posted on Saturday, November 30, 2013 - 05:36 am
1. Is anyone using NHS number as a standard DICOM tag? 2. Is a second UID supported by modalities and PACS (we need to use PAS ID as our primary ID)? 3. There will be times when the NHS number field maybe blank.
I would be grateful for any comments, views etc
posted on Saturday, November 30, 2013 - 03:00 pm
In the Netherlands, IHE-nl and the national healthcare IT organisation Nictiz have indicated the preferred way to add a second Patient ID to DICOM objects. This is done by using the Other Patient ID's Sequence (0010,1002). An item in this sequence contains a Patient ID (0010,0020), and by means of other attributes you can indicate the nature of the ID. This is all described (in English!) in this document:
As Stephen Brown described therein, quoting MIMA, the Other Patient IDs Sequence is the way to go, using Issuer of Patient ID +/- Issuer of Patient ID Qualifiers Sequence.
Note that though there is an OID for (new) NHS Numbers issued by HL7 UK (2.16.840.1.1138184.108.40.206.1), I don't know if anyone has standardized the string to use for Issuer of Patient ID (i.e., is it "NHS" or NHSNMBR", or something else, or should one put the OID there (like the Dutch BSN document requires), or should one always use the OID in the optional Universal Entity ID too, just to be sure ?).
Pim, do your systems deployed in the UK use the OID in Issuer of Patient ID?
All this needs to be mapped reliably to HL7 PID-3, so presumably the PAS/HL7 folks have already figured all this out (what values to send in which field)? Surprisingly, in all the old "NHS Number is great" documentation, I could not find a specific reference to HL7 PID fields and values, despite the detailed functional requirements.
Also, I was surprised by Neelam's comment in the earlier post about using the local PAS number rather than the NHS number in XDS metadata.
This might be expedient from a single site's perspective, but seems to be a step backward and rather defeats the point of the "cross-enterprise" part of XDS, and precludes a "national" registry.
The concept of identification across a single affinity domain with XDS is to use the same identifier across all the sites in that domain. Without that, you need to resort to XCA and XCA-I with gateways to handle mapping of identifiers from domains beyond the scope of one PAS numbering system. By contrast, the Dutch approach uses the BSN identifier for XDS, I gather.
posted on Saturday, November 30, 2013 - 05:09 pm
>Pim, do your systems deployed in the UK use the OID in Issuer of Patient ID?
I'm afraid they don't. The subcomponents 4 and 5 of PID-3 values are used to populate the database. But I'm sure one day we'll include a mapping and go the OID way.
>defeats the point of the "cross-enterprise" part of XDS
This may be caused by the idea that "an XDS infrastructure can be the Vendor Neutral Interface to a VNA". Lots of people are considering standardized archiving of non-DICOM medical images and documents, but if you want to exclude a proprietary system, it's hard to resist the temptation of using XDS.
*** commercial interest*** I have made our PACS accept any document through XDR, the documents can be retrieved through WADO and be found with QIDO when that supplement is Final Text. In this manner you get the good things of PACS (like study root queries) without the bad things (DIMSE as opposed to http). I hope that this type of solution will also be considered in the future when there is a requirement to archive non-DICOM images.
>the Dutch approach uses the BSN identifier for XDS, I gather
That is correct. But this has to be considered form the legal perspective that requires Dutch institutions to use the BSN whenever medical data is exchanged, whether this is done through XDS, a CD with images, or, say, an email.
posted on Saturday, November 30, 2013 - 05:57 pm
"In the Netherlands, IHE-nl and the national healthcare IT organisation Nictiz have indicated the preferred way to add a second Patient ID to DICOM objects. This is done by using the Other Patient ID's Sequence (0010,1002). An item in this sequence contains a Patient ID (0010,0020), and by means of other attributes you can indicate the nature of the ID. This is all described (in English!) in this document: "
Pim and David, Sorry if this is a really stupid question.
I know that our HL7 ORM message to the Rogan DMWL Broker (which provides DMWL to modalities) contains both PAS and NHS nos. However, modalities only query and store PAS no. Hence, images in Fujifilm pACS only have PAS no in the DICOM tags. Although Fujifilm does store NHS no. Within their database -probably through ADT feed. If we wanted the NHS no to be within the DICOM tag do we simply need the modality vendors to change their DMWL query? Or is there something else we need to do?
For anyone wondering, our "to be" RIS provider Rogan is actually providing DMWL. We had a choice to use DMWL from Rogan or Fujifilm --we felt RIS is better placed for providing DMWL.
posted on Saturday, November 30, 2013 - 06:11 pm
>do we simply need the modality vendors to change their DMWL query?
In my experience, it is not always simple to change the behaviour of a modality. Rather than change the queries of N systems that may or may not be easy to change, changing the PACS should be the easier part. Since Fuji receives the NHS numbers in the ADT's, it should be able to add the NHS number in a standardised way to outgoing images that did not receive this number originally from the modality. It's a variation of the pattern suggested by the IHE SWF/PIR profiles: ADT messages from the DSS/OF to the IM/IA influence the attributes of instances going out of the IM/IA.
