Abstract

Background

The triglyceride glucose index (TyG index) is a crucial marker for assessing the risk of chronic diseases, while socioeconomic status (SES), measured by poverty income ratio (PIR) and education level, reflects an individual鈥檚 social standing. Past studies have linked SES to diabetes and cardiovascular diseases, but research on its association with the TyG index is limited. This study aimed to explore the association between SES and the TyG index and assess the mediating role of BMI.

Methods

The cross-sectional study utilized data from the National Health and Nutrition Examination Survey 2007鈥�2016 cycles to explore the relationship between SES and the TyG index in the adult of the USA. Multivariate logistic regression, stratified and interaction analyses were conducted to assess the association between SES and the TyG index. Additionally, parallel mediator analysis estimated the mediated effect of BMI between SES and the TyG index.

Results

Among the 11,358 individuals studied, averaging 49 years of age and with 48.3% males, fully adjusted models revealed negative associations between PIR and education level with the TyG index, while BMI showed a positive correlation. Stratified and interaction analysis indicated consistent findings across subgroups. Mediation analysis revealed that BMI mediated 14.4% and 8.57% of the effects of PIR/Education level on the TyG index, respectively.

Conclusions

SES was negatively associated with the TyG index. Additionally, BMI partially mediate the association between SES and the TyG index. These findings deepen the comprehension of the association between SES and the TyG index.

Introduction

The triglyceride glucose index (TyG index) has garnered considerable interest among researchers due to its simplicity, cost-effectiveness, and efficacy as an assessment tool [1]. Research has demonstrated a strong correlation between the TyG index and insulin resistance, indicating its efficacy as a dependable marker for assessing this physiological condition [2鈥�3] and effective predictor of diabetes mellitus [4]. Several investigations have identified a robust correlation between TyG index and the onset and progression of hypertension [5], metabolic syndrome [6], ischemic stroke [7], atherosclerosis [8鈥�9], cardiovascular disease [10,11,12] and overall health status [13]. Furthermore, a noteworthy association has been observed between the TyG index and overall mortality in critically ill individuals with hemorrhagic stroke, indicating its potential as a novel assessment tool for overall mortality in this patient population [14]. In addition to its association with adverse cardiovascular and metabolic outcomes, an elevated TyG index has been linked to risks of osteoarthritis [15], breast cancer [16], occurrence and recurrence of colorectal adenomas [17], and depression [18鈥�19]. In summary, the TyG index is regarded as a significant marker that has emerged as a potential risk indicator for various chronic diseases. Understanding the influencing factors of the TyG index can facilitate early intervention and prevention of metabolic and chronic diseases.

In the fabric of society, socioeconomic status (SES) delineates an individual鈥檚 position, commonly evaluated through factors such as the poverty income ratio (PIR) and education level. Previous research has established a correlation between SES and the incidence of diabetes and cardiovascular diseases [20, 21], the prevalence of heart failure [22], and overall health status [23]. Previous research has predominantly concentrated on the influence of SES on diseases. To date, there has been no investigation specifically examining the association between SES and the TyG index, which serves as a potential biomarker for numerous chronic diseases.

The study was conducted to explore the possible association between SES and the TyG index. This relationship was examined within a nationally representative cohort of American adults, utilizing data from the National Health and Nutrition Examination Survey (NHANES) spanning from 2007 to 2016. Additionally, the objective was to examine whether specific relevant variables serve as mediators between SES and the TyG index.

Methods

The data source and study population

The NHANES, conducted by the National Center for Health Statistics (NCHS), serves as a crucial epidemiological survey. NHANES seeks to evaluate the nutritional and health condition of the American and offer crucial information on the health status of individuals for policymakers, researchers, and the public. The NHANES was authorized by the NCHS institutional review board. All participants in the project gave their informed consent by signing a consent form. Given that the study utilized publicly available de-identified data, the Ethics Review Committee of the Second Hospital of Shandong University granted an exemption for the study. The study data are derived from the NHANES(2007鈥�2016). Participants were excluded based on predetermined criteria outlined as follows: (1) age less than 20 years (n鈥�=鈥�21387); (2) participants lacking data on triglycerides and glucose (n鈥�=鈥�16599); (3) participants without information on PIR, body mass index (BMI), and educational level (n鈥�=鈥�1244). The study ultimately included 11,358 participant (Fig.听1).

Fig. 1

Flowchart of the sample selection from NHANES (2007鈥�2016)

Study variables

The exposure variable was SES, which consists of two key elements: the PIR and educational level. PIR, a continuous metric, signifies the ratio of reported household income, adjusted for variables like household size, age distribution, and the current year. PIR was segmented into three categories: 鈮�1 (low), 1 to 4 (medium), and 鈮モ��4 (high). Educational level was also categorized into three categories: high school, less than high school, and some college or higher education.

