Associations of the combined C-reactive protein-triglyceride glucose index and frailty index in different dimensions with incident stroke among individuals with stages 0–3 cardiovascular-kidney-metabolic syndrome Authors: Honglin Zheng, Yao Zhang, Li Li, Mingye Zhu, Wei Kong, Jialin Wang, Ruxi Liu, Fang Wang Published: 2026 ABSTRACT BackgroundPrevious studies have revealed the relationships of separated C-reactive protein-triglyceride glucose (CTI) levels and the frailty index (FI) with stroke. However, the impact of combined CTI and FI (CTI-FI) on stroke incidence is unclear, especially among those with cardiovascular kidney metabolic (CKM) syndrome stages 0–3.ObjectiveThis research aimed to validate the associations between different CTI-FI dimensions (baseline CTI-FI, cumulative CTI-FI (cuCTI-FI), and dynamic trajectories of CTI-FI (traCTI-FI)) and stroke risk among individuals with CKM syndrome stages 0–3.MethodsThe enrolled participants and the utilized data were derived from five waves of the China Health and Retirement Longitudinal Study. K-means clustering was used to categorize participants into an appropriate number of clusters. Cox regression analysis, restricted cubic spline (RCS) curves, and Kaplan-Meier (K-M) survival curves were used to evaluate the relationships between different CTI-FI dimensions and stroke risk. Receiver operating characteristic (ROC) curves and the DeLong test were constructed to assess the performance of various dimensions of CTI-FI in predicting stroke.ResultsIn this study, the mean age of the 5293 participants was 57.94 years, and 53.45% were female. During the nearly 9-year follow-up period, 540 (10.20%) stroke events occurred. A 61% and 41% increase in stroke risk was associated with each 1-unit increase in the baseline CTI-FI and cuCTI-FI, respectively. The RCS modeling further revealed significant positive nonlinear associations between baseline CTI-FI (Poverall < 0.001, Pnonlinear = 0.049) and cuCTI (Poverall < 0.001, Pnonlinear = 0.047) and stroke incidence. Compared with the lowest different dimensions CTI-FI level groups, the highest level groups had a greater stroke risk. The fully adjusted HRs (95% CIs) were as follows: baseline CTI-FI (Q4 vs. Q1), 2.36 (1.76, 3.15); cuCTI-FI (Q3 vs. Q1), 4.75 (2.73, 8.27); and traCTI-FI (Cluster 3 vs. Cluster 1), 6.24 (3.72, 10.46). In terms of predicting stroke risk, baseline CTI-FI, cuCTI-FI, and traCTI-FI performed better than CTI and FI, and cuCTI-FI and traCTI-FI performed significantly better than baseline CTI-FI (P < 0.001 and = 0.022, respectively).ConclusionPersistently high CTI-FI is associated with increased stroke risk. CTI-FI, especially cuCTI-FI and traCTI-FI, are potent predictors of stroke. Long-term surveillance of CTI-FI alterations and maintenance of its low levels is clinically important for early stroke detection and prevention in patients with stages 0–3 CKM syndrome.Graphical abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12933-026-03199-9. Research insights What is currently known about this topic? Existing literature has shown that CTI and FI are separately associated with increased stroke risk. They have a bidirectional reinforcing relationship in CKM syndrome progression. However, research on the combined influence of CTI and FI on stroke incidence among patients with CKM syndrome stages 0–3 is limited. What is the key research question? What is the association and predictive value of baseline CTI-FI, cumulative CTI-FI (cuCTI-FI), and dynamic trajectories of CTI-FI (traCTI-FI) with stroke incidence among patients with CKM syndrome stages 0–3? What is new? This study is the first to propose a novel composite index (CTI-FI). It is also the first large-scale cohort study to investigate the relationship between baseline CTI-FI and stroke and to further explore the correlation between changes in CTI-FI (including cuCTI-FI and traCTI-FI) and stroke, as well as their predictive value. How might this study influence clinical practice? CTI-FI provides a more comprehensive risk-stratification tool by integrating inflammatory, metabolic, and biological aging factors. The positive relationship and potential predictive value of CTI-FI with stroke hold important clinical implications for the early prevention and identification of stroke events in populations with CKM stages 0–3. Introduction Stroke, widely recognized as a significant chronic noncommunicable disease that threatens global human health [1], is among the second leading causes of mortality worldwide [2] and is the primary etiology of mortality and disability in China [3]. The incidence of stroke tends to steadily increase with age [4], and early onset has emerged as a concerning feature in recent years [5]. Cardiovascular-kidney-metabolic (CKM) syndrome is a systemically defined disorder initially proposed in October 2023 by the American Heart Association. This condition encompasses the interplay of cardiovascular system dynamics, chronic kidney disease, and metabolic risk factors, along with continuous progression [6]. CKM syndrome is categorized into stages ranging from 0 to 4, wherein stroke is designated as a clinical event in CKM stage 4 [6]. Studies have shown that individuals in CKM stages 0–3 have a higher risk of stroke. This situation is attributed to multiple comorbid conditions, including inflammation, metabolic dysfunction, biological aging, and reduced organ reserve capacity [7, 8]. Additionally, the American Heart Association emphasizes the importance of addressing cardiovascular-related diseases in patients with stages 0–3 CKM syndrome [9]. Consequently, there is an urgent need to explore more comprehensive and accurate risk assessment indicators to improve the early identification of individuals at high risk of stroke among patients with stages 0–3 CKM syndrome. Timely interventions for these patients can effectively prevent or delay disease progression and reduce the clinical burden. The C-reactive protein-triglyceride glucose index (CTI) is a composite biomarker that integrates inflammatory markers and metabolic biomarkers. Such an index can quantify both chronic inflammatory status and insulin resistance in the body [10]. The frailty index (FI) accurately reflects an individual’s biological aging profile, highlighting the decrease in physiological reserve capacity through the cumulative assessment of health deficits across multiple dimensions [11]. The relationship