Study participants were drawn from the China Health and Retirement Longitudinal Study (CHARLS), which is a nationally representative longitudinal survey of residents in rural and urban areas of China of ≥45 years old that commenced in 2011 and their spouse. Detailed information regarding the study design and enrollment criteria have been previously reported (19). The study consists of five waves of surveys conducted between 2011 and 2020, with participants recruited from 23 Chinese provinces using a multistage stratified probability proportional-to-size sampling strategy. The national baseline survey was conducted in 2011, enrolling 17,708 participants, followed by four subsequent waves: Wave 2 (2013–2014), Wave 3 (2015–2016), Wave 4 (2017–2018), and Wave 5 (2019–2020). We applied data from wave 1–5, which are available online at http://charls.pku.edu.cn. The inclusion criteria encompassed individuals aged 18 years or older who had completed follow-up data, including complete data on CMM and METS-IR. The exclusion criteria included: (1) participants with CMM in Wave 1; (2) individuals without complete data on METS-IR at baseline; and (3) Missing data on CMM at baseline or lost to follow-up. Ultimately, a total of 6,977 individuals met the eligibility criteria for subsequent analysis (Figure 1). The original CHARLS study was approved by the Ethical Review Committee of Peking University (IRB00001052-11015), and written informed consent was obtained from all participants at the time of enrollment.

CMM events were defined as the simultaneous presence of at least two cardiometabolic diseases, including diabetes, heart disease and stroke. The diagnoses of heart disease and stroke were confirmed through self-report of a physician’s diagnosis obtained via the questionnaire survey, including “Have you been told by a doctor that you have been diagnosed with a heart attack, Angina, coronary heart disease, heart failure, or other heart problems?” or “Have you been told by a doctor that you have been diagnosed with a stroke?” (20). In addition to self-reporting diabetes, participants were diagnosed with diabetes if they met any of the following criteria: (1) fasting plasma glucose ≥ 7.0 mmol/L; (2) random plasma glucose ≥ 11.1 mmol/L; or (3) HbA1c ≥ 6.5% according to the American Diabetes Association criteria (21). The incidence of CMM was determined at the time of diagnosis of the second CMD, at which point the individuals presented with two distinct types of CMD.

In this study, extensive baseline information was collected by trained.

interviewers following standard procedures. Anthropometric variables including the height (m) and weight (kg) were measured by trained medical staff according to standard protocol. The body mass index (BMI) was calculated as weight divided by height squared (kg/m²). After at least 5 min spent sitting down, the blood pressure of the participants was measured three times by a trained interviewer at 45-s intervals using a digital sphygmomanometer (Omron TM HEM-7200). Blood samples were collected after an overnight fast, then stored at −20°C and transported to Beijing for further measurements. The biochemical parameters included white blood cell count (WBC), platelet count (PLT), hemoglobin, fasting blood glucose (FBG), uric acid (UA), serum creatinine, Cystatin C, high-sensitivity C-reactive protein (hsCRP), and lipid profiles. The METS-IR were calculated according to the following formula: METS-IR = ln [2 × FPG (mg/dL) + fasting serum TG (mg/dL)] × BMI (kg/m²)/ln [HDL − C (mg/dL)] (22).

All participants provided medical history and lifestyle information during face-to-face interviews by trained interviewers. A structured questionnaire was employed to collect the baseline data, including age, gender, hypertension, marital status (married and others), residence (rural, urban), education (elementary school and below, secondary school, and college and above), smoking status, and drinking status. Smoking status was categorized as “never smoking,” “current smoker” and “former smoker.” Participants were diagnosed with hypertension if their systolic blood pressure was ≥140 mmHg, diastolic blood pressure was ≥90 mmHg, or they had a self-reported history of hypertension, or were currently using antihypertensive medications.

