Associations between the cardiometabolic index and atherosclerotic cardiovascular disease across different glucose metabolism statuses: insights from NHANES, 1999–2020

Abstract

Background

The cardiometabolic index (CMI) serves as a significant marker of diabetes mellitus (DM) and may predict the potential for cardiovascular disease development. Nevertheless, the correlation between CMI and atherosclerotic cardiovascular disease (ASCVD) among individuals exhibiting varying glucose metabolism statuses (GMS) continues to be unclear.

Methods

Overall, 24,006 individuals aged 20 and above were enrolled in the research, drawn from the National Health and Nutrition Examination Survey (NHANES) database. Individuals in the study was classified into three distinct categories according to the level of fasting plasma glucose or glycated haemoglobin: normal glucose regulation, prediabetes, and DM. Multivariate logistic regression models and smoothed curve-fitting techniques were applied to investigate the correlation between CMI and ASCVD risk across varying GMS. Additionally, subgroup analyses stratified by relevant factors were performed to identify potential effect modifiers in this relationship.

Results

Overall, 2352 participants (9.8%) with ASCVD were included. An increasing trend in ASCVD risk was observed for each successive CMI tertile. After adjusting for all related covariates, a significantly positive association was observed between CMI and ASCVD (P = 0.0004). Participants with DM in the highest CMI tertile had a 114% higher ASCVD risk compared to those in the lowest tertile (OR = 2.14; 95% CI = 1.30–3.53). Smoothed curve-fitting consistently confirmed the correlation between CMI and ASCVD across diverse GMS. Subgroup analyses and interaction tests highlighted statistically significant differences within the drinking status subgroup (P-interaction = 0.0479) and GMS subgroups (P-interaction = 0.0397).

Conclusion

This research suggests a positive association between ASCVD and CMI in adults in the United States, particularly among individuals with DM.

Background

It is commonly acknowledged that cardiovascular disease (CVD) represents the predominant factor of mortality globally. [1–2]. Despite significant advancements in its diagnosis and treatment, atherosclerotic cardiovascular disease (ASCVD), especially cerebrovascular accident and coronary artery diseases, remains a major health challenge, highlighting the importance of its prevention. Diabetes mellitus (DM) constitutes one of the influential factors for ASCVD development, and the relationship between prediabetes mellitus (pre-DM) and ASCVD has garnered considerable attention [3]. Traditional risk factors are insufficient to predict ASCVD risk [4]. Consequently, there is a growing need for innovative indices that integrate various metabolic parameters to comprehensively evaluate cardiometabolic risk.

The cardiometabolic index (CMI) represents a promising indicator that merges lipid parameters with waist-to-height ratio (WHtR) [5,6,7]. Previous researches have constructed a link between CMI and metabolic syndrome (MetS), as well as diverse cardiovascular manifestations [8–9]. Individuals with elevated CMI levels may have increased systemic inflammation, worsening CVD and DM, leading to a greater risk of complications, poorer prognoses, and higher mortality rate [10]. However, the predictive capability of CMI for ASCVD across various glucose metabolic statuses (GMS) remains underexplored.

This study aimed to address this deficiency by utilizing information derived from the National Health and Nutrition Examination Survey (NHANES) database. The research offers innovative perspectives on the role of CMI in predicting ASCVD, particularly in individuals with varying glucose metabolic statuses. Identifying the predictive value of CMI in these populations is essential for advancing global public health and developing targeted ASCVD prevention strategies.

Methods

Population and study design

The present investigation employed cross-sectional datasets derived from the NHANES, thorough investigation aimed at assessing the nutritional condition of the populace in the U.S. Ethical approval was obtained from the National Center for Health Statistics (NCHS) Review Board, and individuals involved in the study gave their consent [11]. This cross-sectional investigation conformed to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline (Table S1) [12]. The datasets are available to the public via the website at https://www.cdc.gov/nchs/nhanes/.

