The Royal College of Radiologists has opened consultation for the draft Application of the Records Management Code of Practice to Radiology Record Retention Protocols.
Currently, under the NHS Records Management Code of Practice, Radiology records are kept (at a minimum) for varied periods between 8 years to 30 years to “permanently” depending on the type of study, age of patient and situation (legal, deceased, cancer, CJD status, research patient etc.). This can be confusing and awkward to implement automated records retention rules on. No known system implementing these exact rules automatically is available from any PACS vendor in the UK.
Disposal (deletion) of records prior to the current NHS Records Management Code of Practice applicable period expiring is considered unauthorised practice.
In summary, subject to the patient not exercising their rights for earlier removal, the new guidance will recommend to all UK PACS Teams that the retention period for Radiology records moving forward is simplified to the latest of either:
30 years from when the patient was last seen OR
10 years after the death of the patient
This is in recognition that long-term record storage is in the best interests of the patient.
Has there been any consultation with affected suppliers to understand what they can currently do and what they would need to do to make their systems compliant?
This will likely mean an overall increase in the volume of data being stored, and probably also the volume of data being migrated. I agree it is very sensible in terms of simplicity though as many smaller centres can struggle to fulling get to grips with the current rules.
I’m happy to comment unofficially, but it would be good if questions/comments could be submitted formally to the RCR as well, please.
The guidance draws on the underlying Code (https://transform.england.nhs.uk/media/documents/NHSX_Records_Management_CoP_V7.pdf) which all NHS bodies should be adhering to currently. If you already have the systems and/or staff resources in place to calculate the minimum retention date for every record and act accordingly, then that’s great. This does not mandate suppliers implement any new functionality in that regard. That being said, I would be very happy if suppliers did more to support lifecycle management, particularly as regional/national image sharing becomes more prevalent in the near term and we face having copies of copies of copies of referred imaging. That doesn’t however help with getting the initial category flag applied so you/the system can know which of the rules to apply.
On the assumption that most Trusts cannot realistically apply the Code in the detail specified, the proposed simplified ‘rules’ will enable Trusts to meet the standards across all categories of patient record. This doesn’t per-se result in an increase in long-term data storage/migration workload, as you can still choose to apply the shorter periods in the Code instead.
The IG implications are interesting: Minimum retention periods are set in the Code. Maximum periods are a rather grey area, as discussed at lines 43-49.
In terms of other changes to systems, perhaps there needs to be a push by Trusts to include better lifecycle management in future PACS procurement requirements? Data compression is already available, but I believe rarely implemented. Series rationalisation seems a good idea at first, but trying to implement this in a safe way which prevents the loss of key/bookmarked images, measurements/annotations and retains the ability to perform multiplanar reconstructions in future is tricky. And that’s without considering the implications of a source Trust deleting data which another Trust has based a report/annotations/bookmarks on as a result of a referral!
New standards for data lifecycle management is something the College should consider reviewing in future. Having just completed a large data migration, simply trying to preserve audit logs, PACS annotations and teaching metadata has proved extremely difficult/impossible…
@Sean_Key - anyone is welcome to provide feedback and all provided to the College on their proforma is read by the working party. The reaching out to IG personnel was discussed but as there’s no equivalent “Royal College of IG Leads” (as yet) it more relies on them being informed locally. @stephen.gallagher4 - yes - the responses at the time were that none of the ‘usual’ vendors active in the UK had software able to implement the current rules automatically, probably as there is no standardised flagging at an examination level for things such as CJD status, ‘used in research’, cancer patient, litigation hold, unexplained death etc. that the PACS image lifecycle modules could pick up on. They’re all reported to be insufficiently granular due to that lack of information on the existing studies. @John_Hoath - yes it would result in more data definitely being stored (definitely being the key word as many Trusts currently store indefinitely anyway!), but from preliminary research, not a dramatic amount of data at first - the older studies are smaller in size than more modern studies. Some Trusts also did not store thin datasets until recently - the storage burden will probably come later! But then, storage costs keep falling per Tb. The issue I can see though, is increased indexing costs, which don’t tend to reduce wholesale over time. @Stephen.Fenn - I agree - image lifecycle management was always a core part of Radiology in the film days, it’s been oddly neglected by PACS manufacturers in their software despite us in the UK having a longstanding national medical records management strategy!
Thank you for raising this on this forum. This is under ongoing discussion at our Trust. We’re very pleased the RCR are taking the opportunity to provide clarification.
One point that has been raised is the risk of having a policy for medical records (including paper) that doesn’t match the digital imaging retention policy. Imaging that has been kept but without records of clinical assessment to provide context may result in incomplete and/or confusing documentation in event of future investigations. NHS Records Management Code of Practice minimum is 8 years.
Are we mindful of the huge cost of storage for cross sectional imaging over 8 years old that will very rarely be used? Perhaps someone on this group has a close enough relationship with their PACS vendor to look at a large dataset of access audit records for older scans? The vast majority of older scans we keep won’t ever be looked at. As old scans are not used for diagnosis but reference/comparison, I would favour using lossy compression for scans over a certain threshold but this threshold would need to be based on detailed analysis. However, I’ve had reservations from neuroradiologists as degrading the imaging quality may reduce the value for comparison of subtle findings, e.g. MS surveillance.
Hi Mark, very good considerations those. The point about non-Radiological record reconciliation is something I haven’t heard mentioned so useful to feed back. I’ve been exploring the dynamics behind long term storage for a while and had these thoughts on the points which might help your discussions internally:
RE: compression - Scotland (currently a Philips cluster) are a great example, they compressed images (escalating over time). In their tender issued a few years ago now the plan was to discontinue that approach with their new supplier. I would be keen to find any studies that had been written on the clinical experiences during the past ~20 years (and given the compression was across a whole country this would be quite interesting material) but haven’t stumbled on any. The publicly available tender documents give the compression ratios and data volumes if they are of interest.
Costs of bulk long term storage are actually fairly low from a pure I.T. perspective, either for on-prem or cloud. Costs tack on for the indexing (large DB sizes become exponentially more difficult to handle), availability (offline storage or storage on tape / jukebox is ‘cheap’ for Pb’s of data but obviously slow), egress (if in a cloud) and access to the suppliers front-end (some vendors obviously charge, or cost out by total existing archive). There’s also a labour cost in checking the data remains intact and actually retrievable - much like a custodian in a museum!
Do any forum members have any more details on the Scotland experience? It would be useful to understand why they have changed their approach.
For large sites, we still reach a storage threshold every so often and, when this happens, it can be expensive. New storage arrays can run to hundreds of thousands of pounds of investment. If necessary, the money obviously has to be found but I think it would be helpful to recognise the resource implications.
The cloud points you raise will no doubt become more relevant. Egress fees for migration will contribute to vendor lock in and should be considered at the point of implementation (although some PACS that are marketed as “cloud” are really just a server in a third party data centre rather than AWS/Azure/Google Cloud).