Estimated glucose disposal rate outperforms inflammation surrogate markers in predicting cardiometabolic multimorbidity among middle-aged and older adults Authors: Cancan Cui, Xu Wang, Liqiang Wang, Kaiming Xue, Jinghui Cao, Lingyu Zhang, Siqi Yue, Shengtong Chen, Lin Liu, Yanxi Liu, Jianshi Du Published: 2026 ABSTRACT AbstractBackgroundBoth estimated glucose disposal rate (eGDR) and inflammation surrogate (high-sensitivity C-reactive protein [hs-CRP]) have emerged as potential markers reflecting metabolic health. However, their comparative predictive value for cardiometabolic multimorbidity remains unclear, particularly in middled-aged and older populations where both metabolic dysfunction and low-grade inflammation are highly prevalent.MethodsThis prospective cohort study included 7,589 participants aged ≥ 45 years from the China Health and Retirement Longitudinal Study (CHARLS) with complete baseline data. The primary outcome was incident cardiometabolic multimorbidity, defined as the co-occurrence of two or more cardiometabolic conditions, including diabetes, heart disease, and stroke. Cox proportional hazards models were employed to assess separate and joint associations. Concordance index (C-index) was used to evaluate the predictive capacity.ResultsOver a 9-year follow-up period, 2,263 participants developed cardiometabolic multimorbidity. In models adjusting for traditional risk factors, lower eGDR (lowest vs. highest quartile: hazard ratio [HR] 2.07, 95% CI 1.80–2.38) remained significantly associated with increased risk of cardiometabolic multimorbidity. In joint associations, participants with low eGDR exhibited consistently elevated cardiometabolic multimorbidity risk regardless of inflammation surrogate status (HR 1.52 (95% CI 1.33–1.74) in those with low inflammation and HR 1.80 (95% CI 1.58–2.04) in those with high inflammation). Addition of eGDR to the basic model (which included covariates) yielded a significantly higher C-index (0.676) compared to hs-CRP (0.643, P < 0.001).ConclusionsAmong middle-aged and older adults, eGDR is significantly associated with incident cardiometabolic multimorbidity regardless of inflammation status. Low eGDR demonstrates superior discriminative ability compared to hs-CRP. These findings suggest that eGDR may serve as a more comprehensive biomarker for cardiometabolic risk stratification by capturing insulin resistance and metabolic dysfunction.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12933-026-03249-2.Keywords: Estimated glucose disposal rate, hs-CRP, Cardiometabolic multimorbidity, Cohort study Background Cardiometabolic multimorbidity, defined as the co-occurrence of two or more cardiometabolic conditions, has emerged as a major public health challenge worldwide [1]. Recent global assessments estimate that 4.7% of adults aged 60 years and older are affected by two comorbid cardiometabolic diseases, responsible for substantial disability and premature mortality [2]. In China, the burden of cardiometabolic multimorbidity is increasing rapidly, with the prevalence of cardiometabolic multimorbidity nearly doubled within just five years [3]. By 2019, millions of Chinese adults live with multiple cardiometabolic conditions, accounting for a substantial proportion of healthcare utilization and mortality [4]. These statistics underscore the need for scalable biomarkers that capture the metabolic and inflammatory pathways driving cardiometabolic multimorbidity and improve risk stratification in adults. Insulin resistance candidates have gained attention because they reflect complementary axes of metabolic dysfunction [5]. The estimated glucose disposal rate (eGDR) integrates adiposity, glucose metabolism, and hypertension status and is a well-established surrogate of insulin resistance [6]. The eGDR and other insulin resistance indices were recently proposed and has shown associations with cardiometabolic outcomes, including diabetes, coronary heart disease, and stroke in diverse cohorts, as well as frailty [5, 7–9]. In parallel, inflammation surrogates capture the atherogenic burden and correlates with insulin resistance and pro-inflammation phenotypes [10]. Elevated high-sensitivity C-reactive protein (hs-CRP) is linked observationally and genetically to atherosclerotic disease and atherosclerosis, supporting a causal role in cardiovascular events [11,12]. Despite these advances, comparative