Development and Validation of a Prognostic Model for Osteoarthritis Risk in Older Adults with a History of Stroke: A Cross-sectional Analysis Based on NHANES 2011–2018

Background Osteoarthritis (OA) is frequently observed as a comorbidity in individuals with a history of stroke; however, there is currently a lack of widely validated predictive models for individualized risk assessment. The present study aimed to develop t and validate a prognostic model to estimate the likelihood of osteoarthritis occurrence among individuals with a history of stroke. using data from the United States National Health and Nutrition Examination Survey (NHANES). Methods Clinical and demographic data for participants with self-reported stroke (n = 1077) were extracted from the NHANES database covering the period from 2011 to 2018. Candidate predictors of osteoarthritis (OA) identified by univariate analysis were screened using Least Absolute Shrinkage and Selection Operator (LASSO) regression to minimize overfitting. Variables selected by LASSO were entered into a multivariable logistic regression model to identify independent OA predictors, and a nomogram was constructed accordingly. The cohort was randomly divided into training and validation sets (7:3). Model performance was evaluated by receiver operating characteristic (ROC) curves for discrimination, calibration plots for agreement between predicted and observed outcomes, and decision curve analysis (DCA) for clinical utility across risk thresholds. Results Eight potential predictors (asthma, chronic bronchitis, gender, general health, osteoporosis, cancer or malignancy, difficulty standing, and difficulty stooping, crouching, or kneeling) were selected from the candidate variables using LASSO regression. The risk model constructed using these variables demonstrated favorable predictive performance, yielding an area under the receiver operating characteristic curve (AUC) of 0.737 in the training set and 0.721 in the validation set. Calibration analysis revealed a high degree of concordance between predicted probabilities and actual outcomes. Decision curve analysis further indicated that the nomogram offered a net clinical benefit for risk threshold ranges of 0.56–0.90 in the training set and 0.58–0.87 in the validation set. Conclusion A risk monogram incorporating gender, general health, asthma, chronic bronchitis, osteoporosis, cancer or malignancy, difficulty standing, and difficulty stooping, crouching, and kneeling demonstrated good clinical utility for predicting osteoarthritis risk among participants with a history of stroke. This individualized prediction tool may facilitate early identification and targeted prevention in high-risk populations.