The Tromsø Study is a series of population-based, prospective surveys of inhabitants of the municipality of Tromsø, Norway [16]. In 1994/95, 27.158 subjects were screened (77% of eligible subjects). All participants at the age of 55-75 years and 5-10% of the other age groups ≥ 25 years were invited to a second visit including a more comprehensive examination 4-6 weeks later. Of the 9057 individuals who were invited 6862 participated (attendance rate 75%). Persons with known previous MI (n = 402), ischemic stroke (n = 190) or diabetes (n = 308), defined as self-reported diabetes, use of antidiabetic medication, HbA1c > 6.5% or non-fasting plasma glucose ≥10.0 mmol/L, were excluded. Data on uric acid was available in 5700 subjects (Figure 1). The Tromsø Study was conducted by the University of Tromsø in cooperation with The National Health Screening Service. The Regional Committee for Medical Research Ethics approved the study, and all participants gave their written consent.

All measurements and information on risk factors were obtained from baseline data of the 4^th Tromsø Study in 1994 /95. Information about presence of diabetes, smoking habits and physical activity was obtained from a self-administered questionnaire. Blood pressure was recorded in triplet (Dinamap) after 5-min seating, the mean of the second and third measurement was used. Hypertension was defined as systolic blood pressure (SBP) ≥ 140 mmHg and/or diastolic pressure (DBP) ≥ 90 mmHg and/or current use of antihypertensive medication. Physical activity was classified as active (> 1 hour physical activity/week) or inactive (all others). Smoking habits were classified as non-smokers or current smokers. Serum HDL-cholesterol was measured after precipitation of lower-density lipoprotein with heparin and manganese chloride. Serum uric acid was measured by photometry with COBAS® instruments (Roche diagnostics, Switzerland) using an enzymatic colorimetric test, the uricase/ PAP method. Reference values were 140-340 μmol/L (2.4-5.7 mg/100 mL) for females and 200-415 μmol/L (3.4-7.0 mg/100 mL) for males. Creatinine was analyzed by a modified Jaffe reaction, but since creatinine-based estimation of GFR is better validated for enzymatic creatinine measurements, 111 plasma samples from the 1994/95 survey were thawed and reanalysed with an enzymatic method (Modular P/Roche). Values were fitted to a linear regression model, and recalibrated creatinine values were calculated for all participants. Estimated GFR was calculated according to the CKD-EPI formula [17]: eGFR = 141 × min(S_Cr/k,1)^a × max(S_Cr/k,1)^-1.209 × 0.993^age × ([1.018 if female] and × [1.159 if black]) where S_Cr is serum creatinine (mg/dL), k is 0.7 for females and -0.411 for males, min indicates the minimum of S_Cr/k and max indicates the maxiumum of S_Cr/k). Albuminuria was reported as albumin/creatinine ratio (ACR, mg/mmol, urinary albumin and creatinine analyses; kits from ABX Diagnostics, Montpellier, France). For each subject, ACR was measured in fresh samples of each of 3 separate urine specimens and the mean of all 3 was used in the analyses.

The endpoints were death from any cause, first ever non-fatal or fatal myocardial infarction or ischemic stroke. Adjudication of hospitalized and out-of-hospital events was performed by an independent endpoint committee, who thoroughly reviewed data from hospital and out-of-hospital journals, autopsy records and death certificates. Event ascertainment followed a detailed protocol according to established diagnostic criteria. Each case was reviewed separately. Stroke was defined according to the WHO definition, only ischemic strokes were included [18]. Individuals who had died, moved, or emigrated from Tromsø were identified through the population Registry of Statistics Norway. The national 11-digit identification number allowed a linkage to the population Registry of Statistics Norway and ensured a complete follow-up status for all-cause mortality until November 30^th, 2010 (follow-up time 15 years). The cardiovascular (CV) endpoint registry was completed until December 31, 2007 (follow-up time 12 years). Data were censored for date of registered emigration, or deaths from causes other than myocardial infarction and stroke.

