a model for medication strengths

Clinical (archive)
1

I was wondering if anyone had any experience or knowledge of
a useable model for storing computable medication strengths
(and medications forms etc)?

Now I realise this is pretty complex - for instance these are
some of the medication strengths strings listed for some Australian
medications..

0.3mg/mL (0.03%)
0.4mg-10.0mg-2.0mg/mL
0.54g-1.28g/10mL
0.375mg
1% w/w
1 Million KIU/100mL
10 dose
100mcg/capsule

However I actually think that it would be possible to construct
a way of modelling most of these in a regular manner
(certainly if we leave the really hard/obscure ones to the side
it should be possible to cover the _majority_ of medications).
It doesn't even have to be a particularly clever model - I mean
it could have 50+ distinct classes each covering a different style
of medication "strength/form" with no clever OO hierarchy or
anything, and surely it would be better than doing everything
as a string?

I would presume that this is the type of thing they do at
HL7, but whereas I can find models for how medication
components and ingredients relate etc it seems that
strengths and forms are just strings??

From the openehr end, there is some attempt in

the medication archetypes to put it in a computable
form but I'm not sure the archetypes are at
all complete yet - they have a "Strength Per Dose Unit" as a mass
unit, and a "Dose Unit" as a terminology value but this
does not seem to be adequate to me.

Is there any published work that looks at this problem that
can help us flesh out the medication archetypes?

The nearest paper I could find in this area was

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=516250

Andrew

2

Why don't create a PQ object and assign to its units UCUM as
terminology name and the desired units to the code?
Here is a list of the UCUM units list
http://www.hl7.de/download/documents/ucum/ucumdata.html

3

Well, basically, "strength" should be "amount per amount".
Sometimes (as in the examples given above) there is assumed
knowledge involved, say, how many drops are in a milliliter.

So, with each of the "strength" indications one needs to
provide an algorithm and appropriate factors with which to
"reduce" it to a "normalized" form. That way it can be made
computable. Note that the above do not seem to all mean
quite the same things.

Those factors and algorithms could well be encapsulated into
dedicated "classes" as you suggested.

Karsten

4

Hi Andrew

The dm+d data model seems to me to provide some of this. From the dm+d data model V2.3 from this page:

MailScanner has detected a possible fraud attempt from “195.97.218.30” claiming to be MailScanner warning: numerical links are often malicious: http://195.97.218.30/dmd_download.htm

"

pharmaceutical strength

The amount of ingredient substance (as identified by the attribute ingredient substance identifier or basis of strength substance identifier as indicated above).

This attribute indicates the quantity of the substance per defined unit of measure in the Virtual Medicinal Product (e.g. one tablet, one ml) measured by weight or volume per unit or concentration. An ingredient may be present without a strength.

Pharmaceutical strength has 4 components, where a strength is provided the strength value numerator (SVN) and strength value numerator unit (SVNU) are mandatory. Strength value denominator (SVD) and strength value denominator unit (SVDU) are used to fully express ‘per’ strengths.

EXAMPLES:

Paracetamol 500mg tablets

Ingredient SVN SVNU SVD SVDU

Paracetamol 500 mg

Paracetamol 250mg/5ml oral suspension

Ingredient SVN SVNU SVD SVDU

Paracetamol 50 mg 1 ml

Hydrocortisone 1% cream

Ingredient SVN SVNU SVD SVDU

Hydrocortisone 10 mg 1 g

"

would suggest a good starting point, perhaps?

5

Hi Andrew,
The dm+d medication model does contain these concepts and it was used as the basis for constructing the
Australian Medicines Terminology (AMT) - Julie and Hugh (original authors of dm+d)
came to Australia to help with this process.

The result (AMT) was handed to the Australian National eHealth Transition Authority (NeHTA)
by a joint MSIA/DoHA working group over 4 years ago!
Since then NeHTA have “simplified it” and populated the structures at least for PBS medications and are currently working on the rest
of the non-PBS medications

How much has been removed as part of the “simplification” I am unclear

You can access the AMT by going to the NeHTA Web site and registering for a SNOMED licence (free for Australians).
You will then be able to download the AMT

Regards
Vince

Dr Vincent McCauley MB BS, Ph.D
CEO, McCauley Software Pty Ltd www.mccauleysoftware.com
President, Medical Software Industry Association www.msia.com.au
Vice Chair, HL7 Australia
p: +61298186493
f: +61298181435

6

Thank you all for your comments on this,
specific responses below

Diego:

I agree that UCUM might be a useful within
a model - however, I just don't think a single
PQ covers the scope of strengths that I am seeing in
use in Australian medications (they become a lot
more complex).

