Hyperuricemia and coronary heart disease: A systematic review and meta-analysis

Authors


Abstract

Objective

The role of serum uric acid as an independent risk factor for cardiovascular disease remains unclear, although hyperuricemia is associated with cardiovascular disease such as coronary heart disease (CHD), stroke, and hypertension.

Methods

A systematic review and meta-analysis using a random-effects model was conducted to determine the risk of CHD associated with hyperuricemia in adults. Studies of hyperuricemia and CHD were identified by searching major electronic databases using the medical subject headings and keywords without language restriction (through February 2009). Only prospective cohort studies were included if they had data on CHD incidences or mortalities related to serum uric acid levels in adults.

Results

Twenty-six eligible studies of 402,997 adults were identified. Hyperuricemia was associated with an increased risk of CHD incidence (unadjusted risk ratio [RR] 1.34, 95% confidence interval [95% CI] 1.19–1.49) and mortality (unadjusted RR 1.46, 95% CI 1.20–1.73). When adjusted for potential confounding, the pooled RR was 1.09 (95% CI 1.03–1.16) for CHD incidence and 1.16 (95% CI 1.01–1.30) for CHD mortality. For each increase of 1 mg/dl in uric acid level, the pooled multivariate RR for CHD mortality was 1.12 (95% CI 1.05–1.19). Subgroup analyses showed no significant association between hyperuricemia and CHD incidence/mortality in men, but an increased risk for CHD mortality in women (RR 1.67, 95% CI 1.30–2.04).

Conclusion

Hyperuricemia may marginally increase the risk of CHD events, independently of traditional CHD risk factors. A more pronounced increased risk for CHD mortality in women should be investigated in future research.

INTRODUCTION

In humans, uric acid is the end product of purine metabolism due to the nonfunctioning uricase gene leading to elevated serum uric acid levels (1). Although the mechanism and biologic reason for this mutation is still unknown, it has been hypothesized that the loss of uricase activity has evolutionary advantages related to protection from oxidative damage and the prolonged lifespan owing to the antioxidant properties of uric acid (1–3). However, this hypothesis is in conflict with many epidemiologic studies showing that hyperuricemia was frequently noted in patients either with cardiovascular disease or at a high risk of cardiovascular disease such as hypertension, coronary heart disease (CHD), peripheral vascular disease, heart failure, metabolic syndrome, and stroke (4–10). A recent meta-analysis of prospective observational studies (11) for hyperuricemia and risk of stroke demonstrated significantly increased risk for both stroke incidence (risk ratio [RR] 1.47, 95% confidence interval [95% CI] 1.19–1.76] and mortality (RR 1.26, 95% CI 1.12–1.39) based on studies that adjusted for traditional stroke risk factors such as age, sex, hypertension, hypercholesterolemia, and serum glucose. Several pathophysiologic mechanisms have been postulated, including endothelial dysfunction, oxidative metabolism, and platelet adhesiveness and aggregation (12–14). However, the role of hyperuricemia as an independent risk factor for CHD remains controversial (15–20). It is perhaps related to the complex association between hyperuricemia and known CHD risk factors, resulting in methodologic difficulties in some observational studies, particularly with a limited sample size, that elucidate its direct effect on CHD (21). The objective of this study was to systematically review published reports of prospective cohort studies to assess the risk of CHD incidence and mortality in hyperuricemia.

MATERIALS AND METHODS

Literature search

We searched 3 major electronic databases, Medline (through February 2009), EMBase (1980 to February 2009), and the Cochrane Library (through February 2009), using the following medical subject heading terms and keywords: [uric acid OR hyperuricemia OR urate] AND [coronary disease OR myocardial infarction OR coronary artery disease OR angina pectoris OR unstable angina OR cardiovascular disease OR coronary heart disease] (see Appendix A). We also searched bibliographies of identified reports and review articles for additional references. We followed the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) study guidelines (22).

Study eligibility

We only considered 1) prospective cohort studies of adult patients with longer than 1 year of followup and with a sample size of at least 100 subjects, and 2) an inception cohort free of CHD. No geographic or language restrictions were applied. Studies reporting interventional and secondary prevention trials were excluded.

