The modified cardiometabolic index as a simple screening tool for cardiovascular disease risk: a cohort study

Abstract

Cardiovascular disease (CVD) remains a pressing global health issue that poses substantial threats to public well-being. The modified cardiometabolic index (MCMI), a newly developed composite indicator linked to insulin resistance (IR) and central obesity, has not been extensively explored in terms of its association with the occurrence of CVD. To fill this research gap, we carried out a large-scale retrospective cohort study to examine how MCMI is associated with the onset of CVD. Additionally, we compared MCMI with the triglyceride-glucose (TyG) index to evaluate the ability of both indices to predict CVD risk. This retrospective cohort study drew data from the China Health and Retirement Longitudinal Study (CHARLS). It included participants who were 45 years of age or older, with baseline assessments conducted in 2011 and follow-up evaluations in 2020. The diagnosis of CVD was based on self-report. To explore the association between MCMI and the incidence of CVD, we used logistic regression models, restricted cubic splines, and subgroup analyses. Moreover, receiver operating characteristic (ROC) analysis was performed to assess the predictive performance of both MCMI and the TyG index for CVD occurrence. Among the 6,117 participants in the total population, 1,298 eventually developed CVD. MCMI was identified as an independent risk factor for CVD development, with an adjusted odds ratio of 1.18 [95% confidence interval (CI): 1.08–1.29]. A J-shaped non-linear association between MCMI and the incidence of CVD was also discovered. Subgroup analyses showed that MCMI could stably predict CVD. In the comparison of predictive abilities, MCMI demonstrated a statistically superior but still modest predictive power compared to the TyG Index, as evidenced by area under the curve (AUC) values of 0.581 for MCMI and 0.559 for the TyG index (p < 0.001). This study indicated that the MCMI may serve as a preliminary screening indicator for cardiovascular disease risk in middle-aged and elderly populations. Although its independent predictive performance is limited, its simplicity and cost-effectiveness render it potentially useful in primary care settings and large-scale screening initiatives.

Introduction

Cardiovascular disease (CVD) represents a significant public health challenge globally, characterized by rising incidence and mortality rates. Between 1990 and 2019, the number of new cases of CVD increased sharply by 77.12%, from 31.31 million to 55.45 million. Similarly, during this period, the mortality rate attributable to CVD rose by 53.81%, escalating from 12.07 million to 18.56 million¹. It is estimated that in 2019, approximately 330 million people in China were affected by CVD, with CVD accounting for 46.74% and 44.26% of deaths in rural and urban areas, respectively². The American College of Cardiology has emphasized the importance of primary prevention of CVD³. Therefore, it is essential to develop a clinically applicable and cost-effective indicator to identify high-risk populations for CVD and facilitate early intervention.

Insulin resistance (IR) is closely related to the occurrence of CVD^4–6. Some composite biomarkers have been confirmed to be associated with IR, such as the triglyceride-glucose (TyG) index^7,8 and cardiometabolic index (CMI)^9,10. The CMI was calculated by multiplying the waist circumference (WC) to height ratio by the triglyceride (TG) to high-density lipoprotein cholesterol (HDL-C) ratio¹¹. Building on the CMI, Guo et al. proposed the concept of a modified CMI (MCMI) by incorporating blood glucose levels¹². The potential advantage of the MCMI lies in its integration of multiple risk pathways—glucose, lipid, and obesity metrics—leading us to hypothesize that it may outperform existing partial indicators in predicting CVD. To our knowledge, no studies to date have evaluated MCMI for CVD prediction. Given the critical need for simple, cost-effective tools for early CVD risk identification in primary care and large-scale population screening, this study seeks to investigate whether the MCMI can serve as a practical preliminary screening indicator for CVD risk in middle-aged and elderly populations.

Materials and methods

Data source and study population

This cohort study used data from adult participants in the China Health and Retirement Longitudinal Study (CHARLS) database, a national population-based study that started in 2011 and focuses on Chinese adults aged 45 and older (http://charls.pku.edu.cn/). The first survey was conducted in 2011, and follow-up assessments have been carried out every two years, resulting in a total of five rounds of data collection by 2020. Participants were selected using a multi-stage, stratified probability sampling method, which ensured that they were proportionally representative of 150 counties or districts in 28 provinces of China. The initial survey had a high response rate of 80.5%¹³.

