You are viewing the site in preview mode

Skip to main content
Download PDF
Download ePub

Comparing the impact of cumulative insulin resistance surrogates exposure on stroke: optimizing prevention strategies

Lipids in Health and Disease volume 24, Article number: 158 (2025) Cite this article

Abstract

Background

Insulin resistance (IR) plays a major role in increasing the risk of stroke. The objective of this research is to systematically evaluate and compare the impact of cumulative exposure over time to four commonly used IR surrogates—triglyceride-glucose (CumTyG) index, metabolic score for IR (CumMetS-IR), estimated glucose disposal rate (CumeGDR) and triglyceride to high-density lipoprotein cholesterol (CumTG/HDL-C) ratio—on stroke risk, providing insights for optimizing monitoring strategies for primary stroke prevention.

Methods

The study population was sourced from the China Health and Retirement Longitudinal Study (CHARLS2011-2018). Cumulative exposure to IR (CumIR) surrogates was calculated as the mean value of IR surrogates measured in the first and third waves of CHARLS, multiplied by the total exposure duration. The primary endpoint was incident stroke, determined through questionnaires in the third and fourth waves of CHARLS. Multivariable Cox regression models were applied to estimate and compare HRs and 95% CIs for stroke across quartiles of CumIR surrogates.

Results

A total of 4,669 participants with no history of stroke at baseline were included. During a median follow-up of 6 years, 347 new stroke events (7.43%) were recorded. The incidence rates of stroke in the highest quartiles of CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR, as well as the lowest quartile of CumeGDR, were 9.67%, 9.93%, 10.45%, and 13.02%, respectively. In terms of risk assessment, the multivariable Cox regression analysis showed that the highest quartiles of CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR and the lowest quartile of CumeGDR were associated with stroke risk, with corresponding HR (95% CI) of 1.48(1.05–2.10), 1.61(1.15–2.24), 1.72(1.21–2.43), and 3.57(2.25–5.68), respectively. In terms of event prediction, receiver operating characteristic analysis revealed that CumeGDR had the highest predictive accuracy for incident stroke compared with other common IR surrogates.

Conclusions

In assessing stroke risk and predicting events in middle-aged and elderly populations, cumulative exposure to eGDR demonstrates significant advantages over other common IR surrogates. Incorporating eGDR as an IR monitoring marker is recommended for primary stroke prevention.

Graphic abstract

Introduction

The disease burden of stroke is one of the greatest global health challenges of the 21st century [1]. Over the past 30 years, the absolute number of new stroke cases has increased by 70%, and the number of people living with stroke has increased by 85% [2]. In 2019, stroke caused 6.55 million deaths and 143 million disability-adjusted life years worldwide, making it the second leading cause of death and disability globally [2,3,4]. Furthermore, it is important to note that the incidence of stroke increases with age [1]; given the rapid global aging trend, the number of deaths and disabilities worldwide is expected to rise significantly in the future [1]. Therefore, primary prevention should be a priority to mitigate the potential future burden of stroke-related diseases [5].

Insulin resistance (IR) refers to a condition in which insulin-targeted tissues exhibit reduced responsiveness to physiological levels of insulin [6], and it is a key characteristic in stroke patients [7]. Studies have shown that IR can lead to stroke by inducing hemodynamic disturbances, enhancing platelet adhesion, activation, and aggregation, as well as promoting atherosclerosis [7, 8]. There are currently various methods for assessing IR [9,10,11], including the hyperinsulinemic-euglycemic clamp [9], the quantitative insulin sensitivity check index [10], the homeostasis model assessment of IR [11], and IR surrogates such as triglyceride-glucose (TyG) index, triglyceride to high-density lipoprotein cholesterol (TG/HDL-C) ratio, metabolic score for IR (MetS-IR), and estimated glucose disposal rate (eGDR) [12,13,14,15,16]. Among these methods, IR surrogates are the most frequently used worldwide due to their simplicity, convenience, reproducibility, and better potential for widespread application. From a primary prevention perspective, identifying the optimal IR surrogates for stroke risk assessment and event prediction is crucial for reducing the burden of stroke. However, considering the dynamic nature of the metabolic factors involved in IR surrogates, such as blood glucose, lipids, blood pressure, and obesity [17], it is important to further explore the longitudinal changes in IR surrogates and their effects on stroke.

Cumulative exposure has become a widely used approach in recent years for handling repeated measurements in longitudinal studies, reflecting the intensity and duration of exposure to a given parameter over a period of time [18,19,20,21,22]. For example, prolonged high exposure to low-density lipoprotein cholesterol significantly increases the risk of atherosclerosis [23]. This differs from baseline single measurements, as cumulative exposure takes into account both the amount and duration of exposure during follow-up. Given the current severity of stroke and the importance of primary prevention, and the lack of systematic studies evaluating the relationship between cumulative exposure to IR surrogates and stroke, the present study aims to analyze and compare the impact of cumulative exposure to four common IR surrogates (TyG index, TG/HDL-C ratio, MetS-IR, and eGDR) over time on stroke risk using national data from the China Health and Retirement Longitudinal Study (CHARLS). The goal is to provide valuable research data for optimizing monitoring strategies in primary stroke prevention.

Methods

Study population

We utilized data from the CHARLS project, a detailed description of which has been provided in previous publications [24]. The overall design and implementation framework of CHARLS are summarized in supplementary methods. Briefly, the baseline survey of the CHARLS study began in 2011–2012 and recruited 17,008 participants from 28 provinces across China, covering 150 counties, 450 villages, and approximately 10,000 households. The study population was selected through a multi-stage random sampling process, followed by random selection from a sampling frame reflecting all residential units within each village/community. Regarding follow-up, the CHARLS team conducts national tracking surveys every 2–3 years following the baseline survey. As of now, four waves of national tracking surveys have been completed, in 2013, 2015, 2018, and 2020, with blood sample measurements available from the baseline survey and the 2015 survey.

