Comparative Evaluation of Obesity-Related Indices for Predicting Incident Hypertension: Evidence from Chinese and UK Longitudinal Cohorts With Machine Learning Interpretation

Comparative Evaluation of Obesity-Related Indices for Predicting Incident Hypertension: Evidence from Chinese and UK Longitudinal Cohorts With Machine Learning Interpretation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Comparative Evaluation of Obesity-Related Indices for Predicting Incident Hypertension: Evidence from Chinese and UK Longitudinal Cohorts With Machine Learning Interpretation Chunqiang Gu, Dongmei Tang, Fang Yuan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8650506/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 22 Apr, 2026 Read the published version in BMC Cardiovascular Disorders → Version 1 posted 15 You are reading this latest preprint version Abstract Background Hypertension remains a major global health burden, with excess adiposity serving as a key modifiable contributor to its development. However, conventional anthropometric measures, particularly body mass index (BMI), inadequately reflect metabolically harmful fat accumulation. Consequently, the predictive value of emerging obesity-related indices for incident hypertension remains incompletely defined. Methods We systematically evaluated six obesity-related indices—BMI, visceral fat index (VFI), triglyceride–glucose index (TyG), TyG–BMI (TGB), TyG–waist-to-height ratio (TGW), and lipid accumulation product (LA)—in relation to new-onset hypertension using data from two prospective cohorts, CHARLS and ELSA. Cox proportional hazards models, restricted cubic spline (RCS) analyses, and interpretable machine-learning methods were applied to assess associations, nonlinear patterns, and relative predictor importance. Results In both cohorts, all six indices were significantly associated with incident hypertension in univariate analyses, with graded risk increases across quartiles. After mutual adjustment for all indices and covariates, VFI remained the only predictor consistently associated with hypertension risk in both CHARLS and ELSA. RCS analyses identified nonlinear associations for VFI, TGW, and LA in CHARLS, whereas relationships in ELSA were largely monotonic. Machine-learning models showed good discrimination (AUC 0.756 in CHARLS; 0.878 in ELSA), and SHAP analysis consistently ranked VFI, LA, and TGW as the most influential predictors. Conclusion Overall, VFI and related composite adiposity indices, particularly LA and TGW, outperform BMI and isolated metabolic markers in predicting incident hypertension. Population-specific nonlinear patterns highlight the heterogeneity of obesity phenotypes and the limitations of BMI-based risk assessment. Incorporation of these indices into routine screening may improve early identification of individuals at elevated risk, including those with metabolically unhealthy normal weight. Dual Cohort CHARLS ELSA Longitudinal Study Obesity Indices Machine Learning SHAP Analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Hypertension remains the leading modifiable risk factor for cardiovascular disease (CVD) and premature mortality worldwide, posing a substantial burden on public health systems 1 , 2 . As the global population ages, the prevalence of hypertension has surged, particularly among middle-aged and elderly individuals, where arterial stiffening and metabolic dysregulation are prevalent 3 . Chronic elevation of blood pressure (BP) is a primary antecedent for life-threatening complications, including stroke, myocardial infarction, heart failure, and chronic kidney disease 4 , 5 . Although pharmacological interventions have advanced, BP control rates remain suboptimal in many populations due to the asymptomatic nature of the condition and the complexity of its underlying pathophysiology 6 . Therefore, identifying high-risk individuals through easily accessible, non-invasive predictors before the onset of irreversible vascular damage is critical for implementing early preventive strategies and reducing the global burden of hypertension 7 . Obesity is widely recognized as a major driver of hypertension, accounting for 65–75% of the risk for essential hypertension 8 . The accumulation of adipose tissue, particularly visceral fat, triggers a cascade of pathophysiological mechanisms, including sympathetic nervous system overactivation, stimulation of the renin-angiotensin-aldosterone system (RAAS), and induction of systemic inflammation and insulin resistance 9 . Traditionally, Body Mass Index (BMI) has been the gold standard for assessing obesity; however, its limitations are increasingly apparent. BMI fails to distinguish between lean mass and fat mass, nor does it reflect the distribution of body fat—a critical factor since central (visceral) adiposity is more metabolically active and pathogenic than subcutaneous fat 10 , 11 . To address these shortcomings, novel composite indices reflecting visceral adiposity and insulin resistance—such as the Visceral Fat Index (VFI), Triglyceride-Glucose Index (TyG), Triglyceride–Glucose–BMI (TGB), TyG–waist-to-height ratio (TGW) and Lipid Accumulation Product (LA)—have been proposed 12 , 13 . These indicators integrate anthropometric measurements with lipid profiles, potentially offering a more nuanced assessment of the "metabolic obesity" phenotype that drives hypertension risk 14 . Despite the growing repertoire of obesity-related indices, significant gaps remain in the current landscape of research. First, the majority of existing studies have relied predominantly on conventional linear regression models, which often fail to accurately capture the complex non-linear interactions and threshold effects inherent in biological systems 15 . Second, previous literature has largely been confined to single indicators or limited combinations within single-center populations, lacking a systematic comparison of multiple novel indices across diverse ethnic cohorts 16 , 17 . The comparative predictive utility of purely anthropometric versus lipid-combined indices remains a subject of debate, and it is yet to be fully elucidated whether the "obesity paradox"—a phenomenon frequently observed in cardiovascular outcomes—applies to the incidence of hypertension when using these refined markers 18 . Furthermore, while machine learning (ML) excels in handling high-dimensional data and identifying non-linear features, its application in quantifying the contribution weights and ranking the importance of these specific obesity indices remains relatively scarce 19 . To bridge these knowledge gaps, this study leverages data from two large-scale, longitudinal cohorts: The China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA). Our primary objective is to systematically evaluate and compare the predictive performance of six obesity-related indices (BMI, VFI, TyG, TGB, TGW, and LA) for incident hypertension by integrating a multi-modal machine learning approach with rigorous survival analysis. By elucidating the non-linear dose-response relationships and validating findings across distinct East Asian and European populations, this study aims to identify the most robust metabolic predictors to guide personalized risk stratification and hypertension prevention in aging populations 20 . Materials and Methods Data Source and Study Population This study utilized data from two nationally representative prospective cohorts—the CHARLS and the ELSA—to evaluate the longitudinal association between obesity-related indices and the risk of incident hypertension in middle-aged and older adults. CHARLS participants were enrolled at the 2011 baseline and followed through 2018, while ELSA participants were enrolled at the 2004 baseline and followed through 2021. Both cohorts employed standardized protocols to collect detailed information on demographics, lifestyle behaviors, anthropometric measurements, biochemical markers, and health conditions. Individuals aged 45 years or older with complete baseline anthropometric and metabolic data were eligible for inclusion, whereas those with physician-diagnosed hypertension at baseline or with missing key covariates were excluded. After applying these criteria, 5,070 participants from CHARLS and 1,065 participants from ELSA were included in the longitudinal analyses, forming the primary analysis cohort and external validation cohort, respectively. Definition of Outcome The primary outcome of this study was incident hypertension identified during follow-up in each cohort. Hypertension status was determined using a standardized self-reported question administered at every survey wave: “Has a doctor ever told you that you have hypertension?” Participants who answered “yes” at any follow-up wave were classified as having developed hypertension. Individuals who reported hypertension at baseline were excluded to ensure that only new-onset cases were captured. For participants without hypertension throughout follow-up, the censoring time was defined as the date of their last available survey. Time to event was calculated as the interval between the baseline interview and the first wave in which hypertension was reported. Covariate Assessment To control for potential confounding, a comprehensive set of covariates covering demographic, lifestyle, and clinical factors was extracted and harmonized. Demographic Variables: Gender: Categorized as Male or Female. Age: Treated as a continuous variable. Education Level: Education was classified into four categories: Not completed primary school, Primary school, Middle school, and High school and above. Lifestyle Factors: Smoking Status (self-reported): Categorized as "Yes" or "No". Drinking Status (self-reported): Categorized as "Yes" or "No". Biochemical Indicators: Key biomarkers included C-reactive Protein (CRP), Total Cholesterol (TC), and Glycated Hemoglobin (HbA1c), which were analyzed as continuous variables. Comorbidities: History of chronic conditions was assessed based on self-reports, including Diabetes, Stroke, Cancer, and Arthritis. Each condition was coded as a binary variable (Yes/No). Assessment of Obesity-Related Indices Anthropometric measurements, including height, weight, and waist circumference (WC), were collected by trained staff using standard protocols. WC was measured at the midpoint between the lower rib margin and the iliac crest. Fasting blood samples were analyzed to obtain triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and fasting plasma glucose (FPG). Based on these measurements, six obesity-related indices were calculated. BMI was calculated as weight in kilograms divided by the square of height in meters (kg/m 2 ). VFI was estimated using gender-specific equations derived from Chinese populations: For men: $$\:VFI\:=\:-267.93\:+\:0.68·Age\:+\:0.03·BMI\:+\:4.00·WC\left(cm\right)+\:22.00·\text{ln}\left[TG\left(\frac{mmol}{L}\right)\right]-\:16.32·HDL-C\left(\frac{mmol}{L}\right)$$ For women: $$\:VFI\:=\:-187.32\:+\:1.71·Age\:+\:4.23·BMI\:+\:1.12·WC\left(cm\right)+\:39.76·\text{ln}\left[TG\left(\frac{mmol}{L}\right)\right]\--\:11.66·HDL-C\left(\frac{mmol}{L}\right)$$ LA was calculated to assess lipid overaccumulation, using the following formulas: For men: $$\:LA\:=\:\left(WC\left(cm\right)-\:65\right)\times\:\:TG\left(\frac{mmol}{L}\right)$$ For women: $$\:LA\:=\:\left(WC\left(cm\right)-\:58\right)\times\:\:TG\left(\frac{mmol}{L}\right)$$ Three composite metabolic indices were also determined: TyG index: $$\:\text{ln}\left[\frac{TG\left(\frac{mg}{dL}\right)\times\:\:FPG\left(\frac{mg}{dL}\right)}{2}\right].$$ TGB Index: $$\:TyG\:\times\:\:BMI$$ TGW Index: $$\:TyG\:\times\:\:WHtR$$ Waist-to-height ratio (WHtR) was calculated as waist circumference (cm) divided by height (cm). Statistical Analysis All statistical computations were executed using R software (version 4.2.2). A two-tailed P-value of < 0.05 was established as the threshold for statistical significance. Descriptive Statistics and Group Comparisons: Baseline characteristics were stratified by hypertension outcome. We utilized the tableone package to present continuous variables as means with standard deviations (SD) and categorical variables as frequencies (%). Differences between groups were evaluated using Student’s t-tests or Mann-Whitney U tests for continuous data, and Chi-square tests for categorical distributions. Machine Learning and Feature Importance Interpretation: Nine machine learning algorithms were screened using the caret package to identify key predictors, with model performance evaluated by the Area Under the Curve (AUC). The optimal model (Gradient Boosting) was subsequently interpreted using SHAP values via the shapviz package to quantify the global and local contributions of obesity features. Survival Analysis and Cox Regression: To identify predictors of incident hypertension, Cox proportional hazards models were applied. Univariate models were first used to evaluate the crude associations of each obesity-related index with hypertension risk. Multivariable Cox models were then constructed to adjust for demographic, lifestyle, biochemical, and clinical covariates. Proportional hazards assumptions were verified using Schoenfeld residuals. Non-linear Dose-Response and Threshold Analysis: To move beyond linear assumptions, we utilized Restricted Cubic Splines (RCS) with the rms package to model the potential non-linear dose-response relationship between continuous obesity indices and hypertension risk. Subgroup and Interaction Analysis: Stratified analyses were performed to evaluate the stability of the associations across different demographic subgroups (age, gender, smoking, and drinking status). Interaction terms were introduced into the Cox models to statistically test whether the predictive effects of obesity indices were modified by