The study process is shown in Fig. 1. The exclusion criteria were as follows: (1) individuals who already had CVD at the baseline (n = 2130), (2) individuals with a history of stroke (n = 372), (3) lack of data related to heart disease and stroke (n = 251), (4) participants who were not in a fasting state (n = 5983), (5) missing TG data (n = 138), (6) no fasting blood glucose (FBG) measurements (n = 15), (7) no WC data (n = 1273), (8) missing height information (n = 40), and (9) loss to follow-up (n = 1389). Eventually, a total of 6,117 individuals participated in this research.

Fig. 1.

The flowchart of this study.

The CHARLS study was conducted in accordance with the principles of the Declaration of Helsinki and received approval from the Institutional Review Board at Peking University (IRB00001052–11015). All participants provided written informed consent before participating in the study. In addition, the research followed the STROBE guidelines for reporting observational studies in epidemiology.

MCMI and TyG index assessment^12,14

Assessment of endpoint events

All participants were followed up in 2020, and the outcome event was the occurrence of CVD, which included both heart disease and stroke. The diagnoses of heart disease and stroke were made using standardized questions based on previous research. Participants were asked, “Has a doctor ever told you that you had a heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems?” For stroke, the assessment was based on the question, “Has a doctor ever informed you that you had a stroke?” A “Yes” response was considered to indicate the occurrence of the outcome event.

Data collection

The data included in this study encompassed various demographic and health-related variables, specifically age, gender, WC, height, current smoking status, current alcohol consumption, residential area, education level, comorbidities, the TyG index, and the MCMI. Residential areas were classified as either urban or rural, while education levels were categorized into four groups: below primary school, primary school, middle school, and high school or above. All comorbidities were identified based on self-reported histories provided by the participants, which included conditions such as hypertension, Type 2 diabetes mellitus (T2DM), cancer, lung disease, psychiatric disorders, arthritis, dyslipidemia, hepatic disease, kidney disease, digestive system disorders, asthma, and memory-related diseases.

Statistical analysis

The Shapiro - Wilk test was used to assess the normality of the data. The results showed that the continuous variables were not normally distributed, so they were reported as medians (Q1, Q3). Categorical data were presented as counts and percentages [n (%)]. Patients were excluded from the study due to missing values in the primary research variables, MCMI and TyG index. For the remaining variables, where the proportion of missing data was low (at most 2.09%), multiple imputation was employed to handle these gaps.

Univariate and multivariate logistic regression models were used to examine the association between the TyG index, MCMI, and the risk of CVD. In these models, both the TyG index and MCMI were treated as continuous variables and also categorized into quartiles. The models were structured as follows: Model 1 was unadjusted; Model 2 was adjusted for age, gender, smoking, and alcohol consumption; and Model 3 was further adjusted for residential area, education level, and all comorbidities.

Restricted Cubic Spline (RCS) models were established to more intuitively show the associations between the TyG index, MCMI, and the occurrence of CVD. These models were built based on logistic regression analysis, and the Bayesian Information Criterion was used for model selection. The Likelihood Ratio Test was applied to evaluate whether there were non-linear associations in the RCS models. Moreover, the adjustment factors used in these three RCS models were the same as those used in the corresponding logistic regression models.

The predictive abilities of the TyG index, CMI, and MCMI for CVD risk were compared by constructing Receiver Operating Characteristic (ROC) curves. The predictive performance of Models 1, 2, and 3 was also evaluated using ROC analysis. The statistical significance of the differences in Areas Under the Curve (AUCs) among these three indices was assessed using DeLong’s test.

Subgroup analysis was conducted to further evaluate the predictive ability of the TyG index and MCMI for CVD risk in different populations. Tyg index and MCMI were divided into high and low groups based on the optimal cutoff value from the ROC curve. Interaction analysis was performed to verify the heterogeneity of these associations across key subgroups. The variables considered in the subgroup analysis and interaction analysis included gender, hypertension, T2DM, alcohol consumption, smoking, and age.