evidence is limited [13, 14]. A recent study concluded that the eGDR index outperforms six other insulin indices in predicting cardiovascular disease in individuals with cardiovascular-kidney-metabolic syndrome [15]. Another study reported the joint effect between eGDR and systemic inflammation response index on mortality [16]. Our previous study also found that inflammation may mediate the association between other insulin resistance surrogate and cardiovascular disease [17]. However, evidence remains limited regarding the joint associations of eGDR, its components, and inflammatory markers with incident cardiometabolic multimorbidity and their comparative predictive performance. To address these gaps, we analyzed data from the China Health and Retirement Longitudinal Study (CHARLS) to evaluate the separate and joint associations between eGDR and hs-CRP with cardiometabolic multimorbidity. We also aimed to compare the discriminative capacity of eGDR and hs-CRP for predicting cardiometabolic multimorbidity among middle-aged and older populations. Methods Study population We conducted a prospective cohort study using data from the China Health and Retirement Longitudinal Study (CHARLS), an ongoing, nationally representative survey of Chinese adults aged ≥ 45 years. CHARLS employs a multistage, stratified, probability sampling design to select participants from both urban and rural areas across China. The baseline survey was conducted in 2011, with follow-up waves every two years to collect detailed information on sociodemographic characteristics, health status, lifestyle factors, and physical measurements. In addition, biomarker data were obtained from blood samples collected using standardized protocols and analyzed in certified laboratories. Detailed descriptions of the study design, sampling strategy, data collection, and laboratory procedures have been reported elsewhere (https://charls.pku.edu.cn/) [18, 19]. The CHARLS protocol was approved by the Peking University Biomedical Ethics Review Committee, and all participants provided written informed consent. For this analysis, we excluded participants who were aged < 45 years or had missing baseline demographic information; had a history of diabetes, heart disease, or stroke at baseline; had non-fasting blood samples or missing data on hs-CRP or HbA1c; participants with extremely elevated hs-CRP levels (i.e., > 10 mg/L) were also excluded; had missing information on hypertension status or waist circumference; or were lost to follow-up. The final analytic cohort included 7,589 participants (Fig. 1). Inclusion and exclusion flowchart for participants in the study Exposure assessment The eGDR was calculated using the previously validated formula: eGDR = 21.158 − (0.09 × waist circumference (cm)) − (3.407 × hypertension (yes = 1/no = 0)) − (0.551 × HbA1c (%)).20 Waist circumference was measured at the umbilical level with participants standing upright and breathing normally; a non-stretchable tape was placed horizontally around the abdomen at the level of the navel without compressing the skin [21]. Hypertension was defined as a self-reported history or current physician diagnosis of hypertension, or current use of antihypertensive medication [22]. HbA1c and high-sensitivity C-reactive protein (hs-CRP) were measured from fasting venous blood samples using standardized laboratory procedures, including enzymatic methods for HbA1c and high-sensitivity immunoassays for hs-CRP [23]. For regression analyses, eGDR and hs-CRP were categorized into quartiles. For joint analyses, each variable was dichotomized at its median, and participants were classified into four mutually exclusive groups [17]. Outcomes ascertainment Incident cardiometabolic conditions, including heart disease, stroke, and type 2 diabetes, were ascertained through follow-up interviews conducted in each survey wave. Heart disease was defined based on self-reported physician diagnosis of myocardial infarction (heart attack), coronary heart disease or angina, congestive heart failure, or other heart conditions. Stroke was defined as a self-reported physician diagnosis of stroke. Type 2 diabetes was defined as a self-reported physician diagnosis of diabetes or hyperglycemia in response to the question: “Have you been diagnosed with diabetes or hyperglycemia by a doctor?” The primary outcome was incident cardiometabolic