Continuous variables were presented as the mean ± standard deviation or the median with interquartile range, depending on the normality of the data distribution. Categorical variables were presented as frequency and percentage. For comparison of variables between CMM group and non-CMM group, independent t-test was used for normally distributed data and Man-Whitney test were for skewed data. One-way analyses of variance (for normally distributed data), Kruskal-Wallis tests (for skewed data), and chi-square tests (for categorical data) were performed to assess the differences between more than two groups. In our study, the population was divided into four groups according to the quartiles of baseline METS-IR as follows: Q1 (<29.6), Q2 (≥29.6, <34.2), Q3 (≥34.2, <40.0), and Q4 (≥40.0). The Kaplan–Meier analysis was used to assess the cumulative incidence of CMM among METS-IR quartiles, and the differences of curves were examined utilizing the log-rank test. Cox proportional hazards regression was used to estimate the hazard ratios (HRs) and 95% confidence intervals (95% CIs) between METS-IR quartiles and incidence of CMM. Prior to conducting the Cox regression model, we evaluated the proportional hazards assumption using Schoenfeld residuals and found no potential violations. The dose–response relationship between METS-IR and the risk of developing CMM was investigated through restricted cubic splines (RCS) analysis. Subgroup analyses were performed to explore whether the association between METS-IR with CMM differed by gender, age, hypertension, residence, marital status, education, smoking status, and drinking status. Receiver operating characteristic (ROC) curve analysis was used to evaluate the predictive ability of variables for CMM and Delong’s test was used to compare the differences between variables.

All statistical analyses in the present study were performed using RStudio 4.2.1 software, SPSS 26, and an online statistical analysis platform called I STATISTICS.¹ All comparisons were two-sided, and a significance level of p < 0.05 was considered statistically significant.

Figure 1 depicts the overall methodological workflow of the present study. A total of 6,977 participants were included for analysis according to the inclusion and exclusion criteria. During a median 109 (108–109) months of follow-up, 745 (10.7%) participants were identified with new-onset CMM. Compared to individuals without CMM, those with CMM were older, had a higher proportion of females, and exhibited elevated levels of systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, WBC, PLT, hemoglobin, FBG, total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), UA, hsCRP, METS-IR, as well as a higher incidence of hypertension, basal diabetes, basal stroke, and basal heart disease (p < 0.05) (Table 1). Conversely, individuals in this group had a lower prevalence of current smoking and drinking and exhibited reduced level of high-density lipoprotein cholesterol (HDL-C) (p < 0.05).

We also compared the baseline characteristics of the included and lost individuals. Supplementary Table S1 showed that participants who were lost to follow-up were older, more likely to be male and live in urban, less likely to be married and educated, exhibited higher levels of SBP, DBP, FBG, UA, serum creatinine, hsCRP, and had higher prevalence of basal stroke, and basal heart disease (p < 0.05). Conversely, individuals in this group had a lower prevalence of current smoking and drinking, and exhibited reduced level of BMI and hemoglobin (p < 0.05).

The participants were divided into four groups based on the quartiles of baseline METS-IR (Table 2). The incidence of CMM in quartiles Q1, Q2, Q3, and Q4 were 4.99, 7.51, 10.67, and 19.54%, respectively. Participants in the higher METS-IR quantile groups were younger, more likely to be female, and exhibited higher levels of SBP, DBP, BMI, WBC, PLT, hemoglobin, FBG, TC, TG, LDL-C, UA, and hsCRP (p < 0.05). Additionally, individuals in the higher METS-IR quantile groups were more likely to reside in urban areas and to be married, while also exhibiting higher levels of education and a greater proportion of hypertension, basal diabetes, and basal heart disease (p < 0.05). In contrast, the levels of HDL-C, Cystatin C, and the prevalence of current smoking and drinking were lower in the higher METS-IR quantile groups (p < 0.05).

