Avoiding Unread Images with PACS

Abstract

In their article in this issue of Academic Radiology, reporting the effect of a picture archiving and communication system (PACS) on the percentage of unread studies, Evers et al (1) state, “In a perfect world, one would assume that a PACS-RIS solution would result in fewer cases unavailable for follow-up [ie, interpretation], as we had hypothesized at the inception of this study.” We believe that the negative results reported are due to incomplete integration of the PACS with the RIS and to workflow issues. A series of articles written early in the PACS era (2) clearly described the importance of events that occur in the radiology information system (RIS) and that trigger actions in the PACS. For example, the scheduling of a procedure in the RIS should initiate the queuing of a request to retrieve relevant prior studies from the long-term archive (prefetching). Indeed, if the RIS (and therefore the PACS) knows about a current study only after it is acquired, the opportunity is missed to retrieve relevant prior studies in advance (and in off hours), which can further delay or impair proper interpretation until those studies are retrieved. A key workflow and integration step unmentioned by Evers et al is the obligatory RIS scheduling of an examination before its initiation. This step is important for many reasons. First, with proper hospital information system–RIS integration, the accuracy of patient demographics is maintained. With the Digital Imaging and Communications in Medicine (DICOM) Modality Worklist, study and demographic information can be downloaded directly for scheduled procedures for a given imaging device, without keyboard errors (which are frequent), and that information can be placed into the DICOM header. Most of the time, this process guarantees that studies sent to the PACS (preferably automatically) contain header information that is correct. In any case, when the image stream arrives at the entry “gateway” of the PACS, it can be checked or verified against pending orders. In the properly designed system, only studies that pass this verification test are accepted into the PACS for permanent storage and distribution to workstations (3). The small percentage of studies in which RIS verification fails are placed in a holding area (resolve queue) for human verification. While currently only a small percentage of studies fall into this queue at our facility, they are checked and worked toward resolution by PACS quality assurance personnel several times a day. One of the more common reasons for failed verification has been erroneous patient identification at the modality—an increasingly rare occurrence as use of the DICOM Modality Worklist becomes routine. Installation of a PACS follows one of two patterns: phased or all at once. In all imaging departments, regardless of size, the first small step toward a PACS should be the integration of the hospital information system and the RIS. There is no evading this step, even though it can be difficult and expensive. Failure to heed this principle can result in a PACS and an image database that lack integrity and that might eventually become useless. The advantage of completing this step properly is that subsequent, incremental additions of other sections or modalities will be easier and less expensive. In our experience, following these principles results in a PACS that decreases the rate of studies going unread or being tardily read, and that has the desired positive financial effect. Furthermore, a properly integrated PACS with good workstation design can result in increased radioloAcad Radiol 2002; 9:1331–1332