The TyG index is the natural logarithm of the product of triglyceride level (mg/dL) and fasting glucose level (mg/dL), divided by 2 [24]. Fasting blood triglyceride levels were assessed by three separate analyzers: Roche modular P chemistry, Hitachi/Roche Cobas 6000, and Roche Hitachi 717/912. Fasting blood glucose (FBG) levels were measured using two different instruments. From 2007 to 2015, the Roche C501 equipment was deployed, whereas from 2015 to 2016, the Roche C311 device was employed for measurements. Fasting glucose was analyzed using either the Roche Cobas C311 or Roche/Hitachi Cobas C501 chemistry analyzer employing a hexokinase-mediated reaction.

Covariables

The sociodemographic covariables included sex, age (in years), ethnicity and race, and marital status (single, with partner or married). The health-related covariables included BMI(<鈥�25听kg/m2, normal; 25鈥�30听kg/m2, overweight; and 鈮モ��30听kg/m2, obesity), consumption of alcohol (heavy, low to moderate, and none), using tobacco (former, current, and never), and physical activity (inactive, insufficiently active, and sufficiently active); Self-reported diseases including hypertension, hypercholesterolemia, diabetes, and cardiovascular diseases (CVDs) were ascertained based on participants鈥� responses to questions in the questionnaire regarding whether they had informed their doctor of these conditions in the past.

The height and weight of the individuals were documented, and their BMI was computed and categorized accordingly. The smoking status was classified into three categories: nonsmoker (or smoked less than 100 cigarettes), past smoker (smoked more than 100 cigarettes but has since quit), and current smoker. The drinking habits were classified into three groups: heavy drinker, low to moderate drinker, and nondrinker, based on their self-identified mean daily alcohol consumption. A heavy drinker was described as someone who consumes at least one alcoholic beverage per day for women and two or more drinks per day for men. In contrast, a moderate drinker was identified as an individual who consumes less than one alcoholic beverage per day for women and less than two drinks per day for men.

Physical activity was assessed by calculating the total minutes of activity weekly, which included moderate-intensity activity (in minutes) plus twice the minutes of vigorous-intensity activity, and classified into three categories: inactive (no moderate or vigorous-intensity activity), sufficiently active (more than 150听min/week), and insufficiently active (engagement in some activity but not meeting the criteria for being sufficiently active).

Statistical analysis

Continuous variables displaying a normal distribution were presented as mean鈥壜扁�塖D, and subjected to comparison through one-way analysis of variance (ANOVA). Conversely, for continuous variables exhibiting skewed distributions, the Kruskal-Wallis H test was utilized for comparison, and the median (interquartile range) was provided. Categorical or dichotomous variables were expressed as absolute values (percentages), then chi-squared statistics was used for comparison. The collection and evaluation of PIR concentrations and distributions were conducted. To ascertain normality in continuous variables, the Shapiro-Wilk statistical test was utilized.

Linear regression was utilized to compute 尾 coefficients and their corresponding 95% confidence intervals (CIs) for assessing the correlation between SES/BMI and the TyG index. Individuals were stratified according to tertiles of SES/BMI: PIR鈥夆墹鈥�1, 1< PIR<4, PIR鈥夆墺鈥�4; Edu 1( Less than High school), Edu 2 (High school) and Edu 3 (Some college or more); BMI(<鈥�25听kg/m², 25鈥�30听kg/m², and 鈮モ��30听kg/m². Further, the reference group for comparison was established as the lowest tertile. Additionally, Linear regression analyses were conducted, employing PIR/BMI as both a continuous and categorical variable. Model 1 was adjusted for age, sex, and marital status, while Model 2 further incorporated adjustments for drinking status, smoking status, physical activity, hypertension, and CVDs.

Moreover, potential interactions affecting the association between SES and the TyG index were evaluated, considering variables such as sex, age (<鈥�60 vs. 鈮�60 years), physical activity (Irregular/No vs. Regular), hypertension (No vs. Yes), and CVD (No vs. Yes). With the exception of the stratification factor itself, all other variables within each stratum were fully adjusted. The presence of heterogeneity among subgroups was evaluated using multivariate linear regression analysis.

