Of recent times there has been the release of beta versions of AMT (Australian Medicines Terminology) Version 3.0 which is intended as the drug terminology component of SNOMED-CT-AU which is meant to be the terminology system that is used to represent all clinical terms eventually.
Sadly version 3.0 of AMT is deeply flawed and it is scheduled to go into production in June this year - replacing the Version 2.0 releases which have been underway for the last few years and updated regularly.
The problem is that the present plans for AMT V3.0 do not cover the substances that are found within medications so if you are a software designer planning to use this terminology for allergy checking you can’t make sure the substance you are checking for is not present in an individual medicine.
These were included in the Version 2.x models.
When NEHTA is asked they say.
“The exclusion of certain AMT substances in the AMT-SNOMED CT-AU Substance map was due to the exclusion of Pharmaceutical ingredients (PI) in the AMT v3 model. PI’s are included in the AMT v2 model.”
They go on to say that they are now planning to analyse how substances (i.e. ingredients) can be included.
The reason for not having a workable and agreed product for release is simply not explained by anyone. I won't even start about the possible safety issues and backwards compatibility issues that all this potentially raises.
If ever there was an example of the rubbish governance in the e-Health space this is it!
Just junk.
David.
44 comments:
Hear, Hear David, Hear, Hear!
So how much has the Australian Net Taxpayers paid for this dismal failure?
Categorically incorrect David.. the AMTv3 most certainly includes the substances that make up the medicines..
Perhaps the person who fed you this tidbit is talking about the map from AMT substances across into SNOMED substances? (the answer from NEHTA seems to indicate this).
If so, that is quite different, and not at all what you have said in the rest of the article.
This is the kind of stupidity that destroys any confidence of e-health reporting in the journalistic blogging space.
According to people I am talking to you are just categorically wrong.
AMT 3.0 is not suitable for clinical use in interaction detection. It is that simple.
If you know better kindly explain.
The SNOMED to AMT mapping has not been finished and a properly useful product does not exist.
If you can explain just how this is not the case - let's hear it since it seems NEHTA is still working out how to fix the issue according to them...
David.
My main problem is the categorically
demonstrably incorrect lead sentences.
> The problem is that the present plans
> for AMT V3.0 do not cover the
> substances that are found within medications
I'm saying that is wrong. I'm saying you can download the AMT v3.0 file, and look at the substances
contained in each MPUU and MP (and the relative
strength of each ingredient).
I've implemented it multiple times in multiple different organisations. I've sat through chats by NEHTA talking about how it
works.
You can look at the class diagrams that
accompany the AMT v3.0 and SEE THE SUBSTANCES on the frickin diagram.
You seem to be mixing up an AMT->SNOMED
substance map, and the actual AMT release.
Not the same thing..
I think I know what the person giving you the
information is getting at - there may well be
some issue there - but I would guess the issue
is a lot more subtle and nuanced than you
have made out.
OK - you claim all is OK..so why do 3 different sources say the AMT v3.0 is not ready and is NOT fit for purpose?
Which live clinically useful decision support and allergy alert systems are actually using it as of today?
David.
I didn't claim all is ok. I implied your article was crap and didn't make any sense.
paragraph 1 - starts off fine
paragraph 2 - unsubstantiated, pejorative
paragraph 3 - incorrect, provably so
paragraph 4 - correct in v2, and still correct in v3
paragraph 5 (NETHA quote) - no idea if it correct but seems to be a response to some other topic i.e. snomed substance map
paragraph 6 - more from NEHTA, but you paraphrasing
paragraph 7,8,9 - incorrect conclusions based on presumably all the other misunderstandings swirling around in your head
If you want to rewrite the article, removing all the nonsense sentences, then perhaps we could move on to other more nuanced questions such as "what does fit for purpose mean" i.e. fit for whose purpose, "what is preventing AMT adoption" etc..
Which are indeed much more interesting and your correspondants might indeed have valid points to make..
Well I have been told that Substances have been excluded from the V3 model, by NEHTA and I am pulling my hair out about it.
Really it needs a SNOMED-CT map as well, as previously, in V2 there was no substance hierarchy in AMT which makes it useless. I was told that Decision support was "out of scope" for AMT. That statement makes me shake my head as we are left with codes that are only useful to restock pharmacy shelves!!!
