The ChinaHEART, a nationwide public health project, served as the data source. Detailed protocols have been published previously [21]. From 2016 to 2023, we enrolled 190,317 community-dwelling adults aged 35–75 years across 20 sites in Hunan Province, China. Participants completed standardized questionnaires (demographics, lifestyle, medical history, ect.) and underwent physical/laboratory examinations. After excluding individuals with missing key variables (anthropometrics, CMD status, socioeconomic factors, lifestyle characteristics, ect.), 93,801 participants were retained for analysis (Fig. 1).

Fig. 1.

Study enrollment flowchart.

Ethical approval was granted by Fuwai Hospital’s Institutional Review Board (No. 2014-574). All participants provided written informed consent.

Trained staff collected data using standardized protocols: (1) Questionnaires: Demographics, socioeconomic status (annual household income: $\ge$¥10,000 (US $\$$1,408) vs. ¥10,000; education: middle school and above vs. primary school and below), lifestyle (smoking: current/never; alcohol: frequent [more than 4 times per week] vs. non-frequent [never, once or less per month, 2–4 times per month, 2–3 times per week]) [22, 23], physical activity (meeting WHO guidelines [24]: yes/no), and diet (healthy/unhealthy per Chinese dietary guidelines [25]). (2) Anthropometrics: Height and waist circumference (cm) were measured using calibrated stadiometers with participants wearing lightweight clothing and having removed footwear and headwear [21]. RFM was calculated as: RFM = 64 – [20 $\times$ height (m) $÷$ waist circumference (m)] + (12 $\times$ sex), where sex = 1 (female) or 0 (male) [10]. Height and waist circumference were measured in centimeters but converted to meters for the RFM calculation. (3) Laboratory tests and medical history: Hypertension was defined as systolic/diastolic BP $\ge$140/90 mmHg, self-reported diagnosis, or antihypertensive use. Dyslipidemia [26] required TC $\ge$6.2 mmol/L, LDL-C $\ge$4.1 mmol/L, HDL-C 1.0 mmol/L, TG $\ge$2.3 mmol/L, or lipid-lowering medication.

CMD was defined as $\ge$1 of the following [4, 5, 6]: (1) Self-reported diabetes, or with the use of hypoglycemic agents/insulin; (2) Self-reported myocardial infarction; (3) Self-reported stroke.

Baseline characteristics were described for the total study population, young and middle-aged group (35–59 years), and older adult group ($\ge$60 years), including socioeconomic characteristics, lifestyle information, and medical history. Continuous variables (age, RFM) were tested for normality using the Kolmogorov-Smirnov test, which indicated non-normal distributions; therefore, these variables were presented as median (interquartile range) and compared between groups using the Wilcoxon rank-sum test. Categorical variables (sex, annual household income, education level, smoking status, alcohol consumption, physical activity level, diet, hypertension, and dyslipidemia) were presented as frequency (percentage) and compared using chi-square tests.

Multivariable logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between RFM and CMD risk. RFM was analyzed as a continuous variable to calculate ORs and 95% CIs. Then, participants were divided into quartiles (Q1–Q4) based on RFM values, with Q1 as the reference group, to calculate ORs and 95% CIs for Q2, Q3, and Q4 groups. The logistic regression models were adjusted as follows: Model 1 adjusted for age and sex; the full model additionally adjusted for annual household income, education level, smoking status, alcohol consumption, physical activity level, diet, hypertension, and dyslipidemia.

The interaction between RFM and age group (60 vs. $\ge$60 years) was tested in the full model. Stratified analyses were then performed in young and middle-aged, and older adult groups separately to examine the association between RFM and CMD risk, with results presented in forest plots for comparison. Restricted cubic spline functions were used to analyze dose-response relationships between RFM and CMD risk in each age group, adjusting for age, sex, annual household income, education level, smoking status, alcohol consumption, level of physical activity, diet, hypertension, and dyslipidemia.

All statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA) and R 4.4.3 software (R Foundation for Statistical Computing, Vienna, Austria). Two-tailed tests were used, with p 0.05 considered statistically significant.

