Stream: committers
Topic: inventory tracking
Jose Costa Teixeira (Mar 21 2020 at 05:38):
There are several discussions about tracking inventory (the efforts to address pandemic have put a strong focus on operational as well as clinical).
Jose Costa Teixeira (Mar 21 2020 at 05:39):
We did a long analysis in IHE for that - and until now we (HL7) did not have a strong enough need.
Jose Costa Teixeira (Mar 21 2020 at 05:39):
should we draft a "inventory status" resource?
Jose Costa Teixeira (Mar 21 2020 at 05:45):
or we could work up the Product pattern to derive an inventory product...
Jose Costa Teixeira (Mar 21 2020 at 05:47):
First one could look like this:
- InventoryStatus
- inResponseTo (if this is a response to an inventory request)
- dateTime (of status report)
- reporter
- entries
- location
- content
- item (codeableReference)
- quantity
- attributes (lot, serial, etc)
- Additional status info (valid, temperature, etc.)
Jose Costa Teixeira (Mar 21 2020 at 05:48):
second would be similar to the product pattern
David Pyke (Mar 22 2020 at 13:15):
Can this be done as an Extension on Device? Most of the information (location, etc.) is already there. We could add an inventoryStatus element and to handle the specifics then a count query.
Jose Costa Teixeira (Mar 22 2020 at 14:16):
that would be in line with updating the product model
Jose Costa Teixeira (Mar 22 2020 at 14:19):
and it would mean repurposing the device resource IMO. Can be done, but the meaning of the resource becomes different. And we still have an issue: Medication, Device etc still diverge from the pattern. Clinically, an item is either one or the other. From an inventory perspective that would be a bigger issue (in that case you typically don't know/don't care/can't know whether something is a device or medication).
Jose Costa Teixeira (Mar 22 2020 at 14:20):
putting an extension on a device means "by the way, this is stock" and that is not the way stock is handled. It's not product-centric with an inventory extension but the other way around
Jose Costa Teixeira (Mar 22 2020 at 14:21):
I wonder how people are handling this.
Jose Costa Teixeira (Mar 22 2020 at 14:33):
Until now the need for inventory tracking has been excluded from scope. I'd personally like to have something, but it should be aligned with Supply Chain pratices - not that FHIR would suddenly be a supply standard, but for FHIR not to break the treaceability chain.
David Pyke (Mar 22 2020 at 14:43):
Public health is going to want to know how many ventilators are in a clinic/hospital so they can assess the need. So, it might be out of scope but that's a use case that has to be addressed
David Pyke (Mar 22 2020 at 14:44):
Add to that, there's going to need to be a way to route patients to places with open beds. That count will be needed as well
Jose Costa Teixeira (Mar 22 2020 at 14:44):
yes. when I say "it has been excluded from scope" it is obviously not my preference. Until now there was just no interest
Jose Costa Teixeira (Mar 22 2020 at 14:47):
we can rdo something rather quickly, but I would not do it as an extension to device.
David Pyke (Mar 22 2020 at 14:47):
A new resource would take a lot of time to get moving, an extension is quick and we're short of time
Jose Costa Teixeira (Mar 22 2020 at 14:48):
Do we want a standard or a quick patch?
Jose Costa Teixeira (Mar 22 2020 at 14:50):
If we want implementers to quickly patch something, then they can use anything - custom resources, existing standards, extensions...
David Pyke (Mar 22 2020 at 14:50):
Both. A quick patch to get us through this pandemic and a standard to fix this forever
David Pyke (Mar 22 2020 at 14:51):
If we come up with an extension and a IG that show it's use, then it gets to be used in a standard way. Then we can take the long look and make sure we get things done right for the future
Jose Costa Teixeira (Mar 22 2020 at 14:51):
Why not a draft resource?
David Pyke (Mar 22 2020 at 14:52):
Because a resource needs to have a complete understanding of the issue and use case. An extension can be a quick hack to solve an immediate problem
Lloyd McKenzie (Mar 22 2020 at 14:53):
If its's going to be used with existing systems, they won't be able to use a new resource
Jose Costa Teixeira (Mar 22 2020 at 14:54):
well we have a loong white paper, years of discussions and one transaction that was meant for this. I think there will be enough understanding of scope.