For the present study, we included baseline data from the first wave of CHARLS and follow-up data from the third and fourth waves of the national surveys. Cumulative IR (CumIR) surrogates were developed based on data from the first and third waves. The inclusion process for the study population is as follows: Step 1: We included participants aged ≥ 45 years from the baseline data of the first CHARLS wave who were followed up in the third wave. Step 2: We excluded participants who were diagnosed with stroke in the first wave or had an uncertain stroke status in either the first or third wave. Step 3: To calculate IR surrogates, participants were required to have fasting blood samples and obesity measurements, so we excluded those who did not undergo measurements of obesity parameters or blood tests, or who only completed non-fasting blood tests (i.e., did not have Cum-IR measured). A detailed flowchart of the study process was shown in Fig. 1.

Fig. 1
figure 1

Flow chart of study participants

Full size image

Ethics statement

All participants in the CHARLS study provided written informed consent before participation, and subsequent surveys, physical measurements, blood collection, and follow-up were carried out after approval. The implementation of CHARLS was authorized by the Ethics Committee of Peking University (No. 00001052–11015), and all research data were de-identified.

Assessment of covariates and independent variables

Various covariates were assessed from the CHARLS questionnaires and datasets, including basic demographic characteristics (gender, age, education, marital status, and residence), lifestyle factors (smoking and drinking status), comorbidities (hypertension, diabetes, and heart disease), simple body measurements [systolic blood pressure, diastolic blood pressure, height, weight, body mass index (BMI), waist circumference (WC)], and blood biomarkers. Detailed assessments of lifestyle factors, comorbidities, and simple body measurements were summarized in supplementary methods.

Blood samples were collected after overnight fasting. Venous blood samples were collected by trained staff. To maintain sample stability, all samples were stored at -80 °C. Blood sample analysis was performed in the laboratory of Peking University, where biomarkers including glycated hemoglobin, fasting plasma glucose (FPG), HDL-C, low-density lipoprotein cholesterol, total cholesterol, TG, uric acid, and creatinine were measured. The CHARLS team performs weekly quality control checks, ensuring the stability of sample analysis during the study period. Detailed analytical methods, laboratory coefficients of variation, and upper detection limits have been reported in previous studies [25].

The calculation methods for IR surrogates and CumIR surrogates were shown in Table 1 [12,13,14,15, 20,21,22].

Table 1 Calculation method of IR surrogates and CumIR surrogates
Full size table

Outcome determination

The primary outcome in this study was incident stroke, which was assessed based on questionnaires from the third and fourth waves. Participants were asked, “Have you been diagnosed with stroke by a doctor?” and responded with “Yes” or “No.” A response of “Yes” indicated a stroke diagnosis.

Missing data assessment

In this study, the total missing data accounted for 1.91% of the study population, with the highest missing value being 34 (supplementary Table 1). Given the relatively low proportion of missing data, all analyses were performed on the original dataset. Data analysis was conducted between October and November 2024.

Statistical analysis

A two-tailed significance level of 5% was used for all statistical tests. Data analysis was conducted using R language (version 4.2.1) and Empower(R) (version 4.2). Baseline characteristics of the study population are presented as frequencies (percentages), mean ± standard deviation, or median and interquartile range, with appropriate statistical tests selected based on the nature of the data.

Cox regression models were used to estimate the multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals for stroke associated with IR surrogates and CumIR surrogates. Given the large differences in the units of various IR surrogates and CumIR surrogates, we chose to assess the association between the quartiles of these surrogates and stroke risk for easier comparison of results. Following the STROBE guidelines, a stepwise adjustment approach was applied for model building [26]. To minimize the impact of collinearity, the components of the IR surrogates and potential collinear covariates (supplementary Tables 2–9) were excluded from the models [27].

Three adjusted models were constructed for data analysis: Model I adjusted for basic demographic characteristics; Model II further adjusted for potential confounders including lifestyle factors and comorbidities; Model III, the final model, adjusted for all covariates while excluding those covariates with high collinearity.

Receiver operating characteristic (ROC) analysis was applied to assess the predictive value of IR surrogates and their cumulative exposure for stroke events, calculating the corresponding area under the curve, sensitivity, specificity, and optimal threshold. Differences in area under the curve values were compared using the DeLong test [28].

Sensitivity analyses

  1. (1)

    Stratified analyses by age and gender were performed to assess the associations mentioned above, with likelihood ratio tests to detect potential differences across populations.

  2. (2)

    To minimize the impact of medication on the results [29], we excluded participants using antihypertensive, antidiabetic, or lipid-lowering medications at baseline.

  3. (3)

    Multiple imputation was applied to handle missing data [30], and multivariable Cox regression models were run on the imputed datasets.

  4. (4)

    To reduce the influence of cardiovascular disease on stroke incidence [31], the analysis was repeated excluding individuals with cardiovascular disease.

  5. (5)

    To minimize the potential impact of altered insulin secretion patterns in diabetic patients on the measurement of IR [32], the analysis was repeated for non-diabetic participants at baseline.

  6. (6)

    Considering potential reverse causality, the analysis was repeated excluding participants diagnosed with stroke in the third wave survey.

Results

Study participants

A total of 4,669 participants with complete follow-up data were included in this study, with a mean age of 59 years and a male-to-female ratio of 0.82:1. During a median follow-up of 6 years, 347 new stroke events (7.43%) were recorded. Grouping participants by the quartiles of CumIR surrogates revealed a progressive increase in the stroke incidence for CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR, while the stroke incidence for CumeGDR decreased with increasing quartiles (Fig. 2). Notably, the stroke incidence in the highest quartiles of CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR was 9.67%, 9.93%, 10.45%, respectively, while the lowest quartile of CumeGDR had a stroke incidence of 13.02%, suggesting that lower CumeGDR was associated with a higher stroke risk.