Why bother with a medication terminology if decision support is out of scope? This is a monumental waste of taxpayers money, which aligns AMT with all the NEHTA "work". Its time for software vendors to walk away from this mess and do some useful work.
Patto is entirely and factually correct, all other comments offer opinion only.
1. BETA release versions ALWAYS ALWAYS ALWAYS are marked as NOT FOR CLINICAL use. Deployments need to wait for the official final release.
2. NeHTAs response carefully refers to the MAP of substances (between AMT and SNOMED) and NOT AMT content itself.
3. From AMT v3 beta = 4 examples:
2825011000036100
sulfur-precipitated (AU substance)
31372011000036108
hyoscyamine (AU substance)
2227011000036100
infliximab (AU substance)
32039011000036102
tetrahydrozoline (AU substance)
Some independent, unbiased, objective fact-checking wouldn't go astray.
Agree with Patto on this. Andrew M:
"Really it needs a SNOMED-CT map as well, as previously, in V2 there was no substance hierarchy in AMT which makes it useless"
Because decision support engines are based on the Snomed-CT substance heirarchy? (or a more real question: is there any prospect that the Snomed-CT substance heirarchy will ever be suitable for decision support?)
Regarding "Why bother with a medication terminology if decision support is out of scope?": Fair question, but there is an answer: we can use a consistent terminology to compare and aggregate data, even if we can't reliably use it to do decision support. This is considerably more useful than "codes that are only useful to restock pharmacy shelves"
Thanks for all the comments.
My central point - not well made, but now clearer - is that AMT is not yet clinically fit for use as yet after years and years of work.
I wonder when this will ever change?
Sorry if I got the detail wrong. The great thing about the blog is that I rapidly get to know if what I am saying is right or wrong!
David.
It's good to see people discussing information issues, not technology.
However, a few questions:
Why has this not been sorted out before?
Is it a new issue or has there been a sudden realisation that it's still an unresolved problem? Or something else?
FWIW I think the delay and frustration is due to one factor.
Complexity - this stuff is very hard to get right and agreed.
David.
Graham, I am not aware that SNOMED-CT substances are unsuitable for decision support, do you have any evidence for this? But if that is the case they should be fixing it, and have had millions of dollars to do so but have done nothing. Aggregation of data that can't be used for decision support sounds like rearranging deck chairs on the Titanic.
Since AMT was very young I have pushed for more semantics , but keep hearing the out of scope weasel words.
I was told that the substances, not just the map have been removed in V3. So should I believe you or NEHTA?
I've always wondered why NEHTA themselves don't openly engage in discussions on public forums like this. This topic would have been a perfect platform for them to resolve the debate and clarify thee issue - "straight from the horses mouth".
They do themselves a disfavour by not being open and transparent.
I guess their demise now makes the issue redundant. Maybe their successor will get on the front foot!
Andrew:
"SNOMED-CT substances are unsuitable for decision support, do you have any evidence for this?"
Well, I've looked at the heirarchy. That's one kind of evidence ;-)
And I've tried to find a system doing decision support based on the heirarchy before but not found it. That's not evidence, of course.
Counter question: do you have evidence that it is?
"Aggregation of data that can't be used for decision support sounds like rearranging deck chairs on the Titanic."
I think it's a little bit better than that. Really, the ability to at least consolidate a medications list is not merely moving deck chairs around, however much we might like good decision support.
"I was told that the substances, not just the map have been removed in V3. So should I believe you or NEHTA?"
You should believe neither of us, you should download the beta and have a look. The quoted descriptions in the comment above are from the beta, and are true - so substances have not been removed
(and just to clarify Andrew's language - the map has not been removed, it just never existed, at least to my knowledge)
> I was told that the substances,
> not just the map have been removed
> in V3. So should I believe you or NEHTA?
Well neither is true. Both the substances AND the substance map are in v3.
Don't believe Grahame, me or NEHTA. Download the file yourself and see..
