Abstract
OBJECTIVE: This study aimed to compare and evaluate the predictive ability of obesity-related indicators for new-onset hypertension, and to explore causal effects using Mendelian randomization (MR). METHODS: A total of 22, 912 eligible participants were included from the Henan Rural Cohort Study. Logistic regression was used to identify key predictors of hypertension, and gradient boosting machine (GBM) models incorporating ten obesity indices were developed. Model performance was evaluated using the area under the curve (AUC), and indicator importance was assessed with SHapley Additive exPlanations (SHAP). A multi-state Markov model estimated life expectancy (LE) and health-adjusted life expectancy (HALE). Causal associations were examined through MR analysis. RESULTS: The basic GBM model achieved an AUC of 0.835 (95% CI: 0.826-0.844), with body mass index (BMI) showing the highest predictive value (AUC = 0.844). SHAP analysis revealed that all obesity indicators were positively associated with new-onset hypertension but ranked below age and blood pressure level in importance. At age 18, LE was 62.72, 66.11, 68.79 years for individuals with normal weight, overweight, and obesity, respectively. The corresponding HALE was 29.45, 26.23, 22.29 years. MR analysis confirmed causal associations of obesity indicators with hypertension. CONCLUSION: Obesity-related indicators are significantly associated with new-onset hypertension, and they are linked to increased LE and reduced HALE. The findings provide evidence to early recognition and prevention of hypertension risk based on obesity-related indicators.