'Actionability' of orders

Gerard
This area is interesting. At a recent meeting of Standards Australia it was determined that an extract should say if it is for information only or there is, within it, the expectation that some action is required. This should, we felt be boolean ie unambiguous. It took some time to come to this - but an example is an extract containing a medication order that is sent to the pharmacy (action required) and copied to the specialist (no action required).

The openEHR instruction class that is almost fully specified now and will be in version 1, will take this sort of thing to a new level.

We need more debate on this, Sam

Gerard Freriks wrote:

In een bericht met de datum 6-9-2005 22:32:06 West-Europa (zomertijd), schrijft sam.heard@oceaninformatics.biz:

Hi All,

Interesting concepts discussed here: An EHR for a clinician deals with two major classes of information:

  1. Classes that describe the situation of the patient (different levels of detail, changes over time, enourmous sets sometimes)
  2. Classes that describe the behaviour of care professionals ( a single medication description, giving information, asking consent, performing a procedure,a gain many).

Often # 2 are triggered by #1.
And #1 and # 2 can have different time formats (past, present, future).

However, a proper EHR should allow to include ‘future’ things that could or should happen. E.g. in care planning it is important to have a start value of the situation of the patient, compare this with historic description of this situation when available, state a goal of what to expect in an overseable time, and then check for the actual outcome e.g. after care has been carried out.

If an EHR can only store what has already happened, so the past, it becomes useless for supporting clinical practice.

It is interesting that the concepts of trigger is fundamental to HL7 dynamics, and the ‘mood’ of the action / obs (so #1 and # 2) are already available in the RIM classes: INT, RQO, GOL, EVN are very powerful to support the dynamics of monitoring and workflow. Key components of EHR.
See also the ISO TS on the EHR in which both data / structures and dynamics are the first 2 chapters.

Sams distinction between action required versus for your information only is interesting and probably would need further development on message and on EHR level.

And again: from the perspective of clinical documentation and care planning the disctinction between message and EHR is arbitrary: we have systems that are integrated or that are coupled, that have similar functionalities. Key is that we analyse #1 and # 2 appropriately, model it for ICT and only then develop the tech stuff in which it works.

My 3 Euro cents,

William

Hi,

In my parlance:
Messages are to update databases and are facts. eg lab results
Documents are to update humans with professional opinions that are signed.

The EHR and its extract are documents.
Depending on the type of archetype that what is recorded is:
-anamnesis (history)
-observation

  • plan
  • order
    Each of these four types form a family of archetypes and indicate context.

Medication information inside an anamnesis archetype means that this is reported by the patient or others (eg the pharmacist when he sends a copy)
Medication information inside an observation archetype means that this is observed by the healthcare provider
Medication information inside a plan archetype means that this is an intention by the healthcare provider
Medication information inside an order archetype means that this is a prescription on behalf of the patient for others (pharmacist) to dispense/provide

What we miss in this list is the context where the extract reports that something has been executed.
This means that we have to provide for an archetype type that indicates that a plan is executed, or an order is executed.
We have to add the execution type of archetype.
If this is the case we need attributes that allow us to indicate a cancellation, of anamnesis, an observation, a plan, an order and execution.

This makes the complete list:
-anamnesis (history)
-observation

  • plan
  • order
  • execution of a procedure.

And next we need the machinery to indicate the cancellation of these.

I’m just providing some thoughts.
In the AKF project we need to address all this.

Part 1,2,3,4,5 of EN13606 are not affected.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Ok,

Do we also need to “grade” the degree of execution for “orders”?

If I want the patient to take a short course of antibiotics:
I could: a. write a manual prescription
b. make a note in EHR to that effect
OR
use my computer system to: print the prescription
record details
As a general practitioner, my remit is normally served by the further actions of:
signing the prescription
giving it to the patient.

We could also consider recording, somewhere along the line:
when was the prescription given to the pharmacist
when it was dispensed, ready for collection
who the dispensed product was given to (patient or “agent”)
and
when the patient started treatment
when they stopped, (resumed, started again…)
(why they stopped, resumed…)
how many tablets were left after the patient stopped taking the tablets
and maybe even
disposal of those tablets.

I guess most of those are implemented in relation to narcotics, in most countries.

I would humbly submit that Completion of the order is not a simple Boolean. For each of the above, a truly detailed record would retain: date, time, identity of the agent completing that step, identity of the agent recording the step, and possibly a text field to record any further ambiguity not originally compassed in the definition, such as the patient vomiting at some undefined interval after oral administration.

Jan Ravet
Augusta
West Australia

In een bericht met de datum 7-9-2005 0:13:04 West-Europa (zomertijd), schrijft gfrer@luna.nl:

  • execution of a procedure.

