openEHR-technical Digest, Vol 64, Issue 6

Thomas Beale thomas.beale at
Mon Jun 5 12:54:49 EDT 2017

this has to be essentially correct, I think. If you think about it, 
scores (at least well designed ones) are things whose 'questions' have 
only known answers (think Apgar, GCS etc), each of which has objective 
criteria that can be provided as training to any basically competent 
person. When score / scale is captured at clinical point of care, any 
trained person should convert the observed reality (baby's heartrate, 
accident victim's eye movements etc) into the same value as any other 
such person. In theory, a robot could be built to generate such scores, 
assuming the appropriate sensors could be created.

With 'true' questionnaires, the questions can be nearly anything. For 
example, my local GP clinical has a first time patient questionnaire 
containing the question 'have you ever had heart trouble?'. It's pretty 
clear that many different answers are possible for the same physical 
facts (in my case, occasional arrhythmia with ventricular ectopics whose 
onset is caused by stress, caffeine etc; do I answer 'yes'? - maybe, 
since I had this diagnosed by the NHS, or maybe 'no', if I think they 
are only talking about heart attacks etc).

My understanding of questionnaires functionally is that they act as a 
rough (self-)classification / triage instrument to save time and 
resources of expensive professionals and/or tests.

There is some structural commonality among questionnaires, which is 
clearly different from scores and scales. One of them is the simple need 
to represent the text of the question within the model (i.e. archetype 
or template), whereas this is not usually necessary in models of scores, 
since the coded name of the item (e.g. Apgar 'heart rate') is understood 
by every clinician.

Whether there are different types of questionnaires semantically or 
otherwise, I don't know.

- thomas

On 05/06/2017 09:48, William Goossen wrote:
> Hi Heather,
> the key difference is that the assessment scales have a scientific 
> validation, leading to clinimetric data, often for populations, but 
> e.g. Apgar and Barthell are also reliable for individual follow up 
> measures.
> a simple question, answer, even with some total score, does usually 
> not have such evidence base. I agree that in the data / semantic code 
> representation in a detailed clinical model it is not different.

Thomas Beale
Principal, Ars Semantica <>
Consultant, ABD Team, Intermountain Healthcare 
Management Board, Specifications Program Lead, openEHR Foundation 
Chartered IT Professional Fellow, BCS, British Computer Society 
Health IT blog <> | Culture blog 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the openEHR-technical mailing list