Introduction

The World Health Organization reported that depression, one of the most common mental diseases, affected more than 264 million people worldwide, which can be a major health challenge for individuals and an enormous societal burden [1]. The etiologies of depression are multifactorial, including biological, psychological, and social factors [2]. And previous studies have indicated that socioeconomic status (SES) has a significant influence on factors related to depression [3]. Individuals with low SES may be exposed to more adversity but have fewer resources to cope with depression [4]. However, these results are inconsistent and the influencing factors are complex [5].

Cardiovascular health (CVH) is commonly considered a factor that influences both SES and depression. Evidence suggests that less ideal CVH is associated with depression, and interventions targeting diet, physical activity, and sleep may ameliorate depressive symptoms [6, 7]. Meanwhile, individuals with low SES are often exposed to unhealthy lifestyles, which in turn significantly increase their susceptibility to cardiovascular disease [8]. However, important gaps remain. Previous studies tended to use the risk of cardiovascular disease including angina, arrhythmias, and left ventricular dysfunction to represent individual CVH, ignoring a broader, more positive construct: the CVH of individuals without disease [9]. Life’s Essential 8 (LE8) is an approach to measuring and monitoring CVH developed by the American Heart Association [10]. Building on the original metrics (Life’s Simple 7), LE8 updates the algorithm for each metric and adds the sleep-health model to reflect CVH more accurately. Furthermore, it is still unclear whether the association of CVH with SES and depression varies among subpopulations of different age, sex, and racial groups.

Therefore, the present study aimed to examine the intricate relationship between SES and depression among adult participants in the National Health and Nutrition Examination Survey (NHANES) database, and further evaluate the mediating effect of LE8 in this chain.

Methods

Data sources and the study population

Cross-sectional data was collected from three cycles (2013–2018) of the National Health and Nutrition Examination Surveys (NHANES) dataset, a nationwide health survey of the non-institutionalized, civilian, U.S. population. The NHANES sample is drawn in four stages: (a) PSUs (counties, groups of tracts within counties), (b) segments within PSUs (census blocks), (c) dwelling units (households) within segments, and (d) individuals within households. Screening is conducted at the dwelling unit level to identify sampled persons, based on oversampling criteria. All procedures were approved by the Research Ethics Review Board of the National Center for Health Statistics, and written informed consent was obtained from all participants. Out of 29,400 adults who participated in the NHANES 2013–2018, 4745 participants with complete depression-screener data, the values used to calculate Life’s Essential 8 scores, and a family-income-to-poverty ratio were included in the present study (Fig. 1).Missing data associated with the selected variables constituted less than 10% of the full sample (Supplementary Figure S1) and was compensated for by the use of multiple imputations.

Fig. 1
figure 1

Flow chart for the selection of included sample

Full size image

Depression assessment

Depression was assessed using the Patient Health Questionnaire (PHQ-9), which included nine questions about depressive symptoms over the previous two weeks. The responses ranged from 0 (“not at all”) to 1 (“several days”), 2 (“more than half the days”), and 3 (“nearly every day”), with total scores ranging from 0 to 27 [11]. According to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders, PHQ-9 scores of 10 or higher constitute clinically relevant depression (CRD), with a specificity and sensitivity of 88% [12].

Measurement of Life’s Essential 8

LE8 is an enhanced approach used to assess the construct of cardiovascular health (CVH). The components of Life’s Essential 8 include four health behaviors (diet, physical activity, nicotine exposure, and sleep health) and four health factors (body-mass index [BMI], blood lipids, blood glucose, and blood pressure). The detailed algorithms used to calculate LE8 scores can be found in Supplementary Table S1 [13]. The LE8 scores, which range between 0 and 100, represent the average of each of the 8 metrics.

Socioeconomic-status assessment

The ratios of family income to poverty (FIPR), occupation, educational level, and health insurance were used to evaluate SES [14]. FIPR was calculated in accordance with poverty guidelines published by the Department of Health and Human Services (HHS). Participants whose reported income was < $20,000 or ≥ $20,000 were excluded from the sample. The variables were divided into two or three levels, based on a practical interpretation and the sample size within levels (Supplementary Table S2). The SES was created using a latent class analysis to generate an unmeasured variable, based on four categorical variables above. Akaike information criterion (AIC) and Bayesian Information Criterion (BIC) decreased when the latent classes were added; these two indexes reached the bottom and rebounded when the latent class reached 3 (Supplementary Figure S2 A). The G2 statistics continued to go down when the latent classes were added; this decrease leveled off after the three-latent-class solution (Supplementary Figure S2 B). After considering the statistics related to model selection and the meanings of latent classes, we chose the three-latent-class solution. The participants were divided into three grades (High SES, Medium SES, and Low SES) (Supplementary Table S3).

Covariates

The study covariates, including sex, age, race, marital status, family size, alcohol consumption, and stroke are summarized in Supplementary Table S4. These potential confounding factors are presented in the section on demographic data. Stroke was defined as self-reported physician diagnosis of stroke.

