I was wondering if anyone is interested/involved with reporting of cervicovaginal smears. A quite mature and stable micro terminology exists that is aimed to provide effective communication between gynecologist and cyto-pathologists and for reporting of cases: Bethesda System 2001. It had been discussed in 2001 at NCI and published at: http://bethesda2001.cancer.gov/terminology.html
I had modeled it with classical relational method and have been using it since 2002 as an add on to my Anatomic Pathology Information System. The reason I am writing this message is that I personally believe it is a good candidate for Archetype modeling. Well I had done some initial modeling but it is obvious that first it has to go into an appropriate EHR ‘slot’ (i.e. as a specialization of plain pathology/cytology report). And secondly my experience is that it has to be extended so as to include some practical information like type of stain used or number of glass slides and so on. Archetype specialization is a good method to embed such auxilary information items while keeping original structure and semantics of the standard. I see a huge benefit for NCI or other related organizations first to get aware of Archetypes and then adopt the methodology for proper domain knowledge governance.
I would love to get your comments and contributions to design a really usable Archetype for reporting of cervicovaginal smears.
Best regards,
Koray Atalag, MD, Ph.D.
Freelance consultant and developer
Hi,
I am Dr.Shivam and am on the NHS CFH program and very keen on working on the reporting of cervico vaginal smears. My wife is a gynecologist and together we have also been trying to find a good EHR format for reporting the smears…will like to know more about what you have done.
shivam Shivam Natarajan, MS FRCS | perot****systems
Physician Executive, Senior Consultant Transformation
: 07920566081
Mail: Shivam.natarajan@ps.net People - Process - Technology - Results
Well now I am fully convinced that either for cervico-vaginal smears or for modeling of any other mature clinical terminology/ontology openEHR archetypes are number one method of choice. In fact I did some research on implementing Bethesda System 2001 and presented as poster in European Cytology Congress back in 2002. The link to the poster abstract is:
In summary it models the Bethesda System 2001 by using XML and ASTM DTDs which works as an add-on to my Open Source anatomic pathology IS.
Well after completing a Ph.D. thesis on Archetype modeling of such a model (endoscopy), I intend to model it by openEHR Archetypes and RM. Of course I want this work to be utilized in real projects/products.
I will post my Bethesda archetype in the discussion list a little later because I am still trying to recover my files from a defective hard disk
Here is the current candidate archetype for Pap smear. Please contact Heather Leslie (heather.leslie@oceaninformatics.com) to discuss the work going on in the NHS and how to look at and work with the archetypes. openEHR is now being used for clinical modelling in the NHS environment.
Now I remember we had a discussion about this NHS smear archetype and how Bethesda format would go together about a year ago. From a clinical point even here in Turkey the surgeons are pushing cyto-pathologists to report cervicovaginal smears by using this system. So I doubt things wouldn’t be so much different in UK or other.
If it is going to be useful I would be more than happy to redesign it as a specialization of the generic NHS archetype. I am attaching the draft archetype.
Best regards,
Koray Atalag, MD, Ph.D.
Freelance consultant and developer
This is a good example of an area that will probably differ depending on practice nationally. The archetype you have sent is probably not the ideal mix of classification, terminology and structure. It is worth considering that the terms used to describe the findings will vary - but probably not as much as we might think - so we need to rely more on terminology rather than booleans. Also, as the classifications change, ideally we will not need completely new archetypes (unless there is a major change in the manner this is reported).
The openEHR archetype is a specialisation of histology observation - this is probably important as there are a range of findings that might appear that are general in nature and will fit the cytology model.
Finally, the specimen quality issues need to be part of the general histology archetype.
So, it would appear important to find someone who is working on the next generation of cytology reporting (the current archetype is based on recent classifications in Australia) and keep the interface between terminology and archetype optimised for general applicability. One way to do this is to leave the archetype as coded with no data and use templates for national sets. The other is to go for European/International agreement.
The archetype definition is a potential change agent in this space - but we need to get buyin at the appropriate level. This is probably one of the most difficult on the planet!! But why not start there?
I have just joined Ocean to help with some of the NHS work and may be able to help with this area.
Many years ago I designed and developed a cervical smear report and recall module for one of the Scottish GP systems and have recently been working with the Scottish NHS (now radically different from the English NHS) to develop messaging standards for the transmission of cervical smear results to GP systems. We did a fair bit of Read code and SNOMED-CT binding and have experience of some of the tricky aspects of smear reporting like handling recall advice, exclusions and diagnostic information.
Well I am really surprised about the reporting of cervico-vaginal smears in UK and Australia without taking into consideration Bethesda. Indeed it is very interesting to observe such national diversities. To my knowledge, the surgeons are planning the type, timing and extent of the surgeries according to the precise results of this report.
