Introduction

Non-alcoholic fatty liver disease (NAFLD) is a common and rapidly growing chronic condition, with a global prevalence of 32.4% and the highest incidence observed in industrialized nations, particularly the United States1. Characterized by lipid accumulation, insulin resistance, and liver damage induced by metabolic stress, NAFLD is closely associated with metabolic disturbances, including hyperglycemia, dyslipidemia, inflammation, coagulation abnormalities, and hypertension2,3,4. Consequently, NAFLD exerts a persistent disruption on metabolic homeostasis and profoundly impacts target organs, including the heart, brain, kidneys, and peripheral vasculature, ultimately leading to adverse health outcomes, notably an increased risk of cardiovascular mortality5,6,7,8. As the hepatic manifestation of metabolic syndrome, NAFLD is strongly linked to a spectrum of metabolic disorders, most notably type 2 diabetes(T2DM) and cardiovascular diseases (CVDs)9. Conditions including diabetes, hypertension, coronary artery disease, and stroke, commonly referred to as cardiometabolic diseases (CMDs), share overlapping etiologies and are strongly associated with NAFLD. These diseases are major contributors to global health burdens and mortality rates10,11. Furthermore, the coexistence of two or more CMDs, termed cardiometabolic multimorbidity (CMM), represents one of the most severe forms of multimorbidity. CMM is associated with a 3.7-6.9-fold higher risk of all-cause mortality and a 12-15-year reduction in life expectancy at age 60 compared to individuals without CMDs12.Given the high prevalence of NAFLD and its complex relationship with CMM, despite significant efforts in the prevention and treatment of NAFLD, the prognosis for NAFLD patients remains challenging. Therefore, identifying the risk of CMM in NAFLD patients is critical for reducing the incidence and mortality associated with CMM.

The Cardiometabolic Index (CMI) was proposed by Japanese scholar Wakabayashi13 in 2015, it is calculated based on triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and waist-to-height ratio (WHtR). WHtR is primarily used to assess obesity, while TG/HDL-C is employed to evaluate lipid levels. Both WHtR and TG/HDL-C are biomarkers of insulin resistance (IR) and are significantly associated with IR14,15. During IR, the reduced sensitivity of adipose tissue to insulin enhances lipolysis, which releases free fatty acids into the bloodstream. This excess of free fatty acids subsequently drives increased triglyceride synthesis in the liver16. Consequently, IR promotes hepatic fat accumulation, contributing to the onset and progression of NAFLD, as well as the development of atherosclerosis, hypertension, heart failure, and other CVDs17,18. Several studies have demonstrated a relationship between CMI and NAFLD, with the prevalence of NAFLD increasing as CMI rises19,20. Subsequent studies have also highlighted a strong correlation between CMI and diabetes, hypertension, stroke, CVDs, and kidney disease20suggesting that CMI is a novel indicator of visceral fat distribution and metabolic dysfunction with potential value as a marker of metabolic diseases21. However, the relationship between CMI and the risk of CMM in individuals with NAFLD remains unclear. Therefore, using data from the National Health and Nutrition Examination Survey (NHANES), which represents the general U.S. population, this study aims to investigate the association between CMI and the risk of CMM in NAFLD patients aged 18 and older.

Materials and methods

Study design and subjects

The NHANES, initiated by the U.S. Centers for Disease Control and Prevention (CDC) in the 1960s, is a large-scale, nationally representative survey designed to assess the health and nutritional status of the U.S. population. Using a complex, multistage stratified probability sampling approach, NHANES collects extensive cross-sectional data, including information on demographics, diet, physical activity, clinical measures, and laboratory results. Participants give informed consent, ensuring that ethical guidelines are followed and data privacy is maintained. As a crucial resource for public health research, NHANES data play a key role in monitoring health trends, assessing disease prevalence, identifying risk factors, and shaping health policies. The dataset is an invaluable tool for researchers in fields such as epidemiology, nutrition, and environmental health. The data is available to the public at [https://www.cdc.gov/nchs/nhanes].

The data for this study were sourced from the NHANES dataset, spanning from 1999 to 2018, with an initial cohort of 101,316 participants. Given the lack of abdominal ultrasound data in NHANES, the diagnosis of NAFLD was determined using the United States Fatty Liver Index (US FLI), where NAFLD was defined as a US FLI value greater than 3022,23. To ensure the robustness of the study, several exclusion criteria were applied: (1) participants under 18 years of age (N = 42,112); (2) individuals with factors such as excessive alcohol consumption (defined as > 3 drinks per day for men and > 2 for women), a positive hepatitis B or C status, missing components needed to calculate the US FLI, or a US FLI ≤ 30 (N = 52,753)22,23; and (3) those missing CMI data (N = 97) or CMM data (N = 361). After applying these exclusions, the final study population consisted of 5,993 individuals with NAFLD. A detailed flowchart outlining the participant selection process is provided in Fig. 1.