This study used R software (version 4.3.0) and MedCalc (version 20.215) for statistical analysis. All tests were two-sided, and a p-value < 0.05 was considered statistically significant.

Results

Participant characteristics

The baseline characteristics of the 6,117 study participants are summarized in Table 1. The median age of the cohort was 57 years, with a slightly higher proportion of females (55.84%). Most participants resided in rural areas (67.14%) and nearly half had an education level below primary school (47.79%). Regarding lifestyle factors, the majority were non-smokers (70.49%) and non-drinkers (66.44%). Hypertension (21.61%) and arthritis (32.68%) were the most prevalent chronic conditions among the population, while other diseases such as cancer (0.62%) and psychiatric disorders (0.97%) were less common. The median TyG index and MCMI were 8.55 and 2.84, respectively.

Table 1.

Basic characteristics of the population included in this study.

Characteristics	Median (Q1, Q3) or n (%)
Age, years	57.00 (50.00, 63.00)
Gender, n (%)
Female	3416 (55.84)
Male	2701 (44.16)
Waist, cm	84.00 (77.20, 91.20)
Height, cm	157.40 (151.90, 163.90)
Smoking, n (%)
No	4306 (70.49)
Yes	1803 (29.51)
Drinking, n (%)
No	4064 (66.44)
Yes	2053 (33.56)
Residential area, n (%)
Urban	2010 (32.86)
Rural	4107 (67.14)
Education Level, n (%)
Below Primary School	2922 (47.79)
Primary School	1321 (21.61)
Middle School	1259 (20.59)
High School and above	612 (10.01)
Hypertension, n (%)	1315 (21.61)
T2DM, n (%)	278 (4.59)
Cancer, n (%)	38 (0.62)
Lung disease, n (%)	458 (7.50)
Psychiatric disorders, n (%)	59 (0.97)
Arthritis, n (%)	1996 (32.68)
Dyslipidemia, n (%)	461 (7.70)
Hepatic disease, n (%)	172 (2.82)
Kidney disease, n (%)	273 (4.48)
Digestive disease, n (%)	1305 (21.38)
Asthma, n (%)	201 (3.30)
Memory-related disease, n (%)	49 (0.80)
TyG index	8.55 (8.20, 8.98)
MCMI	2.84 (2.43, 3.33)

Data are presented as Median (Quartile 1, Quartile 3) or n (%).

CVD cardiovascular disease, T2DM type 2 diabetes mellitus, TyG index triglyceride-glucose index, MCMI modified cardiometabolic index.

Logistic regression models for assessing the association between TyG index, MCMI, and CVD

TyG was categorized into quartiles based on their TyG values as follows: Q1 ≤ 8.199, 8.199 < Q2 ≤ 8.553, 8.553 < Q3 ≤ 8.983, and Q4 > 8.983. MCMI was categorized into quartiles based on their MCMI values as follows: Q1 ≤ 2.425, 2.425 < Q2 ≤ 2.840, 2.840 < Q3 ≤ 3.326, and Q4 > 3.326. In relation to the TyG index, each 1-unit increase was associated with an elevated risk of CVD, with an odds ratio (OR) of 1.32 [95% Confidence intervals (CI): 1.20–1.44] in Model 1. Model 2 demonstrated a similar OR of 1.31 (95% CI: 1.20–1.44), while Model 3 showed an OR of 1.18 (95% CI: 1.06–1.30). Similarly, for the MCMI, each 1-unit increase was also associated with increased CVD risk, with ORs of 1.35 (95% CI: 1.24–1.47), 1.32 (95% CI: 1.22–1.44), and 1.18 (95% CI: 1.08–1.29) in Models 1, 2, and 3, respectively. When the TyG index and MCMI were analyzed as categorical variables by quartiles, the risk of CVD showed a progressive increase across ascending quartiles in all models (Table 2). It should be noted, however, that in some instances, such as the Q2 of MCMI in Model 3, the confidence interval was very close to 1, suggesting a more modest association.

Table 2.

Logistic regression models for the association between the TyG index, the MCMI and cardiovascular disease risk.

