Introduction

Cardiovascular diseases (CVDs) are a group of disorders affecting the circulatory system, leading to vascular stenosis, myocardial ischemia, and systemic dysfunction. Among these, coronary heart disease (CHD) remains the leading cause of global mortality, responsible for 19.8 million deaths in 20221. In China, CHD prevalence has surged alarmingly, with 11.39 million cases reported in 2023, including over 10% in individuals under 40 years old2. This contrasts sharply with declining trends in high-income nations, highlighting urgent needs for cost-effective risk assessment tools.

The gold standard for diagnosing CHD is coronary angiography. However, its invasiveness, cost, and risk of contrast-related complications limit widespread use3. Insulin resistance (IR) drives atherosclerosis and hypertension, emerging as a pivotal risk factor for CHD4,5,6. While HOMA-IR is widely used for IR estimation, the hyperinsulinemic-euglycemic clamp (HEC) remains the gold standard7. However, HEC’s complexity and cost hinder clinical adoption8. In 2010, Guerrero et al. introduced the TyG index as a simple and accessible surrogate marker for IR. Calculated from fasting triglycerides and glucose, the TyG index correlates strongly with CHD incidence and progression9,10,11. Subsequent studies have demonstrated that the TyG index outperforms HOMA-IR in predicting cardiovascular risk. Despite its potential, existing research focuses on single populations, neglecting ethnic and geographic variations in TyG-CHD associations12.

This study aims to investigate the relationship between TyG index and CHD incidence, with a particular focus on comparing the predictive performance of TyG index in elderly populations from China and UK. By leveraging data from the China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA), we seek to provide insights into the role of TyG index in CHD risk stratification and its potential as a cost-effective tool for early intervention.

Methods

Study population

To ensure cross-national comparability between the China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA), a systematic harmonization protocol was implemented following established multinational cohort guidelines. Core variables — including fasting glucose, triglycerides, and blood pressure-were standardized to international diagnostic criteria. For instance, fasting glucose levels were aligned with WHO thresholds for diabetes diagnosis, and triglyceride measurements were uniformly converted to mg/dL. Blood pressure protocols were reconciled by applying validated calibration factors to adjust for methodological differences between cohorts. Socioeconomic variables such as residential status were contextually adapted, mapping China’s agricultural hukou system to UK rural-urban classifications through occupation and income proxies.

In this study, the baseline wave was wave 2 of CHARLS (2013) and wave 6 of ELSA (2012/13), as the cardiometabolic variables of interest were not measured in the earlier waves. To estimate CHD incidence, we used follow-up data from waves 2 to 4 (2013 to 2015) of CHARLS and waves 6 to 8 (2012/13 to 2016/17) of ELSA.

The study populations were derived from the China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA), two nationally representative cohorts focusing on middle-aged and elderly populations. CHARLS: Participants aged ≥ 45 years at baseline (2011–2012) with complete cardiometabolic data (fasting glucose, triglycerides, and waist circumference) and at least one follow-up visit (2013–2015) were included. ELSA: Participants aged ≥ 50 years at baseline (2012–2013) with complete TyG index components (fasting glucose and triglycerides) and follow-up data (2016–2017) were enrolled. Blood pressure measurements in CHARLS (single reading) were adjusted to match ELSA’s triplicate protocol using validated correction factors (+ 2.1 mmHg systolic, + 1.3 mmHg diastolic) derived from a prior calibration study. The participant selection process for CHARLS and ELSA is detailed in Figs. 1 and 2.

Exclusion Criteria:

  1. 1.

    Individuals with pre-existing coronary heart disease (CHD) at baseline to avoid reverse causality.

  2. 2.

    Participants missing > 20% of key variables (e.g., TyG index components, anthropometric measurements).

  3. 3.

    Those lost to follow-up or with incomplete mortality/CHD event records.

CHARLS is a nationally representative longitudinal survey of the Chinese population aged 45 years and older, conducted by the National Institute of Development Studies at Peking University. Detailed information about the study population is reported in other publications15. To date, national baseline surveys were conducted in 2011–2012 and completed in 2018 (phase 2 in 2013, phase 3 in 2015, and phase 4 in 2018). Blood samples were obtained at baseline (including 11,847 participants) and during phase III (including 13,420 participants). For this analysis, participants who were 45 years of age or older and who provided complete fasting plasma glucose (FBG) and triglyceride (TG) data were included. In the 2013 and 2015 cohorts, 2,405 respondents had complete detailed data and were ultimately deemed eligible to participate.

