Adverse Reaction archetype - review round initiated

Technical (archive)
1

Dear Colleagues,

A review round for the Adverse Reaction archetype was initiated today. This is a significant archetype that requires careful and considered collaboration and I would like to ensure that we have the best team reviewing the specs as we can.

Current reviewers comprise the openEHR Archetype Editorial Group plus existing adopters of the archetype, however we welcome broader expertise from the broader openEHR community.

If you, or one of your colleagues, would like to participate in the review, please log in and adopt the archetype. Instructions and diagrams for adopting are found at http://www.openehr.org/wiki/display/healthmod/Adopt+an+archetype

All adoptors will be included in the ongoing review process.

[If you are a registered CKM user, and/or subscribe to multiple openEHR lists, you may have received this notification several times - my apologies;-)]

Kind Regards

Heather Leslie
(Editor)

2

hi

I’d like to start by asking about the scope of this archetype.

Is the intent to report about a condition of an adverse reaction?

or a concern about? or just a report of a possible one?

Is it a high level clinical summary, or is it supposed to be good

enough to support DSS?

I have considerable interest in this archetype: as well as being

involved with this model through NEHTA and HL7, my daughter

is highly allergic to tree nuts, but (a little unusually), not peanuts

as well.

It seems to me that the current archetype is only good for a

gross point report of a single episode of apparent adverse

reaction. If this is all it’s supposed to be, I won’t have much to say,

but if it’s supposed to be good for more than that….

I’d like the archetype to comment on this. “Recording the

presence of a harmful or undesirable response to an agent

or substance including food, as determined by the

clinician - excluding poisoning and abnormal use” is

ambiguous concerning these questions.

Grahame

3

Hi Grahame,

I have the same thoughts. There are several ways to go and I think we need these sorted out to some extent, before we decide what should be within this particular archetype. Could a possible way be to sort this out as the “report” of a reaction that has taken place and the agent that somehow is suspected or verified to be connected to the reaction? The strength of the presumed connection could also be recorded. Then a number of other risk factors (conditions, treatments) could be recorded in another archetype?

I have lead a project about warning information in EHRs in Sweden and we have been discussing a number of ways forward. To some extent I think it all comes down to being aware of (recorded or feared) risk factors (reactions to a specific agent or conditions and treatments which could give equally serious reactions) at a certain point of clinical care (e.g. prescribing, feeding, anesthesia) where the clinical acting could be altered from a ‘standard procedure’ to avoid a risk of negative conequences for the patient. To be able to connect risk factors to clinical decision points, we also need a high level of structure (e.g. classification of agents) to be able to do the connection.

This is more or less a translation of classic “alert models” to a medical setting.

I agree with you that we will not be much happier if the only archetype we will produce is a rough recording of an apparent reaction, but I think we can make this to do more than that.

To start with, maybe just separate the cases you talked about? Recording an apparent reaction, recording a possible one (could be the first one with less degree of certainty) and then looking at recording risk factors like conditions and treatments later?

I also think this could connect to the “broader concect” that has been introduced by OpenEHR editors of looking at complex archetypes like “diagnosis” and other crucial archetypes for patient safety.

Rikard

2009/7/5 Grahame Grieve <grahame@kestral.com.au>

4

Grahame,

Do it on CKM – not on the list! Then the ideas will not be lost. The proposal certainly covers more than you have noted but would not in itself support a report. This would be a template.

Cheers, Sam

5

hi Sam

happy to do it on CKM, but the interface didn’t suggest itself as suitable for this kind of scoping

discussion

Grahame

6

Hi Grahame,

The current scope for Adverse Reaction is the initial ideal for every archetype at creation - a maximal dataset for a universal usecase.

So the current intent of use of the archetype includes within a report, within DSS, a clinical summary and any other way of utilising it. The archetype can then be constrained to make it ‘fit for purpose’ through context/scenario-appropriate templates.

The current Adverse Reaction archetype is a ‘straw man’ model that needs considerable refining and enhancing - no doubt about it. It represents the thinking of a few people, based on lots of experience and only one reference noted. The purpose of this review is not to gather yet more business requirements but to collate the existing research and thinking done by many learned and expert organisations and national programs into a practical and pragmatic model that we can take forward as the basis for sharing common things about adverse reactions.

As you quite rightly point out, the metadata needs quite a bit of enhancement. The review process should identify any missing data elements, and usecases not currently anticipated. The metadata will be improved to support all of this.

Look forward to seeing your comments through the review. You can comment on each individual data element or metadata item, plus the completeness and/or any missing elements, general issues with regard to the design of this archetype, and other resources that should be considered for this archetype.

Kind regards

Heather

7

One of my daughters is highly allergic to pistachio nuts, mango, and on some other things in the same group - but apparently peanuts are in a different ‘category’ and so she loves peanut butter and jelly sandwiches.

