Stream: implementers
Topic: Blood Test Profiles
Olivia Morrow (Nov 25 2019 at 17:57):
Do you make a separate observation profile for each type of blood test? Or is there a good way to aggregate multiple blood tests into a single profile?
Eric Haas (Nov 25 2019 at 18:09):
what do you mean by a good way?... you can profile each panels and atomic test result and fix the codes and units. You typically will cascade the profiles starting with generic lab profile like US Core and then constraint it further for example for quantitative and qualitative results and then nail down the codes and units and interpretation values in a final profile.
Another way that is newer and less mature is to use ObservationDefinition which kind of a shortcut.
Olivia Morrow (Nov 25 2019 at 18:16):
what do you mean by a good way?... you can profile each panels and atomic test result and fix the codes and units. You typically will cascade the profiles starting with generic lab profile like US Core and then constraint it further for example for quantitative and qualitative results and then nail down the codes and units and interpretation values in a final profile.
Another way that is newer and less mature is to use ObservationDefinition which kind of a shortcut.
I mean, is it necessary and/or acceptable to have a separate blood test observation profile for each type of lab test?
Lloyd McKenzie (Nov 25 2019 at 20:03):
Generally you should have a separate Observation instance for each piece of information that could reasonably be evaluated independently. Yes, that makes for a lot of Observations. However, it provides the freedom to query what data you want without retrieving content that isn't relevant. It also means that the status for each value is managed independently - if you need to correct one measurement in a panel, you don't need to flag every resource in the panel as corrected.
Lloyd McKenzie (Nov 25 2019 at 20:05):
It's possible to combine multiple values into a single Observation using Observation.component, but that's used for elements that are either not useful on their own (e.g. "sensitive" and "to what antibiotic" in a microbiology result need to be conveyed together in a single Observation to be useful. We also allow packaging of data together that is almost universally looked at and interpreted together (e.g. components of an APGAR or the systolic and diastolic values of a blood pressure).
Last updated: Apr 12 2022 at 19:14 UTC
