学术文献库

Understanding the association between injury severity and patients' potential for recovery is crucial to providing better care for patients with traumatic brain injury (TBI). Estimation of this relationship requires clinical information on injury severity, patient demographics, and healthcare utilization, which are often obtained from separate data sources. Because of privacy and confidentiality regulations, these data sources do not include unique identifiers to link records across data sources. Record linkage is a process to identify records that represent the same entity across data sources in the absence of unique identifiers. These processes commonly rely on agreement between variables that appear in both data sources to link records. However, when the number of records in each file is large, this task is computationally intensive and may result in false links. Blocking is a data partitioning technique that reduces the number of possible links that should be considered. Healthcare providers can be used as blocks in applications of record linkage with healthcare datasets. However, providers may not be uniquely identified across files. We propose a Bayesian record linkage procedure that simultaneously performs block-level and record-level linkage. This iterative approach incorporates the record-level linkage within block pairs to improve the accuracy of the block-level linkage. Subsequently, the algorithm improves record-level linkage using the accurate partitioning of the linkage space through blocking. We demonstrate that our proposed method provides improved performance compared to existing Bayesian record linkage methods that do not incorporate blocking. The proposed procedure is then used to merge registry data from the National Trauma Data Bank with Medicare claims data to estimate the relationship between injury severity and TBI patients' recovery.