How to record an inpatient drug infusion?

[Christian_Reich]

Friends: So, after all that conversation I think we can crystallize the points this debate frazzled into:

1. There is no “canonical CDM”. There is only a CDM. I am with @Patrick_Ryan here. If you feel you need to deviate from the scripture please bring it on (like in this debate) and we will find a solution. Only data that is truly local and will never be studied across the network should be put into some kind of an add-on.
2. We do have the need for capturing and studying dosing information that does not anticipate the administration of an entire unit or fixed amount of ingredient, like for example @docsteveharris’ infusion rate. This debate is also happening in another Forum, and I am thinking we could solve it by introducing a new field giving us the required unit (µg/kg/min in this case). Please continue there.
3. The domain assignment of concepts. Again, like @Patrick_Ryan said: We can debate this here and make it better, but we should ultimately have the vocabulary CONCEPT table be the common ground. “Feeling” in which Domain a concept should live is difficult to reproduce. @bailey’s example is a typical case where it is hard to decide, Polyuria and Anuria was another one. We even managed to convince SNOMED to split up the concept into the Measurement and Condition, because having it ambiguously in one Domain screwed up their own hierarchy.

So: Please keep bringing it on, and we will make the CDM better.

[bailey (Charles Bailey)]

Hi Charles. Were you concerned that using quantity to record the total dose would make your version of the CDM further from the original where it has a different meaning? Was there a view point that it would be better to leave this empty in this context and use both the dose_unit_concept_id and an additional “dose_quantity” field? This feels to us like the way to proceed but not sure if we’d lose out in some other way. Thanks.

In the short term, we thought it better to put it into the canonical fields, since it’s not generally possible to study dose otherwise. Still discussing possible long-term solutions as @Christian_Reich mentioned; we’ll get there.

[Andrew (Andrew Williams)]

@Christian_Reich : Others should correct me if this is off base, but I think that it is quite common for infusions to result in unused portions of vials. The amounts are calculated by body weight and the vial units are large enough that the discarded proportion in the last vial is significant. The routine discard of unused portions of vials is common enough to have led to studies and calls for policy changes that will prevent waste: https://www.bmj.com/content/352/bmj.i788
So error in a dose response analysis would come from an attribution based on the assumption of complete infusion.

[zhuk (Oleg Zhuk)]

Well, that’s an old discussion and it has been quiet for three years, but I want to stir it up a little bit now.

As a critical care doctor myself I understand @docsteveharris 's pain with continuous infusion of inotropes and pressors and its representation in OMOP CDM. From the clinical point of view, it doesn’t really matter for me if my patient is on 0,11 mcg/kg/hr epinephrine infusion or 0,13 mcg/kg/hr, as long as I know that the rate is not more than, let’s say, 0,15 mcg/kg/hr epinephrine. The same is for other inotropes, pressors, and prostaglandins with a specific dose range for each drug. And, genuinely speaking, there aren’t many drugs that require the calculation of infusion rates.

I want to propose a simple solution for @docsteveharris 's case: introduction the concepts of low-dose, medium-dose, and high-dose inotropic and vasopressor support, defined somehow like:

• low-dose support: epi 0,01 - 0,05 mcg/kg/min
• medium dose support: epi 0,05 - 0,15 mcg/kg/min
• high-dose support: epi > 0,15 mcg/kg/min

We can make the gradation more specific if needed (each 0,05 mcg/kg/min eg.)

According to this solution, during the ETL process, all you need is to calculate how much support your patient got on that day and write the corresponding concept_id into the Observation table. I think this solution pretty much fits into the current CDM and allows to solve the problem (at least temporarily) with help of standardized vocabularies.

Sorry again for touching old wounds Please correct me if I am missing something.

[clermontg (gilles clermont)]

Hello,

has there been any progress on this? Dr. Harris, how did you end up doing it? Asking the question now as the CHoRUS consortium is trying to line up their approach to the representing of drug infusions, a major domain of critical care delivery.