Hello,
did any off you thought about how a query should look like suppose I want such like this :
“Give me all female patients living in Paris and with no allergies and a the last labresult of type Kreatine is < 20 but within a year”
The hardest part is to combine the demographic information with the medical information. Should you be able to do it with AQL?
An earlier private response to Alessandro before he posted: It might be interesting to find out what other implementers are doing. - thomas
>the first question is: is the patient sex and location stored in the EHR
>in the system? If so, it's just standard EHR-based query.
>
>I'm assuming that the location / address is not however. So a normal query
>can be written to do everything except the 'living in Alkmaar' bit.
That's exactly the boundary between systems "built for"
medical care vs "built for" epidemiology. Both may have
requirements that are contrary to each other, such as this:
>That last item will need a traversal of the reference
>PARTY_PROXY.external_ref <http://www.openehr.org/local/releases/1.0.1/uml/Browsable/_9_5_1_76d0249_1140169202660_257304_813Report.html>,
>if you have it set. But you might not have it set - we don't in most of
>our systems, due to privacy / security.
Karsten
Hi Alessandro
Very good question. I know that some implementers store 'basic query
demographics' in a single persistent composition with an archetype
representing dob, sex, gender etc, to get around this although it to
some extent weakens the ehr / Demographics split.
One approach I would like to see pursued is that we develop a
standardised approach using similar archetypes in a 'demographics
composition' but allowing the backend system to choose whether the
data was sourced from the EHR that composition, or retrieved virtually
from a seaprate Demographics engine as Thomas was suggesting.
That would let us standardise the AQL statements but allow
implementers to meet the needs of different groups as Karsten was
suggesting. Patient registries is an example of where it is more
compelling to hold some demographics data in situ.
I have done a little work on this for the EU PARENT project.
This is the anonymised patient details archetype
http://www.openehr.org/ckm/#showArchetype_1013.1.1745
and seen in use at
http://www.openehr.org/ckm/#showTemplate_1013.26.16
For review purposes, in this template, the anonymised archetype is
carried in the same composition - this would not be the case in a run
time example.
So in this case a standard AQL could be used to retrieve
"Give me all female patients living in Paris and with no allergies and
a the last labresult of type Kreatine is < 20 but within a year"
but how the age, sex, location data is actually resolved is
implementation dependent.
Ian
Dr Ian McNicoll
mobile +44 (0)775 209 7859
office +44 (0)1536 414994
skype: ianmcnicoll
email: ian@freshehr.com
twitter: @ianmcnicoll
Director, freshEHR Clinical Informatics
Director, openEHR Foundation
Director, HANDIHealth CIC
Hon. Senior Research Associate, CHIME, UCL