Abstract
Background
Cardiometabolic multimorbidity (CMM) leads to adverse health outcomes. Based on data from the China Health and Retirement Longitudinal Study (CHARLS), we aimed to explore the cumulative exposure and dynamic change patterns of metabolic syndrome score (MetSS) with CMM progression among middle-aged and older Chinese adults.
Methods
Age and sex specific MetSS was assessed according to equations which were developed for Chinese. K-means clustering analysis was used to classify MetSS changes, and cumulative MetSS (cuMetSS) was calculated as follows: (MetSS2012 + MetSS2015)/2 × time (2015 − 2012). The progression of CMM was defined starting with CMD-free, developing into first CMD (FCMD), further progressing into CMM. Logistic regression analyses and restricted cubic splines (RCS) were performed to evaluate the association of MetSS with CMM progression in 3 models. Subgroup and interaction analyses were subsequently undertaken to investigate the effect of physical activity on the associations of baseline MetSS, cuMetSS and MetSS change patterns with FCMD and CMM risk, respectively.
Results
A total of 3,322 participants were eligible for analysis, of whom 679 experienced FCMD and 101 progressed to CMM. The K-means method classified 4 clusters. Logistic analyses revealed that the risk of CMM both increased with baseline MetSS and cuMetSS increment in all 3 models. Baseline MetSS on continuous scale was not significantly associated with FCMD (all P > 0.05). Yet cuMetSS on continuous scale was significantly associated with increased risk of FCMD when adjusted age and gender (model 1: OR, 95% CI, P: 1.02, 1.01 to 1.03, 0.006), additionally adjusted education, marital status, residence, drinking status, smoking status, BMI and comorbidity (model 2: 1.02, 1.00 to 1.03, 0.008). Further additional adjustment for physical activity (model 3) yielded no statistical significance (P > 0.05). Further subgroup analyses suggested that significance was only noted in subgroups with inactive or vigorous physical activity.
Conclusions
Our findings indicate that cumulative exposure and dynamic change patterns of MetSS were associated with FCMD and CMM, yet there is a modifiable effect of physical activity on the associations of cuMetSS and MetSS change patterns with FCMD risk.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at 10.1186/s40842-026-00289-1.
Keywords: Metabolic syndrome score, Cardiometabolic multimorbidity, Dynamic changes, K-means clustering, Moderate physical activity, CHARLS
Background
Cardiometabolic multimorbidity (CMM), defined as the coexistence of two or three cardiometabolic diseases (CMDs), typically includes diabetes, stroke and heart disease [1, 2]. A population-based study found that CMM was the most prevalent pattern of multimorbidity affecting about 6% middle-aged and older Chinese adults [3]. As reported, CMM increases the risk of cognitive decline [4], dementia [5], depression [6], deterioration in physical disability [7], as well as all-cause mortality [8], leading to adverse health outcomes. It is hence of importance to explore cost-effective and reproducible risk stratification measures to alleviate health burden and improve life quality.
Metabolic syndrome score (MetSS), derived from weighted calculation with five components of metabolic syndrome (MetS) according to specific age, sex and ethnicity [9, 10], can quantify the severity of MetS and adequately capture of the subtle variations in the progression of MetS. Static state of MetSS was widely investigated to be associated with physical abnormalities and functional impairment [11–13]. However, MetSS various over time and its dynamic changes are more precise to reflect altered disease risk and track the progression of diseases [14, 15].
Existing evidence revealed that the long-term monitoring and evaluation of MetSS can predict the new-onset diseases, such as diabetes [16], frailty [17] and cancer [18]. Association between MetSS changes and CMM progression remains inadequately elucidated. And it is necessary to explore the association in a longitudinal design.
Our study, on the basis of national data from the China Health and Retirement Longitudinal Study (CHARLS), targeting middle-aged and older Chinese adults, aims to explore the cumulative exposure and dynamic change patterns of MetSS with CMM progression, provide evidence for the early identification of high-risk CMM individuals.
Methods
Study design and population
CHARLS, a national survey of residents ages 45 and older in China, is supported by Peking University, the National Natural Science Foundation of China, the Behavioral and Social Research Division of the National Institute on Aging and the World Bank. Adopting multi-stage stratified sampling with probability proportional to size, CHARLS gathers a high quality nationally representative sample, including data related to health status and biomarkers of participants. The datasets are available via the official website at https://charls.pku.edu.cn/en/. The Institutional Review Board of Peking University approves CHARLS survey protocols. All participants provided written informed consent in CHARLS.
The present study initially included 1,1847 participants with blood examination in Wave 1 (2011–2012), with follow-up extending through Wave 2 (2013–2014), Wave 3 (2015–2016), Wave 4 (2017–2018) and Wave 5 (2019–2020). The exposure period was Wave 1–3 and outcome period was Wave 4–5, participants with CMD before Wave 4 were excluded in the final analysis [19]. Following specific exclusion criteria were adopted: (1) participants without age data (n = 84) or under 45 years old (n = 346); (2) participants with CMD in Wave 1 (n = 3,207); (3) participants without MetSS value in Wave 1 (n = 1,412); (4) participants with CMD in Wave 3 (n = 1,835); (5) participants without MetSS value in Wave 3 (n = 1,395); (6) participants without CMD data in Wave 4 or Wave 5 (n = 246). Finally, a total of 3,322 participants with eligible information were enrolled in the final analysis (Fig. 1).
Fig. 1.
Flowchart for participants of the present study. Abbreviations: CMD, cardiometabolic disease; MetSS, metabolic syndrome score; FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity
Exposure variable and outcome variables
MetSS was the exposure variable. Age and sex specific MetSS was assessed according to equations [9], which were developed for Chinese, including 5 parameters (Table S1) which were measured only in Wave 1 and Wave 3: waist circumference (WC, cm), triglycerides (TG, mmol/L), high-density lipoprotein cholesterol (HDL, mmol/L), mean arterial pressure (MAP, mm Hg) and fasting blood glucose (FBG, mmol/L). The units of TG and FBG were mg/dL in CHARLS, which were converted properly before calculation of MetSS.
Cumulative MetSS (cuMetSS) was calculated as follows: (MetSS2012 + MetSS2015)/2 × time (2015 − 2012) [17]. MetSS and cuMetSS were analyzed both as the continuous variable and as the categorical variable (based on quartile category for all eligible participants). MetSS changes were classified by K-means clustering using Euclidean distance [20]. Z-score transformation was performed for standardization. Random seed was set as 123 before clustering. The optimal number of clusters (K) was determined by elbow method and four clusters were selected, as this point represents where the curve flattens, indicating that four clusters (K = 4) best balance model simplicity and explanatory accuracy while avoiding overfitting. Moreover, To avoid the impact of extreme value, outliers were removed by density-based spatial clustering of applications with noise (DBSCAN) clustering. Mean silhouette was used to assess validity of clustering (which was 0.52). Stability of clustering was assessed by bootstrapping (B = 1000, Jaccard coefficients of 4 clusters were 0.767, 0.580, 0.678 and 0.648, respectively). In addition, sensitivity analyses were conducted using three (K = 3) or five (K = 5) clusters, as well as clusters without DBSCAN exclusion. R Packages “dbscan”, “Rtsne”, “factoextra”, “cluster” and “fpc” were used.
The progression of CMM was defined starting with CMD-free (Wave 1–3), developing into first CMD (FCMD), further progressing into CMM (Wave 4–5). New-onset FCMD or CMM in Wave 4–5 were the outcome variables. FCMD was diagnosed when a participant had one CMDs, including diabetes, stroke and heart disease. CMM was diagnosed when a participant had two or more CMDs. In present study, diabetes was defined as FBG ≥ 126 mg/dL, glycohemoglobin A1c (HbA1c) ≥ 6.5% (available only in Wave 1 and Wave 3) or self-reported DM (Have you been diagnosed with diabetes or hyperglycemia by a doctor?”). Stroke and heart disease were defined according to self-reported medical conditions (“Have you been diagnosed by a doctor with heart disease, coronary artery disease, angina, congestive heart failure, or other heart problems?” and “Have you been diagnosed by a doctor as having a stroke?”).
