The association between obesity indicators and mortality among individuals with hyperlipidemia: evidence from the NHANES 2003–2018

Abstract

Background

Obesity is linked to a variety of metabolic issues, with hyperlipidemia being a crucial adjustable risk element for cardiovascular diseases (CVD). However, the connection between indicators of obesity with overall and CVD mortality in American adults with hyperlipidemia remains unknown.

Methods

This research employed an extensive cohort drawn from the National Health and Nutrition Examination Survey (NHANES) (2003–2018). Hyperlipidemia was identified through either elevated lipid profiles or self-reported utilization of lipid-reducing medications. Obesity indicators (weight-adjusted waist index (WWI), waist-to-height ratio (WHtR), body mass index (BMI)) were evaluated by physical measurement data. Weighted Cox regression models and restricted cubic splines (RCS) were employed to assess the potential links between obesity indicators and mortality outcomes. Results were further validated through subgroup analyses to ensure robustness and reliability. The receiver operating characteristic (ROC) curve was utilized to evaluate the prognostic capability of obesity indicators for mortality.

Results

This cohort study included data from 12,785 participants with hyperlipidemia. Over an average follow-up period of 8.4 years, a total of 1,454 deaths were documented, 380 of which were related to heart diseases. Cox analysis manifested that, after adjusting covariates, increased WWI was linked to a higher likelihood of overall and CVD mortality (both P < 0.05). RCS analysis illustrated that BMI and WHtR had U-shaped relationships with the overall and CVD mortality. Conversely, a linear positive association was uncovered between WWI and mortality (both P > 0.05 for nonlinearity). Age, alcohol consumption and chronic kidney disease had modifying effects on the relationship between WWI and total mortality among those with hyperlipidemia. The area under ROC indicated that WWI was more effective than for BMI and WHtR in predicting overall and CVD deaths.

Conclusions

In US adults with hyperlipidemia, the connection between BMI, WHtR, with overall and CVD mortality followed a U-shaped pattern, whereas a positive linear correlation was identified between WWI and mortality. WWI has superior predictive capability for the prognosis of individuals with hyperlipidemia compared to BMI and WHtR. These findings provide new insights and targets for the health management of individuals affected by hyperlipidemia.

Introduction

Hyperlipidemia is a prevalent metabolic disorder characterized by elevated concentrations of lipids in the bloodstream, significantly increasing the likelihood of numerous health issues, particularly cardiovascular diseases (CVD) [1]. Notably, hyperlipidemia is one of the few controllable detrimental factors associated with CVD, providing an opportunity for intervention. Given the financial burden and potential adverse effects associated with pharmacological treatments, lifestyle modifications emerge as a crucial and effective strategy for controlling lipid levels and preventing CVD [2]. However, the specific strategies and targets for improving outcomes in hyperlipidemic populations through lifestyle adjustments remain unclear.

Obesity stands as a critical hazard factor for both dyslipidemia and CVD progression [3]. The increase in body fat leads to impaired fatty tissue function and irregular fat distribution, causing detrimental metabolic, biomechanical, and psychological health effects [4]. These manifestations are characterized by an altered atherogenic lipid profile, including elevated triglyceride levels, lowered high-density lipoprotein-cholesterol (HDL-C), higher non-HDL-C and apolipoprotein levels, moderately heightened low-density lipoprotein-cholesterol (LDL-C), raised LDL particle concentration, and a higher proportion of small, dense LDL particles [5]. Additionally, adiposity exacerbates other cardiac hazardous factors such as diabetes, hypertension, and sleep disorders, all of which potentially contribute to an increased probability of CVD [6, 7]. Interventions aimed at weight reduction in obese individuals are typically associated with improvements in lipid profiles and CVD outcomes [8,9,10]. Body mass index (BMI) has long been utilized as a measure of obesity. However, it assesses “total body obesity” and some studies suggest that BMI is not a consistent adverse factor for predicting adverse cardiovascular outcomes [11,12,13]. In contrast to general obesity, central obesity is more strongly associated with various metabolic disorders and CVD [14, 15]. Mohseni et al. also indicates that central obesity and hyperlipidemia have a synergistic effect on the probability of developing hypertension [16]. The waist-to-height ratio (WHtR) and weight-adjusted waist index (WWI) have emerged as more effective tools for assessing central obesity. Despite the utility of these indicators in evaluating adiposity, there has been limited research examining the correlation between these metrics and the outcomes in individuals suffering from hyperlipidemia.

