The association between hyperuricemia and cardiovascular events has been documented in high-risk groups, but is still undetermined in general populations, especially Chinese. This study assessed the temporal association between serum uric acid level, hyperuricemia, and cardiovascular mortality.
A prospective cohort study of 41,879 men and 48,514 women ages ≥35 years was conducted using data from the MJ Health Screening Centers in Taiwan. Mortality from all causes, total cardiovascular disease (CVD), ischemic stroke, congestive heart failure, hypertensive disease, and coronary heart disease were compared according to increasing serum uric acid levels.
A total of 1,151 (21.2%) events of 5,427 total deaths were ascribed to CVD (mean followup 8.2 years). Hazard ratios (HRs) for hyperuricemia (serum uric acid level >7 mg/dl) were estimated with Cox regression model after adjusting for age, sex, body mass index, cholesterol, triglycerides, diabetes, hypertension, heavy cigarette smoking, and frequent alcohol consumption. In all patients, HRs were 1.16 (P < 0.001) for all-cause mortality, 1.39 (P < 0.001) for total CVD, and 1.35 (P = 0.02) for ischemic stroke. In subgroup analysis, the HRs for cardiovascular risk remained significant in patients with hypertension (1.44, P < 0.001) and in patients with diabetes (1.64, P < 0.001). In addition, in a low metabolic risk subgroup, the HRs for all-cause mortality and total cardiovascular morbidity were 1.24 (P = 0.02) and 1.48 (P = 0.16), respectively.
Hyperuricemia was an independent risk factor of mortality from all causes, total CVD, and ischemic stroke in the Taiwanese general population, in high-risk groups, and potentially in low-risk groups.
Elevated serum uric acid levels are associated with gouty arthritis, which has been reported in several well-known studies (1–4) to increase the risk of cardiovascular disease (CVD). In recent decades, although serum uric acid has been recognized as a potential risk factor for CVD, conflicting conclusions about the association between serum uric acid level and CVD are still noted (5–7). Elevated serum uric acid levels were shown to increase the risk of CVD in prospective cohort studies concerning high-risk groups such as patients with gout, hypertension (8), stroke (9), congestive heart failure (CHF) (10), and/or diabetes (11). This observation was ascribed to the association of high serum uric acid levels with metabolic syndrome and insulin resistance (12, 13).
The Framingham Study was the first to examine the independent association between serum uric acid level elevation and cardiovascular outcomes in the general population, but it failed to provide consistent conclusions in 2 separate investigations (1, 14). Additionally, the First National Health and Nutrition Examination Survey (NHANES-I) found an independent relationship between elevated serum uric acid levels and CVD-related mortality that demonstrated a higher relative risk (RR) in women than in men (RR 1.30 versus 1.17) (15, 16). It was argued that the number of patients with pathologic serum uric acid level elevation in some of these studies was not sufficiently large enough to provide power for a conclusive association (6). The difficulty in establishing a solid correlation may also have been a consequence of adjusting for variable numbers of confounders (17, 18). Therefore, it remains to be determined whether an elevated serum uric acid level is an independent risk factor of CVD in the general population.
According to statistics from the Nutrition and Health Survey in Taiwan (19), the prevalences of hyperuricemia in men (serum uric acid level >7.0 mg/dl) and women (serum uric acid level >6.0 mg/dl) were as high as 42.1% and 27.4%, respectively. The mean serum uric acid levels (6.63 mg/dl for men and 5.62 mg/dl for women for representative Taiwanese age >45 years) and prevalence of hyperuricemia were higher in the Taiwanese population as compared with those of other ethnic groups (14, 20) and other world regions (21). Therefore, it is timely and crucial to understand its impact on CVD risk in Taiwan (12, 17).
We used the MJ Health Screening Cohort, one of the largest cohorts established in Taiwan, to identify the association between hyperuricemia (serum uric acid level >7.0 mg/dl), incremental serum uric acid levels, and mortality due to various cardiovascular end points in Chinese adults.