Covariates
The covariates included the following categories: (1) demographics variables: age, gender, education, residence, marital status; (2) lifestyle variables and doctor diagnosed health problems: drinking status, smoking status, physical activity, hypertension, comorbidity condition; (3) biomarkers: height, weight, body mass index (BMI), WC, systolic blood pressure (SBP), diastolic blood pressure (DBP), FBG, HbA1c, TG, total cholesterol (TC), HD and low-density lipoprotein cholesterol (LDL), blood urea nitrogen (BUN), creatinine (Cr), uric acid (UA), Cystatin C, white blood cell (WBC) and C-reactive protein (CRP).
Age was classified into ≥ 60 years old and < 60 years old. Education included three categories: no formal education, < high school, high school and above. Residence comprised rural and urban. Marital status was classified into married/partnered and never married/widowed/divorced/separated. Drinking status was categorized into yes and no. Smoking status was classified into current, former and never. Physical activity stratified into four categories: inactive, moderate and vigorous.
Hypertension was defined as mean SBP/DBP ≥ 140/90 mm Hg or self-reported hypertension (“Have you been diagnosed with hypertension by a doctor?”). Comorbidity condition was defined as if the participants reported at least one of the following doctor diagnosed health problems: kidney disease, liver disease, lung disease or asthma and cancer. BMI was categorized into underweight (< 18 kg/m2), normal weight (18.5–23 kg/m2), overweight (23–25 kg/m2) and obesity (≥ 25 kg/m2) according to the WHO recommendation for the Asian and South Asian population. Abdominal obesity was defined as WC > 85 cm in females or > 90 cm in males according to the classification for Chinese adults.
Statistical analyses
Statistical analyses were done using the R version 4.5.1. Two-sided P value less than 0.05 was considered statistically significant. Categorical variables were presented as count (percentage). Continuous variables were presented as mean (standard deviation). Kruskal-Wallis rank-sum test, t-test, and χ2 test were used to compare variables across groups, where appropriate.
Logistic regression analyses were performed to assess the association of MetSS and cuMetSS with FCMD and CMM after adjustment, respectively. The adjustment comprises of model 1 (age and gender), model 2 (additionally adding education, marital status, residence, drinking status, smoking status, BMI and comorbidity) and model 3 (additionally adding physical activity) adjusted patterns. The associations of the 4 clusters with FCMD and CMM were also analyzed using logistic regression models, respectively. Effect-size estimates are summarized as odds ratio (OR) and 95% CI. Restricted cubic splines (RCS) were conducted to potential nonlinear associations between of cuMetSS with FCMD and CMM, respectively. RCS was established by “rms” R Package, which knot number was 4 (5th, 35th, 65th and 95th percentile), reference value was median and standardization was not required. Sensitivity analyses were adopted: (1) to evaluate the association of MetSS at Wave 3 and ΔMetSS (= MetSS at Wave 3 - baseline MetSS) with FCMD and CMM, respectively; (2) to assess the association of baseline MetSS, MetSS at Wave 3, ΔMetSS and cuMetSS with FCMD and CMM up to Wave 4, respectively. Likelihood ratio (LR) test and Akaike information criterion (AIC) were used to evaluate prediction performance of cuMetSS and clustering.
Missing data of key variables were less than 0.65% except for physical activity, and were imputed using the random forest method. Quantification for missing data of key variables were shown by “VIM” R Package in supplementary materials (Figure S1). As for physical activity, to validate its effect on the associations of baseline MetSS and cuMetSS with FCMD and CMM risk, subgroup and interaction analyses were subsequently undertaken after adjustment of age and gender (model 1) and additional adjustment of education, marital status, residence, drinking status, smoking status, BMI and comorbidity (model 2).
Finally, mediation analyses were applied to assess the mediating effects of inflammatory factors (WBC and CRP) on the association of MetSS and cuMetSS with FCMD and CMM, respectively. Mediating effects were confirmed with results of significant total effect, significant indirect effect and positive proportion of the mediator effect.
Results
Participants characteristics
The participants characteristics of 3,322 participants (mean age 57.62 ± 8.48 years old, 46.48% males) from CHARLS in this study are shown in Table 1, of whom 679 experienced FCMD and 101 progressed to CMM. Compared with CMD-free participants, participants with CMM were more likely to have higher baseline MetSS and cuMetSS (all P < 0.001).
Table 1.
The characteristics of all study participants (n = 3,322)
| Variables | Total (n = 3,322) | CMD | P | ||
|---|---|---|---|---|---|
| CMD-free (n = 2,542) | FCMD (n = 679) | CMM (n = 101) | |||
| Age (years old) | 57.62 ± 8.48 | 57.28 ± 8.50 | 58.68 ± 8.37 | 58.98 ± 8.07 | < 0.001 |
| Age group | < 0.001 | ||||
| ≥60 | 1265 (38.08) | 917 (36.07) | 47 (46.53) | 301 (44.33) | |
| <60 | 2057 (61.92) | 1625 (63.93) | 54 (53.47) | 378 (55.67) | |
| Gender | 0.029 | ||||
| Male | 1544 (46.48) | 1209 (47.56) | 50 (49.50) | 285 (41.97) | |
| Female | 1778 (53.52) | 1333 (52.44) | 51 (50.50) | 394 (58.03) | |
| Residence | 0.390 | ||||
| Rural | 2884 (87.34) | 2199 (86.51) | 86 (85.15) | 599 (88.22) | |
| Urban | 418 (12.66) | 329 (12.94) | 14 (13.86) | 75 (11.05) | |
| Education | 0.187 | ||||
| No formal education | 940 (28.30) | 697 (27.42) | 35 (34.65) | 208 (30.63) | |
| < High school | 2065 (62.16) | 1604 (63.10) | 54 (53.47) | 407 (59.94) | |
| High school and above | 317 (9.54) | 241 (9.48) | 12 (11.88) | 64 (9.43) | |
| Marital | 0.528 | ||||
