SNOMED CT LOINC extension – updates and alpha preview
For those of you who have been following the LOINC and SNOMED CT developments, there is an update on this vital piece of work. Termlex were at the recent SNOMED Business Meeting in London and here is a quick summary of developments so far:
- There is active work on the part of both SNOMED International and Regenstrief Institute, Inc. to align content and to create the SNOMED CT LOINC extension.
- What is more important is that SNOMED International have made some of this work visible in the spirit of openness. We will cover some of this content in the next section – but please note that this is very early work and isn’t for use.
- For those who might have missed our previous posts on the LOINC and SNOMED CT content, the scope and priority of content remains the same. So you can refer to our previous posts here.
- One thing to note is that there were members from different countries including EU representatives that were interested in using SNOMED CT and LOINC in the lab space. Anyone who has looked at the `laboratory medicine` space holistically will recognise that there is a place for both LOINC and SNOMED CT together, instead of just one or the other.
- Stanley Huff‘s slides are a fantastic place to start for those who want more details (see link below). The meeting was also attended by Marjorie Cowan Rallins, Eza Hafeza alongside Stan Huff from the LOINC/Regenstrief Institute, Inc. side.
SNOMED CT LOINC extension
An initial tranche of LOINC concepts have been modelled in SNOMED CT. I heard varying numbers from 60 – 100 concepts. My guess is that 60 is the number of Observables concepts and the other concepts are the ones that have been created to support the modelling.
Here is a screenshot of some of the sample content that SNOMED International have made available.
Please note that we were advised by SNOMED International that this is very early alpha work. So if the screenshots no longer match what you see in the content, then you have been warned… 😬
Things that immediately stand out is that – we have the following broad categories of content:
- Quantitative lab results – which would make sense, since this is often a bulk of lab test results
- Qualitative lab results – which are again are quite typical.
- Note that this includes `presence`, `nominal` and `ordinal` types.
For those of you who are wondering what `ordinals` are – they are often the things that are reported as `+`, `++`, `+++` etc. They are quite often what the `5th part` of a LOINC concept is all about. You might find it useful to look into the basics of knowledge representation for more details.
Descriptions in the SNOMED CT LOINC extension
Here is an example of one of the quantitative result type concepts `Quantitative substance concentration of carbon dioxide in blood at point in time (observable entity)`
- Quite predictably the Fully Specified Name (FSN) and Preferred Term (PT) of this concept have been created to follow the SNOMED CT Editorial principles.
- The other descriptions are equally predictable – since they are the `long common name` and `short name` that every LOINC concept has.
The description with what is seemingly a LOINC ID – is not a cause for alarm. My guess is that for now as an easy way to have all LOINC concept information in a single place. This might eventually make its way to the `RF2 simple map` file at some point.
Relationships in the SNOMED CT LOINC extension
This screenshot tells you most of what you need to know. There is a well recognised alignment between the 6 part LOINC model and the SNOMED CT model for observables as described in an older post. So if you understand that, you can quickly zoom into the only relationship here that is slightly different:
It has an attribute (relationship type) of `Has concept categorization status (attribute)` and value (target) of `Both orderable and observation concept categorization status (qualifier value)`. Both the underlying relationship type and target concepts seem to have been created for the purpose of this work. On the surface of it it would seem to say something is both a `request/orderable` and a `result/reportable`. But given the alpha nature of this work, let us not delve into too much speculation of its purpose and what its implication might be for users of the extension.
When clients use our software (Pathnexus) to create single test catalogues and harmonise their lab tests, they often look for `codes` (tests) that can be on both sides of a request/order <-> result/reportable chain. We have previously discussed some of the issues with this `simplistic view` of a complex process when users refer to `simple tests`. So for now, let us watch this space…
Our thoughts on this work
We are quite excited to see this first tranche of work coming out. We have been quietly working on a 2-way transform of ~3000 concepts between LOINC and SNOMED CT and found some systemic ontological differences. But the principal parts of this early release of the SNOMED CT LOINC extension look sound. The approach to descriptions (automated) and the core LOINC model is similar to how we approached our work… 😅 So we suspect that sooner or later, this work will run into the same ontological design differences that we observed. This would be where both the SNOMED CT and LOINC communities would work together to align these widely used terminologies.
- The notion of `Both orderable and observation concept categorization status (qualifier value)` is one to watch
- The descriptions with `presence` in them used to cause concerns in the UK clinical safety community. So it will be useful to see how these are addressed. We for one would like the `tall man` convention for distinguishing between `presenT` and `presenCe`.
What are your thoughts about this work and some of what you have heard? Were you at the SNOMED Business Meeting and heard other things about this work? If so, please share your thoughts as comments…