And next we need the machinery to indicate the cancellation of these.

Yes, this is exactly as the two attributes in HL7 v3 RIM Acts:

moodcode to indicate the plan versus execution (INT versus EVN)
statuscode to determine whether this Act is active versus cancelled or on hold or similar.

William

William,

The difference is that in CEN it pertains to an archetype
And in HL7 to a much more molecular/atomic construct.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Gérard,

May I propose a contribution on the the question you pose ?

1 - Concerning the cooperation between actors and roles
The two reasons you mention make me think of the Searle’s speech acts, amongst other developped afterwards by Vinograd in his “conversation for action”.
i) - reason one : “illocutionnary” act which can be :
– a request for an action performing (order, instruction, letter where one can find a request, …) expressed by a requester to an offerer and which implies an answer from the offerer : accept (and then the offerer commit to do what is requested), amend, refuse.
– a commitment for an action performing expressed by the offerer. In practice and depending on the cooperation rules, the commitment is sanctionned by an appointment. The appointment can be in the letter …
– a report on the the action performing expressed by the offerer to the requester (professionnal opinion or advice following a request, lab result, …)
– a validation of the report (or of the action performing ?) by the requester

ii) - reason two : “propositional” act which gives the content of the speech act :
– professional content for a request, with reason, target, condition of the patients influencing the performing of the requested action
– professional content for the commitment
– for the report. it is interesting to distinguish there the differences of the professional content between a diagnostic and a terapeuthic action.

The conversation for action is one of the components of the cooperation. Another one is the coodination of the care plan performing, when for instance an actor/role notifies the following one that the action he was responsible for is performed (or is not performed). May-be other component exists ?

Reports and notifications are different : report is sent to the requester within the conversation for action, and notification is sent to the following actors/roles within the workflow coordination, may-be with an another professional content. The information system is intended to support this cooperation messaging.

2 - Concerning the fact that the EHR is the support of cooperation messaging.
In the schema you describe, the EHR is the obliged “documentation device” (or I could say messaging device) between HCPs to communicate within the “conversation for action” or for the coodination. One can imagine that there are 3 dossier elements related to the same action performing

  • in the performer dossier
  • in the cooperation system
  • in the requester dossier

The utilisation of these “versions” inof the same act the care process should be studied.

HI

H. Mensch
GMSIH

In een bericht met de datum 7-9-2005 13:32:50 West-Europa (zomertijd), schrijft hubert.mensch@gmsih.fr:

Dear Hubert,

I am very pleased to read that others relate the activities of professionals to theories as Searle, Vinograd and probably others.

I agree with the analysis below for the need to distinguish between the illocutionary and propositional aspects of speech acts.
I also believe that it has been studied within the HL7 standard, in which for Acts (expanding it to potential other acts, however, if we look carefully all HL7 acts are speech acts because they are communicated between HC professionals) the illocutionary aspects are modelled carefully as request, promise and event. Validation is an interesting addition, which would require more attention. Currently the Care Provision model is becoming more explicit on care planning, for which your comments will be a good contribution. Again, the way of modeling care with the RIM is flexible and allows implementation in the EHR and in messages. Distinction between the two does not really exist from the point of speech acts, there is a messenger system, which can be the language, a letter, a system, or elektronic exchange between system. However, of course this only works if the semantics are tackled, so the propositional part of speech acts needs to be tackles, which has been proven to work with vocubulary etc. Withing the IT solutions the idea of the archetypes is perfectly capable of making explicit what needs to be contended. Yes, difference between diagnostic and therapeutic is important for me too: first describes the patient (thruh the eyes of the professional) and second describes the professional (from prof. point of view).

I really would ask the OpenEHR community to adapt the HL7 RIM models for describing the propositional content and the UML activity diagrams and sequence diagrams for the illocutionary aspects of Acts.
This is an established model that does tackle many of the issues raised. If at implementation level particular attributes or mechanics of these models do not work, they can be adjusted or constrained. The advantage is obvious, we can work out common illocutionary expressions (placing orders with their dynamics does not differ for a battery of lab tests in an laborder , or a set of vital sign observations ordered by a physician to the nurse, or the request of a nurse to a physician to check a patient who’s condition deteriorated. The principles of request, promise, event and validation hold for these examples.

The content can be archetyped / templated to assure broad understanding by clinicians regardless the technical solution to support them in speech acts.

William Goossen

In een bericht met de datum 7-9-2005 8:27:14 West-Europa (zomertijd), schrijft gfrer@luna.nl:

William,

The difference is that in CEN it pertains to an archetype
And in HL7 to a much more molecular/atomic construct.