PATIENTS AND METHODS
Clinical data were collected from privately owned nationwide MJ Health Screening Centers in Taiwan from 1994–1996. Participants from the screening centers of the MJ Health Management Institution were from all regions of Taiwan. They paid to take the health examination in 4 centers located in Taipei (north), Taichung (central west), Kaohsiung (south), and Hualien (east). We obtained data from those who agreed to release the data for research. The socioeconomic distribution of these participants was close to that of Taiwanese people (22). More than one-quarter of the participants were classified as having low socioeconomic status, with average or poor education. In addition, the MJ cohort data have been used in several recent publications on some timely and important research topics (22–24).
The content of the health examination included a structured questionnaire for demographic characteristics, medical history, medication, and lifestyle factors (such as personal habits, food intake, and physical activity), and anthropometric measurement for body weight, height, blood pressure, and fasting blood specimens. Serum uric acid level, cholesterol, triglycerides, and plasma glucose were measured enzymatically using the Hitachi 7150 autoanalyzer (Hitachi, Tokyo, Japan). Statistical analysis was based on one measurement of these clinical chemistry markers and on 3 consecutive measurements of blood pressure at baseline.
Serum uric acid levels were categorized into quartiles: ≤5 mg/dl (reference), 5.1–7 mg/dl, 7.1–9 mg/dl, and >9 mg/dl. Hyperuricemia was defined as serum uric acid levels >7.0 mg/dl. Cholesterol was categorized into 3 levels: ≤200 mg/dl, 201–240 mg/dl, and >240 mg/dl. Triglycerides were also categorized into 3 levels: ≤150 mg/dl, 151–200 mg/dl, and >200 mg/dl. Hypertension was defined as either systolic blood pressure >140 mm Hg, diastolic blood pressure >90 mm Hg, or use of antihypertensive medications. Diabetes was defined as fasting blood sugar >126 mg/dl or use of antidiabetic medications. Body mass index (BMI) was calculated as weight (in kg) divided by height (in m2). Frequent alcohol consumption was defined as those reporting frequent or daily alcohol use versus never, abstaining, or occasional use. The definition of frequent cigarette smoking was also defined as those reporting frequent or daily smoking habits on the patient questionnaire. Overall, there were 146,900 individuals (67,954 men and 78,946 women) that participated in this screening program from 1994–1996. After excluding participants age <35 years, with duplicate identification numbers, or with missing or unconfirmed data, 90,393 individuals (41,879 men and 48,514 women) were followed until December 31, 2003.
Death information in this study was obtained from the National Mortality Registry Office. Because it is mandatory to report death events in Taiwan, the completeness of the registry is nearly 100% (25, 26). In the current study, only 0.46% of the cohort data was excluded due to inability to match with self-reported birth date and social security number. The participant's cause of death was ascertained using the International Classification of Diseases, Ninth Revision (ICD-9). The classification for causes of death included total CVD (ICD-9 codes 390–459), ischemic stroke (ICD-9 codes 433–438), hemorrhagic stroke (ICD-9 codes 430–432), coronary heart disease (ICD-9 codes 410–414), CHF (ICD-9 code 428), hypertensive disease (ICD-9 codes 401–405), and all-cause mortality. This study was approved by the Institutional Review Board of Academia Sinica, Taipei, Taiwan.
Clinical data were analyzed using both univariate and multivariate techniques. The serum biochemical data comparison between sexes was analyzed with either a 2-sample t-test or a chi-square test. Three different ways of categorizing serum uric acid levels were tested in various analyses: as hyperuricemia (>7 mg/dl), as ordered categories, and as quartiles. Kaplan-Meier survival curves were drawn to demonstrate risk of ordered serum uric acid level categories on mortality of CVD. Incidence rates for CVD mortality were presented by serum uric acid level quartiles and by ordered serum uric acid level categories. The Cox proportional hazards model was used to estimate the RR of hyperuricemia on various cardiovascular end points and all-cause mortality. The multivariate Cox regression models included one model adjusted for age and sex and another model adjusted for age, sex, BMI, cholesterol, triglycerides, diabetes, hypertension, heavy cigarette smoking, and frequent alcohol consumption. The incremental risk ratio for every mg/dl increase in serum uric acid levels was also estimated. Furthermore, we examined the risk of hyperuricemia in 2 high-risk subgroups: patients with hypertension and diabetes and a low metabolic risk group. Stratification by sex was also applied to the survival analysis and HRs were adjusted for age, BMI, cholesterol, triglycerides, diabetes, hypertension, heavy cigarette smoking, and frequent alcohol consumption. The statistical software SAS version 9.01 (SAS Institute, Cary, NC) was used for analysis.
