Stream: implementers
Topic: Patient birth date uncertainty
Cooper Thompson (Feb 27 2019 at 21:57):
PA wants to gather input from folks related to GF#20341. Specifically:
1) How common are systems that can document uncertainty thresholds for the patient's date of birth. What options are available?
2) Are there any known realm-specific accuracy thresholds (for example: Australia)
We are trying to acertain whether this should be a standard extension, or if there is enough realm-specific variance that we should push this to realm-specific extensions.
Brian Postlethwaite (Feb 28 2019 at 06:20):
Fyi, the Australian extension supports granularity to mark the day, month and year as estimates independently, e.g. Estimated year, but day and month accurate. Know the birthday, but not sure which year, could be 88 or 89.
Finnie Flores (Feb 28 2019 at 14:26):
In Canadian Institute for Health Information (CIHI), at least 5 of the data holdings captures an indicator if birthday is estimated. https://www.cihi.ca/en/access-data-and-reports/make-a-data-request/data-holdings
Grahame Grieve (Mar 01 2019 at 22:21):
there are a number of systems that allow for catching the Australian uncertainty flag, but I've never seen them actually do anything with it other than display it back to the user
Brian Postlethwaite (Mar 03 2019 at 22:01):
The cases I've been exposed to is for govt reporting. Don't know what they actually use it for...
Brian Postlethwaite (Apr 24 2019 at 19:21):
@Alexander Henket do you know if there is a need for this in the Netherlands?
Alexander Henket (Apr 24 2019 at 20:56):
I have not encountered a need like this but uncertainty is a factor. I’ll ask around.
Brian Postlethwaite (Apr 24 2019 at 21:26):
Thanks.
Stefan Lang (Apr 24 2019 at 22:18):
Certainty is a factor e.g. in epidemiological registries where it is taken into account for statistical computing.
Most of what is typically represented in the systems I'm aware of is covered by the date data type (Y, YM, YMD), but it does not allow for things like "exact year, estimated month".
So IMHO a standard extension would be useful, especially to independently express year/month/day to be an estimate.
But this should not be limited to the birth date. Uncertainty also appears in death dates or patient reported information.
Lloyd McKenzie (Apr 24 2019 at 23:33):
Another argument in favor of it being an operation is that if Account maintains a balance, then every time a new financial transaction resource instance is posted, you'd need to do it in a transaction to ensure that content remained consistent.
Brian Postlethwaite (Jul 17 2019 at 19:42):
A mis-posting @Lloyd McKenzie ?
Brian Postlethwaite (Jul 17 2019 at 19:48):
The Patient Adminstration Workgroup is looking at this issue (GF#20341), and given the lack of interest in it at this stage, we recommend doing local extensions until further feedback is provided.
If you do create your own local extension for this, please let us know, and consider the other ones that have already been allocated to see if are consistent with you usage requirements, and we can re-consider as a standard extension.
At this stage only Australian definition, and Canadian definition, and no detail as to if they are the same usages.
Lloyd McKenzie (Jul 17 2019 at 20:34):
Apparently, but so long ago, no clue where it should have been :(
Craig Newman (Jul 19 2019 at 15:30):
There is a potential use care for a "estimated date" flag extension on different date elements. In the immunization world sometimes the patient doesn't know exactly when they got their vaccination, they just know "it was in October last year". In this case, the typical advice is to select the 15th of the month, but it would be nice to flag it as an estimate. Or sometimes "evidence of immunity" observations have an uncertain date (I don't need my varicella vaccination because I had chicken pox back in 2005). It's not a critical need now for immunization, but it's something that might be useful in the future.
Richard Townley-O'Neill (Jul 22 2019 at 01:22):
@Craig Newman
FHIR dates can be partial e.g. 201810 for October last year.
Brian mentioned the Australian extension date-accuracy-indicator. But it is unclear what real use is made of this.
Katarina (Jul 22 2019 at 08:01):
Patient.birthdate is a partial date. How can I define, that birthdate must be a complete date?
Brian Postlethwaite (Jul 22 2019 at 12:57):
You could set its length to a minimum.
Last updated: Apr 12 2022 at 19:14 UTC
