As some of you know, there is now a renewed cooperation between SNOMED International and Regenstrief to make SNOMED CT and LOINC work more easily together. This article covers some of the information that we know about it and what it might mean to countries, vendors, projects, labs etc. This is the first part of our series – there is a part 2 with a specific focus on the UK and NHS here.

What is the vision?

Broadly the agreement allows for the following:

1. Distribute LOINC content to both LOINC and SNOMED CT users.
2. Create SNOMED CT and LOINC codes for all concepts shared between the terminologies.

Let’s unpack that a bit. For a while those of us who wanted to use both LOINC and SNOMED CT in our products needed to import both content separately in separate distribution formats. More importantly, once we imported them – there was no simple way of saying – this `code` in SNOMED CT or LOINC is the same as this other code in LOINC or SNOMED CT! In fact, the only way to bridge them was to use a `post-coordinated expression` that was distributed as a technical preview file. If you are thinking `huh post-coordinated expression??`, then you’d be right to rejoice after reading the second point above:

Every concept in LOINC (observables) will have an equivalent SNOMED CT concept (observable)

Note the scope `observables` but if you don’t understand what that means then don’t worry – for most practical purposes anything that is a `lab result` is what this covers. The implication here is also interesting (and our reading) is that the SNOMED CT concept equivalent of a LOINC observable will be `fully modeled/defined`. So, this implies:

Implied: Every LOINC part (for observables) will possibly have a corresponding equivalent SNOMED CT concept

If you don’t understand LOINC Parts, those are the things that make up a broad definition of a LOINC concept. For exampl, if you look at ` 3330-8 |Amiodarone [Mass/volume] in Serum or Plasma|` in LOINC, you’ll see it is defined/modeled as follows:

Those entries called `component`, `property`, `time` etc are all Parts. In fact, in LOINC, the concept’s `Fully Specified Name` (FSN) is a combination of the mains parts (based on heuristics).

Since these Parts aren’t actually from the SNOMED CT Observables hierarchy,

If every LOINC part (for observables) will have a corresponding equivalent SNOMED CT concept, then SNOMED CT’s hierarchy will likely be extended

We could imagine at the minimum, the following hierarchies being extended and/or more importantly aligned with LOINC!!

• Substance
• Techniques
• Properties
• Specimen

We will cover more details at a different point in time – but suffice to say this is exciting! For those who have lived in the LOINC SNOMED CT dual world (like us), the `LOINC Part mapping file` that has been around could form the basis. However, don’t read too much into `mapping` here, because you want to read the next sections!

What would this look like?

The proposal is to distribute a LOINC extension for the LOINC concept based on the SNOMED CT model!

Yes, that’s right – a LOINC Extension in SNOMED CT! You would see LOINC content in the `shape` and format of SNOMED CT. If we unpack this, what this could mean in the future is:

• You could subscribe to a LOINC extension in addition to the SNOMED CT International Edition, which would contain all the `Lab related content` in SNOMED CT RF2 format
• You could then combine this LOINC extension with your own national or local extension of interest, to extend or localise if further if you so wish

More importantly, by simply loading all of LOINC content in a SNOMED CT model, you could use the underlying ontology (relationships in SNOMED CT) to perform analytics. More about this later… Some important information to understand right now:

Regenstrief and SNOMED International have signed the agreement, but are still working out the specifics of the work (likely 2023) for visible work. So we should avoid speculating and rather recognise the motivation here is:

• Minimising duplication of work in this space and closer working between both organisations and their respective working groups and committees.
• LOINC extension will be owned and maintained by Regenstrief – including the model and the ontology
• Any concept in the LOINC extension can be promoted to the SNOMED CT International Edition
• Any concept that exists already in SNOMED CT International Edition, could/would be added to LOINC (getting a LOINC identifier).

Proof of concept work will be completed, before the formal project is launched.

For those of us wondering, what about `request` concepts (procedures) in SNOMED CT and would they be added to LOINC, I think we should remember the initial scope of this work is `Observables` (aka results).

Why does this matter – Benefits

When standards bodies make statements, we always wonder what this means and why should we care. So here are some reasons why this is a good thing and why we should be excited about it.

Lab Informatics Knowledge

• LOINC has over the years built a large community of lab specialists/experts and users, all of which can be leveraged by the entire `lab informatics community`
• LOINC has a wealth of knowledge and value-add bits in it (e.g. definitions, LIVD profiles, Units in UCUM, panels, etc) all of which represent a substantial corpus of knowledge in the lab world. By being able to cross-walk from SNOMED CT to LOINC, you’d get all this for free!!

Suppliers

• Lab system vendors (LIMS and other systems) have had to configure and redo a lot of their systems each time they enter a country, where different standards are used. This is actually a cost to the entire ecosystem. not just suppliers. This cost can now be reduced by using the same `format` – depending if they want the LOINC format or the SNOMED CT format but supporting both since they often co-exist in an Electronic Health Record
• This approach continues to provides a 1:1 map for suppliers who use a native LOINC format and want to leverage the SNOMED CT ontology later in the future.

