, n (%)  3 132 (22.3) 1177 (17.5) <0.001a  4-5 12 (2.0) 18 (0.3) <0.001a

ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; HbA1c = glycated hemoglobin; HDL-C = high-density lipoprotein cholesterol; IU = international unit; LDL-C = low-density lipoprotein cholesterol; OR = odds ratio; SBP = systolic blood pressure.

aChi-square test.

bStudent’s t test.

cCalculated by using the Chronic Kidney Disease Epidemiology Collaboration formula.

3.2. ROC curves for prediction of hyperuricemia

Fig. 2 shows the ROC curves of three models for predicting hyperuricemia in vegetarians. Model 1 includes age and gender. The area under the ROC curve of model 1 is 75.72%. In addition, model 2 includes variables in model 1 plus diabetes mellitus, hypertension, hyperlipidemia, and proteinuria. The area under the ROC curve of model 2 is 81.01%. Moreover, model 3 includes variables in model 2 and eGFR, systolic blood pressure, glycated hemoglobin, HDL-C, LDL-C, BMI, albumin, alcohol drinking, and dietary habits (vegan vs lacto-ovo vegetarian diets). The area under the ROC curve of model 3 is 85.52%, the highest among the three models. Fig. 3A illustrates the ROC curves of different prediction models for hyperuricemia in vegetarians without gout history. The area under the ROC curves is 64.59%, 70.17%, and 77.97% for model 1, model 2, and model 3, respectively. With regard to vegetarians with gout history, the area under the ROC curves is 75.16%, 79.81%, and 84.85% for model 1, model 2, and model 3, respectively (Fig. 3B).

Fig. 2:

The ROC curves of prediction models for hyperuricemia in vegetarians. Model 1: adjusted for age and gender. Model 2: adjusted for the variables in model 1 and diabetes, hypertension, hyperlipidemia, and proteinuria. Model 3: adjusted for the variables in model 2 and estimated glomerular filtration rate, systolic blood pressure, glycated hemoglobin, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, body mass index, serum albumin, alcohol drinking, and dietary patterns (lacto-ovo vegetarian vs vegan diet). ROC = receiver operating characteristic.

Fig. 3:

The ROC curves of prediction models for hyperuricemia in vegetarians with (A) or without (B) gout history. Model 1: adjusted for age and gender. Model 2: adjusted for the variables in model 1 and diabetes, hypertension, hyperlipidemia, and proteinuria. Model 3: adjusted for the variables in model 2 and estimated glomerular filtration rate, systolic blood pressure, glycated hemoglobin, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, body mass index, serum albumin, alcohol drinking, and dietary patterns (lacto-ovo vegetarian vs vegan diet). ROC = receiver operating characteristic.

3.3. Multivariable logistic regression models

Table 2 shows the association of each variable with the presence of hyperuricemia in the three models in vegetarians. In model 1, age (adjusted OR: 1.01 per year, p = 0.002) and male gender (adjusted OR: 7.22, p < 0.001) are independent risk factors for hyperuricemia. After adding diabetes mellitus, hypertension, hyperlipidemia, and proteinuria into the model (model 2), age (adjusted OR: 1.02 per year, p = 0.008), male gender (adjusted OR: 15.8, p < 0.001), hypertension (adjusted OR: 1.84, p < 0.001), and hyperlipidemia (adjusted OR: 1.08, p < 0.001) are independent risk factors for hyperuricemia. In the fully adjusted model (model 3), male gender (adjusted OR: 8.29, p < 0.001), hyperlipidemia (adjusted OR: 1.04, p = 0.024), BMI (adjusted OR: 1.13 per kg/m2, p < 0.001), and serum albumin (adjusted OR: 2.18 per g/dL, p = 0.001) are independent risk factors for hyperuricemia in vegetarians. In contrast, proteinuria (adjusted OR: 0.66, p = 0.01) and eGFR (adjusted OR: 0.95 per mL/min/1.73 m2, p < 0.001) are associated with lower risks of hyperuricemia in vegetarians.

Table 2 - Multiple logistic regression models for hyperuricemia in vegetarians Model 1a Model 2b Model 3c Adjusted OR (95% CI) p Adjusted OR (95% CI) p Adjusted OR (95% CI) p Age, per year 1.01 (1.00-1.02) 0.002 1.02 (1.00-1.03) 0.008 0.98 (0.97-1.00) 0.07 Male vs female 7.22 (5.95-8.76) <0.001 15.8 (11.4-22.1) <0.001 8.29 (5.32-12.9) <0.001 Diabetes 0.98 (0.65-1.48) 0.92 0.92 (0.54-1.58) 0.76 Hypertension 1.84 (1.40-2.40) <0.001 1.28 (0.95-1.73) 0.11 Hyperlipidemia 1.08 (1.06-1.10) <0.001 1.04 (1.00-1.07) 0.024 Proteinuria 0.80 (0.60-1.08) 0.14 0.66 (0.48-0.90) 0.01 eGFRd, per mL/min/1.73 m2 0.95 (0.94-0.96) <0.001 HbA1c, % 0.87 (0.72-1.00) 0.13 SBP, per 10 mmHg 1.00 (0.91-1.09) 0.93 HDL-C, per mg/dL 0.99 (0.98-1.00) 0.06 LDL-C, per mg/dL 1.04 (0.99-1.08) 0.1 BMI, per kg/m2 1.13 (1.06-1.20) <0.001 Serum albumin, per g/dL 2.18 (1.37-3.46) 0.001 Alcohol drinking 0.80 (0.50-1.28) 0.36 Lacto-ovo vegetarian vs vegan diet 1.17 (0.83-1.66) 0.36

