"Medication dosage based on..."

I’ve been presented with a medication dosing use case that I haven’t seen before, and that I’m not certain how to model. The requirements are as follows:

  • A structured dosage and timing is recorded, for example “20 mg three times a day”
  • This dosage is based on a mental calculation by the prescriber, based on a parameter of the patient, for example weight, age or body surface.
  • We need to be able to record, in a coded element, which parameter the dose is based on, and a free text comment about it.

I feel like this is a kind of very simple “dose formula”, where you don’t record an actual formula but just the main parameter of an approximate mental calculation. As such it’s related to the “Dose formula” element of the “Dosage” archetype (https://ckm.openehr.org/ckm/archetypes/1013.1.2751), but not quite the same thing.

Comments? Tips? :blush:

1 Like

The future way of doing this will probably be something like in this TP/GDL CHOPS chemo example. You can see a formula for prednisolone dose, which is trivial:

   prednisolone_dose: Quantity
        Result <- prednisolone_dose_per_m2 * Subject.bsa_m2

And for doxorubicin, which is more complex:

doxorubicin_dose: Quantity
        Result <- doxorubicin_dose_per_m2 * Subject.bsa_m2 *
            map Subject.bilirubin.range
                |high|:        0.5
                |very_high|:   0.25
                |crit_high|:   0.0

The above looks a bit cryptic, but essentially it’s just applying the common dose multiplier based on bilirubin level. We are close to having the language that can expressed all this formalised. Formulas like this could in theory be embedded in archetypes (because the Expression Language will be usable there as well), but ideally, we would not create multiple artefacts that represent this same clinical knowledge. I think we should consider these formulae (e.g. all the dose ones belonging to CHOPS) as being a knowledge artefact in its own right that needs to be governed like an archetype.

We already have GDL2 representations of these kinds of things; this next generation will be more semantically powerful and adaptable to more back-ends, EHR environments etc.

This probably doesn’t solve the need in an immediate concrete sense, but we are not too far away from the above approach becoming reality.


I think your example is definitely the future for proper calculation of dose based on more or less stable parameters of the patient. This use case is quite a lot simpler though. :grinning:


I’ve just spotted Dosage justification (at174) in the Medication INSTRUCTION. It is designed to carry the reasoning on why the dose is what it is.
The example is a bit weird - ‘Adjusted for amputation’, which in fact means that it is using the adjusted weight for an amputee to calculate the dose. In this use case it can be more generic ie weight, age or BSA.