Fig. 2
figure 2

Stroke incidence in the study population according to CumIR quartile groupings. CumIR: Cumulative insulin resistance; CumTyG: cumulative triglyceride-glucose; CumMetS-IR: cumulative metabolic score for insulin resistance; CumTG/HDL-C: cumulative triglyceride to high-density lipoprotein cholesterol; CumeGDR: cumulative estimated glucose disposal rate; CumIR: cumulative insulin resistance

Full size image

Table 2 shows the baseline characteristics of the study population, comparing those who had a stroke during follow-up with those who did not. Typically, patients who experienced a future stroke were older at the time of the baseline survey. They were also more likely to be overweight or obese, have poorer insulin resistance status, have more underlying diseases, and exhibit higher levels of blood pressure, blood uric acid, blood glucose, and lipids other than HDL-C.

Table 2 Summary of baseline characteristics of the study population according to occurrence of stroke
Full size table

Association between baseline IR surrogates quartiles and stroke

In the three adjusted multivariable models, higher levels of TyG index, TG/HDL-C ratio, MetS-IR, and lower levels of eGDR were significantly associated with a higher risk of stroke (supplementary Table 10). Among these, the lowest quartile of eGDR had the highest HR for stroke, followed by the highest quartile of MetS-IR, TG/HDL-C ratio, and TyG index (Model III: HR2.42 vs. 1.86 vs. 1.59 vs. 1.55).

Association between CumIR surrogates quartiles and stroke

Supplementary Fig. 1 shows the histogram distribution of CumIR surrogates and the pairwise correlation coefficients between CumIR surrogates. The correlation matrix revealed that CumeGDR was negatively correlated with the other three CumIR surrogates.

As with baseline IR surrogates, the association between CumIR surrogates’ quartiles and stroke showed similar results across all models (Table 3). Higher levels of CumTyG index, CumTG/HDL-C ratio, CumMetS-IR, and lower levels of CumeGDR were significantly associated with increased stroke risk. Specifically, the lowest quartile of CumeGDR had the highest HR for stroke, followed by the highest quartiles of CumMetS-IR, CumTG/HDL-C ratio, and CumTyG index (Model III: HR3.57 vs. 1.72 vs. 1.61 vs. 1.48). Notably, compared to baseline eGDR, CumeGDR showed a higher HR for stroke.

Table 3 Cox regression model analyzes the association between CumIR surrogates and stroke
Full size table

ROC analysis for IR surrogates and CumIR surrogates in predicting stroke

Figure 3 shows the ROC analysis for baseline IR surrogates and CumIR surrogates in predicting stroke events. Among baseline IR surrogates, eGDR showed the highest accuracy for predicting stroke, with an optimal threshold of 9.5555. For CumIR surrogates, CumeGDR demonstrated the highest accuracy for predicting stroke events, with an optimal threshold of 23.7991, outperforming CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR (Table 4).

Fig. 3
figure 3

ROC analysis curves of baseline IR surrogates and CumIR surrogates in predicting stroke. ROC: Receiver operating characteristic; IR: insulin resistance; CumIR: Cumulative insulin resistance; CumTyG: cumulative triglyceride-glucose; CumMetS-IR: cumulative metabolic score for insulin resistance; CumTG/HDL-C: cumulative triglyceride to high-density lipoprotein cholesterol; CumeGDR: cumulative estimated glucose disposal rate; CumIR: cumulative insulin resistance

Full size image
Table 4 ROC analysis for IR surrogates and CumIR surrogates in predicting stroke
Full size table

Sensitivity analysis

In the first sensitivity analysis, we found that the association between CumIR surrogates and stroke did not differ by age or gender (supplementary Table 11, all P-interaction > 0.05). In the second sensitivity analysis, we excluded participants using antihypertensive, antidiabetic, or lipid-lowering medications, and the results remained consistent with the main analysis (sensitivity analysis 2: supplementary Tables 12 and 13). After conducting multiple imputation (sensitivity analysis 3), the results were similar to the main analysis (supplementary Table 14). Repeating the analysis after excluding participants with cardiovascular disease (sensitivity analysis 4: supplementary Tables 15 and 16) and diabetes (sensitivity analysis 5: supplementary Tables 17 and 18) did not lead to any significant changes in the results. Finally, after excluding participants diagnosed with stroke in the third wave survey (sensitivity analysis 6: supplementary Tables 19 and 20), the findings remained highly consistent.

Discussion

Based on the CHARLS national prospective cohort study, we found that the cumulative exposure of IR surrogates—TyG index, TG/HDL-C ratio, MetS-IR, and eGDR—over time is associated with the incidence of stroke. Among these, CumeGDR appears to provide additional risk stratification and is more accurate for predicting stroke events. From a primary prevention perspective, we recommend including the IR surrogate eGDR as a key factor in stroke monitoring.

As global populations age, stroke is becoming an increasingly common condition. By 2050, the absolute number of stroke-related deaths worldwide is expected to reach 9.7 million, placing a heavy burden on families and public health systems [1, 33]. Previous studies have shown that insulin resistance (IR) is an important risk factor for stroke and that there is a causal relationship between the two [7, 34,35,36]. Numerous studies have evaluated various IR surrogates for stroke risk assessment, demonstrating the utility of these surrogates in real-world settings [37,38,39,40,41]. However, a systematic comparative analysis of IR surrogates for stroke risk assessment remains lacking. It is important to note that in a recent longitudinal study based in rural China, Zhao et al. used the TyG index, visceral adiposity index (VAI), lipid accumulation product (LAP), and Chinese VAI (CVAI) as surrogate IR indices to compare their effectiveness in stroke risk assessment [41]. Their results showed that all these surrogates were independently associated with stroke risk, with TyG index providing the most informative risk assessment (each increase in standard deviation). It should be noted that, CVAI, VAI, and LAP are rarely classified as IR surrogate indices, and they are more appropriately classified as obesity indices. Overall, the TyG index has good application potential as an IR alternative index in stroke risk assessment. In the current study, we systematically evaluated and compared the baseline IR surrogates—TyG index, TG/HDL-C ratio, MetS-IR, and eGDR—and concluded that baseline eGDR had the strongest association with stroke, making it the most suitable for early stroke risk assessment.