Study selection

Two authors (SYK and KMK) independently screened each of the potential titles, abstracts, and/or full texts to determine inclusion. Areas of disagreement or uncertainty were resolved by consensus. When multiple articles were published from a single study, we selected the reports that contained the most complete and relevant data on the association between hyperuricemia and CHD.

Data abstraction and quality assessment

All of the data were independently abstracted in duplicate by 2 authors (SYK and KMK) using a data abstraction form to retrieve information on study characteristics, participant information, cutoff levels for hyperuricemia, CHD outcome, analyses, and adjustment. Discrepancies were resolved by consensus. When necessary, we attempted to contact the original authors for additional information, but we were unable to obtain unpublished data. We used the Newcastle-Ottawa Scale to assess the quality of the studies (23). A quality score was calculated on the basis of 3 major components of cohort studies: selection of study groups (0–4 points), comparability of study groups (0–2 points), and ascertainment of the outcome of interest (0–3 points). A higher score represents better methodologic quality (Table 1).

Table 1. Details of the cohort studies on incident coronary heart disease (CHD)*
Study name, year (ref.)Study population (% men)Age, years†Followup, yearsHyperuricemia definition, mg/dlTotal no. of CHD eventsBasis of outcome definitionVariables controlledQuality assessment score‡
  • *

    GRIPS = Gottingen Risk Incidence and Prevalence Study; M = men; EKGs = electrocardiograms; CT = computed tomography; NHANES I = First National Health and Nutrition Examination Survey; C = combined; W = women; BP = blood pressure; CVD = cardiovascular disease; MONICA = Monitoring Trends and Determinants on Cardiovascular Diseases; DM = diabetes mellitus; HDL = high-density lipoprotein; BMI = body mass index; LVH = left ventricular hypertrophy; ARIC = Atherosclerosis Risk in Communities; LDL = low-density lipoprotein; MI = myocardial infarction; FEV1 = forced expiratory volume in 1 second; ICD-9 = International Classification of Diseases, Ninth Revision; MRFIT = Multiple Risk Factor Intervention Trial; CABG = coronary artery bypass graft; KORA = Cooperative Health Research in the Region of Augsburg; CRP = C-reactive protein.

  • b

    Values are the range, mean ± SD, or mean.

  • c

    Based on Newcastle-Ottawa Scale (23).

  • §

    Nested case–control design.