Therefore, this study was carried out to investigate possible links between obesity indicators and mortality among adults suffering from hyperlipidemia derived from National Health and Nutrition Examination Survey (NHANES). This correlation, if confirmed, may be useful in guiding weight control beneficial for lipid management and cardiovascular health.

Methods

Study population

The NHANES, executed by the National Center for Health Statistics (NCHS), stands for an extensive nationwide survey. It utilizes a layered, multilevel probability cluster sampling procedure to assess the health and nutritional condition of the non- institutionalized civilian populace in the US. The research protocols of this study secured approval from the NCHS Ethics Review Committee for Research. Informed consent was acquired from every participating individual.

Data was collected from participants in NHANES from 2003 to 2018. The criteria for exclusion were outlined as follows: age < 20 or age > 80, missing information on hyperlipidemia, body measurements or follow-up data. Finally, this research encompassed a total of 12,785 participants. The diagram illustrating the systematic selection procedure was presented in Fig. 1.

Fig. 1

Flow chart of the study subjects

The measurement of obesity indicators and definition of hyperlipidemia

The obesity indicators consist of BMI, WHtR and WWI. In NHANES, during the physical assessments, measurements of weight, height, and waist were recorded for participants. Then, the WHtR and WWI were calculated using the following formulas:

$$\:WHtR=\frac{waist\left(cm\right)}{height\left(cm\right)}\:WWI=\frac{waist\left(cm\right)}{\sqrt{weight\left(kg\right)}}$$

The original WHtR values range from 0 to 1. To enhance the readability and data visualization, WHtR was multiplied by ten before statistical analysis.

A diagnosis of hyperlipidemia was established when at least one of the subsequent five criteria was fulfilled: total cholesterol ≥ 200 mg/dL, triglycerides ≥ 150 mg/dL, LDL-C ≥ 130 mg/dL, or HDL-C ≤ 50 mg/dL in females and 40 mg/dL in males, or taking cholesterol-lowering drugs.

Ascertainment of death

Mortality data was sourced from the National Death Index database. The survival outcomes of the participants were monitored until Dec 31, 2019. All-cause mortality refers to death resulting from any reason, while CVD mortality was specified as death due to heart diseases (codes 054–064 in NCHS).

Covariates

In order to mitigate the impact of possible confounding variables, a number of covariates were carefully chosen, including gender, age, ethnicity, educational attainment, marital status, family income to poverty ratio (PIR) (< 1.3, ≥ 1.3), BMI, drinking status, smoking status, physical activity (PA), hypertension, diabetes, heart diseases and chronic kidney disease (CKD). Smoking status and alcohol consumption were determined by the survey questions. Smoking is characterized as having consumed over 100 cigarettes throughout one’s lifetime. The assessment of daily alcohol intake was derived from the consumption frequency and volume recorded over the preceding year, and then classified into heavy drinking (≥ 20 g per day for females; ≥30 g per day for males), mild drinking (0–20 g per day for females; 0–30 g per day for males), and never (did not consume a minimum of 12 drinks in the last year). PA data were obtained from a self-reported PA questionnaire and analyzed by calculating the overall metabolic equivalent minutes each week (MET-min/week). According to the Physical Activity Guidelines for Americans [17], PA intensity is categorized into three levels: inactive (< 600 MET-min/week), active (600–1200 MET-min/week), and highly active (> 1200 MET-min/week). Hypertension was defined if any of the subsequent assessment criteria were satisfied: average SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg, a physician diagnosis of hypertension, or the use of hypertensive medication. Diabetes was identified in individuals who either self-reported a diagnosis by a physician or had fasting plasma glucose ≥ 7.0 mmol/L or glycosylated hemoglobin ≥ 6.5%. Heart diseases were defined as having been informed of having at least one of the subsequent four issues: “coronary heart disease” or “congestive heart failure” or “angina” or “heart attack”. CKD was identified based on either an estimated glomerular filtration rate (eGFR) of < 60 mL/min/1.73 m² or a urine albumin-creatinine ratio > 30 mg/g [18]. The eGFR was computed utilizing the CKD-EPI 2021 equations.

Statistical methodology

This study employed a complex sampling design and sample weights in accordance with the guidelines established by the NHANES analysis. The computation of sample weights was performed utilizing the subsequent approach: fasting subsample mobile examination center (MEC) weight is determined by dividing the fasting subsample 2-year MEC weight by 8. In this study, qualitative variables were computed by the Chi-square test and presented as weighted percentages. Continuous variables with a normal distribution were computed by the t-test and shown as weighted means ± standard deviation (SD). Variables with skewed distributions were computed by the nonparametric Wilcoxon rank sum test and presented as weighted medians (interquartile range [IQR]). Weighted Cox regression models were employed to assess the relationship between adiposity indicators and the death rates of hyperlipidemia participants. The results are expressed as hazard ratio (HR) and 95% confidence interval (CI). The restricted cubic splines (RCS) were carried out to determine the linearity between obesity indicators and the death risk. Stratified analyses were used to verify the robustness of results. The receiver operating characteristic (ROC) curve was used to evaluate the forecasting capability of obesity metrics for mortality.

