The Kailuan Study (Registration Number: ChiCTR-TNRC-11001489) is a prospective cohort study based on a community population that investigates cardiovascular diseases and related risk factors through surveys and interventions. The study details can be found in previous publications by our research group (28, 29). In 2006, Kailuan General Hospital and its ten affiliated hospitals conducted initial health examinations for both active and retired employees of the Kailuan Group, with a total of 101,510 participants. Subsequent follow-ups were conducted every two years, maintaining consistency in follow-up content, anthropometric measurements, and biochemical indicators with the initial health examination. This study focused on the metabolically healthy obese (MHO) population, which consisted of obese individuals who participated in the health examinations of the Kailuan Group from 2006 to 2010 and did not have a history of hypertension, diabetes, hyperlipidemia, or cardiovascular disease. The obesity cutoff (BMI ≥28 kg/m²) was defined based on ethnic-specific thresholds accounting for higher visceral adiposity in Asian populations (WHO-recommended BMI ≥27.5 kg/m² for Asians) (30, 31), in alignment with the Chinese Adult Obesity Guidelines (32), while ensuring consistency with previous epidemiological research (33, 34). The specific criteria for inclusion were meeting the obesity criteria (BMI≥28 kg/m²) along with the following: serum total cholesterol (TC) <5.72 mmol/L; serum triglycerides (TG) ≤1.7 mmol/L; serum high-density lipoprotein cholesterol (HDL-C) >1.0 mmol/L (males) or >1.3 mmol/L (females); systolic blood pressure (SBP) <140 mmHg and diastolic blood pressure (DBP) <90 mmHg; fasting plasma glucose (FPG) <7 mmol/L; no history of taking antihypertensive, antidiabetic, or lipid-lowering medications; and no history of CVD (including coronary heart disease, ischemic and hemorrhagic stroke, heart failure, atrial fibrillation) (12, 17). Individuals who did not meet these criteria were excluded. A total of 4,750 participants were ultimately included, and follow-up commenced in 2010 and concluded on December 31, 2021. The inclusion and exclusion processes are illustrated in Figure 1 . The study was performed in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Kailuan General Hospital (Approval Number: 2006-05). All participants provided written informed consent after receiving detailed explanations about the study objectives, data collection procedures, and their right to withdraw at any time without consequences to medical care. This consent process followed the standardized protocol established in the Kailuan cohort, as described in our previous publications (35, 36).

This study used the epidemiological survey content and anthropometric indicators mentioned in previously published articles (28, 37). All measurements were conducted in a quiet room with the temperature controlled between 22°C and 25°C. Each participant completed a face-to-face questionnaire during each health examination, collecting information on demographic characteristics (sex, age, education level, BMI), personal health history (hypertension, diabetes, cardiovascular disease, and the use of antihypertensive, glucose-lowering drugs, and lipid-lowering medications), and lifestyle factors (smoking status, alcohol consumption habits, Physical Activity habits), among others. For anthropometric measurements, the participants were instructed to wear lightweight clothing and remain barefoot. Trained doctors measured weight and height under standardized conditions according to a standardized protocol, with units standardized to 0.1 kilograms and 0.1 centimeters. Body mass index (BMI) was calculated as weight (kg) divided by height in meters squared (m²). During blood pressure measurements, following a 5-minute rest for participants, calibrated mercury sphygmomanometers were used to measure the blood pressure of each participant’s right upper arm. Measurements were repeated at intervals of 1–2 minutes a total of three blood pressure readings. The average of three readings was recorded (38).

All participants were required to fast for 8–12 h, and fasting venous blood samples were obtained after centrifugation to collect the upper serum layer for measurement of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), fasting plasma glucose (FPG), high-sensitivity C-reactive protein (hs-CRP), and serum uric acid (UA) (37). Biochemical indicators were measured using a Hitachi 7600 automatic biochemical analyzer. The procedures strictly followed the instructions provided with the reagent kits, and the analyses were performed by professional laboratory technicians.

The TyG index was calculated using the established formula: ln[TG(mg/dL) × FPG(mg/dL)/2] (37), and it was divided into four groups based on quartiles: Q1 group (<8.18); Q2 group (8.18-8.41); Q3 group (8.42-8.62); Q4 group (≥8.63).