these subgroup variables. Results Baseline Characteristics of Study Participants A total of 5,070 participants from the CHARLS cohort (2011–2018) and 1,065 participants from the ELSA cohort (2004–2021) were included after applying the inclusion and exclusion criteria. Baseline characteristics of participants in the CHARLS and ELSA cohorts are summarized in Table 1 and Table 2 , respectively. Significant differences were observed for six obesity-related indices (BMI, VFI, TyG, TGB, TGW, and LA) across both datasets ( p < 0.05). In the CHARLS cohort, age, education, total cholesterol, HbA1c, diabetes, and arthritis were also significantly associated with hypertension, while in ELSA, age, education, and CRP showed significant group differences. Table 1 Baseline characteristics of participants with and without hypertension in the CHARLS cohort (2011–2018) level Health Control Hypertension p n 3595 1475 Gender (%) Female 1925 (53.5) 769 (52.1) 0.377 Male 1670 (46.5) 706 (47.9) Age (mean (SD)) 57.93 (8.38) 60.27 (8.91) < 0.001 Edu (%) Not completed primary school 1756 (48.8) 803 (54.4) < 0.001 primary school 785 (21.8) 328 (22.2) middle school 702 (19.5) 229 (15.5) high school and above 352 (9.8) 115 (7.8) Smoke (%) No 2455 (68.3) 1001 (67.9) 0.793 Yes 1140 (31.7) 474 (32.1) Drink (%) No 2363 (65.7) 960 (65.1) 0.684 Yes 1232 (34.3) 515 (34.9) CRP (mean (SD)) 2.31 (7.36) 2.44 (5.44) 0.56 TC (mean (SD)) 191.24 (37.11) 196.12 (41.43) < 0.001 HbA1c (mean (SD)) 5.19 (0.66) 5.33 (0.89) < 0.001 Diabetes (%) No 3474 (96.6) 1394 (94.5) 0.001 Yes 121 (3.4) 81 (5.5) Stroke (%) No 3550 (98.7) 1453 (98.5) 0.587 Yes 45 (1.3) 22 (1.5) Cancer (%) No 3570 (99.3) 1463 (99.2) 0.789 Yes 25 (0.7) 12 (0.8) Arthritis (%) No 2419 (67.3) 936 (63.5) 0.01 Yes 1176 (32.7) 539 (36.5) BMI (%) Q1 974 (27.1) 294 (19.9) < 0.001 Q2 944 (26.3) 323 (21.9) Q3 899 (25.0) 368 (24.9) Q4 778 (21.6) 490 (33.2) VFI (%) Q1 1013 (28.2) 255 (17.3) < 0.001 Q2 961 (26.7) 306 (20.7) Q3 898 (25.0) 369 (25.0) Q4 723 (20.1) 545 (36.9) TyG (%) Q1 969 (27.0) 299 (20.3) < 0.001 Q2 924 (25.7) 343 (23.3) Q3 887 (24.7) 380 (25.8) Q4 815 (22.7) 453 (30.7) TGB (%) Q1 993 (27.6) 275 (18.6) < 0.001 Q2 943 (26.2) 324 (22.0) Q3 903 (25.1) 364 (24.7) Q4 756 (21.0) 512 (34.7) TGW (%) Q1 997 (27.7) 271 (18.4) < 0.001 Q2 953 (26.5) 314 (21.3) Q3 901 (25.1) 366 (24.8) Q4 744 (20.7) 524 (35.5) LA (%) Q1 972 (27.0) 296 (20.1) < 0.001 Q2 960 (26.7) 307 (20.8) Q3 904 (25.1) 363 (24.6) Q4 759 (21.1) 509 (34.5) Table 2 Baseline characteristics of participants with and without hypertension in the ELSA cohort (2004–2021) level Health Control Hypertension p n 624 441 Gender (%) Female 341 (54.6) 230 (52.2) 0.459 Male 283 (45.4) 211 (47.8) Age (mean (SD)) 61.00 (6.23) 63.22 (6.81) < 0.001 Edu (%) Not yet in high school 180 (28.8) 174 (39.5) 0.003 High school graduation 155 (24.8) 90 (20.4) Junior college 171 (27.4) 99 (22.4) University degree or above 118 (18.9) 78 (17.7) Smoke (%) No 540 (86.5) 385 (87.3) 0.786 Yes 84 (13.5) 56 (12.7) Drink (%) No 31 (5.0) 35 (7.9) 0.064 Yes 593 (95.0) 406 (92.1) CRP (mean (SD)) 2.56 (3.99) 3.51 (5.10) 0.001 TC (mean (SD)) 6.15 (1.17) 6.22 (1.15) 0.341 HbA1c (mean (SD)) 5.42 (0.47) 5.46 (0.42) 0.143 Diabetes (%) No 617 (98.9) 436 (98.9) 1 Yes 7 (1.1) 5 (1.1) Stroke (%) No 622 (99.7) 439 (99.5) 1 Yes 2 (0.3) 2 (0.5) Cancer (%) No 597 (95.7) 417 (94.6) 0.488 Yes 27 (4.3) 24 (5.4) Arthritis (%) No 457 (73.2) 311 (70.5) 0.366 Yes 167 (26.8) 130 (29.5) BMI (%) Q1 177 (28.4) 90 (20.4) < 0.001 Q2 174 (27.9) 92 (20.9) Q3 142 (22.8) 124 (28.1) Q4 131 (21.0) 135 (30.6) VFI (%) Q1 196 (31.4) 71 (16.1) < 0.001 Q2 166 (26.6) 100 (22.7) Q3 132 (21.2) 134 (30.4) Q4 130 (20.8) 136 (30.8) TyG (%) Q1 180 (28.8) 90 (20.4) 0.002 Q2 164 (26.3) 103 (23.4) Q3 139 (22.3) 123 (27.9) Q4 141 (22.6) 125 (28.3) TGB (%) Q1 186 (29.8) 81 (18.4) < 0.001 Q2 166 (26.6) 100 (22.7) Q3 139 (22.3) 127 (28.8) Q4 133 (21.3) 133 (30.2) TGW (%) Q1 196 (31.4) 71 (16.1) < 0.001 Q2 156 (25.0) 110 (24.9) Q3 140 (22.4) 126 (28.6) Q4 132 (21.2) 134 (30.4) LA (%) Q1 189 (30.3) 78 (17.7) < 0.001 Q2 150 (24.0) 116 (26.3) Q3 152 (24.4) 114 (25.9) Q4 133 (21.3) 133 (30.2) Machine Learning Performance and SHAP-Based Feature Importance Analysis Nine machine learning algorithms, including Gradient Boosting, SVM, Logistic Regression, Partial Least Squares, and Neural Network, were applied to both cohorts. Gradient Boosting exhibited the best classification performance, with AUC = 0.756 in CHARLS and AUC = 0.878 in ELSA (Fig. 1 ). SHAP value analysis ranked the relative contribution of variables to hypertension prediction. In CHARLS, the VFI was the most influential predictor, followed by LA and TGW ( Fig. 2 a, c ) . In ELSA, LA ranked first in importance, followed by VFI and TGW ( Fig. 2 b, d ) . These findings suggest that the predictive contribution of specific obesity indicators may differ by population and regional metabolic profiles. In both cohorts, VFI, LA, and TGW consistently ranked among the top predictors, underscoring their substantial contribution to hypertension development and their potential value as key obesity-related indicators in risk stratification. (a) Mean absolute SHAP values for all variables in the CHARLS model. (b) Mean absolute SHAP values for all variables in the ELSA model. (c) SHAP summary plot for the CHARLS model, showing the direction and magnitude of each feature’s impact on hypertension prediction. (d) SHAP summary plot for the ELSA model, indicating the relative influence and distribution of feature effects on model output. Cumulative Incidence of Hypertension Across Quartiles of Obesity-Related Indices To further evaluate the longitudinal association between obesity-related indices and the risk of developing hypertension, Kaplan–Meier survival analysis was conducted. In the CHARLS cohort ( Figure S1 ), distinct risk stratification was observed across quartiles for all six indices over the 7-year follow-up. The survival curves displayed a clear stepwise separation, where participants in the highest quartiles (Q4) consistently exhibited the steepest decline in hypertension-free probability compared to those in the lowest quartiles (Q1), with log-rank tests revealing highly significant differences for all indices (p < 0.0001). This graded dose-response pattern was robustly replicated in the ELSA cohort ( Figure S2 ), despite the longer follow-up period of 12 years. Participants with higher baseline levels of obesity indices showed a significantly accelerated onset of hypertension. While the majority of indices (BMI, VFI, TGB, TGW, and LA) maintained highly significant associations ( p < 0.0001), TyG showed a slightly less pronounced but still statistically significant separation ( p = 0.0017). These findings collectively indicate that elevated obesity-related indices are strong, long-term predictors of hypertension development across different populations and follow-up durations. Association Between Obesity-Related Indices and Hypertension Risk in Cox Regression Models To further quantify the relationship between obesity-related indices and hypertension risk over time, Cox proportional hazards models were applied. Univariate Cox proportional hazards models demonstrated that higher quartiles of all six indices (BMI, VFI, TyG, TGB, TGW, and LA) were significantly associated with an elevated risk of hypertension in both datasets (HR > 1, p < 0.05; Fig. 3 ). After adjusting for potential confounders, multivariable Cox regression showed that VFI remained independently associated with hypertension risk ( p < 0.05) in both datasets, while other indices (such as BMI and TGW) largely lost their statistical significance (Fig. 4 ). This phenomenon suggests that VFI captures the core risk information more effectively than other markers when evaluated concurrently, and the inverse trend of LA likely reflects statistical suppression due to the inclusion of multiple correlated metabolic indicators. Nonlinear Dose–Response Relationships Between Obesity Indices and Hypertension Risk To further explore potential nonlinear associations and threshold effects between obesity-related indices and hypertension development, RCS models were applied. The RCS models revealed distinct dose-response patterns across the two cohorts. In the CHARLS cohort, BMI, VFI, TGB, TGW, and LA exhibited significant nonlinear associations ( p 1), similar to the risk observed at higher values, whereas the lowest risk was found in the intermediate range. In contrast, lower values of BMI, TGB, and TyG were consistently associated with reduced hypertension risk (protective effects). In the ELSA cohort, while VFI, TGB, TGW, and LA also displayed statistically significant nonlinear relationships ( p < 0.05) and BMI and TyG remained linear, the overall patterns differed from CHARLS. All six indices in ELSA generally showed a monotonic increasing trend, where hazard ratios progressively increased with higher values, confirming a robust positive dose–response association without the paradoxical high risk at lower levels observed in CHARLS ( Fig. S3 ). Subgroup Analyses of Obesity-Related Indices and Hypertension Risk in the ELSA Cohort In the CHARLS cohort, the positive association between obesity indices and hypertension risk remained generally consistent across strata, though specific heterogeneities were observed (Fig. 6 ). Significant interactions were detected for BMI and TGB across age groups ( p < 0.05), with a more pronounced risk gradient observed in younger participants ( 55 years). Furthermore, LA and VFI exhibited significant interactions with gender ( p 0.05), suggesting that their associations with hypertension are stable regardless of demographic or behavioral characteristics. In the ELSA cohort, the associations between obesity-related indices and hypertension risk were largely consistent across most subgroups ( Fig. S4 ). Significant interactions were detected for BMI across gender strata ( p = 0.049), indicating that the predictive effect of BMI on hypertension risk was stronger in women than in men. For the remaining indices, including TGB, LA, VFI, TGW, and TyG, no significant interactions were observed across gender, age, smoking, or drinking subgroups ( p > 0.05), suggesting stable associations regardless of demographic or behavioral characteristics. Within each subgroup, higher quartiles of obesity indices were generally associated with increased hazard ratios for hypertension, reinforcing their robustness as predictors across populations in the ELSA cohort. Forest plots display HRs with 95% CI for each quartile of six obesity indices—(a) BMI, (b) TGB, (c) LA, (d) VFI, (e) TGW, and (f) TyG—stratified by gender, age, smoking, and drinking status. Discussion This study systematically evaluated the predictive performance of six distinct obesity-related indices for incident hypertension, leveraging a robust dual-cohort design (CHARLS and ELSA) and integrating advanced machine learning with traditional survival analysis. Our findings confirm that sophisticated metabolic and body composition indices—particularly those incorporating visceral fat and insulin resistance markers—outperform simple anthropometric measures in forecasting long-term hypertension risk among middle-aged and older adults. The core finding of this study is the consistent, superior predictive performance of VFI, LA, and TGW across both Chinese (CHARLS) and British (ELSA) cohorts. Both the SHAP analysis derived from the Gradient Boosting model and the multivariate Cox regression independently ranked VFI, LA, and TGW among the most potent risk factors, even after extensive adjustment for established confounders. This convergence of evidence highlights the biological primacy of visceral adiposity (VFI), lipotoxicity (LA), and the combined insulin resistance and central obesity (TGW) in driving hypertension onset. Conversely, classical measures like BMI and the isolated metabolic index TyG showed relatively weaker predictive power when competing against these composite markers in multivariate and ML models. This is consistent with recent literature suggesting that BMI, due to its inability to differentiate between metabolically harmful visceral fat and benign subcutaneous or lean mass, dilutes the true risk signal 21 . VFI and LA, by specifically incorporating factors like WC, TG, and HDL-C, capture the essence of ectopic fat deposition and dysregulated adipokine secretion, mechanisms known to directly activate the sympathetic nervous system and the RAAS, leading to sustained blood pressure elevation 9 , 22 . The TGW index further strengthens this prediction by integrating insulin resistance (TyG) with central fat distribution (WHtR), emphasizing the synergistic effect of impaired glucose and lipid metabolism on vascular stiffness and endothelial dysfunction 23 . Our RCS analysis provided critical insights into the dose-response relationship, revealing significant non-linearity for most indices. Notably, the shape of the risk curve showed crucial differences between the CHARLS (Chinese) and ELSA (European) cohorts, underscoring the influence of ethnicity and lifestyle on metabolic phenotypes. In the ELSA cohort, indices like VFI, LA, and TGW generally demonstrated a monotonic, continuous risk increase, suggesting that for this older European population, any level of metabolic deterioration poses a higher risk compared to the reference low group. However, in the CHARLS cohort, indices like VFI, LA, and TGW exhibited a more complex, often J- or U-shaped pattern at the lower end of the distribution. For example, some indices showed a seemingly protective effect (HR 1) even at lower values. This finding may reflect the phenomenon of "lean metabolically unhealthy" or specific clinical cutoffs for Asian populations, where lower absolute levels of visceral fat or lipids, coupled with other