multimorbidity, defined as the co-occurrence of two or more of the above conditions during follow-up. The onset date was assigned as the time point at which the participant first met the definition of cardiometabolic multimorbidity, that is when the second cardiometabolic condition occurred. The outcomes were assessed by trained interviewers and are harmonized to international leading aging surveys in the Health and Retirement Study as widely used in previous studies [24, 25]. Covariates Covariates were collected at baseline using standardized questionnaires, physical examinations, and laboratory assessments. These included age, sex, and residence (urban or rural), as well as socioeconomic factors such as educational attainment (primary school or below vs. secondary school or above) and marital status (married vs. others). Lifestyle factors included current smoking status, current alcohol consumption, and sleep duration (< 7 h, 7–9 h, > 9 h). Anthropometric measures included body mass index (BMI), calculated as weight in kilograms divided by height in meters squared. Medical history variables included physician-diagnosed hypertension and dyslipidaemia. Depressive symptoms were assessed using the validated 10-item Center for Epidemiologic Studies Depression Scale,[26] and renal function was evaluated using the estimated glomerular filtration rates by plasma creatinine levels [27]. The minimal covariate adjustment set was determined a priori based on directed acyclic graphs (DAGs), existing literature, and clinical relevance (Supplementary eFigure 1). Statistical analysis Continuous variables are presented as means with standard deviations (SD), and categorical variables as counts with percentages. Baseline characteristics were described across quartiles of estimated glucose disposal rate (eGDR) and across joint categories of eGDR and hs-CRP and compared using ANOVA test for continuous variables and the chi-square test for categorical variables. Given the low proportion of missing covariate data (< 5%), complete-case analysis was used for the primary analyses. Participants were followed from baseline until the first occurrence of cardiometabolic multimorbidity, death, loss to follow-up, or the end of follow-up, whichever came first. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), with attained age as the underlying time scale [24]. We conducted an additional sensitivity analysis using follow-up time based on survey cycles as the underlying time scale. The proportional hazards assumption was assessed using Schoenfeld residuals, and no substantial violations were observed (Supplementary eTable 1). Kaplan–Meier curves were generated to evaluate the cumulative incidence of cardiometabolic multimorbidity across exposure groups. Multivariable Cox models were fitted to assess the separate and joint associations of eGDR and hs-CRP with incident cardiometabolic multimorbidity. Model 1 adjusted for age and sex, and Model 2 further adjusted for residence, marital status, educational attainment, smoking status, alcohol consumption, sleep duration, BMI category, depressive symptoms, and renal function. Subgroup analyses were conducted stratified by age (< 65 and ≥ 65 years), sex, BMI category (normal weight and overweight/obesity), current smoking status, and residence (rural and urban), because these demographic, lifestyle, and metabolic factors are closely related to eGDR and CMM risk and may plausibly modify the observed associations. Multiplicative interactions were evaluated using likelihood ratio tests comparing models with and without interaction terms. Additive interactions were quantified using relative excess risk due to interaction (RERI). Sensitivity analyses included excluding participants who developed cardiometabolic multimorbidity within the first 1–2 years of follow-up to minimize reverse causation and excluding only participants with two or more of these conditions at baseline. We also considered death as the competing event using Fine-Gray sub-distribution models. To evaluate predictive performance, model discrimination was assessed using Harrell’s concordance index (C-index), brier scores, and calibration plots. A base model including traditional risk factors (minimal adjustment set) was compared with models additionally incorporating eGDR, hs-CRP, or both biomarkers. Bootstrap resampling was used to estimate confidence intervals