Stream: Orders and Observation WG
Topic: Make ObservationDefinition linkable to Patient
Ramūnas Jurgilas (Jan 10 2022 at 14:32):
Currently ObservationDefinition can not be searchable using patient ID. To support this functionality needs to implement/add custom search parameters.
I would like , that ObservationDefinition contain possibility link to patient. As example Observation have subject achieving this functionality.
Rik Smithies (Jan 10 2022 at 14:48):
What would be the use case for that? Definitions are normally independent of any patient. An observation for a patient may be based on an observation definition, but I cannot think of a need for a direct link between a definition and a patient. Search is not the issue here, it is the need (or lack of it) for subject at all.
Ramūnas Jurgilas (Jan 10 2022 at 15:08):
In my case medical device (as example monitoring heart rate) gets assigned to particular patient. Then for particular patient we need to assign high and low heart rate ranges. To achieve it I would use ObservationDefinition with reference to patient and later query high/low values when needed using patient ID.
Where do you think I should store this type of information?
Rik Smithies (Jan 10 2022 at 15:34):
Reference ranges normally go in the Observation itself
Lloyd McKenzie (Jan 10 2022 at 15:58):
The point here is to establish a standard reference range for a particular type of Observation for a particular patient that is independent of the specific Observations. E.g. you could indicate that a particular patient's normal heart range is X. You then might collect 100s of heart rate measurements. The normal ranges shown in those Observations would be grabbed from the patient-specific ObservationDefinition.
Ramūnas Jurgilas (Jan 10 2022 at 16:18):
Thanks @Lloyd McKenzie for putting describing it more clearly.
Rik Smithies (Jan 10 2022 at 16:20):
There is the patient-specific-device reference range - which is what I thought this was. Then there is a patient-specific device-independent "normal" range, which seems different. Which is this?
Lloyd McKenzie (Jan 10 2022 at 16:32):
Ideally, we should have a mechanism to handle both. So that would argue for ObservationDefinition to have a reference to Device as an extension too.
Rik Smithies (Jan 10 2022 at 16:41):
There is also the existing principle that reference ranges for actual observations that have happened (not the patient's general condition) go on the actual observation. I don't see that a personal device would mean that changes, or you would be looking in two places.
Lloyd McKenzie (Jan 10 2022 at 16:46):
The normal ranges for a particular Observation can be determined in two ways:
- they can be patient-independent and instead based just on the equipment, possibly taking into account gender, age or other general patient characteristics
- they can be patient-specific based on historical measurements and/or specific clinical decisions. E.g. "For this patient, only alert as 'abnormal' if they are outside this range".
What we need to do is provide a mechanism to define and store the second type. You would still send the ranges on a per Observation basis because the patient-specific normal ranges might change from month to month or year to year and would want to know on a given Observation what the range(s) are for that Observation.
Rik Smithies (Jan 10 2022 at 16:55):
The first bullet could be split into "from any device" and "from a patient specific device" but the data should be the same in each case. I thought the question was the second of those (otherwise why even mention device in the question) and I was saying there are the same really. But @Ramūnas Jurgilas can confirm which case this one is. But the second bullet could be supported too (seems like https://chat.fhir.org/#narrow/stream/179166-implementers/topic/Defining.20and.20storing.20Individual.20referenceRange.20for.20patients)
Lloyd McKenzie (Jan 10 2022 at 17:08):
Right. What's being asked for here is for ObservationDefinition to support being patient-specific (via an extension) - as is also proposed in the thread you referenced.
Ramūnas Jurgilas (Jan 11 2022 at 06:50):
In my case it is second case: "they can be patient-specific based on historical measurements and/or specific clinical decisions. E.g. "For this patient, only alert as 'abnormal' if they are outside this range"."
Last updated: Apr 12 2022 at 19:14 UTC
