Abstract
Objective
Current literature lacks reliable evidence on the association between Total Cholesterol, High-Density Lipoprotein, and Glucose (CHG) and stroke incidence. Therefore, this study aims to investigate the potential association between CHG and stroke risk in a middle-aged and elderly Chinese population.
Methods
This is a prospective cohort study, and the data is derived from the China Health and Retirement Longitudinal Study. Cox proportional hazards regression models were utilized to explore relationships between CHG and stroke incidence. To identify potential nonlinear relationships, Cox proportional hazards regression models with restricted cubic spline functions were constructed. Additionally, Receiver Operating Characteristic (ROC) curves were developed to explore the predictive ability of CHG for stroke.
Results
This study included a total of 10,396 participants, with female accounting for 53.27%. Through five survey cycles from 2011 to 2020, 953 (9.17%) incident strokes occurred. After multivariate adjustment, Cox regression modeling showed that each 1-unit increase in CHG was associated with a 33.5% increased risk of stroke (HR 1.335; 95% CI 1.063–1.677). A nonlinear relationship was identified, with an inflection point at CHG = 4.556. On the left side of that point, each unit increase in CHG was associated with a statistically nonsignificant increase of 12.4% in stroke risk (HR = 1.126; 95% CI: 0.691–1.835), while on the right side of the point, it was linked to a 58.0% increase in risk (HR = 1.580; 95% CI: 1.097–2.274). Additionally, in predicting stroke risk, the AUC value for CHG (0.5737) was higher than that of TyG (0.5679), TC/HDL-c (0.5619), FPG (0.5513), HDL-c (0.5424), and TC (0.5334).
Conclusions
This study found that CHG is positively related to stroke risk and exhibits a non-linear relationship, with a turning point at 4.556. CHG may serve as an early marker for high-risk populations, and proactive monitoring and interventions, including lifestyle changes and management of other metabolic indicators for those with CHG above 4.556, may help lower stroke risk.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12883-026-04647-5.
Keywords: Stroke, CHG, Non-linear association, Lipid and glucose metabolism, Predicted value
Introduction
Stroke is an acute neurological injury caused by interruption of blood flow to the brain (ischemic stroke) or hemorrhage (hemorrhagic stroke), with pathological manifestations characterized by necrosis (infarction) of central nervous system tissue. This damage can be confirmed by imaging within 24 h [1, 2]. As a critical global health challenge, stroke demonstrates strong correlations with elevated mortality rates, prolonged functional disabilities, and restricted therapeutic interventions [3–5]. Epidemiological data from the Global Burden of Disease Study 2019 indicate that stroke has advanced from the fifth to the third most prominent contributor to worldwide health deterioration since 1990 [6]. Projections suggest that without comprehensive preventive strategies, annual stroke-related mortality may potentially escalate to 7–8 million by 2030 [7]. Stroke also imposes a substantial economic burden on both families and society. Therefore, the identification and management of stroke risk factors are crucial for stroke prevention and for alleviating socioeconomic pressure.
Lipid metabolic disorders manifest through an aberrant profile characterized by reduced high-density lipoprotein cholesterol (HDL-c) alongside elevated concentrations of total cholesterol (TC), triglycerides (TG) and low-density lipoprotein cholesterol (LDL-c), significantly predisposing individuals to cardiovascular and cerebrovascular events [8–10]. Multiple lines of research confirm that elevated TC or reduced HDL-c levels are significantly associated with an increased incidence of stroke. Moreover, the ratio of TC to HDL-c has also been shown to be closely related to the risk of atherosclerosis, cardiovascular events, and stroke [11–13]. Abnormal glucose metabolism is recognized as a common risk factor for stroke, diabetes mellitus (DM), and metabolic syndrome [14–16]. Several studies have shown that higher fasting plasma glucose (FPG) levels are positively linked with the risk of stroke [17]. Therefore, we speculate that a composite indicator integrating TC, HDL-C, and FPG may help capture metabolic statuses that single indicators cannot reveal, and these metabolic statuses may be closely related to the risk of stroke.
In recent years, a novel index called total cholesterol, HDL-c, and glucose (CHG), composed of TC, HDL-c, and FPG, has been proposed by Amin Mansoori et al. [18]. Studies have shown that the CHG index has significant value in predicting the risk of DM and its complications, as well as in the assessment of cardiovascular events [18–20]. However, no studies have yet evaluated the association between the CHG and the risk of stroke. Leveraging longitudinal data from the China Health and Retirement Longitudinal Study (CHARLS) spanning 2011–2020, we conducted a prospective cohort study to assess the relationship between CHG and stroke risk.
Methods
Study design and data source
Data were drawn from the CHARLS, a nationally representative longitudinal cohort aimed at evaluating the economic, social, and health dimensions of the Chinese population. Utilizing stratified multistage probability sampling, researchers recruited participants comprising 10,257 households at baseline, spanning 450 communities within 150 counties across 28 provinces [21]. Participants aged ≥ 45 years at baseline were enrolled during a comprehensive survey conducted from June 2011 to March 2012. Data collection utilized standardized questionnaires administered in face-to-face interviews, with subsequent waves performed at two-year intervals. Ethical clearance was obtained from the Biomedical Ethics Review Board of Peking University, China (IRB00001052-11015). All participants provided informed written consent prior to study involvement. The datasets and supplementary materials related to this study were made publicly available through the official CHARLS project website [21].
Variables
The total cholesterol, high-density lipoprotein, and glucose index
CHG index was evaluated as a continuous measure. The index was computed according to the formula: CHG = Ln [TC*FPG/2*HDL-c], where TC, FPG, and HDL-c were measured in mg/dl [18].
Stroke diagnosis and follow-up
Participants who had not experienced stroke at baseline but reported incident stroke during follow-up were recorded as new cases. Stroke-related information was collected through a questionnaire system, which asked participants whether they had ever been diagnosed with stroke by a physician, the date of diagnosis or the date they became aware of the condition, and whether they were receiving treatment for stroke [21, 22]. During follow-up, reported stroke events were classified as incident cases, with the self-reported date regarded as the event onset time. Researchers calculated the interval from baseline assessment to stroke occurrence to determine the time-to-event. For participants without reported stroke throughout the entire follow-up period, the follow-up duration was defined from initial assessment to the final survey date [21].
Data collection and covariates
Baseline information of the participants was collected by trained researchers using standardized questionnaires and examinations [21]. This process mainly included the following aspects: (1) Demographic and lifestyle characteristics, such as age, sex, drinking status, smoking status, and physical activity; (2) Medical history, recording information on chronic diseases including chronic lung disease (CLD), hypertension, chronic kidney disease (CKD), coronary heart disease (CHD), and DM; (3) Physical assessments, measuring height, weight, waist circumference (WC), systolic blood pressure (SBP), and diastolic blood pressure (DBP); and (4) Biochemical indicators, including LDL-c; platelet count (PLT); uric acid (UA); blood urea nitrogen (BUN); HDL-c; C-reactive protein (CRP); FPG; glycated hemoglobin (HbA1c); hemoglobin (HGB); serum creatinine (Scr); blood cell count (WBC); TC; TG; cystatin C. Covariates were selected based on previous literature and clinical expertise [22, 23].
