Abstract
OBJECTIVE: This study aims to identify potential independent risk factors for rheumatoid arthritis (RA)- related mortality and develop a nomogram model to predict individualized mortality risk. METHODS: This study included 310 RA patients from the National Health and Nutrition Examination Survey (NHANES) during 1999 - 2018. We applied LASSO, univariate, and multivariate logistic regression analyses to determine risk factors in the training cohort and construct a nomogram model. Calibration plots evaluated the nomogram's accuracy. Finally, we established the nomogram's clinical utility through DCA and performed internal validation within the training cohort. RESULTS: Of the 310 patients, 140 experienced RA - related deaths, corresponding to a mortality rate of 45.16%. Within the training cohort, age, heart failure, and systemic inflammatory response index (SIRI) emerged as independent predictors of RA - related mortality. A nomogram model, constructed through multivariable logistic analysis, demonstrated an AUC of 0. 852 (95% CI: 0. 799 - 0. 904) in the training cohort and an AUC of 0. 904 (95% CI: 0. 846 - 0. 963) in the validation cohort. The calibration curve revealed a strong agreement between predicted and actual probabilities. In both training and validation cohorts, DCA highlighted the nomogram's significant net benefits for predicting RA - related mortality risk. CONCLUSIONS: This study demonstrates age, heart failure, and SIRI's ability to predict RA mortality with good discrimination and clinical utility. The model gives clinicians a simple tool to quickly identify high - risk RA patients, promoting early intervention, personalized treatment, and better prognosis.