Refreshing archetypes related to pathology reporting

Clinical
,
1

Hi!

@heather.leslie, @SDubois, I and others have exchanged som emails regarding refreshing archetypes related to pathology reporting and agreed it would be better to continue (and repeat some of) the discussion here on discourse so that more people can join.

Starting question: What is the status of the project Pathology Synoptic Reporting?

Background: Several pathologists in Sweden are interested in shared development of structured data entry form parts for the pathology investigation and reporting process. A Swedish proof of concept (PoC) project is currently investigating how an openEHR CDR (EHRbase) can be used as a shared read/write-documentation hub for pathology lab systems (from TietoEVRY) and digital imaging/analysis/PACS systems (from Sectra) and possibly EHR systems. A national cancer registry is also involved in the project. During this work e.g. both @SDubois and @elham2222 found things that could be improved in the current set of related archetypes. For examle Stefan posted a mindmap at https://ckm.openehr.org/ckm/archetypes/1013.1.386/resourcecentre with parts colleced from different international reporting projects.

2

Erik asked me to post my email response


The Pathology Synoptic Reporting project is ancient (2009) and the models are likely to need considerable revision, given pattern evolution since they were drafted and in unfinished review limbo. We could/should consider setting their status to rejected - to discourage active use and create a clean slate to work on reimagining the domain and patterns without being unduly influenced by the existing work. They can still be found with CKM when searching for archetypes in the ‘rejected’ state if we need them for reference. Then we can rejuvenate the Project name/metadata etc for our purposes.

Stefan’s work, plus others who seem to be working in the area in India etc, will be integral to due diligence on the ‘whole of domain’ that will allow us to distil the clinical concepts and repeating patterns that will inform the archetype design. This is potentially a massive project – even to limit scope to one or two cancers requires a deeper and broader investigation to ensure that we have a plan that can cater for future additions. (At least as best we can do without a crystal ball). There is definitely an ‘art’ to deciding how deep and broad we need to go before doing the deep dive design for one cancer.

The OBSERVATION.lab_test_result was designed to be the anchor and generic framework for all lab-related modelling. The current anticipated pattern will require devising a ‘histopathology/anatomical pathology’ framework as a core/universal CLUSTER to nest inside the OBS. This will support more detailed CLUSTERs of the nature that exist in the existing project. I suspect we will revise the concepts considerably if we cast a wide lens on the variety of histopathology reporting requirements, and the new CLUSTERs will be required. You can see a parallel family of genomics archetypes and example templates, designed on these same principles, here: Clinical Knowledge Manager

The same archetype design pattern is intended for modelling microbiology, serology, cytology etc. subdomains within the Lab test result family – a core framework CLUSTER for each subdomain, that will carry more CLUSTERs containing the specific details. This pattern is also underpinning the Imaging result family of archetypes.

Stefan and other domain experts can guide us, but I would imagine using https://documents.cap.org/documents/synoptic_reporting_definition_examples_v4.0.pdf , or similar, as the basis for the histopath framework CLUSTER to ensure that we get the core reporting framework as standardised as possible. Then we can hopefully discern modelling patterns across all cancers, also considering that lymph nodes and metastases and details such as resection margins etc should be good candidates to model for broad reuse across all cancers.

I’m sure it is much better to continue this discussion on Discourse. My concern is that there are a number of people who have already been working in this area, without understanding how to align with these patterns and we need to create a forum to bring this work into one central communication collaboration (ie at announcement/major communication) point. Slack, or similar, may be better for daily/weekly working interaction.

Just thinking out loud. All thoughts welcome.

Editors/CKAs need to help shape this, but we can support you to drive the detailed design and reviews if you’d like.

Cheers

Heather

1 Like
3

This was discussed again in the Swedish project and we’ve got some pathologists willing to help and will also look into funding of some consultancy hours from international archetyping experts.

We’d suggest trying to create a first verison of a general cluster archetype to be used together with a first specific cluster archetype for at least the “micro” part of breast cancer. Then a second specific cancer “micro” archetype could be added and/or heading over to the “macro” part of breast cancer.

Who else in the openEHR community would be interested in getting this started?

2 Likes
4

I agree that designing a generic micro and macro examination findings would be a really valuable starting point.

It will take quite a bit of research/investigation to discern the common patterns from amongst the variations across the common use cases. However I find it is usually worth putting in the effort in the beginning as it means the review process is more a refining than redesigning.

I’d be interested to participate


Cheers

Heather

2 Likes
5

An update from Swedish pathology work:
There has now been some serious (non-openEHR) modeling work done regarding breast cancer in Sweden (e.g. concept- and information-models) by @sanna.asberg, @LindaAulin and others.

Karolinska University Hospital has been asked to see if/how that can be represented in openEHR so that the coverage and usefulness can be tested/evaluated using e.g. openEHR template based forms.