Categories	Model 1		Model 2		Model 3
	OR	95% CI	OR	95% CI	OR	95% CI
TyG index
Continuous	1.32	1.20 ~ 1.44	1.31	1.20 ~ 1.44	1.18	1.06 ~ 1.30
Quartiles
Q1	1 (Reference)		1 (Reference)		1 (Reference)
Q2	1.38	1.15 ~ 1.66	1.37	1.14 ~ 1.65	1.36	1.12 ~ 1.64
Q3	1.60	1.33 ~ 1.91	1.55	1.29 ~ 1.86	1.44	1.19 ~ 1.75
Q4	1.72	1.43 ~ 2.05	1.69	1.41 ~ 2.03	1.43	1.18 ~ 1.74
MCMI
Continuous	1.35	1.24 ~ 1.47	1.32	1.22 ~ 1.44	1.18	1.08 ~ 1.29
Quartiles
Q1	1 (Reference)		1 (Reference)		1 (Reference)
Q2	1.38	1.14 ~ 1.67	1.36	1.13 ~ 1.65	1.33	0.09 ~ 1.62
Q3	1.61	1.34 ~ 1.94	1.60	1.32 ~ 1.93	1.46	1.20 ~ 1.78
Q4	2.20	1.84 ~ 2.64	2.15	1.79 ~ 2.58	1.75	1.43 ~ 2.14

TyG Quartiles: Q1 ≤ 8.199, 8.199 < Q2 ≤ 8.553, 8.553 < Q3 ≤ 8.983, and Q4 > 8.983.

MCMI Quartiles: Q1 ≤ 2.425, 2.425 < Q2 ≤ 2.840, 2.840 < Q3 ≤ 3.326, and Q4 > 3.326.

Model 1: unadjusted.

Model 2: adjusted for age, gender, smoking and drinking.

Model 3: adjusted for Model 2 plus residential area, education, and all co-morbidities.

TyG index: Triglyceride-Glucose Index; MCMI: Modified Cardiometabolic Index.

OR, Odds Ratio; CI, Confidence Interval.

Analysis of associations between TyG index, MCMI, and CVD risk using RCS models

Figure 2A illustrated an unadjusted RCS model that revealed a positive nonlinear association between the TyG index and CVD risk (p for overall < 0.001; p for nonlinear = 0.012). Even after adjusting for confounding variables, the TyG index continued to demonstrate a nonlinear association with CVD risk, as shown in Fig. 2B and 2C. Furthermore, Fig. 2D indicated a J-shaped nonlinear association between the MCMI index and CVD risk (p for overall < 0.001; p for nonlinear < 0.001). Following adjustments, Fig. 2E and 2F supported these findings.

Fig. 2.

Restricted cubic spline models. Restricted cubic spline models with multivariable-adjusted associations were adopted to demonstrate dose-response associations between TyG index (ABC), MCMI (DEF) and the prevalence of cardiovascular disease.

Comparison of predictive efficacy of TyG index, CMI and MCMI for CVD risk events

Figure 3 compared the predictive abilities of the TyG index, CMI and the MCMI indices for CVD risk over a nine-year period. The results indicated that the MCMI demonstrated superior predictive capability compared to the TyG index and CMI (Table 3).

Fig. 3.

Receiver operating characteristic analysis comparing the predictive ability of TyG index, CMI and MCMI for cardiovascular disease.

Table 3.

The predictive ability of TyG, CMI, and MCMI for the risk of cardiolvascular disease.

Categories	AUC	95% CI	Compared to the MCMI
TyG	0.559	0.546–0.571	P < 0.001
CMI	0.556	0.543–0.568	P < 0.001
MCMI	0.581	0.569–0.594	-

The statistical comparison of the AUCs was employed using DeLong’s test.

TyG index triglyceride-glucose index, CMI cardiometabolic index, MCMI modified cardiometabolic index, AUC area under the curve, CI confidence interval.

Predictive performance across different models

When more patient information was incorporated into the prediction models, their performance improved accordingly, as evidenced by the following AUC (95% CI) values derived from the ROC curves: Model 1: 0.58 (0.56–0.60); Model 2: 0.60 (0.59–0.62); Model 3: 0.66 (0.65–0.68) (Fig. 4).

Fig. 4.

