Stream: implementers
Topic: Reference range
Pranitha Sruthi (Sep 25 2017 at 10:12):
Hi all, I am working on creating profiles for various ECG leads using Observation. I need to add both the upper and lower limits for the amplitude and duration of the respective leads. When I slice the element 'component', the referenceRange in the sliced component does not have upper and lower limits. Where can I specify the limits and units(mV or ms)? Thank you
Lloyd McKenzie (Sep 25 2017 at 14:34):
I know that proper handling of ECGs and similar data remains an open issue. @Eric Haas ?
Eric Haas (Sep 25 2017 at 14:38):
Why doesn't Observation.component.referenceRange work?
Pranitha Sruthi (Sep 26 2017 at 05:01):
@Eric Haas The Observation.component.referenceRange does not have 'high and low' elements.
Pranitha Sruthi (Sep 26 2017 at 05:03):
Pranitha Sruthi (Sep 26 2017 at 05:12):
Unlike Observation.referenceRange, Observation.component.referenceRange doesn't have low and high elements
Lloyd McKenzie (Sep 26 2017 at 05:17):
It uses the same type, so it definitely has those components, but there may be a bug in Forge that's causing them to not be exposed. @Michel Rutten ?
Pranitha Sruthi (Sep 26 2017 at 05:20):
Pranitha Sruthi (Sep 26 2017 at 05:21):
It's not given in the spec
Lloyd McKenzie (Sep 26 2017 at 05:23):
Yes it is. It says "See ReferenceRange" - it's re-using the type defined earlier in the resource.
Pranitha Sruthi (Sep 26 2017 at 05:25):
Ok. But how can I fix values for upper and lower limits?
Pranitha Sruthi (Sep 26 2017 at 05:25):
I also have to fix units for the values
Lloyd McKenzie (Sep 26 2017 at 14:47):
Like I said, there may be a UI issue in Forge.
Michel Rutten (Sep 27 2017 at 11:09):
Hi @Pranitha Sruthi, Forge does not yet support the dynamic expansion of elements that use a contentReference, such as Observation.component.referenceRange. You can manually edit the profile, however if you then open it in Forge, some child constraints may disappear. We still need to implement support in Forge for this part of the FHIR spec.
Pranitha Sruthi (Sep 27 2017 at 11:24):
@Michel Rutten Ok thank you
Pranitha Sruthi (Sep 27 2017 at 11:26):
@Lloyd McKenzie Thank you
Grahame Grieve (Nov 26 2017 at 10:41):
@Melyssa Grieve and I are working to code up a real clinical record in fhir resources. She’s working on the vital signs, and we’ve found a problem
Grahame Grieve (Nov 26 2017 at 10:42):
the chart that she is copying from only has a range, but the profile says it must be a numeric value.
Grahame Grieve (Nov 26 2017 at 12:23):
Grahame Grieve (Nov 26 2017 at 12:24):
@Eric Haas @Rob Hausam thoughts?
Rob Hausam (Nov 26 2017 at 16:00):
It's definitely not the most common case to represent the data this way, but it obviously does occur at times. I think we'll likely need to broaden the Vital Signs profile to also allow valueRange. It's interesting, though, the way this is being charted. The raw measurement of RR is a number, as represented by the dots, but in this case the actual value isn't being preserved and only the range is available. I haven't really considered this before, but I'm not sure if we have a good way of dealing with the precision of data - e.g. recording a point value along with a confidence interval (what we have here is similar, except that the point value itself isn't being recorded). Maybe we should further look at how that aspect should be handled?
Eric Haas (Nov 26 2017 at 16:02):
just to clarify this is a result and not a reference range. The clinician looses count ?
Options :
- we could throw in a range slice for RR, HR
- we could throw in a text slice for all profile?
-
value absent reason + components
@Jenni Syed what do you think?
Eric Haas (Nov 26 2017 at 16:07):
If its a referenceRange then use Observation.referenceRange.text
Eric Haas (Nov 26 2017 at 16:07):
how much of an edge case is this?
Jenni Syed (Nov 27 2017 at 16:39):
If the underlying system actually had the real number, they should send that. I'm assuming in this case the actual data is just the range, and that's not just a display simplification.
Jenni Syed (Nov 27 2017 at 16:40):
They should be able to use the valueRange for these types of cases - so I would prefer a slice. Not sure how broad that slice would need to be?
Grahame Grieve (Nov 27 2017 at 22:41):
I suspect that this would not be an issue in a real world case, where you have electronic tracking from source. But I'm not sure...
Abbie Watson (Dec 14 2017 at 17:10):
I worked for a year or two on a biomarker reference app which dealt pretty exclusively with reference ranges. There's definately situations where people may work exclusively with only reference ranges; particularly in social work and patient education settings, predictive modeling, etc.
Last updated: Apr 12 2022 at 19:14 UTC