Karsten:

Well, basically, "strength" should be "amount per amount".
Sometimes (as in the examples given above) there is assumed
knowledge involved, say, how many drops are in a milliliter.

So, with each of the "strength" indications one needs to
provide an algorithm and appropriate factors with which to
"reduce" it to a "normalized" form. That way it can be made
computable.

I agree that this is useful, but this "normalization" is even beyond
what I require (though would be the obvious phase 2!).
I just need a simple model that lets me
literally "store" the strength designations for common medications
in a way that is more controlled.. (my use of the word
computable was maybe too strong)

So I need to be able to represent that there is a medication
called

Aluminium Hydroxide/Magnesium Hydroxied/Simethicone

that is sold as either a

a tablet 200mg/200mg/20mg
or in a liquid form 200mg-200mg-20mg/5mL

Preferably the model would prevent storage of incorrect data
- dressings would require an 'area' measure, injections would
require 'mass' / 'volume' measures, tinctures would only allow
percentages etc. Ampoules are sometimes measured as
percentages, sometimes ratios, sometimes mass/volume.
The model would need to cope with the fact that sometimes
tablets have multiple medical ingredients and so it must
be able to store N masses etc

I need this to be controlled so that I can do things like
display them in a uniform manner (do I put "-" between
components of a tablet, or "/" as in the example). I don't
know what the answers are but without some formalization
I know that we'll all do it different in an _uncontrolled_
way.

I know that there are some national efforts to put medications
in terminology (dm+d, AMT etc) and if all this information
is in the terminology then perhaps we don't need any
"strength" modelled in the medication archetypes? We
just point to a single term that says what the exact
medication trade product is, and the strength is
implicit from that?

Note that the above do not seem to all mean
quite the same things.

Yes - I was just cutting and pasting various "in use"
strengths from a whole bunch of australian medications.
There appears to be rougly 1000 different strengths
in use (i.e. if we say 0.1mg/mL is different to 0.2mg/mL).
The computer scientist in me wants to bring order
to this universe - surely we can get this down to
under 100 documented "strength" types

InjectionStrength XXmg / YYmL or XXmg[] / YYmL
CapsuleStrength XXmg or XXg or XXmcg
etc

Paul:

I will look into dm+d modelling a bit - thanks for the tip. I
have been using the AMT (which is its Australian
counterpart) but it doesn't seem to have the concepts
you mentioned.

Andrew

7

The dm+d medication model does contain these concepts and it was used as the
basis for constructing the

How much has been removed as part of the "simplification" I am unclear

Thanks Vince - this question arose because I was trying to do some
mappings into the AMT - the simplified version (as of 1.10) doesn't
seem to be populated with any of these concepts (thought the
accompanying UML diagram does seem to indicate the relationships
may exist). I guess it may be a case of waiting till some of this
data appears in the AMT (I imagine actually getting the medication
themselves listed is NEHTAs first priority)

Andrew

8

Diego:

I agree that UCUM might be a useful within
a model - however, I just don't think a single
PQ covers the scope of strengths that I am seeing in
use in Australian medications (they become a lot
more complex).

correct. There's several parts and PQ is only a part
of the whole, as is UCUM

I think that there would be general interest in the
question of information models vs ontologies in
medications. And there has been considerable
discussion in HL7 on this.

One of the big problems is that the discussion is
dominated by our collective weaknesses, and we do
not know how to ask what the proper way to do things
is.

Grahame

9

Hi Andrew,

I have not looked too hard at these, but I would expect DV_PROPORTION
and DV_INTERVAL<DV_PROPORTION> to be the most common types to use, or
else just two DV_QUANTIFIEDs in an archetype (the first one).