In the current (downloadable) AMT pre-prod zip
Refset\Content\xder2_csRefset_SubstanceToSnomedCtauMappingSnapshot_AU1000036_20140331
The gist of the response from NEHTA seems to be that there are now some 'unused' substances that have been removed from the MAP in V3. I have no idea if that is true - but it seems to be a much smaller problem and one that probably doesn't deserve the Chicken Little approach being taken on this blog.
the map between AMT and SNOMED CT substances exists - not sure if it has been included in this tech preview or BETA release of AMT v3, or whether it will be in the official release later this year, or whether it needs further review before release (at some unknown date) given changes in BOTH AMT v3 and SNOMED CT substances (and versioning alignment issues) Any map product for any health vocabulary really needs to be constructed after all official release content is finalised, otherwise we build instability upon instability.
Despite the protestations I do not buy the arguments.
I don't not matter whats in the beta release, I have been told that PIs (Substances) will be remove from the initial V3 release of AMT - Should I not believe what I have been told?
To declare that SNOMED-CT is not fit for use for decision support without evidence is over the top. If it is not then its a major flaw in the international terminology chosen for use in this country and it should be fixed. On our testing and examination it is fit for use, but we have not exhaustively tested it.
The point is that AMT is to little to late and there is no mood for making it fit for purpose. Given that its probably already cost hundreds of millions of tax payer dollars its a disgrace.
Rabid defences of AMT should come with disclaimers about it been s source of consulting income. The sky may have already fallen on NEHTA, but if anything is going to change we need to look at the problems that have occurred under ts management and not deny them. Things like RXNorm in the USA seems to have got it right, why can't our public servants manage to do that?
The AMT could provide some utility for improving basic processing of medicines information, particularly when shared or aggregated across systems - it could provide standardised names for medications as well as provide the link between Trade names and non-proprietary names. I say 'could' because it is as yet, after all these years, not widely implemented in clinical systems in Australia.
The AMT also includes names of the active ingredients (substances) of each medication in its list.
The AMT could provide even more utility if it also contained links to drug class information and useful substance knowledge-bases. The substance knowledge-base most favoured by NEHTA is the one included in SNOMED-CT.
So to use an example cited above 'sulfur-precipitated' is an ingredient of many drugs listed in the AMT. If an electronic allergy checking system wished to check whether 'Clearasil Pimple Treatment Tinted' contained 'sulfur' using the AMT, it could only determine that it contained the substance 'sulfur-precipitated'. Is this the same as 'sulfur' for allergy checking purposes? The AMT, on its own, can't help. The AMT to SNOMED-CT map produced by NEHTA does map 'sulfur-precipitated' in the AMT to 'sulfur' in SNOMED. This map exists in V2 and the V3 beta release.
Notwithstanding the complexities involved in this case ( e.g. see http://www.australianprescriber.com/magazine/31/1/8/10 ), the AMT is of limited value with respect to its substance names. This has been known since its inception, and very little progress has been made to resolve it. One solution is to map each substance name in the AMT to a corresponding substance in SNOMED-CT so that the more powerful SNOMED-CT substance hierarchy could be used.
Producing such a map and maintaining it requires effort. SNOMED-CT does not contain 'sulfur-precipitated' , for example. The IHTSDO also has a long running project to redesign the SNOMED-CT substance hierarchy.
So the move by NEHTA to replace AMT v2 with AMT v3 should have little impact on those few systems that currently implement AMT v2, other than being disruptive. V3 has both advantages and disadvantages. Changing over would be a pain.
NEHTA could have done, and still could do much more to assist organisations adopt the AMT. And it could have done and still could do much more to improve the usefulness of the AMT. It doesn't even provide basic synonyms such as 'amoxycillin' / 'amoxycillin'.
NEHTA continues to put up barriers to accessing basic information about the fundamental e-health building blocks it purports to champion.
It continues the ludicrous claim to:-
"Lead the uptake of eHealth solutions of national significance for Australia
and
Progress and accelerate their adoption through the health system"
Good patient medication information is critical to improving healthcare. Good terminologies can and should help. Good documentation, education and support for these are essential.
Australia continues to wallow in no-man's land.
> I don't not matter whats in the beta release,
> I have been told that PIs (Substances) will
> be remove from the initial V3 release of AMT
Seriously?? There is a 'pre-production' (April) release of AMT v3 that includes substances. It is probably about half of the modelling data included in the release. And you think that they are going to remove all that data in between pre-production (April) and production (July)? And fundamentally change all the modelling data. And do this without telling anyone other than your anonymous mate.
