Prevalence of hyperuricemia and its related risk factors in healthy adults from Northern and Northeastern Chinese provinces

The population-based cross-sectional survey, Chinese Physiological Constant and Health Condition (CPCHC), was conducted during 2008 and 2010. Our research used the samples from Heilongjiang Province (2008) and Inner Mongolian Autonomous Region (2010) in this survey. To maximize the accuracy of prevalence estimates, subjects were selected using a random, multistage cluster sampling scheme in four regions of each province selected according to administrative and economic levels. The four regions of Heilongjiang and Inner Mongolia are Harbin city, Harbin suburbs, Mudanjiang, Hailin and Hohhot, Ulanqab City, Xilin Gol League, Bayannur, which belong to big cities, suburb of big cities, small and medium-sized cities and rural and pastoral areas respectively according to population and economics. Written informed consent was obtained from each participant before data collection. The protocol was approved by the Institutional Review Board of the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences.

The 2008–2010 CPCHC samples included 29,639 apparently healthy participants. Those who suffered from systemic disease involving diabetes mellitus, hypertension or other cardiovascular, renal, gastro-intestinal, pulmonary disease or cancer were excluded. Moreover, participants taking any medication known to affect carbohydrate and lipid metabolism were also excluded. Of the total number of participants, 44.3% (n = 13,140) were selected randomly to complete blood testing. After excluding individuals aged < 18 years, where the important variables was missing, and the results of biochemical tests exceeded four times the standard deviation of the sample, our final sample size was 8,439 adults (3,725 men and 4,714 women), of which 5,859 were Han Chinese, 954 were Korean-Chinese, 1,292 were Mongolian-Chinese, and 334 were of other ethnicity.

A standard questionnaire was used to obtain demographic characteristics (age, gender, and ethnicity), socioeconomic data (education level, marital status, and occupation), lifestyle, past medical history, and lifestyle risk factors (diet, exercise, smoking, and drinking).

Body weight was measured to the nearest 0.1 kg on a calibrated beam scale and height was measured to the nearest 0.1 cm barefoot in triplicate using a wall-mounted stadiometer. Body mass index (BMI, an index for overall obesity) was calculated as body weight (in kilograms) divided by height (in square meters). Waist circumference (WC) was measured midway between the lower rib margin and the iliac crest at the end of a gentle expiration. Blood pressure was measured after the participant had rested quietly for at least 10 minutes, using an electric sphygmomanometer (OMRON, HEM-7000). Blood pressure was measured three times and the measurements were averaged. When subjects were found to be hypertensive, a physician review was performed to exclude secondary hypertension. All personnel who participated in data collection and anthropometry were trained medical personnel, and all devices were calibrated.

All procedures were performed following a 9–12 hour overnight fast and all subjects were told to take bland diet before blood testing. Blood was drawn from the antecubital vein of the arm. Total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) were measured with a Beckman AU Series Automatic Biochemical Analyzer (Japan), using Sekisui Medical (Japan) reagents. Uric acid (UA), Fasting blood glucose (FBG), creatine, and 17 other biochemical tests were measured with the same instrument, using Beckman AU reagents. All laboratories participating in the survey followed the same internal quality control program that was standardized by the Peking Union Medical College Hospital. Estimated Glomerular Filtration Rate (eGFR)was calculated by the CG/BSA formula [16].

HUA was defined as SUA ≥ 416.4 μmol/L (male) or SUA ≥ 356.9 μmol/L (female). Blood pressure was classified as normotensive (SBP < 120 mmHg and DBP < 80 mmHg), pre-hypertensive (SBP: 120–139 mmHg and/or DBP: 80–89 mmHg) and hypertensive (SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg) by the Seventh Report of the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7). Body mass index (BMI: kg/m²) was calculated from measured weight and height. BMI was classified as underweight (< 18.5 kg/m²), normal (18.5–25 kg/m²), overweight (25–30 kg/m²) and obese (> 30 kg/m²) by WHO criteria. Abdominal obesity was defined as WC ≥ 90 cm (male) or WC ≥ 80 cm (female) by IDF consensus. The American Diabetes Association criteria was used to classify FBG as normal glucose (FBG < 5.6 mmol/L), impaired fasting glucose (IFG) (FBG ≥ 5.6 mmol/L ≤ FBG < 7.0 mmol/L), and diabetic (FBG ≥ 7.0 mmol/L). Dyslipidemia was classified according to ATP III, TG: Normal < 1.69 mmol/L, Borderline high 1.69–2.26 mmol/L, High 2.26–5.65 mmol/L, Very high ≥ 5.65 mmol/L; TC: Desirable < 5.17 mmol/L, Borderline high 5.17–6.24 mmol/L, High ≥ 6.24 mmol/L; HDL-C: High 1.56 mmol/L, Optimal 1.03–1.56 mmol/L, Low <l.03 mmol/L; LDL-C: Optimal <2.59 mmol/L, Near optimal 2.59–3.38 mmol/L, Borderline high 3.38–4.16 mmol/L, High 4.16–4.94 mmol/L, Very high ≥4.94 mmol/L.