GRIPS, 1994 (43)5,728 men free of atherosclerotic diseases in Germany40–605M: 6.5M: 107Clinical symptoms, EKGs, cardiac enzymes, angiography, and CTNone4/0/3
NHANES I, 1995 (30)5,421 (46) adults free of CHD at baseline in the US25–7413.5C: 7M: 403; W: 286Hospital records and death certificatesAge, race, cholesterol, diastolic BP, smoking, alcohol, education level, and use of antihypertensive and diuretic medicines4/2/3
Honolulu Heart, 1995 (31)2,710 Japanese-American men free of CVD in the US55–6423M: 6.8M: 352Autopsy reports and/or medical records such as EKGs and cardiac enzymesAge4/1/3
MONICA, 1999 (41)960 middle-aged men free of heart attack and DM in Germany45–648M: 6.3M: 55Medical records such as clinical symptoms, EKGs, cardiac enzymes, and autopsy reportsAge, alcohol, cholesterol:HDL ratio, hypertension, smoking, BMI, education, and use of diuretics4/2/3
Framingham study, 1999 (15)6,763 (45.5) adults free of CVD in the US47 ± 1517.4M: 6.8; W: 5.3M: 394; W: 223Medical records such as clinical symptoms, EKGs, and cardiac enzymesAge, BMI, systolic BP, use of antihypertensive and diuretic meds, DM, cholesterol, alcohol, smoking, LVH, and menopause status4/2/3
ARIC, 2000 (36)13,504 (43.7) healthy middle-aged subjects in the US45–648M: 7.6; W: 6.3M: 264; W: 128Medical records such as clinical symptoms, EKGs, and cardiac enzymes, and data on death certificatesAge, race, ARIC center, smoking, LDL, systolic BP, BMI, HDL, DM, waist:hip ratio, protein, triglycerides, alcohol, and antihypertensive meds4/2/3
Gubbio study, 2001 (44)2,469 (45.2) adults free of CVD at entry in Italy35–746C: 7.3M: 68; W: 41Paper/phone questionnaires, EKGs, and medical recordsAge, sex, systolic BP, cholesterol, glucose, smoking, and BMI4/2/3
Chin-Shan, 2005 (50)3,602 adults free of CVD in Taiwan≥358.5M: 7.7; W: 6.686Death certificates and hospital recordsAge, systolic BP, BMI, DM, cholesterol, smoking, and alcohol4/2/3
Reykjavik study, 2005 (47)§6,042 (70.3) adults without a history of MI in Iceland56 ± 917.5M: 5.7; W: 4.72,080Questionnaires, EKGs, and medical recordsAge, smoking, systolic BP, cholesterol, BMI, triglycerides, FEV1, and DM4/2/3
Rotterdam study, 2006 (39)4,385 (35.4) adults free of CHD in The Netherlands≥558.4M: 6.4; W: 5.4; C: 6.5515ICD-9 codes on medical recordsAge, sex, systolic BP, cholesterol, HDL, DM, smoking, diuretic use, and waist:hip ratio4/2/3
MRFIT, 2006 (32)12,866 men free of CVD at baseline in the US46 ± 66.5M: 7.0M: 1,108Review of medical records such as EKGs and CABG surgeryAge, BP, cholesterol, serum creatinine, DM, smoking, BMI, family history of acute MI, alcohol, aspirin, and diuretic use4/2/3
Atomic bomb study, 2007 (48)2,024 (38.3) atomic bomb survivors free of CHD at baseline in JapanM: 62 ± 9.9; W: 63.2 ± 8.48C: 7.049Self-reports, EKGs, and medical recordsAge, sex, smoking, alcohol, glucose, and fatty liver4/2/3
MONICA/KORA, 2008 (42)3,424 middle-aged men free of heart attack in Germany45–7411.7M: 6.6M: 297The population-based data from MONICA/KORA Augsburg coronary event registry and death certificatesAge, smoking, alcohol, physical activity, hypertension, BMI, DM, CRP dyslipidemia, creatinine, and diuretic use4/2/3

Statistical analysis

Some studies included in our meta-analysis used the Système International d'Unités (μmoles per liter) to report levels of serum uric acid. We therefore converted those to the conventional units (milligram per deciliter), using a conversion rate of 16.81 (1 mg/dl = 59.48 μmoles/liter) (24). The category nearest to 6.8 mg/dl was considered as the hyperuricemia group for both sexes (25).

Pooled estimates of both unadjusted and multivariate RRs were calculated using the DerSimonian and Laird random-effects model (26, 27) for CHD incidence and mortality. This statistical technique weights individual studies by sample size and variance (both within- and between-study variance) and yields a pooled point estimate and a 95% CI. The DerSimonian and Laird technique was considered an appropriate pooling technique because of the relative heterogeneity of the source population in each study. We evaluated the presence of heterogeneity across trials by using the I2 statistic, which quantifies the percentage of variability that can be attributed to between-study differences (28). A stratified analysis by sex was conducted to evaluate sex-related heterogeneity in both unadjusted and multivariate RRs of CHD incidence and mortality. To investigate the impact of study characteristics such as sex, publication year, ethnicity, study location, and cutoff level defining hyperuricemia on the study estimates of RR, we performed a multivariate meta–regression analysis on the log-transformed scale of RR. To assess the potential for publication bias, we performed the Begg test and the Egger test and constructed funnel plots to visualize possible asymmetry (29). All of the statistical analyses were done in Stata, version 10 (StataCorp, College Station, TX).

RESULTS

Description of the studies

The electronic database search identified 6,557 references. Bibliographic lists of relevant review articles were manually searched and elicited 21 additional references. The title and abstract review of these references resulted in 310 original articles. A total of 26 prospective cohort studies representing data from 402,997 participants were finally included in this review. Figure 1 shows the study flow.

Figure 1.

Selection of studies included in the analysis. CHD = coronary heart disease.

