Stream: implementers
Topic: Mapping infusion
Mirjam Baltus (Dec 19 2018 at 08:37):
I have a use case where I need to communicate data about an infusion, including for example information about lumens and lines, and linking to medication that is administered. What would be a good resource type to map this to?
Jose Costa Teixeira (Dec 19 2018 at 12:36):
Are you defining what the infusion should be (like in a catalog), or is this a prescription for infusion, or an administration?
Mirjam Baltus (Dec 20 2018 at 12:13):
It's an administration, I think. The patient has one, and a nurse needs to be able to see the info about it.
It seems like it should be a device, but the actual data I need to put in there doesn't really map to the Device fields. So I was hoping that anyone had already done something like this.
Martin Grundberg (Dec 20 2018 at 12:39):
I'm not sure what you mean with that it should be a Device? Are you thinking of an Infusion pump?
But as for prescription and administration parts, I was looking into this area the other day and posted this question which may be of interest:
https://chat.fhir.org/#narrow/stream/4-implementers/subject/Planned.20Medication.20administrations where also @Jose Costa Teixeira helped me with an answer.
Mirjam Baltus (Dec 20 2018 at 12:54):
What I need to map is the infusion itself, and it will have a link to the medication that is administered. So it's not just the administration of that, but also information about the line or lumen, what type, etc.
To me that feels like an object, so more a device than a medicationadministration.
Mirjam Baltus (Dec 20 2018 at 12:56):
But perhaps a DeviceUseStatement fits better.
Jose Costa Teixeira (Dec 20 2018 at 13:09):
Actually if you want to capture the act of administering medications and all associated stuff (device) you can use medicationadministration, which points to the device.
And your device can be a compound device, with the different components
John Silva (Dec 20 2018 at 13:10):
I've had related questions to this, related to how does one (person/system) even know that an order (MedRequest) or administration (MedAdmin) is an infusion vs a 'regular medication'. The clinical folks I talk to know (instinctively) the difference, but technically what or how does a system or a representation like FHIR convey that difference. One of the suggestions I got was to look at the units for the dose; if it is a rate, then it 'must be an infusion'. I'm not sure if that's clinically correct but it's an obtuse way of determining this. In previous products I've worked on (with HL7 V2) we used the frequency (repeat pattern) to determine (differentiate) an infusion from a 'regular med'; if it was 'C' (Continuous) or 'C' (and PRN) it was considered an infusion. Unfortunately, FHIR doesn't seem to have a concept for "Continuous" in the timing.code property (value set: TimingAbbreviation ). [yes, I can already hear the chorus saying 'use an extension' ;-) -- but why should an extension be needed for something that is so common -- or at least it worked in the V2 world for a long time?] I also don't feel that is should depend on a Device; there are 'old fashioned' gravity drip infusions that do not use any device (per-se) so I don't think we can depend on a device (like infusion pump) to differentiate infusion from medication. Then to your point about multiple parts of the infusion; in the V2 world this would be handled by having the RXC-B (base) segments for the non-active solution (fluid) ingredients and the RXC-A (additive) for the active ingredients added to the base solution. I'm not really sure how that fits into the FHIR Medication model? (Another method that you could possibly use is to use the dosage.method to determine infusion vs med; this seems to be analogous to the RXR-4 - Administration Method, Table 0165 in V2.)
Lloyd McKenzie (Dec 20 2018 at 16:26):
If you won't ever have an infusion that doesn't have a rate and having a rate means it's an infusion, then a flag would be redundant - and we do our best to avoid redundancy where we can. Something being in v2 doesn't mean it'll be represented the same way in FHIR.
John Silva (Dec 20 2018 at 17:29):
@Lloyd McKenzie -- I can imagine cases where an 'unencoded order' (in V2-speak) doesn't yet have the rate information determined because that is done either by the pharmacist or at the delivery time by the nurse, yet we still know that it's an infusion order. How would this be conveyed in FHIR MedReq resource? (certainly don't want to be doing text parsing to determine this)
Lloyd McKenzie (Dec 20 2018 at 19:47):
At that point, it might not be computably determinable. Is there a use-case for it to be?