All statistical analyses were conducted using the Free Statistics analysis (Version 1.9.2) and R Statistical Software (Version 4.3.2). The statistical tests employed were two-tailed, with a significance threshold set at P < 0.05.

Results

Baseline characteristics of study participants

A total of 12,785 individuals met the eligibility criteria for the present study. The baseline characteristics categorized by survival status were featured in Table 1. Overall, the mean age was 49.9 years, with males comprising 47.4%. In comparison to survivors, those non-survivors were typically older, males and exhibited a lower PIR and educational attainment. They were also more inclined to be smokers and possessed a history of hypertension, diabetes, heart diseases and CKD.

Table 1 Baseline characteristics of the study population

Relationships between obesity indicators and overall mortality

Over a median monitoring period of 8.4 years, 1,454 (11.4%) participants with hyperlipidemia died, including 380 (3.0%) cardiovascular deaths. Weighted cox regression analysis in Table 2 revealed that, as a continuous variable, BMI exhibited no statistically significant association with total death risk in hyperlipidemia participants; as a categorical variable, compared to individuals with underweight, the total death is lower in groups with normal BMI, overweight, and adiposity (all P < 0.01). As continuous variables, the initial unadjusted analysis revealed a noteworthy escalation in the risk of overall mortality associated with higher WHtR*10 and WWI. Following multivariable adjustments, it was noted that a unit increase in WWI was associated with a 48% rise in risk for model 1 and a 21% increase for model 2 (both P < 0.001). Nevertheless, in model 2, the correlation between WHtR*10 and mortality failed to attain statistical significance (P = 0.901). When WHtR*10 and WWI were transformed into quartiles, the statistical relationships between WWI and overall mortality exhibited consistency, in comparison to the lowest quartile, elevated quartiles were correlated with a heightened risk of death (all models: P for trend < 0.001). The relationship between WHtR*10 and overall death was not significant in the model 2 (P for trend = 0.631). RCS in Fig. 2A demonstrated the positive linear correlation between WWI and overall mortality (P for non-linearity > 0.05), while BMI and WHtR*10 exhibit U-shaped correlations with overall death in participants with hyperlipidemia (P for non-linearity < 0.05) (Fig S1A and S2A).

Table 2 The association between obesity indicators and all-cause mortality in hyperlipidemia patients

Fig. 2

Association between WWI and all-cause (A) and CVD mortality (B) in patients with hyperlipidemia. The blue line represents OR, and the blue transparent area represents 95% CI

Survival curve analysis in Fig. 3A revealed significantly lower survival rates in higher WWI quartiles relative to the lowest quartile (P < 0.001). Stratified analyses in Fig. 4 were conducted to further validate the relationships between WWI and overall mortality in hyperlipidemia participants. Age, alcohol consumption and CKD history presented modifying effects on the connection between WWI and death risk (P for interaction < 0.05).

Fig. 3

Survival analyses of WWI on all-cause (A) and CVD mortality (B) in patients with hyperlipidemia

Fig. 4

Subgroup analysis of the association between WWI and the mortality of hyperlipidemia

Relationships between obesity indicators and CVD mortality

Cox regression analysis of obesity indicators and CVD death risk in hyperlipidemia sufferers was presented in Table 3. After adjusting for all covariates, irrespective of whether BMI is considered as a continuous variable or a classificatory variable, no statistically significant correlation has been identified between BMI and heart diseases death in the population with hyperlipidemia (P > 0.05). As a continuous variable, univariate analysis indicated CVD death risk significantly increased with higher WHtR*10 and WWI (both P < 0.001). After adjusting for potential confounders, per unit increase in WWI was linked to a 67% rise at risk of CVD death in model 1 and a 27% increase in model 2(both P < 0.05). In contrast, model 2 failed to reveal a statistically significant relationship between WHtR*10 and CVD mortality (P = 0.212). Upon analyzing WHtR*10 and WWI as categorical variables, it was observed that individuals categorized within the upper quartiles of WWI demonstrated an increased death risk in comparison to those situated within the bottom quartile of WWI (P for trend < 0.05). The connection between WHtR*10 and CVD mortality was not significant in the model 2 (P for trend = 0.365). Figure 2B demonstrated a positive correlation between WWI and CVD death risk in hyperlipidemia sufferers by RCS analysis, while BMI and WHtR*10 exhibit nonlinear relationships with CVD mortality in individuals with hyperlipidemia (Fig S1 B and S2 B).