Gerard

Yes, but in HL7 the same ideas work for the CMETS / archetypes / Templates in such a sense that we talk about groupings of observations or groupings of acts (batteries for example).

William

Dear all,

Some personal thoughts about the scope of CEN/tc251 documents plus extracts and HL7v3 messages.
And the basic set of ‘ancestor archetypes’ that we need for the EHR.

The EHR is the place where the healthcare provider documents:

  • the data obtained like: letters received, orders received
  • the data obtained via anamnesis, personal observations, test results
  • the thoughts about it: the evaluation of the data obtained, the production of the professional opinion
  • the resulting action like a plan
  • the optional resulting action like an intervention
  • the optional resulting action like an order
  • the evaluation of the plan

Any archetype (e.g. medication: drug name, dossage, amount, usage, etc) can occur in all 7 bullit points.
In En13606 each of these 7 points provide the context for ‘medication’.
It is not an attribute in the part 1 Document model, or part the Archetype model that will provide the context information.
In EHRcom it has to be a special ‘ancestor’ archetype that provides the intended context.
This is they way in which EHRcom provides the same function as the Mood code in HL7v3.

Again EN13606 is about the documentation of that what happened.
E.g. it documents that an order for medication is given.
The EHRextract might contain this documented fact.
The EN13606 will not provide mechanisms for the ‘transactional’ concepts that one needs to transmit the medication order (prescription). The EN13606 is not a substitute for HL7 v3 messages that are used to transmit a medication order.
The CEN/tc251 HISA norm might provide the definition of this type of function. The state machine involved will have to be aligned with HL7 are the plans as far as I can see.
But this is NOT part of the EN13606 EHRcom norm.

The preceding paragraph indicates the distinction between HL7 v3 messages and EHR documents and EHR extracts.
The scope of the EN13606 is substantially different from that of HL7.
Not withstanding this difference, both HL7 v3 messages and EHR documents or extracts can/should contain the same clinical and non-clinical archetypes/templates.

In to few places this essential distinction between EHRcom and HL7v3 messages is described.
In my view CEN/tc251 and HL7v3 are complementary and not in competition.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Gerard

I have to correct you here - in an openEHR context anyway.

In my parlance:
Messages are to update databases and are facts. eg lab results
Documents are to update humans with professional opinions that are signed.

The EHR and its extract are documents.
Depending on the type of archetype that what is recorded is:
-anamnesis (history)
-observation

  • plan
  • order
    Each of these four types form a family of archetypes and indicate context.

It is important to plan for when information is connected and flows - and not get in a mess about all the possible disconnects. As medication order is an instruction, it is has a state machine - and it is possible to report who recorded this - and the source of the information if required. But it is always an instruction. It can be administered and Act statements (with or without an instruction) can be recorded based on the medication order archetype. But medication orders are always medication orders - even if you record them after the fact, these are entered as completed instructions.

So the following do not fit in open EHR model.

Medication information inside an anamnesis archetype means that this is reported by the patient or others (eg the pharmacist when he sends a copy)
Medication information inside an observation archetype means that this is observed by the healthcare provider
Medication information inside a plan archetype means that this is an intention by the healthcare provider
Medication information inside an order archetype means that this is a prescription on behalf of the patient for others (pharmacist) to dispense/provide

The basis for such an approach is that queries are reliable - a planned medication is a state of medication - as is completed - and are determined by the last Act statement associated with the instruction. This is managed by the workflow engine but can propagate in a distributed environment. So if you order a Pap test bi-annually, no matter where the person actually has the pap test, the Act statement can propogate back to the source of the order. (A bit futuristic but one day this will be the norm!).

What we miss in this list is the context where the extract reports that something has been executed.

This is the openEHR ACT class. We need to get you in touch with the INSTRUCTION class - perhaps Tom will put the latest paper around.

This means that we have to provide for an archetype type that indicates that a plan is executed, or an order is executed.
We have to add the execution type of archetype.
If this is the case we need attributes that allow us to indicate a cancellation, of anamnesis, an observation, a plan, an order and execution.

This makes the complete list:
-anamnesis (history)
-observation

  • plan
  • order
  • execution of a procedure.

And next we need the machinery to indicate the cancellation of these.

I’m just providing some thoughts.
In the AKF project we need to address all this.

Part 1,2,3,4,5 of EN13606 are not affected.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Hi All,

Interesting concepts discussed here: An EHR for a clinician deals with two major classes of information:

  1. Classes that describe the situation of the patient (different levels of detail, changes over time, enourmous sets sometimes)
  2. Classes that describe the behaviour of care professionals ( a single medication description, giving information, asking consent, performing a procedure,a gain many).