Receiver operating characteristic analysis comparing the predictive ability of Model 1–3. Model 1 was unadjusted; Model 2 was adjusted for age, gender, smoking, and alcohol consumption; and Model 3 was further adjusted for residential area, education level, and all comorbidities.

Subgroup analysis of TyG index and MCMI in predicting CVD risk

Figure 5 illustrated the subgroup analyses of the TyG index in predicting CVD risk. The analysis revealed that CVD risk was not significantly influenced by factors such as gender, hypertension, T2DM, alcohol consumption, smoking, or age. In other words, the interactions between these variables and the TyG index were not statistically significant (p > 0.05 for interaction). Similarly, Fig. 6 presented the subgroup analyses for the MCMI, which yielded comparable findings. It was noted that the MCMI and CVD risk were also not significantly affected by the same factors (p > 0.05 for interaction).

Fig. 5.

Subgroup analyses of TyG index were performed to investigate the association between the TyG index and cardiovascular disease across different subgroups.

Fig. 6.

Subgroup analyses of MCMI were performed to investigate the association between the MCMI and cardiovascular disease across different subgroups.

Discussion

In this study, we evaluated the ability of the MCMI to predict CVD risk and compared it with the TyG index. Our results showed that after adjusting for multiple confounding variables, both indices were associated with an increased risk of CVD. MCMI appears superior to TyG, but neither achieves strong discrimination.

CVD is the leading cause of death worldwide and a major cause of disability, imposing a heavy economic burden on society¹⁵. CVD is associated with a variety of risk factors, including hypertension, hyperlipidemia, and others¹⁶. Recent studies have shown that IR is significantly associated with both the incidence and outcomes of CVD^17–20. The TyG index has been found to be useful for assessing IR and is closely related to CVD^21,22. However, the TyG index is easily affected by factors such as FBG and age⁸. Furthermore, in recent years, novel indices such as the atherogenic index of plasma (AIP)—calculated as log(TG/HDL-C)—have been associated with CVD risk^23,24. However, this indicator focuses exclusively on lipid metabolism parameters. Therefore, the ultimate goal is to find a composite indicator that is more accurate and convenient.

CMI was initially introduced a decade ago as a straightforward marker for distinguishing diabetes mellitus, integrating factors of abdominal obesity and dyslipidemia¹¹. It has been shown to be associated with various diseases, including atherosclerosis, metabolic syndrome, and stroke^25–28. Guo et al. proposed a novel measure, the MCMI, based on the formulas for CMI and TyG¹². This new index integrates indicators of IR, abdominal obesity, and dyslipidemia. Guo et al. found that MCMI was positively correlated with the risk of both non-alcoholic fatty liver disease and liver fibrosis¹². To our knowledge, this is among the first studies to investigate the association between MCMI and CVD risk, and it also includes a comparison with the TyG index. Previous research has focused on the association between CMI and CVD risk. Liu et al. found that CMI serves as a reliable and independent predictor of CVD risk within the diabetic population²⁹. Chen et al. found that among U.S. adults, higher levels of CMI were significantly associated with increased odds of CVD prevalence³⁰. Our study yielded comparable findings, suggesting that MCMI serves as an independent risk factor for the development of CVD. The innovation of our metric lies in the integration of FBG into the CMI. Furthermore, we conducted a comparative analysis of the predictive capabilities of MCMI and the TyG index.