for C-indices. To estimate the additional predictive power, the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated. All statistical analyses were performed using R (version ≥ 4.2), and a two-sided P value < 0.05 was considered statistically significant. Results Baseline characteristics A total of 7,589 participants (mean age 59.0 [SD 9.0] years; 52.3% women) were included in the final analysis (Fig. 1). Baseline characteristics stratified by eGDR levels are presented in Table 1. Participants with lower eGDR tended to be older, have higher BMI and waist circumference, and exhibit a higher prevalence of dyslipidemia. Table 1Characteristics of 7,589 participants according to eGDR levelsParticipants, No.Quartile 1Quartile 2Quartile 3Quartile 4P value1902189518951897Age, years, mean(SD)61.08(9.28)58.10(9.08)58.54(9.35)59.03(9.75)< 0.001Sex, n(%)< 0.001Female1064(55.9)1059(55.9)895(47.2)948(50.0)Male838(44.1)836(44.1)1000(52.8)949(50.0)Residence, n(%)< 0.001Rural1531(80.5)1560(82.3)1648(87.0)1695(89.4)Urban371(19.5)335(17.7)247(13.0)202(10.6)Married, n(%)1645(86.5)1701(89.8)1692(89.3)1674(88.2)0.08Educational level, n(%)0.10Secondary or above542(28.5)624(32.9)559(29.5)497(26.2)Primary or below1360(71.5)1271(67.1)1336(70.5)1400(73.8)Current smoking status, n(%)492(25.9)520(27.4)668(35.3)721(38.0)< 0.001Current drinking, n(%)575(30.2)637(33.6)706(37.3)686(36.2)< 0.001Sleep duration, n(%)< 0.001< 7 h793(41.7)759(40.1)766(40.4)837(44.1)> 9 h246(12.9)219(11.6)185(9.8)175(9.2)7–9 h863(45.4)917(48.4)944(49.8)885(46.7)BMI, kg/m225.54(4.18)25.08(2.82)22.36(2.46)20.10(2.13)< 0.001Waist, cm91.36(10.74)91.17(4.58)82.35(3.30)73.74(4.25)< 0.001Dyslipidemia, n(%)260(13.7)91(4.8)65(3.4)41(2.2)< 0.001Data are presented as the mean (SD) or number (%), as appropriateAbbreviations: SD: standard deviation, BMI: body mass index, eGDR: estimated glucose disposal rate, hs-CRP: high-sensitivity C-reactive protein Characteristics of 7,589 participants according to eGDR levels Data are presented as the mean (SD) or number (%), as appropriate Abbreviations: SD: standard deviation, BMI: body mass index, eGDR: estimated glucose disposal rate, hs-CRP: high-sensitivity C-reactive protein Table 2 summarizes baseline characteristics across joint categories of eGDR and hs-CRP. Participants with low eGDR and high hs-CRP had the highest BMI and waist circumference, and approximately 10% reported dyslipidemia; in contrast, those with high eGDR and low hs-CRP had more favorable metabolic profiles. The distribution of eGDR and hs-CRP across joint categories is illustrated in Fig. 2. Table 2Characteristics of 7,589 participants according to joint eGDR and hs-CRP levelsParticipants, No.High GDR & low hs-CRPHigh eGDR & high hs-CRPLow eGDR & low hs-CRPLow eGDR & high hs-CRPP value2252154015622235Age, years, mean(SD)57.76(9.10)60.29(10.00)58.70(9.19)60.21(9.33)< 0.001Sex, n(%)< 0.001Female1169(51.9)674(43.8)877(56.1)1246(55.7)Male1083(48.1)866(56.2)685(43.9)989(44.3)Residence, n(%)< 0.001Rural1986(88.2)1357(88.1)1314(84.1)1777(79.5)Urban266(11.8)183(11.9)248(15.9)458(20.5)Married, n(%)2027(90.0)1339(86.9)1391(89.1)1955(87.5)0.01Educational level, n(%)0.05Secondary or above647(28.7)409(26.6)472(30.2)694(31.1)Primary or below1605(71.3)1131(73.4)1090(69.8)1541(68.9)Current smoking status, n(%)779(34.6)610(39.6)399(25.5)613(27.4)< 0.001Current drinking, n(%)818(36.3)574(37.3)520(33.3)692(31.0)< 0.001Sleep duration, n(%)< 0.001< 7 h924(41.0)679(44.1)656(42.0)896(40.1)> 9 h203(9.0)157(10.2)180(11.5)285(12.8)7–9 h1125(50.0)704(45.7)726(46.5)1054(47.2)BMI, kg/m221.26(2.40)21.19(2.79)24.73(3.43)25.72(3.61)< 0.001Waist, cm77.89(5.73)78.27(5.77)89.65(8.06)92.39(8.22)< 0.001Dyslipidemia, n(%)64(2.8)42(2.7)125(8.0)226(10.1)< 0.001Data are presented as the mean (SD) or number (%), as appropriateSD: standard deviation, BMI: body mass index, eGDR: estimated glucose disposal rate, hs-CRP: high-sensitivity C-reactive protein Characteristics of 7,589 participants according to joint eGDR and hs-CRP levels Data are presented as the mean (SD) or number (%), as appropriate SD: standard deviation, BMI: body mass index, eGDR: estimated glucose disposal rate, hs-CRP: high-sensitivity C-reactive protein Distribution of eGDR and hs-CRP values across the joint groups Separate and joint association of eGDR and hs-CRP During a median follow-up of 9 years, 2,263 participants developed cardiometabolic