Fig. 1
figure 1

Flow-chart of the study samples.

Full size image

Definition of CMI, CMD and CMM

In this study, CMI was treated as the exposure variable, calculated as the product of the TG/HDL-C ratio and the waist-to-height ratio (waist circumference/height), in accordance with prior studies13. The CMDs considered in this analysis included hypertension, diabetes, coronary heart disease (CHD), and stroke24,25.Diabetes was identified through any of the following criteria: (1) self-reported physician diagnosis of diabetes; (2) current use of antidiabetic medications; (3) fasting plasma glucose ≥ 126 mg/dL or glycated hemoglobin ≥ 6.5%26. Hypertension was defined by any of the following: (1) self-reported history of hypertension; (2) use of antihypertensive medications; (3) systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg27. Stroke and CHD diagnoses were based on participants’ self-reported medical history. CMM was defined as the co-occurrence of two or more CMDs24,25.

Definition of covariates

In our analysis, we included several potential confounders as covariates, such as age, sex (female, male), race (non-Hispanic white, non-Hispanic black, Mexican American, and others), Body Mass Index(BMI) categories (< 25, 25–30, ≥ 30 kg/m²)28smoking status (non-smokers, former smokers, and current smokers), alcohol consumption (non-drinkers, light, moderate, and heavy drinkers)29physical activity levels (low, moderate, high, or very high, based on metabolic equivalents [MET-minutes/week])30and the use of medications for hypertension (yes, no), lipid-lowering agents (yes, no), and diabetes (yes, no). Hyperlipidemia was defined as meeting any of the following criteria: (1) triglycerides (TG) ≥ 150 mg/dL, (2) total cholesterol (TC) ≥ 200 mg/dL, (3) low-density lipoprotein cholesterol (LDL-C) ≥ 130 mg/dL, (4) high-density lipoprotein cholesterol (HDL-C) < 40 mg/dL for males or < 50 mg/dL for females, or (5) use of lipid-lowering medications31. The poverty-to-income ratio (PIR) was divided into three categories: low-income (< 1.3), moderate-income (1.3–3.5), and high-income (> 3.5)32.To assess NAFLD, we used the modified US FLI and Fibrosis-4(FIB4) score. The US FLI has been validated as a predictor of hepatic steatosis, with a score greater than 30 suggesting the presence of NAFLD33. The FIB4 score was utilized to evaluate the risk of advanced liver fibrosis34.

The formula is as follows35:

$$\eqalign{& {\text{USFLI}} = ({{\text{e}}^{ - 0.{\text{8}}0{\text{73}} \times {\text{non - Hispanic black}} + 0.{\text{3458}} \times {\text{Mexican American}} + 0.00{\text{93 }} \times {\text{ age }} + 0.{\text{6151}} \times {\text{ln}}({\text{gamma glutamyl transpeptidase)}}}}) \cr & + 0.0{\text{249}} \times {\text{waist circumference }} + {\text{ 1}}.{\text{1792 }} \times {\text{ ln }}\left( {{\text{insulin}}} \right) + 0.{\text{8242 }} \times {\text{ ln }}\left( {{\text{glucose}}} \right) - {\text{ 14}}.{\text{7812}})/ \cr & ~\left( {{\text{1 }} + {{\text{e}}^{\scriptstyle - 0.{\text{8}}0{\text{73 }} \times {\text{ non - Hispanic black}} + 0.{\text{3458}} \times {\text{Mexican American}} + 0.00{\text{93}} \times {\text{age}} + 0.{\text{6151}} \times {\text{ln}}\left( {{\text{gamma glutamyl transpeptidase}}} \right) \hfill \atop \scriptstyle + 0.0{\text{249}} \times {\text{waist circumference}} + {\text{1}}.{\text{1792 }} \times {\text{ ln }}\left( {{\text{insulin}}} \right) + 0.{\text{8242 }} \times {\text{ ln }}\left( {{\text{glucose}}} \right) - {\text{ 14}}.{\text{7812}} \hfill} }} \right) \cr}.$$