between CTI and FI is bidirectionally reinforced during CKM syndrome progression. Specifically, the chronic inflammatory state and insulin resistance exhaust the body’s physiological reserves, driving the onset and progression of frailty. In turn, frailty-related problems, such as reduced physical activity and nutritional metabolic dysregulation, worsen inflammatory reactions and metabolic disruptions [12, 13]. Both CTI and FI are implicated in the risk of new-onset stroke in stages 0–3 of CKM syndrome [11, 14] despite the intricate interactions among inflammatory responses, metabolic dysregulation, and biological aging, which frequently cooccur in the elderly population and exert synergistic pathogenic effects on the progression of CKM syndrome [15, 16]. However, existing research has predominantly focused on the independent use of the CTI or FI for stroke risk assessment and has not addressed their combined effects [17, 18]. Therefore, this study proposes a novel composite index (CTI-FI). This index synergistically integrates three core health risks in the elderly population: inflammation, metabolic dysregulation, and biological aging. It is intended to offer a more comprehensive and accurate risk evaluation framework for stroke prevention in populations at stages 0–3 of CKM syndrome. In this study, cohort data from the China Health and Retirement Longitudinal Study (CHARLS) were used to assess the association between CTI-FI and the incidence of new-onset stroke among individuals in stages 0–3 of CKM syndrome and to evaluate its predictive efficacy. First, the relationship between the baseline CTI-FI and the incidence of new-onset stroke in individuals with stages 0–3 CKM syndrome was examined. Moreover, to illustrate the long-term cumulative effects and dynamic evolution of inflammatory, metabolic, and biological aging burdens on stroke risk, this study further explored the relationships between the cumulative CTI-FI (cuCTI-FI) and the dynamic trajectory in CTI-FI (traCTI-FI) and stroke risk in individuals with CKM syndrome stages 0–3. Ultimately, the study also assessed the effectiveness of CTI-FI in different dimensions as a tool for predicting stroke risk in individuals with stages 0–3 CKM syndrome. The findings of this study present a more comprehensive and precise assessment tool for the early identification of individuals at high risk of stroke, offer a novel perspective on stroke prevention and treatment management, and contribute to reducing stroke incidence among middle-aged and elderly individuals at stages 0–3 of CKM syndrome. Methods Study design and participants Participants were enrolled from CHARLS, a nationally prospective cohort study tracking health status changes among individuals aged 45 and above. In CHARLS, the first survey (wave 1) was carried out from June 2011 to March 2012. Four follow-up surveys, wave 2, wave 3, wave 4, and wave 5, were subsequently conducted in 2013, 2015, 2018, and 2020. Sociodemographic, lifestyle and health-related information, as well as physiological and biochemical indicators, were collected in the CHARLS by trained staff through a combination of face-to-face questionnaires and standardized biomedical tests. Blood samples were collected at wave 1 and wave 3 for some participants, and biochemical indicators were detected by blood tests. Please visit the website (http://charls.pku.edu.cn/en) for more details about the CHARLS cohort. The exclusion flowchart for the study is shown in Fig. 1. A total of 17708 participants from wave 1 of the CHARLS were used as the baseline, and waves 2, 3, 4, and 5 of the CHARLS were selected as the follow-up for this study. We first excluded participants who were under 45 years old (n = 368), who were missing data for baseline CKM stage classification (n = 7958), who were identified as CKM stage 4 at baseline (n = 1171), and those with incomplete baseline information (n = 920). Next, we excluded participants who were missing data for defining stroke or who were lost to follow-up (n = 1998). Finally, 5293 eligible participants were included in the baseline CTI-FI analyses. Afterward, participants who had missing data for calculating CTI-FI at wave 3 (n = 3907) and who were diagnosed with stroke at wave 2 (n = 22) were further excluded. Eventually, 1364 participants were included in the changes in CTI-FI (including cuCTI-FI and traCTI-FI) analyses. Fig. 1Exclusion flowchart for the study design Exclusion flowchart for the study design Data collection The data required in this study were obtained from the five waves of the CHARLS database. The data collected for this study included sociodemographic, lifestyle and health-related information, including basic information such as age, sex, marital status, residence, education, smoking status, alcohol consumption status, and body mass index (BMI). It also included a history of comorbidities such as hypertension, diabetes, and dyslipidemia, and these medical histories. The physiological indicators, including diastolic blood pressure (DBP) and systolic blood pressure (SBP), etc. Biochemical indicators determined by blood tests, including total cholesterol (TC), triglycerides (TG), high‑density lipoprotein cholesterol (HDL‑C), low‑density lipoprotein cholesterol (LDL‑C), glycated hemoglobin (HbA1c), and fasting plasma glucose (FPG), etc. Definitions of the composite index (CTI-FI) The C-reactive protein-triglyceride glucose index (CTI) in this study was calculated according to a previous publication [10] using the following formula: CTI = 0.412×Ln (CRP [mg/L]) + Ln (TG [mg/dL]×FPG [mg/dL])/2. The frailty index (FI) was constructed using the health deficit accumulation model, which is applicable to the CHARLS dataset and reported in the literature [19, 20]. We excluded cardiovascular disease-related items (such as heart failure and stroke). Thirty health-related variables were included on the basis of previous studies, including the core domains of physical functioning, chronic conditions, cognitive performance, and depressive symptoms (Table S1). Cognitive function was assessed using a 0–1 ordinal scale, with lower scores indicating better cognitive function. Other variables were dichotomized and labeled 0 (nondeficient) or 1 (deficient). We first calculated the sum of current health deficits for each subject and then divided it by 30 to determine the FI. The new combined indicator of CTI and FI (CTI-FI) was obtained using a