Often # 2 are triggered by #1.
And #1 and # 2 can have different time formats (past, present, future).

However, a proper EHR should allow to include ‘future’ things that could or should happen. E.g. in care planning it is important to have a start value of the situation of the patient, compare this with historic description of this situation when available, state a goal of what to expect in an overseable time, and then check for the actual outcome e.g. after care has been carried out.

If an EHR can only store what has already happened, so the past, it becomes useless for supporting clinical practice.

It is interesting that the concepts of trigger is fundamental to HL7 dynamics, and the ‘mood’ of the action / obs (so #1 and # 2) are already available in the RIM classes: INT, RQO, GOL, EVN are very powerful to support the dynamics of monitoring and workflow. Key components of EHR.
See also the ISO TS on the EHR in which both data / structures and dynamics are the first 2 chapters.

Sams distinction between action required versus for your information only is interesting and probably would need further development on message and on EHR level.

And again: from the perspective of clinical documentation and care planning the disctinction between message and EHR is arbitrary: we have systems that are integrated or that are coupled, that have similar functionalities. Key is that we analyse #1 and # 2 appropriately, model it for ICT and only then develop the tech stuff in which it works.

My 3 Euro cents,

William

  • If you have any questions about using this list, please send a message to d.lloyd@openehr.org

Jan

This is exactly as the Instruction class is designed - if you have a look on the Ocean website at the medication archetype - you will see that it has a set of pathway steps such as these.

I have asked Tom to provide the latest paper on Instruction to get this discussion moving further.

Sam

Jan Ravet wrote:

Ok,

Do we also need to “grade” the degree of execution for “orders”?

If I want the patient to take a short course of antibiotics:
I could: a. write a manual prescription
b. make a note in EHR to that effect
OR
use my computer system to: print the prescription
record details
As a general practitioner, my remit is normally served by the further actions of:
signing the prescription
giving it to the patient.

We could also consider recording, somewhere along the line:
when was the prescription given to the pharmacist
when it was dispensed, ready for collection
who the dispensed product was given to (patient or “agent”)
and
when the patient started treatment
when they stopped, (resumed, started again…)
(why they stopped, resumed…)
how many tablets were left after the patient stopped taking the tablets
and maybe even
disposal of those tablets.

I guess most of those are implemented in relation to narcotics, in most countries.

I would humbly submit that Completion of the order is not a simple Boolean. For each of the above, a truly detailed record would retain: date, time, identity of the agent completing that step, identity of the agent recording the step, and possibly a text field to record any further ambiguity not originally compassed in the definition, such as the patient vomiting at some undefined interval after oral administration.

Jan Ravet
Augusta
West Australia

From: Gerard Freriks
To: Williamtfgoossen@cs.com
Cc: openehr-clinical@openehr.org ; d.kalra@chime.ucl.ac.uk ; thomas.beale@oceaninformatics.biz
Sent: Wednesday, September 07, 2005 6:12 AM
Subject: Re: Antw: Re: ‘Actionability’ of orders

Hi,

In my parlance:
Messages are to update databases and are facts. eg lab results
Documents are to update humans with professional opinions that are signed.

The EHR and its extract are documents.
Depending on the type of archetype that what is recorded is:
-anamnesis (history)
-observation

  • plan
  • order
    Each of these four types form a family of archetypes and indicate context.

Medication information inside an anamnesis archetype means that this is reported by the patient or others (eg the pharmacist when he sends a copy)
Medication information inside an observation archetype means that this is observed by the healthcare provider
Medication information inside a plan archetype means that this is an intention by the healthcare provider
Medication information inside an order archetype means that this is a prescription on behalf of the patient for others (pharmacist) to dispense/provide

What we miss in this list is the context where the extract reports that something has been executed.
This means that we have to provide for an archetype type that indicates that a plan is executed, or an order is executed.
We have to add the execution type of archetype.
If this is the case we need attributes that allow us to indicate a cancellation, of anamnesis, an observation, a plan, an order and execution.

This makes the complete list:
-anamnesis (history)
-observation

  • plan
  • order
  • execution of a procedure.

And next we need the machinery to indicate the cancellation of these.

I’m just providing some thoughts.
In the AKF project we need to address all this.

Part 1,2,3,4,5 of EN13606 are not affected.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Hi All,

Interesting concepts discussed here: An EHR for a clinician deals with two major classes of information:

  1. Classes that describe the situation of the patient (different levels of detail, changes over time, enourmous sets sometimes)
  2. Classes that describe the behaviour of care professionals ( a single medication description, giving information, asking consent, performing a procedure,a gain many).

Often # 2 are triggered by #1.
And #1 and # 2 can have different time formats (past, present, future).