The definitions of Triglyceride-Glucose (TyG) Index and TG/HDL-c were presented as follows: TyG index was defined as ln[FPG (mg/dL)×TG (mg/dL)/2] [24]. TG/HDL-c was defined as the ratio of triglycerides to high-density lipoprotein cholesterol. BMI was determined by dividing body weight (kg) by height squared (m²).
Missing data processing
In this study, some covariates had missing data, with the number and percentage of missing entries as follows: BUN (1, 0.01%), drinking status (43, 0.41%), CLD (76, 0.73%), CHD (85, 0.82%), HbA1c (86, 0.83%), Hypertension (87, 0.84%), CKD (94, 0.90%), DM (133, 1.28%), PLT (228, 2.19%), HB (229, 2.20%), WBC (231, 2.22%), Smoking status (232, 2.23%), SBP (1521, 14.63%), WC (1536, 14.77%), DBP (1554, 14.94%), BMI (1619, 15.57%), Cystatin C (2500, 20.05%), Physical activity (6518, 62.70%), and MCV (228, 2.19%).Missing data potentially undermined the statistical validity of our analytical sample during model development. To minimize bias from these missing variables, we applied a multiple imputation technique to deal with the missing data [25, 26]. Covariates included in the imputation model were age, sex, drinking status, smoking status, physical activity, WBC, PLT, BUN, Scr, TG, LDL-c, CRP, HbA1c, UA, HGB, Cystatin C, SBP, DBP, BMI, WC, Hypertension, DM, CLD, CHD, CKD. The missing value filling process was executed using a linear regression method with ten iterations. The missing data analysis was based on the missing at random (MAR) assumption [25].
Statistical analysis
Participants were stratified into CHG tertiles to facilitate intergroup comparisons of baseline characteristics. Continuous data following normal distribution were summarized using mean ± standard deviation (SD), whereas those with non-normal distribution were expressed as median with interquartile range (IQR). Categorical data were reported as frequencies and percentages. For group comparisons, one-way analysis of variance (ANOVA) or the Kruskal-Wallis’s test was selected according to the distributional characteristics of continuous variables. Chi-square (χ²) analysis evaluated categorical variable differences across study groups.
Both unadjusted and adjusted Cox proportional-hazards regression models were performed to evaluate the relationship between CHG and stroke incidence. It is crucial to note that, prior to the analysis, the proportional hazards (PH) assumption was assessed using Schoenfeld residuals to validate the effectiveness of the model. The results indicated that the p-value for the PH assumption test was greater than 0.05, confirming the validity of the assumption and thereby supporting the application of the Cox proportional hazards regression analysis. Three models were used: Model I (unadjusted for any covariates), Model II (age, WC, sex, and BMI were adjusted), and Model III (age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status were adjusted). Due to multicollinearity between TG, SBP, and other variables, they were excluded from the multivariate model (see Supplementary Table S1). In addition, the study found that WBC, PLT, and CRP are unequally distributed among the quartiles of CHG, which indicates a significant association between these factors and CHG. Secondly, studies shows that elevated levels of WBC and CRP are associated with enhanced inflammatory responses, which may lead to endothelial dysfunction and atherosclerosis [27–29]. PLT activation is also closely related to the inflammatory process, further increasing the risk of stroke by affecting coagulation and thrombosis [30].Furthermore, there is currently no evidence that CHG directly causes changes in WBC and PLT, thereby affecting the incident of stroke. Therefore, these acute-phase reactants (such as PLT, CRP, and WBC) are included as confounding factors in the Cox regression model.
Significant associations between DM, obesity, hypertension, CKD, alcohol consumption, and stroke had been confirmed by previous studies [31–35]. To verify the robustness of the present findings, multiple sensitivity analyses were performed. First, sensitivity analysis was conducted after excluding participants with a BMI ≥ 28 kg/m² [36]. Additionally, the association between CHG and the risk of stroke was evaluated in participants without CKD, hypertension, or DM and who were never drinkers. Moreover, a generalized additive model (GAM) was introduced in the multivariable Cox proportional-hazards regression model to incorporate continuous covariates as curves into the model. Considering the potential bias introduced during the multiple imputations of covariates, a complete case sensitivity analysis was performed, excluding participants with missing values for any covariates, to reassess the association between CHG and stroke. Additionally, a Cox proportional hazards regression model that does not include the acute phase reactants WBC, PLT, and CRP were constructed to further verify the robustness of the findings. Finally, E-values were calculated to assess the potential impact of unmeasured confounders on the observed association between CHG and the risk of stroke [37].
We conducted subgroup investigations utilizing stratified Cox proportional hazards regression models, with stratification by SBP, CHD, smoking, physical activity, DBP, sex, and age. Continuous variables were categorized in accordance with established clinical thresholds: SBP was stratified at < 140 mmHg versus ≥ 140 mmHg, while DBP thresholds were set at < 90 mmHg versus ≥ 90 mmHg [38]. Multivariate model adjustments incorporated potential confounding variables (DBP, AST, drinking status, TG, physical activity, Scr, age, hypertension, sex, FPG, smoking status, ALT, and SBP), while simultaneously excluding initial stratification parameters. To assess potential interactive effects, we employed likelihood ratio tests by comparing models with and without interaction terms.
To explore the potential nonlinear association between CHG and the risk of stroke, Cox proportional hazards models with restricted cubic spline functions were employed. Upon identifying nonlinear associations, a recursive algorithm determined turning points. Two-segment Cox proportional hazard models were then built flanking this turning point.
Finally, Receiver Operating Characteristic (ROC) curve was constructed to assess the predictive capabilities of TC, FPG, HDL-c, TC/HDL-c, TyG and CHG for stroke risk, and the corresponding area under the curve (AUC), best threshold, sensitivity, and specificity were calculated.
Results
Characteristics and inclusion of study participants
Our study was based on multiple waves of data from CHARLS conducted in 2011, 2013, 2015, 2018,and 2020 [21]. The baseline survey in 2011–2012 included a total of 17,708 Chinese general population aged 45 years and older [21]. To further define the study population, several exclusion criteria were applied. First, 486 participants who had experienced a stroke at the time of the 2011 survey were excluded; next, 59 individuals with incomplete stroke information were excluded. 1,060 individuals lost to follow-up were omitted. Furthermore, 5,200 participants were excluded due to missing FPG, TC, or HDL-c data at baseline. We also excluded 196 participants with abnormal or extreme CHG values. Finally, 311 individuals younger than 45 years were excluded from the study. After applying these criteria, a total of 10,396 participants were included in the final analysis. Details of the participant selection process are shown in Fig. 1.
Fig. 1.
Flowchart illustrating the study participants
Table 1 presents the demographic and clinical characteristics of 10,396 participants, of whom 53.27% were female. CHG values were approximately normally distributed, with a range of 3.43 to 5.63 and a mean (± SD) of 4.49(± 0.27) (Fig. 2). Participants were divided into three groups according to CHG tertiles: T1(< 4.3), T2(4.35–4.53), and T3(≥ 4.53). Compared with T1, higher tertiles exhibited higher levels of age, TC, FPG, WBC, HbA1c, BUN, TG, TyG, TC/HDL-c, UA, SBP, HGB, and Scr, while PLT, and LDL-c levels were relatively lower. In addition, the higher tertiles had greater proportions of participants with hypertension, DM, and those engaging in high physical activity or current drinking, as well as a higher proportion of females, compared to T1.