Future alignment

• LOINC will now be able to leverage the richer ontological representation (and expressivity) in SNOMED CT allowing interesting use cases like analytics and semantic interoperability
• This opens up the area of cross-border interoperation of records/results – where for example one country uses LOINC and another uses SNOMED CT.
• Diagnostic medicine is a large space – with multiple specialities (including genomics). If this Observables scope works well, then LOINC has a lot of other content that could likewise be aligned with SNOMED CT delivering the same benefits.

Note, LOINC extension does not remove to localise the system for user needs, but allows suppliers to use SNOMED CT or LOINC to conform to the relevant national standard.

This is what we do in Pathnexus – where we allow users to standardise lab tests and results in either LOINC or SNOMED CT and interoperate!!

Frequently Asked Questions

Please treat all these as current thinking – again, let us give SNOMED International and Regenstrief space to share more details as things progress!

Question: When will the LOINC extension be available?

• A proof of concept (PoC) will be first undertaken before the formal project work will be done.
• Current plans are for the PoC to be done in Q4 2022 and for the formal work to be done in 2023.

Question: If an organisation requests a new observable concept, who creates the concept?

• If you are using LOINC directly, then expectation is to submit a request to Regenstrief and they’ll likely add it.
• If you are not using LOINC, then exceptation is to submit a request to SNOMED International – who will work with Regenstrief to coordinate how it’ll be added.

Question: What is the scope of what goes into the LOINC extension and what stays in LOINC?

• Would LOINC content like Panels, UCUM etc be included?
• Initially the content will be limited just be Observables content
• Separate conversation would be needed for UCUM
• Would this extension also include the LOINC Radlex playbook?
• There is a need to honour to current agreement with Radlex, so this needs to be explored.
• While initial talks looks promising, this is still early days

Question: Is LOINC extension content to be included in the international release of SNOMED CT?

• The LOINC extension is no different to a classic extension. So the LOINC extension will be just like any other extension that builds on (extends) the International release.

This is us thinking out aloud – the LOINC extension might initially exist in the Community Extension space. Then it might slowy make its way into a formal extension for download as the formal project becomes mature.

Question: As a member with an existing extension – how do I include this LOINC content in my release?

• This would work like every module/extension you want to use. The SNOMED CT way would be to  include this extension in the module dependency refset, just like you would with any extension

Question: How does LOINC extension effect existing observables?

• Official response: The expectation is that it shouldn’t affect this existing Observable content.

This is not the official response – but our guess would be that they would get LOINC identifiers where they exist in SNOMED CT but not in LOINC. Beyond that, they would simply continue to exist. If the underlying question here is:

• Would this mean existing Observable content be moved to the LOINC extension?
• Without speculating, we’d think that is the sort of question that that PoC work will want to flush out. We are sure that both SNOMED International and Regenstrief would want to avoid unnecessary churn in SNOMED CT identifiers and/or introduce dependencies.

Question: How does this work in the context of Electronic Health Records (where both SNOMED CT and LOINC are used)?

• The value is in the benefit of being able to use both as summarised above.

Unofficially we think the value is in leveraging the ontology of SNOMED CT for analytics and other uses, while benefiting from the wider adoption and value-add features of LOINC.

Question: Are there plans to have a `new common format` for this content?

• Initially the LOINC distribution will continue and so will SNOMED CT
• Perhaps other developments might inform this, but it is too early in the day to know.

Question: Would a sub ontology be created for the LOINC Extension?

• It might be created at some point in the future. Too early to tell at this point.

Thoughts for the future

We are incredibly excited about this work, since we have used both LOINC and SNOMED CT in our product Pathnexus, to allow users to standardise their lab data. We currently allow our users to create new standardised lab tests (where they don’t exist) – without being experts in SNOMED CT or LOINC. The cool thing is that we allow users to reuse existing tests in LOINC or SNOMED CT – here is a quick screenshot of (but this is definitely not a topic for a future post).

The second immediate application of this approach we have is using the richer semantic/ontology representation of SNOMED CT for analysis of lab reports irrespective if they are based on LOINC or SNOMED CT. This will deffo be a separate post, but this announcement makes such use cases even far more simpler!

The final and quite interesting prospect is in the area of cross-border care and how this plays out in the context of the International Patient Summary (IPS), where both LOINC and SNOMED CT could be used in different countries.

Are you trying to standaridise your lab data using SNOMED CT or LOINC? Do you have plans to analyse (or build dashboards) for your lab reports using LOINC or SNOMED CT – if so, are you as excited about this announcement as we are? I

f you are unsure how both these use cases can be addressed by this announcement, then get in touch with us.