Blog Details

Background

When we discuss lab tests and data I always come across various repeated conversations. They often focus around similar threads:

• As a GP/doctor I just want to place a request for a test (also called an orderable sometimes) and be able to see the corresponding result.
• As a Lab Specialist, I want to report on what I did — e.g. specimen, units, technique, reference range, etc.
• As a public-health expert/statistician/patient, I would like to know which tests were `normal` vs `abnormal` and the differences between the abnormal tests (e.g. sensitivity, specificity of technique/method, etc).

This plays out for most lab medicine related content in SNOMED CT and can be quite confusing. So here is an example for `serum calcium` bringing back `53` results across three hierarchies in SNOMED CT!! Why?

There is a triad here and they are all linked to something that is simple yet complex. Some of the clues to what the triad is in the way I phrased the requirements above and the representative users.

I hope Tolkein fans won’t be offended by me borrowing the `3 elven rings of power` analogy to talk about lab test results. For in truth, there are three perspectives or sources of `truth` in the laboratory medicine data space. Just like the three Elven rings in Lord of the Rings (LoTR) were all magical and shared certain powers but were eventually different, all these three perspectives share similar use-cases and motivations but are significantly different. Fortunately/Unfortunately there is no `one ring to rule them all` — sorry to disappoint the LoTR fans! However, there is a lesson in LoTR — trying to assert that any one ring rules (or overrules) the other two is dangerous!

Request perspective — the first ring of Power!

This is the perspective of most clinicians who want to request a test be done. They do this by checking a box on a piece of paper or a checkbox on an Order Comms system. From the perspective of these users they have a patient in front of them and need to know the `outcome` of a test to plan their intervention/actions.

Sometimes they want to request a very specific version of a test or on a test on a different specimen, time, etc. However, their principal concern is still the same. They would like the `lab specialists` to do what they do (and know) best and help them plan the next course of action for the patient.

They want to know when a result for the lab test they request is available and be able to view it.

So typically for a primary care practitioner (GP in the UK), they want to see `serum calcium level/measurement` as a `test` that can be requested. When they get results, they want to see something like `serum calcium level` with a `value` attached to it — something like:

From the SNOMED CT perspective, the `request` is a concept in the `Procedure` hierarchy.

If you know the history of SNOMED CT, then you’ll know that lots of content from UK Read (version 3 — Clinical Terms) was merged into it back in the mists of time. So you can now imagine why there is a lot of content around `lab requests` in the SNOMED CT Procedure hierarchy.

In fact, you might be hard pressed not to find commonly requested lab/diagnostic tests in SNOMED CT! This is a good thing, since the more content we get out of the box in SNOMED CT, the better it is for us.

Remember what I said above, that there are times when the requesting clinician wants something a little more specific than a generic test. For example, they might want `ionised calcium` or the `adjusted calcium` etc since both these affect different things in our body. Again, SNOMED CT comes to rescue and has a list of all these `more specific` versions of a `test requests! 👏

Yet again, this is a good thing. I mean, let’s be honest if you are browsing SNOMED CT you are doing something clinically related even if you aren’t a clinician yourself. So you need to have some understanding of the reason for such specific tests.

Results perspective — the second ring of Power!

This perspective is a little more intricate and sometimes hidden to an extent from users. In a way, this is what actually happens when a lab receives a request for the test to be done.