John Silva (Dec 20 2018 at 19:59):
@Lloyd McKenzie - I would expect the clinicians in the group (the Pharmacy group or other clinicians) would know better than me but at a conceptual level a specific order is known to be an infusion, regardless of how it is encoded in FHIR (or V2 for that matter). As I mentioned before, the clinicians know (instinctively) what they define as an infusion, it seems like there needs to be a way to represent this that isn't a 'side effect' of some other property that may or may not be set (i.e. rate). There are reasons why this distinction is important, again, I defer to the clinicians among us, but for one important thing is for charting of administrations. WIth a 'regular med' the expectation is that the charting of admins coincides with the specific times specified by the order, e.g. Q2H, BID, etc. and any deviation from that ordered schedule (frequency) needs to be documented with an exception (the notGiven concept in FHIR and similar in V2). However, for infusions, the expectation is that they are 'continually running' either until the order specifies a specific stop time or until the complete infusion is given (the bottle/bag is empty). The charting of infusions then is very different because of this. (and there are also the more complicated infusions like titrated doses, etc.) Products I've worked on DO differentiate infusions from medications for some of these very same reasons, so yes, there are use cases for this distinction.
Lloyd McKenzie (Dec 20 2018 at 20:05):
The route would generally convey something like "IV infusion"
John Silva (Dec 20 2018 at 20:14):
@Lloyd McKenzie Yes, but there are quite a few routes that are used for "infusions" but they don't necessarily mean it's an 'infusion' (from a clinicians point of view). At least in the V2 world (which probably has equivalent codes/value sets in FHIR) there are things like IV Push, IV Drip, etc. that might or might not be considered infusions (to clinicians -- to technical folks it would seem like anything given with a needle over time must be an infusion but to clinical folks there is a distinction.) I thought of the same idea with Admin Method, that too has the same problems.
BTW, the FHIR SNOMED CT Route Codes ( http://hl7.org/fhir/valueset-route-codes.html ) has a large number of routes that may or may not indicate that a Med Order is an infusion. @Jose Costa Teixeira - can you give us some insight into the distinction between infusions and 'regular meds'?
Found this link that tries to explain what "infusions" are: https://www.healthline.com/health/intravenous-medication-administration
"Infusion therapy involves the administration of medication through a needle or catheter."
So,just because the route is 'IV" doesn't mean that's the only way an infusion is delivered; it can also be delivered via catheters (lines), which would be a different 'route' from a V2 or FHIR perspective.
Again, this is why the association of 'Continuous' frequency with an 'infusion" seems to be a much simpler way to capture the concept. (However, outliers like 'IV Push" beg the question as to if they are infusions or meds, given over a very quick time.)
Lloyd McKenzie (Dec 20 2018 at 20:24):
An IV drip would still be a "continuous" administration, so I don't see that as a distinction.
John Silva (Dec 20 2018 at 22:33):
Yes, an IV Drip is 'continuous' but IV Push isn't (I believe). The point is that there are more than a few route 'codes' that would be needed to determine (or infer really) that something may be an infusion. Knowing that the frequency is 'C' (Continuous) is a simpler and more direct way to know this; no 'guessing' or inferring from side-effects of other properties.
Jose Costa Teixeira (Dec 21 2018 at 09:56):
can you give us some insight into the distinction between infusions and 'regular meds'?
I often err on the side of not trying to find clear boundaries like these. I think that some things are clearly infusions, other things are borderline. And you also write 'regular meds' so that is a sign... I just saw some different definitions and even the word 'infusion' is sometimes challenged..
Jose Costa Teixeira (Dec 21 2018 at 10:00):
Single point in time, continuous, bolus,... those are scenarios which need to be explored and supported (see IHE MMA volume 1), but i don't see a reason why the resource should indicate explicitly "this is an infusion". Even if we did, it is a classification that I would argue is not global.