CVD includes coronary heart disease, ischemic stroke, hemorrhagic stroke, heart failure, and atrial fibrillation (39). Hypertension is defined as systolic blood pressure (SBP) ≥ 140mmHg and/or diastolic blood pressure (DBP) ≥ 90mmHg, or the use of antihypertensive medications, or a history of hypertension (40). Diabetes was defined as fasting plasma glucose (FPG) ≥ 7.0 mmol/L, prolonged use of antidiabetic medications, or a history of a confirmed diabetes diagnosis (41). Hyperlipidemia is defined as total cholesterol (TC) > 5.72 mmol/L or triglycerides (TG) > 1.7 mmol/L, or the prolonged use of statins, niacin, or fibric acid derivatives (42). Participants who smoked an average of ≥1 cigarette per day in the past year were classified as smokers; otherwise, they were considered non-smokers. Participants who had consumed an average of ≥100 mL of alcohol per day (alcohol content >50%) in the past year were classified as drinkers; otherwise, they were considered nondrinkers. Participants engaging in physical exercise ≥3 times per week, with each session lasting ≥30 min, were classified as active exercisers; otherwise, they were classified as non-active exercisers. Educational level was categorized as below high school and high school or above based on the level of education attained (43).

CMD is defined as the occurrence of one or more of the following: diabetes, coronary heart disease, or stroke (3, 4). Using the health examination data from 2010 as the starting point for follow-up and CMD occurrence as the endpoint event, in the case of multiple subtypes of CMD occurring, the subtype of the first event was recorded; for those who did not experience an endpoint event, the follow-up was concluded on December 31, 2021. CMD occurrence was assessed annually by trained medical personnel who reviewed the hospital diagnoses of the study participants in the hospitals affiliated with the Kailuan Group and the designated medical insurance hospitals in Tangshan. All diagnoses were confirmed by professional physicians based on the hospital records.

This study utilized SAS 9.4 statistical software (SAS Institute, Inc., Cary, NC, USA) was used for data analysis. In this study, the missingness of baseline covariates was less than 5%. We used Multiple Imputation by Chained Equations (MICE) to handle missing values for subsequent analyses (44). Metric data with a normal distribution were expressed as mean ± standard deviation, and intergroup data were analyzed using one-way analysis of variance. Skewed distribution metric data were represented as medians and quartiles, and intergroup data were analyzed using non-parametric tests (Kruskal-Wallis). Count data were presented as relative numbers or percentages (%), and intergroup comparisons were made using the chi-square test, which was divided into four groups based on quartiles. Prior to conducting Cox analysis, we performed a test for the proportional hazards (PH) assumption (45). The results indicated that the p-value for the PH assumption test was 0.275 (>0.05), suggesting that the proportional hazards assumption was met. The Kaplan–Meier method was used to calculate the incidence rates of CMD and its subtypes (coronary heart disease, stroke, and diabetes) for each group, with comparisons made using the log-rank test. After meeting the proportional hazards assumption, Cox proportional hazards regression models were used. The risk ratio (HR) and 95% confidence interval (95% CI) were calculated to assess the relationship between the TyG index and risk of CMD and its subtypes. Model 1 was unadjusted for other variables, Model 2 adjusted for age (continuous variable, years) and sex (categorical variable, male or female); Model 3, based on Model 2, further adjusted for smoking status (categorical variable, yes or no), drinking habits (categorical variable, yes or no), physical activity habits (categorical variable, active or inactive), education level (categorical variable, senior high school or above), and family history of CMD (categorical variable, present or absent); Model 4, based on Model 3, further adjusted for SBP (continuous variable, mmHg), FPG (continuous variable, mmol/L), HDL-C (continuous variable, mmol/L), LDL-C (continuous variable, mmol/L), hs-CRP (continuous variable, mg/L), BMI (continuous variable, kg/m²), UA (continuous variable, μmol/L), and estimated glomerular filtration rate (eGFR, continuous variable, ml/min/1.73m²). After adjusting for potential confounding factors, restrictive cubic splines were used to describe the dose-response relationship between TyG index (continuous variables) and the risk of CMD.Subsequently, stratified analyses were conducted based on age (<45 years, ≥45 years), sex (male, female), physical activity habits (active, inactive) and sex-specific median height (above-median, below-median). To ensure the robustness of the results, individuals who developed CMD within 2 years and those with BMI < 30 kg/m² were excluded from the sensitivity analysis of the relationship between the TyG index and risk of CMD. We additionally included waist circumference (≥90 cm for men/≥85 cm for women) as an alternative indicator of central obesity (data from the 2010 follow-up) for sensitivity analysis. Statistical significance was set at P < 0.05 (two-tailed) was considered statistically significant.