However, a proper EHR should allow to include ‘future’ things that could or should happen. E.g. in care planning it is important to have a start value of the situation of the patient, compare this with historic description of this situation when available, state a goal of what to expect in an overseable time, and then check for the actual outcome e.g. after care has been carried out.

If an EHR can only store what has already happened, so the past, it becomes useless for supporting clinical practice.

It is interesting that the concepts of trigger is fundamental to HL7 dynamics, and the ‘mood’ of the action / obs (so #1 and # 2) are already available in the RIM classes: INT, RQO, GOL, EVN are very powerful to support the dynamics of monitoring and workflow. Key components of EHR.
See also the ISO TS on the EHR in which both data / structures and dynamics are the first 2 chapters.

Sams distinction between action required versus for your information only is interesting and probably would need further development on message and on EHR level.

And again: from the perspective of clinical documentation and care planning the disctinction between message and EHR is arbitrary: we have systems that are integrated or that are coupled, that have similar functionalities. Key is that we analyse #1 and # 2 appropriately, model it for ICT and only then develop the tech stuff in which it works.

My 3 Euro cents,

William

  • If you have any questions about using this list, please send a message to d.lloyd@openehr.org

William and Hubert

Thank you for this discerning discussion. As a clinician I have been working to resolve the many issues around workflow in the shared EHR environment. Sistine Barreto and Eric Browne here in Australia have been studying this area. Thomas Beale has pulled this together for the openEHR:

“The speech-act theory on which HL7 is based is interesting, but it is questionable as to whether it is a good basis for modelling recorded data in an information system. The reason for scepticism is that once information structures (very discrete) are tangled with linguistics (very slippery), the semantics of the information is obscured with largely irrelevant linguistic details. The intention of the mood code is to replicate the speech act mood of interlocutors (ultimately humans, mediated via systems, such as a pathology result system and a hospital EMR) in a ‘conversation’ that supposedly takes place about an observation, an order, an appointment etc. In such conversations between humans, there may be nearly infinite shades of tone and voice, and hence intention and commitment. However information systems cannot easily deal with so many shades of meaning particularly to do with future “intention”. And most likely they don’t need to. Usually, the need is for something to be actioned or not. The problem with mood codes comes down to this: they are an attempt to mark every item of information with the mood of the speech act which corresponds to the item. Even assuming that there always are such speech acts (there may not be, e.g. communication between two software packages), this means that we are forced to mark each item with possibly a very subtly different marker from every other item. Consider the shades of commitment: “intent”, “promise”, “proposal”. Which one is it in any circumstance? In HL7, you have to decide at message design time. Originally there were only about 6 mood codes. Now there are 13. It may keep growing. None of the mood codes are of any use in the EHR, and I would argue, they are of little use in messages.”

The Instruction and Act classes in openEHR are the result of this work and have had a long gestation including some PhDs! We have also linked the pathway steps in an instruction (which may involve communication) with the fundamental machine states in a standard workflow engine. This design approach, which I am sure Tom will want to talk about, aims to ensure that the machine ,as well as the person, can discern what is going on (to the extent that it is required to support workflow or decision support).

It is also clear that for safety the state in disconnected systems may be set differently from connected systems for the same step. For example, in community health a medication order may be set to an active state when prescribed (as the dispensing and administration events will not be recorded), where as it is still ‘initial’ (prior to actioning) in a connected inpatient environment (where the first administration record may set it to active.

The sort of ‘promised’, ‘proposed’ and other moods can be set in the archetype if appropriate for that sort of activity.

Cheers, Sam

Without paying attention to EN13606 or OpenEHR I produced the text.

So when you have special classes fine.
Then these classes act as what I called ‘ancestor archetypes’.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Surely the issue here is not merely that of the (binary?) nature of a
Speech Act, but the significance of that act to the enterprise within
which it occurs.
Thus even a delivery 'for information only' might invoke some action on
the part of the receiver, even if it is only the obligation to record its
arrival, e.g. for audit purposes.
What is missing here are definitions of the protocols governing the
performance of these acts by the enterprise's actors. This 'deontic
ontology' would permit the parties suitably to label each transaction and
to relate those labels to the appropriate protocols, which would indicate
the appropriate actions and enable the identification of, and recovery
from, transmission and other errors.
Formal languages, such as LOTOS etc., for defining such protocols hav
ebeen around for years, particularly in the telecoms industry. I suggest
that you investigate the inclusion of a suitable metalanguage.
Bernie

Date: Wed, 07 Sep 2005 05:57:29 +0930
From: Sam Heard <sam.heard@oceaninformatics.biz>
Reply-To: openehr-clinical@openehr.org
To: Gerard Freriks <gfrer@luna.nl>
Cc: Dipak Kalra <d.kalra@chime.ucl.ac.uk>,

     Thomas Beale <thomas.beale@oceaninformatics.biz>,
     openEHR-Clinical <openEHR-Clinical@openehr.org>

Sam Heard wrote:

Gerard Freriks wrote:

Hi,

Reading an HL7 list.

