Both cross-sectional studies were conducted in accordance with local and national regulations and the principles of the Declaration of Helsinki. The first dataset was obtained from the “Chinese Resident Cardiovascular Disease and Risk Factor Surveillance Program,” carried out in Yunnan Province between January and December 2021. The sampling procedures and methods employed have been published previously (17). In summary, eight districts/counties were selected using a stratified multistage random sampling method to screen permanent residents aged 18 years for CVD. The second dataset was obtained from a survey on CVD-related risk factors among highland residents of Yunnan Province, conducted between September 2023 and January 2024. The specific sampling procedure is outlined in Supplementary Material Figure S1. In brief, stratified multistage random sampling was employed to select permanent residents aged 35 years and older from four townships located at altitudes exceeding 2,500 m in Yunnan Province for CVD screening. Both projects were approved by the Ethics Committee of Fuwai Hospital, Chinese Academy of Medical Sciences, and Fuwai Yunnan Cardiovascular Disease Hospital. All respondents provided informed consent. The two studies were integrated into a single database, with further exclusion of participants with a prior history of CVD and those with missing variables for ASCVD risk assessment, resulting in the inclusion of a total of 6,861 study participants aged 35–74 years. The participant inclusion process is presented in Figure 1.

Both cross-sectional studies were conducted in township or community health centers using the same survey protocol and data collection forms. A survey team of 20 uniformly trained internists, sonographers, nurses, pharmacists, and medical students administered questionnaires, performed physical examinations, and collected blood samples from participants. The questionnaires collected information on age, gender, education level, per capita annual income, smoking status, alcohol intake status, physical activity, self-reported medical history (hypertension and diabetes mellitus), and medication use (glucose-lowering, antihypertensive, and lipid-lowering medications).

Education levels were divided into two groups: primary school and below and junior high school and above. Low income was defined as an annual income of ≤ 20,000 RMB. Current smokers were defined as individuals who smoked at least one cigarette a day for more than 6 months, and current drinkers were defined as those who consumed alcohol at least once a week in the last year. According to the World Health Organization (WHO) 2020 Guidelines on Physical Activity and Sedentary Behaviour (18), physical inactivity was defined as engaging in ≤150 min of moderate-intensity physical activity or <75 min of vigorous-intensity physical activity per week. Insufficient intake of vegetables and fruits was defined as consuming them on fewer than 3 days/month (19). Overweight was defined as BMI = 24.0–27.9 kg/m² and obesity as BMI ≥ 28 kg/m² (20). Hyperuricemia was defined as serum uric acid levels of >420 μmol/L in men and >360 μmol/L in women (21).

Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured using an OMRON HBP-1300 electronic sphygmomanometer. Hypertension was defined as SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg, self-reported hypertension, or self-reported use of antihypertensive medications within the last 2 weeks (22). Fasting blood samples were collected by vein puncture in the morning after at least 8 h of fasting. Total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein (LDL), fasting blood glucose, serum creatinine, and blood uric acid levels were measured. The eGFR was calculated using a modified Chinese version of the Modification of Diet in Renal Disease equation: eGFR = 175 × Scr^−1.234 × age^−0.179 [×0.79 if female], where Scr denotes serum creatinine concentration (mg/dL) and age denotes the age of the participant. Chronic kidney disease was defined as eGFR <  mL/min/1.73 m²) (23). Diabetes mellitus was defined as fasting blood glucose ≥7.0 mmol/L or a self-reported history of diabetes mellitus. Hypercholesterolemia (24) was defined as a self-reported history of dyslipidemia or TC > 5.17 mmol/L.

The TyG index was calculated as lnTG [(mg/dL) × FBG (mg/dL)/2] (11, 24). The China-PAR model was employed as recommended by “the Chinese Guidelines for Cardiovascular Disease Risk Assessment and Management.” The model has been demonstrated to exhibit satisfactory internal consistency and external validity (25). The China-PAR model, incorporating suitable variables, was used to calculate the 10-year ASCVD risk stratification. A risk classification of <5% was considered a low risk, 5%–9% a moderate risk, and ≥10% a high risk. Detailed information of the 11 variables included in the model is provided in the Supplementary Material.

Continuous variables were expressed as means and standard deviations (SDs) for normally distributed data and as medians with interquartile ranges for non-normally distributed data. Categorical variables were presented as frequencies and percentages. The chi-square test, ANOVA, or Kruskal–Wallis rank-sum test was utilized to compare characteristics between groups. Multiple logistic regression models were employed to examine the association between different levels of the TyG index, altitude of residence, and high ASCVD risk. By adding covariates (demographic characteristics, socioeconomic factors, and lifestyle factors), we built a four-model analytical framework. Models 1, 2, and 3 were first constructed to test the independent association of the TyG index with altitude. Model 4 was then constructed for mutual adjustment (protocol details are given in the Supplementary Material). Odds ratios (ORs) and 95% confidence intervals (95%CIs) were calculated for each interquantile range (IQR) increment in the TyG index (0.89) and per 1-SD (631.4 m) in altitude. Categorical exposures were calculated with reference to the lowest level of the TyG index or the first tertile of altitude. Potential non-linear associations were calculated using four-knot restricted cubic spline regression. Subgroup analyses were performed using Model 3, stratified by the TyG index (<8.12, 8.12–8.54, 8.54–9.01, and >9.01) or altitude (<2,000 m, 2,000–2,999 m, and ≥3,000 m). To further explore the existence of a linear trend between the TyG index or altitude and the high risk of ASCVD, we assigned a median value to each group.