I must admit that I could not get exactly Sam’s point about mixture of classification, terminology and structure. Maybe you might want to consider that in terms of terminology, whole Bethesda System is already represented in SNOMED (or at least in the version I am using: SNOMED III Intl. Pathology Microglossary). And the structure is depicted explicitly and unambiguously in the standard. And that it had been developed by an international panel of experts and a consensus has been reached. The 2001 version was just a minor revision to the former one.
We have also extended Bethesda System 2001 so as to make it more intuitive and clinically useful - like Archetype specialization without breaking original semantics and constraints. Additional information such as the specimen quality or the type of specimen itself (PAP Smear, liquid based or other) and SNOMED codes are now in it. I can send the working version if requested.
My aim is to create awareness about the use of this standardized ‘micro-ontology’ in reporting of cervico-vaginal smears. I recommend that it should be expressed either as a specialization of generic archetype or as an alternative and be posted into an appropriate public Archetype repository.
It would be nice to hear from others living in countries other than UK, Australia and Turkey.
Best regards,
Koray Atalag, MD, Ph.D.
Freelance consultant and developer
I am continuing to write to this thread because of a related discussion in the technical list. In summary we have formerly discussed here about two archetypes for cervico-vaginal smears, existing NHS and Bethesda. We did not reach any conclusion so I though about having a highly specialized archetype which could conform to both parents. This brought about the issue about multiple inheritance and decision was taken collectively to avoid it.
So here are some key requirements to be considered:
Multiple inheritance of archetypes should be avoided
Different archetypes for same purpose should better be avoided
The variation in the terminology in alternative archetypes should be minimised
The number of nested specializations should better be limited to 2-3
As Sam had already suggested, in fact the two archetypes are indeed very much alike; but not in a way that they can be specialized to the other. However after a thorough examination of the parent Histology archetype which specializes to PAP one, it is possible that it can also specialize to Bethesda. So as far as we have one common parent, the problem resolves easily.
However, I believe that a single Cervico-Vaginal Smear or simply PAP Smear archetype can be designed by incorporating some of the uncovered items from Bethesda to existing one. And I guess by making certain uncompatible nodes (such as Category in PAP and General Categorization in Bethesda) mutually exclusive (i.e. cardinality 0..1 and two nodes with 0..1 occurrences) both can be represented in same archetype. My knowledge on templates is very limited but I assume they can be further constrained for local needs during implementation. This would be useful to limit the number of archetypes in same ontologic domain serving for same purpose. Otherwise I do not see any problem in having alternative archetypes.
There is also another important point: Bethesda archetype is not usable clinically for report generation as is because it lacks certain information such as ones in protocol or quality/adequacy sections. So it means it has to be further specialized. Keeping in mind the expert recommendation to keep number of specializations down to 2-3, it might be better to have them separately.
One last remark is that although aiming to minimize the total number of archetypes, I would recommend to make distinction between at least histo-pathology and cyto-pathology (never mention immunohistochemistry, cytogenetics or electron microscopy Anatomic Pathology might be a more appropriate name to embrace both domains.
Thanks Koray - I am going to narrow the discussion to the clinical…
I believe that a single Cervico-Vaginal Smear or simply PAP Smear archetype can be designed by incorporating some of the uncovered items from Bethesda to existing one. And I guess by making certain uncompatible nodes (such as Category in PAP and General Categorization in Bethesda) mutually exclusive (i.e. cardinality 0..1 and two nodes with 0..1 occurrences) both can be represented in same archetype.
This is an interesting issue - I think these are actually the same things - the broad categories that provide the means of triggering the next level actions. I wonder if we should be allowing renaming in different languages e.g. en-AU and en-US for the labels applied if they are the same - this is what we do in Apgar for instance but there the ordinals 0,1,2 are set. We could have a more extensive set of categories and then use templates to allow choice of the ones required for local use. We could leave it for specialisations to provide the terms within this part of the archetype ie non in the root - this would have the advantage of allowing use of SNOMED or terminologies in the different jurisdictions. The problem with this is consistency across jurisdictions.
My knowledge on templates is very limited but I assume they can be further constrained for local needs during implementation. This would be useful to limit the number of archetypes in same ontologic domain serving for same purpose. Otherwise I do not see any problem in having alternative archetypes.
Yes - it does allow that
There is also another important point: Bethesda archetype is not usable clinically for report generation as is because it lacks certain information such as ones in protocol or quality/adequacy sections. So it means it has to be further specialized. Keeping in mind the expert recommendation to keep number of specializations down to 2-3, it might be better to have them separately.
I think templates would allow this.
One last remark is that although aiming to minimize the total number of archetypes, I would recommend to make distinction between at least histo-pathology and cyto-pathology (never mention immunohistochemistry, cytogenetics or electron microscopy Anatomic Pathology might be a more appropriate name to embrace both domains.
This may be correct - just that our analysis of reports in these two was almost identical.
Anatomic Pathology might be a more appropriate name to embrace both domains.