BMI = body mass index; CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; HbA1c = glycated hemoglobin; HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density lipoprotein cholesterol; OR = odds ratio; SBP = systolic blood pressure.

aAdjusted for age and gender.

bAdjusted for the variables in model 1 and diabetes, hypertension, hyperlipidemia, and proteinuria.

cAdjusted for the variables in model 2 and eGFR, SBP, HbA1c, HDL-C, LDL-C, BMI, serum albumin, alcohol drinking, and lacto-ovo vegetarian vs vegan diet.

dCalculated by using the Chronic Kidney Disease Epidemiology Collaboration formula.

With regard to participants without gout history (Table 3), male gender (adjusted OR: 6.54, p < 0.001), hyperlipidemia (adjusted OR: 1.03, p = 0.046), BMI (adjusted OR: 1.12 per kg/m2, p = 0.001), and serum albumin (adjusted OR: 2.73 per g/dL, p < 0.001) are independent risk factors for hyperuricemia in vegetarians. However, eGFR (adjusted OR: 0.95 per mL/min/1.73 m2, p < 0.001) and HDL-C (adjusted OR: 0.98 per mg/dL, p = 0.001) are independently associated with lower risks for hyperuricemia in vegetarian (model 3). Among those with gout history, male gender (adjusted OR: 9.47, p = 0.012) and HDL-C (adjusted OR: 1.06 per mg/dL, p = 0.012) are independent risk factors for hyperuricemia in vegetarians. In contrast, age (adjusted OR: 0.89 per year, p < 0.001) and eGFR (adjusted OR: 0.94 per mL/min/1.73 m2, p = 0.001) are independently linked to lower risks for hyperuricemia in vegetarians (Table 3).

Table 3 - Multiple logistic regression models for hyperuricemia in vegetarians by gout history Model 1a Model 2b Model 3c Adjusted OR (95% CI) p Adjusted OR (95% CI) p Adjusted OR (95% CI) p Without gout history  Age, per year 1.02 (1.01-1.03) <0.001 1.03 (1.01-1.04) <0.001 1.00 (0.98-1.02) 0.92  Male vs female 6.22 (5.05-7.66) <0.001 13.7 (9.56-19.5) <0.001 6.54 (4.05-10.6) <0.001  Diabetes 1.03 (0.65-1.61) 0.91 0.97 (0.54-1.75) 0.92  Hypertension 1.59 (1.18-2.14) 0.002 1.12 (0.80-1.56) 0.52  Hyperlipidemia 1.08 (1.06-1.10) <0.001 1.03 (1.00-1.07) 0.046  Proteinuria 0.90 (0.66-1.23) 0.52 0.74 (0.53-1.04) 0.08  CKD-Epi eGFR, per mL/min/1.73 m2 0.95 (0.94-0.96) <0.001  HbA1c, % 0.81 (0.66-1.01) 0.06  Systolic blood pressure, per 10 mmHg 0.98 (0.89-1.08) 0.65  High-density lipoprotein, per mg/dL 0.98 (0.97-0.99) 0.001  Low-density lipoprotein, per mg/dL 1.03 (0.99-1.08) 0.15  BMI, per kg/m2 1.12 (1.05-1.19) 0.001  Serum albumin, per g/dL 2.73 (1.64-4.53) <0.001  Alcohol drinking 0.80 (0.47-1.36) 0.41  Lacto-ovo vegetarian vs vegetarian diet 1.17 (0.81-1.70) 0.41 With gout history  Age, per years old 0.98 (0.95-1.00) 0.06 0.96 (0.92-1.00) 0.042 0.89 (0.83-0.94) <0.001  Male vs female 2.21 (1.12-4.36) 0.022 4.02 (1.16-13.9) 0.028 9.47 (1.65-54.3) 0.012  Diabetes 0.63 (0.19-2.11) 0.46 1.56 (0.29-8.34) 0.61  Hypertension 2.09 (0.99-4.43) 0.053 1.29 (0.51-3.28) 0.59  Hyperlipidemia 1.03 (0.98-1.09) 0.24 1.04 (0.94-1.15) 0.44  Proteinuria 0.59 (0.22-1.57) 0.29 0.59 (0.20-1.79) 0.35  eGFRd, per mL/min/1.73 m2 0.94 (0.91-0.97) 0.001  HbA1c, % 0.74 (0.48-1.13) 0.16  Systolic blood pressure, per 10 mmHg 1.11 (0.81-1.53) 0.52  HDL-C, per mg/dL 1.06 (1.01-1.11) 0.012  LDL-C, per mg/dL