The potential mediating effect of BMI on the link between SES (assessed by PIR/Edu) and the TyG index was estimated using parallel mediator analysis. We re-categorized education level into two categorical variables for analysis: 鈥渉igh school/less than high school鈥� and 鈥渟ome college or more.鈥� In the parallel mediation analysis, BMI serve as mediator. The direct effect (DE) represents the impact of PIR/Edu on the TyG index. Indirect effects (IE) denote the effects of PIR/Edu on the TyG index mediated by BMI. The proportion of mediation was calculated by dividing the indirect effects by the total effect (TE). To further validate the stability of our findings, we employed bootstrap analysis with 1,000 resampling iterations [25]. The R 4.3.0 packages were utilized for statistical analysis. Significance was inferred at a p-value of less than 0.05, employing two-tailed testing.

Results

Clinical characteristics of the participants

A total of 11,358 eligible participants from the NHANES dataset spanning from 2007 to 2016 were included for analysis (Fig.听1). The baseline features of the individuals was presented in Table听1. Among the participants, 3,358 individuals were classified as having normal weight, 3,791 were categorized as overweight, and 4,209 were classified as obese. The average age of the participants was 49 years, with males comprising 48.3% of the sample. Variables such as the TyG index, PIR, educational level, age, gender, race and ethnicity, marital status, triglyceride, and fasting glucose exhibited significant differences (P鈥�<鈥�0.001) in baseline characteristics across BMI categories. As BMI increased, triglyceride and fasting glucose levels, as well as the TyG index, showed an upward trend. Additionally, the incidence rates of hypertension, hypercholesterolemia, diabetes, and CVDs increased with rising BMI levels.

Table 1 Characteristics of the study population based on BMI

Association between SES/BMI and TyG index

Table听2 presents the association between SES/BMI and TyG index in multivariate linear regression models. As a continuous variable, PIR was negatively correlated to TyG index in the crude model (尾=-0.03, 95% CI: -0.03~-0.02, P鈥�<鈥�0.001). Following adjustments for covariates, the correlation remained consistent in both Model 1 and Model 2(尾= -0.04, 95% CI: -0.05~-0.03, P鈥�<鈥�0.001; 尾= -0.02, 95% CI: -0.03~-0.02, P鈥�<鈥�0.001) (Table听2).

Table 2 Association between SES and BMI with TyG index in the multiple regression model

When PIR was examined as a categorical variable, individuals in the highest PIR tertile exhibited a 0.1 lower TyG index compared to those in the lowest tertile in the crude model(尾 = -0.10, 95% CI:-0.13~-0.06, P鈥�<鈥�0.001); Following adjustments in both Model 1 and 2, the association between PIR and TyG remained marginally significant (尾 =-0.17, 95% CI:-0.21~-0.13, P鈥�<鈥�0.001 and 尾 = -0.11, 95% CI:-0.15~-0.07, P鈥�<鈥�0.001, respectively).

The relationship between education level and the TyG index was investigated subsequently. A negative connection between the TyG index and education level was discovered during the investigation. Specifically, individuals with the highest level of education exhibited a 0.12 lower TyG index compared to those with the lowest level of education in the fully adjusted Model 2 (尾 = -0.12,95% CI: -0.15~-0.09, P鈥�<鈥�0.001). Simultaneously, the multivariate linear regression analysis results unveiled a robust positive correlation between BMI and the TyG index(尾鈥�=鈥�0.49, 95% CI: 0.46鈥墌鈥�0.52, P鈥�<鈥�0.001). The relationship remained notable and consistent across all models presented in Table听2.

Furthermore, stratified and interaction analyses were conducted to assess the durability of the correlation between SES and the TyG index in the fully adjusted Model 2. Importantly, there was no notable interaction effect observed between SES and the TyG index after stratifying the data by sex, age, physical activity, hypertension, and CVD. This further verifying the associations between PIR or education level and TyG index (Figs.听2 and 3).

Fig. 2

Associations between PIR and TyG index in different subgroups

Except for the stratification component itself, each stratification factor was adjusted for sex, age, marital status, smoking status, drinking consumption, physical activity, hypertension, and CVDs. Stratified and interaction analysis indicated the association between PIR and TyG index was robust in different subgroups

Fig. 3

Associations between education level and TyG index in different subgroups

Except for the stratification component itself, each stratification factor was adjusted for sex, age, marital status, smoking status, drinking consumption, physical activity, hypertension, and CVDs. Stratified and interaction analysis indicated the association between education level and TyG index was robust in different subgroups

Mediation analysis

Further examination of the mediation analysis scrutinized the role of BMI in mediating the association between PIR/Edu and the TyG index. Table听3 presents the TE, depicting the overall impact of SES on the TyG index; While DE illustrating the influence of SES on the TyG index independent of BMI; and the IE, indicating the influence of SES on the TyG index mediated by BMI. Generally, the DE significantly outweighed the IE, although the latter was statistically significant. The proportion of BMI mediating the effects of SES on the TyG index was calculated to be 14.4% and 8.57%, respectively (Fig.听4). Furthermore, we conducted bootstrap mediation analysis with 1,000 resampling iterations, which confirmed the stability of our findings. Detailed results are provided in Supplementary Table 1.