Jose Costa Teixeira (Dec 21 2018 at 10:03):
personally, i care more that our standard supports the various cases as the Pharmacy WG has done in the examples. If there is a need to explicitly state the type (other than medadministration.dosage.method), i would bring @Melva Peters @John Hatem @Jean Duteau
Jose Costa Teixeira (Dec 21 2018 at 10:05):
so, @John Silva for encoding that explicit statement "infusion vs other meds", i would use dosage.method (it is extensible so you can use that)
Mirjam Baltus (Dec 21 2018 at 10:06):
I would not need to map that something is an infusion, but rather: this patient has an infusion, with these lines/lumen and administering system. Linking there to the medication, but that's not too relevant for my mapping.
Jose Costa Teixeira (Dec 21 2018 at 10:06):
yes, @Mirjam Baltus i was mostly addressing the question by John Silva
Jose Costa Teixeira (Dec 21 2018 at 10:07):
For your question about administering systemm I would use medicationadministration.device, if it is about a medication
Mirjam Baltus (Dec 21 2018 at 10:09):
@Jose Costa Teixeira: thanks, that is a good one. I was mainly trying to find the best starting point, with a resource type that points to all of that.
Jose Costa Teixeira (Dec 21 2018 at 10:10):
which brings an interesting side issue: from medicationAdministration you can only point to a single device. And device does not support "children" devices. so it may be worth discussing how you can point to several devices like you want
Jose Costa Teixeira (Dec 21 2018 at 10:12):
if it is administration you are talking about, use medadministration.device
If, in its definition, the medication is associated with a device, there should be other options but I can't figure them out yet
Mirjam Baltus (Dec 21 2018 at 10:17):
Yes, I think that will not work. I need to be able to have at least two medadmins (lock fluid and actual med), and the administering system can consist of multiple devices as well.
Andrew Gledhill (Dec 21 2018 at 12:13):
Are you discussing medication administration 'rate' for iv medicines ?
fixed rate continuous infusion
slow bolus over 2 minutes
variable rate continuous infusion
iv push over 10 seconds
This is an interesting area which imo has poor clinical definition at present.
John Silva (Dec 21 2018 at 12:17):
Thanks @Jose Costa Teixeira -- yes, surely the definition (if there is one) of an infusion is probably not consistent across the healthcare continuum so I guess it makes sense not to prescribe a way to represent it in FHIR (nor was it in V2). However, not having the notion of a 'Continuous' frequency makes it harder for systems (or users) that want to use that as part of the way to make that determination. Using dose rate information as a surrogate for this seems a little bit of a stretch; again, as I point to the use case of 'un-encoded orders' where the rate is not yet set, but it is known that the order is an 'infusion order'. [BTW, in products I've worked on we use the notion of infusion vs medication to separate out the nurse's workflow, a very important aspect of medication delivery so the use case is definitely there for us; whether or not this is a 'global use case' I can't say but our products have been used world-wide and this has been one of the aspects of our products which has been very important to clinicians acceptance of it.]
John Hatem (Dec 22 2018 at 19:07):
A few thoughts: medications orders or administered medications may have an "intravenous" route of administration, and separately may require a "method" to be specified as in the case of IV Push, or Bolus. Infusions ( the word) can be continuous, or bolus, or iv push, but in my clinical experience I didn't really "order" or "administer" or for that matter "document" using the word "infusions". I am not saying you can't use the word but for this discussion - I stay away from the word, because we need more specificity to clarify what folks are talking about. Now as to the current Pharmacy resources, we do support the ability to order or administer medications that may have an intravenous route of administration, or if needed, we also have the ability to specify the "method". Often the method is not needed. As I read this thread, my questions are, is someone asking can you make it clear that you have a continuous rate IV solution running on a patient? is it possible to represent IV Push and Bolus? My answer would be yes. I will check back on the thread, but not for several days, and if needed can point you to the pharmacy examples where these use cases are represented. I hope this is helpful.