The characteristics of the included studies and their participants are shown in Tables 1 and 2. A total of 26 studies (13 for CHD incidence and 13 for CHD mortality) were included. Nine studies (15, 20, 30–36) were carried out in the US, 11 (37–47) in Europe, and 6 (48–53) in Asia. All of the studies except one (37) were written in English. The lengths of followup in the included studies varied between 5 and 24.9 years. The definition of hyperuricemia ranged from 5.6 to 7.7 mg/dl in men and from 4.7 to 7.0 mg/dl in women. In most studies, CHD events were defined using the medical records and/or International Classification of Diseases codes from the hospital records or death certificates. Results from 9 studies for CHD incidence and 8 studies for CHD mortality were fully adjusted for traditional CHD risk factors such as age, sex, hypertension, diabetes mellitus, smoking, and hypercholesterolemia. The majority (85%) of the included studies were of high quality.

Table 2. Details of the cohort studies on coronary heart disease (CHD) deaths*
Study name, year (ref.)Study population (% men)Age, years†Followup, yearsHyperuricemia definition, mg/dlTotal no. of CHD eventsBasis of outcome definitionVariables controlledQuality assessment score‡
  • *

    CHA = Chicago Heart Association; M = men; W = women; ICD-8 = International Classification of Diseases, Eighth Revision; NHANES I = National Health and Nutrition Examination Survey I; ICD-9 = International Classification of Diseases, Ninth Revision; BMI = body mass index; DM = diabetes mellitus; BP = blood pressure; KMIC = Korea Medical Insurance Corporation; ICD-10 = International Classification of Diseases, Tenth Revision; ICD-7 = International Classification of Diseases, Seventh Revision; LVH = left ventricular hypertrophy; EKG = electrocardiogram; MRFIT = Multiple Risk Factor Intervention Trial; CVD = cardiovascular disease; MI = myocardial infarction; VHMPP = Vorarlberg Health Monitoring and Promotion Program; GGT = gamma glutamyl transferase.

  • b

    Values are the range, range (mean), mean ± SD, or mean.

  • c

    Based on Newcastle-Ottawa Scale (23).

CHA, 1979 (35)7,804 (54) white subjects free of CHD at baseline in the CHA Detection project in the US45–645M: 7.0; W: 6.0M: 48; W: 7ICD-8 codes on death certificates; autopsy and hospital reports if availableNone4/0/3
CHA for women, 1989 (34)4,825 white women in the CHA Detection project in the US45–6411.5Per 1 mg/dlW: 23ICD-8 codes on death certificates; autopsy and hospital reports if availableAge, weight, smoking, diastolic BP, cholesterol, and antihypertensive meds4/2/3
NHANES I, 2000 (20)5,926 (45.6) noninstitutionalized adults in the US25–74 (48.1)16.4Per 1 mg/dl increase: M: 7.0; W: 5.6M: 222; W: 172ICD-9 codes on death certificates; hospital records if availableAge, cholesterol, race, BMI, smoking, alcohol, hypertension, DM, and sex4/2/3
Japanese male workers study, 2000 (53)49,413 Japanese male railroad workers25–605.4M: 6.5M: 85ICD-9 codes on health and pension recordsAge4/1/3
Belgian study, 2001 (37)9,701 (53.9) adults in Belgium25–7410M: 7.0; W: 5.4M: 150; W: 51ICD-9 codes on hospital recordsM: age, diastolic BP, education level, smoking, and alcohol; W: age, cholesterol, systolic BP, smoking, BMI, alcohol, and DM4/2/3
KMIC, 2004 (52)22,698 Korean men enrolled in the National Health Insurance Corporation30–779M: 7.0M: 99ICD-9 and ICD-10 codes from hospitalization records and death certificatesAge, hypertension, DM, cholesterol, and smoking4/2/3
Atomic bomb study, 2005 (51)10,615 (36.4) Japanese atomic bomb survivors49 ± 14.824.9M: 7.0; W: 6.0M: 177; W: 250ICD-7 through ICD-10 codes on death certificatesAge, BMI, smoking, alcohol, systolic BP, cholesterol, hypertension, DM, kidney disease, malignant tumor, and estimated radiation dose from the atomic bombs4/2/3
Israeli male study, 2005 (40)9,125 men free of CHD at baseline in Israel49 ± 723M: 5.6M: 830ICD-9 codes on death certificates and hospital recordsAge, BMI, systolic BP, DM, cholesterol, smoking, and LVH on EKG4/2/3
Greek study, 2005 (38)1,198 (42) adults in rural Greece≥2514Per 1 mg/dl increaseM: 34; W: 33ICD-9 codes on death certificatesAge, body weight, smoking, alcohol, glucose, systolic BP, cholesterol, village, triglycerides, and educational level4/2/3
MRFIT, 2008 (33)9,105 men free of CVD at baseline in the US41–6317M: 7.0M: 833ICD-9 and ICD-10 codes on death certificatesAge, systolic/diastolic BP, cholesterol, BMI, triglycerides, serum creatinine, glucose, alcohol, smoking, family history of acute MI, aspirin and diuretic use4/2/3
VHMPP for men, 2008 (46)83,683 Austrian men41.6 ± 14.612.4M: 6.8M: 844ICD-9 and ICD-10 codes on death certificates; autopsy records if availableAge, BMI, cholesterol, systolic/diastolic BP, triglycerides, GGT, smoking, and year of examinations4/2/3
VHMPP for women, 2008 (45)28,613 elderly Austrian women62.3 ± 8.821W: 5.4W: 518ICD-9 and ICD-10 codes on death certificates; autopsy records if availableAge, BMI, cholesterol, systolic/diastolic BP, triglycerides, GGT, smoking, glucose, occupational status, and year of examinations4/2/3
Chinese cohort study, 2009 (49)90,393 (46.3) adults in Taiwan51.5 ± 11.58.2Per 1 mg/dl increase: M/W: 7286ICD-9 codes on death certificatesAge, sex, BMI, cholesterol, DM, triglycerides, hypertension, smoking, and alcohol4/2/3