undiagnosed inflammatory conditions, might still confer a disproportionately high risk for incident disease compared to Western cohorts 24 . These cross-population differences suggest that single, universal threshold guidelines for these complex indices may be insufficient for global clinical application. The concurrent use of both machine learning and Cox proportional hazards models proved to be mutually reinforcing. ML, represented by the high AUC achieved by the Gradient Boosting model (CHARLS AUC = 0.756; ELSA AUC = 0.878), effectively validated the overall predictive accuracy of our feature set. Crucially, the SHAP framework provided transparency, definitively ranking VFI, LA, and TGW as the most influential features. This ranking strongly aligns with the independent hazard ratio estimates derived from the multivariable Cox model, where these same indices maintained significant and highest HRs. This convergence reinforces that the identified composite indices are not merely statistically correlated with hypertension but represent robust, high-impact prognostic factors. Furthermore, the ML approach successfully incorporated non-linear effects and complex feature interactions that traditional Cox models, particularly when treating variables linearly, might underestimate 25 . Our findings carry significant implications for hypertension prevention and management. Firstly, the emphasis on VFI, LA, and TGW suggests that clinical practice should move beyond BMI and prioritize screening strategies that directly assess metabolic obesity and visceral fat accumulation. Simple calculation of LA and TGW, which require only routine blood tests (TG, FPG) and basic anthropometrics (WC, height), can easily be integrated into community health screenings to identify the "metabolically unhealthy normal weight" individual—a high-risk group often missed by BMI-centric approaches 26 . Secondly, the observed interaction between LA/VFI and gender (stronger effect in women in CHARLS) suggests the need for gender-specific risk stratification, potentially reflecting the distinct hormonal and fat distribution patterns in older women 27 . The identification of population-specific non-linear thresholds further advocates for customized, targeted intervention strategies rather than a one-size-fits-all approach to weight and metabolic management. The primary strength of this study lies in its rigorous methodology, combining a dual-cohort longitudinal design for external validation (CHARLS to ELSA) with a sophisticated multi-modal analytical framework (ML + Cox + RCS). This integration provides highly consistent and robust evidence across diverse ethnic and environmental backgrounds. However, several limitations warrant consideration. First, hypertension status was defined by self-report in the final wave, which may introduce misclassification bias compared to direct clinical measurement, though this definition is standard in large cohort studies. Second, while we controlled for numerous established confounders, residual confounding from unmeasured factors (e.g., dietary intake, physical activity intensity) cannot be entirely excluded. Finally, the VFI and TGW indices rely on imputed calculations rather than direct imaging, representing surrogates for true visceral fat mass. Conclusion VFI, LA, and TGW emerge as superior and consistent predictors of incident hypertension compared to traditional measures across both Chinese and European older adults. By utilizing these easily calculated composite indices, clinicians and public health practitioners can significantly enhance the precision of risk stratification, allowing for timely, targeted interventions focused on visceral adiposity and insulin resistance to combat the rising tide of hypertension. Declarations Ethics approval and consent to participate This study was conducted using fully de-identified, publicly available data from CHARLS and ELSA. Both CHARLS and ELSA obtained ethical approval from their respective institutional review boards, and all participants provided informed consent at the time of original data collection. Because the present analysis involved secondary use of anonymized data and did not include any direct interaction with human subjects, additional ethical approval was not required in accordance with national and institutional regulations, including the Measures for the Ethical Review of Life Science and Medical Research Involving Human Subjects (2023, China). Consent for publication Not applicable. Clinical trial number not applicable. Competing interests The authors declare no competing interests. Funding None. Author Contribution Yuan Fang : Conceptualization; Study design; Writing – review & editing; Chunqiang Gu : Conceptualization; Data collection; Data curation; Formal analysis; Interpretation of results; Writing – review & editing; Dongmei Tang : Formal analysis; Data interpretation; Writing – review & editing; All authors : Contributed to manuscript writing, critically revised the manuscript, and approved the final version. Data Availability The data analyzed in this study were obtained from two publicly accessible longitudinal cohorts. CHARLS data are available through the Peking University Open Research Data Platform (http://charls.pku.edu.cn/) upon registration and adherence to data-use guidelines. ELSA data can be accessed through the UK Data Service (https://ukdataservice.ac.uk/) under standard end-user licensing. All derived variables and analytic code used to generate the results are available from the corresponding author upon reasonable request. The curated datasets derived from CHARLS and ELSA that were used for the present analyses are provided in Supplementary File 1 and Supplementary File 2, respectively. These files contain the cleaned variables required to reproduce the main analyses reported in this study. References Worldwide trends in. hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants. Lancet Sep. 2021;11(10304):957–80. 10.1016/s0140-6736(21)01330-1 . Zhou B, Perel P, Mensah GA, Ezzati M. Global epidemiology, health burden and effective interventions for elevated blood pressure and hypertension. Nat Rev Cardiol Nov. 2021;18(11):785–802. 10.1038/s41569-021-00559-8 . Mills KT, Stefanescu A, He J. The global epidemiology of hypertension. Nat Rev Nephrol Apr. 2020;16(4):223–37. 10.1038/s41581-019-0244-2 . Carey RM, Whelton PK, Prevention. Detection, Evaluation, and Management of High Blood Pressure in Adults: Synopsis of the 2017 American College of Cardiology/American Heart Association Hypertension Guideline. Ann Intern Med . Mar 6 2018;168(5):351–358. 10.7326/m17-3203 Roth GA, Mensah GA, Johnson CO, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990–2019: Update From the GBD 2019 Study. J Am Coll Cardiol Dec. 2020;22(25):2982–3021. 10.1016/j.jacc.2020.11.010 . Carey RM, Muntner P, Bosworth HB, Whelton PK. Reprint of: Prevention and Control of Hypertension: JACC Health Promotion Series. J Am Coll Cardiol. Dec 2018;11(23 Pt B):2996–3011. 10.1016/j.jacc.2018.10.022 . Abe M, Hirata T, Morito N, et al. Smartphone application-based interventions for cardiometabolic risk factor management: A systematic review and meta-analysis. Hypertens Res Sep. 2025;3. 10.1038/s41440-025-02365-y . Hall JE, do Carmo JM, da Silva AA, Wang Z, Hall ME. Obesity, kidney dysfunction and hypertension: mechanistic links. Nat Rev Nephrol Jun. 2019;15(6):367–85. 10.1038/s41581-019-0145-4 . Parvanova A, Reseghetti E, Abbate M, Ruggenenti P. Mechanisms and treatment of obesity-related hypertension-Part 1: Mechanisms. Clin Kidney J Jan. 2024;17(1):sfad282. 10.1093/ckj/sfad282 . Dobre MZ, Virgolici B, Timnea O. Key Roles of Brown, Subcutaneous, and Visceral Adipose Tissues in Obesity and Insulin Resistance. Curr Issues Mol Biol May. 2025;9(5). 10.3390/cimb47050343 . Stefan N. Causes, consequences, and treatment of metabolically unhealthy fat distribution. Lancet Diabetes Endocrinol Jul. 2020;8(7):616–27. 10.1016/s2213-8587(20)30110-8 . Papathanasiou KA, Roussos CE, Armylagos S, Rallidis SL, Rallidis LS. Lipid Accumulation Product Is Predictive of Cardiovascular Hospitalizations among Patients with Stable Ischemic Heart Disease: Long-Term Follow-Up of the LAERTES Study. J Cardiovasc Dev Dis Oct. 2024;10(10). 10.3390/jcdd11100316 . Xu AR, Jin Q, Shen Z, Zhang J, Fu Q. Association between the risk of hypertension and triglyceride glucose index in Chinese regions: a systematic review and dose-response meta-analysis of a regional update. Front Cardiovasc Med. 2023;10:1242035. 10.3389/fcvm.2023.1242035 . Ren X, Chen M, Lian L, et al. The triglyceride-glucose index is associated with a higher risk of hypertension: evidence from a cross-sectional study of Chinese adults and meta-analysis of epidemiology studies. Front Endocrinol (Lausanne). 2025;16:1516328. 10.3389/fendo.2025.1516328 . Zhang X, Ye R, Sun L, et al. Relationship between novel anthropometric indices and the incidence of hypertension in Chinese individuals: a prospective cohort study based on the CHNS from 1993 to 2015. BMC Public Health Mar. 2023;6(1):436. 10.1186/s12889-023-15208-7 . Mao L, Lin L, Shi Z, Song H, Zhao H, Xu X. Determinants and prediction of hypertension among Chinese middle-aged and elderly adults with diabetes: A machine learning approach. Heliyon Sep. 2024;30(18):e38124. 10.1016/j.heliyon.2024.e38124 . Zhao G, Zhou Z. Correlation between obesity-related indices and hypertension. Am J Transl Res. 2024;16(8):3842–50. 10.62347/uufg4260 . Thakker J, Khaliq I, Ardeshna NS, Lavie CJ, Oktay AA. The Obesity Paradox of Cardiovascular Outcomes in Patients with Diabetes Mellitus. Curr Diab Rep Jun. 2025;7(1):35. 10.1007/s11892-025-01592-4 . Silva GFS, Fagundes TP, Teixeira BC, Chiavegatto Filho ADP. Machine Learning for Hypertension Prediction: a Systematic Review. Curr Hypertens Rep Nov. 2022;24(11):523–33. 10.1007/s11906-022-01212-6 . Feng X, Zhu J, Hua Z, Yao S, Tong H. Comparison of obesity indicators for predicting cardiovascular risk factors and multimorbidity among the Chinese population based on ROC analysis. Sci Rep Sep. 2024;9(1):20942. 10.1038/s41598-024-71914-1 . Ofstad AP, Sommer C, Birkeland KI, et al. Comparison of the associations between non-traditional and traditional indices of adiposity and cardiovascular mortality: an observational study of one million person-years of follow-up. Int J Obes (Lond). May 2019;43(5):1082–92. 10.1038/s41366-019-0353-9 . Jung JY, Ryoo JH, Oh CM, et al. Visceral adiposity index and longitudinal risk of incident metabolic syndrome: Korean genome and epidemiology study (KoGES). Endocr J Jan. 2020;28(1):45–52. 10.1507/endocrj.EJ19-0008 . Chen Y, Hu P, He Y, Qin H, Hu L, Yang R. Association of TyG index and central obesity with hypertension in middle-aged and elderly Chinese adults: a prospective cohort study. Sci Rep Jan. 2024;26(1):2235. 10.1038/s41598-024-52342-7 . Huang Y, Zhou Y, Xu Y, et al. Inflammatory markers link triglyceride-glucose index and obesity indicators with adverse cardiovascular events in patients with hypertension: insights from three cohorts. Cardiovasc Diabetol Jan. 2025;8(1):11. 10.1186/s12933-024-02571-x . Hwang SH, Lee H, Lee JH, et al. Machine Learning-Based Prediction for Incident Hypertension Based on Regular Health Checkup Data: Derivation and Validation in 2 Independent Nationwide Cohorts in South Korea and Japan. J Med Internet Res Nov. 2024;5:26:e52794. 10.2196/52794 . Kim B, Taniguchi K, Isobe T, Oh S. Triglyceride-glucose index is capable of identifying metabolically obese, normal-weight older individuals. J Physiol Anthropol Feb. 2024;3(1):8. 10.1186/s40101-024-00355-6 . Kaneva AM, Bojko ER. Sex differences in the association between obesity and hypertension. Arch Physiol Biochem Jun. 2023;129(3):682–9. 10.1080/13813455.2020.1861027 . Additional Declarations No competing interests reported. Supplementary Files supplementaryfile1.csv supplementaryfile2.csv Cite Share Download PDF Status: Published Journal Publication published 22 Apr, 2026 Read the published version in BMC Cardiovascular Disorders → Version 1 posted Editorial decision: Revision requested 20 Mar, 2026 Reviewers agreed at journal 18 Mar, 2026 Reviews received at journal 15 Mar, 2026 Reviewers agreed at journal 14 Mar, 2026 Reviewers agreed at journal 13 Mar, 2026 Reviewers agreed at journal 01 Mar, 2026 Reviews received at journal 19 Feb, 2026 Reviewers agreed at journal 15 Feb, 2026 Reviews received at journal 10 Feb, 2026 Reviewers agreed at journal 10 Feb, 2026 Reviewers invited by journal 09 Feb, 2026 Editor invited by journal 29 Jan, 2026 Editor assigned by journal 28 Jan, 2026 Submission checks completed at journal 28 Jan, 2026 First submitted to journal 20 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8650506","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":589404332,"identity":"e84be6fb-a996-4480-86d4-4f51214bbce4","order_by":0,"name":"Chunqiang Gu","email":"","orcid":"","institution":"The First Hospital of Jiaxing, Affiliated Hospital of Jiaxing University","correspondingAuthor":false,"prefix":"","firstName":"Chunqiang","middleName":"","lastName":"Gu","suffix":""},{"id":589404333,"identity":"c825a2a8-65f2-4d35-a8bd-37237d2e980c","order_by":1,"name":"Dongmei Tang","email":"","orcid":"","institution":"The First Hospital of Jiaxing, Affiliated Hospital of Jiaxing University","correspondingAuthor":false,"prefix":"","firstName":"Dongmei","middleName":"","lastName":"Tang","suffix":""},{"id":589404334,"identity":"7221126a-4b4d-447d-9ac2-0d099fd22ca6","order_by":2,"name":"Fang Yuan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2klEQVRIiWNgGAWjYBACNvnDBx98+GEjZ9/e2PggoaKGsBY+CbZkw5k9acYGPIebDR6cOUZYi5wEj5k0D9uhRAOJ9DbJhy3MRDhMui3ZcAbPgQRzhsS2isQGNgb+9u4E/FpkQH6xuJNn2XCw7UbiDhkGiTNnN+DXwpAGsuVZMcPBRqCWM2wMBhK5hLTkgPxyOLHhMGNbQWIbMxFaJKBaNhxjbGMgTgvPMUggS/YwNksknDnGQ9Av8u3NkKjkl3/+8OOPiho5/vZe/FowAA9pykfBKBgFo2AUYAUAiT9QBWrQYrsAAAAASUVORK5CYII=","orcid":"","institution":"Jiaxing Maternity and Child Health Care Hospital","correspondingAuthor":true,"prefix":"","firstName":"Fang","middleName":"","lastName":"Yuan","suffix":""}],"badges":[],"createdAt":"2026-01-20 15:14:55","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8650506/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8650506/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12872-026-05885-8","type":"published","date":"2026-04-22T15:56:58+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":102545187,"identity":"70e44ffa-dce9-4a50-b44e-c9bfe1c251c6","added_by":"auto","created_at":"2026-02-12 20:36:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":28338,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC curve comparison of nine machine learning models for predicting incident hypertension in the CHARLS and ELSA cohorts.