| Married/partnered | 3012 (90.67) | 2312 (90.95) | 92 (91.09) | 608 (89.54) | |
| Never married/widowed/divorced/separated | 310 (9.33) | 230 (9.05) | 9 (8.91) | 71 (10.46) | |
| Alcohol | 0.096 | ||||
| Yes | 1291 (38.90) | 995 (39.14) | 48 (47.52) | 248 (36.52) | |
| No | 2028 (61.10) | 1546 (60.82) | 53 (52.48) | 429 (63.18) | |
| Smoke | 0.029 | ||||
| Never | 2067 (62.39) | 1550 (60.98) | 64 (63.37) | 453 (66.72) | |
| Former | 216 (6.52) | 164 (6.45) | 9 (8.91) | 43 (6.33) | |
| Current | 1030 (31.09) | 824 (32.42) | 28 (27.72) | 178 (26.22) | |
| Physical activity | 0.394 | ||||
| Vigorous | 600 (41.78) | 474 (42.40) | 13 (28.26) | 113 (41.54) | |
| Moderate | 430 (29.94) | 335 (29.96) | 17 (36.96) | 78 (28.68) | |
| Inactive | 406 (28.27) | 309 (27.64) | 16 (34.78) | 81 (29.78) | |
| FBG (mg/dL) | 99.52 ± 11.18 | 99.09 ± 10.98 | 100.62 ± 11.43 | 102.80 ± 13.37 | < 0.001 |
| HbA1c (%) | 5.08 ± 0.38 | 5.07 ± 0.38 | 5.11 ± 0.37 | 5.18 ± 0.46 | 0.003 |
| TG (mg/dL) | 117.70 ± 73.99 | 115.51 ± 72.24 | 121.88 ± 77.01 | 144.68 ± 89.47 | < 0.001 |
| TC (mg/dL) | 191.05 ± 36.58 | 189.52 ± 36.07 | 194.19 ± 36.76 | 208.59 ± 42.25 | < 0.001 |
| HDL (mg/dL) | 52.59 ± 14.65 | 52.85 ± 14.72 | 52.33 ± 14.47 | 47.89 ± 13.44 | 0.003 |
| LDL (mg/dL) | 116.00 ± 32.98 | 114.57 ± 32.41 | 118.92 ± 33.16 | 132.35 ± 39.92 | < 0.001 |
| BUN (mg/dL) | 15.68 ± 4.37 | 15.63 ± 4.40 | 15.87 ± 4.29 | 15.48 ± 4.26 | 0.422 |
| Cr (mg/dL) | 0.76 ± 0.18 | 0.76 ± 0.18 | 0.76 ± 0.17 | 0.79 ± 0.20 | 0.259 |
| UA (mg/dL) | 4.32 ± 1.18 | 4.31 ± 1.16 | 4.34 ± 1.22 | 4.63 ± 1.26 | 0.024 |
| Cystatin_C (mg/L) | 0.99 ± 0.24 | 0.99 ± 0.24 | 1.00 ± 0.20 | 1.03 ± 0.35 | 0.318 |
| CRP (mg/L) | 2.21 ± 6.81 | 2.16 ± 7.02 | 2.37 ± 6.12 | 2.47 ± 5.97 | 0.706 |
| WBC (1000 cells/uL) | 6.10 ± 1.80 | 6.08 ± 1.82 | 6.17 ± 1.77 | 6.37 ± 1.58 | 0.157 |
| Height (cm) | 1.58 ± 0.09 | 1.58 ± 0.09 | 1.58 ± 0.08 | 1.58 ± 0.08 | 0.906 |
| Weight (kg) | 57.84 ± 10.84 | 57.49 ± 10.69 | 58.76 ± 11.24 | 60.48 ± 11.24 | 0.001 |
| WC (cm) | 82.86 ± 11.98 | 82.42 ± 11.54 | 84.12 ± 13.13 | 85.51 ± 13.80 | < 0.001 |
| BMI (kg/m2) | 23.26 ± 8.73 | 23.16 ± 9.76 | 23.50 ± 3.76 | 24.22 ± 3.58 | 0.354 |
| BMI category | 0.004 | ||||
| Underweight | 215 (6.52) | 171 (6.73) | 2 (1.98) | 42 (6.19) | |
| Normal weight | 1522 (46.12) | 1200 (47.21) | 41 (40.59) | 281 (41.38) | |
| Overweight | 716 (21.70) | 543 (21.36) | 21 (20.79) | 152 (22.39) | |
| Obesity | 847 (25.67) | 611 (24.04) | 37 (36.63) | 199 (29.31) | |
| Abdominal obesity | < 0.001 | ||||
| Yes | 1128 (33.96) | 800 (31.47) | 43 (42.57) | 285 (41.97) | |
| No | 2194 (66.04) | 1742 (68.53) | 58 (57.43) | 394 (58.03) | |
| SBP (mm Hg) | 125.91 ± 19.76 | 124.88 ± 19.28 | 128.83 ± 20.85 | 132.06 ± 21.05 | < 0.001 |
| DBP (mm Hg) | 74.00 ± 11.83 | 73.47 ± 11.58 | 75.69 ± 12.40 | 75.94 ± 12.93 | < 0.001 |
| Hypertension | < 0.001 | ||||
| Yes | 1031 (31.04) | 727 (28.60) | 44 (43.56) | 260 (38.29) | |
| No | 2291 (68.96) | 1815 (71.40) | 57 (56.44) | 419 (61.71) | |
| Comorbidity | 0.002 | ||||
| Yes | 532 (16.03) | 375 (14.75) | 20 (19.80) | 137 (20.18) | |
| No | 2787 (83.97) | 2165 (85.17) | 81 (80.20) | 541 (79.68) | |
| MetSS | 0.81 ± 2.53 | 0.72 ± 2.49 | 0.98 ± 2.57 | 1.73 ± 3.04 | < 0.001 |
| MetSS (Wave 3) | 1.18 ± 2.57 | 1.08 ± 2.54 | 1.48 ± 2.63 | 1.66 ± 2.78 | < 0.001 |
| cuMetSS | 2.98 ± 6.87 | 2.70 ± 6.79 | 3.68 ± 6.95 | 5.08 ± 7.86 | < 0.001 |
| New-onset heart disease | < 0.001 | ||||
| Yes | 441 (13.96) | 0 (0.00) | 83 (82.18) | 358 (52.72) | |
| No | 2717 (86.04) | 2398 (94.34) | 18 (17.82) | 301 (44.33) | |
| New-onset stroke | < 0.001 | ||||
| Yes | 233 (7.39) | 0 (0.00) | 67 (66.34) | 166 (24.45) | |
| No | 2922 (92.61) | 2398 (94.34) | 32 (31.68) | 492 (72.46) | |
| New-onset diabetes | < 0.001 | ||||
| Yes | 230 (7.30) | 0 (0.00) | 65 (64.36) | 165 (24.30) | |
| No | 2919 (92.70) | 2399 (94.37) | 33 (32.67) | 487 (71.72) | |
Abbreviations: CMD, cardiometabolic disease; FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; FBG, fasting blood glucose; HbA1c, glycohemoglobin A1c; TG, triglycerides; TC, total cholesterol; HDL, high density lipoprotein cholesterol; LDL, low density lipoprotein cholesterol; BUN, blood urea nitrogen; Cr, creatinine; UA, uric acid; CRP, C-reactive protein; WBC, white blood cell; WC, waist circumference; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; MetSS, metabolic syndrome score; cuMetSS, cumulative metabolic syndrome score. Data are expressed as mean (standard deviation) for continuous variables or count (percent) for categorical variables. P value was calculated by the t-test, Kruskal-Wallis rank-sum test or the χ2 test, where appropriate
Similarly, as illustrated in Table S2 and S3, based on baseline MetSS and cuMetSS quartile category for all eligible participants, participants with higher baseline MetSS and cuMetSS were more likely to suffer CMM (all P < 0.05).
MetSS changes were classified by K-means clustering using Euclidean distance. To avoid the impact of extreme value, outliers (n = 126) were removed by DBSCAN clustering, the result was visualized in Figure S2 (parameter: eps = 0.5, minPts = 5). The optimal number of clusters (K = 4) was determined by elbow method (Fig. 2A). The K-means method classified 4 clusters (Fig. 2B and C): “persistently low level” (cluster 1), “median level increase” (cluster 2), “median level decrease” (cluster 3) and “persistently high level” (cluster 4). Table 2 summarizes participants characteristics in 4 clusters.
Fig. 2.
The K-means clustering. A. The optimal number of clusters (K = 4) was determined by elbow method. B. The K-means method classified 4 clusters: “persistently low level” (cluster 1), “median level increase” (cluster 2), “median level decrease” (cluster 3) and “persistently high level” (cluster 4). C. MetSS change between 2012 and 2015 for the 4 clusters. Abbreviations: MetSS, metabolic syndrome score
Table 2.