Statistical analysis

Given the complex sampling design of NHANES, all analyses in this study accounted for sample weights, clustering, and stratification. And a new sample weight was constructed in accordance with the NHANES analytical guidelines. Missing data were addressed via multiple imputation, using the R package “VIM.” A latent class analysis was conducted using the R package “poLCA.” The tolerance value for judging the point of convergence was set to 1E-10, while the maximum iterations were set to 1000. The model selection was based on the AIC, BIC, and likelihood ratio statistic G2.

For continuous variables, Shapiro-Wilk tests were used to confirm normality. Non-normally distributed data were presented as median (interquartile range, IQR), while Mood’s test was used to compare the CRD and non-CRD group levels. The categorical variables were presented as the number of cases and composition ratio (n [%]); chi-square tests were used to compare the percentages of these variables in different groups.

A multivariate logistic regression analysis was performed to analyze whether SES and LE8 were associated with CRD. The low FIPR group was used as the reference group for higher SES. The results are expressed as an odds ratio (OR) and corresponding 95% confidence interval (CI). Hosmer-Lemeshow test was used to assess the goodness of fit of logistic regression models. A multiple linear regression analysis was used to measure the association between SES and LE8, with the results reported as β and corresponding 95% CI. An F-test was applied to check the assumptions of the linear regression analysis. In this analysis, Model 1 was not adjusted for covariates; Model 2 was adjusted for sociodemographic variables, including sex, age, race, marital status, and family size. Model 3 was further adjusted for variables that likely influenced the results, including alcohol consumption and stroke, based on Model 2.

The mediating effect of LE8 on the association between SES and CRD was determined using the R package “mediation.” The path model in Supplementary Figure S3 indicates the mediating effect of periodontal measures. All statistical analyses were conducted by R 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). Differences with a two-sided P < 0.05 were considered statistically significant when there were more than four latent classes.

Results

Table 1 summarizes the demographic characteristics of the participants, categorized according to the CRD. There were 4745 participants in this study (30.348% aged 20–39 years, 34.542% aged 40–59 years and 35.111% aged ≥ 60 years; female: male ratio was 1:0.906), with 415 CRD participants. In general, participants with CRD are more likely to be female; have higher BMI; spend less time on physical activity; have a lower educational level; no domestic partner and small family size than participants without CRD. In addition, significant differences were also observed among the participants in different groups based on alcohol consumption, smoking, sleep health, occupation, health insurance, diabetes, stroke, and daily dietary intake including energy, protein, dietary-fiber, magnesium, sodium, and potassium (P < 0.05).

Table 1 Characteristics of the included participants. (n = 4745)
Full size table

Table 2 summarizes the data on the association between SES and CRD, and no significant difference of Hosmer-Lemeshow tests were observed in all models (P > 0.05). In contrast to the low SES group, the high SES group were negatively correlated with CRD in all three models including Model 1 (OR = 0.199; 95% CI: 0.132 to 0.302, P < 0.01), Model 2 (OR = 0.206; 95% CI: 0.133 to 0.319, P < 0.01), and Model 3 (OR = 0.21; 95% CI: 0.136 to 0.325, P < 0.01). And the correlation still significant in subgroup analysis stratified by both sex and age. However, after adjusting for all covariates, the results of the subgroup analysis stratified by race showed that SES was inversely associated with CRD in all participants apart from Mexican Americans (MA) (OR = 0.191; 95% CI: 0.027 to 1.338, P = 0.09).

Table 2 Associations of socioeconomic status with clinically relevant depression
Full size table

The association between LE8 and CRD is shown in Table 3 and Supplementary Table S5. In the total sample, LE8 was negatively correlated with CRD in all three models [Model 1 (OR = 0.961, 95% CI: 0.952 to 0.971, P < 0.01); Model 2 (OR = 0. 96, 95% CI: 0.949 to 0.971, P < 0.01); Model 3 (OR = 0.962, 95% CI: 0.951 to 0.972, P < 0.01)]. And the correlation still significant in all subgroup analyses (P < 0.05). The results of Hosmer-Lemeshow test were not statistically significant in all models (P > 0.05).

Table 3 Associations of Life’s Essential 8 score with clinically relevant depression
Full size table

Table 4 presents the results of multivariate linear regression models between SES and LE8. Compared with those with low SES, participants with high SES had a higher LE8 scores in all three models [Model 1 (β = 10.407, 95% CI: 8.595 to 12.22, P < 0.01); Model 2 (β = 10.937, 95% CI: 9.115 to 12.76, P < 0.01); Model 3 (β = 10.852, 95% CI: 9.115 to 12.589, P < 0.01)]. Additionally, in subgroup analyses stratified by sex and age, the trends remained the same in all three models (P < 0.05). However, in subgroup analyses stratified by race, the association of SES with LE8 is significant in all three models in participants except for MA. The F-test results were statistically different in all models (P < 0.05).