The mean ± SD age of the participants was 51.5 ± 11.5 years (Table 1). Hyperuricemia (serum uric acid level >7 mg/dl) was seen in 24.4% of the participants, with a male predominance (39.7% men, 11.3% women). Serum uric acid levels increased with age in women but not in men. Serum uric acid levels in 81.3% of men were between 5 and 9 mg/dl, but most women (88.8%) had serum uric acid levels <7 mg/dl. The mean ± SD followup duration was 8.2 ± 1.3 years. There were 5,427 deaths documented during followup. Among the total deaths, 1,151 (21.2%) events were ascribed to CVD (344 ischemic stroke, 200 hemorrhagic stroke, 286 coronary heart disease, 105 hypertensive disease, 60 CHF, and 156 other CVD) (Table 2).
Table 1. Baseline characteristics of participants from the MJ Health Screening Center, stratified by sex*
Men (n = 41,879)
Women (n = 48,514)
Values are the number (percentage) unless otherwise indicated. All variables were tested by 2-sample t-test or chi-square test, and significant differences were found between sexes (P < 0.05) except when analyzing body mass index (BMI) and diabetes status.
Age, mean ± SD years
51.7 ± 12.0
51.4 ± 11.0
Followup duration, mean ± SD years
8.1 ± 1.4
8.3 ± 1.2
BMI, mean ± SD kg/m2
24.2 ± 3.2
24.2 ± 11.8
Cholesterol, mean ± SD mg/dl
191.1 ± 38.5
193.8 ± 40.4
Triglycerides, mean ± SD mg/dl
146.3 ± 126.1
120.5 ± 96.1
Glucose, mean ± SD mg/dl
102.0 ± 28.4
101.1 ± 29.6
Uric acid, mean ± SD mg/dl
6.8 ± 1.6
5.4 ± 1.4
Serum uric acid level by age group, mean ± SD mg/dl (no.)
6.90 ± 1.49 (9,135)
4.95 ± 1.19 (9,532)
6.85 ± 1.49 (11,355)
5.11 ± 1.27 (13,670)
6.68 ± 1.67 (9,667)
5.56 ± 1.43 (13,981)
6.72 ± 1.66 (11,722)
5.79 ± 1.59 (11,331)
Serum uric acid level, mg/dl
Hyperuricemia (serum uric acid level >7 mg/dl)
Hypercholesterolemia (cholesterol >240 mg/dl)
Hypertriglyceridemia (triglycerides >200 mg/dl)
Frequent alcohol drinker
Heavy cigarette smoker
Table 2. Risk of hyperuricemia or per unit increase in serum uric acid level on mortality of all causes and various types of CVD (n = 90,393)*
Cause of death
No. of events
HR (95% CI) for hyperuricemia
HR (95% CI) per 1 mg/dl change in serum uric acid level
HR (95% CI) for hyperuricemia
HR (95% CI) for hyperuricemia
Hazard ratios (HRs) were adjusted for age, sex, body mass index, cholesterol, triglycerides, diabetes, hypertension, heavy cigarette smoking, and frequent alcohol consumption, and were stratified by sex. CVD = cardiovascular disease; 95% CI = 95% confidence interval; CHF = congestive heart failure.
Coronary heart disease
Effect of hyperuricemia in all patients.