John Silva (Dec 22 2018 at 20:11):
@John Hatem - thanks for your input; it's good to hear from someone with clinical (and FHIR/HL7) experience. Yes, as a technical (non-clinical) person when I hear the word "infusion" I'm not really sure what the clinician is talking about and if they can necessarily state their definition of what it means to them. As I mentioned, one of the 'classifiers' that they use is the frequency (not the rate) as an indicator of 'infusion' (or not). The administration method can come into play as well, but the main differentiator seems to be the frequency of 'continuous'. I've had the clinicians tell me that an IV Push and IV Bolus are not infusions but medications; I believe because it is given over a relatively short period of time, i.e. not continuously. One of the 'strange corner cases' that I really don't understand is the classification of Colloid orders as either infusions or medications, I think base on dose limits. As a technical person I would have thought anything that uses an IV route would be an infusion, but that doesn't seem to be the case. (and that doesn't take into account meds that are given via cathers like PICC or other lines.)
Thanks again!
Melva Peters (Dec 31 2018 at 18:40):
The previous discussion seems to include 2 different use cases (at least my read on it). Our current MedicationAdministration resource includes a method attribute and a route attribute as part of the dosage as well as a reference to the device used to administer the medication.
The method attribute definition is "A coded value indicating the method by which the medication is intended to be or was introduced into or on the body. This attribute will most often NOT be populated. It is most commonly used for injections. For example, Slow Push, Deep IV." It is currently is bound to an exemplar Snomed CT value set which may need to be changed given this discussion. In any event, the value set is exemplar so implementers can bind to a value set that is appropriate for them.
The route attribute allows you to specify the route or path of administration - for example Intravenous. Again, the binding exemplar to SNOMED CT.
We include a reference to the device that was used for the administration. @Mirjam Baltus Have you look a this to see if it meets your needs to include information about the lumens and lines, etc?
As always, if you believe there are changes needed to any Pharmacy resource, we encourage you to add a tracker item to Gforge with your use case (and ideally examples) and the Pharmacy Work Group will consider.
Gayanjith Premalal (Jan 31 2019 at 12:48):
Can we think about business requirement or stakeholder requirements which would require us to have a clear distinction between the infusions and the rest of the medication requests/administrations? At least for our EHR we had solution requirements to differentiate between the infusions and the rest of the medications. for an example, the information that is required to record an infusion administration event are different from the information that is required to record when administering a tablet. And also the presentation of the administration occasions.
We have based the distinction on the route of administration (which I think is problematic after reading this conversation) so I'm interested in finding out the best way to make this distinction.
John Silva (Jan 31 2019 at 13:34):
Yes, our clinical experts need to differentiate "infusions" from "medications" from both an ordering and administration perspective. Part of the reason for this is that there seems to be different kinds of clinical (inpatient environment, by nurses) workflows for administrations of "infusions" vs "medications". For example, nurses (I've been told) want to see what are the scheduled meds for their shift so that they know when to administer them, whereas infusions (not IV Push or Bolus - which are treated as meds because they are more 'one time/short duration') are handed differently. When you think about it, the infusions are "running continuously" (which is how the clinical experts tell me they differentiate these two concepts) and there are different things they do with infusions, obviously like charting volume per unit time (e.g. 100ml in last hour), if the infusion has started/stopped/paused or rate changed. Hey, come to think about it, this infusion concept is spelled out very well in the IHE Technical Framework Supplement Infusion Pump Event Communication (IPEC).
https://wiki.ihe.net/index.php/Infusion_Pump_Event_Communication
Christof Gessner (Jan 31 2019 at 22:02):
I cross-posted this topic to the HL7 pharmacy WG. https://confluence.hl7.org/display/PHAR/Pharmacy Suggestion is to raise a FHIR tracker item on the issue.
Last updated: Apr 12 2022 at 19:14 UTC