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/c61b69e44501945916b835a6.png"},{"id":102746774,"identity":"877e2be7-0b12-48d3-a062-a1d75395cc56","added_by":"auto","created_at":"2026-02-16 09:01:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":78985,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSHAP-based feature importance analysis of machine learning models predicting incident hypertension in the CHARLS and ELSA cohorts.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Mean absolute SHAP values for all variables in the CHARLS model. (b) Mean absolute SHAP values for all variables in the ELSA model. (c) SHAP summary plot for the CHARLS model, showing the direction and magnitude of each feature’s impact on hypertension prediction. (d) SHAP summary plot for the ELSA model, indicating the relative influence and distribution of feature effects on model output.\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/229fd76b311342231e0a9bd0.png"},{"id":102747255,"identity":"d6c0395f-3db3-44ae-a792-42a00b32d0aa","added_by":"auto","created_at":"2026-02-16 09:04:18","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":72750,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eForest plots of univariate Cox proportional hazards models for the association between obesity-related indices and incident hypertension in the CHARLS (a) and ELSA (b) cohorts.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/fb7721402193f969732cffb3.png"},{"id":102545188,"identity":"917f4edc-f590-44eb-8938-135e3c473fc8","added_by":"auto","created_at":"2026-02-12 20:36:23","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":62666,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMultivariable Cox proportional hazards models assessing the independent associations between obesity-related indices and incident hypertension in the CHARLS (a) and ELSA (b) cohorts.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/c06ee10d590f6b407e53f2c3.png"},{"id":102545183,"identity":"d6a24e4c-b310-4972-99df-f2fbaed28248","added_by":"auto","created_at":"2026-02-12 20:36:23","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":55594,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRCS analysis showing nonlinear associations between obesity-related indices and hypertension risk in the CHARLS.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/485327918d5efb74a5d9e13e.png"},{"id":102746934,"identity":"bdab3ef2-eb9f-4fb6-9d04-e2a8a02beb31","added_by":"auto","created_at":"2026-02-16 09:03:00","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":89906,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSubgroup analyses of the associations between obesity-related indices and hypertension risk in the CHARLS cohort.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eForest plots display HRs with 95% CI for each quartile of six obesity indices—(a) BMI, (b) TGB, (c) LA, (d) VFI, (e) TGW, and (f) TyG—stratified by gender, age, smoking, and drinking status.\u003c/p\u003e","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/53f6e1a5dc6a609f32463e68.png"},{"id":107927659,"identity":"260fa651-859d-497a-8d5e-0559d55a0372","added_by":"auto","created_at":"2026-04-27 16:00:40","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1120501,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/6a8b1a84-3c1d-45da-9760-15801b670de6.pdf"},{"id":102746982,"identity":"edf344c5-e9ee-46ab-a73f-91b2619af084","added_by":"auto","created_at":"2026-02-16 09:03:20","extension":"csv","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":987990,"visible":true,"origin":"","legend":"","description":"","filename":"supplementaryfile1.csv","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/8d7fcd5df0d891ecab407a1e.csv"},{"id":102746751,"identity":"4b6ac106-7692-4a59-a316-e4529242b9c0","added_by":"auto","created_at":"2026-02-16 09:01:01","extension":"csv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":159695,"visible":true,"origin":"","legend":"","description":"","filename":"supplementaryfile2.csv","url":"https://assets-eu.researchsquare.com/files/rs-8650506/v1/237c620e1226d793da4d1883.csv"}],"financialInterests":"No competing interests reported.","formattedTitle":"Comparative Evaluation of Obesity-Related Indices for Predicting Incident Hypertension: Evidence from Chinese and UK Longitudinal Cohorts With Machine Learning Interpretation","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHypertension remains the leading modifiable risk factor for cardiovascular disease (CVD) and premature mortality worldwide, posing a substantial burden on public health systems\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. As the global population ages, the prevalence of hypertension has surged, particularly among middle-aged and elderly individuals, where arterial stiffening and metabolic dysregulation are prevalent\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Chronic elevation of blood pressure (BP) is a primary antecedent for life-threatening complications, including stroke, myocardial infarction, heart failure, and chronic kidney disease\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Although pharmacological interventions have advanced, BP control rates remain suboptimal in many populations due to the asymptomatic nature of the condition and the complexity of its underlying pathophysiology\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. Therefore, identifying high-risk individuals through easily accessible, non-invasive predictors before the onset of irreversible vascular damage is critical for implementing early preventive strategies and reducing the global burden of hypertension\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eObesity is widely recognized as a major driver of hypertension, accounting for 65\u0026ndash;75% of the risk for essential hypertension\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. The accumulation of adipose tissue, particularly visceral fat, triggers a cascade of pathophysiological mechanisms, including sympathetic nervous system overactivation, stimulation of the renin-angiotensin-aldosterone system (RAAS), and induction of systemic inflammation and insulin resistance\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Traditionally, Body Mass Index (BMI) has been the gold standard for assessing obesity; however, its limitations are increasingly apparent. BMI fails to distinguish between lean mass and fat mass, nor does it reflect the distribution of body fat\u0026mdash;a critical factor since central (visceral) adiposity is more metabolically active and pathogenic than subcutaneous fat\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. To address these shortcomings, novel composite indices reflecting visceral adiposity and insulin resistance\u0026mdash;such as the Visceral Fat Index (VFI), Triglyceride-Glucose Index (TyG), Triglyceride\u0026ndash;Glucose\u0026ndash;BMI (TGB), TyG\u0026ndash;waist-to-height ratio (TGW) and Lipid Accumulation Product (LA)\u0026mdash;have been proposed\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. These indicators integrate anthropometric measurements with lipid profiles, potentially offering a more nuanced assessment of the \"metabolic obesity\" phenotype that drives hypertension risk\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eDespite the growing repertoire of obesity-related indices, significant gaps remain in the current landscape of research. First, the majority of existing studies have relied predominantly on conventional linear regression models, which often fail to accurately capture the complex non-linear interactions and threshold effects inherent in biological systems\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. Second, previous literature has largely been confined to single indicators or limited combinations within single-center populations, lacking a systematic comparison of multiple novel indices across diverse ethnic cohorts\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. The comparative predictive utility of purely anthropometric versus lipid-combined indices remains a subject of debate, and it is yet to be fully elucidated whether the \"obesity paradox\"\u0026mdash;a phenomenon frequently observed in cardiovascular outcomes\u0026mdash;applies to the incidence of hypertension when using these refined markers\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Furthermore, while machine learning (ML) excels in handling high-dimensional data and identifying non-linear features, its application in quantifying the contribution weights and ranking the importance of these specific obesity indices remains relatively scarce\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo bridge these knowledge gaps, this study leverages data from two large-scale, longitudinal cohorts: The China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA). Our primary objective is to systematically evaluate and compare the predictive performance of six obesity-related indices (BMI, VFI, TyG, TGB, TGW, and LA) for incident hypertension by integrating a multi-modal machine learning approach with rigorous survival analysis. By elucidating the non-linear dose-response relationships and validating findings across distinct East Asian and European populations, this study aims to identify the most robust metabolic predictors to guide personalized risk stratification and hypertension prevention in aging populations\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData Source and Study Population\u003c/h2\u003e \u003cp\u003eThis study utilized data from two nationally representative prospective cohorts\u0026mdash;the CHARLS and the ELSA\u0026mdash;to evaluate the longitudinal association between obesity-related indices and the risk of incident hypertension in middle-aged and older adults. CHARLS participants were enrolled at the 2011 baseline and followed through 2018, while ELSA participants were enrolled at the 2004 baseline and followed through 2021. Both cohorts employed standardized protocols to collect detailed information on demographics, lifestyle behaviors, anthropometric measurements, biochemical markers, and health conditions. Individuals aged 45 years or older with complete baseline anthropometric and metabolic data were eligible for inclusion, whereas those with physician-diagnosed hypertension at baseline or with missing key covariates were excluded. After applying these criteria, 5,070 participants from CHARLS and 1,065 participants from ELSA were included in the longitudinal analyses, forming the primary analysis cohort and external validation cohort, respectively.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDefinition of Outcome\u003c/h3\u003e\n\u003cp\u003eThe primary outcome of this study was incident hypertension identified during follow-up in each cohort. Hypertension status was determined using a standardized self-reported question administered at every survey wave: \u0026ldquo;Has a doctor ever told you that you have hypertension?\u0026rdquo; Participants who answered \u0026ldquo;yes\u0026rdquo; at any follow-up wave were classified as having developed hypertension. Individuals who reported hypertension at baseline were excluded to ensure that only new-onset cases were captured. For participants without hypertension throughout follow-up, the censoring time was defined as the date of their last available survey. Time to event was calculated as the interval between the baseline interview and the first wave in which hypertension was reported.\u003c/p\u003e\n\u003ch3\u003eCovariate Assessment\u003c/h3\u003e\n\u003cp\u003eTo control for potential confounding, a comprehensive set of covariates covering demographic, lifestyle, and clinical factors was extracted and harmonized.\u003c/p\u003e \u003cp\u003eDemographic Variables:\u003c/p\u003e \u003cp\u003eGender: Categorized as Male or Female.\u003c/p\u003e \u003cp\u003eAge: Treated as a continuous variable.\u003c/p\u003e \u003cp\u003eEducation Level:\u003c/p\u003e \u003cp\u003eEducation was classified into four categories: Not completed primary school, Primary school, Middle school, and High school and above.\u003c/p\u003e \u003cp\u003eLifestyle Factors:\u003c/p\u003e \u003cp\u003eSmoking Status (self-reported): Categorized as \"Yes\" or \"No\".\u003c/p\u003e \u003cp\u003eDrinking Status (self-reported): Categorized as \"Yes\" or \"No\".\u003c/p\u003e \u003cp\u003eBiochemical Indicators:\u003c/p\u003e \u003cp\u003eKey biomarkers included C-reactive Protein (CRP), Total Cholesterol (TC), and Glycated Hemoglobin (HbA1c), which were analyzed as continuous variables.\u003c/p\u003e \u003cp\u003eComorbidities:\u003c/p\u003e \u003cp\u003eHistory of chronic conditions was assessed based on self-reports, including Diabetes, Stroke, Cancer, and Arthritis. Each condition was coded as a binary variable (Yes/No).