In univariate analysis, survival curves in the absence of cardiovascular death significantly differed among 4 groups of participants categorized by their serum uric acid levels (Figure 1). The lower serum uric acid level group yielded better survivorship results. Serum uric acid level and mortality exhibited a dose-response relationship. All-cause or total CVD mortality increased 8% or 11% per mg/dl increase of serum uric acid level, respectively (Table 2). The age- and sex-adjusted risk of hyperuricemia (serum uric acid level >7 mg/dl) for total cardiovascular mortality (no tabulated data) was statistically significant with an HR of 1.63 (95% confidence interval [95% CI] 1.44–1.84, P < 0.001). The RRs of hyperuricemia for different types of cardiovascular death were as follows: HRs were 1.59 (95% CI 1.27–1.99, P < 0.001) for ischemic stroke, 1.74 (95% CI 1.02–2.96, P = 0.04) for CHF, 1.94 (95% CI 1.30–2.89, P = 0.001) for hypertensive disease, and 1.62 (95% CI 1.27–2.06, P = 0.001) for coronary heart disease. To examine whether hyperuricemia was an independent risk factor for CVD mortality, other conventional factors such as obesity, hypertension, diabetes, hyperlipidemia, smoking, and alcohol consumption were adjusted (Table 2). The multivariate adjusted HR for total cardiovascular death was statistically independent, but was reduced to 1.39 (95% CI 1.20–1.60, P < 0.001). With the exception of coronary heart disease, the adjusted HRs of hyperuricemia for ischemic stroke (1.35, 95% CI 1.04–1.76; P = 0.02) and CHF (2.05, 95% CI 1.11–3.78; P = 0.02) remained significant. Hyperuricemia had no significant impact on mortality of hemorrhagic stroke (HR 1.18, 95% CI 0.83–1.67; P = 0.36).
Effect of hyperuricemia in high-risk (patients with hypertension and diabetes) and low-risk subgroups.
Hypertension and diabetes were significantly associated with all-cause and cardiovascular mortality, with HRs of 1.47 (95% CI 1.37–1.58) and 1.89 (95% CI 1.75–2.04), respectively, for total deaths, and 2.82 (95% CI 2.37–3.35) and 1.67 (95% CI 1.42–1.98), respectively, for total cardiovascular deaths after multivariate adjustment (no tabulated data). There were 29,421 patients with hypertension (13,475 men and 15,946 women) and 7,899 patients with diabetes (3,623 men and 4,276 women) allocated to the high-risk subgroup analysis. The mean ages for both subgroups were ∼58–59 years. During followup, 882 (27.0%) cardiovascular events occurred in 3,272 total deaths among patients with hypertension and 278 (21.9%) cardiovascular events occurred in 1,271 total deaths among patients with diabetes. We found that the RRs of hyperuricemia (serum uric acid level >7 mg/dl) for all-cause mortality and total CVD mortality in both sexes combined were statistically significant in either patients with hypertension or diabetes (Figure 2).
An analysis of participants with low metabolic risk was further studied. We defined low-risk patients as those with triglycerides <150 mg/dl, glucose <100 mg/dl, BMI <27 kg/m2, systolic blood pressure <130 mm Hg, diastolic blood pressure <85 mm Hg, and with no history of receiving medications for blood pressure or diabetes. Among the low-risk population (n = 36,003, mean ± SD serum uric acid level 5.48 ± 1.46 mg/dl), the HR for hyperuricemia (serum uric acid level >7 mg/dl) in all-cause mortality (762 events) was 1.24 (95% CI 1.03–1.48, P < 0.02). Although the hyperuricemic risk of CVD mortality (78 events) was 1.48 (95% CI 0.86–2.54, P = 0.16) times that of the reference group, it was not statistically significant.
