Martin Hurrell wrote: When we were on the telecon the other day I suggested that ET tubes provide an example of missing concepts in SNOMED CT. This probably arises because tubes can have a number of attributes, armoured/reinforces, cuffed, blockers etc. some of which can exist in the same physical device. So to capture all possibilities would mean a proliferation of terms which I suspect is avoided by authors. An example would be an ‘armoured, cuffed ET tube’.
Rather than trying to find every example of this issue it seems more helpful to look at why it arises. I think it’s because SNOMED CT’s ontology is designed to support the classification of concepts more from the POV of terminologists than of users (understandable of course!). In many cases it would helpful to use simpler concept descriptions and more attributes (qualifiers) rather than using pre-coordinated terms. Since SNOMED CT isn’t going to be re-written in the near future a practical (SNOMED compatible) alternative could be to use the Compositional Grammar (CG) to do the job.
Taking the airway example just given, I can’t find a SNOMED term for ‘Cuffed ET tube’, nor come to that, ‘Uncuffed ET tube’ (although there is a term for ‘Uncuffed tube’, tube type undefined). There are, however, useful terms under the concept ‘Device cuff’ (SCTID 272188009) all of which are in IOTA. So, using terms in IOTA, a description of a cuffed, armoured, endotracheal tube could be rendered in the CG as:
309801005 | Armored endotracheal tube | + 2248009 | endotracheal tube cuff |
(a real example is the ‘PRO-Breathe® ArmourFlex® Cuffed ET Tube’ (FDG880) www.proactmedical.co.uk/proshop_support_docs/ETtubes/ET_brochure.pdf).
So, I think that it might be helpful to review the SNOMED CT qualifiers and to identify ones that are needed and sort out some inconsistencies. I don’t think there would be that many but there are some. For example, why have ‘uncuffed tube’ but not ‘cuffed tube’? Is it necessary to distinguish between the physical device components ‘endotracheal tube cuff’ and ‘endobronchial tube cuff’ when they always exist as part of an ET or bronchial tube, respectively - why not just use the generic term ‘tube cuff’ (SCTID 257373001) in association with a description of the parent object?. To go further, why not make ‘armored’, ‘X-ray opaque’, ‘Pre-formed’, ‘steerable tip’ etc. qualifiers that can be attributes of airway devices? This could be done by adding relatively few terms to the SNOMED qualifiers and then using the CG to define any required complex object description.
An obvious advantage of pre-coordinated terms is that, hopefully, they're all valid. Using the CG to associate attributes with concepts it would be possible to define things that couldn't exist e.g. a tube that was simultaneously 'cuffed' and 'uncuffed'. I suppose the ideal would be to bite the bullet, define all valid pre-coordinated terms, but at the same time define associated qualifiers / attributes so that terms could be derived automatically from simple user input.
5 Comments
Zac Whitewood-Moores
Much as I hate 'over' pre-coordination, in the case of devices it is hard for the concept to be represented any other way that is practical. However I am not sure this describes the situation far enough. Size would also appear to be a critical attribute which I do not see mentioned.
In the UK, the dictionary of medicines and devices focuses mainly on prescribe-able devices meaning that many hospital devices are superficially covered or absent all together. Within the hospital deployments it is important that a similar level of granularity is available to the primary care, or it will be challenging to interrogate records (e.g. device failure reports) and use electronic stock control.
I agree that a template approach with modelling of attributes would be entirely beneficial, and perhaps focussing on those features that are critical to choice of tube might be the first place to start, which is part of what I see above. This is perhaps going to be even more useful in the critical care environment than the theatre environment as renewal of devices during the stay is more likely to manage infection risk.
Martin Hurrell
I agree, the example doesn't explicitly refer to these but certainly size (length and gauge) would be important attributes to record. The CG (I believe) supports this. Dependent on the device type many other physical properties might be recorded such as tip types, calibration marks, insulation etc. That would be one advantage of a composition where the inclusion of attributes might vary with need whilst the anchor remained the primary concept.
James Palmer
Hello all,
As I see it this problem will have occurred in other areas (e.g. hip arthroplasty implants) and we should follow the example that they have used rather than reinventing the wheel.
Cheers,
J
Steven Dain
One has to look at the differing uses of an ontology.
If one is using the ontology to store information from an EHR, one does not have to use the SNOMED terms exactly. I don't like pre-coordinated terms if possible in a vocabulary.
One could in the EHR GUI could use a phrase "portex, armoured, x-ray marked. cuffed, 8 mm x 28 cm ETT" and then store that description in a database as Andrew has described above with the other attributes attached eg something like
|armoured endotracheal tube|+|endotracheal tube cuff|+|x-ray marked|+|8.0 mm ID|+|manufacturer=Smiths|
(numeric codes left out for clarity)
Zac Whitewood-Moores
I think it would be useful to have Stuart Abbott input on this or others with device terminology experience. He has so far been unable to comment unfortunately due to permissions. Can Ian Green or Fleur McBriar help maybe sort the permissions? Thanks