Epidata 3.2 was used for data entry and verification. Data were entered twice by two different people, and inconsistencies were checked with the original records. SAS 9.2 was used for data processing and statistical analysis. For univariate analyses, SUA levels and the prevalence of HUA were described by gender, age and regional groups. Hyperuricemia was considered as the outcome variable, and 7 demographic indicators, 18 habits (including smoking, drinking, labor work, exercise, diet habit, sleeping, and sitting time etc.), 37 parameters on health status (including medical history, sub-health state) and 9 metabolic markers obtained through physical examination and biochemical tests were set as predictor variables. In multivariate analyses, factors that may affect uric acid or HUA were introduced into a logistic regression model. Sub-health and past medical history factors were not included in the model because of a large amount of missing data. The main factors include demographic information, metabolic indicators, and behavioral factors of life including eating habits, physical exercise, smoking and drinking status, sleep time and other factors. The impact of sub-health and medical history on HUA was confirmed by univariate analysis and expert judgment. Mean ± standard deviation was used to describe the levels of uric acid, which is continuous variable with a normal distribution. Prevalence (%) indicates the percentage of healthy Chinese men and women with a condition at the time of data collection. Quantitative indicator comparison was conducted by t-test, and analysis of variance (ANOVA) with the Student-Newman-Keuls (SNK) post-test was used for group comparisons. Results between men and women were compared using the χ2 test. Qualitative data were compared by chi-square partitioning to allow α correction; P < 2α / n (n - 1) was considered statistically significant. The research applied binary multi-factor and non-conditional logistic regression models with stepwise regression. The entering criterion for the logistic regression models was 0.05, and the exclusion criterion was 0.10, to obtain the optimal model for the evaluation of the risk factors for HUA. Odds ratios (OR) and 95% confidence intervals (CI) were calculated to study the associations between various risk factors and HUA. P <0.05 was considered statistically significant.

In men, the average uric acid level over the age of 55 was lower than that under 55, but there were no statistically significant differences between average uric acid levels aged 18–24, 25–34, 35–44, 45–54 years and aged 55–64, 65–74, over 75 years respectively (Table 1). In women, the average uric acid levels gradually increased over 45 years of age. Uric acid levels in all age groups were significantly higher in men than that in women (P <0.01), and the total mean uric acid levels in men was higher than that in women (357.7 μmol/L versus 263.2 μmol/L, P < 0.0001) (Table 1 and Additional file 1: Figure S1). Among four different regions, uric acid levels in suburbs of big cities were highest in men, women and the entire study population (Table 1). Participants from small and medium-sized cities and agricultural and pastoral areas have lower average uric acid levels (294.5 μmol/L and 295.6 μmol/L) than that from the center of big cities and suburbs of big cities (309.2 μmol/L and 325.7 μmol/L, P < 0.05). There were no significant differences between uric acid levels in participants from small and medium-sized cities and those from rural and pastoral areas (Table 1). In different ethnic groups, there were significantly differences between uric acid levels in women and the entire study population but no statistical significance in men by SNK test.

