CKM editorial policies

I’m looking for a place to read up on ckm editorial policies/decisions.

From reviewing medication summary I noticed all summary archetypes (alcohol use, smoking etc) are evaluations, while they mostly/solely describe observed data points (e.g. amount taken of a medication) so I wondered why they are not observations?

I found:

It states: “The Observation class is for recording information that varies over a short period of time.“
From that statement it makes sense to model a summary of lifetime medication use as EVALUATION.

But from the spec I would say summaries like medication usage are observations: EHR Information Model

I’m not looking to change current modelling practices as much as understanding the history behind the current practice.

Hi Joost,

I suspect that you’re not so much seeking CKM policies but guidance on how to design an archetype, starting with the correct CLASS?
Editors use the EHR Information Model specification document that you have already found as the source of truth.

CLUSTERs are reusable fragment models. ENTRYs are standalone models:

  • EVALUATION archetypes are the simplest model and the one we usually consider first by default. In practice, the archetypes that are developed as EVALUATIONs are summaries or opinions about the data recorded using OBSERVATIONs. These are commonly used to capture data that evolves over time, as an evolving summary eg a medication, tobacco smoking or alcohol consumption.
  • OBSERVATIONS record data, the state of the patient and event timing. These are usually used to record measurements, history and examination findings, scores and scales. This aligns with a use case for a medication statement, a tobacco smoking diary or alcohol consumption diary - what we are taking, smoking or consuming on a specific day or over a specified period of time.
  • INSTRUCTIONs and ACTIONs are usually self-explanatory.

There are a number of instances where it can be argued that an archetype could be either an EVAL or an OBS. In general, if the data will never need to record the state of the patient to assist interpretation, or specified event timing (point in time or interval), then it is modelled as an EVALUATION.

This page on the wiki provides further information -

There are a couple of Youtube videos that might also be helpful (via Region Östergötland):

1 Like