multiplicative model, which has been widely applied in epidemiological analyses [21, 22], as follows: CTI-FI = CTI × FI. This model can capture potential interactions among metabolic, inflammatory, and functional deficits through the product term. Assessment of the three dimensions of the CTI-FI This study included three CTI-FI dimensions: baseline CTI-FI, cumulative CTI-FI (cuCTI-FI), and dynamic trajectories of CTI-FI (traCTI-FI). The cuCTI-FI showed the long-term cumulative state of metabolic, inflammatory, and functional deficits and was calculated using a linear model by the following formula [18, 23]: cuCTI-FI = (CTI-FI2012 + CTI-FI2015)/2 × time period. In this formula, the time period was the interval duration (in years), which was 3 years in this study. The traCTI-FI reflects the long-term evolutionary characteristics of the CTI-FI. On the basis of the CTI-FI values in 2012 and 2015, the participants were divided into three trajectory clusters according to their overall level and temporal trend using the K-means clustering algorithm. Cluster 1 (persistent low-level of CTI-FI): Both the initial and terminal values of CTI-FI remained at low levels, ranging from 0.53 in 2012 to 0.73 in 2015, with a slight elevation trend; Cluster 2 (persistent moderate-level of CTI-FI): Both the initial and terminal values of CTI-FI were at moderate levels, with a slight increase from 1.10 in 2012 to 1.40 in 2015; and Cluster 3 (stable high-level of CTI-FI): Both the initial and terminal values of CTI-FI were at high levels, with an obvious increase from 1.56 in 2012 to 2.65 in 2015 (Fig. 2). Fig. 2A K values based on the elbow method. B Three clusters determined by the K-means method. C Three trajectories of CTI-FI CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index. Cluster 1: persistent low-level of CTI-FI (marked in green); Cluster 2: persistent moderate-level of CTI-FI (marked in blue); and Cluster 3: stable high-level of CTI-FI (marked in red). A K values based on the elbow method. B Three clusters determined by the K-means method. C Three trajectories of CTI-FI CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index. Cluster 1: persistent low-level of CTI-FI (marked in green); Cluster 2: persistent moderate-level of CTI-FI (marked in blue); and Cluster 3: stable high-level of CTI-FI (marked in red). Classification of CKM syndrome stages 0 to 4 In this study, the stages 0–4 of CKM syndrome were classified according to the presidential advisory statement published by the American Heart Association (AHA) [6] and previous studies based on the CHARLS [24–26]. No CKM risk factors were classified into Stage 0. Dysfunctional adiposity or overweight was classified into Stage 1. Appearance of metabolic disorders or moderate- to high-risk CKD was classified into Stage 2. Presence of subclinical cardiovascular disease (CVD) in CKM or its risk equivalents was classified into Stage 3. Due to the lack of indicators for classifying subclinical CVD in the CHARLS, risk equivalence criteria were used for classification. Existing clinical cardiovascular diseases in CKM were classified into Stage 4. Detailed classification criteria of CKM stages 0–4 were provided in Appendix A. Assessment of outcomes The primary outcome of this study was new-onset stroke in patients with CKM syndrome stages 0–3. In the CHARLS, as in previous studies [27, 28], participants who self-reported stroke during follow-up were determined as new-onset stroke events when they were asked, “Have you been told by a doctor that you have been diagnosed with a stroke?” Statistical analysis In this study, the basic characteristics are presented as the means ± standard deviations (SDs) and medians (interquartile ranges) or as numbers (percentages). T tests, the Mann-Whitney U test, and chi-square tests were used to identify differences between continuous or categorical variables. Participants enrolled in the baseline CTI-FI analyses were divided into four groups (Q1-Q4) on the basis of the quartiles of continuous baseline CTI-FI. Additionally, in the changes in CTI-FI analyses, participants were divided into three groups (Q1-Q3) on the basis of the tertiles of continuous cuCTI-FI. These participants were divided into 3 clusters (persistent low-level, persistent moderate-level, and stable high-level clusters) using the K-means clustering algorithm. As a classic unsupervised clustering method, K-means clustering partitions samples into groups with high intracluster homogeneity and intercluster heterogeneity by minimizing the within-cluster sum of squares (WCSS). The method identifies the point at which the reduction in total squared distance begins to slow by plotting the WCSS curve against the number of clusters. The optimal number of clusters was three in this study, for which the elbow method was adopted (Fig. 2). The relationships between baseline CTI-FI, cuCTI-FI, and traCTI-FI and stroke incidence were assessed using hazard ratios (HRs) and 95% confidence intervals (CIs), which were calculated using Cox proportional hazards (CoxPH) models. Three models were considered for each Cox regression analysis: Model 1 (unadjusted), Model 2 (multivariable adjusted), and Model 3 (fully adjusted). Detailed information on the adjusted confounding factors is provided in Appendix A. Restricted cubic spline (RCS) regression was used to explore the potential linear or nonlinear dose-response relationships of baseline CTI-FI and cuCTI-FI with stroke risk. Kaplan-Meier (K-M) survival curves were used to assess the relationships between the three CTI-FI dimensions and stroke risk. The receiver operating characteristic (ROC) analysis and the DeLong test were used to evaluate the predictive ability of three CTI-FI dimensions for stroke. Moreover, interactive and stratified analyses were conducted according to age, sex, marital status, place of residence, educational attainment, smoking status and alcohol consumption status, as well as a history of cardiometabolic conditions such as hypertension, diabetes mellitus, and dyslipidemia, to examine whether these factors modified the associations between different dimensions of the CTI-FI and stroke risk. Seven types of sensitivity analyses were performed to verify the robustness of the primary findings in our study: (1) Participants who had baseline CTI-FI values beyond 3 standard deviations were excluded. (2) We converted the form of some variables, which include education