The characteristics of study participants classified according to K-means clustering (n = 3,196)
| Variables | Total (n = 3,196) | Cluster | P | |||
|---|---|---|---|---|---|---|
| 1 (n = 1,100) | 2 (n = 352) | 3 (n = 1,221) | 4 (n = 523) | |||
| Age | 57.69 ± 8.50 | 58.80 ± 8.70 | 56.89 ± 8.44 | 57.54 ± 8.42 | 56.24 ± 8.02 | < 0.001 |
| Age group | < 0.001 | |||||
| ≥60 | 1966 (61.51) | 622 (56.55) | 216 (61.36) | 766 (62.74) | 362 (69.22) | |
| <60 | 1230 (38.49) | 478 (43.45) | 136 (38.64) | 455 (37.26) | 161 (30.78) | |
| Gender | < 0.001 | |||||
| Male | 1680 (52.57) | 397 (36.09) | 251 (71.31) | 654 (53.56) | 378 (72.28) | |
| Female | 1516 (47.43) | 703 (63.91) | 101 (28.69) | 567 (46.44) | 145 (27.72) | |
| Residence | 0.102 | |||||
| Rural | 2779 (87.47) | 973 (88.45) | 314 (89.20) | 1046 (85.67) | 446 (85.28) | |
| Urban | 398 (12.53) | 124 (11.27) | 35 (9.94) | 165 (13.51) | 74 (14.15) | |
| Education | 0.047 | |||||
| No formal education | 1991 (62.30) | 703 (63.91) | 202 (57.39) | 763 (62.49) | 323 (61.76) | |
| < High school | 304 (9.51) | 109 (9.91) | 25 (7.10) | 120 (9.83) | 50 (9.56) | |
| High school and above | 901 (28.19) | 288 (26.18) | 125 (35.51) | 338 (27.68) | 150 (28.68) | |
| Marital | 0.057 | |||||
| Married/partnered | 2893 (90.52) | 992 (90.18) | 316 (89.77) | 1095 (89.68) | 490 (93.69) | |
| Never married/widowed/divorced/separated | 303 (9.48) | 108 (9.82) | 36 (10.23) | 126 (10.32) | 33 (6.31) | |
| Alcohol | < 0.001 | |||||
| Yes | 1939 (60.73) | 549 (49.91) | 237 (67.33) | 782 (64.05) | 371 (70.94) | |
| No | 1254 (39.27) | 551 (50.09) | 115 (32.67) | 437 (35.79) | 151 (28.87) | |
| Smoke | < 0.001 | |||||
| Never | 1000 (31.38) | 469 (42.64) | 68 (19.32) | 357 (29.24) | 106 (20.27) | |
| Former | 212 (6.65) | 79 (7.18) | 13 (3.69) | 98 (8.03) | 22 (4.21) | |
| Current | 1975 (61.97) | 551 (50.09) | 271 (76.99) | 760 (62.24) | 393 (75.14) | |
| Physical activity | < 0.001 | |||||
| Vigorous | 580 (41.94) | 244 (52.14) | 59 (35.54) | 209 (39.73) | 68 (30.49) | |
| Moderate | 419 (30.30) | 115 (24.57) | 64 (38.55) | 166 (31.56) | 74 (33.18) | |
| Inactive | 384 (27.77) | 109 (23.29) | 43 (25.90) | 151 (28.71) | 81 (36.32) | |
| FBG (mg/dL) | 99.28 ± 11.05 | 98.10 ± 11.00 | 98.12 ± 11.36 | 99.72 ± 10.80 | 101.49 ± 11.16 | < 0.001 |
| HbA1c (%) | 5.08 ± 0.38 | 5.08 ± 0.38 | 5.05 ± 0.38 | 5.08 ± 0.37 | 5.08 ± 0.38 | 0.468 |
| TG (mg/dL) | 111.59 ± 58.00 | 72.68 ± 23.92 | 94.59 ± 34.33 | 118.38 ± 45.82 | 189.02 ± 64.67 | < 0.001 |
| TC (mg/dL) | 190.40 ± 36.26 | 186.61 ± 34.02 | 196.22 ± 36.22 | 188.20 ± 37.04 | 199.61 ± 37.08 | < 0.001 |
| HDL (mg/dL) | 52.90 ± 14.15 | 63.18 ± 13.74 | 55.03 ± 11.75 | 47.99 ± 10.06 | 41.32 ± 9.08 | < 0.001 |
| LDL (mg/dL) | 116.45 ± 32.35 | 111.35 ± 29.34 | 124.34 ± 32.51 | 117.67 ± 33.37 | 119.02 ± 34.26 | < 0.001 |
| BUN (mg/dL) | 15.67 ± 4.35 | 16.27 ± 4.45 | 16.52 ± 4.57 | 15.20 ± 4.25 | 14.92 ± 3.96 | < 0.001 |
| Cr (mg/dL) | 0.77 ± 0.18 | 0.78 ± 0.17 | 0.74 ± 0.14 | 0.77 ± 0.20 | 0.75 ± 0.17 | < 0.001 |
| UA (mg/dL) | 4.32 ± 1.18 | 4.32 ± 1.14 | 4.15 ± 1.03 | 4.32 ± 1.23 | 4.45 ± 1.19 | 0.003 |
| Cystatin_C (mg/L) | 1.00 ± 0.23 | 1.02 ± 0.20 | 0.96 ± 0.18 | 1.01 ± 0.28 | 0.94 ± 0.21 | < 0.001 |
| CRP (mg/L) | 2.21 ± 6.91 | 2.12 ± 7.61 | 2.60 ± 7.61 | 2.08 ± 5.07 | 2.45 ± 8.42 | 0.497 |
| WBC (1000 cells/uL) | 6.09 ± 1.79 | 5.92 ± 1.72 | 5.89 ± 1.68 | 6.21 ± 1.81 | 6.30 ± 1.90 | < 0.001 |
| Height (cm) | 1.58 ± 0.09 | 1.59 ± 0.09 | 1.56 ± 0.11 | 1.58 ± 0.08 | 1.57 ± 0.08 | < 0.001 |
| Weight (kg) | 57.78 ± 10.90 | 54.29 ± 9.48 | 57.68 ± 10.14 | 58.72 ± 10.92 | 63.00 ± 11.62 | < 0.001 |
| WC (cm) | 82.76 ± 11.93 | 78.06 ± 10.68 | 82.96 ± 13.24 | 84.23 ± 11.16 | 89.11 ± 11.42 | < 0.001 |
| BMI (kg/m2) | 23.21 ± 8.86 | 21.40 ± 2.89 | 24.78 ± 24.56 | 23.41 ± 3.45 | 25.50 ± 3.93 | < 0.001 |
| BMI category | < 0.001 | |||||
| Underweight | 211 (6.65) | 129 (11.73) | 14 (3.98) | 63 (5.16) | 5 (0.96) | |
| Normal weight | 1476 (46.50) | 690 (62.73) | 149 (42.33) | 514 (42.10) | 123 (23.52) | |
| Overweight | 789 (24.86) | 85 (7.73) | 95 (26.99) | 346 (28.34) | 263 (50.29) | |
| Obesity | 698 (21.99) | 187 (17.00) | 91 (25.85) | 293 (24.00) | 127 (24.28) | |
| Abdominal obesity | < 0.001 | |||||
| Yes | 2137 (66.86) | 963 (87.55) | 218 (61.93) | 761 (62.33) | 195 (37.28) | |
| No | 1059 (33.14) | 137 (12.45) | 134 (38.07) | 460 (37.67) | 328 (62.72) | |
| SBP (mm Hg) | 125.88 ± 19.82 | 122.47 ± 18.46 | 126.18 ± 20.62 | 126.76 ± 19.98 | 130.77 ± 20.48 | < 0.001 |
| DBP (mm Hg) | 73.93 ± 11.87 | 71.40 ± 10.99 | 74.01 ± 11.47 | 74.71 ± 12.04 | 77.38 ± 12.40 | < 0.001 |
| Hypertension | < 0.001 | |||||
| Yes | 2207 (69.06) | 859 (78.09) | 229 (65.06) | 823 (67.40) | 296 (56.60) | |
| No | 989 (30.94) | 241 (21.91) | 123 (34.94) | 398 (32.60) | 227 (43.40) | |
| Comorbidity | 0.203 | |||||
| Yes | 2685 (84.09) | 907 (82.45) | 301 (85.51) | 1026 (84.03) | 451 (86.23) | |
| No | 508 (15.91) | 192 (17.45) | 51 (14.49) | 194 (15.89) | 71 (13.58) | |
| MetSS | 0.61 ± 2.05 | -1.29 ± 0.92 | 0.16 ± 1.09 | 1.12 ± 1.07 | 3.69 ± 1.59 | < 0.001 |
| MetSS (Wave 3) | 0.95 ± 2.09 | -0.85 ± 0.95 | 3.57 ± 1.87 | 0.67 ± 0.93 | 3.58 ± 1.42 | < 0.001 |
| cuMetSS | 2.33 ± 5.57 | -3.20 ± 2.26 | 5.59 ± 4.11 | 2.69 ± 2.14 | 10.91 ± 3.35 | < 0.001 |
| New-onset heart disease | 0.117 | |||||
| Yes | 2615 (86.08) | 928 (84.36) | 282 (80.11) | 982 (80.43) | 423 (80.88) | |
| No | 423 (13.92) | 125 (11.36) | 51 (14.49) | 176 (14.41) | 71 (13.58) | |
| New-onset stroke | 0.004 | |||||
| Yes | 2814 (92.72) | 996 (90.55) | 301 (85.51) | 1066 (87.31) | 451 (86.23) | |
| No | 221 (7.28) | 55 (5.00) | 36 (10.23) | 89 (7.29) | 41 (7.84) | |
| New-onset diabetes | 0.001 | |||||
| Yes | 2808 (92.70) | 998 (90.73) | 308 (87.50) | 1057 (86.57) | 445 (85.09) | |
| No | 221 (7.30) | 50 (4.55) | 27 (7.67) | 97 (7.94) | 47 (8.99) | |
| FCMD | 0.001 | |||||
| Yes | 2546 (79.66) | 916 (83.27) | 265 (75.28) | 961 (78.71) | 404 (77.25) | |
| No | 650 (20.34) | 184 (16.73) | 87 (24.72) | 260 (21.29) | 119 (22.75) | |
| CMM | 0.037 | |||||
| Yes | 3100 (97.00) | 1080 (98.18) | 341 (96.88) | 1175 (96.23) | 504 (96.37) | |
| No | 96 (3.00) | 20 (1.82) | 11 (3.12) | 46 (3.77) | 19 (3.63) | |
Abbreviations: FBG, fasting blood glucose; HbA1c, glycohemoglobin A1c; TG, triglycerides; TC, total cholesterol; HDL, high density lipoprotein cholesterol; LDL, low density lipoprotein cholesterol; BUN, blood urea nitrogen; Cr, creatinine; UA, uric acid; CRP, C-reactive protein; WBC, white blood cell; WC, waist circumference; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; MetSS, metabolic syndrome score; cuMetSS, cumulative metabolic syndrome score; FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity. Data are expressed as mean (standard deviation) for continuous variables or count (percent) for categorical variables. P value was calculated by the t-test, Kruskal-Wallis rank-sum test or the χ2 test, where appropriate
In addition, sensitivity analyses were done using three (K = 3) or five (K = 5) clusters, as well as clusters without DBSCAN exclusion, which illustrated poor performance of clustering. To be more specific, there are 2321 participants in cluster 1 when K = 3 (Figure S3A), account for more than 70% of all participants, indicating poor discrimination. And only 77 participants were classified to cluster 4 when K = 5 (Figure S3B), which were less than 5% of the total participants, suggesting poor statistical significance and clinical applicability. Similarly, few participants were classified to cluster 4 (Figure S3C) or cluster 5 and 6 (Figure S3D) impacted by extreme value without DBSCAN exclusion.