Hyperuricemia and CHD incidence

The pooled estimate of unadjusted RRs for CHD incidence based on 13 studies (15, 30, 31, 36, 39, 41–44, 47–50) was 1.34 (95% CI 1.19–1.49; comparing hyperuricemic with normouricemic patients). A significant heterogeneity between studies was noted (I2 = 64.4%, P = 0.001). Based on 9 studies (15, 32, 36, 39, 41, 42, 47, 48, 50) that fully adjusted for traditional risk factors of CHD, the overall risk of incident CHD related to hyperuricemia had a pooled multivariate RR of 1.09 (95% CI 1.03–1.16), with mild heterogeneity (I2 = 17.6%, P = 0.27).

The pooled multivariate RR of incident CHD was 1.04 (n = 7 studies; 95% CI 0.90–1.17) (15, 32, 36, 41, 42, 47, 50) for men and 1.07 (n = 4 studies; 95% CI 0.82–1.32) (15, 36, 47, 50) for women. There was moderate heterogeneity between studies with respect to outcomes among men (I2 = 42%, P = 0.11), but no heterogeneity between studies was noted among women (I2 = 0.0%, P = 0.76). The forest plot of multivariate RRs and 95% CIs for CHD incidence is shown in Figure 2A.

Figure 2.

Random-effects analysis of multivariate risks of coronary heart disease (CHD) associated with hyperuricemia. A, CHD incidence, B, CHD mortality. ES = effect size; 95% CI = 95% confidence interval; MONICA = Monitoring Trends and Determinants on Cardiovascular Diseases; ARIC = Atherosclerosis Risk in Communities; MRFIT = Multiple Risk Factor Intervention Trial; KORA = Cooperative Health Research in the Region of Augsburg; KMIC = Korea Medical Insurance Corporation; VHMPP-M = Vorarlberg Health Monitoring and Promotion Program for men; VHMPP-W = Vorarlberg Health Monitoring and Promotion Program for women.

Hyperuricemia and CHD mortality

The pooled unadjusted RR for CHD mortality based on 9 studies (20, 33, 35, 40, 45, 46, 51–53) was 1.46 (95% CI 1.20–1.73). A significant heterogeneity between studies was noted (I2 = 77.6%, P = 0.001). The RR based on 8 studies (33, 37, 40, 45, 46, 49, 51, 52) that fully adjusted for traditional CHD risk factors was 1.16 (95% CI 1.01–1.30), with mild heterogeneity (I2 = 29.4%, P = 0.17).