status, smoking status, drinking status, and BMI categories (see Appendix A for detailed information). (3) We changed the follow-up endpoint from 2020 to 2018 to minimize potential bias from the COVID-19 pandemic and test the robustness of our findings across different follow-up durations. (4) Participants were divided into 3 groups based on the baseline CTI-FI tertiles. (5) Participants were divided into 4 groups based on the cuCTI-FI quartiles. (6) Participants were clustered into 4 clusters. (7) We used ROC analysis and the DeLong test based on a Random Survival Forest (RSF) [29] model to identify the robustness of CTI-FI’s predictive capacity for stroke (Detailed information and parameters of RSF were provided in Appendix A). The statistical software R (version 4.4.1) were used for the statistical analyses, and P-values < 0.05 (two-sided) were considered statistically significant in this study. Results Baseline characteristics of the participants In this study, in accordance with specific inclusion and exclusion criteria, 5293 participants (mean age: 57.94 years) who did not have stroke and were in stages 0–3 of CKM syndrome at baseline were enrolled for baseline CTI-FI analyses, 2829 (53.45%) of whom were females. The detailed baseline characteristics of these participants stratified by CTI-FI quartiles (Q1-Q4) are summarized in Table 1. Compared with individuals in the lower CTI-FI quartile group, individuals in the higher quartile group were significantly more likely to be older, female, and urban residents and have lower educational levels, with a greater proportion of never-smokers and never-drinkers. They had a significantly higher prevalence of hypertension, diabetes, and dyslipidemia. With respect to physical measurements and laboratory parameters, higher CTI-FI quartiles were associated with higher levels of BMI, SBP, TC, TG, LDL-C, and HbA1c but with lower HDL-C levels. The mean baseline CTI-FI value across all participants was 0.75. A comparison between the Q1 and Q4 groups revealed an increase in the mean CTI-FI value from 0.23 to 1.44. Table 1Baseline characteristics of participants according to baseline CTI-FI quartilesCharacteristicsOverall(n = 5293)Q1(n = 1324)Q2(n = 1323)Q3(n = 1323)Q4(n = 1323)P-valueaAge (years)57.94 ± 8.4056.08 ± 8.1357.42 ± 8.2457.95 ± 8.4160.32 ± 8.28< 0.001Sex, n (%)< 0.001Male2464 (46.55)723 (54.61)658 (49.74)607 (45.88)476 (35.98)Female2829 (53.45)601 (45.39)665 (50.26)716 (54.12)847 (64.0)Marital status, n (%)< 0.001Unmarried481 (9.09)104 (7.85)104 (7.86)106 (8.01)167 (12.62)Married4812 (90.91)1220 (92.15)1219 (92.14)1217 (91.99)1156 (87.38)Residence, n (%)< 0.001Urban3632 (68.62)860 (64.95)888 (67.12)901 (68.10)983 (74.30)Rural1661 (31.38)464 (35.05)435 (32.88)422 (31.90)340 (25.70)Education, n (%)< 0.001< Middle school3645 (68.86)806 (60.88)851 (64.32)914 (69.09)1074 (81.18)≥ Middle school1648 (31.14)518 (39.12)472 (35.68)409 (30.91)249 (18.82)Smoking status, n (%)< 0.001Non-smoker3248 (61.36)752 (56.80)788 (59.56)817 (61.75)891 (67.35)Current smoker2045 (38.64)572 (43.20)535 (40.44)506 (38.24)432 (32.65)Drinking status, n (%)< 0.001Non-drinker1825 (34.48)540 (40.79)485 (36.66)443 (33.48)357 (26.98)Current drinker3468 (65.52)784 (59.21)838 (63.34)880 (66.52)966 (73.02)Hypertension, n (%)< 0.001Yes1038 (19.61)84 (6.34)182 (13.76)316 (23.89)456 (34.47)No4255 (80.38)1240 (93.66)1141 (86.24)1007 (76.11)867 (65.53)Diabetes, n (%)< 0.001Yes226 (4.27)6 (0.45)22 (1.66)64 (4.84)134 (10.12)No5067 (95.73)1318 (99.55)1301 (98.34)1259 (95.16)1189 (89.87)Dyslipidemia, n (%)< 0.001Yes383 (7.24)57 (4.31)76 (5.74)92 (6.95)158 (11.94)No4910 (92.76)1267 (95.69)1247 (94.26)1231 (93.05)1165 (88.06)BMI (kg/m2)23.42 ± 3.8023.06 ± 3.2623.31 ± 3.6123.53 ± 3.9623.80 ± 4.27< 0.001SBP (mm Hg)128.77 ± 20.77127.18 ± 19.33127.02 ± 20.30129.78 ± 21.43131.12 ± 21.68< 0.001DBP (mm Hg)75.23 ± 11.9875.29 ± 11.7974.54 ± 11.8975.66 ± 12.3375.43 ± 11.890.034TC (mg/dL)193.55 ± 37.77191.35 ± 38.26191.83 ± 37.53193.88 ± 37.55197.14 ± 37.51< 0.001HDL (mg/dL)51.83 ± 15.3552.30 ± 15.1052.90 ± 15.5951.14 ± 15.1050.96 ± 15.52< 0.001LDL (mg/dL)116.38 ± 34.50115.65 ± 35.35115.11 ± 33.01116.92 ± 34.93117.84 ± 34.620.192HbA1c5.25 ± 0.775.17 ± 0.685.17 ± 0.765.26 ± 0.765.39 ± 0.95< 0.001CTI-FI0.75 ± 0.510.23 ± 0.100.52 ± 0.070.80 ± 0.101.44 ± 0.45< 0.001SD: standard deviation; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; HbA1c: glycated haemoglobin; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index. The values are the means ± SDs or n (percentages). a Continuous and categorical variables were compared using analyses of t-tests and chi-square tests, respectively Table 1Baseline characteristics of participants according to baseline CTI-FI quartilesCharacteristicsOverall(n = 5293)Q1(n = 1324)Q2(n = 1323)Q3(n = 1323)Q4(n = 1323)P-valueaAge (years)57.94 ± 8.4056.08 ± 8.1357.42 ± 8.2457.95 ± 8.4160.32 ± 8.28< 0.001Sex, n (%)< 0.001Male2464 (46.55)723 (54.61)658 (49.74)607 (45.88)476 (35.98)Female2829 (53.45)601 (45.39)665 (50.26)716 (54.12)847 (64.0)Marital status, n (%)< 0.001Unmarried481 (9.09)104 (7.85)104 (7.86)106 (8.01)167 (12.62)Married4812 (90.91)1220 (92.15)1219 (92.14)1217 (91.99)1156 (87.38)Residence, n (%)< 0.001Urban3632 (68.62)860 (64.95)888 (67.12)901 (68.10)983 (74.30)Rural1661 (31.38)464 (35.05)435 (32.88)422 (31.90)340 (25.70)Education, n (%)< 0.001< Middle school3645 (68.86)806 (60.88)851 (64.32)914 (69.09)1074 (81.18)≥ Middle school1648 (31.14)518 (39.12)472 (35.68)409 (30.91)249 (18.82)Smoking status, n (%)< 0.001Non-smoker3248 (61.36)752 (56.80)788 (59.56)817 (61.75)891 (67.35)Current smoker2045 (38.64)572 (43.20)535 (40.44)506 (38.24)432 (32.65)Drinking status, n (%)< 0.001Non-drinker1825 (34.48)540 (40.79)485 (36.66)443 (33.48)357 (26.98)Current drinker3468 (65.52)784 (59.21)838 (63.34)880 (66.52)966 (73.02)Hypertension, n (%)< 0.001Yes1038 (19.61)84 (6.34)182 (13.76)316 (23.89)456 (34.47)No4255 (80.38)1240 (93.66)1141 (86.24)1007 (76.11)867 (65.53)Diabetes, n (%)< 0.001Yes226 (4.27)6 (0.45)22 (1.66)64 (4.84)134 (10.12)No5067 (95.73)1318 (99.55)1301 (98.34)1259 (95.16)1189 (89.87)Dyslipidemia, n (%)< 0.001Yes383 (7.24)57 (4.31)76 (5.74)92 (6.95)158 (11.94)No4910 (92.76)1267 (95.69)1247 (94.26)1231 (93.05)1165 (88.06)BMI (kg/m2)23.42 ± 3.8023.06 ± 3.2623.31 ± 3.6123.53 ± 3.9623.80 ± 4.27< 0.001SBP (mm Hg)128.77 ± 20.77127.18 ± 19.33127.02 ± 20.30129.78 ± 21.43131.12 ± 21.68< 0.001DBP (mm Hg)75.23 ± 11.9875.29 ± 11.7974.54 ± 11.8975.66 ± 12.3375.43 ± 11.890.034TC (mg/dL)193.55 ± 37.77191.35 ± 38.26191.83 ± 37.53193.88 ± 37.55197.14 ± 37.51< 0.001HDL (mg/dL)51.83 ± 15.3552.30 ± 15.1052.90 ± 15.5951.14 ± 15.1050.96 ± 15.52< 0.001LDL (mg/dL)116.38 ± 34.50115.65 ± 35.35115.11 ± 33.01116.92 ± 34.93117.84 ± 34.620.192HbA1c5.25 ± 0.775.17 ± 0.685.17 ± 0.765.26 ± 0.765.39 ± 0.95< 0.001CTI-FI0.75 ± 0.510.23 ± 0.100.52 ± 0.070.80 ± 0.101.44 ± 0.45< 0.001SD: standard deviation; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; HbA1c: glycated haemoglobin; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index. The values are the means ± SDs or n (percentages). a Continuous and categorical variables were compared using analyses of t-tests and chi-square tests, respectively Baseline characteristics of participants according to baseline CTI-FI quartiles SD: standard deviation; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; HbA1c: glycated haemoglobin; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index. The values are the means ± SDs or n (percentages). a Continuous and categorical variables were compared using analyses of t-tests and chi-square tests, respectively In accordance with additional exclusion criteria, 1364 participants (mean age: 57.35 years; female: 54.03%) who did not have a stroke before wave 3 (2015) were included in this study for dynamic changes in CTI-FI analyses. The mean CTI-FI values for these participants in 2012 and 2015 were 0.81 and 1.11, respectively. Table S2 and Table S3 present the detailed baseline characteristics of these participants stratified by cuCTI-FI tertiles (Q1-Q3) and three trajectories of traCTI-FI (Cluster 1 to Cluster 3), respectively, which are similar to the baseline characteristics of participants stratified by CTI-FI quartiles (Q1-Q4) in the baseline CTI-FI analyses. Associations of different CTI-FI dimensions (baseline CTI-FI, cuCTI-FI, and traCTI-FI) with stroke risk A total of 540 (10.20%) stroke events occurred among the 5293 participants included in the baseline CTI-FI analyses during almost 9 years of follow-up. Additionally, in the changes in CTI-FI analyses (including cuCTI-FI and traCTI-FI analyses), 125 (9.16%) stroke events among 1364 participants occurred during follow-up (Table S4). Three Cox proportional hazards regression models were constructed to assess the relationships between the three dimensions of the CTI-FI and stroke incidence (Table 2). Significant positive relationships between baseline CTI-FI and cuCTI-FI as continuous variables and stroke risk were identified in all the models. Each 1-unit increase in baseline CTI-FI or cuCTI-FI was significantly associated with 61% and 41% higher stroke risk, respectively, in the fully adjusted model 3, with HRs (95% CIs) of 1.61 (1.38, 1.88) and 1.41 (1.29, 1.53), respectively. Restricted cubic spline (RCS) modeling based on model 3 further demonstrated a significant positive nonlinear association between baseline CTI-FI and stroke risk (P for overall < 0.001, cutoff value = 0.50; P for nonlinear = 0.049) and between cuCTI-FI and stroke risk (P for overall < 0.001, cutoff value = 2.08; P for nonlinear = 0.047) (Fig. 3A, B). Table 2Associations of baseline CTI-FI, cuCTI-FI, and traCTI-FI with the risk of strokeCharacteristicModel 1Model 2Model 3HR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueCTI-FI (per 1 unit)1.91 (1.68, 2.18)< 0.0011.86 (1.62, 2.13)< 0.0011.61 (1.38, 1.88)< 0.001 CTI-FI (Quartiles) Q11.0 (Reference)1.0 (Reference)1.0 (Reference)Q21.57 (1.16, 2.12)0.0031.54 (1.14, 2.08)0.0051.49 (1.10, 2.02)0.009Q32.27 (1.71, 3.01)< 0.0012.23 (1.68, 2.97)< 0.0011.94 (1.45, 2.59)< 0.001Q43.16 (2.41, 4.14)< 0.0013.01 (2.28, 3.98)< 0.0012.36 (1.76, 3.15)< 0.001P for trend< 0.001< 0.001< 0.001cuCTI-FI (per 1 unit)1.47 (1.36, 1.58)< 0.0011.44 (1.33, 1.56)< 0.0011.41 (1.29, 1.53)< 0.001 cuCTI-FI (Tertiles) Q11.0 (Reference)1.0 (Reference)1.0 (Reference)Q21.63 (0.87, 3.04)0.1241.58 (0.85, 2.96)0.1491.59 (0.85, 2.98)0.149Q35.61 (3.28, 9.58)< 0.0015.17 (2.99, 8.95)< 0.0014.75 (2.73, 8.27)< 0.001P for trend< 0.001< 0.001< 0.001 traCTI-FI (Clusters) Cluster 11.0 (Reference)1.0 (Reference)1.0 (Reference)Cluster 23.54 (2.32, 5.42)< 0.0013.43 (2.22, 5.29)< 0.0013.15 (2.04, 4.89)< 0.001Cluster 37.81 (4.74, 12.85)< 0.0017.00 (4.19, 11.69)< 0.0016.24 (3.72, 10.46)< 0.001P for trend< 0.001< 0.001< 0.001HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; Model 1: Unadjusted model. Model 2: Adjusted for age, sex, marital status, place of residence, educational attainment, smoking and alcohol consumption status. Model 3: Further adjusted for hypertension, diabetes mellitus, and dyslipidemia, BMI, SBP, DBP, TC, TG, LDL-C, HDL-C and HbA1c based on Model 2 Table 2Associations of baseline CTI-FI, cuCTI-FI, and traCTI-FI with the risk of strokeCharacteristicModel 1Model 2Model 3HR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueCTI-FI (per 1 unit)1.91 (1.68, 2.18)< 0.0011.86 (1.62, 2.13)< 0.0011.61 (1.38, 1.88)< 0.001 CTI-FI (Quartiles) Q11.0 (Reference)1.0 (Reference)1.0 (Reference)Q21.57 (1.16, 2.12)0.0031.54 (1.14, 2.08)0.0051.49 (1.10, 2.02)0.009Q32.27 (1.71, 3.01)< 0.0012.23 (1.68, 2.97)< 0.0011.94 (1.45, 2.59)< 0.001Q43.16 (2.41, 4.14)< 0.0013.01 (2.28, 3.98)< 0.0012.36 (1.76, 3.15)< 0.001P for trend< 0.001< 0.001< 0.001cuCTI-FI (per 1 unit)1.47 (1.36, 1.58)< 0.0011.44 (1.33, 1.56)< 0.0011.41 (1.29, 1.53)< 0.001 cuCTI-FI (Tertiles) Q11.0 (Reference)1.0 (Reference)1.0 (Reference)Q21.63 (0.87, 3.04)0.1241.58 (0.85, 2.96)0.1491.59 (0.85, 2.98)0.149Q35.61 (3.28, 9.58)< 0.0015.17 (2.99, 8.95)< 0.0014.75 (2.73, 8.27)< 0.001P for trend< 0.001< 0.001< 0.001 traCTI-FI (Clusters) Cluster 11.0 (Reference)1.0 (Reference)1.0 (Reference)Cluster 23.54 (2.32, 5.42)< 0.0013.43 (2.22, 5.29)< 0.0013.15 (2.04, 4.89)< 0.001Cluster 37.81 (4.74, 12.85)< 0.0017.00 (4.19, 11.69)< 0.0016.24 (3.72, 10.46)< 0.001P for trend< 0.001< 0.001< 0.001HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; Model 1: Unadjusted model. Model 2: Adjusted for age, sex, marital status, place of residence, educational attainment, smoking and alcohol consumption status. Model 3: Further adjusted for hypertension, diabetes mellitus, and dyslipidemia, BMI, SBP, DBP, TC, TG, LDL-C, HDL-C and HbA1c based on Model 2 Associations of baseline CTI-FI, cuCTI-FI, and traCTI-FI with the risk of stroke HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; Model 1: Unadjusted model. Model 2: Adjusted for