For PACS, archiving is an important part of the acronym, and equally important to effective system management. PACS calls for changes in the way patient information is handled, and the life-cycle management of images gains greater significance as imaging becomes more integrated with the electronic medical record. My experience of more than 20 years in PACS application and implementation has taught some lessons worth sharing.

Redundancy Matters
For years, radiology has maintained a file room for the storage of film, with no redundancy. This has probably been due to the realities of the prohibitive media cost and storage space needed to duplicate film. Some would argue that PACS, in and of itself, provides a level of redundancy; in a sense, that is true. As images are acquired, they are stored in a short-term cache for immediate accessibility. The short-term cache is then written to the long-term archive for permanent storage, so there is redundancy. One needs, however, to be aware of the PACS event that triggers repetition to the long-term archive. In many PACS, images are written to archive as soon as the study is validated. In others, images may not be written to the archive until the study is read. If this is the case, there is a risk of image loss in the event of a catastrophic failure of the short-term cache.

Another key aspect of image redundancy involves the application of the part of the DICOM standard know as DICOM Storage Commit Service. If supported, once the PACS accepts ownership of the examination, the imaging device can automatically delete the study from its cache. The consequences of a modality not supporting Storage Commit Service is the inefficient use of its local memory—and the potential for a lost study if the modality deletes it and the PACS has not accepted it.

As PACS become more sophisticated and enterprise oriented, the importance of a redundant long-term archive increases. The most sophisticated (and expensive) approach is to maintain redundant storage devices in two locations, so that if one is damaged, the other can continue as if nothing had happened. Less sophisticated approaches usually involve backing up the primary archive to less expensive media, such as DVDs or high-density tape. As an alternative, a remote service can back up images in the background so that they can be recovered in the event of a failure.

Plan for Disaster Recovery
There are many factors to consider in the event of a disaster, and it is important to have a policy and procedure for dealing with such an event. This is particularly important given the security policies established as part of HIPAA in April 2005. A strict interpretation of these policies implies that a facility must have a policy for handling data recovery, and that the policy must be periodically tested.

Depending on the type of redundant archive, this may be a simple or more complex task. In the case of redundant spinning media, it can be as straightforward as deleting a study off one disk and ensuring that the redundant copy is rewritten. In the case of off-line media, it means finding the correct media and reintroducing them to the system so they can be rewritten to the spinning disk.

Part of the disaster-recovery policy should also be specification of media identification, if off-line media are used. This may seem obvious, but I have seen facilities that did not have proper discipline in identifying optical disks; now, when the system requests retrieval from certain media, it is nearly impossible to identify the correct media.

Plan for Media Migration
Prior to PACS, various policies were deployed with respect to the maintenance of modality archive media. Some sites may have kept media for several months and then recycled them for reuse, as the film was the permanent record. Others of an academic nature may have had reasons for keeping media permanently for research purposes.

With the advent of a PACS, the site will eventually need to decide what to do with prior modality media. This is probably a bigger issue for those sites that kept studies around for significant periods of time. In those instances, it makes sense to transition the archives to PACS, if possible. This can be time consuming and has limitations, in terms of ensuring that patient demographics match in the PACS. Often, this can be done with temporary resources in off hours in order to have the least impact on operations. The advantages of migrating these data are obvious: better accessibility, additional redundancy, and the elimination of multiple sources of media to manage.

Transition From Film Purging
An idiosyncrasy of the implementation of a PACS is the need to revisit study-purging policies. In a purely film-based environment, there may be a concerted effort to purge film after the legal retention period. Should a patient return just prior to when the folder would have been purged, however, many facilities will review its contents, reset the purge date, and continue to keep the folder.

Why, then, do some assume that with the advent of PACS, the film library will disappear in, say, five years? The assumption is that all new studies are acquired digitally, so each year, the oldest year’s worth of folders can be purged. In reality, the policy should be no different from what it was in the days of film. If a patient returns prior to a purge, the film folder should be similarly reviewed and relevant contents preserved for additional time. Of course, selective digitization can be applied to preserve old relevant studies electronically, but the principle is the same: any new study for a prior patient in the PACS should trigger a folder review to ensure proper retention of comparative studies.

The proper management of images in an electronic archive is a key part of PACS. Establishing good policies and procedures for maintaining a PACS archive is tantamount to good PACS management.

Joseph L. Marion, MBA, jmarion@wi.rr.com is principal of Healthcare Integration Strategies, Waukesha, Wis, a health care information consulting company, and has been involved with PACS since its conception in the 1970s.