Associations of MetSS with FCMD and CMM
Table 3 presents the association of baseline MetSS and cuMetSS (on both continuous and categorical scales) and risk of FCMD and CMM by Logistic analyses, respectively. As illustrated in Table 3, the risk of CMM both increased with baseline MetSS and cuMetSS increment in all 3 models, while the risk of FCMD showed different pattern.
Table 3.
Association of baseline MetSS and cuMetSS (on both continuous and categorical scales) with FCMD and CMM by using the Logistic regression analyses
| Variables | OR, 95% CI, P | ||||
|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | |||
| FCMD | |||||
| Baseline MetSS | 1.03, 1.00 to 1.06, 0.074 | 1.03, 0.99 to 1.06, 0.104 | 1.01, 0.95 to 1.07, 0.784 | ||
| Baseline MetSS quartile | |||||
| Q1 | Reference group | Reference group | Reference group | ||
| Q2 | 1.22, 0.95 to 1.57, 0.127 | 1.12, 0.86 to 1.45, 0.399 | 1.03, 0.69 to 1.55, 0.871 | ||
| Q3 | 1.18, 1.04 to 1.34, 0.009 | 1.14, 1.00 to 1.30, 0.047 | 0.97, 0.79 to 1.19, 0.744 | ||
| Q4 | 1.11, 1.02 to 1.21, 0.018 | 1.11, 1.02 to 1.22, 0.015 | 1.08, 0.95 to 1.24, 0.239 | ||
| P for trend | 0.024 | 0.024 | 0.025 | ||
| cuMetSS | 1.02, 1.01 to 1.03, 0.006 | 1.02, 1.00 to 1.03, 0.008 | 1.01, 0.99 to 1.03, 0.274 | ||
| cuMetSS quartile | |||||
| Q1 | Reference group | Reference group | Reference group | ||
| Q2 | 1.24, 0.96 to 1.61, 0.096 | 1.15, 0.88 to 1.50, 0.318 | 1.12, 0.73 to 1.72, 0.594 | ||
| Q3 | 1.21, 1.06 to 1.37, 0.004 | 1.17, 1.02 to 1.34, 0.023 | 1.24, 1.00 to 1.53, 0.050 | ||
| Q4 | 1.16, 1.06 to 1.27, 0.001 | 1.16, 1.06 to 1.28, 0.001 | 1.15, 1.00 to 1.33, 0.059 | ||
| P for trend | 0.020 | 0.020 | 0.019 | ||
| CMM | |||||
| Baseline MetSS | 1.13, 1.06 to 1.19, < 0.001 | 1.13, 1.06 to 1.19, < 0.001 | 1.21, 1.09 to 1.34, < 0.001 | ||
| Baseline MetSS quartile | |||||
| Q1 | Reference group | Reference group | Reference group | ||
| Q2 | 1.34, 0.67 to 2.70, 0.408 | 1.26, 0.63 to 2.58, 0.516 | 1.82, 0.55 to 6.57, 0.335 | ||
| Q3 | 1.56, 1.15 to 2.16, 0.005 | 1.47, 1.07 to 2.07, 0.020 | 1.86, 1.11 to 3.37, 0.026 | ||
| Q4 | 1.34, 1.09 to 1.67, 0.007 | 1.64, 1.18 to 2.41, 0.006 | 1.64, 1.18 to 2.41, 0.006 | ||
| P for trend | 0.007 | 0.007 | 0.009 | ||
| cuMetSS | 1.05, 1.02 to 1.07, < 0.001 | 1.05, 1.02 to 1.07, < 0.001 | 1.07, 1.03 to 1.11, < 0.001 | ||
| cuMetSS quartile | |||||
| Q1 | Reference group | Reference group | Reference group | ||
| Q2 | 1.24, 0.63 to 2.48, 0.532 | 1.20, 0.60 to 2.43, 0.611 | 1.92, 0.60 to 6.78, 0.283 | ||
| Q3 | 1.53, 1.13 to 2.11, 0.007 | 1.41, 1.02 to 1.97, 0.041 | 2.00, 1.16 to 3.72, 0.018 | ||
| Q4 | 1.31, 1.06 to 1.63, 0.015 | 1.65, 1.18 to 2.45, 0.006 | 1.65, 1.18 to 2.45, 0.006 | ||
| P for trend | 0.008 | 0.008 | 0.010 | ||
Abbreviations: MetSS, metabolic syndrome score; cuMetSS, cumulative metabolic syndrome score; FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; OR, odds ratio; 95% CI, 95% confidence interval. Under model 1, age and gender were adjusted. Under model 2, education, marital status, residence, drinking status, smoking status, BMI and comorbidity were additionally adjusted. Under model 3, physical activity were additionally adjusted
Baseline MetSS on continuous scale was not significantly associated with FCMD (all P > 0.05). Yet cuMetSS on continuous scale was significantly associated with increased risk of FCMD when adjusted age and gender (model 1: OR, 95% CI, P: 1.02, 1.01 to 1.03, 0.006), additionally adjusted education, marital status, residence, drinking status, smoking status, BMI and comorbidity (model 2: 1.02, 1.00 to 1.03, 0.008), further additional adjustment for physical activity (model 3) yielded no statistical significance (P > 0.05). Baseline MetSS and cuMetSS quartile increments were consistently associated with increased risk of FCMD after multiple adjustment in model 1 and model 2 (both P < 0.05), but no observable significance in model 3 (P > 0.05). RCS (Fig. 3) indicated a linear increase in CMM risk with increasing cuMetSS (P for nonlinear > 0.05) in all 3 models.
Fig. 3.
Restricted cubic spline for the associations of cuMetSS with FCMD and CMM. Under model 1, age and gender were adjusted. Under model 2, education, marital status, residence, drinking status, smoking status, BMI and comorbidity were additionally adjusted. Under model 3, physical activity were additionally adjusted. Abbreviations: FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; cuMetSS, cumulative metabolic syndrome score; OR, odds ratio; 95% CI, 95% confidence interval
In terms of MetSS changes (Table 4), compared with participants in cluster 1 (persistently low level), participants in cluster 4 (persistently high level) had a higher risk of CMM (model 1: 2.38, 1.24 to 4.58, 0.009; model 2: 2.41, 1.25 to 4.64, 0.009 and model 3: 4.17, 1.45 to 12.01, 0.008, respectively) after multiple adjustment. The significance of cluster 3 and cluster 4 with FCMD risk persisted in model 1 and model 2, but not in model 3.
Table 4.