The FIB-4 score formula is as follows36:

$$\eqalign{& {\text{FIB}} - {\text{4 score }} = {\text{ Age }}\left( {{\text{year}}} \right) \times {\text{ AST }}\left( {{\text{IU}}/{\text{L}}} \right) \cr & /\left( {{\text{platelet count }}\left( {{\text{1}}{0^{\text{9}}}/{\text{L}}} \right) \times {\text{ square}} - {\text{root of ALT }}\left( {{\text{IU}}/{\text{L}}} \right)} \right) \cr} .$$

Statistical analysis

This study followed the NHANES guidelines and utilized a stratified, non-random sampling approach. We applied the appropriate NHANES sampling weights: the wtmec4 year weight for survey cycles from 1999 to 2002 and the wtmec2 year weight for cycles from 2003 to 2018. These weights were consistently used in all statistical analyses, including descriptive statistics and regression models, to ensure the accuracy and generalizability of our findings. Continuous variables were expressed as weighted means ± standard error (SE) and analyzed using weighted linear regression, while categorical variables were presented as weighted percentages ± weighted proportions and examined with weighted Rao-Scott chi-square tests. To evaluate the independent relationship between CMI levels and the risk of CMM/CMD in individuals with NAFLD, multivariable logistic regression models were used, adjusting for relevant clinical confounders. The clinical covariates were incorporated into the adjusted regression models if they met any of the following criteria: (1) a P-value < 0.10 in univariate analysis for the association between the covariate and the outcome variable, (2) inclusion of the covariate resulted in a greater than 10% change in the regression coefficients, or (3) the covariate was identified as a potential confounder based on previous studies37,38. As shown in Table S1, Table S2, Table S3 and Table S4 in Supporting Information Appendix A, the first two criteria were applied to select variables. Three adjustment models were applied: Model 1 (unadjusted), Model 2 (adjusted for age, sex, and race), and Model 3 (further adjusted for PIR, BMI, smoking and alcohol consumption, physical activity, and the use of lipid-lowering medications). Additional stratified analyses and interaction tests were conducted based on age, sex, race, and BMI to explore potential effect modifications. Restricted Cubic Spline (RCS) models were used to examine potential non-linear associations between CMI levels and the odds of CMM in NAFLD patients. We constructed RCS models using different knot numbers and selected the optimal model based on the lowest Akaike Information Criterion (AIC), which balances model fit and complexity. A significance level of P ≤ 0.05 was considered, and all statistical analyses were performed using R software (version 4.2.2) and Empower (R) (version 2.0) software.

Results

Basic characteristic of study participants

Table S5 in Supporting Information Appendix A shows the baseline characteristics of those patients who were excluded due to missing partial information of CMI and CMM. The study cohort consisted of a large sample of 5,993 participants. Based on the quartiles of CMI, participants were divided into four groups, with their baseline characteristics summarized in Table 1. Significant differences were observed across the groups. Compared to the Q1 group, individuals in higher CMI quartiles demonstrated a greater prevalence of CMM, higher BMI, elevated FLI, and an increased incidence of diabetes and dyslipidemia, and a higher proportion of middle-aged adults (40–59 years old). Furthermore, they had higher rates of smoking, a greater proportion of males, and a higher percentage of non-Hispanic White individuals. In terms of laboratory measures, most parameters, including fasting blood glucose, glycated hemoglobin, ALT, AST, GGT, serum cholesterol, triglycerides (TG), LDL-C, and uric acid, exhibited an increasing trend with higher CMI levels. In contrast, HDL-C levels showed a declining trend (p < 0.001). Figure 2 presents the overlap of the prevalence rates of the four major CMDs.

Table 1 Weighted baseline characteristics of four groups based on the quartiles of CMI.
Full size table
Fig. 2
figure 2

Venn diagram showing the overlap of prevalence of four cardiometabolic diseases (diabetes, CHD, hypertension, and stroke). Abbreviations: CHD, coronary heart disease.

Full size image

Association between CMI and CMM in NAFLD

We investigated the association between CMI levels and the risk of CMM in patients with NAFLD, presenting the results using three different multivariable logistic regression models in Table 2. The findings revealed a statistically significant positive correlation between CMI levels and CMM risk in NAFLD patients across all three models.

Table 2 Association between CMI and the risk of prevalent CMM in NAFLD participants.
Full size table

We further explored the association between CMI and the presence of one, two, or three CMDs in the population. The results, presented in Table 3 using three different multivariable logistic regression models, revealed that CMI was significantly associated with the presence of two or three CMDs across all three models.