Table 3 BMI as a mediator in the associations of SES with TyG index

Fig. 4

Estimated proportion of the association between SES and TyG index mediated by BMI

Model was adjusted for: Sex, age, marital status, drink status, smoke status, physical activity, Hypertension, and CVDs. Convert education level into 2 categorical variables for analysis, high school/less than high school and some college or more. Mediation proportion鈥�=鈥塈E/DE鈥�+鈥塈E

Discussion

This study conducted an analysis on a nationally representative subset of Americans to assess the association between SES and the TyG index. The results indicate that in the cross-sectional survey involving 11,358 participants, lower SES among American adults is independently associated with higher TyG index. Furthermore, the results of this research indicate that BMI mediates the association between SES and the TyG index, with the proportion of mediation in PIR and educational level being 14% and 8.57%, respectively. In all adjusted models, these associations are independent of major underlying risk factors, including sex, age, marital status, drink status, smoke status, physical activity, hypertension, and CVDs.

The profound influence of SES on non-communicable disease risk factors, as well as the occurrence and fatality rates of cardiovascular and cerebrovascular diseases, has been firmly established [20, 21, 22, 23, 26鈥�27]. The current findings are in accordance with prior research, revealing an independent negative correlation between SES and the TyG index, which holds implications for numerous chronic conditions. Therefore, comprehending the mechanisms by which SES impacts the TyG index is essential for informing public policy towards effective and timely prevention strategies. Presently, the precise manner in which SES influences the TyG index remains uncertain.

Education, as a core indicator of SES, has been widely utilized in the field of epidemiological research. Nonetheless, the correlation between education and health remains a topic of debate. Several studies suggest that elevated education levels are frequently correlated with enhanced health outcomes. Consistent with the study findings, Stephens et al. noted that individuals with higher education levels tend to exhibit improved metabolic health and significantly reduce the risks of waist circumference, systolic blood pressure, blood glucose, glycosylated hemoglobin, triglycerides, high-density lipoprotein abnormalities, and metabolic syndrome compared to those with lower education levels [28]. Furthermore, another study indicated a notable inverse correlation between education and the incidence of heart disease among non-Hispanic white populations in the United States [29]. However, in a large-scale study of middle-aged individuals in China, education was identified as a potential risk factor for dyslipidemia and heart disease [30]. Similarly, research on middle-aged individuals in Japan [31] and male individuals in South Korea [32] also found a notable positive correlation between education level and the Incidence of hyperlipidemia. The inconsistency in these research findings may stem from differences among countries, regions and races.

Income, as a core indicator of SES, is closely linked to a variety of significant health issues. Existing study has confirmed a clear causal relationship between higher household income and improved cardiovascular biomarkers, which in turn contributes to reducing the incidence of cardiovascular diseases [33]. However, It is also crucial to acknowledge that the relationship between low income and heart disease is significant [34]. Further multivariate meta-analyses have revealed positive correlations between middle-low income levels and elevated risk of coronary artery disease, stroke, cardiovascular events, and cardiovascular-related mortality [35]. The findings of this study imply an independent negative correlation between PIR and the TyG index. These findings not only underscore the important role of income status in maintaining individual health but also provide important evidence for the formulation of public health policies to further reduce the negative impact of socioeconomic inequality on health.

The specific mechanisms underlying the negative correlation between SES and the TyG index are not yet fully understood, but might involve several factors. Firstly, education level significantly influences individuals鈥� dietary behaviors [36,37,38]. Studies have shown that individuals with higher levels of education tend to prioritize health and price factors in their dietary choices [37鈥�38], which may indirectly lead to changes in TG and/or blood glucose indicators [39], subsequently affecting the TyG index. Additionally, lower-income individuals may opt for low-cost foods due to economic pressures, resulting in reduced intake of fruits and vegetables [40鈥�41], which could also have a negative impact on the TyG index. Furthermore, another study found a clear correlation between food insecurity and food addiction in middle- to low-income adults [42], further impacting overall physical health. Secondly, high SES can provide individuals with more important resources such as better environmental conditions, healthy foods, and health insurance [43]. Lastly, Individuals with higher incomes often experience lower levels of psychological stress, are more likely to maintain healthy dietary patterns and lower weight [44,45,46], contributing to maintaining a healthy TyG index. However, the current understanding of these mechanisms is not comprehensive, and more investigation is needed to delves deeply into and clarify the specific relationships among them.