Hyperuricemia is strongly associated with sex (Table 1), and the risk of hyperuricemia in men was 5.19 times (odds ratio) that of women (95% CI 5.01–5.37). However, the HRs of hyperuricemia for all-cause mortality and total CVD were greater in magnitude in women than in men when a unified cut-off of serum uric acid level 7 mg/dl was used (Figure 2). The multivariate adjusted HRs (Table 2) of hyperuricemia (serum uric acid level >7 mg/dl) for most mortality end points were also larger in women than in men. Nevertheless, when we tested the difference between sexes, we found that only the HR for all-cause mortality was statistically different (P = 0.04 between men and women), but the HR for CVD mortality was at borderline (P = 0.07). Similarly, the HR estimates were higher in female patients with hypertension than in male patients with hypertension for both all-cause and CVD mortality, but they were only marginally significant for the latter (P = 0.08 for all-cause mortality and P = 0.05 for CVD mortality).
Specifically, the mortality rate of total CVD in women was lower than that of men in every serum uric acid level quartile (P < 0.001 for all, P = 0.16 for quartile 4) (Figure 3A). In women, the higher the serum uric acid level quartile, the higher the mortality rate was noted in total CVD mortality. Although this uprising trend was less apparent in men, a clear dose-response curve was seen for both men and women. The HRs for CVD mortality comparing the third and fourth serum uric acid level quartiles with the first were all statistically significant in both men and women (Figure 3C).
To further compare the serum uric acid level and CVD mortality relationship between sexes by the serum uric acid level increments of ≤5 mg/dl, 5.1–7 mg/dl, 7.1–9 mg/dl, and >9 mg/dl, the sex-specific pattern was similar to that according to serum uric acid level quartile, except that CVD mortality rates were closer between sex at 3 of 4 serum uric acid levels and there was a crossover at serum uric acid levels ∼7 mg/dl (Figure 3B). Dose-response curves were seen in both sex groups (Figure 3D). Confidence intervals for all HR estimates overlapped substantially between men and women (P > 0.1 for all).
In this comprehensive Taiwanese cohort, hyperuricemia (serum uric acid level >7 mg/dl) was an independent risk factor for all-cause and cardiovascular mortality not only in all patients, but also in patients with hypertension and diabetes and potentially in low-risk subgroups. The adjusted risks increased 8–13% with every mg/dl increase of serum uric acid level for these 2 major end points and various CVDs, including ischemic stroke, CHF, and hypertensive disease. The RR was greater for women than for men, statistically significant for all-cause mortality, and borderline significant for CVD morality.
According to the Framingham Study (1), patients with gout are 2 times more likely than patients without gout to have coronary heart disease. The Health Professionals Follow-up Study reported that patients with a history of gout have a higher CVD mortality risk than patients with coronary heart disease history (3), and patients with gout are more likely to have metabolic syndromes (27). The Multiple Risk Factor Intervention Trial suggests that other than the proinflammatory effect of uric acid, the slow but persistent inflammation of gout expedites atherosclerosis and thrombogenesis, which pronounces a higher risk of gout than hyperuricemia in the development of CVD (2, 28). Additionally, although positive associations have been previously reported between serum uric acid levels and CVD risk, some of these fail to show statistical significance after adjusting for additional confounders (2, 14, 29). Lack of statistical significance can be attributable to small sample size and insufficient power (2). In the current study with a large sample size, the association between hyperuricemia and total cardiovascular mortality was statistically significant after multivariate adjustment.
Previous studies have shown that an elevated serum uric acid level is an independent predictor of coronary heart disease mortality (16, 18). However, hyperuricemia in this study did not significantly contribute to coronary disease–related mortality. Instead, hyperuricemia was found to affect ischemic stroke, as in the Rotterdam Study and other studies (17, 18). This could be due to the relatively lower risk of coronary heart disease and higher stroke mortality rate in Taiwan (17). In addition, we also demonstrated the significant associations between hyperuricemia and mortality from CHF and hypertensive disease that, to our knowledge, have not been previously reported.