Table 1.
The baselined characteristics of participants
| CHG tertile | T1(< 4.35) | T2(4.35–4.53) | T3(≥ 4.53) | P-value |
|---|---|---|---|---|
| N | 3465 | 3465 | 3466 | |
| age (years, mean ± SD) | 58.63 ± 9.26 | 59.24 ± 9.32 | 59.60 ± 9.38 | < 0.001 |
| WBC(109/L, mean ± SD) | 6.09 ± 1.80 | 6.16 ± 1.84 | 6.38 ± 1.93 | < 0.001 |
| PLT (109/L, mean ± SD) | 217.11 ± 74.43 | 210.76 ± 71.88 | 206.02 ± 71.51 | < 0.001 |
| BUN (mmol/L, mean ± SD) | 15.59 ± 4.35 | 15.73 ± 4.61 | 15.93 ± 4.71 | 0.008 |
| FPG (mg/dL, mean ± SD) | 92.27 ± 10.82 | 103.07 ± 9.50 | 128.26 ± 38.50 | < 0.001 |
| Scr (mg/dL, mean ± SD) | 0.78 ± 0.19 | 0.78 ± 0.21 | 0.79 ± 0.27 | 0.008 |
| TC (mg/dL, mean ± SD) | 187.10 ± 37.61 | 191.43 ± 35.21 | 196.66 ± 38.96 | < 0.001 |
| TG (mg/dL, median, quartile) | 94.69 (71.68–126.56.68.56) | 102.66 (73.46–144.26.46.26) | 131.87 (82.31–213.06.31.06) | < 0.001 |
| HDL-c (mg/dL, mean ± SD) | 51.62 ± 18.26 | 52.88 ± 14.21 | 49.78 ± 11.91 | < 0.001 |
| LDL-c (mg/dL, mean ± SD) | 134.49 ± 32.35 | 117.47 ± 28.10 | 100.09 ± 31.62 | < 0.001 |
| CRP (mmol/L, median, quartile) | 0.98 (0.54–1.98) | 0.97 (0.52–2.03) | 1.09 (0.57–2.40) | < 0.001 |
| HBA1C (%, mean ± SD) | 5.08 ± 0.44 | 5.15 ± 0.48 | 5.49 ± 1.06 | < 0.001 |
| UA (mg/dL, mean ± SD) | 4.38 ± 1.20 | 4.40 ± 1.21 | 4.57 ± 1.31 | < 0.001 |
| HGB (g/dL, mean ± SD) | 14.29 ± 2.29 | 14.33 ± 2.17 | 14.43 ± 2.24 | 0.024 |
| Cystatin C (mg/L, mean ± SD) | 1.01 ± 0.25 | 1.01 ± 0.26 | 1.00 ± 0.29 | 0.191 |
| SBP(mmHg, mean ± SD) | 134.22 ± 55.18 | 134.52 ± 57.03 | 137.82 ± 56.89 | 0.013 |
| DBP(mmHg, mean ± SD) | 77.42 ± 20.58 | 77.02 ± 21.39 | 78.49 ± 21.28 | 0.011 |
| BMI (kg/m2, mean ± SD) | 23.46 ± 3.79 | 23.39 ± 3.73 | 23.77 ± 3.97 | < 0.001 |
| WC (cm, mean ± SD) | 84.07 ± 12.05 | 83.57 ± 12.45 | 85.24 ± 12.97 | < 0.001 |
| TyG | 8.39 ± 0.44 | 8.57 ± 0.48 | 9.02 ± 0.70 | < 0.001 |
| TC/HDL-c | 3.84 ± 1.15 | 4.22 ± 1.11 | 4.04 ± 1.62 | < 0.001 |
| Hypertension (N, %) | 799 (23.06%) | 878 (25.34%) | 1002 (28.91%) | < 0.001 |
| Sex | < 0.001 | |||
| Feamle | 1848 (53.33%) | 1939 (55.96%) | 1751 (50.52%) | |
| Male | 1617 (46.67%) | 1526 (44.04%) | 1715 (49.48%) | |
| DM (N, %) | 77 (2.22%) | 108 (3.12%) | 393 (11.34%) | < 0.001 |
| CLD (N, %) | 305 (8.80%) | 333 (9.61%) | 352 (10.16%) | 0.155 |
| CHD (N, %) | 408 (11.77%) | 391 (11.28%) | 450 (12.98%) | 0.082 |
| CKD (N, %) | 205 (5.92%) | 212 (6.12%) | 181 (5.22%) | 0.244 |
| Physical | 0.045 | |||
| Low | 903 (26.06%) | 868 (25.05%) | 901 (26.00%) | |
| Moderate | 1205 (34.78%) | 1120 (32.32%) | 1132 (32.66%) | |
| High | 1357 (39.16%) | 1477 (42.63%) | 1433 (41.34%) | |
| Drinking status (N, %) | < 0.001 | |||
| Current | 1051 (30.33%) | 1119 (32.29%) | 1275 (36.79%) | |
| Ever | 271 (7.82%) | 270 (7.79%) | 297 (8.57%) | |
| Never | 2143 (61.85%) | 2076 (59.91%) | 1894 (54.65%) | |
| Smoking status (N, %) | 0.012 | |||
| Current | 1035 (29.87%) | 1004 (28.98%) | 1094 (31.56%) | |
| Ever | 286 (8.25%) | 268 (7.73%) | 317 (9.15%) | |
| Never | 2144 (61.88%) | 2193 (63.29%) | 2055 (59.29%) |
Abbreviations: CHG Total Cholesterol, High-Density Lipoprotein, and Glucose, WBC White Blood Cell, PLT Platelet, BUN Blood Urea Nitrogen, FPG Fasting Plasma Glucose, Scr Serum Creatinine, TC Total Cholesterol, TG Triglyceride, HDL-c High-Density Lipoprotein Cholesterol, LDL-c Low-Density Lipoprotein Cholesterol, CRP C-Reactive Protein, HBA1c Hemoglobin A1c, UA Uric Acid, HGB Hemoglobin, SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, BMI Body Mass Index, WC Waist Circumference, DM Diabetes Mellitus, CLD Chronic Lung Disease, CHD Coronary Heart Disease, CKD Chronic Kidney Disease, SD standard deviation, TC/HDL-c Total Cholesterol to High-Density Lipoprotein Cholesterol Ratio, TyG Triglyceride-Glucose Index
Fig. 2.
CHG distribution profile. The data exhibited a near-normal distribution, extending across the interval 3.43–5.63, with a mean (± SD) of 4.49(± 0.27)
The incidence of stroke
As indicated in Table 2, a total of 953 stroke events occurred among the study participants. The overall incidence of stroke was 113.97 cases per 10,000 person-years. The incidence of stroke in the CHG tertile groups were 106.66, 107.53, and 129.19 per 10,000 person-years for T1, T2, and T3, respectively. The overall cumulative incidence of stroke was 9.17%. The cumulative incidences of stroke in the CHG tertile groups were 8.66% for T1,8.66%for T2, and 10.18% for T3.
Table 2.