Table 3 Association between CMI and the risk of prevalent cardiometabolic diseases in NAFLD participants.
Full size table

Dose-response relationships between CMI and CMM in NAFLD

We used RCS to explore the potential non-linear relationship between CMI and the risk of CMM in patients with NAFLD (Fig. 3). The model with four knots placed at the 5th, 35th, 65th, and 95th percentiles produced the lowest AIC value (Table S6), indicating the best model fit. The resulting dose–response curve visually depicted the association between CMI and CMM risk. The RCS analysis revealed a statistically significant nonlinear relationship (P for nonlinearity < 0.001). Specifically, CMM risk rose rapidly with increasing CMI levels until around 2.3, after which the increase became more gradual.

Fig. 3
figure 3

Multivariable odds ratio (or) for CMM based on CMI stratified by sex, age, race, BMI. Each stratification adjusted for all the factors (age, sex, race, PIR, BMI, smoking behavior, drinking behavior, physical activity, and lipid-lowering therapy) except the stratification factor itself. CMM cardiometabolic multimorbidity, CMI cardiometabolic index, PIR the poverty-to-income ratio, BMI body mass index.

Full size image

Subgroup analysis

To further elucidate the complex relationship between CMI and the risk of CMM in patients with NAFLD, we conducted subgroup analyses and interaction tests within predefined subgroups (Fig. 4). Among these subgroups, the association between CMI and CMM prevalence was notably stronger in female NAFLD patients (P for interaction = 0.011). In contrast, interaction tests in other subgroups did not show significant results (P for interaction > 0.05).

Fig. 4
figure 4

Dose-response relationship between CMI and CMM. Values represent difference in predicted response in reference to CMI of mean. Red solid lines represent restricted cubic spline models. The black vertical dashed line indicates the CMI median, and the red vertical dashed line indicates the changepoint. Adjusted for age, sex, race, PIR, BMI, smoking behavior, drinking behavior, physical activity, and lipid-lowering therapy. CMM cardiometabolic multimorbidity, CMI cardiometabolic index, PIR the poverty-to-income ratio, BMI body mass index.

Full size image

Discussion

This study investigated the potential relationship between CMI and the risk of CMM in the general adult NAFLD population in the United States. This association remained statistically significant after adjusting for all confounding variables (adjusted OR: 1.196, 95% CI: 1.077–1.328). RCS analysis demonstrated that the levels of CMI displayed a nonlinear relationship with CMM. Furthermore, Subgroup analysis revealed that the association between CMI and CMM prevalence was significantly stronger in female NAFLD patients. Finally, further investigation revealed that CMI remained positively associated with the presence of two or three CMDs. To our knowledge, this is the largest sample size study to date examining the relationship between CMI and CMM in the adult NAFLD population in the United States.

CMDs are a leading cause of death worldwide, and with the aging population, the incidence of CMD is rising significantly. This trend is unsurprising, as many CMDs share overlapping risk factors and etiologies, with complex bidirectional interactions between these diseases 10. The coexistence of multiple CMDs (CMM) significantly increases the risk of mortality 12, yet the precise pathophysiological mechanisms remain under investigation. It is currently believed that the development of CMM results from the combined effects of genetic, environmental, and lifestyle factors, which promote the onset and interrelationship of individual CMDs. Common pathophysiological pathways include chronic low-grade inflammation, IR, and dyslipidemia39. Furthermore, studies have shown that individuals with one CMD or CMM are 1.41 and 1.89 times more likely, respectively, to experience higher levels of psychological stress compared to those without CMDs40. Thus, CMM has a more profound negative impact on health compared to individual CMDs41. Previous studies have demonstrated a significant association between the CMI and increased CMDs risk in the general population13,42, and CMI is also considered a potential predictor of NAFLD. However, research regarding whether CMI can predict the risk of CMM in NAFLD patients remains limited. Therefore, this study aims to investigate the relationship between CMI and the risk of CMM in NAFLD patients. Our results reveal that elevated CMI levels are significantly associated with an increased risk of CMM in NAFLD patients.Meanwhile, we found that they exhibit a significant nonlinear relationship, consistent with previous studies in hypertensive populations43.This association remains statistically significant even after adjusting for various confounders. This study addresses a significant gap in the existing literature and provides valuable insights for early risk assessment and intervention in the clinical management of NAFLD patients.

