In the pre-digital age, healthcare providers took a pragmatic approach to dealing with old film.
Storage space was costly, and the silver recovered from the back of film could be sold, returning money to the health service. So as a general rule, if the film was more than eight years old, and the patient hadn’t made any more visits to hospital in that period, the film was destroyed.
If there was a need to keep an image for longer (because it related to an oncology patient, for example), it could be dealt with by a phone call to the film filing clerk. They would put a sticker on the film to indicate that it was not to be destroyed, and that was that.
The implementation of digital picture archiving and communications systems should have changed things. But because most trusts implemented PACS through the National Programme for IT, they could put off worrying about how to manage old images.
Now that the National PACS Programme’s ten year contracts are expiring, and trusts are being faced with putting in place their own contracts and migrating their old images to new systems, the issue has resurfaced.
Although digital storage has fallen in cost, it once again makes no financial sense to store images in perpetuity.
As Neelam Dugar, consultant radiologist at Doncaster and Bassetlaw NHS Trust, says: “There’s no point in keeping images if they’re not of value to patient care. It’s a huge expense to the taxpayer at no benefit to anybody.”
When is it safe to delete images?
Trusts need to address two questions; how long should images be stored for, and how should their deletion be managed?
Updated guidance issued by the Royal College of Radiologists in 2008 says general patient records can be deleted eight years after the conclusion of treatment (in Scotland, six years after the date of the last entry, or three years after death).
There are certain exceptions in which the RCR advises keeping images for much longer, including records belonging to children, mental health patients, and oncology patients.
Dr Dugar feels the eight-year cut-off point worked well in the film packet era, and there is no good reason to change it in the digital age.
Shannon Werb, global director of clinical strategy for Perceptive Software says that clinicians rarely need to access images that are more than six months old, and that after six months “the access rate falls off a cliff.”
After three years, he says, the recall rates go down to nearly zero. Data protection legislation states that personal records should be kept no longer than necessary – a good reason not to keep images beyond the period recommended in the RCR guidelines.
Werb, who has discussed this issue with healthcare providers in different countries, says there is no universal agreement, however.
Typically, clinicians tend to favour keeping images for several years, arguing that this will enable them to provide the best care for their patients, while legal teams often argue for a relatively early cut-off point.
Werb points out that diagnostic tools improve all the time, so in five years’ time it could be possible to identify a problem on an image that was not visible originally; leaving a hospital vulnerable to legal action.
“Hospitals are grappling with IT and legal drivers towards the reasons to cull the data versus the clinical reasons to keep it,” he says, adding that when hospitals make policy on information lifecycle management, all parties (clinical, legal and IT) should be involved.
John Davidson, market development manager at GE Healthcare IT, concurs: “Getting an agreement between users is the hardest part.”
The technical challenge
Agreeing policy is only the first part of the challenge. The second is working out how to implement it.
The Scottish National PACS Clinical Advisory Group, for example, currently has a retention and deletion policy that cannot be fully implemented.
The reason, says, Andrew Downie, consultant radiologist at NHS Greater Glasgow and Clyde, is that the PACS used in Scotland cannot handle complicated rules that allow for protection of the images of regular attenders, or paediatric and cancer patients. Neither does it have any mechanism for identifying patients who have died.
PACS vendors have been apparently reluctant to build information lifecycle management into their products.
The solution may instead lie in vendor neutral archives, which brings together images and other unstructured data (such as referral letters) from different disciplines.
VNA vendors are now incorporating the ability to set deletion rules within their software. According to Davidson, GE’s Centricty VNA allows the creation of up to a 100 different permutations of rules that will make sure the data is deleted at the right time, provided the correct information is maintained in the DICOM record.
The advantage of using a VNA to manage deletion rules is that it can implement centralised policies for the whole trust, allowing for a much more coherent approach than attempting to implement individual policies in separate PACS systems.
This centralised approach has been helped by the Imaging Object Change Management profile, developed by Integrating the Healthcare Enterprise, which enables one imaging system (such as a VNA) to send a message to another imaging system (such as a PACS), giving the status of a particular image.
So if a clinician tries to call up a deleted image on the PACS, instead of an error message, they will see a message stating that the image was deleted on a particular date, in accordance with policy.
In Doncaster and Bassetlaw, says Dr Dugar, one of the requirements of a new VNA supplier was that the VNA should pass a message to the PACS giving the proposed date of deletion.
“The clinicians can always see the date of deletion, which is displayed on the PACS system,” she says. “They know that an image is going to be deleted on such-and-such a date, so if they have a patient with a slow-growing tumour and they would like the images to be kept for longer, then they can ring up the PACS office and ask to extend the date of deletion to another 20 years.”
Thinking about lifecycle management
The systematic deletion of old data doesn’t have to be the whole story when it comes to information lifecycle management.
Trusts can also adopt tiered storage, enabling them to archive old data to cheaper storage media, which allows for the possibility of accessing it at a later date.
Or they can do what NHS Scotland is doing, and compress older data, freeing up storage space. Scotland is looking at using “lossy” image compression, which means that some image data is lost permanently, but not to such a degree that the images will be rendered clinically useless.
Considerations about how data is going to be managed in the long term need to inform the discussions with the vendor at the beginning of a new contract.
Davidson says that some trusts on managed service contracts negotiate a fixed price for a number of years, and therefore find that, even if they decide to delete old data, they continue paying for the storage.
Information lifecycle management is a problem that isn’t going to go away, and trusts need to begin thinking about how to address the clinical, legal and technical challenges.
Werb’s key piece of advice is to think about the management of data from an enterprise perspective, rather than in terms of decentralised silos. “If you think about it that way, it becomes easier to imagine what it would take to build policies to manage the protection, retention and ultimate culling of your data.”
There will be a workshop on lifecycle management as part of the RCR Imaging Informatics Group Annual Meeting that will be co-located with the EHI Live 2013 conference and exhibition at the NEC in Birmingham from 5-6 November. The meeting will discuss many aspects of PACS, RIS, image sharing, and storage, and is open to members of all disciplines with an interest in the burgeoning area of imaging informatics. This year’s conference is free for all visitors to attend.