age, sex, marital status, place of residence, educational attainment, smoking and alcohol consumption status. Model 3: Further adjusted for hypertension, diabetes mellitus, and dyslipidemia, BMI, SBP, DBP, TC, TG, LDL-C, HDL-C and HbA1c based on Model 2 Fig. 3Multidimensional analyses of the associations of baseline CTI-FI, cuCTI-FI, and traCTI-FI with Stroke Risk. RCS analysis for baseline CTI-FI A and cuCTI-FI B. K‑M survival analysis compared stroke cumulative incidence by baseline CTI-FI C, cuCTI-FI D, and traCTI-FI E subgroups. ROC curves analysis for individual indicators F. HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; All analyses were based on the fully adjusted model 3 Multidimensional analyses of the associations of baseline CTI-FI, cuCTI-FI, and traCTI-FI with Stroke Risk. RCS analysis for baseline CTI-FI A and cuCTI-FI B. K‑M survival analysis compared stroke cumulative incidence by baseline CTI-FI C, cuCTI-FI D, and traCTI-FI E subgroups. ROC curves analysis for individual indicators F. HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; All analyses were based on the fully adjusted model 3 This study also revealed positive associations between baseline CTI-FI (Q1-Q4), cuCTI-FI (Q1-Q3), and traCTI-FI (Clusters 1–3) and the risk of stroke. First, compared with participants in the lowest baseline CTI-FI quartile (Q1), the risk of stroke increased with increasing quantiles. The fully adjusted HRs (95% CIs) were as follows: Q2: 1.49 (1.10, 2.02), Q3: 1.94 (1.45, 2.59), and Q4: 2.36 (1.76, 3.15). Additionally, compared with the participants in the lowest cuCTI-FI tertile (Q1), those in the highest cuCTI-FI tertile (Q3) had a higher stroke risk, with an HR (95% CI) of 4.75 (2.73, 8.27) in Model 3. Finally, compared with participants in Cluster 1 (persistent low-level of CTI-FI), those in Clusters 2 and 3 had significantly increased stroke risk, with fully adjusted HRs (95% CIs): 3.15 (2.04, 4.89) and 6.24 (3.72, 10.46), respectively. The results of the trend tests were consistent with the above positive association results (all P trends < 0.001) (Table 2). Kaplan-Meier (K-M) survival analysis The Kaplan-Meier cumulative incidence curve revealed that stroke incidence progressively increased with increasing baseline CTI-FI quartiles from Q1 to Q4, cuCTI-FI tertiles from Q1 to Q3, and traCTI-FI from Cluster 1 to Cluster 3, with statistically significant intergroup differences (all log-rank P < 0.001) (Fig. 3C-E). These K-M survival results further revealed a positive association among the three dimensions of the CTI-FI and stroke risk. ROC curve analysis The cuCTI-FI had the highest predictive value for stroke (AUC = 0.758, 95% CI: 0.716–0.801), followed by the traCTI-FI (AUC = 0.750, 95% CI: 0.706–0.795), the baseline CTI-FI (AUC = 0.715, 95% CI: 0.671–0.760), the baseline FI (AUC = 0.714, 95% CI: 0.669–0.758), and the baseline CTI (AUC = 0.684, 95% CI: 0.636–0.731) (Fig. 3F and Table S5). Table S6 shows that for predicting stroke risk, the baseline CTI-FI, cuCTI-FI, and traCTI-FI significantly outperformed the baseline CTI (P = 0.045, < 0.001, and < 0.001, respectively) and that the cuCTI-FI and traCTI-FI outperformed the baseline FI (P < 0.001 and = 0.017, respectively). The predictive value of the cuCTI-FI and traCTI-FI significantly exceeded that of the baseline CTI-FI (P < 0.001 and = 0.022, respectively). Subgroup and interaction analyses To evaluate the reliability of the association between the three CTI-FI dimensions and stroke risk and whether certain factors modulate these relationships, subgroup analyses and interaction tests were conducted within subgroups in this study on the basis of the fully adjusted model 3. As shown in Fig. 4A and Table S7, positive associations between baseline CTI-FI and stroke risk were observed across most stratification factors (HRs > 1, P < 0.05), and interactions were found with age (Pinteraction = 0.047) and educational attainment (Pinteraction = 0.037). A positive association was also found between cuCTI-FI and traCTI-FI and stroke across most stratification factors (HRs > 1, P < 0.05). In addition, educational attainment (Pinteraction = 0.038) and drinking status (Pinteraction = 0.021) showed significant interactions for cuCTI-FI and traCTI-FI, respectively (Fig. 4B-C and Table S8, S9). Fig. 4Subgroup and interaction analyses of baseline CTI-FI A, cuCTI-FI B and traCTI-FI C in association with stroke incidence. HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; All analyses were based on the fully adjusted model 3 Subgroup and interaction analyses of baseline CTI-FI A, cuCTI-FI B and traCTI-FI C in association with stroke incidence. HR: hazard ratio; CI: confidence interval; CTI: C-reactive protein-triglyceride glucose index; FI: Frailty index; cuCTI-FI: cumulative CTI-FI; traCTI-FI: dynamic trajectories of CTI-FI; All analyses were based on the fully adjusted model 3 Sensitivity analyses Table S10 presents the results of the sensitivity analyses, which confirm the robustness of our primary findings regarding the associations between CTI-FI and stroke. The association between baseline CTI-FI and the increased risk of stroke persisted in model 3, fully adjusted for confounding factors, after excluding participants with baseline CTI-FI values beyond 3 standard deviations, converting the form of the confounding factors, and changing the follow-up endpoint from 2020 to 2018. Furthermore, consistent results were found when we divided participants into 3 groups based on the baseline CTI-FI tertiles, 4 groups based on the cuCTI-FI quartiles, and 4 clusters using the K-means clustering method. Finally, in terms of predictive performance, the findings of cuCTI-FI and traCTI-FI are superior to baseline CTI-FI and to CTI or FI remaining, based on the RSF model (Table S11, S12). Discussion The composite indicator CTI-FI was first proposed in this study, which provides a novel biomarker perspective for precise stroke risk stratification in the population with CKM syndrome stages 0–3. The findings of our study revealed that both baseline CTI-FI and cuCTI-FI were significantly positively correlated with stroke risk, with a clear dose-response relationship. Additionally, the risk of stroke increases with higher quantiles of baseline CTI-FI and cuCTI-FI. Moreover, the risk further increased among participants with stable high-level traCTI-FI. Notably, the primary findings remained stable across