In addition to comparing baseline IR surrogates for stroke risk, our study also investigated and compared the relationship between the cumulative exposure of IR surrogates over time and incident stroke—a relationship that has not been systematically examined before. Cumulative exposure is a new concept used in the analysis of repeated measures data, reflecting the sustained intensity of exposure to a given parameter over a specified period [18,19,20]. Several recent studies have explored the association between individual cumulative IR surrogates such as CumTyG index, CumMetS-IR, and CumeGDR with stroke risk. Specifically, studies from the Kailuan cohort reported HRs of 1.30–1.35 for high CumTyG index and 1.70 for high CumMetS-IR in relation to stroke risk [42, 43]. Our findings are consistent with these results, with CumMetS-IR showing a slightly stronger association with stroke risk compared to CumTyG index. Regarding the association between CumeGDR and stroke, a recent study by Yao et al. reported findings where the HRs associated with stroke across CumeGDR quintiles were Q1:1, Q2:0.78, Q3:0.78, Q4:0.43, and Q5:0.37, respectively [22]; these recalculated results align with our findings. Compared to previous studies that individually analyzed these IR surrogates [21, 22, 42, 43], our study has the advantage of systematically evaluating the cumulative exposure of multiple IR surrogates in a homogeneous study population, including CumTG/HDL-C ratio. We used quartiles for group categorization and calculated the HRs, which allowed us to identify CumeGDR as the most optimal surrogate for stroke risk assessment.

Regarding eGDR and its cumulative exposure being significantly superior to other IR surrogates in stroke risk assessment and event prediction, we attribute this advantage to eGDR’s incorporation of metabolic components more closely linked to stroke. Compared to indices like the TyG index, TG/HDL-C ratio, and MetS-IR, the eGDR calculation incorporates more stable glycemic parameters such as HbA1c [44] and central obesity indicator WC, while additionally integrating hypertension– a critical risk factor for cerebrovascular and cardiovascular diseases [45]. From the perspective of stroke risk factors, blood pressure, blood glucose, and obesity all contribute significantly to stroke, whereas elevated cholesterol levels typically play a lesser role [46, 47]. Notably, in the context of blood glucose measurement, a single FPG measurement often exhibits high within-individual variability, making it a less-than-ideal indicator for reflecting chronic hyperglycemia; by contrast, HbA1c, as an integrated measure of blood glucose concentration, demonstrates superior temporal stability at the individual level [44, 48]. According to the study by Selvin et al., a multivariable model incorporating only FPG revealed associations similar to those observed for HbA1c-Stroke, yet the effect size of FPG was significantly smaller [46]. Regarding obesity assessment, prior research evidence confirms that central obesity contributes more substantially to stroke risk than general obesity [49, 50]. In the Atherosclerosis Risk in Communities study, Mozafar et al. employed targeted maximum likelihood estimation to compare the impacts of BMI and central obesity on stroke risk in individuals with and without diabetes. After adjusting for confounders, their findings revealed that the relative risk for stroke associated with BMI ranged from 1.04 to 1.11, whereas the relative risk associated with central obesity ranged from 1.10 to 1.15 [49]. Another study from Northeast China focusing on a middle-aged and elderly population reported comparable findings: they demonstrated that obesity assessed via BMI was associated with an odds ratio of 1.79 for stroke, while central obesity evaluated using WC showed an odds ratio of 1.94 [50]. Finally, compared to the TyG index, TG/HDL-C ratio, and MetS-IR, eGDR additionally incorporates hypertension as an evaluation factor, which may represent the most critical reason why eGDR significantly outperforms other IR surrogates in stroke risk assessment. As is well-established, hypertension represents the most important risk factor for stroke (far exceeding elevated LDL-C, hyperglycemia, and obesity in impact), and serves as the primary driver of the current and future increase in stroke-related disease burden [47, 51]. According to global stroke risk factor analysis data, approximately 39% of stroke events are attributable to hypertension [52]. In conclusion, we propose that the significant advantage of eGDR and its cumulative exposure in stroke risk assessment over other IR surrogates may primarily stem from the fact that the components of eGDR play a more critical role in promoting stroke risk. Furthermore, when multiple metabolic factors co-accumulate, the risk of stroke incidence is further elevated [47, 51, 53].