Jose Costa Teixeira (Mar 22 2020 at 14:56):
which means we can deliver something. We can see what data we need to exchange, and we can follow the 2 paths: make an extension AND prepare a new resource/profile
David Pyke (Mar 22 2020 at 14:58):
Right, so let's get the extension written and start the doc review for a new resource.
Jose Costa Teixeira (Mar 22 2020 at 14:58):
@Lloyd McKenzie is this for discussion in the WF call tomorrow? (i mean - is it a candidate for discussion?)
Jose Costa Teixeira (Mar 22 2020 at 15:00):
extension on product? Device, DeviceDefinition, Medication, Medknowledge, all the products...
or on list/composition - because inventory report is a list of items?
Jose Costa Teixeira (Mar 22 2020 at 15:02):
and we should exclude / clarify inventory displacement (which can either be a supplyrequest/supplyDelivery or a simple location=in transport)
Lloyd McKenzie (Mar 22 2020 at 15:04):
At this point, I think we'd primarily be looking at extensions to R2 and R3 resources that have been semi-widely implemented - which would exclude DeviceDefinition and MedicationKnowledge
Lloyd McKenzie (Mar 22 2020 at 15:05):
It seems more OO than workflow, but I'm happy to use workflow time on it.
Jose Costa Teixeira (Mar 22 2020 at 15:07):
(I think it his is OO too and I'll check when we have the Product discussions in OO as well. But would be good to align early. )
Jose Costa Teixeira (Mar 22 2020 at 15:08):
instead of an extension inside the product resources, why not an extension around those resources? i.e. an inventory extension that contains a product?
Jose Costa Teixeira (Mar 22 2020 at 15:09):
could be a profile on list/composition.
Lloyd McKenzie (Mar 22 2020 at 15:10):
If Device is being used to represent a 'kind', no reason not to put the count on the Device - that's where model number and other relevant info will be
David Pyke (Mar 22 2020 at 15:15):
How would you address "open bed" counts? Would a bed be a Device or something else?
Jose Costa Teixeira (Mar 22 2020 at 15:16):
Lloyd McKenzie said:
If Device is being used to represent a 'kind'
that is valid for R3 and R2 indeed. Yes, we can add count (which is in the product pattern), and we can add "who reported / last seen by"
Jose Costa Teixeira (Mar 22 2020 at 15:16):
bed could be a device or healthcareservice.
David Pyke (Mar 22 2020 at 15:17):
Wait, a bed is a Location according to the description. NEvermind
Jose Costa Teixeira (Mar 22 2020 at 15:17):
what about meds? I expect we'll see drug shortages
Jose Costa Teixeira (Mar 22 2020 at 15:18):
@David Pyke which description?
We need to know how many beds we have in city region X.
David Pyke (Mar 22 2020 at 15:19):
The Location resource lists a bed as a use case and has an element that uses the V2 bed status valueset
Jose Costa Teixeira (Mar 22 2020 at 15:20):
ok but I don't think that is what we need here. Unless we want an inventory of locations - which I think we don't. I think we want an inventory of services.
David Pyke (Mar 22 2020 at 15:20):
"Examples of Locations are:
Building, ward, corridor, room or bed"
David Pyke (Mar 22 2020 at 15:22):
You could do a count of all Locations where physicalType = "bd" and get the count for that system
Jose Costa Teixeira (Mar 22 2020 at 15:22):
that doesn't seem like an extension
David Pyke (Mar 22 2020 at 15:22):
A hospital with no open beds would need ambulances routed a hospital that does
David Pyke (Mar 22 2020 at 15:23):
Nope, that's a simple query for count. No extension needed.
David Pyke (Mar 22 2020 at 15:23):
Except it's not a search parameter for Location
Jose Costa Teixeira (Mar 22 2020 at 15:24):
in that case you can not query a hospital's available beds unless the hospital keeps track of all their beds and statuses (as instances of Location).