The pooled multivariate RR for CHD mortality was 1.09 (n = 7 studies; 95% CI 0.98–1.19) (33, 37, 40, 46, 49, 51, 52) among men and 1.67 (n = 4 studies; 95% CI 1.30–2.04) (37, 45, 49, 51) among women. There was no statistically significant heterogeneity between studies with respect to outcomes (I2 = 0.0% for both sexes). The forest plot of multivariate RRs and 95% CIs for CHD mortality is shown in Figure 2B.

For each increase of 1 mg/dl in uric acid level, the overall pooled multivariate RR for CHD mortality was 1.12 (n = 4 studies; 95% CI 1.05–1.19) (20, 34, 38, 49) (Figure 3). The sex-specific relative risks for each increase of 1 mg/dl in serum uric acid level were similar, but no longer statistically significant.

Figure 3.

Random-effects analysis of multivariate risks of coronary heart disease (CHD) mortality associated with an increase of 1 mg/dl in serum uric acid level. ES = effect size; 95% CI = 95% confidence interval; NHANES I = National Health and Nutrition Examination Survey I; CHA-W = Chicago Heart Association for women.

Publication bias assessment

Some evidence of publication bias for both CHD incidence and mortality was noted in the funnel plots (Figure 4), Begg's tests (P = 0.06 and 0.08, respectively), and Egger's tests (P = 0.03 and 0.12, respectively).

Figure 4.

Begg's funnel plots for publication bias in studies for coronary heart disease (CHD) incidence and mortality. CI = confidence interval; s.e. = standard error; RR = risk ratio.

Sensitivity analyses

A multivariate meta–regression analysis to investigate the impact of several covariates on the study estimates of RR found that none of the covariates, sex, year of publication, ethnicity (Asian versus non-Asian), study location, and cutoff levels defining hyperuricemia, modified the association between hyperuricemia and CHD risk. On the other hand, earlier publication year (β 0.03, P = 0.002) and female sex (β 0.55, P < 0.001) were significantly associated with a greater risk estimate for CHD mortality.

DISCUSSION

This systematic review and meta-analysis of prospective cohort studies shows a significant, modest association between hyperuricemia and CHD events, independent of traditional CHD risk factors. The overall risk of CHD death increased by 12% for each increase of 1 mg/dl of uric acid level. In the subgroup analyses, hyperuricemia appeared to significantly increase the risk of CHD deaths in women (approximately 70%), but not in men. Although this sex difference requires further research, our results support previous findings of a stronger association between hyperuricemia and cardiovascular disease in women (20, 30, 54, 55).

Our results are consistent with a previous meta-analysis of 16 observational studies that examined the association between hyperuricemia and CHD (47). It showed a pooled RR of 1.13 (95% CI 1.07–1.20) with significant heterogeneity (P = 0.02) (47). In their subgroup analyses, the RR for CHD was 1.12 (95% CI 1.05–1.19) in men and 1.22 (95% CI 1.05–1.40) in women. Eight of 16 studies used in the previous meta-analysis were not included in our review. Two studies (47, 56) were not prospective cohort studies and 3 (43, 57, 58) did not present the outcomes of our interest in their texts. Unfortunately, we could not obtain the relevant, unpublished data from the authors of the original studies. Newer studies (31, 37, 40) were included in our meta-analysis, in place of 3 studies (59–61) that used the same cohorts.