Incidence of stroke (% or per 10,000 person-year)
| Participants(n) | Stroke events(n) | Incidence rate (95% CI) (%) | Per 10,000 person-year | |
|---|---|---|---|---|
| Total | 10,396 | 953 | 9.17(8.61–9.72) | 113.97 |
| T1 | 3465 | 300 | 8.66(7.72–9.59) | 106.66 |
| T2 | 3465 | 300 | 8.66(7.72–9.59) | 107.53 |
| T3 | 3466 | 353 | 10.18(9.18–11.19) | 129.19 |
| P for trend | < 0.001 |
Abbreviations: CI confidence, n number
When stratified by age (< 50 years,50–60 years,60–70 years, and ≥ 70 years), the incidence of stroke was higher in men than in women across all age groups (Fig. 3).Furthermore, the incidence of stroke increased with age in both males and females, except in the ≥ 70 years age group.
Fig. 3.
Incidence of stroke according to sex and decadal age categories
Relationship between CHG and the risk of stroke
To examine the potential relationship between CHG and stroke incidence, we employed three Cox proportional hazards regression models. The initial unadjusted Model I revealed that each 1-unit increment in CHG corresponded to a 64% increase in stroke risk (HR = 1.640, 95% CI: 1.316–2.044). After adjustment for sex, age, WC, and BMI, Model II consistently demonstrated a significant association, with a one-unit CHG elevation linked to a 46.0% increased risk of stroke (HR = 1.460, 95% CI: 1.169, 1.823). Within the comprehensively adjusted Model III, the observed relationship remained statistically robust, demonstrating that every one-unit elevation in CHG was linked to a 33.5% increased risk of stroke (HR = 1.335, 95% CI: 1.063–1.677) (Table 3).
Table 3.
The relationship between CHG and the risk of stroke in different models
| Exposure | Model I(HR,95%CI) | Model II(HR,95%CI) | Model III(HR,95%CI) | Model IV(HR,95%CI) |
|---|---|---|---|---|
| CHG per 1-unit | 1.640 (1.316, 2.044) < 0.001 | 1.460 (1.169, 1.823) <0.001 | 1.335 (1.063, 1.677) 0.013 | 1.326 (1.055, 1.668) 0.016 |
| CHG quartile | ||||
| T1 | Ref | Ref | Ref | Ref |
| T2 | 1.008 (0.859, 1.183) 0.922 | 1.003 (0.855, 1.177) 0.971 | 1.003 (0.854, 1.178) 0.968 | 1.001 (0.852, 1.175) 0.997 |
| T3 | 1.225 (1.050, 1.428) 0.009 | 1.153 (1.008, 1.345) 0.048 | 1.122 (0.958, 1.313) 0.153 | 1.107 (0.945, 1.297) 0.209 |
| P for trend | < 0.05 | < 0.05 | < 0.05 | < 0.05 |
Abbreviations: CHG Total Cholesterol, High-Density Lipoprotein, and Glucose, CI confidence, HR Hazard Ratio
Model I: No covariates were adjusted.
Model II: BMI, WC, sex, and age were adjusted
Model III: Variables adjusted for included age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model IV: Variables adjusted for included age(smooth), DM, CHD, drinking status, CKD, CRP (smooth), DBP (smooth), HbA1c(smooth), PLT (smooth), WC (smooth), WBC (smooth), Scr (smooth), sex, BMI (smooth), physical activity, and smoking status were adjusted
In addition, CHG was converted from a continuous variable to a tertile-based categorical variable and re-entered into the Cox proportional hazards regression models. In Table 3 Model III, using the lowest tertile (T1) as the reference group, after multivariable adjustment, the HRs for stroke risk were: 1.003 (95% CI: 0.854–1.178) for T2 and 1.122 (95% CI: 0.958–1.313) for T3. This indicated that compared with T1, the stroke risk in T2 showed no significant change, while T3 showed a 12.2% increase in stroke risk that was not statistically significant. However, the overall trend of stroke risk from T1 to T3 was upward (P for trend < 0.05).
Sensitivity analysis
Result stability was assessed through several sensitivity analyses. Initially, GAM was applied to incorporate continuous covariates as smooth functions, yielding broadly aligned outcomes with the comprehensively adjusted Model III (Table 3, Model IV comparison). Specifically, 1-unit CHG increments elevated stroke risk by 32.6% (HR = 1.326 95%CI: 1.055–1.668). When focusing exclusively on participants with BMI < 28 kg/m², following covariate adjustment, CHG rises maintained significant positive associations with stroke (HR = 1.342,95%CI: 1.038–1.736).Similarly, after excluding participants with DM, similar results were observed: a 1-unit increase in CHG was associated with an HR of 1.313 (95% CI:1.020–1.689) for stroke risk. Moreover, a significant positive association between CHG (per-unit increase) and stroke risk was still observed in participants without hypertension (HR = 1.398,95% CI:1.012–1.932). This association also persisted among participants without CKD, with an HR of 1.289 (95% CI:1.014–1.639) for stroke risk per unit increase in CHG. Similarly, among never-drinkers, the HR for stroke risk associated with a per-unit increase in CHG was 1.626 (95% CI: 1.201, 2.203). Furthermore, the Cox proportional hazards regression model that did not include the acute phase reactants WBC, PLT, and CRP showed results consistent with the fully adjusted model III, indicating that for each 1-unit increase in CHG, the stroke risk increased by 36.0% (HR = 1.360, 95% CI: 1.083–1.707). Finally, a sensitivity analysis was conducted after excluding participants with missing values for any covariates in the complete case analysis. The results indicated that the association between CHG and stroke had an HR of 1.641 (95% CI: 1.034–2.607), which was consistent with the findings from the analysis of multiple imputation data (Table 4).
Table 4.
CHG in relation to stroke risk in different sensitivity analyses
| Exposure | Participants(n) | Stroke-events(n) | HR (95%CI) P-value |
|---|---|---|---|
| Model I | 9117 | 792 | 1.342 (1.038, 1.736) 0.025 |
| Model II | 9818 | 872 | 1.313 (1.020, 1.689) 0.034 |
| Model III | 7717 | 518 | 1.398 (1.012, 1.932) 0.042 |
| Model IV | 9804 | 883 | 1.289 (1.014, 1.639) 0.038 |
| Model V | 6113 | 538 | 1.626 (1.201, 2.203) 0.002 |
| Model VI | 10,396 | 953 | 1.360 (1.083, 1.707) 0.008 |
| Model VII | 3206 | 278 | 1.641 (1.034, 2.607) 0.036 |
Abbreviations: CI confidence, HR Hazard Ratio, n number
Model I was a sensitivity analysis in participants with BMI<28kg/m2; Variables adjusted for included age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model II was a sensitivity analysis in participants without DM; Variables adjusted for included age, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model III was a sensitivity analysis in participants without hypertension; Variables adjusted for included age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model IV was a sensitivity analysis in participants without CKD; Variables adjusted for included age, DM, CHD, drinking status, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model V was a sensitivity analysis in participants with never drinking; Variables adjusted for included age, DM, CHD, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Model VI is a sensitivity analysis of adjusted variables that do not involve acute phase reactants such as CRP, PLT, and WBC. The adjusted variables include age, DM, CHD, drinking status, CKD, DBP, HbA1c, WC, Scr, sex, BMI, physical activity, and smoking status
Model VII is a complete case sensitivity analysis(before multiple imputation), with adjusted variables including age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Additionally, an E-value of 2.0 was calculated, which was found to be greater than the relative risk for the association between CHG and potential unmeasured confounders (1.6) but less than the relative risk for the association between unmeasured confounders and stroke (3.02). It was thereby indicated that unidentified or unaccounted confounders were unlikely to have exerted a substantial impact on the association between CHG and stroke risk. The reliability and robustness of our findings were strengthened by these sensitivity analyses.