The detailed population screening process for CHARLS is described below. (Refer to Fig. 1)

Fig. 1
figure 1

Flow chart of population screening in CHARLS. Data are presented as n; TyG, Triglyceride-glucose.

Full size image

ELSA is a longitudinal study of population ageing based in UK, collecting data from individuals aged 50 years and over to understand various aspects of ageing in UK. More than 18,000 people have been enrolled in the study since 2002, with follow-up interviews conducted every two years. Information collected by ELSA includes physical and mental health, welfare, financial status, attitudes towards ageing, and how these aspects have changed over time. Data from ELSA participants provide policy references for various aspects of ageing, including health and social care, retirement and pension policies, and social and civic engagement. Based on the conditional screening for analysis in this survey, 3,038 respondents in the 2012/13 and 2016/17 cohorts had complete data and were ultimately considered eligible to participate.

The detailed population screening process for ELSA is described below. (Refer to Fig. 2)

Fig. 2
figure 2

Flow chart of population screening in ELSA. Data are presented as n; TyG, Triglyceride-glucose.

Full size image

Statistical assessment

Coronary heart disease evaluation

In both the CHARLS and ELSA databases, CHD assessment primarily relied on self-reported physician diagnoses of coronary heart disease, acute myocardial infarction, angina, or other clinically confirmed cardiovascular events such as heart failure and ischemic stroke. In addition to self-reports, both databases considered biomarkers and physiological measures related to CHD risk, such as blood glucose levels, blood pressure, lipid profiles, and inflammatory markers. In the ELSA database, participants also underwent regular physical examinations and were asked to provide their medical records, including information on past diagnoses, treatments, and medication use.

TyG index evaluation

Exposure variables in this study included TyG indices collected from 2013 to 2015 (CHARLS) and 2012/13 to 2016/17 (ELSA). In both databases, body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). The TyG index was obtained using the formula: TyG = [TG (mg/dL) × FBG (mg/dL)/2]. The TyG index was further adjusted by multiplying it with BMI and waist circumference to produce TyG-BMI and TyG-WC, respectively.

Covariates evaluation

In CHARLS, baseline demographic information (age, height, weight, waist circumference, sex, education level, marital status), basic anthropometric measurements of systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, and potential risk factors, such as diabetes mellitus(DM), high blood pressure(HBP), dyslipidemia, smoking, and alcohol consumption, were retrospectively evaluated. Laboratory tests included fasting blood glucose (FBG), total cholesterol (TCHO), triglycerides (TG), high-density lipoprotein cholesterol (HDL), low-density lipoprotein (LDL), glycated hemoglobin A1c (HbA1c), and C-reactive protein (CRP).

In ELSA, baseline demographic information (age, sex, race [white/non-white], education level, marital status, employment status, exercise habits), SBP, DBP, BMI, and potential risk factors, such as DM, HBP, hypercholesterolemia, smoking, and alcohol consumption, were retrospectively evaluated. Laboratory tests included FBG, TC, TG, HDL, LDL, CRP, and fasting blood glucose score (FGS).

Statistical analysis

In the CHARLS and ELSA, the TyG of the baseline wave was divided into Q1, Q2, Q3, and Q4 according to the interquartile range.

This study utilized data from public databases, employing a multistage sampling approach. To account for attrition rates and poststratification, sample weights were assigned to participants to ensure representativeness. Descriptive statistics, including mean and standard deviation (SD) for continuous data and percentages for categorical data, were used to report basic characteristics. For continuous variables, T-tests or Mann-Whitney U tests were applied, while chisquare tests or Fisher’s exact tests were used for categorical variables to analyze differences in baseline characteristics between groups.

Univariate logistic regression identified crude associations between variables and CHD, while multivariate models adjusted for confounders (e.g., age, BMI, smoking) to estimate independent effects of TyG index quartiles. Subsequently, a restricted cubic spline model was established to analyze the relationship between the triglyceride-glucose (TyG) index and CHD, with knots placed at the 25th, 50th, and 75th percentiles. Additionally, several subgroup and interaction analyses were performed to identify potential influencing factors. Additive interactions between TyG index and covariates were quantified using the synergy index (SI), with values > 1 indicating synergistic effects, complemented by the relative excess risk (RERI) and attributable proportion (AP).

Multivariate logistic regression was used to calculate odds ratios (ORs), estimating the likelihood of CHD incidence per TyG quartile increase, with 95% confidence intervals (CIs) indicating the precision of these estimates. Restricted cubic spline (RCS) regression with three knots was employed to model nonlinear dose-response relationships between TyG index and CHD risk, allowing identification of population-specific risk thresholds. Statistical analyses were conducted using R version 4.0.3 and SPSS software. A two-sided P-value of less than 0.05 was considered statistically significant.