multiple models adjusted for confounding factors, as well as in stratified and sensitivity analyses. In conclusion, these findings highlight the potential clinical relevance of CTI-FI and further underscore the significant value of its dynamic monitoring dimension for stroke risk prediction, providing key clinical evidence for the early prevention and control of stroke in the population with stages 0–3 CKM syndrome. In this study, drawing upon the data from participants aged 45 and above in the CHARLS with stages 0–3 CKM syndrome, a notable positive correlation was identified between CTI-FI and stroke risk. Although this study is the first to propose this composite indicator, direct comparisons with studies on other health effects using the same indicators are limited. However, the following evidence also supports the scientific validity and rationality of this study. First, previous studies have consistently demonstrated a positive correlation between CTI and stroke. Huo et al. first focused on individuals with different blood glucose statuses and reported a significant positive linear correlation between the CTI and stroke [14]. Xu et al. extended the study to the CKM syndrome population and reported a positive association between CTI and stroke risk in this population [30]. Tang et al. validated its predictive role for stroke in hypertensive populations [31]. These findings highlight the importance of CTI as an integrated biomarker combining inflammatory and metabolic risk factors. Additionally, the frailty index (FI) is a precise indicator reflecting an individual’s biological aging characteristics and highlights the decline in physiological reserve capacity. The positive correlation between the FI and stroke risk has consistently been verified. Shrauner et al. validated the predictive value of FI for stroke, confirming a positive dose-dependent relationship between the degree of frailty and stroke risk [32]. Zhang et al. reported that worsening frailty status increases stroke risk [33]. Huang et al. identified that individuals with persistent or progressive frailty had the highest risk of stroke [34]. All of these findings provide theoretical support for the combined application of the CTI and FI in stroke risk prediction. From a mechanistic perspective, CTI and FI are closely interconnected through a bidirectional regulatory mechanism [35]. The chronic inflammatory state reflected by CTI can activate nuclear factor-κB signaling pathway in vivo [36] and promote the release of numerous inflammatory factors [37]. These inflammatory factors further inhibit the protein synthesis pathway in skeletal muscle cells [38], accelerate muscle tissue loss and deplete physiological reserve functions [39], and ultimately contribute to the onset and progression of frailty [40]. On the other hand, frailty, represented by FI, exacerbates insulin resistance and systemic inflammatory responses through mechanisms such as reduced physical exercise load [41] and disrupted nutritional metabolic balance [42], thereby forming a vicious cycle of “inflammation-metabolic disorder-biological aging”. This vicious cycle is particularly pronounced in individuals with CKM syndrome stages 0–3 [43], which accelerates the formation and rupture of atherosclerotic plaques, leading to acute tissue ischemia [44] and thereby inducing ischemic stroke [45]. The overlap of these mechanisms provides a biological basis for the combined evaluation of CTI and FI. In this study, the CTI-FI was calculated using a multiplication model, which is widely applied in epidemiological analyses and reflects potential interactions among metabolism, inflammation, and functional defects. It can better reflect the body’s overall health status. Xu et al. found that, compared with the TyG and CRP indices, the CTI has a superior ability to predict cardiovascular disease [46]. David Sacks et al. demonstrated that the TyG-waist-to-height ratio (TyG-WHtR) and TyG-waist circumference (TyG-WC) outperformed the TyG index alone in assessing cardiovascular disease risk [47]. Consistent with the findings of previous studies, this study revealed that CTI-FI has better predictive performance for stroke than either the CTI or the FI alone. This study fills a gap in literature by evaluating inflammation and metabolic indicators and by considering the development of new indicators of functional health status. These advantages highlight the originality and applicability of CTI-FI as a multidimensional risk marker, providing critical data to support its incorporation into precision stroke prevention strategies. The cuCTI-FI serves as a cumulative indicator that reflects long-term inflammation, metabolic processes, and biological aging. The findings of this study indicate that continuously high cumulative CTI-FI exposure gradually increases stroke risk, which is consistent with the findings of other studies that have constructed cumulative exposure indicators, such as cuCTI, cuTyG-WWI, and cuAIP [48–50]. These findings can be explained by the acknowledged pathophysiological mechanisms reported in previous studies. Chronic mild inflammation, metabolic disorders, and biological aging together accelerate stroke progression [51]. Meanwhile, dynamic CTI-FI trajectories can better reflect disease progression. Persistent low levels indicate relative stability and low risk, whereas a slight increase suggests a relatively minor impact. A sustained high trajectory may signal the onset of a vicious cycle among inflammation, metabolic disorders, and biological aging and corresponds to the highest risk of stroke in this study. Moreover, compared with the baseline CTI-FI, the cuCTI-FI and traCTI-FI exhibit superior performance in predicting stroke risk. Consistent with our study, a previous study also confirmed that the accumulating and dynamic trajectories in CTI indices are more effective than baseline CTI in predicting stroke risk [18]. Increasing evidence emphasizes the need to pay closer attention to the cumulative exposure and longitudinal changes in biomarkers. Subgroup and interaction analyses revealed that younger age and higher educational level strengthened the association between baseline CTI-FI and stroke risk. In addition, higher educational level and nondrinking status strengthened the associations of cuCTI-FI and traCTI-FI with stroke incidence, respectively. First, younger populations have greater metabolic reserve capacity, which