Association of cluster category with FCMD and CMM by using the Logistic regression analyses
| Variables | OR, 95% CI, P | ||||
|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | |||
| FCMD | |||||
| Cluster category | |||||
| Cluster 1 | Reference group | Reference group | Reference group | ||
| Cluster 2 | 1.60, 1.19 to 2.15, 0.002 | 1.63, 1.20 to 2.20, 0.002 | 1.88, 1.19 to 2.98, 0.007 | ||
| Cluster 3 | 1.34, 1.09 to 1.66, 0.006 | 1.35, 1.09 to 1.68, 0.007 | 1.36, 0.96 to 1.92, 0.082 | ||
| Cluster 4 | 1.45, 1.11 to 1.90, 0.006 | 1.48, 1.13 to 1.95, 0.005 | 1.54, 1.00 to 2.37, 0.052 | ||
| CMM | |||||
| Cluster category | |||||
| Cluster 1 | Reference group | Reference group | Reference group | ||
| Cluster 2 | 2.00, 0.94 to 4.29, 0.073 | 1.96, 0.91 to 4.22, 0.086 | 3.94, 1.25 to 12.43, 0.019 | ||
| Cluster 3 | 2.28, 1.33 to 3.89, 0.003 | 2.30, 1.34 to 3.95, 0.003 | 3.40, 1.32 to 8.78, 0.011 | ||
| Cluster 4 | 2.38, 1.24 to 4.58, 0.009 | 2.41, 1.25 to 4.64, 0.009 | 4.17, 1.45 to 12.01, 0.008 | ||
Abbreviations: FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; OR, odds ratio; 95% CI, 95% confidence interval. Under model 1, age and gender were adjusted. Under model 2, education, marital status, residence, drinking status, smoking status, BMI and comorbidity were additionally adjusted. Under model 3, physical activity were additionally adjusted
Sensitivity analyses of MetSS at Wave 3 were similar with that of MetSS at Wave 1, yet ΔMetSS showed no significant association with FCMD or CMM in all models (Table S4). Sensitivity analyses up to Wave 4 showed similar findings (Table S5). Part of ORs were remarkably elevated due to few CMM events (n = 48) with relatively short outcome period.
The prediction performance was obviously improved by adding cuMetSS or clustering in the full model to the basic model (including age, gender, education, marital status, residence, drinking status, smoking status, BMI, physical activity and comorbidity) (Table S6).
Subgroup analyses
To validate the effect of physical activity on the associations of baseline MetSS and cuMetSS with FCMD and CMM risk, subgroup and interaction analyses were subsequently undertaken and results were provided in Table 5. Significance was only noted in subgroups with inactive (model 1: 1.09, 1.02 to 1.15, 0.005; model 2: 1.09, 1.02 to 1.17, 0.008) and vigorous (model 1: 1.10, 1.02 to 1.18, 0.010; model 2: 1.10, 1.02 to 1.18, 0.014) physical activity (model 1: P for interaction = 0.046; model 2: P for interaction = 0.028) on the associations of cuMetSS with FCMD. Similar findings were observed on the associations of baseline MetSS and cuMetSS with CMM. Yet significance was only noted in subgroups with inactive (model 1: 0.89, 0.82 to 0.97, 0.011; model 2: 0.90, 0.82 to 0.98, 0.019) physical activity on the associations of baseline MetSS with FCMD (though P for interaction > 0.05 for the two models).
Table 5.
Subgroup analyses for the modifiable effect of physical activity on the associations of baseline MetSS and cuMetSS with FCMD and CMM risk
| Variables | OR, 95% CI, P (FCMD) | OR, 95% CI, P (CMM) | ||
|---|---|---|---|---|
| Model 1 | Model 2 | Model 1 | Model 2 | |
| Baseline MetSS | ||||
| Physical activity | ||||
| Inactive | 0.89, 0.82 to 0.97, 0.011 | 0.90, 0.82 to 0.98, 0.019 | 0.79, 0.66 to 0.95, 0.014 | 0.78, 0.65 to 0.93, 0.005 |
| Moderate | 1.04, 0.94 to 1.15, 0.446 | 1.03, 0.93 to 1.15, 0.529 | 1.06, 0.82 to 1.38, 0.653 | 1.00, 0.80 to 1.26, 0.978 |
| Vigorous | 0.96, 0.87 to 1.05, 0.349 | 0.95, 0.86 to 1.04, 0.283 | 0.66, 0.52 to 0.83, < 0.001 | 0.65, 0.52 to 0.81, < 0.001 |
| P for interaction | 0.057 | 0.060 | 0.005 | 0.008 |
| cuMetSS | ||||
| Physical activity | ||||
| Inactive | 1.09, 1.02 to 1.15, 0.005 | 1.09, 1.02 to 1.17, 0.008 | 0.92, 0.87 to 0.97, 0.004 | 0.91, 0.84 to 0.97, 0.006 |
| Moderate | 1.00, 0.92 to 1.08, 0.924 | 0.98, 0.89 to 1.08, 0.670 | 1.00, 0.92 to 1.08, 0.984 | 1.02, 0.93 to 1.13, 0.662 |
| Vigorous | 1.10, 1.02 to 1.18, 0.010 | 1.10, 1.02 to 1.18, 0.014 | 0.91, 0.85 to 0.98, 0.010 | 0.92, 0.85 to 0.99, 0.027 |
| P for interaction | 0.046 | 0.028 | 0.056 | 0.026 |
Abbreviations: MetSS, metabolic syndrome score; cuMetSS, cumulative metabolic syndrome score; FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; OR, odds ratio; 95% CI, 95% confidence interval. Under model 1, age and gender were adjusted. Under model 2, education, marital status, residence, drinking status, smoking status, BMI and comorbidity were additionally adjusted
Mediation analyses
Mediation analyses were applied to examine the mediating effects of inflammatory factors (WBC and CRP) on the association of baseline MetSS and cuMetSS with FCMD and CMM. However, no statistical significance was observed (P for indirect effect > 0.05).
Discussion
Based on the longitudinal analysis of data from CHARLS, we find significant associations between cumulative exposure and dynamic change patterns of MetSS with FCMD and CMM. Notably, the association of cuMetSS and MetSS change patterns with FCMD can be modified by moderate physical activity. These results provide evidence for the potential application of cuMetSS and MetSS change patterns as a risk stratification measure to prevent CMM progression among middle-aged and older adults.
Previous investigations have demonstrated that MetSS is able to predict future health conditions [13], and MetSS was significantly associated with cardiometabolic risk [21], such as impaired glucose regulatory status [22], stroke [23], cardiovascular disease [24, 25] and cardiometabolic mortality [26, 27]. However, MetSS was assessed only in baseline or in cross-sectional design.
Further investigations regarded MetSS as a dynamic factor and was assessed in longitudinal design, exploring its trajectory patterns with cardiometabolic risk. For example, a prospective study identified 3 MetSS trajectory patterns of “low”, “medium” and “high”, and found normoglycemic individuals within the high MetSS trajectory pattern had an over seven-fold increased risk of diabetes [28]. Similarly, 4 MetSS trajectory subgroups were clustered in another study and participants with “highly fluctuating elevated” MetSS exhibited significantly higher stroke risk compared with those with “stable low” MetSS [29].
In present study, we also longitudinally accessed MetSS trajectory and classified 4 clusters of MetSS, but the tendency of the MetSS changes were different from past investigations [28, 29]. Our cluster results reflected 4 change patterns including “persistently low level” (cluster 1), “median level increase” (cluster 2), “median level decrease” (cluster 3) and “persistently high level” (cluster 4). Consistent with past investigations, the highest cluster presented the worst risk of CMM. Whereas physical activity can modify the association of MetSS trajectory cluster 3 and cluster 4 with FCMD, but not modify the association of cluster 2 with FCMD, suggesting median level increase of MetSS is even more harmful than persistently high level of MetSS.
The modifiable effect of physical activity or exercise to MetS severity have been verified by trials [30, 31] and observational and Mendelian randomization study [32], while also improving mental health and work ability. Optimal exercise was recommended for enhancing healthy longevity in aging populations [33], and the use of exercise was emphasized in CMD prevention [34]. Our results revealed that early investigation of moderate physical activity was optimal to prevent CMM progression in middle-aged and older adults, accordant with previous report [35]. It also indicated that dynamic MetSS changes were more sensitive to the modifiable factor than static MetSS status in early stage of CMM.
