The unknown substance conundrum…

Is it safe and sensible to record that a patient has had an adverse/allergic reaction to an unknown substance?

This is a question that is causing quite a stir in the HL7 Patient Care email list and I have no doubt that it will be a hot topic in the current FHIR/openEHR Adverse Reaction archetype review. Some are advocating to record a reaction without offering a possible causal substance.

I think we can all agree that it makes absolute sense to record the DIAGNOSIS of an obviously <allergic manifestation> (eg urticaria) to an unknown substance as “<allergic manifestation> of unknown origin”. I think most can comfortably agree with that statement.

Now for the controversial bit…

However while a patient can have a reaction to an unknown substance, I will assert that it does NOT make sense to record an ADVERSE REACTION manifestation to an unidentified substance. I am certainly not denying that a clinician who observes a rash that is clearly of allergic origin, such as an urticaria, will absolutely be thinking in terms of ‘this is clearly of allergic origin and I must investigate further to identify the substance’. This is clearly good clinical practice. However, as clinicians, I am suggesting that sometimes we need to differentiate between what we observe or think from what we might record in a health record at the same given point in time .

In all my years of clinical practice, I cannot remember any paper record where I have ever seen ‘urticaria’ or ‘anaphylaxis’ entered into the Allergy List without a proposed substance being associated. I can see absolutely no reason why we should start doing this for electronic health records.

Both the substance and the manifestation are required to enable a clinician to avoid future exposures in future clinical care. That is the logic behind the existence of the ubiquitous Adverse Reaction or Allergy List.

Until one or more causal substances is postulated and/or confirmed, an isolated clinical manifestation of urticaria or anaphylaxis is, in reality, just a simple (maybe allergy-related) sign, symptom or diagnosis of unknown cause. The act of recording a reaction to ‘nothing’ or ‘unknown substance’ adds no absolutely no value to the Adverse Reaction or Allergy List as it cannot trigger any decision support by a clinical system.

I think I can also safely assert that all clinicians ideally want to be able to prevent a future reaction for their patient, especially in the situation where the substance may not be identifiable at the time of the reaction. This is a key clinical activity that we need to be able to record clearly in EHRs – we cannot afford for an identified allergic reaction to be ‘lost’ in the health record before we can identify the substance.

To that end, the HL7 Patient Care approach has been to record a Reaction Event in the absence of a Substance, while the openEHR approach has been to record the manifestation as part of the health record – symptoms, signs, test results, diagnosis etc, of unknown origin. However for both, a clear imperative is that the clinician can able to flag that this patient has an unresolved puzzle: that there has been a known allergic reaction that may cause future harm if the cause is not identified.

So collectively we need to determine a way for this outstanding health puzzle to be tracked and not lost in the depths of a long ago consultation note. This aligns with the design of Lawrence Weed’s Problem Oriented Medical Records (POMR). In HL7, this approach is captured in their concept of ‘Concern Tracker’, and in the draft Contsys standard (ISO 13940) it is known as a ‘Health Issue Thread’. openEHR needs to develop EVALUATION archetype/s that will support this abstract concepts and support the POMR approach.

Archetypes/DCMs MIA in SDOs

Curious to note that there is very little apparent interest in detailed clinical information models (DCMs) of any brand or flavour in the major Standards Development Organisations (SDO’s) – they are effectively Missing In Action when compared to the likes of CDA and IHE profiles.

The ISO 13972 DCM specification took a long and tortuous time to travel through the ISO TC 215 processes. Engagement with 13972 during its development, and from what I can observe now it is on the verge of publication has been rather sparse.  A further piece of work is now starting in ISO regarding quality criteria for DCMs but it also seems to be struggling to find an audience that understands it, or even cares.

I don’t quite understand why the concept of DCMs has not been a big ticket item on the radar of the SDOs for a long time as it is a major missing piece of any standards-based framework. Groups like CIMI are raising awareness, alongside the openEHR work, so momentum is gathering, but for some reason it seems to keep a very understated profile compared to new opportunities like FHIR in HL7.

The work of messages, documents, profiles and terminologies are clearly important for interoperability, but standardisation of clinical content models working closely with terminologies can potentially make the work required to develop messages, documents, and profiles orders of magnitude easier.

Let me test a metaphor on you. Think of each message, document or profile as a sentence and each archetype or DCM as a word, a building block that is one component of each sentence. By focussing on the building a specific sentence, we are working backwards by trying to determine the components, and the outcome is still just that single sentence. However if we start by standardising the words/archetypes, then once they are stable it is relatively simple to construct not only one sentence for a specific purpose, but the potential is a much greater output in which many more additional sentences can be created using a variety of words in different combinations. If we manage the words (archetypes) as core building blocks and get them right, then we allow a multitude of possible sentences (messages/documents/profiles) to proliferate.

The ‘brand’ of archetype/DCM solution does not concern me so much as raising awareness that clinician-led, standardised clinical content is a significant missing and overlooked piece of the international eHealth foundations puzzle.