multimorbidity. Kaplan–Meier curves showed clear separation across quartiles of eGDR and hs-CRP, as well as across their joint categories (Fig. 3). In multivariable-adjusted models, lower eGDR was strongly associated with a higher risk of cardiometabolic multimorbidity (lowest vs. highest quartile: hazard ratio [HR] 2.07, 95% CI 1.80–2.38; Table 3). Higher hs-CRP was also associated with increased risk, with participants in the highest quartile having an 18% higher risk compared with those in the lowest quartile (HR 1.18, 95% CI 1.05–1.34). The dose-response relationships of eGDR and hs-CRP with cardiometabolic multimorbidity was shown in Supplementary eFigure 2, which demonstrates a linear relationship between eGDR and multimorbidity risk. Cumulative incidence of cardiometabolic multimorbidity (CMM) according to separate and joint eGDR and hs-CRP values Table 3Separate associations between eGDR and hs-CRP with risk of cardiometabolic multimorbidityExposuresEvents/Total numberAge and sex adjusted modelMultivariable-adjusted modelHR (95% CI)P valueHR (95% CI)P valueeGDRQuartile 4376/18971 (ref)Quartile 3435/18951.18(1.03–1.35)0.021.14(0.99–1.31)0.06Quartile 2596/18951.67(1.47–1.90)< 0.0011.47(1.27–1.69)< 0.001Quartile 1856/19022.47(2.18–2.79)< 0.0012.07(1.80–2.38)< 0.001Hs-CRPQuartile 1500/19391 (ref)Quartile 2498/18751.01(0.89–1.14)0.870.94(0.83–1.07)0.37Quartile 3617/18821.27(1.13–1.43)< 0.0011.10(0.98–1.24)0.11Quartile 4648/18931.37(1.22–1.54)< 0.0011.18(1.05–1.34)0.01HR: hazard ratio, CI: confidence interval, BMI: body mass index, eGDR: estimated glucose disposal rateAge, sex, residence, household expend, marriage, education level, BMI level, smoking status, alcohol consumption, sleep duration, depressive symptoms, and renal function measured by eGFR were adjusted in the multivariable-adjusted models Separate associations between eGDR and hs-CRP with risk of cardiometabolic multimorbidity HR: hazard ratio, CI: confidence interval, BMI: body mass index, eGDR: estimated glucose disposal rate Age, sex, residence, household expend, marriage, education level, BMI level, smoking status, alcohol consumption, sleep duration, depressive symptoms, and renal function measured by eGFR were adjusted in the multivariable-adjusted models In joint analyses, participants with both low eGDR and high hs-CRP had the highest risk of cardiometabolic multimorbidity (HR 1.80, 95% CI 1.58–2.04) compared with those with high eGDR and low hs-CRP (Table 4). Notably, individuals with low eGDR but low hs-CRP also had a significantly elevated risk (HR 1.52, 95% CI 1.33–1.74), whereas those with high eGDR but high hs-CRP did not show a statistically significant increase in risk (HR 1.06, 95% CI 0.93–1.22). We did not detect significant effects of multiplicative (P for interaction = 0.24) or additive interaction (RERI 0.10 95% CI -0.17, 0.36). Table 4Associations between joint eGDR and hs-CRP levels with risk of cardiometabolic multimorbidityExposuresEvents/Total numberAge and sex adjusted modelMultivariable-adjusted modelHR (95% CI)P valueHR (95% CI)P valueHigh eGDR & low hs-CRP470/22521 (ref)High eGDR & high hs-CRP341/15401.07(0.93–1.23)0.321.06(0.93–1.22)0.38Low eGDR & low hs-CRP528/15621.71(1.51–1.94)< 0.0011.52(1.33–1.74)< 0.001Low eGDR & high hs-CRP924/22352.12(1.90–2.37)< 0.0011.80(1.58–2.04)< 0.001HR: hazard ratio, CI: confidence interval, BMI: body mass index, eGDR: estimated glucose disposal rateAge, sex, residence, household expend, marriage, education level, BMI level, smoking status, alcohol consumption, sleep duration, depressive symptoms, and renal function measured by eGFR were adjusted in the multivariable-adjusted models Associations between joint eGDR and hs-CRP levels with risk of cardiometabolic multimorbidity HR: hazard ratio, CI: confidence interval, BMI: body mass index, eGDR: estimated glucose disposal rate Age, sex, residence, household expend, marriage, education level, BMI level, smoking status, alcohol consumption, sleep duration, depressive symptoms, and renal function measured by eGFR were adjusted in the multivariable-adjusted models Subgroup analyses stratified by age, sex, residence, BMI, and smoking status showed no significant effect modification (Supplementary eTable 2). Additional sensitivity analysis using follow-up time based on survey cycles as the underlying time scale yielded consistent results (Supplementary eTable 3). The findings