Table 3 The association between obesity indicators and CVD mortality in hyperlipidemia patients

Figure 3B revealed the significantly reduced survival rates in higher WWI quartiles in comparison with the lowest quartile (P < 0.001). Subgroup analyses in Fig. 4 further verified the relationships between WWI and CVD death risk in hyperlipidemia sufferers. Findings indicated that age presented a modifying impact on the connection between WWI and CVD death risk (P for interaction < 0.05).

ROC curves of obesity indicators

Figure 5 manifested that WWI predicts all-cause and CVD mortality more effectively than BMI and WHtR in individuals with hyperlipidemia. According to the ROC curve, for overall mortality, WWI AUC value: 0.655, WHtR: 0.548, BMI: 0.538; for CVD mortality, WWI AUC value: 0.648, WHtR: 0.566, BMI: 0.505. A sensitivity analysis was conducted by excluding individuals who died within one year, and the subsequent ROC analysis revealed consistent trends with the previous findings (Fig S3).

Fig. 5

ROC curves of obesity indicators and all-cause (A) and CVD mortality (B) in patients with hyperlipidemia

Discussion

This is the first study to elaborate on the relationship between obesity indicators and mortality among individuals with hyperlipidemia. The main findings were as followings, (1) A positive linear correlation was identified between WWI and the risk of both overall and CVD mortality in individuals with hyperlipidemia. In contrast, the correlations of BMI and WHtR with death risks were nonlinear. (2) The ROC curves revealed that WWI demonstrated a better predictive ability regarding the prognosis of individuals suffering from hyperlipidemia when compared to BMI and WHtR. (3) Age, alcohol consumption, and CKD history had modifying effects on the connection between WWI and overall mortality in individuals with hyperlipidemia.

Dyslipidemia is a primary detrimental factor for atherosclerotic CVD. While adiposity is a chronic, progressive, and recurrent condition marked by an excessive buildup or unusual distribution of body fat [4]. Obesity, influenced by both genetic and environmental factors, can lead to various metabolic disorders and complications, causing damage to multiple tissues and organs throughout the body. Obesity management is crucial for overall and cardiovascular health, yet it continues to face numerous challenges. Firstly, treatment options for adiposity are limited, primarily focusing on lifestyle interventions. Secondly, long-term management goals remain unclear, with a lack of high-level evidence-based management targets. BMI has extensive clinical evidence and is relatively simple to measure, but relying solely on BMI to reflect adiposity management status and predict adverse outcomes has limitations [19, 20]. WWI and WHtR are relatively new obesity indicators, which can better reflect the overall distribution of fat. Previous studies have demonstrated their links with the occurrence and prognosis of various diseases [21,22,23]. The findings from this study suggested that, after adjusting for multiple covariates, elevated WWI correlated with increased risks of overall and CVD mortality among individuals with hyperlipidemia. Furthermore, RCS demonstrated a direct linear correlation between WWI and the risks of both total and CVD mortality. Additionally, the ROC curves indicated that WWI provides a more accurate prediction of the prognosis for individuals with hyperlipidemia compared to BMI and WHtR. Therefore, WWI can serve as a valuable monitoring indicator for obesity management in individuals with hyperlipidemia, potentially offering improved risk stratification and guiding more targeted interventions in this high-risk population.