Subgroup analysis
Through prespecified subgroup evaluations (Table 5), we observed no meaningful interaction effects between CHG and the variables SBP, DBP, age, sex, physical activity, smoking, and CHD (all P for interaction > 0.05). The results suggest that these factors did not significantly influence or modify the association between CHG and stroke risk.
Table 5.
Stratified associations between CHG and the risk of stroke by age, sex, DBP, physical activity, smoking, CHD, and SBP
| Characteristic | No of participants(n) | HR (95% CI) | P value | P for interaction |
|---|---|---|---|---|
| Age (years) | 0.996 | |||
| < 50 | 2002 | 1.437 (0.763, 2.707) | 0.262 | |
| 50–60 | 3802 | 1.332 (0.910, 1.950) | 0.141 | |
| 60–70 | 3001 | 1.408 (0.978, 2.027) | 0.065 | |
| ≥ 70 | 1591 | 1.402 (0.818, 2.403) | 0.219 | |
| Sex | 0.083 | |||
| Male | 4858 | 1.668 (1.225, 2.271) | 0.001 | |
| Female | 5538 | 1.134 (0.820, 1.567) | 0.447 | |
| SBP (mmHg) | 0.819 | |||
| < 140 | 6767 | 1.334 (0.973, 1.828) | 0.073 | |
| ≥ 140 | 3629 | 1.403 (1.023, 1.925) | 0.036 | |
| DBP (mmHg) | 0.441 | |||
| < 90 | 8618 | 1.278 (0.971, 1.682) | 0.080 | |
| ≥ 90 | 1778 | 1.533 (1.047, 2.243) | 0.028 | |
| Physical Activity | 0.996 | |||
| Low | 2672 | 1.395 (0.955, 2.038) | 0.085 | |
| Moderate | 3113 | 1.362 (0.917, 2.025) | 0.126 | |
| High | 2622 | 1.382 (0.948, 2.014) | 0.093 | |
| Smoking | 0.583 | |||
| Current | 2672 | 1.255 (0.846, 1.863) | 0.260 | |
| Ever | 3457 | 1.110 (0.574, 2.145) | 0.756 | |
| Never | 4267 | 1.518 (1.127, 2.045) | 0.006 | |
| CHD | 0.881 | |||
| No | 9147 | 1.393 (1.078, 1.799) | 0.011 | |
| Yes | 1249 | 1.339 (0.844, 2.124) | 0.215 |
Abbreviations: n number, HR Hazard ratios, CI confidence, SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, CHD Coronary Heart Disease
Above model adjusted for age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
In each case, the model is not adjusted for the stratification variable
Non-linear relationship between CHG and the risk of stroke
A Cox proportional hazards regression model with restricted cubic spline functions was employed, and a nonlinear association between CHG and the risk of stroke was identified (
P for nonlinearity < 0.05; Fig. 4). An inflection point of CHG at 4.556 was determined by a recursive algorithm. Above the inflection point, there were 3,139 participants, with 328 stroke events occurring; below the inflection point, there were 7,257 participants, with 625 stroke events occurring. A two-piece Cox regression model was applied to estimate the HR and CI on either side of the inflection point. Above the inflection point, each unit increase in CHG was associated with a 58.0% increase in stroke risk (HR = 1.580, 95% CI: 1.097–2.274). Conversely, below the inflection point, each unit increase in CHG was associated with a statistically nonsignificant increase of 12.6% in stroke risk (HR = 1.126, 95% CI: 0.691–1.835) (Table 6).
Fig. 4.
Nonlinear relationship between CHG and the risk of stroke
Table 6.
The result of two-piecewise Cox regression model
| Outcome: Incident stroke | Participants(n) | Stroke-events(n) | HR (95%CI) P-value |
|---|---|---|---|
| Inflection points of CHG | 4.556 | ||
| < 4.556 | 7257 | 625 | 1.126 (0.691, 1.835) 0.632 |
| ≥4.556 | 3139 | 328 | 1.580 (1.097, 2.274) 0.014 |
Adjustments included age, DM, CHD, drinking status, CKD, CRP, DBP, HbA1c, PLT, WC, WBC, Scr, sex, BMI, physical activity, and smoking status
Abbreviations: n number, CHG Total Cholesterol, High-Density Lipoprotein, and Glucose, HR Hazard ratios, CI confidence
ROC analysis of the predictive value of CHG, TC, HDL-c, FPG, TC/HDL-c, and TyG for stroke
The predictive capabilities of CHG, TC, HDL-c, FPG, the TC/HDL-c ratio, and TyG for stroke risk were assessed by constructing ROC curves (Supplementary Fig. S1). The results showed the AUC values for each variable in the following order: CHG: 0.5737 > TyG: 0.5679 > TC/HDL-c: 0.5619 > FPG: 0.5513 > HDL-c: 0.5424 > TC: 0.5334. The Youden index values for CHG, TC, HDL-c, FPG, the TC/HDL-c ratio, and TyG were 0.1200, 0.0616, 0.0679, 0.0758, 0.1058, and 0.1100, respectively, with corresponding best cutoff values of 399.7645, 188.8541, 56.6369, 110.4300, 3.4633, and 8.5876 (Supplementary Table S2). These findings indicate that CHG has the highest Youden index and AUC values compared to TC, HDL-c, FPG, the TC/HDL-c ratio, and TyG, demonstrating superior the predictive capability for stroke risk.
Discussion
This study evaluated the association between CHG and stroke incidence in the middle-aged and elderly population in China. The results show that there are an independent positive relationship and a non-linear relationship between CHG and them. A inflection point at 4.556 was identified, where the association between CHG and stroke risk differed on either side of this point. Additionally, the ROC curve analysis demonstrated that CHG is superior to any of its individual components in predicting stroke risk.
