Titolo della tesi: A Bayesian probabilistic record linkage method to perform survival analysis for joint prostheses when the operated side is not available
Surveillance on implantable joint prostheses is crucial to assess performance of devices and ensure safety of patients. In data collection, unique identifiers for patients and information about the operated side are necessary to correctly follow implanted devices over time and perform accurate survival analysis. Hospital Discharge Data is the largest source of information on joint replacement surgery in Italy; however, it does not include the operated side variable, making follow-up impossible and survival estimates unreliable. This work provides a Bayesian Probabilistic Record Linkage model that allows to link, with good accuracy, primary interventions and corresponding revisions (i.e. implant failure) where a unique device is involved, making follow-up possible and precise enough to perform survival analysis with results close to the ones obtained when the operated side is known.