This study enrolled a total of 93,801 participants (Table 1), with a median age of 59 years (interquartile range [IQR]: 52, 67). The cohort comprised 47,004 (50.11%) middle-aged and young participants and 46,797 (48.89%) older participants. No significant difference was observed in RFM distribution between the two age groups (p = 0.739), with median [IQR] values of 34.44 [27.26, 38.75] and 34.06 [26.12, 39.78], respectively.

Compared with the older group, the middle-aged and young group demonstrated significantly higher proportions of female participants (64.18% vs. 56.48%), annual household income $\ge$¥10,000 (92.20% vs. 81.32%), educational attainment of middle school and above (60.36% vs. 27.70%), and insufficient physical activity (77.40% vs. 76.12%) (all p 0.001). The older group exhibited a significantly higher prevalence of alcohol consumption (9.21% vs. 5.15%), and current smoking (24.51% vs. 22.80%) compared to the middle-aged and young group (all p 0.001).

In the total population, RFM demonstrated significant associations with increased CMD risk (Table 2). In the unadjusted model, a 1-unit increase in RFM was associated with a 3% higher risk of CMD (p 0.001). After adjusting for age and sex, the OR was 1.10 (95% CI: 1.10–1.11, p 0.001), which remained significant in the multivariable-adjusted model (1.09, 1.08–1.09, p 0.001). When RFM was categorized by quartiles, compared with the Q1 group, the Q2 (1.68, 1.59–1.77, p 0.001), Q3 (2.56, 2.34–2.80, p 0.001), and Q4 groups (4.02, 3.68–4.39, p 0.001) all exhibited significantly higher risk of CMD (p for trend 0.001).

Significant interaction effects were observed between RFM (p = 0.001) and RFM quartiles (p 0.001) with age groups in the multivariable logistic regression model (Fig. 2). A 1-unit increase in RFM was associated with CMD risk elevation in the middle-aged and young group (1.10, 1.09–1.11, p 0.001), as well as the older group (1.08, 1.07–1.08, p 0.001). When stratified by RFM quartiles, both age groups showed progressively increased CMD risks across higher quartiles (vs. Q1): Q2 (young and middle-aged: 1.69, 1.56–1.84 vs. older: 1.64, 1.53–1.75), Q3 (2.87, 2.50–3.28 vs. 2.27, 2.01–2.55), and Q4 (4.66, 4.06–5.35 vs. 3.45, 3.07–3.87) (all p 0.001).

Fig. 2.

Forest plot of subgroup analysis for the association between relative fat mass and cardiometabolic disease risk. Young: age 60 years group; Old: age $\ge$60 years group. Abbreviations: CI, confidence interval; CMD, cardiometabolic disease; OR, odds ratio; RFM, relative fat mass; Q1–Q4, quartile groups stratified by RFM interquartile ranges. Model adjusted for age, sex, annual household income, educational attainment, alcohol consumption, smoking status, physical activity, diet, hypertension, and dyslipidemia.

Restricted cubic spline analyses adjusted for age, sex, household income, education, smoking, alcohol use, physical activity, diet, hypertension and dyslipidemia revealed distinct patterns across populations (Fig. 3). In the total population, RFM exhibited a J-shaped association with CMD risk (p for overall 0.001, p for non-linear 0.001). Similar increasing trends were observed in both age groups, with differing curve morphologies. Both the young and middle-aged group (p for overall 0.001, p for non-linear = 0.030) and older group (p for overall 0.001, p for non-linear 0.001) displayed J-shaped associations. The older group demonstrated a steeper risk elevation gradient, with CMD risk (OR $>$1) emerging at RFM $>$30, whereas the inflection point occurred earlier (RFM $\approx$ 34) in the young and middle-aged group.

Fig. 3.

Dose-response relationship between relative fat mass and cardiometabolic disease risk. Young: age 60 years group; Old: age $\ge$60 years group. Abbreviations: CI, confidence interval; CMD, cardiometabolic disease; OR, odds ratio; RFM, relative fat mass. Restricted cubic spline models adjusted for sex, annual household income, educational attainment, alcohol consumption, smoking status, physical activity, diet, hypertension, and dyslipidemia.