MCMI may serve as a valuable tool for predicting the risk of CVD and can be explained from several perspectives. Firstly, the MCMI comprehensively reflects IR and disorders in glucose and lipid metabolism. The core component of the MCMI—ln [TG × FBG/HDL-C]—essentially represents an improvement of the TyG index (ln [TG × FBG/2]), as it directly captures the critical features of IR, namely high TG³¹, elevated FBG¹⁴, and low HDL-C³². IR is a key driver of the occurrence and progression of CVD, as it impairs vascular health through multiple mechanisms, including endothelial dysfunction³³, inflammatory responses³⁴, and oxidative stress³⁵. Consequently, this aspect of the MCMI provides a quantitative assessment of the extent of glucose and lipid metabolic disorders and IR, forming the basis for its predictive capacity regarding CVD risk. Second, the MCMI reflects central obesity. A significant innovation of the MCMI is its integration of the WC/height as a measure of central obesity. Compared to WC or body mass index alone, the WC/height ratio provides a more effective adjustment for body type variations and offers a more accurate representation of visceral fat distribution³⁶. Central obesity is not only a key contributor to IR but also leads to the secretion of numerous pro-inflammatory factors, adipokines, and free fatty acids. These substances can directly or indirectly exacerbate IR, promoting atherosclerosis, cardiac hypertrophy, fibrosis, and thrombosis^37–39. Consequently, the predictive capacity of the MCMI for CVD risk is closely related to its ability to reflect central obesity. Third, the MCMI reflects an atherogenic dyslipidemia phenotype. A high TG/HDL-C ratio is widely recognized as a hallmark of atherogenic dyslipidemia, characterized by an increase in TG-rich lipoproteins and their remnants, an elevated presence of small dense low-density lipoprotein particles, and impaired HDL functionality^40–43. Therefore, MCMI is a simple and integrated metabolic indicator with potential utility for initial risk screening.

However, it is essential to contextualize our findings within the broader landscape of existing CVD risk prediction models. The field of cardiovascular risk prediction is saturated with numerous models of varying complexity and accuracy. Some of these models, such as the FRS, SCORE, and QRISK models, have undergone extensive validation and frequently report superior discriminative ability⁴⁴. These models typically integrate demographic, lifestyle, and multiple clinical variables, with their predictive performance based on more comprehensive information. In contrast, the AUC value for MCMI derived from this study (0.58), while statistically superior to the TyG index, demonstrates limited absolute discriminative ability as a standalone indicator and cannot directly replace the aforementioned well-established comprehensive risk assessment models. Therefore, the primary potential value of MCMI may not lie in pursuing “predictive superiority” over these composite models, but rather in its unique application advantages: its extreme simplicity of construction, rapid calculation, and low cost. This positions MCMI as a promising preliminary screening tool in resource-limited settings (e.g., primary care institutions) or for large-scale population-based initial screening.

Specifically, in primary care or public health screening, a threshold for MCMI could be established to identify a “potentially high-risk” population. For individuals identified as high-risk through this screening, the next step in clinical practice should not be initiating intensive treatment directly based on the MCMI result. Instead, they should be referred for more comprehensive assessment, such as formal risk stratification using composite models like FRS or SCORE, or for in-depth clinical evaluation by a specialist. This “simple initial screening to comprehensive assessment” two-step strategy can improve the detection rate of high-risk individuals while balancing screening breadth and resource efficiency.

Exploring the potential of MCMI within a multi-marker strategy is an important direction for future research. Our study suggested that the value of MCMI, as an indicator reflecting core metabolic risk, may lie in serving as a foundational element combined with other key variables. When we adjusted for variables such as age, gender, and comorbidities, the AUC for MCMI increased to 0.66. This combination of a “core metabolic indicator add key readily available variables” aims to judiciously add minimal complexity in exchange for a significant improvement in predictive performance, thereby potentially finding a more optimal balance between predictive accuracy and clinical ease of use.

This study has several advantages. First, it employs a large sample retrospective cohort design (n = 6,117) based on a national database (CHARLS), providing strong representativeness. Second, the statistical methods used in this study are comprehensive, utilizing logistic regression, RCS analysis, ROC analysis, and subgroup analysis, which enhances the overall robustness of the findings. Finally, the clinical significance is prominent. We believe this study represents an initial effort to propose and evaluate the MCMI for predicting CVD risk. By integrating multidimensional indicators such as blood glucose, blood lipids, and central obesity, it offers a promising, low-cost screening tool for primary healthcare. However, it is important to note that the modest AUC values for both indices indicate that their standalone predictive power is limited. Therefore, future research should explore the potential of integrating MCMI into a multimarker strategy, combining it with other established risk factors or novel biomarkers, to develop more robust and clinically useful predictive models for CVD.