Participants in the cross-sectional study were collected from different waves spanning 1999 to 2020, before the pandemic. To ensure the validity and dependability of the findings, particular exclusion standards were enforced, which included: (1) participants younger than 20 years (N = 48,878); (2) participants without a recorded CMI value (N = 34,615); (3) participants lacking complete questionnaire data pertaining to ASCVD (N = 119), and (4) individuals without a diagnosis of GMS (N = 4). In total, 24,006 individuals were recruited (Fig. 1).

Fig. 1

Flowchart of the population

Definition

CMI

Stringent laboratory evaluations and examinations were performed to guarantee the integrity and comparability of the statistics. Typically, blood specimens obtained from mobile survey units or predefined sampling sites are generally processed and analyzed in a controlled laboratory environment. In mobile health-screening units, licenced healthcare practitioners measured the stature and abdominal circumference of the participants. The WHtR was calculated using the aforementioned indicators as the waist circumference-to-height ratio (cm/cm). The CMI was computed using the following formula [13]:

$$\:CMI=\frac{TG\:(mmol/L)}{HDL‐C,\:mmol/L}\times\:WHtR$$

ASCVD

The major outcome of the study was ASCVD, characterised by the occurrence of at least one diagnosis of angina, myocardial infarction, cerebrovascular accident, or coronary heart disease, as delineated in the guidelines founded by the American College of Cardiology and the American Heart Association [14]. The participants’ self-reported history of ASCVD was estimated using the question: “Have you ever been advised by a physician or other healthcare professionals that you are diagnosed with coronary heart disease, cerebrovascular accident, angina, or myocardial infarction?” Individuals were classified as having ASCVD if they responded positively to any of the enquiries.

Glucose metabolism status

Pre-DM is characterized by a glycated haemoglobin (HbA1c) level that falls within the range of 5.7–6.4%, or the concentrations of fasting blood glucose (FBG) spanning from 5.6 mmol/L to 6.9 mmol/L. DM was diagnosed based on an HbA1c level of 6.5% or more or a FBG concentration of 7.0 mmol/L or above. Individuals without a diagnosis are considered to have normal glucose regulation (NGR) [15].

Other variables

The following variables were included in the current investigation: race, family poverty-to-income ratio (PIR), educational attainment, alcohol consumption, smoking habit, sex, weight, height, waist circumference, age, TG, total cholesterol (TC), diastolic blood pressure (DBP), systolic blood pressure (SBP), HbA1c, FBG, low-density lipoprotein cholesterol (LDL-C), HDL-C, and history of hypertension and hyperlipidaemia. The level of educational achievement was categorised into less than high school, high school or equivalent, and higher than high school. Body Mass Index (BMI) is determined by dividing the weight (kg) by the square of the height (m²).

The determination of smoking status was contingent upon the responses provided to the following question: “Have you smoked at least 100 cigarettes over the course of your life? “The participants were grouped as either smokers or non-smokers. Participants’ alcohol consumption was sorted into two groups: drinkers and non-drinkers. In the past 12 months, a person with average drinking ≤ 1 drink/day was considered a non-drinker; those with average drinking > 1 drink/day were considered drinkers. Hypertension was identified based on self-reported medical diagnoses and an average SBP/DBP of 140/90 mmHg or higher [16]. Hyperlipidaemia is characterised by TC levels of 200 mg/dL or higher, LDL-C levels of 160 mg/dL or higher, HDL-C levels lower than 40 mg/dL, or a prior diagnosis made by a physician [17].

Statistical analysis

Consistent with the intricate sampling framework of the NHANES, this study employed suitable weights to guarantee that the sample accurately represented the demographics of the United States national population. Statistical data were analysed through Empower Stats (version 4.0) and R (version 4.3.1). The weighted student’s t-test or chi-square analysis was adopted for continuous or categorical variables, separately, to assess variations in fundamental characteristics between the non-ASCVD and ASCVD groups. Potential sources of bias were considered to enhance the precision of outcomes. Specifically, Model 1 didn’t incorporate any supplementary adjustment variables. Considerations for demographic factors such as race, gender, and age were integrated into the adjustment in Model 2. Model 3 encompassed a thorough array of adjustments for ethnicity, gender, PIR, alcohol consumption, smoking habit, education, age, BMI, hypertension, and hyperlipidaemia. Furthermore, the CMI was categorised into tertiles for detailed analysis. To explore the correlation between CMI and ASCVD, both univariate and multivariate logistic regression analyses were applied. Meanwhile, smoothed curve fitting was performed. A subgroup analysis was conducted with stratified factors, including gender, race, educational attainment, alcohol consumption, smoking habit, age, BMI, hypertension, hyperlipidaemia, and GMS, to identify potential moderating effects. Additionally, heterogeneity in correlations among distinct subgroups was evaluated through interaction analyses. Additionally, likelihood ratio tests were adopted to investigate the interplay between CMI and stratification factors. The level of statistical significance was confirmed at a P-value threshold below 0.05.