Results

Baseline characteristics of study participants

In the CHARLS study, 2,405 participants were included, with a mean age of 70.46 ± 8.85 years, of whom 41% were male. Analysis of the CHARLS database indicated that the average TyG index in the diseased group in 2013 was 8.89 ± 0.65, while in the non-diseased group, it was 8.96 ± 0.66. Comparative analysis revealed that the diseased group had elevated levels of fasting plasma glucose, TG, HDL, and glycosylated hemoglobin. (Refer to Tables 1 and 2)

Table 1 Baseline characteristics of basic indicators in middle-aged and elderly people in CHARLS.
Full size table
Table 2 Univariate analysis of basic indicators in middle-aged and elderly people in CHARLS.
Full size table

In the ELSA study, there were 3,038 participants, with a mean age of 65.11 ± 7.43 years, of whom 42.3% were men. Analysis of ELSA database data from 2012/13 to 2016/17 showed that the TyG index of the diseased group was 6.87 ± 0.49, whereas that of the non-diseased group was 6.97 ± 0.46. Comparative analyses indicated that the diseased group had a higher proportion of males, as well as higher BMI, TyG, and TyG-BMI values, and a higher prevalence of hypercholesterolemia, and exhibited a higher prevalence of hypertension, diabetes, and hyperlipidemia. Furthermore, the diseased group exhibited elevated levels of fasting blood glucose, serum triglycerides, and C-reactive protein. (Refer to Tables 3 and 4)

Table 3 Baseline characteristics of basic indicators in middle-aged and elderly people in ELSA.
Full size table
Table 4 Univariate analysis of basic indicators in middle-aged and elderly people in ELSA.
Full size table

These comparative analyses suggest that environmental risk factors such as smoking and alcohol consumption, along with anthropometric measurements (BMI, systolic blood pressure, diastolic blood pressure), are closely associated with CHD. Laboratory tests indicating hypertension, diabetes, hypercholesterolemia, fasting blood glucose, and LDL levels are crucial indicators directly influencing the occurrence of heart disease. These factors, through their impact on the TyG index, may subsequently affect the prevalence of CHD.

Odds ratios for Coronary heart disease

Table 5 describes the correlation between the risk of CHD occurrence and the quartiles of TyG in CHARLS. The analysis showed that after considering factors such as gender, height, weight, BMI, alcohol consumption, history of smoking, current smoking status, hypertension, and diabetes, the risk of CHD significantly with the increase of TyG quartiles (OR = 2.111, 95% CI:1.514–2.888). Moreover, Q4 was the most sensitive to the susceptibility of CHD compared with Q1. TyG was significantly associated with the occurrence of CHD cases. (Refer to Table 5)

Table 6 describes the correlation between the risk of CHD occurrence and the quartiles of TyG in ELSA. The analysis showed that after consideringounding factors such as gender, height, weight, BMI, alcohol consumption, history of smoking, current smoking status, hypertension, and diabetes, the risk of CHD significantly with the increase of TyG quartiles (OR = 1.356, 95% CI:1.093–1.901). (Refer to Table 6)

These findings suggest that higher TyG index values are associated with an increased risk of CHD in both the CHARLS and ELSA populations. Compared with the Chinese population, the British population in Q4 of TyG had a lower susceptibility to CHD, and there was a significant difference the two countries.

Table 5 Multivariate analysis of TyG and CHD in middle-aged and elderly people in CHARLS.
Full size table
Table 6 Multivariate analysis of TyG and CHD in middle-aged and elderly people in ELSA.
Full size table
Fig. 3
figure 3

The results of RCS analysis between TyG and CHD in middle-aged and elderly people in CHARLS. Note: Controlling for gender, BMI, alcohol consumption, history of smoking, current smoking status, hypertension, and diabetes in CHARLS.

Full size image
Fig. 4
figure 4

The results of RCS analysis between TyG and CHD in middle-aged and elderly people in ELSA. Note: Controlling for gender, BMI, alcohol consumption, history of smoking, current smoking status, hypertension, and diabetes in ELSA.

Full size image

Figure 3 depicts the linear dose-response relationship between TyG and CHD in CHARLS (overall = 0.004, nonline = 0.538; overall = 0.007, nonlinearity = 0.870), with the survey determining a gradual in CHD risk as TyG continues to rise. The TyG threshold is 9.72. Figure 4 depicts the linear dose-response relationship between Ty and CHD in ELSA (overall = 0.007, nonlinearity = 0.870), also showing a gradual increase CHD risk as TyG continues to rise, with a TyG threshold of 8.51, which is lower than that in China (p < 0.05). Ultimately, it is shown that TyG is a decisive factor for CHD events.(Refer to Figs. 3 and 4).