remained robust after excluding only participants with two or more of these conditions at baseline (Supplementary eTable 4), after excluding 382 participants who developed cardiometabolic multimorbidity within the first 2 years of follow-up (Supplementary eTable 5) and considering death as the competing risk using Fine-Gray sub-distribution regression (Supplementary eTable 6). Discriminative performance Models incorporating eGDR demonstrated improved discrimination compared with the base model including traditional risk factors (C-index 0.676 [95% CI 0.657–0.695] vs. 0.606 [0.587–0.625]). Adding hs-CRP to the base model also improved discrimination (C-index 0.643, 95% CI 0.625–0.662), although to a lesser extent than eGDR alone (Fig. 4). Notably, adding hs-CRP to a model already including eGDR did not further improve model discrimination. The calibration plot showed good agreement between predicted and observed risk across risk deciles (Supplementary eFigure 3). The time-dependent C-index remained stable across follow-up, with eGDR consistently showing higher discriminatory performance than hs-CRP over time (Supplementary eFigure 4). Moreover, we found that eGDR achieved a higher C-index than the model combining its individual components (Fig. 5). The eGDR also demonstrated higher reclassification performance (NRI 0.187, 95% CI 0.098–0.201; IDI 0.019, 95% CI 0.013–0.039) (Supplementary eTable 7). Comparison of the predictive value of separate and joint eGDR and hs-CRP for cardiometabolic multimorbidity (CMM). Receiver operating characteristic curves were used to evaluate the CMM prediction. The basic model was adjusted for age, sex, residence, marital status, educational level, BMI, smoking status, current drinking, sleep duration, depressive symptoms, and renal function Comparison of the predictive value of eGDR and its components (waist circumference, hypertension, and HbA1c) for cardiometabolic multimorbidity (CMM). Receiver operating characteristic curves were used to evaluate the CMM prediction. The basic model was adjusted for age, sex, residence, marital status, educational level, BMI, smoking status, current drinking, sleep duration, depressive symptoms, and renal function Discussion In this nationally representative cohort of middle-aged and older Chinese adults, we evaluated the separate and joint associations of eGDR and hs-CRP with the risk of cardiometabolic multimorbidity. We found that lower eGDR was strongly and consistently associated with higher risk of cardiometabolic multimorbidity. In joint analyses, low eGDR conferred elevated risk regardless of inflammation status, whereas elevated hs-CRP in the presence of preserved eGDR was not associated with increased risk. In addition, eGDR demonstrated superior discriminative capacity compared with hs-CRP, and adding hs-CRP to eGDR did not materially improve prediction. Together, these findings highlight the central role of metabolic dysfunction, as captured by eGDR, in the development of cardiometabolic multimorbidity. eGDR is a well-validated surrogate of insulin resistance derived from readily available clinical parameters [28]. Lower eGDR reflects greater insulin resistance and impaired metabolic health. A growing body of epidemiologic evidence has linked lower eGDR to adverse cardiometabolic outcomes, including type 2 diabetes, cardiovascular disease, heart failure, stroke, and mortality across diverse populations [6–8. 16, 29–32] Prior studies using population-based cohorts have consistently shown that lower eGDR, measured at baseline or the trajectory, is associated with cardiometabolic multimorbidity [33, 34]. In the CHARLS cohort specifically, eGDR has been shown to outperform other insulin resistance indices in predicting cardiovascular disease,15 and longitudinal changes in eGDR have been associated with cardiometabolic risk [35]. Consistent with and extending these findings, our study demonstrates that eGDR is a strong predictor of multimorbidity, capturing the clustering and progression of cardiometabolic conditions rather than single disease endpoints. Compared with prior studies that focused on individual cardiometabolic outcomes, our study addresses an important gap by examining cardiometabolic multimorbidity, which better reflects the real-world burden of coexisting chronic diseases, especially among the older population [36]. While previous research has