Study population
The CARDIA multi-center randomized, prospective cohort study was conducted from 1985 to 1986 (year 0), enrolling 5115 African-American and Caucasian aged from 18 to 30 from across the general population or selected census areas within four research centers in the USA. All participants were investigated at years 2, 5, 7, 10, 15, 20, 25 and 30, respectively. The institutional review committee from each research center accepted the research scheme and informed consent from all individual cohort participants was obtained in writing.The baseline data for this study used the 0-year examination data, for a total of 5,114 participants (one patient withdrew consent). Following the exclusion of patients with incomplete clinical data (missing fasting blood glucose, triglyceride, insulin, missing endpoint records), a total of 4992 patients formed part of the final analytical queue (Additional file: Fig. S1). Patientss were grouped into four groups, depending on TyG index quartiles.
TyG index, HOMA-IR and CHF
Participants at 0 year fasted for at least eight hours, immediately followed by blood collection using an EDTA vacuum vessel. Consequrently, plasma was isolated and frozen at -70°C prior to shipping to the laboratory using dry ice. Glucose was determined at baseline using hexokinase UV, calibrated, and followed by enzymatic analysis of triglyceride levels [22]. The TyG index was determined as:Ln (fasting triglycerides (mg/dL) × fasting blood glucose (mg/dL)/2) [23]. HOMA-IR was determined as: fasting blood glucose (mmol/L) × fasting serum insulin (µU/mL)/22.5 [24].
Diagnostic validation of CHF necessitated a finalized CHF diagnosis by a physician, together with the implementation of CHF clinical management protocols during the patient hospitalization period (i.e. diuretic/s + digoxin / Glycerin tri-nitrate, hydralazine, ACE-inhibitor/s or angiotensin receptor blocker/s). All patients were monitored until an endpoint of August 2017.
Covariates
Covariates included in the present analysis were obtained through established protocols / quality assurance processes throughout all centers involved [25]. Education level was stratified as ≤ 12 years (up to high school degree), 13 to 16 years (up to graduate educational level), and ≥ 13 years (representing>high school education). Smoking-status was stratified as present and present non-smoking (including past and never smoking). Hypertension was deemed present upon a systolic blood pressure of ≥ 130 mmHg, diastolic blood pressure of ≥ 90 mmHg, or current consumption of anti-hypertension drug/s [26]. Obesity was deemed present upon a body mass index (BMI) ≥ 30 [27]. The dietary modification study equation for renal disease diet was implemented in this study in order to estimate the glomerular filtration rate (eGFR) within serum creatinine: eGFR (mL/min/1.73m2) = 175 × standardized Scr−1.154 × age−0.203 × 1.212 [if African-American] × 0.742 [if female].Participants with eGFR < 60mL/min/1.73m2 were deemed to have chronic kidney disease (CKD) [28, 29]. Detailed descriptions of measurements for total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides, serum creatinine, and fasting plasma glucose for all participants were previously published.
Statistical analyses
Normally distributed continuous data were represented by mean±SD, while non-normally distributed continuous data were represented by median (inter-quartile range). Categorical variables reported percentage frequency. Participants were classified into four groups according to the quartiles of the TyG index. Wilcoxon or Kruskal–Wallis test were employed for analyzing group variations for continuous variables, while Chi-square test was employed for categorical variables. Smooth curve fittings and scatter plots were used to address the relationship between TyG index and HOMA-IR. The Cox proportional-hazard regression model was employed to determine HR and 95% CI for CHF events by quartiles of TyG index, and HOMA-IR, respectively. The proportional hazard assumption was evaluated by visualization of Schoenfeld residuals, where such analytical outcomes indicated no evidence of assumption breaches (Additional file: Table S1). Multi-collinearity was investigated using variance inflation factors, while TC was removed as a significant variance inflation factor (≥ 5) . Three models were fitted: model 1 was not adjusted; model 2 was adjusted for age, sex, and race; model 3 was adjusted for variables included in model 2 and education level, smoking status, hypertension, diabetes mellitus, hypercholesteremia, systolic and diastolic blood pressure, obesity, CKD, HDL-C and LDL-C. Trend P values were evaluated by a median value within each quartile, as a continuous variable. Kaplan–Meier curve data outcomes were employed for determining cumulative incidence of CHF events through both TyG index and HOMA-IR quartiles, with estimation variations being comparatively analyzed through log-rank protocols. The ROC curve and area under the curve were used to assess both TyG index-based and HOMA-IR-based capacity for predicting CHF event risk during follow-up. The participants were divided into subgroups according to sex, race, education, obesity, smoking status, hypertension and CKD status.The results were scrutinized following adjustments for age, sex, race, education, obesity, smoking status, hypertension, diabetes mellitus, hypercholesteremia, CKD, LDL-C, HDL-C, except for the subgroup variable. All statistical analyses were conducted using R® software (version 4.0.3, http://www. R-project.org/). The study deemed that P values less than 0.05 (bilateral) conferred statistical significance.