But the Rogan MWL server could indeed serve the NHS number to interested modalities.
posted on Saturday, November 30, 2013 - 06:22 pm
"Also, I was surprised by Neelam's comment in the earlier post about using the local PAS number rather than the NHS number in XDS metadata."
Hi David, This was based on advice from our technical architects. From what I remember there was a good reason for this
We want to walk before we run . 1. Ist step--DICOM images with HL7 v2 messages reports linked reports. Move from Agfa PACS to Fujifilm PACS. Milestone success. 2. 2nd step--replace RIS from McKesson to Rogan-Canon. Due to happen soon. 3. 3rd step--Link DICOM images with pdf encapsulated CDA request cards and reports within VNA using XDS standards. I.e move from messaging to document standards. 4. 4th step include other specialty images and document --with an Enterprise XDS/XDS-I viewer for Multispecialty display. 5. 5th step -transmit images and documents to other NHS hospitals using XD* standards. I can see IEP being the logical way forward -they have incorporated XDS registry and a PIX manager within their architecture. Multi-specialty enterprise wide sharing of images and documents is hugely important for patient care and quality.
posted on Saturday, November 30, 2013 - 06:44 pm
Thanks Pim. Your suggestion will not help.
We need the NHS number to in the DICOM tag within Fujifilm PACS. We wish to the "Common View" feature of Fujifilm.
We have a Breast Screening PACS from Visbion which uses NHS no as Primary ID. Fujifilm PACS uses PAS no as primary ID. Our breast radiologists would like to use the common view feature of Fujifilm to see all the imaging instances saved in Visbion within Synpase. I am told that this requires the PID within the DICOM tags.
for the SWASH consortium (Salisbury, IOW, Southampton and Portsmouth) we also use the hospital number as the primary identifier. However, we also recognised the need to have the NHS number available when imaging is displayed in PACS, so we elected to use the 'Other Patient ID' tag in DICOM. Our MWL supplier populates (0010, 1000) with the NHS number, and that is returned from *most* modalities. This aligns with the data from our LSP PACS, which also used that same tag for the NHS number, so when displaying migrated imaging as well as newly acquired imaging, the NHS number can be displayed.
posted on Tuesday, December 03, 2013 - 05:53 pm
...works very well in Scotland (CHI number)
In Sunderland we used a social security DICOM tag for NHS no, and mapped the ORM messages to PACS to support this (well it is an American product!)
The days of local ID's are over....
posted on Tuesday, December 03, 2013 - 05:55 pm
Should have added, not all PACS systems support more than 1 master Patient Index...
posted on Wednesday, December 04, 2013 - 03:52 pm
The NHS number is not a DICOM problem, it’s not a Radiology problem, it’s not even an enterprise problem –it is resistance to change at a national level. Most LSPs trust the RIS ID is the PAS ID (with a ODS prefix), simple for the clinician, in the wider enterprise, to mentally strip off the leading characters to find the equivalent PAS ID on another system. But, that does not work for a patient ID at a national level. The RIS ID is just a number; it could simply be replaced by a different unique number (where available the NHS number). That is the first of the challenges – what to do in the absence of a NHS number. Internally, the RIS has a mechanism to ensure that it only generates unique (incremental IDs). It now has to apply some intelligence to swap around the numbers and send updates to the downstream systems or, more significantly, the upstream systems. The second issue is that now Radiology is operating with a different number to the rest of the hospital (the very reason why RIS ID were slowly replaced with PAS IDs as the primary ID in the past). A clinician, within the enterprise, will have to do a look up between PAS and RIS ID. Some systems may be able to display and sort by either number, but this is not always the case. Before you know it, the preferred filter becomes the date of birth and last name – easy but not 100% fool proof. So, the PAS ID is the root of the problem. Again, it is just a number; with a quick database update, it could be replaced by the NHS number (where available). Then the change could be cascaded to all systems as their one and only ID (the PAS ID), just as it is now, except in the case of A&E, NM, breast screening, obstetrics, etc. As it is for RIS and the temporary PAS ID, in the absence of the (validated) NHS number, the PAS has to generate an alternative unique ID and perform updates and merges as the NHS number becomes available, and cascade this change to all systems. This is why the simple option is to internally generate an incremental PAS ID as the Trust global patient ID. The NHS number is no longer an indication of ‘eligibility for care’. In theory, it is available to all but, in practice, there must be a mechanism to quickly issue an ID to overseas visitors that require urgent care or to other patients where the NHS number has not been issued, traced or validated. This may include Armed Forces Personnel and Service Dependents. Tracing can be performed by the Personal Demographics Service (PDS) in batches, or for a single record using the Summary Care Record Application (SCRa). This article is not about issuing the NHS number, it is about using it. So, my proposal is not to try and find ways to make DICOM fit the UK problem of two primary IDs (local PAS and NHS), but to replicate what is already common practice between PAS and RIS IDs. When an urgent non-scheduled case comes in to Radiology, an order cannot be placed if the patient does not exist on RIS; a request is made to create a PAS ID and pass that through to RIS. This may not be possible out-of-hours and so a common solution is to have a small number of holding PAS IDs on the system. When required, one of these numbers is used (on RIS) and the demographics passed to the PAS team to update the PAS ID (the updates are then cascaded down to all systems). Why is it not possible to issue a block of temporary NHS numbers to each Trust, thus ensuring 100% availability of a NHS number to be used as the PAS number? When these temporary numbers are used the details are sent to the PDS where they will either identify the correct NHS number or create a new one and send a merge to the Trust to merge the temporary number into the validated NHS number. Simple – all systems use the PAS number, no change here. The PACS (and DICOM) has only one primary patient ID, the PAS number. Most importantly, it is a nationally unique, yet common across all systems and all enterprises. What DICOM might need is a ‘NHS number status’ tag to pick up the NHS number validation status from the HL7 message, or at least a flag to indicate that it has been validated, but that is another topic.