Earlier studies demonstrated that the relationship between obesity and various diseases could be affected by variables like age and sex. A meta-analysis encompassing populations from multiple countries revealed a U-shaped correlation between BMI and overall mortality, with slight variations across gender, age and racial groups [24]. In recent years, several investigations have introduced the notion of the “obesity paradox”, which suggests that overweight elderly individuals and those with certain specific illnesses may have better outcomes compared to those with normal weight or underweight [25, 26]. Cox regression analysis exposed that, in the hyperlipidemic population, the risks of overall and CVD death are lower for individuals with normal BMI, obesity, and overweight compared to those who are underweight. RCS analysis demonstrated a U-shaped correlation between BMI and mortality, which is consistent with previous research findings. These results indicate the necessity of strictly control BMI within an appropriate range for the hyperlipidemic population. Stratified analyses were also conducted to investigate the impact of gender, age, education level, drinking and smoking status, and common chronic diseases on the relationships between WWI and mortality among individuals with hyperlipidemia. Results indicated that age, alcohol consumption and CKD history disturbed the relationship between WWI and total mortality in hyperlipidemia sufferers. Compared to younger individuals, per unit increase in WWI among older individuals was associated with a smaller increase at risk of overall and CVD death, further supporting the evidence for the obesity paradox. A large-scale clinical study has discovered that higher BMI and lower waist are associated with reduced mortality in individuals older than 80 years [27]. This indicates that guidelines and goals for adiposity management in the elderly population should be determined with caution, and further research is needed to validate whether WWI is a suitable indicator. Additionally, heavy drinkers with hyperlipidemia exhibited increased overall mortality risk with higher WWI relative to light drinkers. This finding can be attributed to alcohol’s disruption of lipid metabolism through increased lipolysis and fatty acid efflux from fatty tissue [28]. Moreover, acetaldehyde, an intermediate product of ethanol metabolism, can induce endoplasmic reticulum stress and mitochondrial dysfunction, leading to oxidative stress and increased vulnerability of various systemic diseases [29, 30]. Interestingly, CKD history moderated the relationship between WWI and all-cause mortality in hyperlipidemia sufferers; those without CKD showed higher overall death risk as WWI increased. This may be explained through several factors. The adaptive responses of the body may enable individuals with CKD to better combat the effects of diseases in certain situations, which could, to some extent, delay mortality. Additionally, individuals with CKD may be more likely to receive proactive medical management and interventions, thereby lowering the likelihood of certain complications and reducing mortality. However, it is important to note that this does not imply that CKD serves as a protective factor for obese individuals with hyperlipidemia. More clinical research and long-term observations are necessary to achieve a thorough comprehension of the interaction between CKD and mortality in hyperlipidemia individuals.

The impact of obesity on outcomes in hyperlipidemic populations likely involves multiple potential mechanisms. Primarily, adiposity leads to a series of direct pathological changes. Adipose tissue is recognized as a secretory organ that significantly contributes to the regulation of metabolism, endocrine function, immunity, and inflammation. Obesity enhances fatty acid storage in key circulatory and metabolic organs, leading to lipotoxicity and causing systemic oxidative stress, inflammation, and metabolic disturbances. Fatty tissue in individuals with obesity is linked to increased secretion of pro-inflammatory proteins such as interleukin-6 (IL-6), tumor necrosis factor-α, and IL-18. Conversely, the fatty tissue of lean individuals predominantly produce anti-inflammatory molecules like transforming growth factor-β, IL-4, IL-10, and IL-13 [31]. Moreover, excessive epicardial fatty tissue surrounding the heart can cause local dysregulated adipokine secretion, pro-inflammatory cytokines synthesis, and modifications in genetic expression. Such alterations result in endothelial dysfunction, cardiac functional and structural remodeling, facilitating the progression of coronary vascular disease, atrial arrhythmia, and heart failure [32,33,34]. Furthermore, adiposity serves as a significant contributing factor to other known cardiovascular detrimental factors, including diabetes, hypertension, and CKD. The presence of these comorbidities in obese individuals with hyperlipidemia can significantly exacerbate their cardiovascular risk and worsen outcomes. Certainly, with the increase in the population suffering from obesity, a substantial amount of foundational research on metabolism and diseases is currently being conducted, which will further elucidate the precise mechanisms linking obesity and diseases.

Advantages and limitations

This study has several remarkable advantages. Firstly, it is derived from NHANES data, which employs stratified multistage probability sampling and offers a prolonged follow-up duration, enhancing the representativeness and persuasiveness of these findings. Secondly, confounding variables were adjusted to minimize bias and obtain more reliable results. Additionally, stratified analyses were executed to demonstrate the reliability of the relationship between WWI and overall and CVD mortality among individuals with hyperlipidemia. However, this study also presents some shortcomings. Given the source of NHANES data, this sample population is primarily based on the US population, which may not be generalizable to global populations. Beyond the ROC analysis, no additional analyses were performed after excluding samples where deaths occurred within a specified period. Furthermore, despite adjusting for various potential confounding variables, it is possible that additional unmeasured confounders exist that may affect the connection between adiposity-related factors and mortality in individuals with hyperlipidemia. Despite these limitations, the findings offer significant understanding regarding the relationship between obesity metrics and the risk of death among individuals suffering from hyperlipidemia.

Conclusions

In summary, this study demonstrated the positive correlation for the first time between WWI and overall and CVD mortality in participants suffering from hyperlipidemia using the NHANES database. This finding offers more advantageous targets for lifestyle improvements and obesity management that contribute to long-term health in the hyperlipidemic population.