Monitoring the accuracy of a PACS image database

"What you don't know won't hurt you" is a proverb that does not apply to a database on which patient lives depend. One of the core components of the picture Archiving and communications system (PACS) is an image database that contains the location and state of images and their corresponding demographic information. An image associated with the wrong patient name has potentially devastating implications, especially if the error is not caught early. This article describes how Texas Children's Hospital addresses the challenge of ensuring the accuracy of data of our PACS image database. It presents the routine checks that our PACS analysts perform every three hours and on a daily and monthly basis. These include steps involved in "fixing" exam data when an examination has missing or incorrect demographic information. In addition, this report compares how our institution assured accuracy of our film-based archive before PACS. Human error is the source of most inaccuracies in the image database, and automation such as bar code scanners and DICOM Modality Work List management have decreased their frequency, but without totally eliminating them. Some errors are routinely generated by the nature of radiology imaging operations and the limitations of devices for automating those operations. For example, computed tomography exams of the head, chest, and abdomen are routinely acquired during the same scan but must be separated into different exams for interpretation by different physicians. Factors contributing to inaccurate information include major system modifications, such as software or hardware upgrades, service interruptions, inaccurate problem descriptions provided by the user to the PACS analyst, and steps taken by the PACS analyst or technologist to correct errors.