I wondered if the confusion is the application of the archetype - you mentioned one gross point but there would be one ‘instance’ of the archetype per allergy you wanted to record.

So I would have

Each column is an identifier to Archetype.Adverse reaction
Onset Agent Specific Substance Reaction category Date of Exposure Reaction description
01/20/2009 18:30 Mango Fruit drink with mango Intolerance 01/02/2009 17:30 Rash face and neck
03/13/2009 12:05 Pistachio Pistachio nut Intolerance 03/13/2009 12:00 Rash face and neck

Of course capturing that level of detail through anything but a patient portal seems unlikely and then there are the specific allergy test results and I assume using the openEHR-EHR-EVALUATION.excluded-adverse.v1 to record not being allergic to peanuts - or is it the above with reaction category ‘no reaction’ ?

8

Hi Greg

Although it has not been the style of doctors to capture the information in the form you suggest, I guarantee you they will be very interested when it is available. We have had a single box on a form to write a word or two in the past.

No reaction is required in the adverse reaction space to record an exposure that caused no reaction amongst others. To say there is no allergy you would use the exclusion statement.

I am interested to know about your daughter as it does sound like an allergy although this group is usually a poisonous effect and not IgE mediated. Maybe you can let me know off line.

Cheers, Sam

9

Hi Sam,

Your comment about the difficulty of telling what the reaction is/was (allergy? toxic effect?), could actually highlight a basic problem with recording this easily. I think we need a bit of terminology input or decisions here. When the clinician notices the reaction, it is often not possible to tell what the diagnosis of the reaction is. Still we need a quick way of recording this reaction for the clinician even to bother. I have figures to show this is a tricky part: less than one percent of assumed side effects of drugs were actually recorded as such in an emergency department in a big swedish hospital (I will find the reference if you need it). So when even thinking of recording Adverse Reaction, I believe we need clear guidelines and an well adapted user interface. More or less like: point at the agent, point at what should be connected to the agent, and make the connection. Possibly one or two more factors are really essential even to make the Adverse Reaction recording usable, but it is a thin line. The more we make mandatory, the higher the risk of users not complying. We need to consider this kind of useability aspects, too, when trying to make well structured and efficient systems.

A possible way to move forward concerning the diagnostics could be that the clinician starts with the notion that the current reaction is out of the ordinary (or just not expected in the setting) - and that this should be observed in future. After this first stage (roughest diagnostic level) of already useful information, it could always be detailed further. Then we could start with a hierarchical terminologic tree with maybe ‘hypersensitivity’ as the highest concect in the hierarchy and then divide this further down, ending up with concepts like specific allergic reactions, toxic effects, metabolic intolerances or similar types of conditions. This “going down the tree” could actually be facilitated by guidelines or CDSSs.

And yes, I agree that this should be pointed out in the CKM review and to some extent I have already pointed out these thoughts there.

Maybe someone says this hierarchy should be handled by SNOMED CT? I have discovered two problems with that approach so far. One is that (at least in the version I saw) there is a high (maybe too high) degree of detailed diagnostics in SNOMED CT. I noticed about twenty diagnoses of hand eczema where the difference was the agent. I would have prefered do have only ‘hand exczema’ as the single SNOMED CT term and then connect this to a separate classification of the possible agents. Otherwise you will not keep it clear and you will risk not to get the whole picture.

The second is that I have difficulties finding the evidence of the hierarchies made in SNOMED CT. Who decides, where is the reference literature, is really the hierarchical terminology tree correct in all diagnostic terms? Sorry if I just lost that part somewhere in the exploring.

However, I haven’t seen any better system than SNOMED CT either (it is certainly more exact and clearly structured than ICD), so I still believe that some things could be added to the SNOMED CT concept to improve it even further. I suppose this is part of the work being performed by the member countries working with SNOMED CT.

/Rikard

2009/7/6 Sam Heard <sam.heard@oceaninformatics.com>

10

Hi Heather

I have to admit that the difference between gathering more business requirements and collating existing

research is clearer in theory than in practice

I will comment in the review

Grahame

11

Hi Rikard

Thanks for your very thoughtful reply. It would be interesting to determine if this is a tree or not. I suspect that the words will have many interpretations. At least a tree could help people see it as a tree.

I guess the lowest level of reporting in my view is intolerance – but this is misleading as it implies that there is only expected or slightly amplified side-effects, whereas the reaction could be IgE mediated and bode very poorly for the future.

Lets see where it goes with the review and make sure your ideas are there.

SNOMED is a definitional hierarchy (which is partly implicit) - your idea of a diagnosis ‘eczema’ with a specific cause satisfies my view of the world. Let’s make sure the canonical form of the representation of diagnosis does not include cause as we know there will always be new ones not in the terminology.

It would be good to link this to the discussion in the CKM.

Cheers, Sam