Patient note deidentification using a find-and-replace iterative process.

Abstract

Medical research relies on access to clinical data. Health Insurance Portability and Accountability Act regulations require that patient information required for clinical research not have data that can be used to identify the patients from whose medical records the information has been derived. The only exception would be an institutional review board (IRB)-approved study for which the researcher has obtained a waiver to use patient data for a research study. Before requesting an IRB waiver, however, the researcher may want to search the clinical data for particular characteristics or determine whether the quantity of data warrant obtaining IRB approval. The application, the Simple PAtient Note Scrubber, or SPANS, reviews and changes line content through an iterative process. At each iteration, SPANS analyzes changes made during the previous pass and reviews changes in relation to terms adjacent to the newly altered data. Knowledge of the document domain, encompassing the different types of documents to be scrubbed, is the key to making this type of process effective.