\u003c/p\u003e\n\u003ch3\u003eAssessment of Obesity-Related Indices\u003c/h3\u003e\n\u003cp\u003eAnthropometric measurements, including height, weight, and waist circumference (WC), were collected by trained staff using standard protocols. WC was measured at the midpoint between the lower rib margin and the iliac crest. Fasting blood samples were analyzed to obtain triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and fasting plasma glucose (FPG).\u003c/p\u003e \u003cp\u003eBased on these measurements, six obesity-related indices were calculated. BMI was calculated as weight in kilograms divided by the square of height in meters (kg/m\u003csup\u003e2\u003c/sup\u003e).\u003c/p\u003e \u003cp\u003eVFI was estimated using gender-specific equations derived from Chinese populations:\u003c/p\u003e \u003cp\u003eFor men:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:VFI\\:=\\:-267.93\\:+\\:0.68\u0026middot;Age\\:+\\:0.03\u0026middot;BMI\\:+\\:4.00\u0026middot;WC\\left(cm\\right)+\\:22.00\u0026middot;\\text{ln}\\left[TG\\left(\\frac{mmol}{L}\\right)\\right]-\\:16.32\u0026middot;HDL-C\\left(\\frac{mmol}{L}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eFor women:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:VFI\\:=\\:-187.32\\:+\\:1.71\u0026middot;Age\\:+\\:4.23\u0026middot;BMI\\:+\\:1.12\u0026middot;WC\\left(cm\\right)+\\:39.76\u0026middot;\\text{ln}\\left[TG\\left(\\frac{mmol}{L}\\right)\\right]\\--\\:11.66\u0026middot;HDL-C\\left(\\frac{mmol}{L}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eLA was calculated to assess lipid overaccumulation, using the following formulas:\u003c/p\u003e \u003cp\u003eFor men:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:LA\\:=\\:\\left(WC\\left(cm\\right)-\\:65\\right)\\times\\:\\:TG\\left(\\frac{mmol}{L}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eFor women:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:LA\\:=\\:\\left(WC\\left(cm\\right)-\\:58\\right)\\times\\:\\:TG\\left(\\frac{mmol}{L}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThree composite metabolic indices were also determined:\u003c/p\u003e \u003cp\u003eTyG index:\u003cdiv id=\"Eque\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Eque\" name=\"EquationSource\"\u003e\n$$\\:\\text{ln}\\left[\\frac{TG\\left(\\frac{mg}{dL}\\right)\\times\\:\\:FPG\\left(\\frac{mg}{dL}\\right)}{2}\\right].$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eTGB Index:\u003cdiv id=\"Equf\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equf\" name=\"EquationSource\"\u003e\n$$\\:TyG\\:\\times\\:\\:BMI$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eTGW Index:\u003cdiv id=\"Equg\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equg\" name=\"EquationSource\"\u003e\n$$\\:TyG\\:\\times\\:\\:WHtR$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWaist-to-height ratio (WHtR) was calculated as waist circumference (cm) divided by height (cm).\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eAll statistical computations were executed using R software (version 4.2.2). A two-tailed P-value of \u0026lt;\u0026thinsp;0.05 was established as the threshold for statistical significance.\u003c/p\u003e \u003cp\u003eDescriptive Statistics and Group Comparisons: Baseline characteristics were stratified by hypertension outcome. We utilized the tableone package to present continuous variables as means with standard deviations (SD) and categorical variables as frequencies (%). Differences between groups were evaluated using Student\u0026rsquo;s t-tests or Mann-Whitney U tests for continuous data, and Chi-square tests for categorical distributions.\u003c/p\u003e \u003cp\u003eMachine Learning and Feature Importance Interpretation: Nine machine learning algorithms were screened using the caret package to identify key predictors, with model performance evaluated by the Area Under the Curve (AUC). The optimal model (Gradient Boosting) was subsequently interpreted using SHAP values via the shapviz package to quantify the global and local contributions of obesity features.\u003c/p\u003e \u003cp\u003eSurvival Analysis and Cox Regression: To identify predictors of incident hypertension, Cox proportional hazards models were applied. Univariate models were first used to evaluate the crude associations of each obesity-related index with hypertension risk. Multivariable Cox models were then constructed to adjust for demographic, lifestyle, biochemical, and clinical covariates. Proportional hazards assumptions were verified using Schoenfeld residuals.\u003c/p\u003e \u003cp\u003eNon-linear Dose-Response and Threshold Analysis: To move beyond linear assumptions, we utilized Restricted Cubic Splines (RCS) with the rms package to model the potential non-linear dose-response relationship between continuous obesity indices and hypertension risk.\u003c/p\u003e \u003cp\u003eSubgroup and Interaction Analysis: Stratified analyses were performed to evaluate the stability of the associations across different demographic subgroups (age, gender, smoking, and drinking status). Interaction terms were introduced into the Cox models to statistically test whether the predictive effects of obesity indices were modified by these subgroup variables.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eBaseline Characteristics of Study Participants\u003c/h2\u003e \u003cp\u003eA total of 5,070 participants from the CHARLS cohort (2011\u0026ndash;2018) and 1,065 participants from the ELSA cohort (2004\u0026ndash;2021) were included after applying the inclusion and exclusion criteria. Baseline characteristics of participants in the CHARLS and ELSA cohorts are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, respectively. Significant differences were observed for six obesity-related indices (BMI, VFI, TyG, TGB, TGW, and LA) across both datasets (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). In the CHARLS cohort, age, education, total cholesterol, HbA1c, diabetes, and arthritis were also significantly associated with hypertension, while in ELSA, age, education, and CRP showed significant group differences.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of participants with and without hypertension in the CHARLS cohort (2011\u0026ndash;2018)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003elevel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHealth Control\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHypertension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003en\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3595\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1475\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1925 (53.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e769 (52.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.377\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1670 (46.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e706 (47.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e57.93 (8.38)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e60.27 (8.91)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEdu (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot completed primary school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1756 (48.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e803 (54.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eprimary school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e785 (21.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e328 (22.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003emiddle school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e702 (19.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e229 (15.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ehigh school and above\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e352 (9.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e115 (7.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoke (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2455 (68.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1001 (67.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1140 (31.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e474 (32.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrink (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2363 (65.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e960 (65.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.684\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1232 (34.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e515 (34.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRP (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.31 (7.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.44 (5.44)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTC (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e191.24 (37.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e196.12 (41.43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHbA1c (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.19 (0.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.33 (0.89)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3474 (96.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1394 (94.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e121 (3.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e81 (5.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStroke (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3550 (98.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1453 (98.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.587\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e45 (1.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e22 (1.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCancer (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3570 (99.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1463 (99.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.789\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25 (0.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12 (0.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArthritis (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2419 (67.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e936 (63.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1176 (32.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e539 (36.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e974 (27.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e294 (19.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e944 (26.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e323 (21.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e899 (25.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e368 (24.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e778 (21.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e490 (33.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVFI (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1013 (28.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e255 (17.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e961 (26.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e306 (20.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e898 (25.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e369 (25.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e723 (20.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e545 (36.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTyG (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e969 (27.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e299 (20.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e924 (25.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e343 (23.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e887 (24.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e380 (25.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e815 (22.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e453 (30.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTGB (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e993 (27.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e275 (18.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e943 (26.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e324 (22.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e903 (25.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e364 (24.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e756 (21.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e512 (34.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTGW (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e997 (27.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e271 (18.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e953 (26.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e314 (21.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e901 (25.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e366 (24.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e744 (20.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e524 (35.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLA (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e972 (27.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e296 (20.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e960 (26.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e307 (20.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e904 (25.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e363 (24.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e759 (21.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e509 (34.