Spatial heterogeneity and its influencing factors of cardiometabolic multimorbidity in a natural community population: a study based on Lingwu city, rural Northwest China.

Abstract

Objective: Cardiometabolic multimorbidity (CMM) significantly contributes to the economic burden in China, particularly in rural areas. This study aimed to analyze the spatiotemporal distribution of CMM and identify its primary influencing factors in different townships in Lingwu City, Ningxia, to inform public health policies in Northwest China.

Methods: The standardized prevalence of CMM was investigated using data from Cardiovascular Disease High-Risk Group Early Screening and Comprehensive Intervention Program (2017-2022) conducted in Lingwu City, Ningxia. We applied spatial autocorrelation, cluster analysis, and spatiotemporal scanning to explore the spatiotemporal distribution characteristics of CMM and identify high-risk clusters. Four machine learning algorithms, logistic regression (LR), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost) were developed using 15 major cardiovascular disease influence factors. The performance of these models was evaluated based on accuracy, precision, recall, and AUC to determine their applicability across different townships in Lingwu City. The optimal model was selected for further analysis using interpretable machine learning algorithms (SHAP analysis) to identify common and key influence factors influencing CMM prevalence across townships.

Results: Among the 11,353 participants, 1,334 individuals (11.8%, 95% CI: 11.2-12.4%) were diagnosed with CMM, with significant variations in influence factors observed across townships (P < 0.05). Trend surface analysis revealed a parabolic geographic distribution of CMM prevalence in Lingwu City, increasing from north to south. Dongta Township exhibited the highest prevalence (16.6%), followed by Chongxing Township and Wutongshu Township. Spatiotemporal scanning identified four high-incidence clusters. The random forest algorithm outperformed others in predicting CMM prevalence across townships. SHAP analysis highlighted differences in the geographic distribution of 15 influence factors. Age, waist circumference, and hypertension were significant influence factors across Lingwu City. Township-specific influence factors included TG and BMI in Dongta; HDL and TG in Chongxing, Haojiaqiao; HDL and TC in Wutongshu and TC, TG, HDL and BMI in Baitugang.

Conclusion: CMM prevalence shows significant geographic variation within Lingwu City, with distinct risk factors across townships. Tailored interventions, based on local needs, should be implemented to reduce CMM prevalence effectively, optimize health resource allocation, and inform public health policies in rural areas of Northwest China.