The Kailuan study is a prospective cohort study in the Kailuan community in Tangshan, China. The detailed study design and procedures have been described previously.[17-19] During June 2006 to October 2007, a total of 101,510 participants (81 110 men and 20 400 women; aged 18 to 98 yeas) were enrolled in the first survey (baseline) and underwent a comprehensive biennial health examination. All participants were followed up until their death or December 31, 2017. Changes in TyG index was developed from 2006 to 2010 to predict CVD risk from 2010 to 2017 (Figure S1). We excluded 3,669 and 2,042 participants with MI or stroke in or prior 2010, 30,971 participants who did not finish the survey at 2010, 1,282 and 1,103 participants with missing data on FBG or TG at baseline or the survey at 2010. Ultimately, a total of 62,443 participants were enrolled in the present study (Figure S2). The study was performed according to the guidelines of the Helsinki Declaration and was approved by the Ethics Committee of Kailuan General Hospital (approval number: 2006-05) and Beijing Tiantan Hospital (approval number: 2010-014-01). All participants were agreed to take part in the study and provided written informed consent.
Data collection and definitions
Information on demographic characteristics, lifestyle factors (smoking status, drinking status, and physical activity), and medical history were collected via standardized questionnaire by trained staffs. Education was classified as illiteracy or primary school, middle school, and high school or above. Income was categorized into > 800 and ≤ 800 yuan/month. Smoking and drinking status were stratified into never, former or current. Physically active was classified as ≥4 times per week and ≥20 minutes at a time, <80 minutes per week, or none. Body mass index (BMI) was calculated by dividing body weight (kg) by the square of height (m2). Blood pressure was measured in the in the seated position using a mercury sphygmomanometer, the average of 3 readings were calculated as systolic blood pressure (SBP) and diastolic blood pressure (DBP). All the blood samples were analyzed using an auto-analyzer (Hitachi 747, Hitachi, Tokyo, Japan) on the day of the blood draw. The biochemical indicators tested included fasting blood glucose, serum lipids, serum creatinine, and high-sensitivity C-reactive protein (hs-CRP).
Hypertension was defined as SBP ≥140 mm Hg or DBP ≥90 mm Hg, any use of the antihypertensive drug, or self-reported history of hypertension. Diabetes was defined as FBG≥7.0mmol/L, any use of glucose-lowing drugs, or any self-reported history of diabetes. Dyslipidemia was defined as any self-reported history or use of lipid-lowering drugs, or TC ≥ 5.17 mmol/L.
Calculation of changes in TyG index
The TyG index was calculated as ln (fasting TG [mg/dl] × FBG [mg/dl]/2) as previous done.[20, 21] Changes in TyG index was calculated as TyG index value at 2010 minus that at baseline (2006).
Assessment of outcomes
The outcome in the present study was the first occurrence of CVD events. The types of CVD included stroke and MI. We defined CVD events as described previously.[17, 22, 23] The database of CVD diagnoses was obtained from the Municipal Social Insurance Institution and Hospital Discharge Register and was updated annually during the follow-up period. An expert panel collected and reviewed annual discharge records from 11 Kailuan hospitals to identify patients who were suspected of CVD. Incident stroke was diagnosis based on neurological signs, clinical symptoms, and neuroimaging tests, including computed tomography or magnetic resonance, according to the World Health Organization criteria. MI was diagnosed according to the criteria of the World Health Organization on the basis of clinical symptoms, changes in the serum concentrations of cardiac enzymes and biomarkers, and electrocardiographic results.[17, 25]
Participants were divided into four categories according to quartiles of changes in TyG index. The baseline characteristics were presented as mean±standard deviation (SD) or frequency with percentage as appropriate. Tests of differences in characteristics across changes in TyG index categories were performed using analysis of variance or the Kruskal-Wallis test for continuous variables according to distribution and chi-square for categorical variables. The person-years were determined from the date when the message was collected at baseline to either the date of MI onset, death, or the date of participating in the last examination in this analysis, whichever came first. Kaplan-Meier methods were performed to evaluate the incidence rate of CVD and its subtypes, and differences among groups were evaluated by log-rank test.
Cox proportional hazard regression model was applied to calculated hazard ratio (HR) and 95% confidence interval (CI) for CVD and its subtypes. The proportional hazard assumption was evaluated with visualization of Schoenfeld residuals and no potential violation was observed. Two models were constructed. Model 1 was adjusted for age, sex, and TyG index at baseline. Model was additionally adjusted for education, income, smoking status, drinking status, physical activity, BMI, SBP, DBP, a history of hypertension, diabetes mellitus, and dyslipidemia, antidiabetic drugs, lipid-lowering drugs, antihypertensive drugs, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and hs-CRP at baseline. P-values for trend were computed using quartiles as ordinal variables. To capture the dose-response relationship between changes in TyG index and CVD, restricted cubic splines with four knots at the 5th, 35th, 65th, and 95th percentiles of TyG index change distribution with median of the Q1 group as the reference point.
Additional analyses were performed to validate the robustness of the results. First, competing risk model was applied to assess the association between changes in TyG index and the outcomes considering non-CVD death as a competing risk event. Second, restricted analysis was conducted by excluding participants with abnormal FBG level (≥7.0 mmol/L) or abnormal TG level (≥1.7 mmol/L) at baseline. Third, to explore the potential impact of reverse causality, we repeated the primary analysis using a 2-year lag period by excluding incident stroke cases from the first 2 years of follow-up. Subgroup analyses were conducted stratified participants by age (< 60 and ≥ 60 years), sex (women and men), BMI (<25 and ≥ 25 kg/m2), and FBG (<5.6, 5.6-7.0, and ≥ 7.0 mmol/L) to assess the possible effect modification by these variables, interactions between subgroups were tested using likelihood ratio tests comparing models with and those without multiplicative interaction terms. Additionally, we used C statistics, integrated discrimination improvement (IDI), and net reclassification index (NRI) to evaluate the incremental predictive value of change in TyG index beyond conventional risk factors.
All analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina) and R software version 3.6.1 (R Core Team, Vienna, Austria). All statistical tests were 2-sided, and P < 0.05 was considered statistically significant.