As shown in Table 2, the total prevalence of HUA is 13.7%. Men had a higher prevalence of HUA than women (21% vs. 7.9%; P < 0.0001). In men, the HUA prevalence decreased with increasing age. In women, HUA prevalence increased in those over 45 years old. In participants over 65 years old, HUA prevalence in men and women was similar (P > 0.05). In those over 75 years old, the prevalence in women was higher than men, but not significantly higher (P > 0.05) (Table 2). In different regions, the prevalence of HUA in men was significantly higher than women and the suburbs of big cities have the highest HUA prevalence in male (28.0%), female (14.0%) and total (20.3%). In men, the HUA prevalence in agricultural and pastoral areas (18.5%) versus small and medium-sized cities (17.1%) was not significantly different (P > 0.05). However, HUA prevalence in the center of big cities (22.6%) and suburbs of big cities (28.0%) was significantly higher than agricultural and pastoral areas, as well as significantly higher than small and medium-sized cities. In women, the prevalence tendency of all regions was similar with men (Table 2). It was also found that, in areas with higher HUA prevalence, the proportion of female patients was significantly higher, which were 25.7% and 28.7% in small and medium-sized cities and rural and pastoral areas respectively, while in the metropolitan central area and suburban, the HUA prevalence ratio in female are 34.1% and 37.8% respectively (Figure 1).

After stepwise logistic regression screening, 12 variables remained in the model. These were gender, area of residence, age group, hypercholesterolemia, hypertriglyceridemia, impaired fasting glucose, hypertension, obesity, CKD, abdominal obesity, drinking, and sleeping time. Pearson goodness of fit showed a good model fitness (P = 0.9896). In the fully adjusted model with these 12 co-variables, the odds of women having HUA was 0.52 (0.42–0.64) times the odds of men (Table 3). Participants in the suburbs of big cities have an odds of HUA 2.14 (1.71–2.68) times that of the agricultural and pastoral areas, while those in the center of big cities have an odds of HUA 1.57 (1.25–1.98) times that of the agricultural and pastoral areas. With an increase in lipid levels, the odds of HUA also increased. Compared with the control group, the odds of HUA in the hypercholesterolemia group was 1.48 (1.12–1.94). Similarly, compared with the control group the odds ratio of HUA in the hypertriglyceridemia group was 4.30 (2.69–6.86). The fasting glucose impairment group had an odds of HUA 1.27 (1.05–1.54) times higher than the odds of the respective control group, while the diabetic group had an odds of HUA 0.59 (0.40–0.86) times of the normal glucose group. The odds of HUA in the hypertensive group was 1.53 (1.18–1.99) times the odds of the normotensive group. The odds of HUA increased along with an increase in obesity; the odds of HUA in the obese group was 2.42 (1.73–3.39) times the odds of the normal weight group. Compared with the respective control group, the odds ratio of the abdominal obesity group 1.82 (1.46–2.26). As the CKD classification increased, the odds of HUA also increased. The CKD3 class had an HUA odds 6.34 (3.96–10.13) times higher than that of the control group. The odds ratio of HUA in participants who drink group was 1.36 (1.11–1.67) times the non-drinking group. The group with less than four hours of sleep had an odds of HUA 2.48 (1.39–4.42) times the control group.

Odds ratios for various risk factors were also calculated separately for men and women (Additional file 1: Table S1 and Table S2). After stepwise logistic regression screening for men, 13 variables remained in the model. These were residence, work status, age group, cholesterol level, triglycerides, fasting glucose, obesity, CKD, abdominal obesity, diet, sleep time, drinking, and physical exercise. The fully adjusted model indicates that abdominal obesity, hypertriglyceridemia and less sleep time are associated with higher odds of HUA. Oppositely, bland diet group has less odds of HUA. For women, eight variables remained in the model: age group, area of residence, triglycerides, low-density lipoprotein cholesterol, high blood pressure, obesity, CKD, and abdominal obesity. Menopause status was included as a possible co-variable, but was not selected by the stepwise regression for inclusion in the final model. Unlike the general population and men, impaired fasting glucose and total cholesterol level were not selected for inclusion into the model for women. Compared with LDL-C < 2.59 mmol/L, the odds ratio of HUA of LDL-C > 3.38 mmol/L was elevated. The impact of residence was also significant for women: compared with the odds of HUA of participants living in agricultural and pastoral areas, the odds ratio of HUA in the suburbs of big cities was 3.52 (2.35–5.28) and in the center of big cities was 2.24 (1.47–3.40).