I wonder how other conditions with lateralities are generally dealt with, such as in ENT, or MSK e.g. limb disorders. Is laterality typically pre- or post-coordinated in those cases?
I also do not think we necessarily need to pre-coordinate SNOMED terminology just because some OMOP providers prefer it. Can be a mapping between a single OMOP-related code and a post-coordinated group of SNOMED codes? I know there is not a single correct answer here. I suppose it is the purest in me that prefers post-coordination, as it removes redundancy and can communicate the same information with less codes required.
I'd much prefer ophthalmology SNOMED terms to all have laterality as part of the code. The two eyes will often each have different conditions and thus different signs and diagnoses. If there is no laterality specified (or its not available for all significant findings and diagnoses), then it greatly reduces the value of having SNOMED terms. For example, if all we have is that the patient has "314028008 | Post-traumatic angle recession (disorder)", "77075001 | Primary open angle glaucoma", "15462009 | Arcuate scotoma (finding)", "246886000 | Drainage bleb, flat (disorder)", and information on "370937003 | Vertical cup disc ratio (observable entity)", then it's not possible to tell which eye had the trauma, which had the failed trab, and which has the more severe visual field defect.
Would baking-in laterality into the terms be made easier with pre-coordinating terminology? Or are those two orthogonal concerns?
Thanks Eric. In my mind, you would have two codes, one for the diagnosis, and one for laterality. The two would always be provided together. Or, for something like visual acuity, you will have separate codes for laterality, distance from chart, unaided/presenting/BCVA/PH, LogMAR/Snellen, rather than a massive list of codes that accounts for all of the combinations of the above. Having a separate code for all of these is a definite possibility, but seems less elegant/parsimonious/logical to me. But I know there are strong arguments on both sides!
Thanks all. It's an interesting and important discussion. My current standpoint is similar to Anthony. I think having laterality as a property of diagnoses / procedures etc would potentially work well. That is how our EPR OpenEyes codes things for instance.
For certain diagnoses there may be even more potential details of properties that could be added - for example BRVO - the specific quadrant affected, or for eye lid pathology - as well as laterality there may be a desire to specify upper or lower lid and even further location details of medial, lateral etc. The number of codes needed to capture all this if using pre-coordination could become unmanageable.
Would be useful to get advice from Elaine or any of the other SNOMED team members about how this could be handled from their experience.
Interesting and valuable discussion. Pre-coordinating laterality risks combinatorial explosion of codes as Anthony and Ian point out. Use of a separate laterality codes is supported by the EHR system I use (Oracle (Cerner) Health Millennium) but I know that we have incomplete data here as emphasised by Eric. I think the route to data quality is always accelerated by data extraction from the clinical database, exposure to the clinical group involved and challenge to complete data recording comprehensively. The journey is aided by training, team work, and the organisation explicitly valuing the time spent by healthcare practitioners recording data for care. I will discuss with Ian Green and others.
7 Comments
Sophia Wang
I wonder how other conditions with lateralities are generally dealt with, such as in ENT, or MSK e.g. limb disorders. Is laterality typically pre- or post-coordinated in those cases?
Anthony Khawaja
That's a great question. Elaine Wooler Jane Millar Ian Green - any ideas?
Anthony Khawaja
I also do not think we necessarily need to pre-coordinate SNOMED terminology just because some OMOP providers prefer it. Can be a mapping between a single OMOP-related code and a post-coordinated group of SNOMED codes? I know there is not a single correct answer here. I suppose it is the purest in me that prefers post-coordination, as it removes redundancy and can communicate the same information with less codes required.
Eric Brown
I'd much prefer ophthalmology SNOMED terms to all have laterality as part of the code. The two eyes will often each have different conditions and thus different signs and diagnoses. If there is no laterality specified (or its not available for all significant findings and diagnoses), then it greatly reduces the value of having SNOMED terms. For example, if all we have is that the patient has "314028008 | Post-traumatic angle recession (disorder)", "77075001 | Primary open angle glaucoma", "15462009 | Arcuate scotoma (finding)", "246886000 | Drainage bleb, flat (disorder)", and information on "370937003 | Vertical cup disc ratio (observable entity)", then it's not possible to tell which eye had the trauma, which had the failed trab, and which has the more severe visual field defect.
Would baking-in laterality into the terms be made easier with pre-coordinating terminology? Or are those two orthogonal concerns?
Anthony Khawaja
Thanks Eric. In my mind, you would have two codes, one for the diagnosis, and one for laterality. The two would always be provided together. Or, for something like visual acuity, you will have separate codes for laterality, distance from chart, unaided/presenting/BCVA/PH, LogMAR/Snellen, rather than a massive list of codes that accounts for all of the combinations of the above. Having a separate code for all of these is a definite possibility, but seems less elegant/parsimonious/logical to me. But I know there are strong arguments on both sides!
Ian Rodrigues
Thanks all. It's an interesting and important discussion. My current standpoint is similar to Anthony. I think having laterality as a property of diagnoses / procedures etc would potentially work well. That is how our EPR OpenEyes codes things for instance.
For certain diagnoses there may be even more potential details of properties that could be added - for example BRVO - the specific quadrant affected, or for eye lid pathology - as well as laterality there may be a desire to specify upper or lower lid and even further location details of medial, lateral etc. The number of codes needed to capture all this if using pre-coordination could become unmanageable.
Would be useful to get advice from Elaine or any of the other SNOMED team members about how this could be handled from their experience.
Charles Gutteridge
Interesting and valuable discussion. Pre-coordinating laterality risks combinatorial explosion of codes as Anthony and Ian point out. Use of a separate laterality codes is supported by the EHR system I use (Oracle (Cerner) Health Millennium) but I know that we have incomplete data here as emphasised by Eric. I think the route to data quality is always accelerated by data extraction from the clinical database, exposure to the clinical group involved and challenge to complete data recording comprehensively. The journey is aided by training, team work, and the organisation explicitly valuing the time spent by healthcare practitioners recording data for care. I will discuss with Ian Green and others.