The underlying mechanisms between MetSS and CMM still remain far from fully understood. Low-grade systematic inflammation may play an important role [36–38]. It is widely accepted that elevated inflammation increased the risk of incident diabetes and cardiovascular events [39–41]. However, no significant difference of CRP and WBC were found between participants with CMD-free, FCMD and CMM and mediation analyses showed no significance. The possible reason may be limited types of inflammatory biomarkers, more sensitive inflammatory biomarkers like high-sensitivity CRP or neutrophil-lymphocyte ratio were not available in CHARLS, further mechanism research can be expand to those inflammatory biomarkers.
Strengths and limitations
Strengths of our study include a longitudinal design exploring not only static status of MetSS, but also dynamic MetSS changes in a large-scale nationally representative population. The application of advanced statistical modeling techniques, such as K-means clustering, enabled the classification of distinct trends of MetSS changes. Additionally, our study provides evidence for the early identification of high-risk CMM individuals to prevent CMM progression in middle-aged and older adults.
Some limitations should be acknowledged when interpreting our findings. Firstly, blood examination were adopted only in Wave 1 and Wave 3 (two time points) in CHARLS, which may not fully capture the cumulative exposure of MetSS and lead to misclassification of MetSS changes. And the association between MetSS changes and CMM progression could be influenced by the limited temporal resolution. Secondly, diagnosis of stroke and heart disease relies on self-reported health condition in CHARLS and diagnosis of diabetes was also based on self-reported status in outcome period, which may result in reporting biases. Thirdly, our study was based on middle-aged and older Chinese adults, so the results may not generalizable to the other populations. Fourthly, as cluster was a categorical variable (participants were divided into 4 groups), further subgroup analyses (participants were further divided into 12 groups) resulted in inadequate sample size in each group (especially for Cluster 2 with 352 participants). The number of participants in each group was insufficient for robust statistical inference, so subgroup analyses for clusters were not undertaken. Fifthly, given that CMM events were relatively inadequate (n = 101), we regarded it as an endpoint of progression, further study with longer follow-up period is necessary to validate the findings of our study. Finally, CMM is a complex multimorbidity affected by numerous factors, there still can be confounding factors that are not considered in our study.
Conclusions
Taken together, our findings indicate that cumulative exposure and dynamic change patterns of MetSS were associated with FCMD and CMM, yet there is a modifiable effect of physical activity on the associations of cuMetSS and MetSS change patterns with FCMD risk. Dynamic screenings targeting MetSS changes and early identification of CMM risk are needed to prevent CMM progression among middle-aged and older adults. Further study to explore underlying mechanisms is warranted.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors are deeply grateful to all the participants and staff of the CHARLS.
Abbreviations
- MetSS
Metabolic syndrome score
- cuMetSS
cumulative metabolic syndrome score
- CMM
Cardiometabolic multimorbidity
- FCMD
First cardiometabolic diseases
- CHARLS
China Health and Retirement Longitudinal Study
- BMI
Body mass index
- SBP
Systolic blood pressure
- DBP
Diastolic blood pressure
- HbA1c
Glycohemoglobin A1c
- TG
Triglycerides
- TC
Total cholesterol
- HDL
High density lipoprotein cholesterol
- LDL
Low-density lipoprotein cholesterol
- BUN
Blood urea nitrogen
- Cr
Creatinine
- UA
Uric acid
- WBC
White blood cells
- CRP
C-reactive protein
- OR
Odds ratio
- 95% CI
95% confidence interval
Author contributions
Y.Y., A.D. and L.W. planned and designed the study and directed its implementation. Y.Y., H.Z. and Z.S. did the data preparation and quality control. Y.Y., X.D. and J.C. conducted statistical analyses. Y.Y., J.Z. and Q.X. wrote the manuscript. All authors read and approved the final manuscript prior to submission.
Funding
The study was supported by Shandong Province Traditional Chinese Medicine Science and Technology Project (Q-2023116, M-2022047 and M-20240617) and Jointly-built Science and Technology Project of the Science and Technology Department of the National Administration of Traditional Chinese Medicine (GZY-KJS-SD-2023-018).
Data availability
The datasets are available via the official website at https://charls.pku.edu.cn/en/.
Declarations
Ethics approval and consent to participate
The Institutional Review Board of Peking University approves CHARLS survey protocols in accordance with the Declaration of Helsinki. All participants provided written informed consent in CHARLS.
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.
Jingjing Zhang and Xiaoyu Ding contributed equally to this work.
Contributor Information
Anye Du, Email: anyee4359@163.com.
Yuan Yuan, Email: yyparttime@163.com.
References
- 1.Emerging Risk Factors C, Di Angelantonio E, Kaptoge S, Wormser D, Willeit P, Butterworth AS, Bansal N, O’Keeffe LM, Gao P, Wood AM, et al. Association of Cardiometabolic Multimorbidity With Mortality. JAMA. 2015;314(1):52–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Han Y, Hu Y, Yu C, Guo Y, Pei P, Yang L, Chen Y, Du H, Sun D, Pang Y, et al. Lifestyle, cardiometabolic disease, and multimorbidity in a prospective Chinese study. Eur Heart J. 2021;42(34):3374–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fan J, Sun Z, Yu C, Guo Y, Pei P, Yang L, Chen Y, Du H, Sun D, Pang Y, et al. Multimorbidity patterns and association with mortality in 0.5 million Chinese adults. Chin Med J (Engl). 2022;135(6):648–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jin Y, Liang J, Hong C, Liang R, Luo Y. Cardiometabolic multimorbidity, lifestyle behaviours, and cognitive function: a multicohort study. Lancet Healthy Longev. 2023;4(6):e265–73. [DOI] [PubMed] [Google Scholar]
- 5.Liu C, Liu R, Tian N, Fa W, Liu K, Wang N, Zhu M, Liang X, Ma Y, Ren Y, et al. Cardiometabolic multimorbidity, peripheral biomarkers, and dementia in rural older adults: The MIND-China study. Alzheimers Dement. 2024;20(9):6133–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yang W, Li W, Wang S, Qi X, Sun Z, Dove A, Xu W. Association of cardiometabolic multimorbidity with risk of late-life depression: a nationwide twin study. Eur Psychiatry. 2024;67(1):e58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zhang Y, Zhou Y, Kivimaki M, Holt-Lunstad J, Carrillo-Larco RM, Xu X. Cardiometabolic multimorbidity, social activity, and joint trajectories of physical disability, depressive symptom, and cognitive function in mid-to-late life: a multicohort study. BMC Med. 2025;23(1):577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lee S, Ahn C, Abe SK, Rahman MS, Islam MR, Saito E, An S, Sawada N, Shu XO, Koh WP, et al. Association Between Cardiometabolic Multimorbidity and 15-year Mortality in the Asia Cohort Consortium. J Epidemiol. 2025;35(7):321–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yang S, Yu B, Yu W, Dai S, Feng C, Shao Y, Zhao X, Li X, He T, Jia P. Development and validation of an age-sex-ethnicity-specific metabolic syndrome score in the Chinese adults. Nat Commun. 2023;14(1):6988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gurka MJ, Lilly CL, Oliver MN, DeBoer MD. An examination of sex and racial/ethnic differences in the metabolic syndrome among adults: a confirmatory factor analysis and a resulting continuous severity score. Metabolism. 2014;63(2):218–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yang Y, Li Q, Long Y, Yuan J, Zha Y. Associations of metabolic syndrome, its severity with cognitive impairment among hemodialysis patients. Diabetol Metab Syndr. 2023;15(1):108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Winn M, Karra P, Haaland B, Doherty JA, Summers SA, Litchman ML, Gunter MJ, Playdon MC, Hardikar S. Metabolic dysfunction and obesity-related cancer: Results from the cross-sectional National Health and Nutrition Examination Survey. Cancer Med. 2023;12(1):606–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wu M, Shu Y, Wang L, Song L, Chen S, Liu Y, Bi J, Li D, Yang Y, Hu Y, et al. Metabolic syndrome severity score and the progression of CKD. Eur J Clin Invest. 2022;52(1):e13646. [DOI] [PubMed] [Google Scholar]
