Bayesian hierarchical spatio-temporal models for the analysis of interregional mobility of patients undergoing joint arthroplasty in Italy

Gaining insight into the drivers of patient mobility in joint replacement procedures is crucial for evidence-based healthcare planning and policy evaluation. However, traditional indices commonly used to describe mobility, such as the Attraction Index and the Escape Index, provide only static, ratio-based measures and fail to capture the spatio-temporal dynamics that shape patient flows. This work develops and applies Bayesian hierarchical spatio-temporal models to study patient mobility in joint arthroplasty, with a particular focus on total ankle replacement in Italy. The proposed approach integrates spatial modeling, temporal trends, and relevant covariates, offering a more robust framework than descriptive indices alone. Furthermore, it incorporates a soft constraint in the likelihood to align the estimated annual totals with observed or externally estimated values. The proposed models demonstrate promising accuracy, effectively capture the spatio-temporal dynamics underlying patient mobility, and can be employed to develop forecasting strategies to inform future healthcare planning.