Despite the strengths of this study, several limitations should be considered when interpreting the results. First, there are statistical concerns. The exclusion of participants due to existing CVD, missing data, or loss to follow-up may have introduced selection bias, potentially affecting the generalizability of our findings. Furthermore, while multivariable adjustment was employed to control for confounders, there is a risk of over-adjustment in the fully adjusted model (Model 3), particularly from the inclusion of comorbidities that might be mediators on the causal pathway. Although the overall trend remained consistent, this could have attenuated the observed effect sizes. Additionally, the multiple comparisons conducted in the extensive subgroup analyses increase the risk of type I error, implying that some isolated subgroup findings should be interpreted with caution and require future validation. Second, the ascertainment of the primary outcome is subject to potential misclassification. The diagnosis of CVD was based on self-report, which relies on participants’ accurate recall and understanding of a physician’s diagnosis. This could lead to under- or over-reporting of CVD, potentially biasing the observed associations between the MCMI and CVD risk, though the large sample size may have mitigated this to some extent. Third, and importantly, residual confounding remains a possibility due to unmeasured factors. Most notably, crucial information on the use of medications such as antihypertensive, lipid-lowering, and hypoglycemic agents was not available for analysis. These medications can profoundly influence the levels of the metabolic indices (e.g., by lowering TG, FBG) and the risk of CVD outcomes. Our inability to account for this key confounder represents a significant limitation and may have influenced the observed associations.

Fourth, the generalizability of our findings may be influenced by the baseline characteristics of the CHARLS cohort. The prevalence of key cardiometabolic conditions such as diabetes, hypertension, and hyperlipidemia in our analytical sample was relatively low. While this reflects the community-based nature of the cohort, it may limit our ability to fully assess and compare the predictive utility of MCMI and the TyG index in populations with a higher burden of these conditions, which are the primary targets for insulin resistance-focused risk stratification. Future studies in cohorts enriched with these risk factors are warranted to validate and extend our findings. Fourth, the generalizability of our findings may be influenced by the baseline characteristics of the CHARLS cohort. The prevalence of key cardiometabolic conditions such as diabetes, hypertension, and hyperlipidemia in our analytical sample was relatively low. While this reflects the community-based nature of the cohort, it may limit our ability to fully assess and compare the predictive utility of MCMI and the TyG index in populations with a higher burden of these conditions, which are the primary targets for insulin resistance-focused risk stratification. Fifth, the CHARLS study predominantly comprises Chinese adults. Given that the distributions and predictive values of cardiometabolic risk markers can vary across ethnic and genetic backgrounds, the performance of MCMI observed in our study may not be directly generalizable to other ethnic populations. In conclusion, large-scale clinical studies with more rigorous methodologies are needed to confirm these results.

Conclusion

This study indicated that the MCMI may serve as a preliminary screening indicator for cardiovascular disease risk in middle-aged and elderly populations. Although its independent predictive performance is limited, its simplicity and cost-effectiveness render it potentially useful in primary care settings and large-scale screening initiatives.

Abbreviations

CVD

Cardiovascular disease

IR

Insulin resistance

TyG

Triglyceride-glucose

CMI

Cardiometabolic index

WC

Waist circumference

TG

Triglyceride

HDL-C

High-density lipoprotein cholesterol

MCMI

Modified cardiometabolic index

CHARLS

China Health and Retirement Longitudinal Study

FBG

Fasting blood glucose

T2DM

Type 2 diabetes mellitus

MICE

Multiple imputation by chained equations

RCS

Restricted Cubic Spline

ROC

Receiver Operating Characteristic

AUCs

Areas Under the Curve

OR

Odds ratios

CI

Confidence intervals

Author contributions

S.Y. and C.Z was responsible for the design and conceptualization of the study, in addition to drafting and revising the manuscript. Y.D., D.C., and S.Z contributed to data collecting, statistical analysis, and result interpretation. S.Y. and C.Z was responsible for the data results visualization, read and revised the manuscript. All authors read and approved the final manuscript.

Data availability

The datasets produced and analyzed in this research are accessible in the CHARLS database (http://charls.pku.edu.cn/). The datasets generated during and/or analysed during the current study are available from the corresponding author (Chengsen Zhang) on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The CHARLS study was conducted in line with the principles stated in the Declaration of Helsinki and received approval from the Institutional Review Board of Peking University (IRB00001052-11015). Prior to their involvement in the CHARLS study, all participants gave their written informed consent. The research adhered to the STROBE guidelines for reporting observational studies in epidemiology.