- 14.DeBoer MD, Filipp SL, Gurka MJ. Use of a Metabolic Syndrome Severity Z Score to Track Risk During Treatment of Prediabetes: An Analysis of the Diabetes Prevention Program. Diabetes Care. 2018;41(11):2421–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Gui MH, Ling Y, Liu L, Jiang JJ, Li XY, Gao X. Effect of Metabolic Syndrome Score, Metabolic Syndrome, and Its Individual Components on the Prevalence and Severity of Angiographic Coronary Artery Disease. Chin Med J (Engl). 2017;130(6):669–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gurka MJ, Golden SH, Musani SK, Sims M, Vishnu A, Guo Y, Cardel M, Pearson TA, DeBoer MD. Independent associations between a metabolic syndrome severity score and future diabetes by sex and race: the Atherosclerosis Risk In Communities Study and Jackson Heart Study. Diabetologia. 2017;60(7):1261–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zeng P, Li M, Cao J, Zeng L, Jiang C, Lin F. Association of metabolic syndrome severity with frailty progression among Chinese middle and old-aged adults: a longitudinal study. Cardiovasc Diabetol. 2024;23(1):302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Deng L, Liu T, Liu CA, Zhang Q, Song MM, Lin SQ, Wang YM, Zhang QS, Shi HP. The association of metabolic syndrome score trajectory patterns with risk of all cancer types. Cancer. 2024;130(12):2150–9. [DOI] [PubMed] [Google Scholar]
- 19.Yang Y, Liu A. Associations of cumulative exposure and dynamic trajectories of the C-reactive protein-triglyceride-glucose index with incident stroke in middle-aged and older Chinese adults: a longitudinal analysis based on CHARLS. Cardiovasc Diabetol. 2025;24(1):386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Huang X, Wen S, Huang Y, Huang Z. Gender differences in the association between changes in the atherogenic index of plasma and cardiometabolic diseases: a cohort study. Lipids Health Dis. 2024;23(1):135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Liu Y, Xin Y, Wang W, Feng W, Zhang L, Zhao Y, Zhu Y. The association between metabolic syndrome severity and frailty risk in patients with rheumatoid arthritis: a cross-sectional study. Arthritis Res Ther. 2025;27(1):145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Merry TL, Metcalf P, Scragg R, Gearry R, Foster M, Krebs JD. Metabolic syndrome severity score (MetSSS) associates with metabolic health status in multi-ethnic Aotearoa New Zealand cohorts. Diabetes Res Clin Pract. 2022;192:110088. [DOI] [PubMed] [Google Scholar]
- 23.Jang YN, Lee JH, Moon JS, Kang DR, Park SY, Cho J, Kim JY, Huh JH. Metabolic Syndrome Severity Score for Predicting Cardiovascular Events: A Nationwide Population-Based Study from Korea. Diabetes Metab J. 2021;45(4):569–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lin JJ, Dai PY, Zhang J, Guan YQ, Gong WW, Yu M, Fang L, Hu RY, He QF, Li N, et al. Association between metabolic syndrome severity score and cardiovascular disease: results from a longitudinal cohort study on Chinese adults. Front Endocrinol (Lausanne). 2024;15:1341546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gurka MJ, Guo Y, Filipp SL, DeBoer MD. Metabolic syndrome severity is significantly associated with future coronary heart disease in Type 2 diabetes. Cardiovasc Diabetol. 2018;17(1):17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Schlesinger N, Elsaid MI, Rustgi VK. The relationship between metabolic syndrome severity and the risk of mortality in gout patients: a population-based study. Clin Exp Rheumatol. 2022;40(3):631–3. [DOI] [PubMed] [Google Scholar]
- 27.Honarvar M, Mehran L, Masoumi S, Agahi S, Khalili S, Azizi F, Amouzegar A. Independent association between age- and sex-specific metabolic syndrome severity score and cardiovascular disease and mortality. Sci Rep. 2023;13(1):14621. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Amouzegar A, Honarvar M, Masoumi S, Khalili D, Azizi F, Mehran L. Trajectory patterns of metabolic syndrome severity score and risk of type 2 diabetes. J Transl Med. 2023;21(1):750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Li Q, Liu Y, Gao X, Fang Q, Zeng L, Xu Y, Huang J. Dynamic Changes in Metabolic Syndrome Scores and New-Onset Stroke Risk in Middle-Aged and Older Adults: A Nationwide Prospective Cohort Study in China Aligned With Predictive, Preventive, and Personalized Medicine. J Am Heart Assoc. 2025;14(22):e041833. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Gallardo-Alfaro L, Bibiloni MDM, Bouzas C, Mascaro CM, Martinez-Gonzalez MA, Salas-Salvado J, Corella D, Schroder H, Martinez JA, Alonso-Gomez AM, et al. Physical activity and metabolic syndrome severity among older adults at cardiovascular risk: 1-Year trends. Nutr Metab Cardiovasc Dis. 2021;31(10):2870–86. [DOI] [PubMed] [Google Scholar]
- 31.Haufe S, Kerling A, Protte G, Bayerle P, Stenner HT, Rolff S, Sundermeier T, Kuck M, Ensslen R, Nachbar L, et al. Telemonitoring-supported exercise training, metabolic syndrome severity, and work ability in company employees: a randomised controlled trial. Lancet Public Health. 2019;4(7):e343–52. [DOI] [PubMed] [Google Scholar]
- 32.Wang S, Wu J, Liu R, Zhao Q, Feng Y, Zhao L, Zhang Y, Jiao X, Tarimo CS, Wu J, et al. Association between physical activity and sedentary behavior with cardiometabolic multimorbidity in the elderly hypertensive population: an observational and Mendelian randomization study. Psychol Sport Exerc. 2025;79:102869. [DOI] [PubMed] [Google Scholar]
- 33.Izquierdo M, de Souto Barreto P, Arai H, Bischoff-Ferrari HA, Cadore EL, Cesari M, Chen LK, Coen PM, Courneya KS, Duque G, et al. Global consensus on optimal exercise recommendations for enhancing healthy longevity in older adults (ICFSR). J Nutr Health Aging. 2025;29(1):100401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Wang J, Zheng Y, Jiang Y, Suo C, Zhang T, Chen X, Xu K. Association between physical activity-related metabolic signature and cardiometabolic diseases and multimorbidity: A cohort study from UK biobank. Prev Med. 2025;191:108211. [DOI] [PubMed] [Google Scholar]
- 35.Liu Y, Yang Y, Wu H, Yang H, Chen L, Sun F, Xia Y. Intensity-specific physical activity measured by accelerometer and the risk of mortality among individuals with cardiometabolic diseases: A prospective study from the UK Biobank. Int J Nurs Stud. 2024;156:104786. [DOI] [PubMed] [Google Scholar]
- 36.Ladeiras-Lopes R, Teixeira P, Azevedo A, Leite-Moreira A, Bettencourt N, Fontes-Carvalho R. Metabolic syndrome severity score is associated with diastolic dysfunction and low-grade inflammation in a community-based cohort. Eur J Prev Cardiol. 2020;27(19):2330–3. [DOI] [PubMed] [Google Scholar]
- 37.Zhao L, Hu H, Zhang L, Liu Z, Huang Y, Liu Q, Jin L, Zhu M, Zhang L. Inflammation in diabetes complications: molecular mechanisms and therapeutic interventions. MedComm (2020). 2024;5(4):e516. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ferrucci L, Fabbri E. Inflammageing: chronic inflammation in ageing, cardiovascular disease, and frailty. Nat Rev Cardiol. 2018;15(9):505–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Nie Y, Zhou H, Wang J, Kan H. Association between systemic immune-inflammation index and diabetes: a population-based study from the NHANES. Front Endocrinol (Lausanne). 2023;14:1245199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Cheng W, Bu X, Xu C, Wen G, Kong F, Pan H, Yang S, Chen S. Higher systemic immune-inflammation index and systemic inflammation response index levels are associated with stroke prevalence in the asthmatic population: a cross-sectional analysis of the NHANES 1999–2018. Front Immunol. 2023;14:1191130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ridker PM, Bhatt DL, Pradhan AD, Glynn RJ, MacFadyen JG, Nissen SE, Prominent R-I, Investigators S. Inflammation and cholesterol as predictors of cardiovascular events among patients receiving statin therapy: a collaborative analysis of three randomised trials. Lancet. 2023;401(10384):1293–301. [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 datasets are available via the official website at https://charls.pku.edu.cn/en/.