This study included a total of 4,750 MHO participants as the observational population, with an average age of 53.70 ± 12.93 years, and a male proportion of 75.0%. The baseline clinical and biochemical characteristics of the MHO participants, based on the quartiles of the TyG index, are shown in Table 1 . Stratified by quartiles of the TyG index, compared to the Q1-Q3 groups, the Q4 group had the highest proportion of male participants, highest values for BMI, SBP, DBP, and heart rate indicators, and the poorest metabolic profile (TC, HDL-C, LDL-C, TG, FPG, UA, and hs-CRP). Additionally, the Q4 group exhibited the highest proportion of unhealthy lifestyle habits (smoking, drinking, and lack of exercise) (P < 0.01) ( Table 1 ).

During a median follow-up period of 11.4 years (IQR, 10.3, 11.2), 826 (17.39%) MHO participants developed CMD. Among them, 131 MHO participants experienced coronary heart disease, 215 participants had a stroke, and 542 participants developed diabetes. The cumulative incidence rates of CMD subtypes in different TyG quartiles were compared using the log-rank test, which showed statistically significant differences (P < 0.01; Figure 2 ). The incidence density of CMD and its subtypes (coronary heart disease, stroke, and diabetes) exhibited an increasing trend across TyG quartiles. The incidence densities (per 1000 person-years) of CMD in the Q1-Q4 groups were 19.02, 26.13, 27.70, and 45.86, respectively. For coronary heart disease, the densities were 3.36, 3.48, 4.64, and 7.22; for stroke, 4.89, 6.89, 6.99, and 9.78; for diabetes, 11.11, 16.82, 18.43, and 31.27. After adjusting for confounding factors in Model 4, compared to the Q1 group of TyG quartiles, the adjusted HR values (95% CI) for CMD in the Q2-Q4 groups were 1.33 (1.03, 1.65), 1.37 (1.04, 1.82), and 2.04 (1.56, 2.68) (P < 0.0001). The adjusted HR values (95% CI) for coronary heart disease were 1.16 (0.88, 1.53), 1.23 (0.92, 1.63), and 1.78 (1.36, 2.33), respectively (P < 0.0001). The adjusted HR values (95% CI) for stroke were 1.17 (0.89, 1.55), 1.23 (0.93, 1.64), and 1.75 (1.34, 2.29) (P < 0.0001). The adjusted HR values (95% CI) for diabetes were 1.66 (1.18, 2.34), 1.74 (1.23, 2.46), and 2.59 (1.85, 3.63) (P < 0.0001). In model 4, for each SD increased, the HR (95% CI) for CMD, coronary heart disease, stroke, and diabetes mellitus was 1.41 (1.27,1.57), 1.35 (1.22,1.50), 1.36 (1.22,1.51) and 1.51 (1.34,1.70), respectively ( Table 2 ). Additionally, restrictive cubic spline analysis revealed a linear dose-response relationship between the TyG index and risk of CMD ( Figure 3 ).

Stratified analysis revealed an interaction between the TyG index and sex, whereas no interactions were found with age, physical activity, and height. Across different age, sex, physical activity, and sex-specific median height subgroups, higher TyG index levels were consistently associated with an increased risk of CMD. Moreover, the risk of CMD in females was higher than that in males. However, in the physically active group, the association between the TyG index and CMD did not show statistically significant differences among the subgroups ( Table 3 ).

To eliminate potential confounding effects on the study results, 132 individuals who experienced CMD events within the first 2 years of follow-up (132 individuals) were excluded, and the results were consistent with the main findings ( Supplementary Table S1 ). Additionally, sensitivity analysis was performed by repeating the main Cox regression analysis with a focus on individuals with BMI ≥ 30 kg/m², and the results were consistent with the main findings ( Supplementary Table S2 ). Waist circumference, as an alternative indicator of MHO to BMI, showed a consistent association pattern with TyG-CMD compared to BMI ( Supplementary Table S4 ).