established insulin resistance as a shared upstream risk factor of multiple cardiometabolic conditions, few studies have quantified its role in the accumulation of multiple diseases accounting for biological inflammation levels. Our findings suggest that eGDR, as an integrative marker of insulin resistance and metabolic dysfunction, may be particularly useful for identifying individuals at high risk of transitioning from single conditions to multimorbidity among the older population, where both metabolic dysfunction and low-grade inflammation are highly prevalent. Inflammation and insulin resistance are closely intertwined biological processes [37]. Chronic low-grade inflammation, as reflected by hs-CRP, correlates with endothelial dysfunction, oxidative stress, and atherosclerosis, while insulin resistance promotes dyslipidemia, hyperglycemia, and vascular injury [38]. Prior studies have examined inflammatory markers and insulin resistance separately, and some have explored their joint effects; however, evidence remains limited and inconsistent for cardiometabolic multimorbidity. For instance, Zhang et al., explored the association between eGDR and incident cardiovascular diseases among individuals without diabetes [39]. Chen et al., reported that eGDR and systemic inflammation response index have joint effect on mortality in Cardiovascular-Kidney-Metabolic (CKM) syndrome [40]. In our study, although hs-CRP was independently associated with cardiometabolic multimorbidity, its effect was modest and did not persist when eGDR was preserved. In contrast, the association of low eGDR with cardiometabolic multimorbidity remained robust regardless of hs-CRP levels. These findings suggest that metabolic dysfunction may play a more dominant role than inflammation alone in driving cardiometabolic disease clustering. It is also possible that eGDR captures downstream consequences of both insulin resistance and inflammation, thereby serving as a more comprehensive integrative marker. Similarly, a prior study found that eGDR was more strongly associated with all-cause and cardiovascular mortality regardless of inflammation status using NHANES data [41]. Importantly, we found that eGDR had the higher discriminative capacity for predicting cardiometabolic multimorbidity compared with hs-CRP and traditional risk factors. While hs-CRP modestly improved prediction beyond traditional covariates, its incremental value was limited when eGDR was already included. This finding underscores the clinical utility of eGDR as a simple, cost-effective, and readily available metric for risk stratification in population settings. Consistently, a prior study found that eGDR index may outperform other insulin resistance indices in predicting cardiovascular events in individuals with CKM syndrome [42]. However, it is noteworthy that the overall C-index values remained moderate, suggesting that eGDR should complement, rather than replace, existing risk prediction tools. Although the individual components of eGDR are already routinely assessed in clinical practice, eGDR may still have practical value as an integrated marker of insulin resistance–related metabolic dysfunction. By summarizing these related cardiometabolic domains into a single score, eGDR may facilitate risk stratification and help identify individuals at elevated risk for CMM who may benefit from closer monitoring or more intensive preventive strategies. In addition, our prediction analyses showed that eGDR yielded a higher C-index than a model including its three individual components separately, suggesting that this composite measure may capture risk information more efficiently than the components considered individually. Although eGDR incorporates HbA1c, which is closely related to diabetes risk, CMM in this study was defined as the co-occurrence of diabetes, heart disease, and stroke. Therefore, the predictive value of eGDR should be interpreted as reflecting an integrated measure of insulin resistance–related metabolic dysfunction rather than direct overlap with all outcome components. Notably, eGDR yielded a higher C-index than a model including waist circumference, hypertension, and HbA1c as separate components, suggesting that the composite score may capture clinically relevant risk information more efficiently than its individual components considered separately. Several strengths of this study should