Data are given as mean ± SD or median and interquartile range. Uric acid was categorized into gender-specific tertiles. Crude and age-adjusted incidence rates were calculated as events per 1000 person years at risk. Age-adjustment of incidence rates was performed on 10 year age groups with the population of Tromsø in 1995 as the standard population. Multiple linear regression analyses were performed with uric acid as the dependent variable. Covariates were systolic blood pressure, body mass index (BMI), high density lipoprotein cholesterol (HDL), total cholesterol, smoking status and physical activity. Renal covariates (estimated GFR and ACR) and use of diuretics or antihypertensive drugs were also added to the models in the multivariable analyses. ACR was logarithmically transformed. Cox proportional hazard models were used to investigate associations of uric acid with cardiovascular outcomes and mortality, calculated per 1 SD (87 μmol/L) change in uric acid, in unadjusted, age-adjusted and multivariable analyses. The proportional hazard assumption was checked by visual inspection of the -log-log survival curves. Tests for interactions and non-linearity (by quadratic terms) were assessed in separate models. Non-linear effects were also explored in fractional polynomial regression models. P values < 0.05 were considered statistically significant. Most analyses were run using SPSS software version 15.0 (SPSS, INC, Chicago, Illinois), fractional polynomial regression models were performed with STATA/MP 12.1 (Stata Corp LP, College Station, Texas).

Mean serum uric acid was 357 ± 84 μmol/L for men and 276 ± 70 μmol/L for women. Figure 2 shows serum uric acid concentrations according to gender and age. In both genders, increasing uric acid was associated with a poorer risk profile in terms of elevated BMI, blood pressure, ACR, and proportion of hypertensive persons. Physical activity, renal function and HDL were lower with increasing uric acid concentrations. Only a few subjects were using diuretics and anti-hypertensive medication at baseline in 1994/95 (Table 1).

The correlation between uric acid and age differed in men and women (men; r = -0.09, p < 0.001, women; r = 0.17, p < 0.001, Pearson’s correlation coefficients), and tested significant for gender interaction (P < 0.001). Results of the multiple linear regression analyses are given in Table 2. In both genders, standardized beta-coefficients were highest for BMI, HDL-cholesterol, total cholesterol, and GFR. The addition of renal factors (eGFR and ACR; model 5) into the model including drug intake and traditional cardiovascular risk factors, contributed significantly to variation of serum uric acid; adjusted R² increased from 0.19 to 0.23 in men, from 0.20 to 0.27 in women, and from 0.20 to 0.40 in pooled analyses for men and women together. ACR was independently associated with serum uric acid in women, but not in men, and there was a significant interaction with gender (P = 0.005).

Number of events during follow-up were 659 first ever cases of fatal or non-fatal MI, 430 fatal or non-fatal ischemic strokes, and 1433 deaths from all causes. Median observation time was 12.5 years for myocardial infarction and ischemic stroke, and 15.7 years for all-cause mortality. Crude and age-adjusted incidence, for MI and ischemic stroke, picturing absolute risk rates according to increasing tertiles of serum uric acid, are shown in Figure 3. Age-adjusted incidence rate for myocardial infarction was significantly higher in the upper serum uric acid tertile among men, but not in women. Increasing crude and age-adjusted incidence rates with higher levels of uric acid were observed in both genders for ischemic stroke.

In both genders, P-values for linear, increasing trend were <0.001 for all endpoints in multivariable, Cox proportional hazard models where the lowest serum uric acid tertile was reference (data not shown). In multivariable analyses calculated per 1 SD (87 μmol/L) increase in uric acid (Table 3), uric acid was independently associated with all-cause mortality in men (HR, 95% CI; 1.11, 95% CI 1.02-1.20), women (HR, 95% CI; 1.16, 1.05-1.29), and in the total population (men and women; HR, 95% CI: 1.13, 1.06-1.21). In fully adjusted, multivariable models, serum uric acid was not independently associated with MI in any gender or in the mixed population. In both genders, uric acid lost significance when lipids were introduced as covariates. Serum uric acid was strongly associated with ischemic stroke in men after multivariable adjustments (HR, 95% CI: 1.31, 1.14-1.50). In contrast, significance was lost among women after adjustments for blood pressure and BMI. Test for gender interaction was not significant (P = 0.19). In multivariable analysis of the mixed population, adjusted for sex, 87 μmol/L increase in uric acid was associated with a 24% increase in risk of future ischemic stroke (Table 3). Tests for non-linearity in fractional polynomial models for associations of serum uric acid with the three endpoints were not significant in any gender. Cox regression analyses were also performed without exclusion of persons with diabetes and previous cardiovascular disease. Risk estimates for uric acid and myocardial infarction in fully adjusted analyses did not change substantially (HR and 95% CI for men were 1.06, 0.99-1.14, p = 0.06 and for women 1.04, 0.92-1.14, p = 0.5). We also stratified the study population according to serum uric acid levels higher or lower than the hospital laboratory’s upper reference range. Belonging to the group with higher uric acid levels was not significantly associated with myocardial infarction in men (P = 0.26) or women (P = 0.27) after multivariable adjustments.