I can declare that the sum total of money I have made from AMT consulting is exactly $513, which I once got for taking a day off work and going to a meeting.
As a counterpoint to being labelled an AMT cheerleader - I agree with pretty much everything Eric says.
I have nuanced opinions on where things are good and bad about NEHTA and AMT.
But if you put factually incorrect statements into a story I'm going to jump on them, and if that makes me a rabid defender then so be it.
Patto said:
"But if you put factually incorrect statements into a story I'm going to jump on them, and if that makes me a rabid defender then so be it"
Nope - I for one value the learning that I get from all the experts around here. I also know if I hear nothing I am close to the mark!
It really is a pity AMT is not properly sorted - as Eric makes pretty clear.
David.
This has been a fascinating learning experience about the difficulties and misunderstandings of AMT and SNOMED, mappings and all.
If you scan these comments you start to see how hard it is for us meagre people at the coalface trying to implement interoperable meaningful systems - and yes we would love to use standard terminologies but we can't understand how they work and if and how to use them. Isn't that NEHTA's job to help us?
Dear Eric - your wonderful explanation and examples help us to understand how it all fits together.
Oh NEHTA - why can't you explain things like this? All that money spent on NEHTA, and then Eric comes along and explains it all simply, coherently and completely free of charge.
Dispute the fact there is a lot of research and evidence that poor leadership contributes more to project failure than poor project management lets blame the projects and those working hard within them and force a ridged date on them rather than allowing a little more time get it right so the the next stage has strong foundation and not an inherent debt.
Look at the top layer and you will find a pattern
"Look at the top layer and you will find a pattern"
Well no suprise given they put a trolley dolly in charge of it all!
"Look at the top layer and you will find a pattern"
Google 'National Design Manager and do a history check, try and find a rationale reason why such a success rate is tolerated
It seems there are some clear take-away messages from this post and the ensuing dialog:
1. AMT V3 (just as AMT V2 does) contains links from medicines to substances
2. David's three sources are at best mistaken, and at worst spreading misleading and false information; he should be wary of their future claims (at least with respect to AMT V3) in future and seek input from others here that have a better handle on the facts
3. there exists a map between AMT V2 substances and SNOMED CT substances
4. Patto is correct in his clarification that it is planned to remove some substances that are unused in AMT medications
I will add the following:
1. A "map" is not magic; it is just a lookup table and it will be subject to certain limitations based on the intended use at its creation and these constraints may (usually do) limit its use in other scenarios. The SNOMED CT-AMT substance map suffers from this because the SNOMED CT substances have ambiguous meanings. For example, it is not clear whether certain concepts represent classes of compounds (e.g. a base and various salt forms) or just specific compounds (the base itself). AMT is clear on this, but it impacts the interpretation of the map. Until SNOMED CT Substances is properly fixed (and we are now seeing progress on this), the SNOMED CT Substances (including substance classes) will not be a good basis for decision support.
2. Decision support requires a knowledge base. Most GP systems use MIMS for this. MIMS has a map between AMT codes and MIMS codes to enable their decision support rules to be driven by AMT codes.
3. AMT V3 adds significant new content -- machine processable data on ingredient strength and quantity. SNOMED CT does not have this at all, but the AMT work is preparing the path for the capabiliy to move to SNOMED CT (full disclosure - I have been involved in contracts to supply expertise and the technology that underpins this).
There are a number of short comings, limitations and compromises in AMT V3, but it is an improvement over V2 (formal modelling of strength & quantity information is a major boost for QA, for example).
Remember, perfection is the enemy of good.
I hope you are correct, but I would expect to get the correct answer from NEHTA, rather than a AMT user on a blog. It remains to be seen if that is the situation or not. Andrew Patterson feels they have removed the link to Substances from the trade product, but not the Medicinal Product and you can navigate from the trade product to the medicinal product to get to the substance. If this is the case then the map can then be used to get to the SNOMED-CT Heirarchy for allergy checking.
I would have expected this answer from the AMT team, but got the allergy checking is decision support and its out of scope response.