Prior research demonstrates that raised TC/HDL-C ratios correlate with higher stroke incidence [11]. The CHARLS cohort investigation examined 10,184 Chinese participants ≥ 45 years, revealing per-unit TC/HDL-c increments corresponded to stroke HR of 1.05(95% CI:1.00–1.10; P = 0.0410) [11]. Another prospective study analysis of 27,937 American women ≥ 45 years stratified TC/HDL-C into quintile groups: following multivariable corrections, the uppermost quintile compared with the lowest demonstrated a 65% elevated stroke risk (HR:1.65,95%CI:1.06–2.58) [13]. In addition, FPG is also a common risk factor for stroke [17]. Recently, a meta-analysis (involving 2,555,666 participants) showed that, compared with the lowest FPG group, the highest group had a stroke relative risk(RR) of 1.79(95% CI: 1.68–1.91); in non-DM populations, this association still existed, with a RR of 1.16 (95% CI: 1.11–1.21) [17]. However, the composite index of integrated lipids and blood glucose—CHG index—may provide a more comprehensive reflection of the body’s metabolic status and its assessment of stroke risk. By combining TC, HDL-c, and FPG, CHG not only considers these known risk factors but also emphasizes the interactions among them. Therefore, we hypothesized that elevated levels of CHG are associated with an increased risk of stroke and that CHG may have important incremental value in predicting stroke risk compared to single indicators. Notably, CHG is linked to elevated cardiovascular disease risk [19]. In particular, a Chinese cohort analysis encompassing 6,249 adults aged ≥ 45 years demonstrated that each 1-unit increment in CHG corresponded to a 20% higher risk of cardiovascular disease following multivariable adjustments (HR = 1.20, 95% CI: 1.07–1.34) [19]. However, there are currently no studies on the relationship between CHG and the risk of stroke. Our study confirms this hypothesis: elevated CHG levels are significantly positively associated with stroke risk. ROC curve analysis results also indicate that CHG has slightly better predictive efficacy for stroke compared to its single indicators and commonly used insulin resistance indicators (such as TyG and TC/HDL-c ratio). To minimize information loss and more accurately quantify the relationship between CHG and stroke, we employed both categorical and continuous variable analyses. Meanwhile, we performed a series of sensitivity analyses to verify the robustness of our findings, including analyses restricted to participants with BMI < 28 kg/m², without hypertension, DM, or CKD, and those who had never consumed alcohol. Additionally, a complete case sensitivity analysis was performed. In summary, the confirmed relationship between high CHG and stroke has significant clinical implications. Incorporating CHG into routine clinical assessments may assist healthcare professionals in identifying high-risk populations for stroke and evaluating their metabolic status. As a convenient and cost-effective biomarker derived from standard laboratory tests (such as TC, HDL-c, and FPG), CHG has the potential to serve as a screening tool for stroke, particularly in resource-limited areas where specialized services are restricted. Furthermore, this indicator may optimize screening strategies, enabling clinicians to prioritize identification of individuals who require enhanced monitoring or early referral.
When stratified by CHG tertiles, the multivariable-adjusted model showed that, compared with T1, the HRs for the risk of stroke in T2 and T3 were 1.003 and 1.122, respectively. This indicates that the trend in HR values remained largely stable from the T1 to T2, but showed an upward shift in the T3, suggesting a potential nonlinear relationship between CHG and stroke risk. For hypothesis validation, Cox hazards regression modeling incorporating cubic splines was utilized, finding a non-linear relationship linking CHG to stroke risk and identifying an inflection point at 4.556. When CHG was below 4.556, each 1‑unit increase was associated with a 12.6% increase in stroke risk, which was not statistically significant. When CHG exceeded 4.556, each 1‑unit increase in CHG was associated with a 58.0% increased risk of stroke. This indicates that the inflection point of 4.556 represents a risk stratification point: as CHG increases, the risk of stroke does not significantly rise; however, once this inflection point is exceeded, the risk of stroke increases significantly with rising CHG levels. Additionally, based on this study data, it was found that participants with CHG values near the inflection point of 4.556 exhibited typical characteristics of the CHG components: TC ≈ 195–200 mg/dL, FPG ≈ 110–115 mg/dL, and HDL-c ≈ 51–53 mg/dL. This suggests that participants at this inflection point display mildly elevated FPG levels, TC approaching the upper limit, but maintain relatively low HDL-c levels. Furthermore, we compared the baseline characteristics of participants with CHG < 4.556 and CHG ≥ 4.556. The results showed that participants with CHG ≥ 4.556 had higher levels of age, WBC, WC, TG, CRP, UA, SBP, DBP, and BMI, and a higher proportion of hypertension, DM, low physical activity, alcohol consumption and smoking behaviors (Supplementary Table S3). This further indicates that participants with CHG above this inflection point are often accompanied by other metabolic abnormalities, such as elevated HbA1c, blood pressure, and UA levels, and typically exhibit a lack of physical activity and unhealthy lifestyle habits. Based on these findings, it is recommended that participants with CHG greater than 4.556 receive more proactive monitoring and intervention, including lifestyle adjustments (such as increasing physical activity and improving dietary habits) and management of other relevant metabolic indicators, which may help reduce the risk of stroke. Additionally, incorporating CHG as a routine health check indicator may assist clinicians in identifying potential high-risk populations for stroke at an earlier stage. This risk stratification strategy provides practical guidance for clinical practice and aids in formulating personalized prevention measures.
The specific association between CHG and the pathogenesis of stroke has not been fully elucidated, but it may be related to abnormalities in lipid and glucose metabolism. Elevated blood glucose promotes the generation of harmful products such as advanced glycation end products (AGEs) and reactive oxygen species (ROS), inducing oxidative stress and inflammation, which lead to endothelial dysfunction and vascular wall inflammation, thereby promoting plaque formation and increasing the risk of stroke [39–41]. Concurrently, dyslipidemia is also a key driver of atherosclerosis: through pathways including endothelial dysfunction, foam cell formation and inflammatory responses caused by oxidized LDL, as well as platelet activation and enhanced coagulation, it promotes plaque progression and induces plaque rupture/erosion, thereby increasing the risk of stroke [42, 43]. Therefore, as a composite indicator integrating blood glucose and lipid levels, CHG to some extent reflects the pathophysiological processes of stroke occurrence. It is noteworthy that at lower levels of CHG, the glucose and lipid metabolism in the body are relatively balanced, and metabolic abnormalities have a limited impact on the blood vessels. However, as CHG levels increase, especially when exceeding the threshold (4.556), insulin resistance and inflammatory states may rapidly worsen, leading to the accumulation of ROS, AGEs, and other substances in the body to a certain threshold. This accumulation can further lead to endothelial dysfunction, significantly increasing the risk of stroke. Additionally, persistently elevated blood glucose and lipid levels may trigger the proliferation and migration of vascular smooth muscle cells, further exacerbating atherosclerosis. Furthermore, research indicates that females may exhibit different physiological responses in lipid and glucose metabolism [44]. For example, the estrogen in women’s bodies serves a protective role in lipid metabolism, which may lower the risk of atherosclerosis and related strokes [45]. At the same time, there are differences in body fat distribution between men and women [46]. Although our study did not observe interactions between sexes, this may be due to limitations in sample size, specific CHG thresholds, or other physiological factors that constrain the impact of sex on the relationship between CHG and stroke risk. Future research could further explore how sex differences may affect stroke risk under different environmental and physiological conditions.
This study has the following strengths: (i) To our understanding, this investigation represents the inaugural study exploring CHG’s relationship with stroke risk. Furthermore, CHG underwent dual analytical approaches—continuous and categorical—minimizing data loss while facilitating more thorough and precise evaluation of its association with stroke risk. (ii) We found a nonlinear relationship between CHG and stroke incidence, which represents a significant advancement. (iii) We used multiple imputations to handle missing data, thereby increasing statistical power and reducing bias due to missing covariate information. (iv) Multiple sensitivity analyses were performed for result validation, comprising CHG transformation into categorical variable, incorporating continuous covariates as curves in a GAM, computing E-values for assessing influence from unknown and unmeasured confounders, and re-evaluating the association between CHG and stroke risk among participants with BMI < 28 kg/m², never drinkers, and those without DM, hypertension, or CKD.