This study is the first to report the association between RFM and CMD risk in Chinese adults, demonstrating a significant positive correlation. Previous studies have linked RFM to risks of coronary heart disease [11], stroke [12], diabetes [13, 14, 15], and metabolic syndrome [16], however, evidence on its association with CMD as a composite outcome remains scarce. CMD includes diseases with shared pathophysiological mechanisms and therapeutic targets, often presenting as comorbidities. For instance, metabolic inflammation serves as a central mechanism linking obesity, insulin resistance, and cardiovascular diseases. Adipose tissue—particularly visceral fat—secretes inflammatory cytokines (e.g., IL-6, TNF-$\alpha$), activating the TLR4/NF-$\kappa$B pathway, which induces insulin resistance, endothelial dysfunction, and atherosclerotic plaque formation [27, 28, 29, 30]. This systemic inflammation is not only responsible for the progression of diabetes, but also accelerates coronary heart disease and stroke via oxidative stress and lipid peroxidation. SGLT2 inhibitors, while improving glycemic control, also reduce cardiovascular mortality [31], underscoring potential shared therapeutic targets in CMD. Clinical evidence further supports the numerous co-morbidities associated with CMD: the CAPTURE multinational study found that 32.2% of type 2 diabetes patients had comorbid atherosclerotic cardiovascular disease (ASCVD), including coronary heart disease (16.0%) and cerebrovascular disease (7.7%) [32]. These findings justify analyzing CMD as an integrated entity to optimize comorbidity management and explore common therapeutic strategies. Building on prior research, this study provides critical evidence on the association between RFM and CMD as a composite outcome.

Previous investigations into the associations between RFM and CMD with diabetes, coronary heart disease, and stroke primarily focused on Western populations. This study fills a gap in Chinese evidence while accounting for potential confounders such as physical activity. Zwartkruis et al. [11] identified RFM as superior to BMI and waist circumference in predicting the risk of coronary heart disease in a Norwegian cohort of 95,000 adults. Zheng et al. [12] reported a positive association between RFM and stroke risk in the U.S. NHANES population, with the highest RFM quartile exhibiting a 44% increased stroke risk (OR = 1.44, 95% CI: 1.09–1.90) compared to the lowest quartile. However, their analysis lacked adjustment for physical activity, a known modifier of cardiometabolic risk [33, 34, 35, 36, 37], potentially influencing outcomes. Cacciatore et al. [15] demonstrated RFM’s superior predictive value over BMI for diabetes risk in 1900 older Italian adults. Similarly, Cichosz et al. [13] and Suthahar et al. [14] found RFM outperformed BMI, waist circumference, and waist-to-hip ratio in predicting diabetes risk in U.S. NHANES and Dutch cohorts, respectively. The present study extends these previous findings by providing robust evidence from a large Chinese population, systematically accounting for physical activity and other potential confounders, thereby offering more generalizable and refined insights into the RFM–CMD relationship.

This study identified significant age-related differences in the RFM-CMD risk association between young and middle-aged and older adults, addressing a critical gap in prior research that lacked comparative analyses across age groups. Although Suthahar et al. [14] reported stronger associations between RFM and the risk of type 2 diabetes in younger populations (based on higher hazard ratios in the 40-year group), they did not validate the statistical significance of this age-dependent association through formal interaction analyses. Similarly, Zheng et al. [12] observed increased RFM-stroke risk correlations in the 20–59-year subgroup (vs. non-significant associations in the 60–85-year group) but provided no mechanistic explanation.

Although the effect sizes for RFM increments were numerically similar between age groups, the significant interaction term, coupled with the different dose-response relationship and inflection points, indicates that the nature of the RFM-CMD association is fundamentally age-dependent. Combined with the higher baseline CMD risk in older adults, even a marginally greater OR per unit increase in RFM can translate into a more substantial increase in absolute risk at higher RFM levels. Therefore, the statistical interaction highlights a critical vulnerability in the elderly: their risk begins to escalate earlier and may compound more rapidly, underscoring the potential value of earlier and more vigilant RFM monitoring in this demographic.