Additionally, our study found that the relationship between CMI and CMM was more pronounced in female NAFLD patients, which is consistent with previous research. Previous studies have shown that females exhibit more significant gender differences in CMI, such as stronger associations with hyperglycemia and diabetes, Furthermore, the relationships between IR, TG, and the TG/HDL-C ratio are also more pronounced in women. Other studies have suggested that CMI has a stronger predictive ability for NAFLD in women, younger individuals, and non-obese subjects13,44,45. Similarly, gender differences in CMI regarding cerebrovascular diseases and hypertension have been confirmed, with female CMI showing stronger correlations with stroke and hypertension than in males46,47.One possible explanation is that women are more prone to the breakdown of visceral fat, producing non-esterified fatty acids (FFA), which, when transported to the liver, have a more significant impact on lipid metabolism. In addition, sex hormones may further influence the relationship between CMI and NAFLD by regulating fat distribution48,49.Therefore, gender-specific assessment is crucial in clinical management. Future studies should further investigate the interaction mechanisms between sex hormones and lipid metabolism to optimize prevention and treatment strategies.

Although the association between NAFLD and CMD has been extensively studied, the specific mechanisms underlying the role of CMI in the risk of CMM in NAFLD patients remain incompletely understood. Research has indicated that NAFLD is an important predictor of adverse cardiovascular outcomes and is closely associated with an increased risk of CMD, independent of traditional risk factors50,51,52. Dyslipidemia may be one of the key mechanisms, manifested by elevated TG, reduced HDL-C, and excessive circulating very-low-density lipoprotein cholesterol (VLDL-C), VLDL-C is considered the central factor in the dysregulation of TG and HDL-C, and can further lead to other lipid abnormalities, such as elevated LDL-C, which accelerate atherosclerotic changes53,54. Therefore, the dysregulation of triglyceride-rich lipoproteins is considered a key factor in the cardiovascular risk of NAFLD patients55. In this context, various lipid parameters and lipid ratios have been widely explored, with CMI appearing as a better predictor of NAFLD, potentially related to IR56,57. Adipose tissue IR promotes lipolysis, releasing large amounts of FFAs that are taken up by the liver and converted into LDL-C, thereby accelerating atherosclerosis progression58,59. The inflammatory response and the generation of reactive oxygen species can exacerbate IR, creating a vicious cycle that triggers systemic inflammation60. Notably, IR plays a significant role in the excessive production of VLDL-C in NAFLD patients. Abnormal expression of key molecules in the insulin signaling pathway, such as phosphatases and protein kinases, leads to hepatic IR, which in turn causes the overexpression of key proteins involved in lipid metabolism. This stimulates the liver to produce large amounts of VLDL-C, subsequently inducing dyslipidemia61. Moreover, IR impairs the liver’s ability to clear VLDL-C, leading to excessive accumulation of triglyceride-rich lipoprotein remnants62. IR may also be linked to an imbalance in adipokines and cytokines, characterized by reduced levels of anti-inflammatory adipokines and increased levels of pro-inflammatory cytokines (including interleukins IL-8, IL-6, IL-1b, interferon(IFN)-γ, etc.). The proliferation and dysfunction of adipose tissue, particularly visceral adipose tissue, downregulate the synthesis of anti-inflammatory adipokines while upregulating pro-inflammatory cytokines. This inflammatory environment in NAFLD promotes the formation of atherosclerosis63,64,65. In conclusion, the interplay between IR dyslipidemia, and inflammation may contribute to the association between CMI and CMM in NAFLD. However, further research is needed to elucidate these mechanisms and provide a stronger foundation for targeted interventions.

The main strength of this study lies in its novel use of large-scale cross-sectional NHANES data to investigate the relationship between CMI levels and CMM risk in NAFLD patients. However, several limitations should be acknowledged. First, the cross-sectional design of the study precludes the establishment of a causal relationship between CMI levels and CMM risk, and further validation through longitudinal studies or randomized controlled trials is needed. Second, the diagnosis of NAFLD was based on the USFLI score rather than the gold standard in clinical practice-liver biopsy. Third, laboratory results were based on a single measurement, which did not account for potential long-term variations in these biomarkers. Fourth, although numerous confounding variables were controlled for, the possibility of residual confounding cannot be fully excluded. Fifth, the diagnosis of CMM was partially based on baseline questionnaires and self-reports, which may introduce information bias or underreporting. Finally, the study cohort consisted only of adult NAFLD patients from the United States, limiting the generalizability of the findings. Future multi-center, prospective studies are needed to further elucidate these relationships.

Conclusions

Our cross-sectional study found a nonlinear relationship between CMI and CMM in adult NAFLD patients. These findings indicate a potential association between CMI and cardiometabolic burden in this population, highlighting the relevance of CMI in metabolic health assessment. This underscores the importance of monitoring CMI levels in NAFLD patients and exploring early strategies to mitigate CMM risk.