This study has several noteworthy limitations. First, the study population was limited to middle‑aged and older adults in China, which constrains the generalizability of the findings to younger populations and to groups differing by ethnicity or geographic region. Therefore, further validation is needed in more heterogeneous populations. Furthermore, because this was a observational study, many variables were not included, and adjustment for potential or unmeasured confounding factors such as dietary habits, medication use, and family history of stroke was limited. However, the calculated E-values suggest that unknown or unmeasured confounding is unlikely to have a substantial impact on the association between CHG and the risk of stroke. Future studies should incorporate more relevant variables (e.g., lifestyle, medication use, and dietary habits) to enable a more comprehensive analysis of the relationship between CHG and the risk of stroke and to further validate our findings. Finally, as an observational study, although an independent association between CHG and stroke risk was observed, causality cannot be established.
Conclusion
The study found an independent positive association between higher CHG levels and increased stroke risk among middle‑aged and older Chinese adults. Additionally, a threshold effect curve was observed between them, with an inflection point at CHG = 4.556. Additionally, this study found that CHG has relatively superior predictive value for stroke risk compared to its individual components. This indicator is expected to become a marker for identifying populations at high risk of early stroke, providing new ideas for optimizing risk stratification and preventive measures.
Supplementary Information
Acknowledgements
We would like to express our appreciation to the researchers and staff involved in the China Health and Retirement Longitudinal Study (CHARLS), as well as the study’s participants. Their contributions have been essential in providing the data and methodological framework that underpin our study.
Abbreviations
- CHG
Total Cholesterol, High-Density Lipoprotein, and Glucose
- WBC
White blood cell
- PLT
Platelet
- BUN
Blood urea nitrogen
- FPG
Fasting plasma glucose
- Scr
Serum creatinine
- TC
Total cholesterol
- TG
Triglyceride
- HDL-c
High-Density Lipoprotein Cholesterol
- LDL-c
Low-Density Lipoprotein Cholesterol
- CRP
C-Reactive protein
- HBA1c
Hemoglobin A1c
- UA
Uric acid
- HGB
Hemoglobin
- SBP
Systolic blood pressure
- DBP
Diastolic blood pressure
- BMI
Body mass index
- WC
Waist circumference
- DM
Diabetes mellitus
- CLD
Chronic lung disease
- CHD
Coronary heart disease
- CKD
Chronic kidney disease
- SD
Standard deviation
- CI
Confidence
- n
Number
- HR
Hazard ratio
- ROC
Receiver Operating Characteristic
- TC/HDL-c
Total Cholesterol to High-Density Lipoprotein Cholesterol Ratio
- TyG
Triglyceride-Glucose Index
Authors’ contributions
Min Wang was responsible for running statistical analyses and preparing the initial draft. Min Wang and Jiaqian Zhu participated in critical revision of the manuscript and contributed to the conceptual framework. All contributors have reviewed and approved the final version.
Funding
This study did not receive any funding.
Data availability
The data for this study can be accessed online at http://www.isss.pku.edu.cn/cfps/or https://cfpsdata.pku.edu.cn/#/home. To obtain the data, you will need to register as a user on the website. After your registration is reviewed and approved, you can follow the instructions provided to download the dataset.
Declarations
Ethics approval and consent to participate
This study was ethically conducted and approved by the Biomedical Ethics Review Committee of Peking University in accordance with the principles of the Helsinki Declaration. Moreover, all participants provided written informed consent to take part in the study (IRB approval number IRB00001052–11015).
Consent for publication
Not applicable.
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.
References
- 1.Virani SS, Alonso A, Aparicio HJ, Benjamin EJ, Bittencourt MS, Callaway CW, et al. Heart disease and stroke statistics-2021 update: a report from the American Heart Association. Circulation. 2021;143(8):e254–743. [DOI] [PubMed] [Google Scholar]
- 2.Hilkens NA, Casolla B, Leung TW, de Leeuw F. Stroke LANCET. 2024;403(10446):2820–36. [DOI] [PubMed] [Google Scholar]
- 3.Lloyd-Jones D, Adams R, Carnethon M, De Simone G, Ferguson TB, Flegal K, Ford E, Furie K, Go A, Greenlund K, et al. Heart disease and stroke statistics–2009 update: a report from the American heart association statistics committee and stroke statistics subcommittee. Circulation. 2009;119(3):e21–181. [DOI] [PubMed] [Google Scholar]
- 4.Ishak B, Abdul-Jabbar A, Singla A, Yilmaz E, von Glinski A, Ramey WL, et al. Intraoperative ischemic stroke in elective spine surgery: a retrospective study of incidence and risk. Spine. 2020;45(2):109–15. [DOI] [PubMed] [Google Scholar]
- 5.Feigin VL, Nguyen G, Cercy K, Johnson CO, Alam T, Parmar PG, Abajobir AA, Abate KH, Abd-Allah F, Abejie AN, et al. Global, Regional, and Country-Specific lifetime risks of Stroke, 1990 and 2016. NEW ENGL J MED. 2018;379(25):2429–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.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(10258):1204–1222. [DOI] [PMC free article] [PubMed]
- 7.Strong K, Mathers C, Bonita R. Preventing stroke: saving lives around the world. Lancet Neurol. 2007;6(2):182–7. [DOI] [PubMed] [Google Scholar]
- 8.Olamoyegun MA, Akinlade AT, Fawale MB, Ogbera AO. Dyslipidaemia as a risk factor in the occurrence of stroke in nigeria: prevalence and patterns. PAN AFR MED J. 2016;25:72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zhang H, Kwapong WR, Shao M, Yan J, Lin X, Chen B, Chen K. Predictors of the prevalence of dyslipidemia and influencing factors for young health examination cohort: A Cross-Sectional survey. FRONT PUBLIC HEALTH. 2020;8:400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Djelilovic-Vranic J, Alajbegovic A, Zelija-Asimi V, Niksic M, Tiric-Campara M, Salcic S, Celo A. Predilection role diabetes mellitus and dyslipidemia in the onset of ischemic stroke. Med Arch. 2013;67(2):120–3. [DOI] [PubMed] [Google Scholar]
- 11.Xiao B, Cao C, Han Y, Yang F, Hu H, Luo J. A non-linear connection between the total cholesterol to high-density lipoprotein cholesterol ratio and stroke risk: a retrospective cohort study from the China health and retirement longitudinal study. Eur J Med Res. 2024;29(1):175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Elshazly MB, Nicholls SJ, Nissen SE, St John J, Martin SS, Jones SR, Quispe R, Stegman B, Kapadia SR, Tuzcu EM, et al. Implications of total to High-Density lipoprotein cholesterol ratio discordance with alternative lipid parameters for coronary atheroma progression and cardiovascular events. AM J CARDIOL. 2016;118(5):647–55. [DOI] [PubMed] [Google Scholar]