To me it seems there is exchange of the same data for two reasons.
1- an order, an instruction, an advice, a professional opinion, a letter, a lab result, itself, etc
2- the documentation of , the provision of information about, the order, the instruction, the advise, the opinion,the letter, the lab result, etc.

Eg: the prescription on paper handed to the patient, the prescription electronically that is sent and serves as a kind of trigger event to make acts happen at the receiver side.
And the documentation of acts in a system about the real world.

What they have noticed is one of the fundamental criticisms on the RIM by the philosophical ontologists.

In CEN it is simple we deal only with the EHR and considers the EHR a documentation device that deals with reason two only: documentation of what has happened.

A lab test that is sent to an EHR system as data (information about the patient) is received and stored in the in-tray of the system.
The physician reads the data and accepts it to the healthcare record at that moment he documents the reception of the lab results and make them part of the EHR. Then he is able to formulate a comment about them or act up on them. The comments or the actions will be documented again in the EHR.

In CEN we deal with the EHRextract for exchange only. The extract is an abstract of that what has been documented.

Question: Is data that is not yet documented part of the extract?
If so. How is that indicated?
If not what should we do?
I think it should not, because the EHR is there to document what has happened in the view of the healthcare provider.
Orders, instruction, letters them selves are outside of the scope of the EHR.

I don’t agree. I think that while this is what most extracts will be about, this is not in principle true. The Extract is capable of sending any piece of an EHR. The requested piece (e.g. ‘everything to do with a cystic fibrosis patient’s care for the last 2 years’) could easily have Observations, Evaluations (i.e. any kind of opinions, including diagnoses, risk analyses, plans, goals etc), Instructions (and a CF patient will have a lot of those), and Acts (records of actions taken due to Instructions). THere is no reason not to be able to transport this kind of information in the Extract.

The openEHR (and CEN) approach is simple. The information contains the “triggers”; you send the information to the right person/place and it gets acted on. Just like writing a letter: you put a stamp and address on the front, and you drop it in teh post box. You don’t have to ring up the post office to ask them to send the letter; they already know what to do. If you write in that letter to your aunt that you would like some dried flowers from her garden to give to your daughter on her birthday, your aunt knows what to do, once she has read the letter, and barring difficulties, she does it. If she fails, she writes back and tells you she cannot. If she has fallen down teh stairs and is in hospital, someone from there or your family will ring you.

Personally I do not believe the obsession with “trigger events” is meaningful in health information. “Triggers” are part of protocol design; protocols are for sending information, and should be kept separate from the requests/replies of the information. So, like the post office example, the protocol of how mail is sent around the world is a ‘post-office protocol’ which you, the letter writer know nothing about. You just know about the request you made to your aunt, and her reply.

I think that next to the EHR extract we need something that indicates that it is not an extract of what is documented but something that is an order, instruction, advice, a letter, a lab result, that is data to the receiver until he documents the reception and accepts it to his record system.
I prefer a new construct that carries the notion that it is a trigger.

In an openEHR environment, the receiver and sender systems will both understand the order, since it will be based on an appropriate, shared archetype of an Instruction (e.g. an order for a drug), and it will be at the “prescribed” step (prior to the “dispensing” step). What is not in the Instruction is the information that some particular pharmacy X is being asked to dispense the drug. But consider what normally happens: a patient presents a printed prescrption to the pharmacist and recieves a package in return. In e-health, the pharmacy receives a request for a drug; the request itself does not need to indicate which pharmacy is to dispense the drug; only the envelope in which the request is carried (like the postal example above). In other words, the pharmacy processes all requests for drugs (assuming apporpriate details included in the message) directed to it, rather than some other pharmacy.

How does this get done? Consider the following possibilities:

  • the pharmacy acts like a provider, and is able to see the patient EHR. In this case, all the data required for dispensing will be in the Instruction in patient B’s record. The trigger for actually dispensing will be a notification from the physician’s sytem to the pharmacy system asking for Instruction 0494585 to be acted upon.
  • the pharmacy is just a service, and not privy to health records of patients. It therefore has no reason or capability to receive or view pieces of an EHR. In this case it needs to be sent message which acts as a notification, and carries the content. Such a message can be constructed from the relevant Instruction.
    Now, in the second case, the easiest solution in principle is to generate a copy of the appropriate part of the Instruction, and send it off to the pharmacy by some communication servce, such as a call to their webservice.