Results

Fundamental characteristics

Table 1 provides an overview of the demographic features of the participants, comprising 2352 individuals in the ASCVD group and 21,654 individuals in the non-ASCVD group. Specifically, participants in the ASCVD group exhibited a mean age of 64.04 years, with a sex composition of 55.90% male and 44.10% female. Conversely, the non-ASCVD group exhibited a mean age of 45.30 years, characterised by a sex distribution of 47.86% male and 52.14% female. Subjects with ASCVD exhibited a tendency to be male, older, had lower educational attainment and PIR, and had significantly increased waist circumference measurements, weight, WHtR, BMI, CMI, TG, and glucose indices, such as FBG and HbA1c, compared with those without ASCVD (all P < 0.01). Furthermore, the occurrence of pre-DM, DM, hypertension and hyperlipidaemia was greater in participants with ASCVD than those without ASCVD.

The general features of the individuals, classified according to their GMS, are presented in Table 2. The transition from the NGR group to the pre-DM group, followed by the DM group, demonstrated a sustained elevated CMI (P < 0.0001).

Table 1 Weighted general characteristics of subjects classified by ASCVD

Table 2 Weighted baseline characteristics of subjects classified by GMS

Association between CMI and ASCVD

The data presented in Table 3 revealed a significantly positive association between CMI and ASCVD in Model 1 and Model 2 [odds ratios (ORs) 1.11, 95% confidence intervals (CIs) 1.07–1.14; OR 1.13, 95% CI 1.09–1.16]. After adjustment, a one-unit rise in the CMI was related to a 10% elevation in the likelihood of developing ASCVD (OR 1.10, 95% CI 1.04–1.16) in Model 3. Moreover, when the CMI was categorized into tertiles, the association between the highest tertile and ASCVD remained significant (P = 0.0016). In comparison to subjects in the lowest tertile of CMI, those situated in the highest tertile of CMI presented a 38% elevated risk of ASCVD (OR 1.38, 95% CI 1.23–1.69). In addition, the smoothed-fit curves depicted in Fig. 2A provide additional support for the positively linear correlations between CMI and ASCVD risk.

Table 3 ORs (95% CIs) of ASCVD according to CMI in three models

Fig. 2

Correlation between CMI and ASCVD in subjects with various GMS

The y-axis shows the probability of ASCVD occurrence, where a value of 0 signifies no occurrence, and a value of 1 indicates the occurrence of ASCVD. The x-axis represents the CMI, which ranged from 0 to 30. Figure 2A illustrates the linear correlation between CMI and ASCVD among the whole participants. The red spline curve exhibits a linear correlation between the CMI and ASCVD (P for trend = 0.001). The region indicated by dashed blue lines represents the 95% CI. Each red dot represents the concentration of CMI, which contributes to the formation of a continuous fitted curve. Figure 2B shows the relationship between the CMI and ASCVD in individuals with three various GMS. The red spline curve represents the risk of ASCVD in the NRG, the green spline curve represents individuals with pre-DM, and the blue spline curve represents individuals with DM. Each data point indicated the concentration of CMI, resulting in a continuously fitted curve. The adjusted variables included race, education, PIR, sex, age, BMI, alcohol consumption, smoking status, hypertension, and hyperlipidaemia.