The associations between SES and metabolic health outcomes are well established, but the biological mechanisms underlying these associations are less understood. Low SES is commonly linked to inactivity and psychosocial stress, which in turn impact the blood levels of endogenous molecules such as sugars, lipids, inflammatory proteins, and reactive oxygen species, thereby significantly increasing the risks of chronic diseases [47]. Consistent with the findings, Kunz-Ebrecht found that low SES has been associated with elevated morning cortisol levels and diurnal cortisol secretion [48], lending support to the notion that lower SES is related to various work-related stressors, which contribute to a higher incidence of mental disorders and health-detrimental behaviors [49]. Another study found significant relationships between SES and stool 16听S rRNA microbiota composition in a large sample of British twins. Associations were observed in models adjusting for health-related factors known to impact the microbiome, with suggestion that individual-level SES attenuate microbiota鈥揾ealth associations [50].

Multiple studies have demonstrated a notable association between low SES and elevated BMI [28, 51,52,53]. Additionally, BMI has long been recognized as closely related to the TyG index [2, 54]. Given these known associations, this study analyzed whether BMI potentially exerts a moderating influence on the relationship between SES and the TyG index. These findings indicate that although the proportion of BMI as a mediating factor is relatively small, it does partially mediate the link between SES and the TyG index. This suggests that when exploring the relation between SES and the TyG index, the influence of BMI needs to be fully considered. However, there may also be other important factors beyond BMI, such as genetic factors, dietary intake, and systemic inflammation levels, which may also play significant roles in the relationship between socioeconomic status and the TyG index. Therefore, subsequent investigations should delve into these potential factors to garner a more thorough comprehension of how SES affects individual health.

Our study demonstrates a correlation between SES, BMI, and the TyG index. To effectively mitigate this health disparity, comprehensive public health intervention measures are imperative. These measures include, but are not limited to, improving public education levels, promoting healthy lifestyles, and actively controlling BMI. Primary healthcare services, as a critical pillar of the national health system, play an indispensable role. They are not only responsible for delivering basic medical services but also serve a key function in advancing health education and chronic disease prevention efforts, particularly among populations with lower socioeconomic status and higher metabolic health risks [55]. Therefore, public health policymakers should fully recognize the unique value of primary healthcare services in addressing health disparities and chronic disease prevention. By doing so, they can enhance the overall metabolic health of the population.

Strengths and limitations

This study has certain limitations. Primarily, the study is based on data from the United States, it remains uncertain whether these results are generalizable to other countries and populations. Secondly, given the cross-sectional design of the NHANES dataset, this study cannot capture the temporal order of events, thus precluding the establishment of a causal link between SES/BMI and the TyG index; causality remains undetermined. Additionally, despite rigorous methods employed to ensure data accuracy in NHANES, self-reported SES measures remain prone to recall bias. Thirdly, despite employing regression models and stratified analyses to account for potential confounders, there may still be residual confounding effects from unmeasured or unknown variables, such as dietary intake, drug use, psychological stress and genetic predispositions. Fourly, Although sensitivity analyses were conducted on the data, the issue of reverse causality, specifically the potential relationship between the TyG index and SES, remains fundamentally unresolved due to limitations in data structure and study design. Moreover, this study did not perform multiple comparison correction analysis, and these shortcomings affected the accurate of statistical significance. In the future, it is imperative to adopt more rigorous statistical analysis methods(instrumental variables, structural equation models, and sensitivity analysis, etc [56,57,58,59]) to ensure the reliability and validity of the results, and to verify these findings through longitudinal studies and randomized controlled trials. However, in spite of these constraints, the study has some advantages. It is the pioneering investigation into the correlation between SES and the TyG index. Additionally, the data utilized in this research is sourced from a comprehensive and representative cross-sectional survey. Robust covariate adjustments were applied, bolstering the reliability of the study outcomes.

Conclusion

In conclusion, a notable negative association was found between SES and the TyG index in the study. Moreover, BMI served as a partial mediator in the relation between SES and the TyG index. These findings deepen the comprehension of the association between SES and the TyG index. However, future high-quality longitudinal studies are needed to validate this relationship in a broader population. The study indicated that changes in the TyG index should be monitored in clinical settings, and more importantly, SES should be regarded as a focal point for regional as well as global health programs, health risk monitoring, actions, and regulations.