Table 4 β (p) of Life’s Essential 8 score according to socioeconomic status
Full size table

Table 5 reveals the mediation effect of LE8 on the association between SES and CRD. After all covariates adjustment, LE8 was estimated to mediate 22.13% of the association of SES with CRD. Meanwhile, the mediation effect of LE8 was statistically significant in the subgroup analyses divided by age and gender and the mediating effect ranged from 17.95 to 41.45% in Model 3. While in the subgroup analysis stratified by race, the mediating effect of LE8 could only be found in participants except for MA. In different race species, the mediating effect of LE8 ranged from 13.23 to 33.98% in fully adjusted models.

Table 5 The mediating proportion of Life’s Essential 8 score on the association between socioeconomic status and clinically relevant depression among participants
Full size table

Discussion

The present study investigates the association between SES and CRD and the mediating role of LE8. Community-dwelling adults in the United States with low SES exhibit more severe depressive states than those with higher SES, and the association could be influenced by race of the participants. This relationship between SES and CRD was significant in participants except for MA. In addition, LE8 significantly mediated the association between SES and CRD in participants apart from MA.

Socioeconomic inequity in depression has been widely discussed. A cross-sectional study involving 5969 Korean participants aged 60 or older found that the deleterious effect of a low material standard of living on social cohesion could indirectly influence depression in older adults [15]. Evidence based on the Iranian Prospective Epidemiological Research Studies suggests that participants with low SES are more likely to experience anxiety and depressive symptoms [16]. Furthermore, similar conclusions were also observed in a European collaborative research on ageing examined individuals aged 18 or older [17]. Inconsistent with previous studies tended to use single variables, we construct a comprehensive SES in this study and confirm the association between SES and CRD. However, this association could be influenced by race and no significant association were observed in MA participants.

The association of SES with CRD could be partially mediated by CHV. Existing research has demonstrated that cardiovascular mortality was significantly higher in the low-medium SES group than in the high SES group in the National Health Insurance Service national sample cohort of South Korea [18]. The increased cardiovascular disease burden in populations with low SES is associated with biologic, behavioral, and psychosocial risk factors, which are more prevalent among disadvantaged populations [19, 20]. Mechanistically, individuals with low SES encounter difficulties in accessing abundant resources including knowledge, wealth, power, prestige, medical services, positive social relationships, and recreational facilities [21,22,23,24]. And these factors can further impact the cardiovascular disease of individuals [14]. In addition, previous study also indicated that depression is significantly correlated with poor CHV assessed by the American Heart Association 2010 [6]. At present, there is no consensus on the underlying mechanism of depression in relation to CHV. From a behavioral perspective, individuals with depression often engage in unhealthy lifestyle choices including smoking, excessive alcohol consumption, poor diet, and lack of exercise, all of which are risk factors for cardiovascular health [25]. In this study, we chose LE8, a more comprehensive approach, to measuring CVH and found that approximately 20% of the association between SES and CRD can be explained by LE8. Meanwhile, the mediating effect of LE8 was still significant in different gender or age groups.

In our study, we also found that the mediating effect of LE8 does not significantly impact the association between SES and CRD in MA participants. However, the underlying mechanisms for this race-based difference are intricate and multifactorial. One possible explanation could be dietary factors which linked to both CVD [26] and depression [27]. Evidence suggests that, among children and adults, non-Hispanic white and black Americans consume more junk food than Mexican Americans [28]. Such differences in dietary patterns may have confounded our findings. Furthermore, minority ethnic group participants exhibit higher levels of anhedonia compared to non-Hispanic white participants [29]. For instance, individuals of Latino descent exhibit higher rates of anhedonia compared to African Americans and Chinese Americans [30]. Nevertheless, it worth to known that urgent and necessary measures must be taken to actively reduce socioeconomic inequalities in order to promote mental health.

The present study has several strengths. First, the sample size is large enough to support subgroup analyses with sufficient statistical power. Second, we constructed an overall SES variable to comprehensively evaluate the complex relations of SES with CVH and CRD. In addition, LE8 offers a comprehensive and scientifically backed framework to evaluate the CVH of populations including those without cardiovascular disease.

Nevertheless, we also acknowledge several limitations. First, most indicators were measured once and thus could not provide a complete representation of the average level at different times. In addition, the longitudinal relationship between SES, LE8, and CRD could not be analyzed, due to the cross-sectional research design. Second, some measurement errors were inevitable, as the information on SES and LE8 included a self-report component. Third, we were unable to use a highly detailed group of occupations to calculate SES scores, due to the ambiguous delineation of occupation in two cycles of the NHANES dataset (2015–2016 and 2017–2018).

Conclusion

The results of this study indicate that SES is negatively associated with CRD and this association could be influenced by race. Meanwhile, LE8 largely mediates the relationship between SES and the risk of CRD in participants except for MA. Appropriate SES should be provided not only for a more reasonable allocation of social resources, but also for effectively protecting the CVH and mental health of the population. And it can further reduce the public health burden. Based on the reasoned findings and limitations of the present study, these results should be further confirmed via a large prospective cohort study.