The study further investigated the potential interactions between the TyG index and altitude exposure on both additive and multiplicative scales. For additive interaction, we first categorized study participants into four groups based on their TyG index. Combined with altitude levels (three groups), we then created a new term with 12 categories representing 12 combinations (4 × 3) of TyG index exposure levels and altitude levels for testing additive interaction (26). Next, we calculated the relative excess risk due to interaction (RERI) and corresponding confidence intervals (Cls) using participants in the lowest level of TyG index quartile (Q1) and altitude <2,000 m as the reference group. RERI measures the combined excess risk in both exposed groups that arises from their interaction. A RERI score of 0 denotes no additive interaction (the combined excess risk is the sum of their individual excess risks, calculated as OR-1). A RERI greater than 0 indicates a positive interaction (meaning that the combined excess risk is more than the sum of their individual excess risks), while a RERI less than 0 implies a negative interaction (meaning that the combined excess risk was less than the sum of their individual excess risks. Therefore, a positive RERI value in the present study indicates that residential altitude amplifies the ASCVD risk associated with the TyG index. On the multiplicative scale (27), we added a product term between the TyG index and altitude levels. Likelihood tests were employed to assess the significance of this interaction by comparing the model with and without the interaction term. A p-value of less than 0.05 for the interaction term was considered indicative of a multiplicative interaction.

(1) We excluded participants who were treated with glucose-lowering medications. (2) We also excluded participants treated with lipid-lowering medications. (3) We calculated E-values to evaluate the potential impact of unmeasured confounders on conclusions in observational studies (28, 29). Details can be found at https://www.evalue-calculator.com/.

All analyses were performed using R version 4.3.0. A two-sided p value of less than 0.05 was considered statistically significant.

A total of 6,861 participants were included in the study. The mean age of participants was 52.2 ± 10.5 years and 3,191 (46.5%) were men. After assessment using the China-PAR model, the mean 10-year ASCVD risk for the 6,861 study participants was 2.7 (1.1,5.7), of which 559 (8.1%) were classified as high risk, 1,484 (21.6%) as medium risk. The characteristics of participants stratified by ASCVD risk scores are presented in Table 1. Participants in the high-risk group were predominantly Han Chinese, male, current smokers, and exhibited higher rates of alcohol consumption and physical inactivity. They also tended to have lower education and lower income levels. Additionally, this group showed a higher prevalence of hypertension, diabetes mellitus, chronic kidney disease, and hypercholesterolemia, as along with higher levels of SBP, DBP, FBG, and TC. The distribution of ASCVD risk across participants with TyG and altitude level changes is presented in Supplementary Material Figure S2 and Supplementary Table S1.

Table 2 shows the associations between the TyG index level and a high risk of ASCVD. A higher TyG index level was associated with an increased risk of high ASCVD. These associations remained robust after adjustment for various covariates, including altitude level. In Model 4, compared with the Q1 TyG index level, the ORs (95% CIs) for high ASCVD risk were 1.04 (95% CI: 0.79,1.38) for Q2, 1.43 (95% CI: 1.10,1.87) for Q3, and 2.19 (95% CI: 1.67,2.82) for Q4. The non-linear associations between the TyG index and high risk of ASCVD are shown in Figure 2. In subgroup analyses, the positive association between the TyG index level and high risk of ASCVD remained stable across participants residing at different altitude levels (Table 3).

Altitude was also positively associated with a higher risk of ASCVD, and this correlation remained unchanged after adjusting for the TyG index and other covariates (Table 2). Every 1-SD increase in altitude was associated with a 46.0% higher risk of ASCVD. In the tertile-based analysis, compared with participants residing at altitudes <2000 m, the ORs for a high risk of ASCVD were 1.51 (95% CI: 1.23,1.85) and 2.86 (95% CI: 1.97,3.41) for those living at 2,000–2,900 and ≥3,000 m (Table 2), respectively. The non-linear associations between altitude and a high risk of ASCVD are presented in Figure 2. The positive association persisted across subgroups with different TyG index levels, although some results were statistically non-significant (Table 4).

Table 5 presents the combined health impact of TyG index and altitude levels on ASCVD risk. Participants with higher TyG index levels and high altitude exposure generally exhibited a higher risk of ASCVD. Using participants with low TyG index levels and low altitude exposure as the reference group, those with high TyG index levels and high altitude exposure demonstrated the highest risk of ASCVD (OR: 4.95,95% CI: 2.84,8.63). The results of the interaction between TyG index levels and altitude levels on the high risk of ASCVD are presented in Table 5. On the additive scale, little evidence of interaction was found. Similar results were also observed on the multiplicative scale, with p-values >0.05 for the interaction term (Supplementary Material Table S2).

Subgroup analysis indicated that the association between the TyG index, altitude, and ASCVD risk remained consistent, but the effects were differential across different subpopulations and no significant interactions were observed (Supplementary Material Tables S3, S4). In sensitivity analyses, the associations of TyG index levels and altitude levels with high risk of ASCVD were not materially altered after excluding participants who received treatment with lipid-lowering medications or glucose-lowering medications (Supplementary Material Tables S5, S6). The E-values demonstrated that the associations of TyG index levels and altitude levels with a high risk of ASCVD were robust (Supplementary Material Table S7).