The Kailuan Study is an ongoing prospective community-based cohort study conducted in Tangshan, China. All participants in the Kailuan Study are employees and retirees of the Kailuan Group. Details of the study design and procedure have been described elsewhere . At baseline, 101,510 participants (81,110 males and 20,400 females; aged 18-98 years) were recruited, underwent clinical and laboratory examinations, and completed a questionnaire interview (June 2006 to October 2007) at 11 hospitals affiliated with the Kailuan Group. Subsequent examinations involving anthropometric, cardiovascular risk factor measures, and self-reported questionnaires (including income, educational level, drinking and so on) occurred approximately biennially after baseline until December 2017 (2008/09, 2010/11, 2012/13, 2014/15 and 2016/17). For our investigation, participants were eligible if they attended three consecutive examinations between the second (2008/09), the third (2010/11) and the fourth (2012/13) examinations. Participants were excluded if they had prevalent myocardial infarction (MI), stroke or cancer, respectively, in or prior to 2012, or missing data on FBG or TG at any of the examinations, or BMI >45 kg/m2 at any of the examinations. Ultimately, a total of 44,087 participants were enrolled in the present study (Fig. 1).
The study was conducted in accordance with the guidelines of the Declaration of Helsinki and was approved by the Kailuan General Hospital Ethics Committee. All the participants agreed to take part in the study and provided written informed consent.
Data collection and definitions
Information on demographic and clinical characteristics (age, sex, lifestyle, and past medical history, etc.) were collected using a self-reported questionnaire, as detailed elsewhere . Education level was classified as primary school or below, middle school, and high school or above. Smoking and drinking status were classified as yes or no. Active physical activity was defined as “>4 times per week and 20 min at time”. BMI was calculated as the weigh (kg)/height2 (m2).
Elbow venous blood samples of 5 mL were collected into an anticoagulant tube containing EDTA between 7:00–9:00 am after overnight fasting for at least 8 h, and the serum was collected after centrifugation at 3000 × g for 10 min. The supernatant was measured within 4 h. All biochemical measurement including TG, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), high-sensitive C-reactive protein (Hs-CRP), FBG, and Uric acid (UA), and etc. was measured on the Hitachi 747 autoanalyzer (Hitachi, Tokyo, Japan).
Hypertension was defined as SBP≥140 mmHg or DBP≥90 mmHg, a self-reported history of hypertension, or any use of antihypertensive medication. Diabetes was defined as FBG≥7.0 mmol/L, a self-reported history of diabetes, or use of antidiabetic medication.
Cumulative TyG index
The TyG index was calculated as ln (fasting TG [mg/dL]×FBG [mg/dL]/2), as previously reported . Cumulative TyG index was defined as the summation of average TyG index for each pair of consecutive examinations multiplied by the time between these two consecutive visits in years:
[(TyG index2006+ TyG index2008)/2*time1-2]+[( TyG index2008+ TyG index2010)/2*time2-3]+[( TyG index2010+ TyG index2012)/2*time3-4]
where TyG index2006, TyG index2008, TyG index2010, and TyG index2012 indicate the TyG index at the baseline, second, third and fourth examinations, and time1-2, time2-3 and time3-4 indicate the participant-specific time intervals between consecutive visits in years. The means of time1-2, time2-3 and time3-4 were 2.06 years, 1.95 years and 2.22 years. The participants were stratified by quartile of cumulative TyG index: Q1 group, ≤50.65 (as reference group), Q2 group, 50.65-53.86, Q3 group, 53.86-57.44, Q4 group, ＞57.44.
According to previous studies, adults with a higher TyG index experience an increased risk of CVD, in the current analysis, participants with a fourth quartile of TyG index (＞9.02) at each examination were identified to be in the higher TyG index exposure group. TyG index exposure duration was defined as the times of examinations with higher TyG index among the 4 examinations, quantified as 0 years (never had higher TyG index), 2 years (had higher TyG index once), 4 years (had higher TyG index twice), 6 years (had higher TyG index third), and 8 years (had higher TyG index at all 4 study examinations).
Assessment of CVD
Follow-up ended at the first record of CVD event, all-cause death or at the end of follow-up on 31 December 2019, whichever came first. The types of CVD included MI and stroke. We used ICD-10th revision codes to identify CVD cases (I21 for MI, I6 for stroke)[14,15]. 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 discharges records from 11 local hospitals to identify patients who were suspected of CVD. The diagnosis of MI was determined by the patient’s clinical symptoms, electrocardiogram, and dynamic changes of myocardial enzyme following the World Health Organization’s Multinational Monitoring of Trends and Determinants in Cardiovascular Disease criteria. Stroke was diagnosed based on neurological signs, clinical symptoms, and neuroimaging tests, including computed tomographic or magnetic resonance imaging, in line with the World Health Organization criteria.
Continuous variables were compared using analysis of variance or the Kruskal-Wallis test according to distribution, and categorical variables were compared with the chi-square test.
Kaplan-Meier method was used to compute cumulative incidence of CVD and subgroup of CVD. Cox proportional hazard models were used with age as the time scale to estimate the hazard ratios (HRs) for incident CVD by cumulative TyG index, and were adjusted for baseline confounders, including sex, income (categories of high, intermediate, and low), educational level, drinking (yes or no), smoking (yes or no), diabetes, hypertension, lipid‐lowering medication, BMI, resting heart rate(RHR), HDL-C, LDL-C, UA, Hs-CRP. Missing covariates were imputed by multiple imputation using the fully conditional specification method SAS MI procedure. The results were consistent from analyses that excluded participants with missing covariates. The proportional hazard assumption was examined by Schoenfeld residuals.
To examine the robustness of our results, we performed several sensitivity analyses. First, we excluded events occurring in the first 2 years of follow-up to minimize potential reverse causation. Second, we excluded participants with diabetes or received treatment with lipid lowering medication and repeated analysis. Third, additional adjustment for TyG index at baseline. All analyses were done with SAS (version 9.4), at a two-tailed alpha level of 0.05.