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of participants with and without hypertension in the ELSA cohort (2004\u0026ndash;2021)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003elevel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHealth Control\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHypertension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003en\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e624\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e441\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e341 (54.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e230 (52.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.459\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e283 (45.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e211 (47.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e61.00 (6.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e63.22 (6.81)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEdu (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot yet in high school\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e180 (28.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e174 (39.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHigh school graduation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e155 (24.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e90 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eJunior college\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e171 (27.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99 (22.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUniversity degree or above\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e118 (18.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e78 (17.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoke (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e540 (86.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e385 (87.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.786\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84 (13.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e56 (12.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrink (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31 (5.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e35 (7.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.064\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e593 (95.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e406 (92.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRP (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.56 (3.99)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.51 (5.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTC (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.15 (1.17)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.22 (1.15)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.341\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHbA1c (mean (SD))\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.42 (0.47)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.46 (0.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.143\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e617 (98.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e436 (98.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7 (1.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (1.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStroke (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e622 (99.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e439 (99.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (0.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (0.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCancer (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e597 (95.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e417 (94.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.488\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27 (4.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e24 (5.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArthritis (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e457 (73.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e311 (70.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.366\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e167 (26.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e130 (29.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e177 (28.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e90 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e174 (27.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e92 (20.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e142 (22.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e124 (28.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e131 (21.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e135 (30.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVFI (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e196 (31.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e71 (16.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e166 (26.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100 (22.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e132 (21.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e134 (30.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e130 (20.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e136 (30.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTyG (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e180 (28.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e90 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e164 (26.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e103 (23.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e139 (22.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e123 (27.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e141 (22.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e125 (28.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTGB (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e186 (29.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e81 (18.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e166 (26.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100 (22.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e139 (22.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e127 (28.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e133 (21.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e133 (30.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTGW (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e196 (31.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e71 (16.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e156 (25.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e110 (24.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e140 (22.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e126 (28.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e132 (21.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e134 (30.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLA (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e189 (30.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e78 (17.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e150 (24.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e116 (26.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e152 (24.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e114 (25.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e133 (21.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e133 (30.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMachine Learning Performance and SHAP-Based Feature Importance Analysis\u003c/h3\u003e\n\u003cp\u003eNine machine learning algorithms, including Gradient Boosting, SVM, Logistic Regression, Partial Least Squares, and Neural Network, were applied to both cohorts. Gradient Boosting exhibited the best classification performance, with AUC\u0026thinsp;=\u0026thinsp;0.756 in CHARLS and AUC\u0026thinsp;=\u0026thinsp;0.878 in ELSA (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). SHAP value analysis ranked the relative contribution of variables to hypertension prediction. In CHARLS, the VFI was the most influential predictor, followed by LA and TGW \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea, c\u003cb\u003e)\u003c/b\u003e. In ELSA, LA ranked first in importance, followed by VFI and TGW \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb, d\u003cb\u003e)\u003c/b\u003e. These findings suggest that the predictive contribution of specific obesity indicators may differ by population and regional metabolic profiles. In both cohorts, VFI, LA, and TGW consistently ranked among the top predictors, underscoring their substantial contribution to hypertension development and their potential value as key obesity-related indicators in risk stratification.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e(a) Mean absolute SHAP values for all variables in the CHARLS model. (b) Mean absolute SHAP values for all variables in the ELSA model. (c) SHAP summary plot for the CHARLS model, showing the direction and magnitude of each feature\u0026rsquo;s impact on hypertension prediction. (d) SHAP summary plot for the ELSA model, indicating the relative influence and distribution of feature effects on model output.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCumulative Incidence of Hypertension Across Quartiles of Obesity-Related Indices\u003c/h2\u003e \u003cp\u003eTo further evaluate the longitudinal association between obesity-related indices and the risk of developing hypertension, Kaplan\u0026ndash;Meier survival analysis was conducted. In the CHARLS cohort (\u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e), distinct risk stratification was observed across quartiles for all six indices over the 7-year follow-up. The survival curves displayed a clear stepwise separation, where participants in the highest quartiles (Q4) consistently exhibited the steepest decline in hypertension-free probability compared to those in the lowest quartiles (Q1), with log-rank tests revealing highly significant differences for all indices (p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001). This graded dose-response pattern was robustly replicated in the ELSA cohort (\u003cb\u003eFigure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e), despite the longer follow-up period of 12 years. Participants with higher baseline levels of obesity indices showed a significantly accelerated onset of hypertension. While the majority of indices (BMI, VFI, TGB, TGW, and LA) maintained highly significant associations (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), TyG showed a slightly less pronounced but still statistically significant separation (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0017). These findings collectively indicate that elevated obesity-related indices are strong, long-term predictors of hypertension development across different populations and follow-up durations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eAssociation Between Obesity-Related Indices and Hypertension Risk in Cox Regression Models\u003c/h2\u003e \u003cp\u003eTo further quantify the relationship between obesity-related indices and hypertension risk over time, Cox proportional hazards models were applied. Univariate Cox proportional hazards models demonstrated that higher quartiles of all six indices (BMI, VFI, TyG, TGB, TGW, and LA) were significantly associated with an elevated risk of hypertension in both datasets (HR\u0026thinsp;\u0026gt;\u0026thinsp;1, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05; Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). After adjusting for potential confounders, multivariable Cox regression showed that VFI remained independently associated with hypertension risk (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in both datasets, while other indices (such as BMI and TGW) largely lost their statistical significance (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This phenomenon suggests that VFI captures the core risk information more effectively than other markers when evaluated concurrently, and the inverse trend of LA likely reflects statistical suppression due to the inclusion of multiple correlated metabolic indicators.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eNonlinear Dose\u0026ndash;Response Relationships Between Obesity Indices and Hypertension Risk\u003c/h2\u003e \u003cp\u003eTo further explore potential nonlinear associations and threshold effects between obesity-related indices and hypertension development, RCS models were applied. The RCS models revealed distinct dose-response patterns across the two cohorts. In the CHARLS cohort, BMI, VFI, TGB, TGW, and LA exhibited significant nonlinear associations (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05), whereas TyG showed a linear trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). notably, VFI, TGW, and LA displayed a J-shaped or U-shaped relationship: lower values were linked to an elevated risk (harmful range, HR\u0026thinsp;\u0026gt;\u0026thinsp;1), similar to the risk observed at higher values, whereas the lowest risk was found in the intermediate range. In contrast, lower values of BMI, TGB, and TyG were consistently associated with reduced hypertension risk (protective effects). In the ELSA cohort, while VFI, TGB, TGW, and LA also displayed statistically significant nonlinear relationships (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and BMI and TyG remained linear, the overall patterns differed from CHARLS. All six indices in ELSA generally showed a monotonic increasing trend, where hazard ratios progressively increased with higher values, confirming a robust positive dose\u0026ndash;response association without the paradoxical high risk at lower levels observed in CHARLS (\u003cb\u003eFig. S3\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eSubgroup Analyses of Obesity-Related Indices and Hypertension Risk in the ELSA Cohort\u003c/h2\u003e \u003cp\u003eIn the CHARLS cohort, the positive association between obesity indices and hypertension risk remained generally consistent across strata, though specific heterogeneities were observed (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Significant interactions were detected for BMI and TGB across age groups (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05), with a more pronounced risk gradient observed in younger participants (\u0026lt;\u0026thinsp;55 years) compared to older adults (\u0026gt;\u0026thinsp;55 years). Furthermore, LA and VFI exhibited significant interactions with gender (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Specifically, the predictive effects of LA and VFI on hypertension were stronger in women than in men. No significant interactions were observed for TGB, TGW, or TyG across any subgroups (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05), suggesting that their associations with hypertension are stable regardless of demographic or behavioral characteristics.