Recent studies of losartan and atorvastatin showed that uric acid reduction contributes to attenuation of cardiovascular risk (62, 63). Fenofibrate has also shown a uricosuric effect in healthy subjects and subjects with diabetes mellitus (64, 65). These medications are useful for the management of patients with metabolic syndrome, identified as a multiplex risk factor for cardiovascular disease by the National Cholesterol Education Program's Adult Treatment Panel III report (66). In a small randomized clinical trial (67), allopurinol treatment in newly diagnosed, hypertensive adolescents was associated with significant reductions in casual and 24-hour ambulatory blood pressure compared with placebo. Interestingly, a recent cohort study of hyperuricemic patients enrolled in Veterans Affairs Medical Centers in the Pacific Northwest reported that the use of allopurinol was associated with a 23% lower all-cause mortality rate (68). Several observational studies reported that gout was associated with multiple risk factors for cardiovascular disease and with cardiovascular morbidities and mortalities (33, 34, 69, 70). Whether gout directly or indirectly increases the risk of cardiovascular disease through hyperuricemia remains uncertain, but current data suggest more aggressive cardiovascular risk management in patients with gout (9). Nevertheless, larger clinical trials with longer followup periods are still needed to determine the safety and efficacy of urate-lowering therapy such as allopurinol in cardiovascular disease.

Several potential limitations to this study are inherent to meta-analyses. First, even with our comprehensive search strategy and lack of language restriction, statistical assessment and a funnel plot examination did suggest the possibility of publication bias. Studies with null results are generally less likely to be published and, therefore, more likely to be missed in a database search. However, the majority of the included studies in our meta-analysis reported null results. Our study relied exclusively on published data. There were different definitions of hyperuricemia across the studies; therefore, we chose the category nearest to 6.8 mg/dl in each study for the hyperuricemia group. For men, the cutoff level to define hyperuricemia was between 6.5 and 7.0 mg/dl in 55% and 90% of the studies for CHD incidence and mortality, respectively. Most of the studies used a lower cutoff level to define hyperuricemia for women. However, different cutoff levels for hyperuricemia did not modify the study estimates of CHD risk based on our meta–regression analysis. Statistical methods and the degree of adjustment differed slightly in each study. We utilized the best adjusted RR per individual study and performed separate analyses for unadjusted and multivariate RRs. Second, unmeasured confounding is also a common problem in observational studies, including prospective cohorts. In our review, the use of concomitant medications such as diuretics and the presence of medical comorbidities were not always adjusted for in the included studies. Third, for the CHD outcome data, there is a possibility of misclassification bias because many of the included studies used death certificates or diagnostic codes to define their outcomes.

Our study has several important strengths. We selected only large prospective studies with inception cohorts free of disease, which helped increase the precision of estimates while minimizing heterogeneity. Assessment of the quality of individual studies is a necessary component for a systematic review of both randomized and observational studies. There has not been a consensus on which way is the best to measure the quality of observational studies. Indeed, a recent review identified more than 80 quality assessment tools and noted the lack of a single best generic tool for observational studies (71). We chose to use the Newcastle-Ottawa Scale (23) for quality assessment because this tool appropriately evaluates the 3 most important domains of prospective cohort studies: selection of study participants, measurement of exposures and outcomes, and control of confounding. We performed sex-specific subgroup analyses of the studies fully adjusting for traditional CHD risk factors. Multivariate meta–regression analysis further examined several potential sources of heterogeneity between the studies such as sex, ethnicity, study location, year of publication, and cutoff level for hyperuricemia.

In conclusion, there was a modestly increased risk for CHD associated with hyperuricemia in our meta-analysis. A more pronounced increased risk for CHD mortality in women should be confirmed with future research. It would be particularly important to design further large, long-term studies that determine the effect of urate-lowering therapy on cardiovascular disease.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Seo Young Kim 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 conception and design. Seo Young Kim, Guevara, Choi, Albert.

Acquisition of data. Seo Young Kim, Kyoung Mi Kim.

Analysis and interpretation of data. Seo Young Kim, Guevara, Choi, Heitjan.

APPENDIX A

SEARCH STRATEGY

Medical subject headings term search

  • 1Uric Acid
  • 2Hyperuricemia
  • 3Coronary disease
  • 4Myocardial infarction
  • 5Coronary artery disease
  • 6Angina pectoris
  • 7Angina, unstable
  • 8Cardiovascular Diseases

Direct keyword search

  • 9urate
  • 10hyperuric$
  • 11coronary heart disease
  • 12“1 OR 2 OR 9 OR 10”
  • 13“3 OR 4 OR 5 OR 6 OR 7 OR 8 OR 11”
  • 14“12 AND 13”

Ancillary