Jose Costa Teixeira (Mar 22 2020 at 15:27):
which seems against what we need - we need some mechanisms that can be used now. I wouldn't go to a hospital create one instance of Location for every bed. And today beds are being shuffled around the hospitals..
In this case I would rather stick to a Device or HealthcareService ("Bed", ICU Bed") and put a Quantity on it.
Jose Costa Teixeira (Mar 22 2020 at 15:27):
(makes sense?)
David Pyke (Mar 22 2020 at 15:30):
But Location is the standard and doing an _count tells you how many are available. If you need ICU beds, you filter on partOf -> Location.name=ICU
Lloyd McKenzie (Mar 22 2020 at 15:30):
Hospitals typically will have a Location for every bed - that's how they track who's where. Beds are certainly 'moveable' Locations (just like ambulances).
Jose Costa Teixeira (Mar 22 2020 at 15:31):
So we're counting locations instead of equipment?
David Pyke (Mar 22 2020 at 15:31):
Well, yeah, that's how it's defined
David Pyke (Mar 22 2020 at 15:31):
Hospitals aren't likely to have many beds in storage.
Jose Costa Teixeira (Mar 22 2020 at 15:32):
they have been defined for a specific purpose in FHIR.. I don't think that is what is happening in reality nowadays beds are being repurposed, invented...
Lloyd McKenzie (Mar 22 2020 at 15:32):
Having to detect ICU beds based on name of parent location seems less than ideal. Unfortunately there's nothing on Location that talks about capabilities - that might be a good place for an extension.
Lloyd McKenzie (Mar 22 2020 at 15:33):
Sure - but they'll still be captured as Location instances
Jose Costa Teixeira (Mar 22 2020 at 15:33):
in the middle of a crisis?
Lloyd McKenzie (Mar 22 2020 at 15:33):
You can't move a patient to a bed if it isn't captured as a Location - unless you've given up on using your EHR entirely, in which case, we're not going to get useful information out of the system anyhow.
David Pyke (Mar 22 2020 at 15:33):
I guess if Location.type as well as physicalType is populated, you'd get ICU beds?
Lloyd McKenzie (Mar 22 2020 at 15:34):
The pop-up tents are going to be locations and every cot in them will be a location
Jose Costa Teixeira (Mar 22 2020 at 15:35):
from my perspective this is simpler: We can count HealthcareService instances. Seems better than counting locations.
Lloyd McKenzie (Mar 22 2020 at 15:36):
Better = whatever existing systems are doing. My biggest concern w/ expending time on this topic is that we don't have people telling us what's actually needed.
David Pyke (Mar 22 2020 at 15:36):
I have someone asking me about it. I may have more info soon
David Pyke (Mar 22 2020 at 15:36):
The timing of this conversation couldn't be better
Jose Costa Teixeira (Mar 22 2020 at 15:38):
agree there, so we should be careful about repurposing what we have without input.
To me a bed is not a location in this context. it is not about "where something is" but "what do we have"
Jose Costa Teixeira (Mar 22 2020 at 15:38):
I think @Keith Boone worked on a profile for bed availability
Jose Costa Teixeira (Mar 22 2020 at 15:40):
https://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_Suppl_BED.pdf
seems about admission and assigning people to beds
David Pyke (Mar 22 2020 at 15:41):
I'm not sure more than Location.physicalType=bd, Location.operationalStatus=U (unoccupied) with a search _count is different that what would be needed. And, like I said, if Location.type is populated, you can filter on bed type
Jose Costa Teixeira (Mar 22 2020 at 15:45):
this would serve a purpose to create a list of available beds, which is one of the purposes. And it hasoperational dependencies.
- hospitals track each bed as a resource instance.
- each instance=1 bed (if we have more, how does that work? add a .quantity?)
- this info must be up to date.
Jose Costa Teixeira (Mar 22 2020 at 15:45):
anyway, respirators. those are not locations.
Do we want to track drug availability?
What about hand sanitizer?
Jose Costa Teixeira (Mar 22 2020 at 15:47):
and masks?
