View Source

I've been thinking about how to archetype the concept of severity - it is really much more complex than first thought. There is nothing helpful in SnomedCT that I can see. From my research, it seems to me that there are a number of ways clinicians think about severity - including intensity, impact (functional and physiological), extent, frequency, and persistence. I've started a rough mindmap which I've attached, and my first impression is that each of these 5 areas will need to be archetyped separately as clusters and included in an overarching 'severity' cluster. Again, I would appreciate your feedback on this before archetyping commences, and if you can direct me to any useful resources/definitions etc, I'd be grateful - its been a struggle to gather this much together in one place so far.

Ian 1 Feb 2008
Why is the world so blooming complicated!! I'm not sure I would regard all of these notions of severity as being severity of symptom. Severity of burn is a different 'thing' . I think we may have to restrict this to a separate 'severity of symptom' a concept which is in Snomed.
One of the things to bear in mind is that these are all pretty arbitrary measures and for decision support purposes will probably have to be mapped to simple Yes/No
- Knut 1 Feb 2008
  I agree that we should have a look into SNOMED CT (SCT) to get some ideas.
  SCT has almost 1000 different types of assessment scales, which some are similar types as covered in Heathers proposal.
  For this to be valuable, we need to define which scale to use for what symptoms.
  We should also ensure that we cover the ones used by ICF (International Classification of Functioning, Disability and Health). Most of these are general, but are intended to be used specifically on i.e. body functions or Activity & Participation:
  body functions
  0 No impairment means the person has no problem
  1 Mild impairment means a problem that is present less than 25% of the time, with an intensity a person can tolerate
  and which happens rarely over the last 30 days.
  2 Moderate impairment means that a problem that is present less than 50% of the time, with an intensity, which is
  interfering in the persons day to day life and which happens occasionally over the last 30 days.
  3 Severe impairment means that a problem that is present more than 50% of the time, with an intensity, which is
  partially disrupting the persons day to day life and which happens frequently over the last 30 days.
  4 Complete impairment means that a problem that is present more than 95% of the time, with an intensity, which is
  totally disrupting the persons day to day life and which happens every day over the last 30 days.
  8 Not specified means there is insufficient information to specify the severity of the impairment.
  9 Not applicable means it is inappropriate to apply a particular code (e.g. b650 Menstruation functions for woman in
  pre-menarche or post-menopause age).
  
  Activity & Participation:
  0 No difficulty means the person has no problem
  1 Mild difficulty means a problem that is present less than 25% of the time, with an intensity a person can tolerate
  and which happens rarely over the last 30 days.
  2 Moderate difficulty means that a problem that is present less than 50% of the time, with an intensity, which is
  interfering in the persons day to day life and which happens occasionally over the last 30 days.
  3 Severe difficulty means that a problem that is present more than 50% of the time, with an intensity, which is
  partially disrupting the persons day to day life and which happens frequently over the last 30 days.
  4 Complete difficulty means that a problem that is present more than 95% of the time, with an intensity, which is
  totally disrupting the persons day to day life and which happens every day over the last 30 days.
  8 Not specified means there is insufficient information to specify the severity of the difficulty.
  9 Not applicable means it is inappropriate to apply a particular code
  On the other hand...
  If we are only looking into one dimension like the one in the severity list in the AT, we should ensure a clean mapping to SNOMED (SCT) i.e. like this:

AT	SCT
1: trivial
2: mild	mild
	mild to moderate
5: moderate	moderate
	moderate to severe
8: severe	severe
9: very severe	fatal

Sam 1 Feb 2008
It is essential that we have a clean mapping to SNOMED and we need to make sure we have a comprehensive approach to symtoms that will stand the test of time. Heather's set of qualifiers on symptoms (under the heading of severity) is a good set and shows that people want to say all sorts of things about different symptoms. It is interesting to consider which of them are useful to have in a structured form and which are quite different notions.
The idea of fetal movements is something that will definitely need to be modelled explicitly for maternity so we can leave that out of the mix. I do think visual analogue scales for pain and perhaps some other symptoms might be worthwhile. Which of the scales Heather has put on the table do people think should be an explicit cluster associated with symptom (we can have a slot and allow different scales to be used as required).
- Heather 1 Feb 2008
  I've been thinking about severity a lot;-( Maybe too much! My evolving thoughts are below, and I'd like your comments:
  
  Modelling severity is like trying to model a moving target - it means different things to different people and in different context. Yet, interestingly, clinicians can make sense of it in context. Sometimes intensity, sometimes frequency, sometimes extent, and sometimes other things. So my initial approach was to archetype all these supposed component concepts within an overarching 'severity' archetype.
  
  But the model is not clean if we suggest that all intensity, frequency, extent etc also correlate with severity etc. For example, the fetal movement intensity is right for intensity but does not carry across to severity. So my original thought to draw clusters for each of intensity, extent etc could be drawn in to a single severity cluster won't work.
  
  Perhaps not all clinical concepts can be modelled succinctly, and without causing too much complexity. In fact, perhaps severity is something cannot be modelled as a unique concept by itself, as it cannot easily be pinned down.
  
  The converse is also important - if we do actually succeed in modelling severity in one archetype, then each time severity is mentioned in an archetype, there will be a number of elements/components that are not relevant in the context and which will have to be removed from the template out EVERY time eg extent won't work with severity of pain. This approach doesn't work well - wrong methodology, I think.
  
  So, another alternative - what about going back to basics here. What about making generic element archetypes for the types of data representation that we want to use for measuring various representations of severity? Then these can be used/re-used where appropriate for the context and effectively bypasses the need for us to 'define' severity concretely in an archetype.
  
  So by this I mean separate generic archetypes that can be re-used all over, including representations of severity, intensity, frequency etc etc, eg commonly used choice selections or scales, which can be specialised as needed. Each of these scales should be represented as a cluster (multi-element) or element (single element) archetype, as appropriate:
  
  -         Quantity/Length - 100mm - for marking a point on a Visual Analogue scale - for pain etc.
  -         Generic Scale 0-10 - ordinal. Can be specialised eg for recording the concept marked as 'degree' in the attached image - 0 = not present; through to 10 = extreme.
  -         Generic Scale 0-5 - ordinal that suits other smaller scales. Can be specialised eg first to fourth degree tears (and template out the 5^th), Grade I - V, or First to Third degree for burns. [Additional question - Should this be a separate smaller archetype comprising only 5 elements should the 6^thto 10^th elements be removed from the 0-10 scale and re-use the 0-10 archetype instead?]
  -         Text - comprising commonly used internal code sets eg
            o   complete/partial/incomplete;
            o   worse/same/better;
            o   acute/chronic;
            o   minor/major;
            o   ordinary activity/slight limitation of ordinary activity/marked limitation of ordinary activity; inability to perform ordinary activity.
            o   Not at all/slightly/moderately/considerably/extremely
            o   Low grade/high grade
  These examples are from the mindmap, but all other suggestions are gratefully received - as per Sam's request, below.
  
  In practice, sometimes these scales might be used to represent severity, but at other times can be used to represent other concepts as well.