> feels they have removed the link to Substances from
> the trade product, but not the Medicinal Product
> and you can navigate from the trade product
> to the medicinal product
Well, more specifically (and in slightly more technical detail) - as I understand it - they have removed the modelling of Pharmaceutical Ingredients (PI's) on the trade side (TPUU). Historically in AMTv2, the TPUU has its own completely independent modelling of substances in a medicine IN ADDITION to the modelling of substances you could get from the corresponding 'generic' side (MPUU).
So now, the only stated substance relationships are off the MPUU/MP. And so yes, you can get to them by navigating from TPUU to MPUU - but as one of the bonus features of the work the CSIRO and NEHTA have done is that they run a proper ontology classifier over the whole AMTv3 - and that classifier will actually now detect the MPUU->TPUU relationship and duplicate all the appropriate substance relationships onto the TPUU.
So to give a very concrete example
the MPUU
aspirin 100mg tablet
will state a relationship that it contains 100mg of the AU substance 'aspirin'.
There will also be a TPUU (in fact lots of TPUUs for each brand of aspirin)
Astrix 100 mg tablet
with a stated relationship saying that its parent is the aspirin MPUU.
What the classifier will do is look at all these relationships and realise that the Astrix 100mg tablet should also have a relationship saying it contains the AU substance 'aspirin'. And it will insert that into the resulting AMT release file.
And so based on what I can work out from various emails I've been sent (but with no reference to anyone actually on the AMT team) - I think the removal of the PI modelling has meant there are a few substances that are no longer referred to anywhere - and so NEHTA have chosen to remove them from the corresponding SNOMED map, and maybe indeed from the actual AMT release.
But I think that the talk of removing Pharmaceutical Ingredients and substances was misunderstood to mean that _all_ the modelling of substances/ingredients was being removed, even from MPUU side.
AnonymousMay 24, 2014 11:18 PM
I think you mean the Nehta national design manager, not sure why you relate him to AMT unless he is holding progress back? I do agree though, he seems to have led, and still does in many cases, the PCEHR, NASH and HI services, guess being the CEO's mate carries weight
Two years on - NEHTA has faded into the pages of history, have the concerns raised here been addressed or should the ADHA be guided towards progressing this important work?
ADHA is led by people who believe they know they are doing, they believe they know what they are doing is right and they will not stop until they have done it.
Problem is, they don't know that what they are doing is wrong.
That's the trouble with belief - you can't argue against it.
"Two years on - NEHTA has faded into the pages of history, have the concerns raised here been addressed or should the ADHA be guided towards progressing this important work?"
Unfortunately, NEHTA has not faded, but has simply been rebadged as ADHA. ADHA can no more be guided towards anything than NEHTA was. One only has to read the incoherent nonsense on their website under the heading "What is digital health?" in order to appreciate how depressingly dire the whole NEHTA to ADHA transition saga has been.
Yes the Australian Digital Health Agency's new website is definitely fascinating. Here's how it describes 'digital health':
"What is digital health?
Digital health is about electronically connecting up the points of care so that health information can be shared securely. This is the first step to understanding how digital health can help deliver safer, better quality healthcare.
Features of the My Health Record system
Digital health offers a range of products designed to help you deliver safer, quality healthcare…."
Fascinating: in order to understand the nature and purpose of digital health, you first need to electronically connect the points of care. OK….. so now I need to distract you and so will talk about the features of the My Health Record System….
Also, the new site exhorts the reader to 'Register for Agency Membership!' and 'Join the My Health Record Developer Community!' and 'Register for your My Health Record today! (Click here to take control of your health…)' All with exciting exclamation marks!
Check it out! http://www.digitalhealth.gov.au/get-started-with-digital-health/what-is-digital-health
Thanks for "sharing".
Lets start at the top, paragraph 1, definition:
"Digital health is about electronically connecting up the points of care so that health information can be shared securely. This is the first step to understanding how digital health can help deliver safer, better quality healthcare."
Some quibbles.
1/ The first and second sentences are logically incompatible. Not a bad start then. The first defines digital health as 'about' connecting and sharing. The second says that sharing will then help us understand digital health is. Either digital health is connecting and sharing or it is not.