In a messaging environment, one might assume an HL7 or EDI message. Generating an appropriate HL7v2 message would not be hard. But for v3, the situation may be more problematic: v3 RMIMs contain predefined models of much clinical content (replicating clinical models and information models of EHR content), mixed in with the schema representing the message; it may or may not be easy to generate the appropriate message, since there are many attributes which are required, which are not present in the EHR. How this can be solved will remain to be seen.

Since the content of this new construct will be the same information as an EHR-extract I foresee an envelope for the EHR indicating that it is an extract of a document and an envelope indicating that it is a trigger.

This is the solution-in-principle above; it still requires sensible service models for both sender and receiver; these service models encapsulate the protocols for sending and receiving, acknowledgement and all such things.

The signature placed on the extract of the document indicates: I sign this and take responsibility for the content of the extract.
The signature placed on the trigger indicates: I order you, I instruct you to do something and I take responsibility for that order, instruction.

  • thomas beale

  • If you have any questions about using this list, please send a message to d.lloyd@openehr.org

Hubert Mensch wrote:

Gérard,

May I propose a contribution on the the question you pose ?

1 - Concerning the cooperation between actors and roles
The two reasons you mention make me think of the Searle’s speech acts, amongst other developped afterwards by Vinograd in his “conversation for action”.
i) - reason one : “illocutionnary” act which can be :
– a request for an action performing (order, instruction, letter where one can find a request, …) expressed by a requester to an offerer and which implies an answer from the offerer : accept (and then the offerer commit to do what is requested), amend, refuse.
– a commitment for an action performing expressed by the offerer. In practice and depending on the cooperation rules, the commitment is sanctionned by an appointment. The appointment can be in the letter …
– a report on the the action performing expressed by the offerer to the requester (professionnal opinion or advice following a request, lab result, …)
– a validation of the report (or of the action performing ?) by the requester

all of this is part of the ‘protocol’ of communication

ii) - reason two : “propositional” act which gives the content of the speech act :
– professional content for a request, with reason, target, condition of the patients influencing the performing of the requested action
– professional content for the commitment
– for the report. it is interesting to distinguish there the differences of the professional content between a diagnostic and a terapeuthic action.

personally, I find it perverse to call information “acts” when it is clearly information! But in any case, this is the content of the request, sent under the control of the protocol which is implemented in the relevant services.

The conversation for action is one of the components of the cooperation. Another one is the coodination of the care plan performing, when for instance an actor/role notifies the following one that the action he was responsible for is performed (or is not performed). May-be other component exists ?

in the new openEHR Instruction, this is shown by an Act (= ‘observation of an act having been performed’) instance being committed to the record; the Act indicates which Instruction activity it was based on, who performed it, when, and what differences there were with respect to the specification of the Act (the ‘activity’ as we call it in openEHR). THere is an overall Instruction_execution which tracks all happenings due to an Instruction.

Reports and notifications are different : report is sent to the requester within the conversation for action, and notification is sent to the following actors/roles within the workflow coordination, may-be with an another professional content. The information system is intended to support this cooperation messaging.

If the executor of some act, e.g. a drug administration in a hospital, creates the appropriate Act instance in the EHR, any other physicians (e.g. the GP and specialist) concerned with that patient and the problem in question, can be notified by their system that a new Act has been committed.

2 - Concerning the fact that the EHR is the support of cooperation messaging.
In the schema you describe, the EHR is the obliged “documentation device” (or I could say messaging device) between HCPs to communicate within the “conversation for action” or for the coodination. One can imagine that there are 3 dossier elements related to the same action performing

  • in the performer dossier
  • in the cooperation system
  • in the requester dossier

The utilisation of these “versions” inof the same act the care process should be studied.

I agree. This is more or less what we are trying to do with Instruction, Act, Instruction_execution and so on - clarify this threads of information and activity.

  • thomas

  • If you have any questions about using this list, please send a message to d.lloyd@openehr.org

Williamtfgoossen@cs.com wrote:

If an EHR can only store what has already happened, so the past, it becomes useless for supporting clinical practice.

Well, certainly the openEHR idea of an EHR includes all past (observational), opinion/decisional, and future (instructional) information. What other EHRs have in them may vary. The EHR Extract is just a copy of some parts of the EHR (+ appropriate demographic information), suitably packaged for sending.

