Check if BeObservation is a parent of the observation needed in this context

[bdc-ehealth]

Make BeObservation more lax if needed

[bdc-ehealth]

WG: if a LOINC code is used, there are no changes to BeObservation. The decision on whether a LOINC code or a SNOMED CT should be used will be presented to the NRC (David Op de Beeck - Tom Fiers).

[bdc-ehealth]

@tomfiers Asked question to Tom Fiers en David Op de Beeck op 5/12/2022 per mail.

[bdc-ehealth]

answer from Tom Fiers and David Op de Beeck in https://be-ehealth-standards.atlassian.net/browse/ESI-141

From: Fiers Tom <Tom.Fiers@uzgent.be> 
Sent: 15 December 2022 11:52
To: 'David Op De Beeck (SPF Santé Publique - FOD Volksgezondheid)' <david.opdebeeck@health.fgov.be>; Bart Decuypere (eHealth) <Bart.Decuypere@ehealth.fgov.be>
Subject: RE: FHIR: (Labo)Observaties gemaakt door het CvKO   

Het antwoord is wat dubbel, maar lijkt me vooral snomed 

Voor labanalyses, bv colonkanker zit detest bloed in faeces in loinc, anderzijds kan je ook opperen dat snomed screening voor colon neoplasma ook de lading dekt. 

Borstlkanker hoort in snomed of als mammography, of als screening voor breast neoplasm lijkt me meest aannemelijk gezien RX 

Baarmoederhals : lijkt me ook de snomed screening gezien APD 

Mvg 

Tom

[bdc-ehealth]

Proposal for SNOMED CT codes:

• 275978004 (screening for colon cancer)
• 268547008 (screening for breast cancer)
• 171149006 (screening for cervix cancer)

An issue for the NRC was created here: https://jira.ihtsdotools.org/browse/BERMP-628

[bdc-ehealth]

BeObservation allows both LOINC and SNOMED CT, so no loosening of restrictions is required.

[bdc-ehealth]

Additional questions from the medical terminologist:

Hi Bart,

There is insuffiscient background in your tickets (jira and github) here to give an answer to your question. Could you provide a more complete description of the project and what it aims to capture?

Do you wish to capture the question (the investigation made) or the answer (result of the investigation) or both? LOINC and SNOMED CT are usually meant to be used together in question-answer pairs, the first for the question (to code the test that is run) and the other to render the (non numeric) answer in tests as agreed between Regenstierf and SNOMED international. LOINC can code about any test made to make a clinical observation without taking a piece of the patient and all the lab tests made on specimens taken from the patient. SNOMED CT concepts can render the results, what normal measure, pathology or abnomaly is observed.

For terminologists to define the correct value sets, you need to append to your current question what clinical tests would be recorded in your project as stated by the demanding clinicians (not the SNOMED CT codes you think might be the best ones) and what would be the possible answers to each test that those clinicians expect to encounter. "Cancer screening" can mean about anything. You need to specify how this screening will be made, for each cancer type: blood test, stool test, breast radiology, cervical swab, breast needle biopsy and how the results are going to be recorded (ex: C1 to C5 pathological assesment rating of breast fine needle aspirate).

If you would kindly complete the needed information, at least the tests that will be part of this project if not each clinical answer expected to those tests, we'll be happy to provide you with more insight, eventually by querying our specialists about what are the possible answer values to these tests should you not dispose of that information.

Best regards M-A

Dr LAMBOT Marie-Alexandra Adjointe à la direction médicale

[costateixeira]

1. VAZG will ask Patrick Martens for the actual concept that is used: is it colon cancer screening (which should be 275978004) or large intestine cancer screening (which should be 268548003).
2. In any case we will provide the 3 concepts to the terminology experts to confirm that these codes are correct to start.
3. We still have the "public comment" phase of publication to get further feedback on the terminologies

[vmichie]

The director of CvKO (Patrick Martens) confirms that we have to use 275978004 as Snomed CT code for colon cancer screening.

[bdc-ehealth]

WG: we will reach out to M-A Lambot to discuss with Patrick Martens the codes needed.

[bdc-ehealth]

screeningsCvKO_v1.0.docx

Dear Bart,

Enclosed you will find a first draft analysis of how screenings tests and screening results could be represented using SNOMED CT. I have worked on the breast cancer screening use-case only as it took already a considerable time (more then 6h of my time) to make this analysis for that one use-case and it raises fundamental questions which need to be answered first by the medical end-users, by the requesting authority in charge of the CvKO project, and by the BeSafeshare team before we can "clone" efficiently the chosen methodology to the other two screening test of this project (or any similar request be it at national or regional level).

After discussing the Observation resource with experts from the SNOMED on FHIR group, I have chosen to work only on the representation using SNOMED CT codes as the use of LOINC codes to represent the test in the Observation FHIR resource is only an US authorities recommandation, not followed by other countries closer to us like UK, and it leads to mapping needs and more yet complication.

Your question is definitely not as simple as giving assent on the two and a three concepts refsets you have proposed. As you said in your email, we need to make data exchange interoperable. I would add also that we need to respect the Only Once rule. This implies putting any and every health data standardization project into the global Care set data exchange picture, which I tried to do. Please Anne and Sarah do correct me if I overlooked something in my draft analysis.

Please note, Bart, if you are not familiar with how SNOMED CT extensions work, that every Belgian refset should sit in the national BE extension, published by the Belgian NRC, even if they are intended for the use of only one community. To say it in very few words, SNOMED CT extensions dependencies can only work in "cascade" and never allow for parallel loading into a terminology server of two extensions that would both have the same direct "parent" dependency. In other words, a clinical EHR would not be able to work at the same time in the same local terminology server with a French community extension and a Flemish community extension both built to depend on the national BE extension (itself depending on the international SNOMED CT core edition). Most hospitals get patients from both communities so I believe the issue is evident. Only the most local content that one does not plan ever to share or temporary proof of concept content to be later promoted to the BE extension like we did for the AllergyIntolerance refsets may be housed outside the national BE extension. Anything else will be a technical nightmare for EHRs hospital providers on top of multiplicating content maintenance load and the need for expert manpower we are already short of. David, Katrien and Arabella can definitely explain to you in more details the technical constraints of creating extensions and correct any approximation I had to make here to keep it short.

To you all, this document is but a draft version, offered to your consideration to the best of my current availability on my personal time. Please feel free to comment, correct and enrich it in the online version here (https://docs.google.com/document/d/1ozQ_0-EQvDwkWsyMBquxE4pdD7wBbrHJ/edit?usp=share_link&ouid=116297299763859012982&rtpof=true&sd=true) so can work on it simultaneously more easily.

Best regards

Dr LAMBOT Marie-Alexandra Adjointe à la direction médicale +32 2 535 49 80

[bdc-ehealth]

WG: because we do not get any feedback from the physician in this project, and because we would like to go on, we will use the screening procedure codes in the results. We are aware that these are not the correct codes, but as they are for internal use now we will use these. If these data will be used for CDS or other secondary use, we will need to correct these codes.