- 13.Kurth T, Everett BM, Buring JE, Kase CS, Ridker PM, Gaziano JM. Lipid levels and the risk of ischemic stroke in women. Neurology. 2007;68(8):556–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kaynak N, Kennel V, Rackoll T, Schulze D, Endres M, Nave AH. Impaired glucose metabolism and the risk of vascular events and mortality after ischemic stroke: A systematic review and meta-analysis. CARDIOVASC DIABETOL. 2024;23(1):323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Richter B, Hemmingsen B, Metzendorf M, Takwoingi Y. Development of type 2 diabetes mellitus in people with intermediate hyperglycaemia. COCHRANE DB SYST REV. 2018;10(10):CD12661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.McCracken E, Monaghan M, Sreenivasan S. Pathophysiology of the metabolic syndrome. Clin Dermatol. 2018;36(1):14–20. [DOI] [PubMed] [Google Scholar]
- 17.Shi H, Ge Y, Wang H, Zhang Y, Teng W, Tian L. Fasting blood glucose and risk of stroke: a dose-response meta-analysis. Clin Nutr. 2021;40(5):3296–304. [DOI] [PubMed] [Google Scholar]
- 18.Mansoori A, Nosrati M, Dorchin M, Mohammadyari F, Derakhshan-Nezhad E, Ferns G, et al. A novel index for diagnosis of type 2 diabetes mellitus: cholesterol, high density lipoprotein, and glucose (CHG) index. J Diabetes Invest. 2025;16(2):309–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mo D, Zhang P, Zhang M, Dai H, Guan J. Cholesterol, high-density lipoprotein, and glucose index versus triglyceride-glucose index in predicting cardiovascular disease risk: a cohort study. CARDIOVASC DIABETOL. 2025;24(1):116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Çatak M, Konuk ^G, Hepsen S. The cholesterol-HDL-glucose (CHG) index and traditional adiposity markers in predicting diabetic retinopathy and nephropathy. J DIABETES INVEST. 2025;16(8):1487–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.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(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wei G, Lin F, Cao C, Hu H, Han Y. Non-linear dose-response relationship between body mass index and stroke risk in middle-aged and elderly Chinese men: a nationwide longitudinal cohort study from CHARLS. FRONT ENDOCRINOL. 2023;14:1203896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chen X, Liu S, Chu J, Hu W, Sun N, Shen Y. Joint effect of elevated-c-reactive protein level and hypertension on new-onset stroke: A nationwide prospective cohort study of CHARLS. FRONT PUBLIC HEALTH. 2022;10:919506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ramdas Nayak VK, Satheesh P, Shenoy MT, Kalra S. Triglyceride glucose (TyG) index: a surrogate biomarker of insulin resistance. J Pak Med Assoc. 2022;72(5):986–8. [DOI] [PubMed] [Google Scholar]
- 25.White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377–99. [DOI] [PubMed] [Google Scholar]
- 26.Groenwold RHH, White IR, Donders ART, Carpenter JR, Altman DG, Moons KGM. Missing covariate data in clinical research: when and when not to use the missing-indicator method for analysis. Can Med Assoc J. 2012;184(11):1265–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Gusev E, Sarapultsev A. Atherosclerosis and inflammation: insights from the theory of general pathological processes. Int J Mol Sci. 2023;24(9):7910. [DOI] [PMC free article] [PubMed]
- 28.Elkind MSV. Inflammatory markers and stroke. Curr Cardiol Rep. 2009;11(1):12–20. [DOI] [PubMed] [Google Scholar]
- 29.Raggi P, Genest J, Giles JT, Rayner KJ, Dwivedi G, Beanlands RS, et al. Role of inflammation in the pathogenesis of atherosclerosis and therapeutic interventions. Atherosclerosis. 2018;276:98–108. [DOI] [PubMed] [Google Scholar]
- 30.Stark K, Massberg S. Interplay between inflammation and thrombosis in cardiovascular pathology. Nat Rev Cardiol. 2021;18(9):666–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mosenzon O, Cheng AY, Rabinstein AA, Sacco S. Diabetes and stroke: what are the connections? J Stroke. 2023;25(1):26–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Masson P, Webster AC, Hong M, Turner R, Lindley RI, Craig JC. Chronic kidney disease and the risk of stroke: a systematic review and meta-analysis. Nephrol Dial Transplant. 2015;30(7):1162–9. [DOI] [PubMed] [Google Scholar]
- 33.Wang X, Huang Y, Chen Y, Yang T, Su W, Chen X, et al. The relationship between body mass index and stroke: a systemic review and meta-analysis. J Neurol. 2022;269(12):6279–89. [DOI] [PubMed] [Google Scholar]
- 34.Buonacera A, Stancanelli B, Malatino L. Stroke and hypertension: an appraisal from pathophysiology to clinical practice. Curr Vasc Pharmacol. 2019;17(1):72–84. [DOI] [PubMed] [Google Scholar]
- 35.Zhang C, Qin Y, Chen Q, Jiang H, Chen X, Xu C, et al. Alcohol intake and risk of stroke: a dose-response meta-analysis of prospective studies. Int J Cardiol. 2014;174(3):669–77. [DOI] [PubMed] [Google Scholar]
- 36.Zhou B. Predictive values of body mass index and waist circumference for risk factors of certain related diseases in Chinese adults–study on optimal cut-off points of body mass index and waist circumference in Chinese adults. Biomed Environ Sci. 2002;15(1):83–96. [PubMed] [Google Scholar]
- 37.Haneuse S, VanderWeele TJ, Arterburn D. Using the e-value to assess the potential effect of unmeasured confounding in observational studies. JAMA J Am Med Assoc. 2019;321(6):602–3. [DOI] [PubMed] [Google Scholar]
- 38.Program NHBP. The seventh report of the joint National committee on Prevention, Detection, Evaluation, and treatment of high blood pressure. Bethesda (MD): National Heart, Lung, and Blood Institute (US); 2004. [PubMed] [Google Scholar]
- 39.Brownlee M. The pathobiology of diabetic complications: a unifying mechanism. Diabetes. 2005;54(6):1615–25. [DOI] [PubMed] [Google Scholar]
- 40.Luitse MJA, Biessels GJ, Rutten GEHM, Kappelle LJ. Diabetes, hyperglycaemia, and acute ischaemic stroke. Lancet Neurol. 2012;11(3):261–71. [DOI] [PubMed] [Google Scholar]
- 41.Koutsaliaris IK, Moschonas IC, Pechlivani LM, Tsouka AN, Tselepis AD. Inflammation, oxidative stress, vascular aging and atherosclerotic ischemic stroke. Curr Med Chem. 2022;29(34):5496–509. [DOI] [PubMed] [Google Scholar]
- 42.Libby P, Buring JE, Badimon L, Hansson GK, Deanfield J, Bittencourt MS, Tokgözoğlu L, Lewis EF. Atherosclerosis. NAT REV DIS PRIMERS. 2019;5(1):56. [DOI] [PubMed] [Google Scholar]
- 43.Davignon J, Ganz P. Role of endothelial dysfunction in atherosclerosis. Circulation. 2004;109(23 Suppl 1):III27–32. [DOI] [PubMed] [Google Scholar]
- 44.Varlamov O, Bethea CL, Roberts CTJ. Sex-specific differences in lipid and glucose metabolism. Front Endocrinol. 2014;5:241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Xie F, Li X, Xu Y, Cheng D, Xia X, Lv X, Yuan G, Peng C. Estrogen mediates an Atherosclerotic-Protective action via Estrogen receptor Alpha/SREBP-1 signaling. FRONT CARDIOVASC MED. 2022;9:895916. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Blaak E. Gender differences in fat metabolism. Curr Opin Clin Nutr. 2001;4(6):499–502. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data for this study can be accessed online at http://www.isss.pku.edu.cn/cfps/or https://cfpsdata.pku.edu.cn/#/home. To obtain the data, you will need to register as a user on the website. After your registration is reviewed and approved, you can follow the instructions provided to download the dataset.