The observed disparities may be partly explained by age-related physiological changes. Aging is associated with ectopic fat deposition in organs such as the liver and muscles, which may exert greater metabolic impact than visceral adipose tissue [18]. Older adults also tend to exhibit elevated baseline inflammatory markers (e.g., IL-6, CRP) [19] and distinct diabetes pathophysiology—primarily driven by $\beta$-cell dysfunction in older populations versus insulin resistance in younger groups [20]. These factors, together with RFM’s established correlation with visceral adiposity, may collectively contribute to the divergent RFM–CMD risk patterns across age groups. However, as our study did not include direct biomarker measurements, these mechanisms remain speculative. This study fills a critical evidence gap by demonstrating that while RFM-CMD risk correlations remain positive in both age groups, older adults exhibit steeper risk escalation with RFM elevation and higher absolute CMD risk at elevated RFM levels compared to young and middle-aged individuals.

This study provides novel insights into the non-linear dose-response relationship between RFM and CMD risk across multiple age groups, a previously underexplored area. Prior investigations primarily focused on ROC curve analyses comparing RFM’s predictive value against BMI for diabetes [15] and coronary heart disease [13]. While Zheng et al. [12] identified non-linear RFM-stroke risk associations using smoothing curve fitting, they did not employ restricted cubic spline analyses for formal dose-response characterization. Through restricted cubic spline modeling, this study revealed significant non-linear associations between RFM and CMD risk in both age groups. These findings advance our understanding of age-specific RFM-CMD risk dynamics, demonstrating distinct inflection points and risk gradients between young and middle-aged and older populations. These results highlight the potential value of establishing age-specific RFM thresholds for risk stratification and suggest that such thresholds may be warranted; however, future validation studies are needed to define clinically applicable cut-offs. This methodology overcomes the limitations of previous approaches by quantifying non-linear relationships while adjusting for confounders, establishing a robust framework for future investigations.

This study provides scientific evidence for the association between RFM and the risk of CMD, including age-specific patterns, offering a precise, simple, and usable metric for CMD risk assessment while providing age-stratified personalized intervention strategies. RFM, calculated using height, waist circumference, and sex, has demonstrated strong correlations with body fat percentage measured by DXA and BIA [38]. Its cost-effectiveness and ease of implementation compared to DXA/BIA make it a practical tool for evaluation of adiposity. Furthermore, RFM has been shown to outperform traditional anthropometric indices (e.g., BMI, waist circumference) in predicting cardiovascular risk factors [15, 16], cardiometabolic diseases [14, 39], and cardiovascular mortality [40]. The observed age-specific differences in RFM-CMD risk associations underscore the need for age-adapted intervention thresholds, suggesting stricter RFM control targets and intensified CMD risk management for older adults with elevated RFM. These findings provide a scientific foundation for precision prevention and control strategies for CMD.

This study has several limitations. First, a key limitation is the reliance on self-reported CMD without independent clinical validation, which may introduce misclassification bias. This potential bias may be more pronounced in older adults, who are more susceptible to under-reporting due to factors such as decreased awareness of asymptomatic conditions or barriers to healthcare access. If present, such non-differential misclassification would likely lead to an underestimation of the true association between RFM and CMD, meaning our observed significant associations are likely conservative estimates of the actual effects. It is important to note that several factors enhance the reliability of our data: the use of trained staff, standardized data collection procedures, and the fact that self-reported medical information was based on prior physician diagnosis. Additionally, the large sample size helps to mitigate the impact of random error. Second, as a cross-sectional study, our design precludes causal inference, and the observed associations should be interpreted as correlations rather than causal effects. Residual confounding may persist despite multivariable adjustments. Third, the absence of inflammatory markers and fat deposition data limits mechanistic exploration. Future research should prioritize incorporating such measures—for instance, using medical imaging to quantify ectopic fat or assays to profile inflammatory cytokines—to validate the proposed hypotheses and elucidate the biological pathways linking adiposity to CMD risk across the lifespan. Future validation studies are needed to define and evaluate age-specific RFM thresholds before they can be considered for clinical implementation. Furthermore, well-designed prospective cohorts, randomized controlled trials, and molecular-level studies are warranted to confirm these associations and elucidate underlying mechanisms and therapeutic targets.