Our study has important implications for public health and clinical practice. First, given the significant pathogenic role of IR in stroke, we believe it is essential to incorporate IR assessments into stroke clinical practice, particularly for middle-aged and older populations [1, 7, 8]. Despite the availability of multiple methods for assessing IR [9,10,11,12,13,14,15], for primary prevention in the general population, IR surrogates remain the most suitable due to their simplicity, reproducibility, and wide applicability [16]. Based on our findings, we recommend eGDR as the optimal IR surrogate for stroke monitoring. From a data-driven perspective, CumeGDR incorporates both baseline eGDR values and longitudinal changes during follow-up. It not only preserves the predictive value of eGDR for stroke risk assessment but also provides more risk warnings and predictive information compared to baseline eGDR alone. Therefore, theoretically, maintaining lower CumeGDR levels during follow-up may help control stroke risk. From another perspective, for the middle-aged and elderly populations in China, whether undergoing single or multiple health assessments, focusing on eGDR and CumeGDR can provide effective prediction and risk evaluation for future stroke events. Considering the significant advantages of eGDR and CumeGDR in stroke assessment—along with their simplicity, convenience, and cost-effectiveness—they are highly suitable for widespread use. Based on our findings, we recommend a baseline eGDR threshold of 9.5555 for individuals undergoing single-time health checks. For those who undergo annual checks, we recommend maintaining eGDR below 9.5555 and limiting the cumulative exposure to eGDR over three years to 23.7991. For the control of eGDR and its cumulative exposure levels, we believe that appropriate lifestyle interventions and the application of medications may be significantly helpful [2, 54,55,56]. This is because lifestyle interventions and drug treatments significantly improve metabolism, which may lead to a reduction in eGDR and its cumulative exposure levels and thus lower the risk of stroke. When an individual exceeds the threshold for eGDR and its cumulative exposure, implementing the following clinical actions may be beneficial: (1) Healthy Diet: Adopt a balanced diet emphasizing vegetables, fruits, whole grains, and low-fat dairy products while reducing intake of saturated fats, trans fats, cholesterol, and salt. Avoid sugary beverages and processed foods, and stay adequately hydrated by drinking moderate amounts of water [57, 58]. (2) Moderate Exercise: Engage in activities such as brisk walking, jogging, swimming, or cycling, combined with strength training to enhance muscle strength [2, 59]. (3) Smoking Cessation and Alcohol Moderation: Quitting smoking and limiting alcohol consumption significantly improve metabolic parameters and are among the critical measures for stroke prevention [2, 60]. (4) Management of blood pressure, blood glucose, lipids, and weight: these indicators can be controlled through a combination of lifestyle modifications and pharmacotherapy interventions [2, 54,55,56]. However, whether lifestyle interventions or drug therapies (e.g., hypoglycemic, antihypertensive, or statin medications) are suitable for specific CumeGDR thresholds cannot be determined in the current study. This issue still requires further interventional research for validation. In addition to its direct application in stroke prevention, the comparative findings of the current study can also serve as a valuable reference and guide for future related research. Furthermore, these findings offer valuable resources for future model development and outcome validation across other ethnic populations.

Study strengths and limitations

The strengths of this study include the following: (1) This study evaluated for the first time the association between CumTG/HDL-C ratio and stroke. (2) In terms of innovation, this study systematically analyze the effects of various CumIR surrogates on stroke, providing crucial research data for primary stroke prevention. (3) From a clinical perspective, this study provides both baseline and cumulative exposure data, which intuitively illustrates the importance of controlling cumulative IR exposure and offers significant threshold recommendations. This provides a direct and easily understandable approach for primary stroke prevention in middle-aged and elderly populations. (4) In terms of result robustness, the study performed six different sensitivity analyses, with highly consistent findings.

Similarly, there are certain limitations of the current study that need to be reported: (1) As the data in this study were based on the CHARLS registry questionnaire, where stroke diagnoses relied on self-reported information from participants and medical diagnoses, there may be recall bias, potentially leading to misclassification of some participants [61]. However, according to the latest evidence from repeated checks in the Health and Retirement Study, Glymour et al. suggested that misreporting of stroke is typically non-systematic, and the potential for misclassification bias is minimal [62]. (2) While the incidence of stroke increases with age, the increasing trend of stroke in younger populations also requires attention [1, 63]. The current study’s findings are applicable only to middle-aged and elderly populations, and future research should further explore the role of CumIR surrogates in younger populations. (3) Although the CHARLS study is prospective, observational studies do not involve active intervention protocols for the study population [64]; therefore, this study cannot assess the impact of interventions on improving stroke risk associated with CumIR surrogates. (4) It is unavoidable that a large proportion of participants with missing blood parameters were excluded from the analysis, which led to a relative reduction in sample size. (5) Stroke is a multifactorial, complex disease, and although the current analysis accounted for many confounding factors, there may still be unmeasured or unconsidered confounders [65].

Conclusions

In this national cohort study, we found that, compared to CumTyG index, CumTG/HDL-C ratio, and CumMetS-IR, cumulative exposure to eGDR offers significant advantages in stroke risk assessment and prediction in middle-aged and elderly populations. From a primary prevention perspective, we recommend incorporating eGDR as an IR surrogate for stroke monitoring.

Data availability

The data used in this study were provided by the CHARLS research team; all research data can be downloaded from the official CHARLS website (http://charls.pku.edu.cn/en).

Abbreviations

CHARLS:

China Health and Retirement Longitudinal Study

CI:

Confidence intervals

CumIR:

Cumulative exposure to IR

eGDR:

Estimated glucose disposal rate

HRs:

Hazard ratios

IR:

Insulin resistance

MetS-IR:

Metabolic score for insulin resistance

ROC:

Receiver operating characteristic

TG/HDL-C:

Triglyceride to high-density lipoprotein cholesterol

TyG:

Triglyceride-glucose

References

  1. Feigin VL, Owolabi MO, World Stroke Organization–Lancet Neurology Commission Stroke Collaboration Group. Pragmatic solutions to reduce the global burden of stroke: a World Stroke Organization-Lancet Neurology Commission. Lancet Neurol. 2023;22:1160–206.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Owolabi MO, Thrift AG, Mahal A, Ishida M, Martins S, Johnson WD, et al. Primary stroke prevention worldwide: translating evidence into action. Lancet Public Health. 2022;7:e74–85.

    Article  PubMed  Google Scholar 

  3. GBD 2019 Stroke Collaborators. Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the global burden of disease study 2019. Lancet Neurol. 2021;20:795–820.

    Article  Google Scholar 

  4. GBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the global burden of disease study 2019. Lancet. 2020;396:1204–22.

    Article  Google Scholar 

  5. Bushnell C, Kernan WN, Sharrief AZ, Chaturvedi S, Cole JW, Cornwell WK 3rd, et al. 2024 guideline for the primary prevention of stroke: a guideline from the American heart association/American stroke association. Stroke. 2024;55:e344–424.