Subgroup analysis

Subgroup analyses revealed significant heterogeneity in the TyG-CHD association across demographic and clinical strata between Chinese and British populations (Figs. 5 and 6; Tables 7 and 8). In the Chinese cohort, young males with pronounced TyG-CHD correlations observed among smokers (OR = 1.28, 95% CI: 1.03–1.59; P = 0.026) and alcohol consumers (OR = 1.31, 95% CI: 1.06–1.62; P = 0.013). Notably, individuals without diabetes or dyslipidemia demonstrated stronger TyG-CHD associations in China (OR = 1.98, 95% CI: 1.51–2.59) compared to the UK (OR = 1.32, 95% CI: 1.02–1.71), suggesting minimal confounding by glucose-lowering therapies. In contrast, the UK cohort displayed attenuated associations overall, with a paradoxical inverse TyG-CHD relationship in non-diabetic individuals (OR = 0.37, 95% CI: 0.14–0.96; P = 0.041), potentially influenced by widespread lipid-lowering interventions. Diabetes status emerged as a critical modifier in the UK (P for interaction = 0.016), significantly weakening the TyG-CHD association (OR = 0.22, 95% CI: 0.08–0.66; multiplicative interaction), whereas no such effect was observed in China. Additive interaction analyses further indicated negligible biological synergy between TyG and sociodemographic factors (e.g., education, marital status; SI ≈ 1.0), though diabetes exhibited potential antagonistic effects (UK: SI = 0.07, 95% CI: 0.005–1.101), likely reflecting differential metabolic management practices. These findings underscore the importance of contextualizing TyG-driven cardiovascular risk within population-specific metabolic profiles and healthcare frameworks to optimize risk stratification strategies.

Fig. 5
figure 5

Forest plot of TyG and CHD subgroup analysis in middle-aged and elderly people in China. Data are presented as n(%); DM, diabetes mellitus; HBP, high blood pressure.

Full size image
Fig. 6
figure 6

Forest plot of TyG and CHD subgroup analysis in middle-aged and elderly people in UK. Data are presented as n(%); DM, diabetes mellitus; HBP, high blood pressure.

Full size image
Table 7 The analysis of multiplicative and additive interactions between TyG and CHD in middle-aged and elderly people in China.
Full size table
Table 8 The analysis of multiplicative and additive interactions between TyG and CHD in middle-aged and elderly people in UK.
Full size table

Discussion

Our analysis of CHARLS data reveals that sustained elevation of the TyG index independently predicts CHD risk in adults ≥ 45 years, extending prior findings of its linear association with ischemic stroke19,20. The dose-dependent relationship identified via restricted cubic spline models (TyG threshold: China = 9.72 vs. UK = 8.51) provides actionable thresholds for population-specific interventions, consistent with Cui et al.‘s stroke progression studies17,21. While outcome variation within TyG quartiles suggests modulating factors-akin to hypertensive cohorts showing negligible gender differences14,22,23, the contradictory null association in non-diabetic PCI patients24 underscores population-specific TyG utility.

Although TyG-BMI and TyG-WC combinations showed non-significant predictive gains in our cohort, their integration could refine risk stratification in populations with obesity-driven metabolic profiles, particularly in Asia where central adiposity disproportionately elevates CHD risk26,39. For instance, Chinese clinics might prioritize TyG-WC screening for individuals with BMI ≥ 23 kg/m²-a threshold lower than Western guidelines27 - to capture early endothelial dysfunction28. Conversely, in the UK, where lipid-lowering therapies are widespread, TyG-BMI may better identify residual risk in treated patients14.Recent domestic studies report TyG-WC’s superior discriminative capacity (AUC = 0.680 vs. TyG-alone 0.669)39, consistent with international recommendations to integrate TyG with obesity metrics39,40. Our nonsignificant results may stem from sample size limitations rather than biological irrelevance, a hypothesis supported by meta-analytic data from multinational cohorts24,25. Subgroup analyses revealed attenuated TyG-CHD associations in diabetes/dyslipidemia subgroups, potentially due to therapeutic interventions altering TyG dynamics. Conversely, the persistent positive association in metabolically stable subgroups underscores TyG’s value in primary prevention contexts.