\u003c/p\u003e \u003cp\u003eIn the ELSA cohort, the associations between obesity-related indices and hypertension risk were largely consistent across most subgroups (\u003cb\u003eFig. S4\u003c/b\u003e). Significant interactions were detected for BMI across gender strata (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.049), indicating that the predictive effect of BMI on hypertension risk was stronger in women than in men. For the remaining indices, including TGB, LA, VFI, TGW, and TyG, no significant interactions were observed across gender, age, smoking, or drinking subgroups (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05), suggesting stable associations regardless of demographic or behavioral characteristics. Within each subgroup, higher quartiles of obesity indices were generally associated with increased hazard ratios for hypertension, reinforcing their robustness as predictors across populations in the ELSA cohort.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eForest plots display HRs with 95% CI for each quartile of six obesity indices\u0026mdash;(a) BMI, (b) TGB, (c) LA, (d) VFI, (e) TGW, and (f) TyG\u0026mdash;stratified by gender, age, smoking, and drinking status.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study systematically evaluated the predictive performance of six distinct obesity-related indices for incident hypertension, leveraging a robust dual-cohort design (CHARLS and ELSA) and integrating advanced machine learning with traditional survival analysis. Our findings confirm that sophisticated metabolic and body composition indices\u0026mdash;particularly those incorporating visceral fat and insulin resistance markers\u0026mdash;outperform simple anthropometric measures in forecasting long-term hypertension risk among middle-aged and older adults.\u003c/p\u003e \u003cp\u003eThe core finding of this study is the consistent, superior predictive performance of VFI, LA, and TGW across both Chinese (CHARLS) and British (ELSA) cohorts. Both the SHAP analysis derived from the Gradient Boosting model and the multivariate Cox regression independently ranked VFI, LA, and TGW among the most potent risk factors, even after extensive adjustment for established confounders. This convergence of evidence highlights the biological primacy of visceral adiposity (VFI), lipotoxicity (LA), and the combined insulin resistance and central obesity (TGW) in driving hypertension onset.\u003c/p\u003e \u003cp\u003eConversely, classical measures like BMI and the isolated metabolic index TyG showed relatively weaker predictive power when competing against these composite markers in multivariate and ML models. This is consistent with recent literature suggesting that BMI, due to its inability to differentiate between metabolically harmful visceral fat and benign subcutaneous or lean mass, dilutes the true risk signal\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. VFI and LA, by specifically incorporating factors like WC, TG, and HDL-C, capture the essence of ectopic fat deposition and dysregulated adipokine secretion, mechanisms known to directly activate the sympathetic nervous system and the RAAS, leading to sustained blood pressure elevation\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. The TGW index further strengthens this prediction by integrating insulin resistance (TyG) with central fat distribution (WHtR), emphasizing the synergistic effect of impaired glucose and lipid metabolism on vascular stiffness and endothelial dysfunction\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOur RCS analysis provided critical insights into the dose-response relationship, revealing significant non-linearity for most indices. Notably, the shape of the risk curve showed crucial differences between the CHARLS (Chinese) and ELSA (European) cohorts, underscoring the influence of ethnicity and lifestyle on metabolic phenotypes. In the ELSA cohort, indices like VFI, LA, and TGW generally demonstrated a monotonic, continuous risk increase, suggesting that for this older European population, any level of metabolic deterioration poses a higher risk compared to the reference low group. However, in the CHARLS cohort, indices like VFI, LA, and TGW exhibited a more complex, often J- or U-shaped pattern at the lower end of the distribution. For example, some indices showed a seemingly protective effect (HR\u0026thinsp;\u0026lt;\u0026thinsp;1) at very low values before a steep risk increase, while others (VFI, LA, TGW) in CHARLS showed an unexpected elevated risk (HR\u0026thinsp;\u0026gt;\u0026thinsp;1) even at lower values. This finding may reflect the phenomenon of \"lean metabolically unhealthy\" or specific clinical cutoffs for Asian populations, where lower absolute levels of visceral fat or lipids, coupled with other undiagnosed inflammatory conditions, might still confer a disproportionately high risk for incident disease compared to Western cohorts\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. These cross-population differences suggest that single, universal threshold guidelines for these complex indices may be insufficient for global clinical application.\u003c/p\u003e \u003cp\u003eThe concurrent use of both machine learning and Cox proportional hazards models proved to be mutually reinforcing. ML, represented by the high AUC achieved by the Gradient Boosting model (CHARLS AUC\u0026thinsp;=\u0026thinsp;0.756; ELSA AUC\u0026thinsp;=\u0026thinsp;0.878), effectively validated the overall predictive accuracy of our feature set. Crucially, the SHAP framework provided transparency, definitively ranking VFI, LA, and TGW as the most influential features. This ranking strongly aligns with the independent hazard ratio estimates derived from the multivariable Cox model, where these same indices maintained significant and highest HRs. This convergence reinforces that the identified composite indices are not merely statistically correlated with hypertension but represent robust, high-impact prognostic factors. Furthermore, the ML approach successfully incorporated non-linear effects and complex feature interactions that traditional Cox models, particularly when treating variables linearly, might underestimate\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOur findings carry significant implications for hypertension prevention and management. Firstly, the emphasis on VFI, LA, and TGW suggests that clinical practice should move beyond BMI and prioritize screening strategies that directly assess metabolic obesity and visceral fat accumulation. Simple calculation of LA and TGW, which require only routine blood tests (TG, FPG) and basic anthropometrics (WC, height), can easily be integrated into community health screenings to identify the \"metabolically unhealthy normal weight\" individual\u0026mdash;a high-risk group often missed by BMI-centric approaches\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Secondly, the observed interaction between LA/VFI and gender (stronger effect in women in CHARLS) suggests the need for gender-specific risk stratification, potentially reflecting the distinct hormonal and fat distribution patterns in older women\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. The identification of population-specific non-linear thresholds further advocates for customized, targeted intervention strategies rather than a one-size-fits-all approach to weight and metabolic management.\u003c/p\u003e \u003cp\u003eThe primary strength of this study lies in its rigorous methodology, combining a dual-cohort longitudinal design for external validation (CHARLS to ELSA) with a sophisticated multi-modal analytical framework (ML\u0026thinsp;+\u0026thinsp;Cox\u0026thinsp;+\u0026thinsp;RCS). This integration provides highly consistent and robust evidence across diverse ethnic and environmental backgrounds. However, several limitations warrant consideration. First, hypertension status was defined by self-report in the final wave, which may introduce misclassification bias compared to direct clinical measurement, though this definition is standard in large cohort studies. Second, while we controlled for numerous established confounders, residual confounding from unmeasured factors (e.g., dietary intake, physical activity intensity) cannot be entirely excluded. Finally, the VFI and TGW indices rely on imputed calculations rather than direct imaging, representing surrogates for true visceral fat mass.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eVFI, LA, and TGW emerge as superior and consistent predictors of incident hypertension compared to traditional measures across both Chinese and European older adults. By utilizing these easily calculated composite indices, clinicians and public health practitioners can significantly enhance the precision of risk stratification, allowing for timely, targeted interventions focused on visceral adiposity and insulin resistance to combat the rising tide of hypertension.\u003c/p\u003e"},{"header":"Declarations","content":" \u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003eThis study was conducted using fully de-identified, publicly available data from CHARLS and ELSA. Both CHARLS and ELSA obtained ethical approval from their respective institutional review boards, and all participants provided informed consent at the time of original data collection. Because the present analysis involved secondary use of anonymized data and did not include any direct interaction with human subjects, additional ethical approval was not required in accordance with national and institutional regulations, including the Measures for the Ethical Review of Life Science and Medical Research Involving Human Subjects (2023, China).\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003eNot applicable.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eClinical trial number\u003c/h2\u003e \u003cp\u003enot applicable.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eNone.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eYuan Fang : Conceptualization; Study design; Writing \u0026ndash; review \u0026amp; editing; Chunqiang Gu : Conceptualization; Data collection; Data curation; Formal analysis; Interpretation of results; Writing \u0026ndash; review \u0026amp; editing; Dongmei Tang : Formal analysis; Data interpretation; Writing \u0026ndash; review \u0026amp; editing; All authors : Contributed to manuscript writing, critically revised the manuscript, and approved the final version.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data analyzed in this study were obtained from two publicly accessible longitudinal cohorts. CHARLS data are available through the Peking University Open Research Data Platform (http://charls.pku.edu.cn/) upon registration and adherence to data-use guidelines. ELSA data can be accessed through the UK Data Service (https://ukdataservice.ac.uk/) under standard end-user licensing. All derived variables and analytic code used to generate the results are available from the corresponding author upon reasonable request. The curated datasets derived from CHARLS and ELSA that were used for the present analyses are provided in Supplementary File 1 and Supplementary File 2, respectively. These files contain the cleaned variables required to reproduce the main analyses reported in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWorldwide trends in. hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants. Lancet Sep. 2021;11(10304):957\u0026ndash;80. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s0140-6736(21)01330-1\u003c/span\u003e\u003cspan address=\"10.1016/s0140-6736(21)01330-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou B, Perel P, Mensah GA, Ezzati M. Global epidemiology, health burden and effective interventions for elevated blood pressure and hypertension. Nat Rev Cardiol Nov. 2021;18(11):785\u0026ndash;802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41569-021-00559-8\u003c/span\u003e\u003cspan address=\"10.1038/s41569-021-00559-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMills KT, Stefanescu A, He J. The global epidemiology of hypertension. Nat Rev Nephrol Apr. 2020;16(4):223\u0026ndash;37. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41581-019-0244-2\u003c/span\u003e\u003cspan address=\"10.1038/s41581-019-0244-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarey RM, Whelton PK, Prevention. Detection, Evaluation, and Management of High Blood Pressure in Adults: Synopsis of the 2017 American College of Cardiology/American Heart Association Hypertension Guideline. \u003cem\u003eAnn Intern Med\u003c/em\u003e. Mar 6 2018;168(5):351\u0026ndash;358. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7326/m17-3203\u003c/span\u003e\u003cspan address=\"10.7326/m17-3203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoth GA, Mensah GA, Johnson CO, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990\u0026ndash;2019: Update From the GBD 2019 Study. J Am Coll Cardiol Dec. 2020;22(25):2982\u0026ndash;3021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jacc.2020.11.010\u003c/span\u003e\u003cspan address=\"10.1016/j.jacc.2020.11.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarey RM, Muntner P, Bosworth HB, Whelton PK. Reprint of: Prevention and Control of Hypertension: JACC Health Promotion Series. J Am Coll Cardiol. Dec 2018;11(23 Pt B):2996\u0026ndash;3011. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jacc.2018.10.022\u003c/span\u003e\u003cspan address=\"10.1016/j.jacc.2018.10.022\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbe M, Hirata T, Morito N, et al. Smartphone application-based interventions for cardiometabolic risk factor management: A systematic review and meta-analysis. Hypertens Res Sep. 2025;3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41440-025-02365-y\u003c/span\u003e\u003cspan address=\"10.1038/s41440-025-02365-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHall JE, do Carmo JM, da Silva AA, Wang Z, Hall ME. Obesity, kidney dysfunction and hypertension: mechanistic links. Nat Rev Nephrol Jun. 2019;15(6):367\u0026ndash;85. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41581-019-0145-4\u003c/span\u003e\u003cspan address=\"10.1038/s41581-019-0145-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParvanova A, Reseghetti E, Abbate M, Ruggenenti P. Mechanisms and treatment of obesity-related hypertension-Part 1: Mechanisms. Clin Kidney J Jan. 2024;17(1):sfad282. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ckj/sfad282\u003c/span\u003e\u003cspan address=\"10.1093/ckj/sfad282\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDobre MZ, Virgolici B, Timnea O. Key Roles of Brown, Subcutaneous, and Visceral Adipose Tissues in Obesity and Insulin Resistance. Curr Issues Mol Biol May. 2025;9(5). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cimb47050343\u003c/span\u003e\u003cspan address=\"10.3390/cimb47050343\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStefan N. Causes, consequences, and treatment of metabolically unhealthy fat distribution. Lancet Diabetes Endocrinol Jul. 2020;8(7):616\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s2213-8587(20)30110-8\u003c/span\u003e\u003cspan address=\"10.1016/s2213-8587(20)30110-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePapathanasiou KA, Roussos CE, Armylagos S, Rallidis SL, Rallidis LS. Lipid Accumulation Product Is Predictive of Cardiovascular Hospitalizations among Patients with Stable Ischemic Heart Disease: Long-Term Follow-Up of the LAERTES Study. J Cardiovasc Dev Dis Oct. 2024;10(10). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/jcdd11100316\u003c/span\u003e\u003cspan address=\"10.3390/jcdd11100316\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu AR, Jin Q, Shen Z, Zhang J, Fu Q. Association between the risk of hypertension and triglyceride glucose index in Chinese regions: a systematic review and dose-response meta-analysis of a regional update. Front Cardiovasc Med. 2023;10:1242035. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fcvm.2023.1242035\u003c/span\u003e\u003cspan address=\"10.3389/fcvm.2023.1242035\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRen X, Chen M, Lian L, et al. The triglyceride-glucose index is associated with a higher risk of hypertension: evidence from a cross-sectional study of Chinese adults and meta-analysis of epidemiology studies. Front Endocrinol (Lausanne). 2025;16:1516328. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fendo.2025.1516328\u003c/span\u003e\u003cspan address=\"10.3389/fendo.2025.1516328\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, Ye R, Sun L, et al. Relationship between novel anthropometric indices and the incidence of hypertension in Chinese individuals: a prospective cohort study based on the CHNS from 1993 to 2015. BMC Public Health Mar. 2023;6(1):436. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12889-023-15208-7\u003c/span\u003e\u003cspan address=\"10.1186/s12889-023-15208-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMao L, Lin L, Shi Z, Song H, Zhao H, Xu X. Determinants and prediction of hypertension among Chinese middle-aged and elderly adults with diabetes: A machine learning approach. Heliyon Sep. 2024;30(18):e38124. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.heliyon.2024.e38124\u003c/span\u003e\u003cspan address=\"10.1016/j.heliyon.2024.e38124\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao G, Zhou Z. Correlation between obesity-related indices and hypertension. Am J Transl Res. 2024;16(8):3842\u0026ndash;50. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.62347/uufg4260\u003c/span\u003e\u003cspan address=\"10.62347/uufg4260\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThakker J, Khaliq I, Ardeshna NS, Lavie CJ, Oktay AA. The Obesity Paradox of Cardiovascular Outcomes in Patients with Diabetes Mellitus. Curr Diab Rep Jun. 2025;7(1):35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11892-025-01592-4\u003c/span\u003e\u003cspan address=\"10.1007/s11892-025-01592-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSilva GFS, Fagundes TP, Teixeira BC, Chiavegatto Filho ADP. Machine Learning for Hypertension Prediction: a Systematic Review. Curr Hypertens Rep Nov. 2022;24(11):523\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11906-022-01212-6\u003c/span\u003e\u003cspan address=\"10.1007/s11906-022-01212-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng X, Zhu J, Hua Z, Yao S, Tong H. Comparison of obesity indicators for predicting cardiovascular risk factors and multimorbidity among the Chinese population based on ROC analysis. Sci Rep Sep. 2024;9(1):20942. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-024-71914-1\u003c/span\u003e\u003cspan address=\"10.1038/s41598-024-71914-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOfstad AP, Sommer C, Birkeland KI, et al. Comparison of the associations between non-traditional and traditional indices of adiposity and cardiovascular mortality: an observational study of one million person-years of follow-up. Int J Obes (Lond). May 2019;43(5):1082\u0026ndash;92. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41366-019-0353-9\u003c/span\u003e\u003cspan address=\"10.1038/s41366-019-0353-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJung JY, Ryoo JH, Oh CM, et al. Visceral adiposity index and longitudinal risk of incident metabolic syndrome: Korean genome and epidemiology study (KoGES). Endocr J Jan. 2020;28(1):45\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1507/endocrj.EJ19-0008\u003c/span\u003e\u003cspan address=\"10.1507/endocrj.EJ19-0008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Y, Hu P, He Y, Qin H, Hu L, Yang R. Association of TyG index and central obesity with hypertension in middle-aged and elderly Chinese adults: a prospective cohort study. Sci Rep Jan. 2024;26(1):2235. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-024-52342-7\u003c/span\u003e\u003cspan address=\"10.1038/s41598-024-52342-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang Y, Zhou Y, Xu Y, et al. Inflammatory markers link triglyceride-glucose index and obesity indicators with adverse cardiovascular events in patients with hypertension: insights from three cohorts. Cardiovasc Diabetol Jan. 2025;8(1):11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12933-024-02571-x\u003c/span\u003e\u003cspan address=\"10.1186/s12933-024-02571-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHwang SH, Lee H, Lee JH, et al. Machine Learning-Based Prediction for Incident Hypertension Based on Regular Health Checkup Data: Derivation and Validation in 2 Independent Nationwide Cohorts in South Korea and Japan. J Med Internet Res Nov. 2024;5:26:e52794. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/52794\u003c/span\u003e\u003cspan address=\"10.2196/52794\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim B, Taniguchi K, Isobe T, Oh S. Triglyceride-glucose index is capable of identifying metabolically obese, normal-weight older individuals. J Physiol Anthropol Feb. 2024;3(1):8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s40101-024-00355-6\u003c/span\u003e\u003cspan address=\"10.1186/s40101-024-00355-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaneva AM, Bojko ER. Sex differences in the association between obesity and hypertension. Arch Physiol Biochem Jun. 2023;129(3):682\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/13813455.2020.1861027\u003c/span\u003e\u003cspan address=\"10.1080/13813455.2020.1861027\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Dual Cohort, CHARLS, ELSA, Longitudinal Study, Obesity Indices, Machine Learning, SHAP Analysis","lastPublishedDoi":"10.21203/rs.3.rs-8650506/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8650506/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eHypertension remains a major global health burden, with excess adiposity serving as a key modifiable contributor to its development. However, conventional anthropometric measures, particularly body mass index (BMI), inadequately reflect metabolically harmful fat accumulation. Consequently, the predictive value of emerging obesity-related indices for incident hypertension remains incompletely defined.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe systematically evaluated six obesity-related indices\u0026mdash;BMI, visceral fat index (VFI), triglyceride\u0026ndash;glucose index (TyG), TyG\u0026ndash;BMI (TGB), TyG\u0026ndash;waist-to-height ratio (TGW), and lipid accumulation product (LA)\u0026mdash;in relation to new-onset hypertension using data from two prospective cohorts, CHARLS and ELSA. Cox proportional hazards models, restricted cubic spline (RCS) analyses, and interpretable machine-learning methods were applied to assess associations, nonlinear patterns, and relative predictor importance.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eIn both cohorts, all six indices were significantly associated with incident hypertension in univariate analyses, with graded risk increases across quartiles. After mutual adjustment for all indices and covariates, VFI remained the only predictor consistently associated with hypertension risk in both CHARLS and ELSA. RCS analyses identified nonlinear associations for VFI, TGW, and LA in CHARLS, whereas relationships in ELSA were largely monotonic. Machine-learning models showed good discrimination (AUC 0.756 in CHARLS; 0.878 in ELSA), and SHAP analysis consistently ranked VFI, LA, and TGW as the most influential predictors.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eOverall, VFI and related composite adiposity indices, particularly LA and TGW, outperform BMI and isolated metabolic markers in predicting incident hypertension. Population-specific nonlinear patterns highlight the heterogeneity of obesity phenotypes and the limitations of BMI-based risk assessment. Incorporation of these indices into routine screening may improve early identification of individuals at elevated risk, including those with metabolically unhealthy normal weight.\u003c/p\u003e","manuscriptTitle":"Comparative Evaluation of Obesity-Related Indices for Predicting Incident Hypertension: Evidence from Chinese and UK Longitudinal Cohorts With Machine Learning Interpretation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-12 20:36:12","doi":"10.21203/rs.3.rs-8650506/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-21T02:21:32+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"11896194470239586153327370528002887283","date":"2026-03-18T11:46:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-15T12:37:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"60520276312051848541191930360439449094","date":"2026-03-15T03:44:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"326998853021626697295747525078890523622","date":"2026-03-13T14:46:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"238036245897104575444389135013347294143","date":"2026-03-01T09:07:32+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-19T15:28:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"280671126640884142244001659212828403691","date":"2026-02-15T12:11:52+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-10T07:46:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"292840287210024461471272196148763458331","date":"2026-02-10T07:44:26+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-10T01:28:48+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-01-29T10:51:03+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-29T02:47:07+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-29T02:46:11+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cardiovascular Disorders","date":"2026-01-20T13:35:58+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9dd1755f-b996-4512-a08a-ac0ef8788b10","owner":[],"postedDate":"February 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-27T16:00:28+00:00","versionOfRecord":{"articleIdentity":"rs-8650506","link":"https://doi.org/10.1186/s12872-026-05885-8","journal":{"identity":"bmc-cardiovascular-disorders","isVorOnly":false,"title":"BMC Cardiovascular Disorders"},"publishedOn":"2026-04-22 15:56:58","publishedOnDateReadable":"April 22nd, 2026"},"versionCreatedAt":"2026-02-12 20:36:12","video":"","vorDoi":"10.1186/s12872-026-05885-8","vorDoiUrl":"https://doi.org/10.1186/s12872-026-05885-8","workflowStages":[]},"version":"v1","identity":"rs-8650506","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8650506","identity":"rs-8650506","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}