    Article  PubMed  Google Scholar 

  6. Lee SH, Park SY, Choi CS. Insulin resistance: from mechanisms to therapeutic strategies. Diabetes Metab J. 2022;46:15–37.

    Article  PubMed  Google Scholar 

  7. Ding PF, Zhang HS, Wang J, Gao YY, Mao JN, Hang CH, et al. Insulin resistance in ischemic stroke: mechanisms and therapeutic approaches. Front Endocrinol (Lausanne). 2022;13:1092431.

    Article  PubMed  Google Scholar 

  8. Deng XL, Liu Z, Wang C, Li Y, Cai Z. Insulin resistance in ischemic stroke. Metab Brain Dis. 2017;32:1323–34.

    Article  CAS  PubMed  Google Scholar 

  9. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol. 1979;237:E214–23.

    CAS  PubMed  Google Scholar 

  10. Borai A, Livingstone C, Ferns GA. The biochemical assessment of insulin resistance. Ann Clin Biochem. 2007;44:324–42.

    Article  CAS  PubMed  Google Scholar 

  11. Minh HV, Tien HA, Sinh CT, Thang DC, Chen CH, Tay JC, et al. Assessment of preferred methods to measure insulin resistance in Asian patients with hypertension. J Clin Hypertens (Greenwich). 2021;23:529–37.

    Article  CAS  PubMed  Google Scholar 

  12. Guerrero-Romero F, Simental-Mendía LE, González-Ortiz M, Martínez-Abundis E, Ramos-Zavala MG, Hernández-González SO, et al. The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp. J Clin Endocrinol Metab. 2010;95:3347–51.

    Article  CAS  PubMed  Google Scholar 

  13. McLaughlin T, Reaven G, Abbasi F, Lamendola C, Saad M, Waters D, et al. Is there a simple way to identify insulin-resistant individuals at increased risk of cardiovascular disease? Am J Cardiol. 2005;96:399–404.

    Article  CAS  PubMed  Google Scholar 

  14. Bello-Chavolla OY, Almeda-Valdes P, Gomez-Velasco D, Viveros-Ruiz T, Cruz-Bautista I, Romo-Romo A, et al. METS-IR, a novel score to evaluate insulin sensitivity, is predictive of visceral adiposity and incident type 2 diabetes. Eur J Endocrinol. 2018;178:533–44.

    Article  CAS  PubMed  Google Scholar 

  15. Williams KV, Erbey JR, Becker D, Arslanian S, Orchard TJ. Can clinical factors estimate insulin resistance in type 1 diabetes? Diabetes. 2000;49:626–32.

    Article  CAS  PubMed  Google Scholar 

  16. Borai A, Livingstone C, Kaddam I, Ferns G. Selection of the appropriate method for the assessment of insulin resistance. BMC Med Res Methodol. 2011;11:158.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Bagheri P, Khalili D, Seif M, Rezaianzadeh A. Dynamic behavior of metabolic syndrome progression: a comprehensive systematic review on recent discoveries. BMC Endocr Disord. 2021;21:54.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Wang A, Liu J, Li C, Gao J, Li X, Chen S, et al. Cumulative exposure to high-sensitivity C-reactive protein predicts the risk of cardiovascular disease. J Am Heart Assoc. 2017;6:e005610.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Min Q, Wu Z, Yao J, Wang S, Duan L, Liu S, et al. Association between atherogenic index of plasma control level and incident cardiovascular disease in middle-aged and elderly Chinese individuals with abnormal glucose metabolism. Cardiovasc Diabetol. 2024;23:54.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Zheng H, Chen G, Wu K, Wu W, Huang Z, Wang X, et al. Relationship between cumulative exposure to triglyceride-glucose index and heart failure: a prospective cohort study. Cardiovasc Diabetol. 2023;22:239.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Tian X, Chen S, Xu Q, Xia X, Zhang Y, Wang P, et al. Magnitude and time course of insulin resistance accumulation with the risk of cardiovascular disease: an 11-years cohort study. Cardiovasc Diabetol. 2023;22:339.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Yao J, Zhou F, Ruan L, Liang Y, Zheng Q, Shao J, et al. Association between estimated glucose disposal rate control level and stroke incidence in middle-aged and elderly adults. J Diabetes. 2024;16:e13595.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Ference BA, Braunwald E, Catapano AL. The LDL cumulative exposure hypothesis: evidence and practical applications. Nat Rev Cardiol. 2024;21:701–16.

    Article  CAS  PubMed  Google Scholar 

  24. Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS). Int J Epidemiol. 2014;43:61–8.

    Article  PubMed  Google Scholar 

  25. Chen X, Crimmins E, Hu PP, Kim JK, Meng Q, Strauss J, et al. Venous blood-based biomarkers in the China health and retirement longitudinal study: rationale, design, and results from the 2015 wave. Am J Epidemiol. 2019;188:1871–7.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the reporting of observational studies in epidemiology (STROBE): explanation and elaboration. Epidemiology. 2007;18:805–35.

    Article  PubMed  Google Scholar 

  27. Kim JH. Multicollinearity and misleading statistical results. Korean J Anesthesiol. 2019;72:558–69.

    Article  PubMed  PubMed Central  Google Scholar 

  28. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–45.

    Article  CAS  PubMed  Google Scholar 

  29. Boden-Albala B, Sacco RL. Lifestyle factors and stroke risk: exercise, alcohol, diet, obesity, smoking, drug use, and stress. Curr Atheroscler Rep. 2000;2:160–6.

    Article  CAS  PubMed  Google Scholar 

  30. Austin PC, White IR, Lee DS, van Buuren S. Missing data in clinical research: a tutorial on multiple imputation. Can J Cardiol. 2021;37:1322–31.