For a first (too tighly time-limited) proof-of-concept (POC) version we’ll likely have to include the “paused” openEHR-EHR-CLUSTER.microscopy_breast_carcinoma.v1 archetype and some other not so stable/finished or homegrown archetypes, but we hope that we via the POC can feed some comments back and gain experience regarding what would be gereral vs specialiced for different cancers. (sterting with breast cancer). Some of that division/generalisation has already been considered in the national (non-openEHR-based) work.

We’ll get back with info here when we have something more to share. Meanwhile it would be interesting to hear about similar work being done elsewhere.

1 Like
6

Definitely,

There is work in Wales @johnmeredith and other places on the same track @varntzen @bna

see Cancer treatment in general, Particle therapy, OMOP, DigiOne, European Particle Therapy Network - #9 by ian.mcnicoll

Personally I would start at the other end @Erik and work down from a broad pattern for anatomic pathology, which is largely understood but which we have never formally agreed on. Then tackle some other generic pattern issues - how to do staging in a way that works for both national reporting and MDT support. That may help guide patterns around direct invasion, tumour margins etc. After that the very detailed requirements that are e.g. unique to Breast cancer microscopy will fall out naturally.

Or perhaps a twin track - one to see how much the clinical requirements for microscopy still line up internationally (essentially a clinical exercise) vs. the work above which is really more informatics.

7

Yes @ian.mcnicoll working from the “broad end” would be our preferred way too, but I suspect we’ll have to do things partly in the “twin track” way in order to get our first POC verison out in time. Then we’ll want to iterate towards more generic patterns when there is some international consensus (hopefully before filling our CDR with real patient data based on this).

We hope to be able to contribute a bit to that consensus at the same time as POC:ing.

2 Likes
8

Hello everyone,

My name is Marlene, I’m fairly new to the community so I have tons of questions and no answers, but I would like to collaborate with your project @erik.sundvall. I am currently working in Wales where we are also trying to model our ‘Cancer Datasets’ for reporting and we have come across the same difficulty that you have.

Currently, we have created a generic cluster archetype for microscopy details and we are trying to get it into the best shape possible but we know we re working based on a reduced number of cancers at the moment, and might need expansion. I don’t know how to share the archetype here to see if it could be a starting point.

I looked at the mind map, and I was wondering if you were planning on creating a specific cluster for each tumour site or build a generic and plug in smaller clusters with very specific details? As well, I see that you have included bilateral as a laterality, which isn’t in the openEHR laterality archetype. How are you handling that?

Out of what we have analysed, there are two points that we have identified as most challenging. First is the differentiation between solid tumours and haematological tumours as these have different pathology, location, staging, treatment. etc; the other are sarcomas, which can also be located anywhere and their classification varies. These also happen to be the most frequent tumours in childhood. Have you done any specific work on either of these?

From a pathological point of view, would it make sense to have Elements for nucleic markers, cytoplasmatic markers, etc. to be able to reuse a generic cluster with whichever tumour? In our model we have used ‘runtime constrains’ for nuclear expression, but it might be ‘extendable’ for other uses.

As well, we are starting to do some work on tumour progression/metastasis (including lymph nodes)/transformation. Regarding this, it would be interesting to discuss if that ‘classification’ would need to be done at a higher-level archetype and then reuse the same microscopy clusters under that heading, or if it warrants a separate archetype to record how it was detected, when, etc.

Thanks for attaching: Clinical Knowledge Manager (openehr.org).

@ian.mcnicoll knows of our struggles :joy:. It would be amazing to work on this together to align how we record this.

Thanks,
Marlene

4 Likes
9

Welcome Marlene, Good to see you here!!

To make your work visible , there are a couple of things you can do


  1. Just upload an archetype exported as adl. nTHe only issue here is that people cannot easily visualise it.

  2. Send a link to the archetype or template from AD using the share button to the left of te archetype or template name

image
image714×336 52.7 KB

THis allows you to share a read-only view
e.g.Archetype Designer

@erik.sundvall - Perhaps we could setup a shared Github repository and associated documentation?

10

Hi all,

Hope this helps:
openEHR-EHR-CLUSTER.microscopy_dhcw.v0.adl (github.com)

I cannot share it that way, I guess because we are using GitHub, but that is the link to the repository.

And this is a screenshot of the model so far:

image
image718×864 37.4 KB

These are the current runtime constrains:

image
image693×523 13 KB

And this is the adl:
openEHR-EHR-CLUSTER.microscopy_dhcw.v0.adl (6.0 KB)

Cheers,
Marlene

2 Likes
11

Hi Marlene,

Thank you for sharing. I noticed you created a Cluster Archetype from scratch. I found this archetype: Clinical Knowledge Manager, which might be suitable for IHC - Its purpose is:

“To record a single value laboratory analyte result, commonly found in clinical pathology testing.”

