Missing ism_transition of state for "proposed" and "completed and continued"


I am missing a state here, like "Proposed"(Use case: As a nurse on training, I like to propose a procedyre to be performed). I am also missing a state like "Completed and Continued".(Use case: Procedyre can change timing, from 2 x daily > 3 x daily)


software plattform


Thomas Beale
January 5, 2018, 9:55 PM

Planned, as an Instruction state is a bit more narrow - it just means that this intervention (i.e. order) has been created and is planned to really occur. Whether it is mentioned in a care plan or any kind of other note etc it another matter.

I would normally understand the word 'proposed' to mean: suggest to do X, under Y circumstances. But don't actually do start it. So in fact, don't yet create any order or order-related process.

Remember, the state machine just represents states of an Instruction (i.e. order for intervention, including medication) unfolding through time, not the states of the patient's overall health state, or even of a care plan - even that is a higher level entity, that could easily have many Instructions (medications, physical therapy, education etc), each with their own state (represented by individual state machines).

Thomas Beale
January 5, 2018, 10:04 PM

Action(s) represent what has been done, yes. Instruction represents the formal order, i.e. clinical statement for a specific thing to be done, that can be acted upon by appropriate trained professionals - no more decisions needed by the consultant or primary physician.

So it's not a statement like: 'consider left hip replacement in next 12 months; review based on mobility and pain', which might appear in a care plan, which doesn't cause anything to happen, without the physician having another look later on. But you might create an instruction to monitor / review something, e.g. mobility, cancer (for a patient that has had surgery etc and is provisionally clear).

Again, actionable plans are not really modelled properly by Instruction - you need the Task Planning approach for that. The Instruction is just a clinical 'formula' like 'Amoxyycillin 3td po x 7d' or 'turn patient 2x daily, once after breakfast, once after dinner'. To really see the difference, consider the prescriptions for a typical chemo regime like R-CHOP, versus the detailed administration plan, which contains all the directions like calculating dosages based on body surface area, how to admin each drug, what time, what day etc.

Phuong Pedersen
January 6, 2018, 1:26 PM

I have uploaded some pictures of how we currently are using the instruction/activity model. The procedure.action and the request.instruction are used as underlying datastructure. As you can see we are using the state machine actively.

The pictures show an actionable plan, that helps the nurses to see, what to do on a patient. When we add the timing clusters to the activity, we can even show this in an timeline, or sort the list of activities so that, activities that should be performed in the nearby timeframe comes first. All this we can achieve easily by pure aql queries. I see the possibilities in task planning, but for now the instruction/action model cover our needs. The design and the ideas behind has in a way been proved, but there are also room for improvement. Discussing this with nurses internally, also gives us the impression, that the modell is not wrong. When you say that it is not really modelled properly what do you mean? Do you know of anybody else using instruction/action in they implementation and using the possibilities the state machine gives for decision support? Or implementation of the task planning in a real life scenario?

Ian McNicoll
January 8, 2018, 10:39 AM

Hi Phuong,

We met simialr requirements in the Medicaiton managment ACTION archetype http://openehr.org/ckm/#showArchetype_1013.1.123
using the existing states

A. Medication recommended
Careflow step

The medication has been recommended but no steps have been taken to initiate prescribing.
Current state: planned

B. Minor change to order
Careflow step

The medication order has been changed in a manner which does not require a new instruction/order to be issued, according to local clinical rules.
Current state: active

These seem to closely equate to your examples for Procedure.

There may be some linguistic confusion re the use of 'Procedure planned'. I would have taken this to mean ' we are proposing to do this procedure' rather than 'we have formally planned this procedure'. English is used loosely for both.

Perhaps we need 2 new careflow_steps in Procedure.

1. 'Procedure proposed' (planned). This could be a simple flag in a GP system that the GP intends to send a referral, for a junior member of staff to make a proposal without the authority of an actionable order, or for a similar situation where a senior member of staff is recommending a procedure to a colleague but does not have the authority to make this a formal order. (This was the original purpose of 'Medication Recommended' in Medication ACTION.

2. 'Minor order change' (active) - as per the medication use-case. the definition of 'a minor change' is going to be somewhat local and contextual but this does allow us to imply continuity of the same order where appropriate.
3. 'Major order change' (aborted) - implies some sort of continuity of the original order but where the change is such that it is only safe to re-issue a new order.

I don't think we need a new 'state', just some new care_flowteps that we can propose through CKM. These would be non-breaking changes and in line with other archetypes so should not be too contentious.

Phuong Pedersen
January 11, 2018, 8:15 AM

ok, thanks, then we can continue the discussion on slack , and this issue can be closed.


Phuong Pedersen




Affects versions