posted on Wednesday, December 04, 2013 - 09:40 pm
John, Our RIS--old and new do send NHS no. as ORM messsage to PACS. NHS no. is present in Synapse as a searchable ID. However, this forms part of the PACS database--similar to what you describe in Scotland. It is not sent by modalitities so it is no a DICOM tag-this what Nick is doing in Salisbury.
Shaun is right. The real problem is lack of temporay NHS numbers --which make it unsuitable to be used as a Primary ID. We need sensible strategic thinkers nationally for such changes to happen
For us Visbion is going to add our PAS no to their Breast Screening PACS database--via an ADT feed. So we can then see all the attendances in Visbion from within Synapse for the same patient--using PAS id as the common ID. Thanks for the comments. They are appreciated.
In Scotland the RIS sends the CHI (Scottish NHS number) to the modality, and the modality sends it on to PACS. So the NHS number is the primary ID in the national PACS, allowing us to share images across the country.
Where the RIS doesn't know the NHS number, it sends something else (for CRIS, the CRIS ID prefixed "CRIS") and that is used as the primary ID. The secondary ID is usually the local (originating site) RIS number.
Where the NHS number is later added to CRIS, it sends an update message to the local PACS, which is also passed to the national PACS, replacing the temporary CRIS ID with the NHS number.
When order comms goes live, CRIS will take the NHS number from the HIS, and where the HIS doesn't know the NHS number (should be rare as its also linked to the NHS number database, SCI Store), it will send a temporary number HIS generates. Again when NHS number becomes available all the systems, including PACS will update.
I think I have described it correctly. And it works well, with the occasional exception.
I don't understand why a lack of temporary NHS numbers is an issue - why not use a different ID, so long as it can be updated later?
The Scottish Health Minister decreed that the NHS number will be the prime identifier in Scotland from 2006, and everything has followed from that wise decision. It's a shame this still hasn't happened in England.
posted on Thursday, December 05, 2013 - 07:50 pm
Hi Andrew, So it would seem that you have included NHS number equivalent --CHI as your primary ID within DICOM tag as well as the PACS database. That seems a good strategy.
I remain optimistic here. IEP allows DICOM based image sharing throughout England. Burnbank have incorporated IHE PIX manager within their IEP system from my discussions with them in UKRC. This will allow for matching of patient ID across multiple systems with different primary IDs. I think this will be useful for cross border sharing between hospitals with different PAS ID, whether hospitals use English NHS number or Scottish CHI number.
Our PAS ID will allow for cross speciality sharing of clinical information within our Enterprise through the VNA and context links. Sharing with other hospitals we will use IEP which hopefully become an image and document exchange portal in future---IDEP
Andrew totally agree with you that NHS should be the primary ID.
Our PACs supports Issuers for the PID fields such that we can store unique multiple primary PIDs ie NHS number and Local issuer and external Issuer differentiating between these with a different issuer.
We have been using this as a way to match patients to their records when they store. It works incredibly effectively and is able to check if a valid NHS is provided without a issuer similar to CHI number. It is effectively the same system used in scotland
However there remains an issue
1) Other sites do not send NHS number on outgoing IEP - please please can we all agree to do this and setup our PACs Systems to do so.
2) By some perverse logic IEP do not transfer the NHS number from their database (as required to be input to initiate transfer) nor is this checked against the spine to the DICOM header.
Can everyone please urge IEP/SECTRA to do so.
posted on Thursday, December 01, 2016 - 05:35 pm
Every PACS, IEP, VNA or any other DICOM system must support multiple IDs. PAS ID, NHS no, CHI no using OtherPatientIDSequence DICOM tag.
Hi Dimitri - the IEP has no connection to the Personal Demographics Service part of the Spine so does not check NHS numbers on studies being transferred by hospitals are actually correct (it does not build a database itself from the information submitted on the dashboard - the NHS number or local ID input on the transfer screen is used to query the local PACS and then provide a list of matching patients / studies to choose from). As there is no 'IEP database', including NHS numbers in both the original images DICOM header and also the IEP transfer ticker information is a decision / configuration for the endpoints involved. The IEP team are based in Ipswich and like many others I'm sure would love to have a single identifier used consistently as it would make programming easier! Quite a few major hospitals still use their local ID as a primary identifier for patients though, even when transferring.