be noted. These include the use of a large, nationally representative cohort with long-term follow-up, standardized data collection, and comprehensive assessment of cardiometabolic conditions. The joint-exposure analysis allowed us to disentangle the relative contributions of metabolic dysfunction and inflammation, and the focus on multimorbidity provides clinically relevant insights beyond single-disease outcomes. Nevertheless, several limitations warrant consideration. First, because blood pressure and biomarker measurements were not consistently available across all follow-up visits, cardiometabolic conditions ascertainment was based on self-reported physician diagnoses, which may introduce misclassification, although such misclassification is likely non-differential and has been widely accepted in large epidemiologic studies [43]. Second, exposures were assessed at baseline only, and we were unable to account for temporal changes in eGDR or hs-CRP, which may influence risk estimates. Third, eGDR is an indirect measure of insulin resistance and may not fully capture its complexity compared with gold-standard methods such as the hyperinsulinemic-euglycemic clamp. Particularly, the performance of eGDR against insulin resistance warrants validation in Chinese populations. Fourth, residual confounding cannot be entirely excluded despite adjustment for a wide range of covariates. Finally, although CHARLS is nationally representative of the Chinese population, generalizability to other populations with different genetic, lifestyle, and healthcare contexts should be interpreted with caution. In addition, the primary analytic sample may represent a relatively healthier subset of the CHARLS population because participants with baseline cardiometabolic conditions were excluded, which should be considered when interpreting the generalizability of our findings. Conclusions In conclusion, our findings highlight the central role of eGDR in the development of cardiometabolic multimorbidity and eGDR appears to outperform inflammation markers in risk prediction. The findings suggest that eGDR may serve as a practical tool for identifying high-risk individuals in the general population. Future studies incorporating longitudinal assessments and intervention designs are needed to further elucidate causal pathways and evaluate whether improving metabolic health can reduce the burden of cardiometabolic multimorbidity. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Abbreviations body mass index China Health and Retirement Longitudinal Study concordance index estimated glucose disposal rate high-sensitivity C-reactive protein relative excess risk due to interaction Author contributions XW, LW, and CC conceived and designed the study and wrote the main manuscript text. CC, KX, JC, and LZ analyzed the data. SY, SC, and LL conducted the literature search and prepared figures. YL, and JD performed the manuscript review and supervision. All authors reviewed and approved the final manuscript. Funding National Natural Science Foundation of China (32471535); The “Medicine + X” cross-innovation construction project of China (2025JBGS05); The Science and Technology Capacity Enhancement Program of Jilin Provincial Health Commission (2025WS-ZA002); Beijing Medical Reward Foundation (YXJL-2025-0483-0322); Special Project for the Application of Specialized Disease Database and Clinical Research Platform (ZBK2025019). Data availability Data from the CHARLS project is publicly available to researchers on application (http://charls.pku.edu.cn/). The final datasets and codes generated for this study are available on reasonable request to corresponding author. Declarations Ethics approval and consent to participate This study protocol was approved by the Ethics Review Committee of Peking University (IRB00001052-11015). All participants provided written informed consent at baseline. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cancan Cui PhD, Xu Wang PhD and Liqiang Wang PhD contributed equally to this work. Contributor Information Yanxi Liu, Email: liuyx@jlu.edu.cn. Jianshi Du, Email: dujs@jlu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data from the CHARLS project is publicly available to researchers on application (http://charls.pku.edu.cn/). The final datasets and codes generated for this study are available on reasonable request to corresponding author.