Kaplan–Meier survival curve was utilized to analyze the cumulative incidence of CMM. Figure 2 illustrates that the cumulative incidence of CMM increased gradually with increasing METS-IR across the quartiles (log-rank = 223, p < 0.001). Subsequently, we performed multicollinearity analysis and determined that the variance inflation factor (VIF) for all variables (including METS-IR, age, gender, hypertension, smoking status, drinking status, rural residence, marital status, education, SBP, DBP, WBC, PLT, hemoglobin, serum creatinine, and hsCRP) were less than 5 (Supplementary Table S2), indicating the absence of collinearity among these variables. We assessed the proportional hazards assumption using the Schoenfeld residual test and found no violations (p > 0.05). Table 3 summarizes the association between METS-IR quantile and the risk of CMM using Cox proportional hazards regression model analysis. Model 1 was an unadjusted model; Model 2 adjusted for age, gender; Model 3 further adjusted for hypertension, smoking status, drinking status, rural residence, marital status, education, SBP, DBP, WBC, PLT, hemoglobin, serum creatinine, and hsCRP. Among participants without CMM at baseline, multi-adjusted Cox regression models revealed that the HRs and 95% CIs of CMM were 1.52 (1.15–2.00) for participants in Q2, 2.02 (1.56–2.63) for those in Q3, and 3.61 (2.80–4.64) for those in Q4, compared to participants in Q1 (Table 3). We subsequently categorized the population into four groups based on the types of CMD: individuals without any CMDs at baseline, those with heart disease at baseline, those with diabetes at baseline, and those with stroke at baseline. Table 3 showed that METS-IR quantiles were significantly associated with the risk of CMM, regardless of adjustment for other confounding factors in the group without any CMDs at baseline, consistent with the results obtained prior to grouping. Additionally, a significant correlation between METS-IR quantile and CMM was also observed in groups with heart disease (Q4 vs. Q1: HR = 2.54, 95% CI = 1.55–4.14, p < 0.001; Q3 vs. Q1: HR = 1.82, 95% CI = 1.12–2.96, p = 0.016) and diabetes (Q4 vs. Q1: HR = 2.21, 95% CI = 1.17–4.19, p = 0.015) at baseline.

Restricted cubic spline is a flexible and powerful method for data fitting, especially when dealing with time-varying covariant effects or non-proportional risks. Conversely, fractional polynomials may be overly restrictive, often assuming linear or logarithmic time relationships, which might oversimplify and fail to capture complex data variations. Therefore, we performed RCS to model and visualize the relationship between METS-IR and the risk of CMM. Figure 3 presented a positive nonlinear association between METS-IR and the risk of CMM, even after adjusting for confounding covariates including age, gender, hypertension, smoking status, drinking status, rural residence, marital status, education, SBP, DBP, WBC, PLT, hemoglobin, serum creatinine, and hsCRP (nonlinear p < 0.05).

We performed weighted interaction tests and subgroup analyses to assess whether the association between METS-IR and the risk of developing CMM differs across different subgroups. The result showed that individuals in quartiles Q3 and Q4 were associated with the risk of developing CMM across all subgroups, except for individuals who are not married and former smokers (Figure 4), when compared to those in quartiles Q1. Additionally, the difference in the risk of developing CMM between Q2 and Q1 was statistically significant in subgroups of male, individuals with age ≥ 60 years, without hypertension, be married, urban residence, with primary education, current smoking and drinking (p < 0.05). Furthermore, we observed no significant interactive effect between these subgroups and METS-IR on the risk of CMM in Cox proportional hazards regression analysis (p for interaction > 0.05).

Previous studies have identified several other alternative indicators of non-insulin–based IR, including the ratio of TG to HDL-C (TG/HDL-C) and triglyceride glucose index (TyG index) (23, 24). In this study, we conducted ROC analysis to compare the predictive abilities of TG/HDL-C, TyG index, and METS-IR. As shown in Table 4 and Figure 5, the average AUCs for TG/HDL-C, TyG index, and METS-IR were 0.611 (95% CI = 0.600–0.623), 0.633 (95% CI = 0.621–0.644), and 0.669 (95% CI = 0.657–0.680), respectively. The METS-IR demonstrated significantly higher AUCs for predicting the risk of CMM compared to TG/HDL-C (p < 0.001) and TyG index (p < 0.001).