Data were obtained from the Kailuan study, which is a community-based prospective cohort study conducted in Tangshan, China. The details of the study design and procedures have been described previously.[23-25] Since June 2006 (the baseline survey), a total of 101,510 participants (81,110 men and 20,400 women, aged 18-98 years) were enrolled from 11 hospitals in the Kailuan community, and underwent questionnaire assessments, clinical examinations and laboratory tests. Then all participants were followed up every 2 years and incidence of chronic diseases (e.g. cardiovascular disease) was recorded annually. In the present study, SUA trajectories were developed from 2006 to 2012 to predict MI and all–cause mortality risk after 2012. Participants were excluded if they had MI or died in or prior 2012 or if they had at least 2 measurements of SUA during the 2006-2012. Following these exclusions, we included 85,530 participants in the current analysis (Figure 1). The baseline characteristics of included participants and excluded participants due to missing data on SUA is showed in Table S1.
Assessment of SUA
Fasting blood samples were collected in the morning after an 8 to 12 h overnight fast and transfused into vacuum tubes containing ethylene diamine tetra-acetic acid. The concentration of SUA was examined with a commercial kit (Ke Hua Biological Engineering Corporation, Shanghai, China) using an automatic biochemical analyzer (Hitachi 7600, Tokyo, Japan) according to the manufacturer’s instructions.
Assessment of outcomes
The primary outcomes of our study were incident MI and all–cause mortality. The database of MI diagnoses was obtained from the Municipal Social Insurance Institution and Hospital Discharge Register and was updated annually during the follow-up period. We used the International Classification of Disease, 10th Revision code for the identification of potential MI (I21). Diagnosis of MI was based on a combination of chest pain symptoms, electrocardiographic signs, and cardiac enzyme levels. A panel of 3 cardiologists reviewed the medical records of potential MI cases to confirm diagnosis. All–cause mortality data were gathered from provincial vital statistics offices and reviewed by physicians.
Assessment of covariates
Demographic and clinical characteristics, including age, sex, education, income, smoking status, alcohol use, physical activity, and medical history were collected via standardized questionnaires. Educational attainment was categorized as illiteracy or primary school, middle school, and high school or above. Income level was categorized as <800 yuan and ≥800 yuan. Smoking status and alcohol use were classified as never, former, or current. Active physical activity was defined as ≥80 minutes activity per week. Body mass index was calculated by dividing body weight (kg) by the square of height (m). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured 3 times with the participants in the seated position using a mercury sphygmomanometer, and the average of 3 readings was used in the analyses. All blood samples were tested using a Hitachi 747 auto-analyzer (Hitachi, Tokyo, Japan) at the central laboratory of the Kailuan Hospital. The biochemical indicators tests included fasting blood glucose (FBG), serum lipids, serum creatinine, and high sensitivity C-reactive protein (hs-CRP). Estimated glomerular filtration rate (eGFR) was calculated using the creatinine-based Chronic Kidney Disease Epidemiological Collaboration (CKD-EPI 2009) equation.
Hypertension was defined as any self-reported hypertension or use of antihypertensive drug, or blood pressure ≥140/90 mm Hg. Diabetes mellitus was defined as any self-reported diabetes mellitus or use of glucose-lowering drugs, or FBG ≥7 mmol/L. Dyslipidemia was defined as any self-reported history or use of lipid-lowering drugs, or serum total cholesterol (TC) ≥5.17 mmol/L.
SUA trajectories were identified by group-based trajectory modelling using SAS PROC TRAJ. This method can automatically divide the study population into classes, in such a way that participants in the same class tend to have similar trajectories of SUA change. We used a censored normal model appropriate for continuous outcomes. Model fit was assessed using Bayesian information criterion, with smallest negative number indicating the best fit model. In this study, the model assigned individuals to 1 of the 3 categories of low-stable, moderate-stable, and high-stable SUA levels based on their longitudinal trajectory of SUA over the four examinations (Figure 2).
Baseline characteristics were described as mean ± standard deviation (SD) for continuous variables and percentages for categorical variables. Difference in Means and proportions between groups were compared using Student’s t-test, ANOVA, or chi-squared test, as appropriate. Person–years was computed from the date of the 2012 survey to the date of MI diagnosis, mortality, or the end of the follow up (December 31, 2019), whichever came first. The MI and all–cause mortality probabilities were estimated by Kaplan–Meier method and compared by log–rank test.
Cox proportional hazard regression was used to examine the association between SUA trajectories from 2006 to 2012 and the risk of MI and all-cause mortality by calculating hazard ratios (HRs) and 95% confidence intervals (CIs). The models met the proportional assumption criteria according to Schoenfeld residuals and log-log inspection. Three models were constructed. Model 1 was unadjusted. Because age and sex are strong determinants of exposure and outcomes, these factors was adjusted in Model 2. We further adjusted for education, income, smoking status, drinking status, physical activity, history of hypertension, diabetes and dyslipidemia, use of antihypertensive agents, hypoglycemic agents, lipid-lowering agents, BMI, SBP, DBP, FBG, TC, eGFR, hs-CRP at baseline and baseline SUA in Model 3.
Sensitivity analyses were performed to test the robustness of our findings. First, to control the regression-to-then mean influence, we adjusted average BMI, SBP, DBP, FBG, eGFR, hs-CRP and SUA during the exposure period. Second, to reduce the possibility of reverse causality, we conducted a lag-analysis by excluding incident MI or death, separately, with onset during the first 2 years of follow-up. Third, we used the Fine-Gray competing risk model considering non-MI deaths as competing risk events to assess the association between SUA trajectory and MI. Forth, we excluded participants with cardiovascular or cerebrovascular disease at baseline or during the follow-up to assess the association between SUA trajectory and all-cause, giving other common diseases may have additional effects on all–cause mortality. Subgroup analysis stratified by age (<60 or ≥60 years), sex, BMI (<25 or ≥25 kg/m2), and history of hypertension (no or yes) were performed to evaluate whether SUA trajectories exhibit different effect on the outcomes in special populations, interaction between stratified variables and SUA trajectories were tested using likelihood ratio.
All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). A two-sided P<0.05 was considered statistically significant.