The new archetype could be named for instance “MLH1 nuclear expression”, each IHC would be an archetype and these could be clustered in for instance “Immunohistochemistry results” and added to the respective Microscopic finding e.g. Clinical Knowledge Manager.

Since I am new to openEHR as well, I would appreciate feedback from the community.

Best,
Maurice

12

Hi Maurice! Just as a quick note, I’m not sure you saw the misuse section of the Laboratory analyte result archetype:

Not to be used to record anatomical pathology macroscopic/microscopic findings, other than for additional testing such as cytometric flow studies.

In general, modelling for anatomical pathology has been immature and not a core focus for a number of years. It looks like a lot of different people are focussing on it now though, which may be a perfect chance to improve the situation and create some published models which can be used by everyone :smile:

1 Like
13

Thank you, @siljelb you are right, I overlooked it.

An archetype that I’ve frequently observed being used is OBSERVATION.lab_test.histopathology. However, I can’t seem to find this in CKM or Archetype Designer. It only appears in the “use/misuse” section of other archetypes.

14

The regular OBSERVATION.laboratory_test_result is intended to be used for anatomical pathology results also. I haven’t seen this specialisation you’re referring to, but initially there shouldn’t be a need for one (I’d be happy to be proven wrong though). Details about the tests/examinations and results can be modelled as CLUSTER archetypes and inserted into the standard OBSERVATION in a template.

1 Like
15

Screenshot 2023-08-31 at 11.45.28
Screenshot 2023-08-31 at 11.45.281753×1240 250 KB

16

Hi both,

That archetype has been rejected Clinical Knowledge Manager (openehr.org)

You can find it by clicking on ‘Lifecycle’ →

image
image462×617 23.3 KB

I was looking for this archetype in the CKM a little while ago because I could download it to the AD but I kept getting an error message, and I was pointed to this section.

It is odd that the AD allowed me to import it, but the warning and that the template kept ‘braking’ prompted the question.
image

1 Like
17

@mar.perezcolman, a tip whenever finding an deprecated or rejected archetype, go to the Archetype history
image
and see the Log message when it was rejected/deprecated. It should tell why, and also point to another archetype. In this case the Log message says: “Old pattern, superceded by OBSERVATION.laboratory_test.”

3 Likes
18

Well, if one would work with it to improve or correct it, then not allowing to import it to AD would be wrong. But maybe AD could give a warning? “You are about to import an rejected archetype. Proceed?” or similar.

1 Like
19

As one of the original authors I just want to confirm that the histopathology archetype mentioned is not the direction of travel and I agree with @Silje and @Heather that the Anatomic pathology work should be based on the current generic OBSERVATION.laboratory archetype. There is a pattern applicable to anatomic path that can be added re Macrosopic, macroscopic, intraoperative that we can build from, and I think is still an open question on whether this should be a specialization and not just a looser templated pattern. But we can build from there.

Also, just a note on the background to the archetype that Marlene posted.

It was built to reflect the needs of Welsh Cancer Standards reporting so is a somewhat ‘flatter/simpler’ view of the world than we will need for Anatmic path reports and Cancer MDT support, where e.g staging and ‘nodes’ is way more complex and tumour specific. This seems to be a pretty universal issue.

We tried to solve it with the original set of Anatomic path archetypes by using largely generic representations of e.g ‘nodal spread’ then templating out of specific cancer types but this became pretty messy. I think we might be better to accept that whilst there may be some gneric ideas like nodes examined/nodes positive, that these will only really ever be used for reporting, and that further detailed per-cancer nodal spread clusters will be required.

I think there is enough joint interest and momentum her to perhaps arrange a joint bit of a ‘show and tell’ call perhaps in latter part of September?

4 Likes
20

Wonderful to see all the action in this thread!

If you want some scandinavian input and don’t mind translating via Google Lens or similar transalation tools you can have a look at:

  • The Norwegian template at https://arketyper.no/ckm/templates/1078.60.792/orgchart that seems to have a pretty nice structure for creating a complete report and clarifies where smaller clusters etc may belong in a larger structure. Does anybody know of similar template examples?
  • A work in progress snapshot of Swedish models, currently focused on the microscopy part, see below:

@ian.mcnicoll a a ‘show and tell’ call sounds like a nice idea! Does the Clinical Program Board (ping @joostholslag @vanessap @Paulmiller) and/or super-experienced modelers like @heather.leslie @siljelb or others be involved somehow?

Suggestions/thoughts:

  1. refreshing/creating a general archechetype for microscopic findings might be a good start
  2. and then (or simultaneously) testing compositional design patterns and/or specialiations to cater for breast cancer specific needs and perhaps some other cancer type (depending on participants’ interests).
  3. refreshing/creating a general archechetype for macroscopic findings
  4. 

2 Likes