2/ It is not. Digital health is about the information services one can provide to support health care. Not worth mentioning in the foundational definition I guess. "Sharing" information is a specific decision to allow those services to be distributed. A critical feature in a complex system like health care, but not the end game. And we have been 'connecting' now for 20 years?
Petty criticism of a new agency? Maybe. Or just maybe, the community is expecting this to be a serious and professional agency, run by people with e-health expertise, and not have its foundational text written by someone who read all they know about e-health from the back of a corn flakes packet (retrieved no doubt from the old NEHTA kitchen).
We expect a lot, because this is so important, and many of us have lost faith in government's ability to deliver. If you want us to treat you differently, be different.
Not sure the question was regarding a website critique but something more fundamental.
On the website, it looks like a simple find NEHTA - Replace with Digital Health Agency, I am happy to give them a break on cosmetics, it is a leaderless ship at the moment and drawing on the collective imagination of the old elite so I hear.
Why this term 'digital health'? What happened to eHealth? I've heard of ecommerce, and even egovernment, but not digital commerce nor digital government. What is the purpose of this rebranding???
Could this be ADHA planting blog posts trying to divert attention from the actual questions? For the record, yes all this has been resolved and it just needs the Jurisdictions to insist on AMT being implemented and some minor tweaks to MyHR
Anonymous said "For the record, yes all this has been resolved and it just needs the Jurisdictions to insist on AMT being implemented and some minor tweaks to MyHR".
I don't think that is true at all. According to the most recent AMT documentation:
Substance concepts used in the AMT are unique to the AMT, and are a different set of concepts than used by the International Edition of SNOMED CT or the Australian extension SNOMED CT-AU.
This is partially due to the AMT’s parallel development to SNOMED CT, and partially due to a concern at the time of creation of the AMT that anticipated future changes to the SNOMED CT Substance hierarchy may destabilise the AMT.
Implementers of the AMT should note this separation, and use AMT substances for AMT-related activity. The NCTIS publishes a map between AMT substances and SNOMED CT substances. Refer to Section 2.3.5.6 for further information on this map.
Note that use of the map to integrate AMT with SNOMED CT substances may result in incorrect inferences. This is due to known modelling issues in the SNOMED CT Substance hierarchy which are being addressed by an IHTSDO project. However, until this work is completed and the International Edition of SNOMED CT is updated, care must be taken when using the AMT Substance to SNOMED CT-AU mapping."
There is a fundamental difference in the way AMT substances are modelled vs the way SNOMED CT substances are modelled. The AMT does not model substances, it just contains a flat list of 2157 substances. The SNOMED CT substance hierarchy (62879 descriptions) is highly structured. e.g. "amoxicillin" is nested some 14 concepts deep and the hierarchy can be used to determine that "amoxicillin" is a "penicillin".
Thank you Eric (July 13, 2016 11:37 AM) for the illuminating quote "According to the most recent AMT documentation: .......... "
You have confirmed my long held views that after reading it very very carefully I now know why so very little has been achieved in eHealth by NEHTA and the Health Department, why confusion reigns far and wide, why the bureaucrats have no idea what they are on about and why no sane consumer or their doctors would ever contemplate using the Health Record system being thrust upon them by the Health Department.
Can someone tell me please where do I go in My Health Record to get a printout of My Health record? I don't think I'm stupid but for the like of me I have been unable to just print My Health Record. Also, please tell me I'm not stupid. Thanks
That will be feature in the next release, an app will be provided to allow you to take a screenshot or for mobiles an innovative app that takes a photograph your tablet using a smart phone, however, it will be a beta release so only available on Android.
Oh, that's interesting. My sources told me that there is a new service coming where they would visit your home, and a professional photographer would set up a lights and a camera, to take photos of the MyHealhtRecord screens you were interested in. This was believed to be easier than developing a very complex software solution for printing - you know - with so many different computers, operating systems, web browsers and such. A technical nightmare!
Although the service will add about $150 million to the budget, use is initially on an opt-in basis. However my source thinks the eventual plan will be to roll it out as an opt-out service, where photographers simply arrive at your doorstep and take photos though the window if you are not home.
very funny. Hopefully one of the major newspapars will pick this up and run with it as "ADHA is a dismal failure"
Post a Comment