It is interesting that the concepts of trigger is fundamental to HL7 dynamics, and the 'mood' of the action / obs (so #1 and # 2) are already available in the RIM classes: INT, RQO, GOL, EVN are very powerful to support the dynamics of monitoring and workflow. Key components of EHR.

the mood codes should be artifacts of conversations between systems (this is their speech act basis). Unfortunately they are mixed up in an information model of content, i.e. the protocol definition of communications is the same as the definition of the information being sent. And this is further confused by whether the protocol-related parts of HL7 relate to the interaction of two systems processing messages, or the communications between two health professionals.

See also the ISO TS on the EHR in which both data / structures and dynamics are the first 2 chapters.

Sams distinction between action required versus for your information only is interesting and probably would need further development on message and on EHR level.

I think logically there is a need for a "notification"; however, remember that in any protocol, the sending of a message containing a request for a drug (say) to a receiver which is capable of dispensing drugs, is normally taken as being the notification. You don't send a request for a drug X to pharmacy A if you really want pharmacy B to dispense it. Clearly there are more elements to the protocol than just this. The key is not to confuse the protocol design with the design of the information, nor with who the interlocutors are for any given level of 'message'.

- thomas beale

Thomas,

I don’t agree.

We agree more than you think.

In an openEHR environment, the receiver and sender systems will both understand the order, since it will be based on an appropriate, shared archetype of an Instruction (e.g. an order for a drug),

You write: “since it will be based on the shared archetype of an instruction”.

This is correct: semantic notions are expressed as archetypes.
But what is the EHR? Answer: the record that documents what the healthcare provider received as data, his thoughts, his professional opinion, the plan, the orders, and what he wants exchanges with others.
The EHR is a document that documents.
Then the EHR is the record of what happened.
In that context it records that an order was given to dispense medication.
But this is not the same as the medication order itself.

When this information is sent to the pharmacist he will interpret it as an instruction for him to dispense.
When the same piece of information is sent to the patient it means that this is the record of what he was advised to do, or is the meaning that this is an prescription he must bring to the pharmacist,

In other words.
An EHR with a record of something (e.g. an instruction) needs something extra that indicates what the purpose is.
A record of an order is not the same as the order itself.

My question was: how will OpenEHR (EHRcom) indicate what the purpose of the order is?
Will we indicate this inside the EHRextract?
Will we indicate this in the envelope?
Will we leave this completely to the HL7 machinery of messages.

Gerard

– –

Gerard Freriks, arts

Huigsloterdijk 378

2158 LR Buitenkaag

The Netherlands

T: +31 252 544896

M: +31 654 792800

Gerard Freriks wrote:

Thomas,

I don't agree.

We agree more than you think.

In an openEHR environment, the receiver and sender systems will both understand the order, since it will be based on an appropriate, shared archetype of an Instruction (e.g. an order for a drug),

You write: "since it will be based on the shared archetype of an instruction".

This is correct: semantic notions are expressed as archetypes.
But what is the EHR? Answer: the record that documents what the healthcare provider received as data, his thoughts, his professional opinion, the plan, the orders, and what he wants exchanges with others.
The EHR is a document that documents.
Then the EHR is the record of what happened.
In that context it records that an order was given to dispense medication.
But this is not the same as the medication order itself.

In openEHR, we side-step all such problems and keep it simple. An instance of an Instruction documents roughly what might be in a prescription - the description of the drug or therapy(ies), the administration timing, route, etc, procedure if relevant, and other activities like monitoring, conditions for suspending etc. This is enacted by human or computer workflow executors. Each thing they do they record using an instance of Act or Instruction_act (new Entry subtypes in openEHR). It does not have to be part of the data who is being asked to enact it (but it can be - done with particpations); they will be asked via some appropriate notification (e.g. the patient coming to their hospital acts as a notification). They will obtain permission to see the relevant part of the EHR, and will start giving care, i.e. performing activities in the Instruction, which they will document with Act_instructions (Acts are used to document activities undertaken but not 'prescribed', e.g. ad hoc self-administration by patient). These Entry types by the way indicate the current state (in the state machine sense) of the Instruction (e.g. "active", "suspended" etc).

So we don't record 'orders' as such; we record what is to be done, and usually by whom. Then, notifications (including humans presenting themselves) cause things to happen, and those happenings are documented.

When this information is sent to the pharmacist he will interpret it as an instruction for him to dispense.
When the same piece of information is sent to the patient it means that this is the record of what he was advised to do, or is the meaning that this is an prescription he must bring to the pharmacist,

In other words.
An EHR with a record of something (e.g. an instruction) needs something extra that indicates what the purpose is.
A record of an order is not the same as the order itself.

My question was: how will OpenEHR (EHRcom) indicate what the purpose of the order is?

as you can see from above, we don't work this way!

Will we indicate this inside the EHRextract?
Will we indicate this in the envelope?
Will we leave this completely to the HL7 machinery of messages.

Gerard

- thomas