    Article  PubMed  Google Scholar 

  31. Opotowsky AR, Webb GD. Population-based data on congenital heart disease and stroke. J Am Heart Assoc. 2016;5:e003257.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Hirose T. Significance of insulin secretion pattern lectured by glinides in the treatment of postprandial hyperglycemia. Nihon Rinsho. 2005;63:2237–44. Japanese.

    PubMed  Google Scholar 

  33. Xu W, Guo Y, Zhao L, Fu R, Qin X, Zhang Y, et al. The aging immune system: a critical attack on ischemic stroke. Mol Neurobiol. 2025;62:3322–42.

    Article  CAS  PubMed  Google Scholar 

  34. Kernan WN, Inzucchi SE, Viscoli CM, Brass LM, Bravata DM, Horwitz RI. Insulin resistance and risk for stroke. Neurology. 2002;59:809–15.

    Article  CAS  PubMed  Google Scholar 

  35. Zhou M, Li H, Wang Y, Pan Y, Wang Y. Causal effect of insulin resistance on small vessel stroke and Alzheimer’s disease: a Mendelian randomization analysis. Eur J Neurol. 2022;29:698–706.

    Article  PubMed  Google Scholar 

  36. Chen W, Wang S, Lv W, Pan Y. Causal associations of insulin resistance with coronary artery disease and ischemic stroke: a Mendelian randomization analysis. BMJ Open Diabetes Res Care. 2020;8:e001217.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Yang Y, Huang X, Wang Y, Leng L, Xu J, Feng L, et al. The impact of triglyceride-glucose index on ischemic stroke: a systematic review and meta-analysis. Cardiovasc Diabetol. 2023;22:2.

    Article  PubMed  PubMed Central  Google Scholar 

  38. Zhang S, Cao C, Han Y, Hu H, Zheng X. A nonlinear relationship between the triglycerides to high-density lipoprotein cholesterol ratio and stroke risk: an analysis based on data from the China health and retirement longitudinal study. Diabetol Metab Syndr. 2024;16:96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Cai X, Hu J, Zhu Q, Wang M, Liu S, Dang Y, et al. Relationship of the metabolic score for insulin resistance and the risk of stroke in patients with hypertension: a cohort study. Front Endocrinol (Lausanne). 2022;13:1049211.

    Article  PubMed  Google Scholar 

  40. Zabala A, Darsalia V, Lind M, Svensson AM, Franzén S, Eliasson B, et al. Estimated glucose disposal rate and risk of stroke and mortality in type 2 diabetes: a nationwide cohort study. Cardiovasc Diabetol. 2021;20:202.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Zhao Y, Zhang J, Chen C, Qin P, Zhang M, Shi X, et al. Comparison of six surrogate insulin resistance indexes for predicting the risk of incident stroke: the rural Chinese cohort study. Diabetes Metab Res Rev. 2022;38:e3567.

    Article  CAS  PubMed  Google Scholar 

  42. Wang X, Feng B, Huang Z, Cai Z, Yu X, Chen Z, et al. Relationship of cumulative exposure to the triglyceride-glucose index with ischemic stroke: a 9-year prospective study in the Kailuan cohort. Cardiovasc Diabetol. 2022;21:66.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Cui H, Liu Q, Wu Y, Cao L. Cumulative triglyceride-glucose index is a risk for CVD: a prospective cohort study. Cardiovasc Diabetol. 2022;21:22.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Lenters-Westra E, Schindhelm RK, Bilo HJ, Slingerland RJ. Haemoglobin A1c: historical overview and current concepts. Diabetes Res Clin Pract. 2013;99:75–84.

    Article  CAS  PubMed  Google Scholar 

  45. Yusuf S, Joseph P, Rangarajan S, Islam S, Mente A, Hystad P, et al. Modifiable risk factors, cardiovascular disease, and mortality in 155 722 individuals from 21 high-income, middle-income, and low-income countries (PURE): a prospective cohort study. Lancet. 2020;395:795–808.

    Article  PubMed  Google Scholar 

  46. Selvin E, Coresh J, Shahar E, Zhang L, Steffes M, Sharrett AR. Glycaemia (haemoglobin A1c) and incident ischaemic stroke: the atherosclerosis risk in communities (ARIC) study. Lancet Neurol. 2005;4:821–6.

    Article  CAS  PubMed  Google Scholar 

  47. Fan J, Li X, Yu X, Liu Z, Jiang Y, Fang Y, et al. Global burden, risk factor analysis, and prediction study of ischemic stroke, 1990–2030. Neurology. 2023;101:e137–50.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Meigs JB, Nathan DM, Cupples LA, Wilson PW, Singer DE. Tracking of glycated hemoglobin in the original cohort of the Framingham heart study. J Clin Epidemiol. 1996;49:411–7.

    Article  CAS  PubMed  Google Scholar 

  49. Mozafar Saadati H, Mehrabi Y, Sabour S, Mansournia MA, Hashemi Nazari SS. Estimating the effects of body mass index and central obesity on stroke in diabetics and non-diabetics using targeted maximum likelihood estimation: atherosclerosis risk in communities study. Obes Sci Pract. 2020;6:628–37.

    Article  PubMed  PubMed Central  Google Scholar 

  50. Chen MQ, Shi WR, Wang HY, Sun YX. Sex differences of combined effects between hypertension and general or central obesity on ischemic stroke in a middle-aged and elderly population. Clin Epidemiol. 2021;13:197–206.

    Article  PubMed  PubMed Central  Google Scholar 

  51. Xia F, Yu X, Li Y, Chen Y, Zhang W, You C, et al. Geographic variations of stroke incidence in Chinese communities: an 18-year prospective cohort study from 1997 to 2015. J Stroke. 2020;22:345–56. https://doiorg.publicaciones.saludcastillayleon.es/10.5853/jos.2020.02383

    Article  PubMed  PubMed Central  Google Scholar 

  52. O’Donnell MJ, Chin SL, Rangarajan S, Xavier D, Liu L, Zhang H, et al. Global and regional effects of potentially modifiable risk factors associated with acute stroke in 32 countries (INTERSTROKE): a case-control study. Lancet. 2016;388:761–75.