Mechanistically, TyG elevation promotes CHD via endothelial dysfunction and platelet hyperactivity. Imaging studies confirm that elevated TyG correlates with coronary calcification and plaque progression28,29,30,31. These processes are likely amplified by TyG-associated metabolic disturbances including elevated BMI, hypertension, and dyslipidemia-and inflammatory activation via pathways such as hsCRP-mediated cascades32,33,41. Clinically, serial TyG monitoring could track metabolic trajectory shifts, enabling timely interventions (e.g., dietary modification) before overt CHD manifests. While insulin resistance’s role requires further study, its link to cardiac autonomic impairment16,17,18,19 underscores TyG’s potential as a holistic risk marker beyond lipid-glucose metrics alone.

The substantial disparity in CHD prevalence between China (37.84%) and UK (24.13%, p < 0.001) corresponds to marked differences in TyG profiles, nationwide TyG screening paired with lifestyle counseling could mitigate urbanization-driven risks. Chinese participants exhibited higher TyG indices (8.89 ± 0.65 vs. 6.87 ± 0.49) and TyG-BMI scores (218.42 vs. 187.40), potentially reflecting population-specific metabolic vulnerabilities. Asian populations demonstrate heightened cardiometabolic risk at lower BMI thresholds due to genetic predispositions and phenotypic traits5,26,27, compounded by rapid urbanization-driven lifestyle changes. Concurrently, UK’s earlier adoption of lipid-lowering therapies and standardized glycemic protocols may explain its attenuated TyG-CHD associatio, suggesting TyG’s utility for monitoring residual risk in treated populations14. Behavioral factors further contribute to this divergence: higher smoking rates (40.4% vs. 26.6%) and distinct alcohol consumption patterns in China likely exacerbate TyG-related endothelial damage34,35,36,37,38, whereas UK’s public health initiatives have progressively reduced modifiable risks over the past decade11.

Several limitations merit consideration. First, while the TyG index correlates with established insulin resistance markers, the absence of direct comparisons with gold-standard measures such as continuous glucose monitoring may affect diagnostic precision17,19. Second, excluding participants with incomplete glucose or triglyceride data (approximately 20% of the initial cohort) could introduce selection bias, potentially skewing results toward healthier populations. Third, reliance on self-reported CHD diagnoses without universal angiographic validation raises concerns about outcome misclassification. Additionally, biannual TyG measurements limited our ability to capture short-term metabolic fluctuations, and the exclusive focus on Chinese and British populations restricts generalizability to other ethnic groups with distinct metabolic profiles, such as African or Hispanic cohorts. Future studies should integrate continuous insulin resistance assessments, enroll high-risk subgroups, and validate findings in multi-ethnic cohorts with longitudinal TyG monitoring to address these gaps.

Conclusions

This study is the first to establish population-specific TyG thresholds (China = 9.72 vs. UK = 8.51) and demonstrate their utility in cross-national CHD risk stratification. In studies of Chinese and British populations, we have identified that elevated baseline TyG index and longitudinal TyG index progression are associated with elevated CVD risk, particularly CHD. Individuals with sustained TyG elevation, especially those with diabetes or dyslipidemia, require close monitoring and early intervention. The TyG index is an effective tool for CHD risk stratification, especially in clinical settings with limited resources. Its predictive power is particularly notable in the Chinese population, which may be related to the high prevalence of metabolic syndrome. Integrating TyG with TyG-BMI or hsCRP improves risk prediction, prioritizing patients with dual TyG-hsCRP elevations for intensive management.

Additionally, the stark contrast in CHD prevalence between China and UK underscores the interplay of biological, behavioral, and systemic factors. China’s higher TyG-driven CHD burden calls for urgent public health interventions, including nationwide metabolic screening and lifestyle education. Meanwhile, UK’s experience highlights the value of integrated healthcare systems in mitigating cardiovascular risk. Future studies should prioritize multi-ethnic validation of population-specific TyG thresholds (e.g., 9.72 in China vs. 8.51 in UK) across diverse healthcare systems, while elucidating how genetic predispositions (APOA5 variants) and cultural determinants (e.g., dietary patterns, urbanization gradients) modulate TyG-CHD associations. Mechanistic investigations into TyG-driven pathways (endothelial dysfunction, autonomic impairment) should inform targeted interventions, particularly in high-risk subgroups identified in this study (e.g., Chinese males with agricultural hukou). Ultimately, longitudinal trials are needed to evaluate whether TyG-guided lifestyle or pharmacological interventions reduce cardiovascular events, translating these findings into precision prevention strategies tailored to ethnic and geographic contexts.