Privacy-Preserving Epidemiological Analysis for a Distributed Database of Hospitals


The Diagnosis Procedure Combination (DPC) database is a large-scale claim-based database of Japanese hospitals that uses a standardized case-mix patient-classification system for the evaluation of clinical procedures and hospital performance. The dataset does not include confidential information, although content details may allow identification of individual patients.

We have developed a secure scheme for preserving patient privacy, which also enables epidemiological analysis that can adjust for patient characteristics to enable unbiased comparisons. We tested its performance with a DPC database sample of 80 hospitals.