    Article  PubMed  Google Scholar 

  53. Chen Z, Mo J, Xu J, Wang A, Dai L, Cheng A, et al. Effects of individual and integrated cumulative burden of blood pressure, glucose, low-density lipoprotein cholesterol, and C-reactive protein on cardiovascular risk. Eur J Prev Cardiol. 2022;29:127–35.

    Article  CAS  PubMed  Google Scholar 

  54. Chen ZM, Mo JL, Yang KX, Jiang YY, Wang CJ, Yang X, et al. Beyond low-density lipoprotein cholesterol levels: impact of prior Statin treatment on ischemic stroke outcomes. Innov (Camb). 2024;5:100713.

    CAS  Google Scholar 

  55. Zhu Z, Yang J, Zhong C, Xu T, Wang A, Bu X, et al. Abnormal glucose regulation, hypoglycemic treatment during hospitalization and prognosis of acute ischemic stroke. J Neurol Sci. 2017;379:177–82.

    Article  CAS  PubMed  Google Scholar 

  56. Lutsep HL. Medical and lifestyle interventions, not procedures, should be the primary focus in stroke prevention. Expert Rev Neurother. 2013;13(8):877–8.

    Article  CAS  PubMed  Google Scholar 

  57. Baden MY, Shan Z, Wang F, Li Y, Manson JE, Rimm EB, et al. Quality of plant-based diet and risk of total, ischemic, and hemorrhagic stroke. Neurology. 2021;96:e1940–53.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  58. Spence JD. Diet for stroke prevention. Stroke Vasc Neurol. 2018;3:44–50.

    Article  PubMed  PubMed Central  Google Scholar 

  59. Edward JA, Cornwell WK 3rd. Impact of exercise on cerebrovascular physiology and risk of stroke. Stroke. 2022;53:2404–10.

    Article  PubMed  PubMed Central  Google Scholar 

  60. Yang W, Kang DW, Ha SY, Lee SH. Drinking patterns and risk of ischemic stroke in middle-aged adults: do beneficial drinking habits indeed exist? Stroke. 2021;52:164–71.

    Article  PubMed  Google Scholar 

  61. Mechera-Ostrovsky T, Gluth S. Memory beliefs drive the memory bias on value-based decisions. Sci Rep. 2018;8:10592.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Glymour MM, Avendano M. Can self-reported strokes be used to study stroke incidence and risk factors? Evidence from the health and retirement study. Stroke. 2009;40:873–9.

    Article  PubMed  Google Scholar 

  63. Ekker MS, Boot EM, Singhal AB, Tan KS, Debette S, Tuladhar AM, et al. Epidemiology, aetiology, and management of ischaemic stroke in young adults. Lancet Neurol. 2018;17:790–801.

    Article  PubMed  Google Scholar 

  64. Hamlin R. The relative merits of observational and experimental research: four key principles for optimising observational research designs. Nutrients. 2022;14:4649.

    Article  PubMed  PubMed Central  Google Scholar 

  65. Colditz GA. Overview of the epidemiology methods and applications: strengths and limitations of observational study designs. Crit Rev Food Sci Nutr. 2010;50(Suppl 1):10–2.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We acknowledge the physicians and nurses who participate in the data collection for China Health and Retirement Longitudinal Study.

Funding

This work was supported by the National Natural Science Foundation of China [81670370 and 82360073]; the Natural Science Foundation of Jiangxi Province [20232BAB216004 and 20224ACB206004], the Outstanding Youth Foundations of Natural Science Foundation of Hebei Province [H2017206381]; and the Office of Education Foundation of Hebei Province of China [SLRC2017046].

Author information

Authors and Affiliations

  1. Department of Cardiology, Jiangxi Provincial People’s Hospital, The First Affiliated Hospital of Nanchang Medical College, Nanchang, Jiangxi, China

    Dezhi Hong & Guotai Sheng

  2. Department of Cardiology, Chongren County People’s Hospital, Fuzhou, Jiangxi, China

    Xiaohui Li

  3. Discipline Construction Office, Jiangxi Provincial People’s Hospital, The First Affiliated Hospital of Nanchang Medical College, Nanchang, Jiangxi, China

    Hongyi Yang

  4. Jiangxi Cardiovascular Research Institute, Jiangxi Provincial People’s Hospital, The First Affiliated Hospital of Nanchang Medical College, Nanchang, Jiangxi, China

    Wei Wang & Yang Zou

Authors
  1. Dezhi Hong
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Xiaohui Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Guotai Sheng
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Hongyi Yang
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Wei Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Yang Zou
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

YZ and WW: Conceptualization, project administration, supervision, and YZ and DZ-H: methodology, writing-original draft preparation. XH-L, GT-S, HH-Y, WW: writing-reviewing and editing. YZ: Software. DZ-H, WW: formal analysis and validation. YZ and GT-S: data curation. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Wei Wang or Yang Zou.

Ethics declarations

Ethics approval and consent to participate

All participants in the CHARLS study provided written informed consent before participation, and subsequent surveys, physical measurements, blood collection, and follow-up were carried out after approval. The implementation of CHARLS was authorized by the Ethics Committee of Peking University (No. 00001052–11015), and all research data were de-identified.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hong, D., Li, X., Sheng, G. et al. Comparing the impact of cumulative insulin resistance surrogates exposure on stroke: optimizing prevention strategies. Lipids Health Dis 24, 158 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12944-025-02579-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12944-025-02579-6

Keywords

Lipids in Health and Disease

ISSN: 1476-511X