Relationships between CMI and ASCVD differentiated by individual glucose metabolic States

Table 4 displays the outcomes of multivariate logistic regression analysis examining the correlation between CMI and ASCVD across various GMS. In individuals with NGR, a notably positive association between CMI and ASCVD was evident (Model 1: OR 1.22, 95% CI 1.12–1.33; Model 2: OR 1.28, 95% CI 1.17–1.40; Model 3: OR 1.18, 95% CI 1.04–1.35). Besides, a notably positive association between CMI and ASCVD among individuals with DM (OR 1.06, 95% CI 1.00–1.12) in Model 3 was shown. Upon stratifying CMI into tertiles, individuals in the higher CMI tertiles showed a greater trend of ASCVD than those in the lowest tertile in the pre-DM and NGR groups. Furthermore, in Model 3, participants with DM in the highest CMI tertile demonstrated a 114% elevated risk of developing ASCVD relative to those in the lowest CMI tertile (OR 2.14, 95% CI 1.30–3.53, P for trend = 0.0028). As a result of a previous multivariate analysis revealing an association between CMI and ASCVD, this research utilized smooth curve fitting to investigate this correlation further. After full adjustment, the smoothed plots revealed increased associations between CMI and ASCVD in participants with NGR or DM, and a greater correlation was found in individuals with DM than in those with NGR (Fig. 2B). However, the plot revealed a relatively stable risk of ASCVD as CMI increased in individuals with pre-DM.

Table 4 Relationship between CMI and ASCVD risk as classified by GMS

Correlation of CMI with ASCVD risk among diverse subpopulations

Subgroup analyses were conducted across a range of subpopulations, including race, education, sex, age, BMI, drinking, smoking, hypertension, hyperlipidaemia, and GMS, to estimate the stability and dependability of the association between CMI and ASCVD. Age stratification was conducted according to the classification standards of World Health Organization. The results are shown in Fig. 3. Following adjustment for all covariates, the findings aligned with those derived from prior analyses conducted on the entire population, indicating that elevated CMI is correlated with a higher OR for ASCVD risk. The relationship is consistently observed across the sex, age subgroups or those with education above high school, race of non-Hispanic white, those with BMI ≥ 24, smokers, or those with hypertension, hyperlipemia, or DM. Interaction analysis showed a significantly difference in the drinking or GMS subgroups (P-interaction = 0.0479, 0.0397). In addition, subgroup analyses and interactions for the associations between CMI and ASCVD in the various GMS groups are shown in the supplemental tables (Table S2–4). The analysis demonstrated a notable association between CMI and ASCVD across subgroups stratified by the presence of hyperlipidaemia (P < 0.05) in the DM group (Supplementary Table S4).

Fig. 3

Subgroup analyses and interaction tests between CMI and ASCVD

The adjusted variables included sex, education, age, race, smoking habit, alcohol consumption, PIR, BMI, hypertension and hyperlipidaemia

Discussion

The CMI was first proposed in 2015 as a tool for diagnosing DM and has been shown to be strongly correlated with hyperglycaemia [7]. Subsequent investigations have examined the relationship between CMI and DM in Japanese adults [18]. Similarly, individuals exhibiting elevated levels of CMI demonstrate a markedly increased likelihood of new-onset DM [19]. Furthermore, studies have broadened their significance to CVD and other metabolic conditions. A constructive correlation between CMI and CVD in patients with hypertension has been proven [20]. Higashiyama et al. illustrated an elevated risk of ischaemic CVD in participants without metabolic syndrome who presented with an elevated CMI [21]. CMI may also serve as a significant predictor of NAFLD [20], particularly in Chinese women [22]. Research has proved the association between CMI and unfavorable prognosis. However, few studies have illustrated the association between CMI and ASCVD across various GMS.