- thomas beale

Andrew Patterson wrote:

10

Hello!
If you look for a medication modeling workforce:
At the moment we do see a big amount of work happening in the ISO world, and
also involving CEN and WHO, with EMEA and others:

Here is a list of some of the work items:

Individual Case Safety Report (ICSR) DIS (erwartet für Februar 2009):
* Part 1. ICSR Framework Reference Model (prEN ISO 27953-1)
* Part 2. ICSR Human pharmaceuticals (prEN ISO 27953-2), will be used
by the "International Conference on Harmonisation of Technical Requirements
for Registration of Pharmaceuticals for Human Use (ICH)"

- Identification of Medicinal Products (IDMP) CD (announced for mid
January 2009):
* prEN ISO 11615: Data Elements and Structures for the Exchange of
Regulated Medicinal Product Information for Drug Dictionaries (MPID)
* prEN ISO 11616: Structures and Controlled Vocabularies for
Pharmaceutical Product Identifiers (PhPIDs)
* prEN ISO 11238: Structures and Controlled Vocabularies for
Ingredients (substances)
* prEN ISO 11239: Structures and Controlled Vocabularies for
Pharmaceutical Dose Forms, Units of Presentation and Routes of
Administration
* prEN ISO 11240: Structures and Controlled Vocabularies for Units of
Measurement
* prEN ISO 11595: Structures and Controlled Vocabularies for
Laboratory Test Units for the Reporting of Laboratory Results

Has somebody looked at those / are we connected to this ?

I will also relay this list to the IEEE 11073 Personal Healthcare Devices
(PHD) group, they are at the moment working on a medication dispensing
device, and have the same problem of "techies needing some clinical input".

Greetings from Vienna,
Stefan Sauermann

Von: openehr-clinical-bounces@openehr.org
[mailto:openehr-clinical-bounces@openehr.org] Im Auftrag von
Grahame Grieve
Gesendet: Mittwoch, 21. Jänner 2009 02:15
An: For openEHR clinical discussions
Betreff: Re: a model for medication strengths

> Diego:
>
> I agree that UCUM might be a useful within a model -
however, I just
> don't think a single PQ covers the scope of strengths that
I am seeing
> in use in Australian medications (they become a lot more complex).

...

11

I have not looked too hard at these, but I would expect DV_PROPORTION
and DV_INTERVAL<DV_PROPORTION> to be the most common types to use, or
else just two DV_QUANTIFIEDs in an archetype (the first one).

I think, as Grahame noted, it will come down to where the balance lies
between putting stuff in the information model vs ontology. But my
initial impressions are that a simple information model
(ala just a DV_PROPORTION, or interval of them) will be
inadequate to represent the complexity of medication strengths
and forms that are used in real products (in Australia).

When it comes time to review the medication archetypes we should
review this issue. I don't think I have enough expertise to
progress this any further, though I will play around a bit more with
the AMT to see if it has any insights..

Andrew

12

* prEN ISO 11238: Structures and Controlled Vocabularies for
Ingredients (substances)
* prEN ISO 11239: Structures and Controlled Vocabularies for
Pharmaceutical Dose Forms, Units of Presentation and Routes of
Administration
* prEN ISO 11240: Structures and Controlled Vocabularies for Units of
Measurement

Thanks Stefan,

I think these 3 would be very relevant - does anyone know
how far this work has progressed?

Andrew

13

Yes Andrew, I was at the last ISO WG6 meeting and have a copy of the latest
drafts to go to a Committee Draft.

Evelyn

14

prEN ISO 11238: Structures and Controlled Vocabularies for
Ingredients (substances) -
prEN ISO 11239: Structures and Controlled Vocabularies for
Pharmaceutical Dose Forms, Units of Presentation and Routes of
Administration
prEN ISO 11240: Structures and Controlled Vocabularies for Units of
Measurement

These documents are under revision as Committee Drafts on WG 6 of ISO TC 215 and should be available with the new revisions on the ISO TC 215 shared point documents for revision by the Committee. In Edinburgh, in the last week of April, during the ISO TC 215 meeting these documents will be discussed.
Since this a new work item a lot can change.
Beatriz

Andrew Patterson wrote:

15

HI Andrew

I do not think that you will find that anyone will record this sort of
information in the health record. If they do it will be a research
environment or a complex formulation archetype which will have the vehicle
and each active agent with details about each separately.

You cannot force all that into one data value.

Therefore the ontology/terminology will be the place where the strengths are
described.

Cheers, Sam