The simple/naive way of looking at this is — someone pushes a button and a result pops out. But from the perspective of a lab specialist, they actually get very detailed about exactly how that result was derived. I will not repeat things from previous posts about all the things that are important for a lab specialist, but you often need to understand that the lab specialist (and eventually the other users) also care about the specific details about how the test was done. Here is a brief list (incomplete — so please refer to older posts…)

• What was actually measure (e.g. ionised calcium vs adjusted calcium)
• What technique was used to measure this analyte/component
• What about the analyte was measured — presence, concentration, etc

In fact, the lab specialists are very precise about what they have done, so they want to capture as much information about this `reportable` item as possible. This is often why, when you often see a lot of different variants of the same test but measuring different things using different techniques, etc

Similarly, look at the variants of how many different ways our recent acquaintance `SARS-CoV2` from the lab results perspective — antigen, RNA, sputum, etc

Once you understand how a result is produced (with other information about how the test was performed), the lab issues a result/report. Often what comes out of a lab does not capture all the additional information about what the lab specialist did. Rather it comes out with a numerical result with or without an interpretation of the result. So this would look like:

Note: You might or might not have the `interpretation` column, depending on the lab’s practice and also the test.

Here comes the SNOMED CT bit — and it is a perspective because in SNOMED CT anything that can be `observed` and `measured` belongs to a hierarchy called `Observable Entity`. Since SNOMED CT is an ontology and things in an ontology are `entities`, anything that can be `observed` is in the hierarchy of `Observable Entity`. With that out of the way, you can now imagine that SNOMED CT allows `additional information` like `technique`, `property`, `specimen` etc about a lab result to be represented. In fact, there is a generic model here for such things that is common between both SNOMED CT and LOINC to an extent. Please refer to our previous posts about this `common representation` and how SNOMED CT and LOINC can be made to work for representing lab results.

Screenshot of an example calcium result from the LOINC showing similarities in model below:

These screenshots are from Pathnexus (our lab data harmonisation platform), that allows you to adopt SNOMED CT or LOINC as the basis of your LIMS (Laboratory Information Management System) or EHR (Electronic Health Record) system. Want to find out more?

Using LOINC and SNOMED CT for Lab Results

You might notice that given the commonality, that the names of things in both SNOMED CT and LOINC are somewhat similar even if LOINC tries to use `shorter` forms of things. This is actually down to their editorial principles for how they `name` things.

Now there are a few other considerations for users when representing lab results. Without getting into too much detail, the coverage of results in LOINC is far greater than what you will find in SNOMED CT. When you search LOINC for `serum calcium` you get `140` matches! Yes, of course you can whittle some of this down by `category` but still that is a lot! It also is often a lot more specific/detailed in nature, which is actually what the lab results world likes.

This is one reason why sometimes you hear it being said, `use LOINC for lab results and SNOMED CT for requests`. I am generally reluctant to repeat such `approximations` since they can be taken out of context. You should remember that there are exceptions/caveats to this:

• SNOMED CT is excellent if you want to standardise results in Microbiology particularly around organisms, specimens
• SNOMED CT is excellent if you want to standardise results in Cellular Pathology, particularly around morphology, grading etc.
• LOINC is excellent if you want to represent Panels (also called profiles, batteries, etc) and Assessment scales
• LOINC has a lot of `non-lab content` that you might also want to look at, to see what you are getting into.
• There are other practical considerations like national direction, licensing, etc that you want to consider too before making such a decision.

Given the recent collaboration agreement between SI and Regenstreif, we might start to see this situation change.

So we could indeed have the best of both worlds — broader coverage and other added utility information in LOINC, combined with the ontological soundness + association with other hierarchies in SNOMED CT!

What next — Third ring of power and more?

With those two perspectives (rings) out of the way, it’s time to delve into the somewhat confusing perspective — of `findings` or the third ring of power. Because it is this perspective that often carries the immediately relevant information to the clinicians and even patients. This is also where we understand:

How do we know if a test result is `positive`, `negative`, `high`, `low`, `normal` etc, combing what the lab world sees with what others see!

But for that, we have to take a tiny digression to understand the types of results that can come out of a lab test. We will cover all this in the next series of this article, because this one is already longer than intended!

Want to discuss using SNOMED CT or LOINC in LIMS, EHR or personal health records?

We hope you found this article interesting. We also offer services for migrating existing data to SNOMED CT and LOINC. If you have any questions or feedback, please shoot us an email at [email protected] or contact us!