In the NANHES database, the CMI has been extensively studied, particularly for its association with insulin resistance and DM. Liu et al. indicated a relationship between CMI and glucose metabolic biomarkers, namely HbA1c, fasting insulin, FBG, and HOMA-IR [23]. Song et al. provided further evidence of the relationship between CMI and glucose categories, which is crucial for understanding the implications of their findings [24]. Consistent with previous research, this study revealed that both the prediabetic and diabetic populations exhibited higher CMI levels than the NGR group. This study demonstrated a notable association between CMI and ASCVD. The results also indicated that participants with ASCVD had elevated CMI in comparison with those without ASCVD. After adjustment, the incidence of ASCVD was greater in individuals in the highest CMI tertile among individuals with DM than those in the lowest tertile, which was not observed in the NGR or prediabetes populations.

The comprehensive CMI serves as an integrative measure that combines the lipid metabolism index with the WHtR. Traditionally, a decreased LDL-C level has been regarded as advantageous for preventing CVD, with statins being frequently employed. However, patients with elevated TG levels are at increased risk of ischaemic events, even during statin therapy [25]. Recent research has further supported the notion that lowering TG levels may be an efficacious approach to preventing CVD [26]. Consequently, it is reasonable to propose that a reduction in TG, which correlates with a decrease in CMI, may be involved in mitigating the risk of CVD. Additionally, a WHtR threshold of 0.5 was identified as an appropriate threshold for predicting CVD and DM [27]. Collectively, these findings exhibit that the CMI may provide a more comprehensive reflection of metabolic changes.

The explanation for the connection between ASCVD and CMI is still elusive; however, some plausible mechanisms have been suggested. First, elevated TG levels may enhance inflammatory reactions in vascular endothelial cells, particularly in individuals with DM [28–29]. TG-rich lipoproteins were proven to contribute to the formation of atherosclerotic lesions [30–31]. The results also verified higher TG levels in individuals with DM. Subgroup and interaction analyses revealed the role of hyperlipidaemia. Individuals with hyperlipidaemia and DM have a higher probability of developing ASCVD than those without hyperlipidaemia among individuals with DM. Second, the WHtR, an effective measurement of central adiposity, has been recognized as a better indicator of CVD among individuals of diverse nationalities and ethnic backgrounds [32] and a risk factor for DM [27]. Obesity induces dysfunction within the adipose tissue, and the consequent dysregulation of adipokine concentrations fosters a chronic, systemic inflammatory milieu, which is primary to CVD development [33]. CMI, as a composite measure incorporating the two aforementioned key indicators, can potentially emerge as a superior marker for ASCVD. However, additional studies are required to elucidate these intricate mechanisms.

Strengths and limitations

The research represents the most comprehensive, population-centric exploration into the association between CMI and ASCVD in individuals with different GMS. The information was gathered through a meticulously designed, nationally representative survey conducted across diverse regions of the United States. To ensure the robustness of the findings, potential confounding variables were incorporated to mitigate any interference in the results. The results suggest that higher CMI levels correlate with increased ASCVD risk, particularly in participants with DM. Furthermore, given that standard assays for lipids parameters are commonly used clinically, and that the CMI can be easily calculated, it is prudent to advocate CMI as an effective and convenient metric for evaluating ASCVD risk in patients with DM.

There are some limitations. First, owing to the lack of objective measures of atherosclerotic disease in the NHANES, the primary outcome was determined based on the NHANES questionnaire responses, which may have introduced a recall or self-report bias. Second, the logistic model did not incorporate baseline medications, including antihypertensive, lipid-lowering, and glucose-lowering agents, which may have influenced the outcomes. Third, although the diagnostic criteria for GMS are well-defined, there remains a possibility of misclassification among participants with borderline glycaemic status. Fourth, the exclusion of certain participants from the NHANES database owing to stringent exclusion criteria resulted in a reduced sample size, potentially introducing attrition bias. Fifth, the study neglected to explore temporal variations in CMI across the entire follow-up period. Ultimately, certain variables may have been overlooked. More high-quality and prospective researches are required to confirm these results in the future investigations.

Conclusions

The research suggests a positive association between ASCVD and CMI among U.S. adults, particularly among those with DM. This index integrates multiple metabolic parameters, including adiposity and lipid profiles, and offers a more holistic assessment of cardiometabolic risk than the traditional measures. By incorporating the CMI into clinical practice, clinicians may better stratify patients for ASCVD risk and prioritise preventive interventions.