Jose Costa Teixeira (Mar 22 2020 at 15:48):
I think it is rather simple to bring up something for inventory of item, where item=product, device..,
David Pyke (Mar 22 2020 at 15:48):
Right. REspirators/ventilators are Devices and would need an extension.
Medications would need someone from Parmacy to speak up
David Pyke (Mar 22 2020 at 15:49):
Masks and hand sanitizer, I have no idea
Jose Costa Teixeira (Mar 22 2020 at 15:50):
this seems difficult. I am trying not to anchor on "let's make an extension".
We have already seen how funny this gets because we have so many different concepts.
Jose Costa Teixeira (Mar 22 2020 at 15:56):
I'll see when OO has product discussions.
Jose Costa Teixeira (Mar 22 2020 at 16:05):
in any case, this has 2 fronts: one is the very short term - we don't need to rifles there, we just do what works ( and make sure it is operational).
The other is the longer-term - for which I hope these discussions help our resources support the supply concerns (we have mostly focused on clinical aspects so far)
Lloyd McKenzie (Mar 22 2020 at 16:29):
Masks would be devices. Hand sanitizer would technically be a medication, though it typically wouldn't be tracked that way. I think immediate focus needs to be on short term that works. We'll worry about the long-term later.
Jose Costa Teixeira (Mar 22 2020 at 16:56):
For short term we can create extension that includes quantity and location (which are already in the product pattern)
Josh Mandel (Mar 22 2020 at 22:05):
This discussion is great -- the kind of stuff I was hoping to get a handle on at #Covid-19 Response > Tracking hospital operations . Does someone want to write up a quick plan + examples in a github repo? Would love to contribute and point to this, but also I don't want to step on toes.
David Pyke (Mar 22 2020 at 22:11):
Talk to Keith, he's way ahead of us.
Josh Mandel (Mar 22 2020 at 23:37):
We just got to catch up over the phone. It sounds like Keith has got a great internal start and is working to share what he's written so far. The focus up to this point if I understand correctly has been on the use of a Group resource to count groups of things (like ICU beds) rather than tracking the individual things themselves (though of course there is a path to doing that as well).
Jose Costa Teixeira (Mar 23 2020 at 04:49):
@Josh Mandel I started this question to separate between from "urgent need" (the Covid-19 track), and whether we want to do something on inventory tracking in the core in the mid term (and I realize I was not successful. Perhaps it's not the time).
For an immediate need I'd say - go with whatever works, and I do not know how people are using it.
Tracking beds can be a Group of Locations, or a List of Devices, or anything. If we need to create an extension that applies to all products, we can create one - I can, if nobody is doing it. Let's see what Keith has and I'm ready to jump in on that too if needed.
Keith Boone (Mar 23 2020 at 14:45):
I did indeed. There will me more on what I'm doing NOW wrt to this and FHIR shortly.
Keith Boone (Mar 23 2020 at 14:46):
I'm US based, and my project is focused on requirements that are very US focused b/c of how our health system uniquely operates, but I'll happily have your help.
Jason Walonoski (Mar 23 2020 at 16:53):
Any thought to using SupplyRequest to track a supply list required/used for the various type of encounters/procedures?
Jason Walonoski (Mar 23 2020 at 16:54):
Seems more appropriate to use that for disposable things like gloves and masks, rather than Device. But that is just me.
Jose Costa Teixeira (Mar 23 2020 at 17:05):
gloves required = SupplyRequest OK
gloves used = SupplyDelivery
Jose Costa Teixeira (Mar 23 2020 at 17:07):
We had a brief discussion in Sydney, whether we materialize it depends on implementer feedback, I guess - around a "product consumption" or generic "ProductDispense" (noting that Dispense is not the same as Consumption but we can make it so)
Keith Boone (Mar 23 2020 at 17:29):
Nope, not really interested in "tracking" data right now. It's an approach that could support reporting, but the key thing we are hearing is that public health needs reporting. They don't want to handle aggregating that information themselves. They just want numbers.
Last updated: Apr 12 2022 at 19:14 UTC
