Full Bayesian Procedure for File Linking to Analyze End-of-Life Medical Costs

• Activity Number: 402
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Roee Gutman*+ and Alan Zaslavsky
• Companies: Harvard University and Harvard Medical School
• Keywords: File Linking ; Bayesian Analysis ; Multiple Imputations

Abstract

We studied end-of-life medical expenses using costs from Medicare claims and cause of death (CoD) from death certificates. In the absence of a unique identifier linking the files, common variables identified matches for 33% of deaths. The remaining cases formed cells with multiple cases. We sampled from the joint posterior distribution of model parameters and the permutations that link cases from the 2 files within the cells. The linking models included the regression of death location on CoD and other parameters, and the regression of cost measures with a monotone missing data pattern on CoD and other demographics. Permutations were sampled by enumerating the exact distribution (small cells) or by the Metropolis algorithm (large cells). Sparse matrix data structures enabled efficient calculations despite the large files (1.7MM). Our approach can be used for other file linking analyses.