Questionnaire data is common in the eHealth environment, really capturing workflow and ensuring information is captured that is known to be highly relevant. One problem is interoperability and extensibility of information captured in this manner.
Free text questions
The first approach is to take a very generic questionnaire with a repeatable element 'question' and an answer of any data type. This allows virtually any questionnaire to be built using a template, typing the text of the question and setting the type of answer expected. This has the advantage that any simple questionnaire can be designed at the level of a template and added to data entry. Clearly, the information will not be translated nor is it likely that it will be reliable outside the system where it was created. However, there is no doubt that such an approach is valuable for a range of documentation purposes.
Terminology based questions
It is possible, with more design, to provide a base question and then provide the 'tail' elements from terminology. An example might be "Have you ever suffered from:" with the tail elements "Asthma", "Angina" etc. It is then possible to restrain the name of these elements to a subset of terminology from, for example, SNOMED or ICD10. This might be a specialisation of the free text questionnaire archetype which will allow queries to find all questionnaires. This approach will allow translation (using terminology translation) and will provide computable responses to queries and inferencing.
Archetype based questions
Finally, a questionnaire can be archetyped. This provides complete control of the form of the questionnaire and the results and ensures that these will always be the same. Such an approach is obviously well suited to a questionnaire that is used internationally (The WHO has some well tested check lists) or within a jurisdiction when very tight control of the data collection is required (perhaps for automatic processing).
Comments (1)
08-Sep-2008
Gerard Freriks says:
In like to think in documentation patterns. These Documentation Patters will res...In like to think in documentation patterns.
These Documentation Patters will result in Generic Archetypes ( sometimes called by me Ur-Archetypes) that get used and specialised in the Observation Archetypes.
In Templates using archetypes it is possible to construct a questionnaire.
Templates are for LOCAL use only. They are good to catch and display local information requirements but can NEVER play a role in semantic interoperability.
For QUESTIONS the pattern is:
<Question> < Topic> <Possible answers>
Questions:
Some can be standardised like: 'Have you suffered from ...', 'Do you ...', 'How many ...'
Topics:
Can be any complaint, diagnosis, situation, substance use, etc
Sometimes these topics can be found in a coding system, many times not.
Possible answers:
Boolean types, Ordinal types, Counts types,
and the null flavor types like: skipped and not answered, skipped not asked,
In addition there is the problem to code for a modifier for a negative question: 'Do you have NOT smoked?' or ' Do have smoked NOT 5 cigarettes' in order to differentiate this from 'Do you have smoked?' "Do you have smoked 5 cigarettes?'
An other interesting modifier is the certainty, uncertainty, as a qualification by the patient or the healthcare provider.
This suggests at least a documentation pattern:
<modifier> <Question>, <modifier> <Topic>, <modifier> <Possible answers>