results in a greater susceptibility to early metabolic disturbances [52], and demonstrate greater sensitivity to environmental perturbations [53]. Additionally, biochemical abnormalities may progress relatively rapidly to stroke risk in younger individuals [54], thereby strengthening the association between the CTI-FI and stroke risk in younger individuals. Second, individuals with higher levels of education typically possess stronger health awareness and self-management consciousness and are more inclined to actively participate in systematic health screening and metabolic marker monitoring [55]. Moreover, they are typically exposed to greater occupational psychological stress [56] and are more prone to adopting lifestyle habits that are detrimental to metabolic health, such as westernized dietary patterns and sedentary behavior [57]. These behavioral factors may synergistically contribute to the CKM syndrome dysfunction reflected by CTI-FI, thereby further strengthening the biological association between higher education levels and stroke risk. Finally, alcohol consumption directly interferes with lipid metabolism, insulin sensitivity, and chronic inflammatory pathways, creating superimposed and confounding effects on core pathological signals such as metabolic disturbances, inflammatory activation, and vascular damage, which are integrated by CTI-FI [58, 59]. Additionally, low to moderate alcohol consumption may exert certain protective effects by improving lipid profiles [60]. Furthermore, among alcohol consumers, drinking behavior often coexists with unhealthy lifestyle factors such as inadequate intake of vegetables and fruits and physical inactivity, as well as multiple chronic comorbidities [61, 62], which may independently or synergistically increase the risk of stroke [63]. Accordingly, these factors may further weaken the association between CTI-FI and stroke in the drinking population. Notably, factors such as gender, marital status, place of residence, and smoking did not significantly modulate the association between the three dimensions of CTI-FI and stroke risk. These findings suggest that the predictive value of CTI-FI for stroke risk demonstrates good cross-population consistency and stability. Furthermore, although hypertension, diabetes, and dyslipidemia are recognized classic risk factors for stroke [64–68], results of this study indicate that these factors did not significantly modulate the relationship between different dimensions of CTI-FI and stroke. The results suggest that CTI-FI may exert a stroke predictive role independently of these traditional risk factors, or participate in the pathogenesis of stroke through pathophysiological pathways that partially overlap with but fundamentally differ from those of conventional risk factors. Undeniably, this study has limitations. First, it might have a misclassification bias as stroke was identified via self-report based on physician-diagnosed records. However, a previous study has shown that self-reported stroke identification can be used to explore stroke risk factors and incidence [69]. Second, in this study, strokes cannot be classified into ischemic strokes and hemorrhagic strokes based on the CHARLS data. These two categories of strokes exhibit distinct pathophysiologies. Consequently, the associations between CTI-FI and the risks of different stroke categories may vary. Third, although the classification method for CKM stages 0–4 in this study was based on published literature [24–26] and AHA criteria [6], the inability to fully align with AHA standards for CKM stages 3 and 4 due to the lack of data, such as cardiovascular imaging and biomarkers, may introduce certain selection bias. Fourth, the study focuses on middle-aged and elderly (45+) individuals in the CHARLS. Yet, it has ethnic and age limitations, limiting the generalizability of the findings. Fifth, despite strict adjustment for confounding factors, residual confounding persisted due to unmeasured factors such as dietary patterns and genetic susceptibility. Even with the above limitations, the following advantages of this study cannot be ignored. First, this study was based on the national prospective cohort CHARLS, which enrolled 5293 subjects and completed nearly 9 years of follow-up. The large sample size and extended follow-up period ensured adequate statistical power. Second, multiple statistical methods were employed to validate the associations and predictive power of CTI-FI with stroke risk. Simultaneously, multi-dimensional sensitivity analyses, stratified analyses, and interaction analyses were conducted to confirm the stability and reliability of the primary results fully. Third, this is the first study to develop the CTI-FI composite index, integrating chronic inflammation, insulin resistance, and biological aging, thereby overcoming limitations of single indicators in assessing stroke risk. It offers a new biomarker perspective for stroke risk stratification in CKM 0–3 stages populations. Fourth, the CTI and FI are derived from routine testing/evaluation and can be obtained without additional expensive or invasive procedures, demonstrating good clinical usability. Fifth, this study focuses on the population in CKM stages 0–3, who were often overlooked but had clinically significant symptoms. Finally, this study established cuCTI-FI and traCTI-FI to systematically investigate the impact of long-term accumulation and dynamic changes on stroke risks. It transcends the single-point effect analysis of baseline CTI-FI, aligning more closely with clinical practice and enhancing the clinical reference value. Conclusions In individuals with CKM syndrome stages 0–3, the baseline CTI-FI was significantly positively associated with stroke incidence. Moreover, the risk increased among participants with persistently elevated cuCTI-FI levels and those with stable high-level traCTI-FI. Moreover, CTI-FI has greater predictive efficacy than CTI or FI alone. Notably, the cuCTI-FI and traCTI-FI have higher predictive capacity. These results indicate that the CTI-FI serves as a practical and reliable composite biomarker for identifying high-risk individuals with new-onset stroke in CKM syndrome 0–3 stages populations. In clinical practice, continuous monitoring of CTI-FI levels and their dynamic changes can provide reliable evidence for stroke risk stratification. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1