Stream: Orders and Observation WG
Topic: Report Interpretation advice
Grahame Grieve (Oct 25 2018 at 14:30):
I'm just preparing a presentation around laboratory medicine and FHIR. It seems to me that looking forward, it would be very good to be able to put a link into a DiagnosticReport for a CDS Hooks server that you could use to get interpretation advice about the report.
Grahame Grieve (Oct 25 2018 at 14:31):
I'm thinking about when reports start becoming available to a patient, the ability to offer an active interface for understanding the results and seeking help to make sense of them. E.g. a business opportunity for a lab
Grahame Grieve (Oct 25 2018 at 14:32):
is this worth defining and extension for?
Lloyd McKenzie (Oct 25 2018 at 15:18):
Hooks are initiated by a particular workflow step. I don't understand what it would mean for a hook to be associated with a data record. How would a system know when to invoke the hook? Or on behalf of what user?
Kevin Power (Oct 25 2018 at 15:21):
What about a link to a SMART App instead?
Rob Hausam (Oct 25 2018 at 18:29):
A SMART App might make sense.
Eric Haas (Oct 25 2018 at 18:51):
this sounds blue button like
Grahame Grieve (Oct 26 2018 at 13:18):
I don't know how you link to a smart app. How can you run an arbitrary smart app? but I can imagine calling the hook 'view-report' with an id from any application, providing some context, and getting some cards back .
Grahame Grieve (Oct 26 2018 at 13:18):
the workflow step is clear; it's the appropriate server that's not clear
Lloyd McKenzie (Oct 26 2018 at 14:49):
If the extension makes the hook clear, then I suppose so. But there are other workflow points that could be invoked around a report. I'm not sure that an EHR would ever invoke a hook on an unknown server though.
Kevin Power (Oct 26 2018 at 15:25):
Just thinking about complex reports where information or even suggestion cards may not be sufficient, but could be presented much better in an app. I don't think SMART has defined a way to link to an app in this way, but I think it would be a reasonable thing to consider. On the other hand, if doing it has a hook that returns an App Link card is better aligned, that still meets the requirements I have in my head. Either way, seems like the CDS Hooks Server/app needs to be known to the EHR ahead of time.
Grahame Grieve (Oct 27 2018 at 07:52):
I'm not thinking about EHR. I'm thinking about PHR.
Lloyd McKenzie (Oct 27 2018 at 14:55):
What are the workflow points for hooks in a PHR?
Lloyd McKenzie (Oct 27 2018 at 14:55):
PHRs would still (presumably) need to trust the hook endpoints
Grahame Grieve (Oct 27 2018 at 17:02):
well, this is case where there'd probably no pre-fetch, no further access - the PHR assumes that the claimed source knows the report, so just provides an Id, and the source assumes that the PHR has access to the report, since it's providing an id (and patient details of some kind).
Lloyd McKenzie (Oct 27 2018 at 19:16):
CDS Hooks potentially provides access to the hook server to other data - and it's conceivable that a renderer might want access to other data - e.g. to highlight risks, to show trending, etc. No PHR should be providing such access to data they don't know and trust - even if a link to such a server happens to be included in a report that somehow gets loaded into the PHR.
Grahame Grieve (Oct 27 2018 at 19:23):
you're making a lot of policy assumptions there. Also, there's a big difference between knowing what the server is for a report, and deciding whether you trust it
Lloyd McKenzie (Oct 27 2018 at 19:41):
I'm trying to understand utility in a real-world space. If the situation is one where there's 5-10 possible hooks servers that can do the rendering, the PHR supports all of them and the report just indicates which one to use - ok. But in that case, you could just as easily fire the hook to all the servers and the one that knows how to render it would. I don't see how - in the real world - an EHR, PHR or any other system would automatically trust a server that was mentioned in a report. And without that, I'm not terribly clear on the utility of including a link on the report. That doesn't mean there isn't utility - just that I'm not understanding the business case well enough to see it.
Grahame Grieve (Oct 27 2018 at 19:44):
5-10? what about 50k?
Grahame Grieve (Oct 27 2018 at 19:45):
I do not see why trust is a binary decision in this regard. A PHR could well contact the mentioned server for cards, and display the cards.... you seem to think that's somehow bizarre, but I don't know why
Lloyd McKenzie (Oct 27 2018 at 19:52):
A PHR could do all sorts of things. I'm having trouble imagining a situation where and EHR or PHR would contact a server it hadn't been specifically configured to trust. That's certainly not something I'm aware of any system doing now - or having indicated an intention to do. But perhaps my circle of awareness isn't broad enough.
Grahame Grieve (Oct 27 2018 at 19:55):
problem: what server would I contact about this report? (Then I could decide whether I can trust it some grounds).
Grahame Grieve (Oct 27 2018 at 19:55):
you seem to be completely focused on the wrong thing
Lloyd McKenzie (Oct 27 2018 at 20:16):
But with CDS Hooks, you generally don't know what server cares about something - you fire to everyone who offers the service. I'm not sure why we're injecting a difference here. The only difference is automatic registration of a recipient, which is the piece I'm questionning.
Grahame Grieve (Oct 27 2018 at 20:16):
all 50k servers in the world?
Grahame Grieve (Oct 27 2018 at 20:16):
and I haven't said anything at all about registration
Kevin Power (Oct 27 2018 at 23:55):
Regarding what server to contact, would it be the lab where the report came from? Seems like if the lab offer this “patient friendly” capability, perhaps they have to offer the server/app link with the report in some way, like part of this extension you are considering? I don’t know that either CDS Hooks or SMART would support that as they exist today or not. It would be up to the PHR if they want to trust it.
Grahame Grieve (Oct 28 2018 at 00:21):
Yes it would be the lab. That’s the purpose of the extension. It would be up to the phr to decide. But it can’t decide if it doesn’t know
Last updated: Apr 12 2022 at 19:14 UTC