In the current study, we demonstrated serum uric acid level as a predictor of total death in patients with hypertension and diabetes, which was recently reported in the Cleveland cohort (30). Our data provided an angle to examine the connection between hyperuricemia and metabolic syndrome. According to reports by NHANES-III in the US (31) and the Nutrition and Health Survey in Taiwan Elementary School Children from 2001 to 2002 in Taiwan (32), the prevalence of metabolic syndrome increases substantially with increasing levels of serum uric acid and vice versa. Abdominal obesity, particularly in women, contributes significantly to hyperuricemia in elderly individuals who also have metabolic syndrome (33). Persistent hyperuricemia (serum uric acid level >6.6 mg/dl) in postmenopausal women in the Kinmen Study in Taiwan is associated with subsequent development of diabetes (12). Shared risk factors between serum uric acid level elevation and metabolic syndrome are worth noting (31). In addition to providing a unique facet of hyperuricemia in metabolic syndrome, we demonstrated a significant risk of elevation of hyperuricemia in all-cause mortality in a subgroup with low metabolic risk. This may further support the potential risk of hyperuricemia in CVD development.
In the current study, hyperuricemia was less prevalent in women than in men, and the overall mortality rate of women was lower than that of men, particularly at low serum uric acid levels (Figure 3). Therefore, the RR of hyperuricemia was higher in women and more statistically significant than that in men using the same cut-off point (serum uric acid level 7 mg/dl). Elevated serum uric acid levels in women were associated with a higher cardiovascular RR than that in men according to an NHANES-I study (15, 16). Our analysis ascribed this phenomenon to a difference in the serum uric acid level and CVD relationship between men and women. Estrogen plays a cardioprotective role in women who have an overall lower cardiovascular risk than men (14, 34). Among women in the NHANES-I study, the serum uric acid level and CVD association increased with a growing number of cardiovascular risk factors. In contrast, findings among men showed that there was no association between serum uric acid level and cardiovascular mortality at high-risk status (21). Hyperuricemia in women could possibly be a hallmark of escape from estrogen protection.
Uric acid has been reported to be an antioxidant that may prevent stress-induced cell transformation and oxidant-induced cardiac and renal toxicity (35). There is relatively more evidence to suggest that uric acid is a pro-oxidant that can increase oxygen radicals in circulation, which may in turn promote the lipid oxidation, leading to vascular endothelial dysfunction, inflammation, nitric oxide production impairment, atherosclerosis, and thrombogenesis (35). In animal models, the proinflammatory and proliferative effect of soluble uric acid influences vascular smooth muscle cells (36). Hyperuricemia activates the renin–angiotensin system, which increases sodium resorption (35). Hypertension develops thereafter in association with intrarenal vascular disease. Losartan, an angiotensin receptor II blocker, reportedly reduces serum uric acid levels in addition to its antihypertensive effects (37). Hypertension and vascular change can be alleviated by treatment with allopurinol (38), which is a xanthine oxidase inhibitor.
This study has several limitations. At the time of data collection there may have been some recall biases regarding history of hypertension, diabetes, and frequency of cigarette and alcohol consumption. Hyperuricemia may be associated with chronic renal dysfunction (39); however, information for serum creatinine and glomerular filtration rate was not provided by this original patient data set. The effect of these factors as confounders to the independent predictability of serum uric acid level on CVD death was considered minimal, according to previous population-based studies (2, 7, 16). Information regarding the repeated serum uric acid levels was not available in the original data set. A certain degree of misclassification among cardiovascular events was also possible since ICD-9 codes were obtained from death certificates (26). However, these limitations are presumably nondifferential biases (25) that can be overcome by a large sample size and a reasonably long observation period.
In conclusion, our study shows that hyperuricemia acts as an independent risk factor for all-cause and CVD mortality not only in all patients representing the general public, but also in patients with hypertension and diabetes and potentially low-risk subgroups. Our study also demonstrates a dose-response effect of increasing serum uric acid levels on mortality, particularly cardiovascular mortality.
Dr. Pan 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. Jiunn-Horng Chen, Pan.
Acquisition of data. Hsin-Jen Chen, Yeh, Pan.
Analysis and interpretation of data. Jiunn-Horng Chen, Chuang, Pan.