Quote Of The Year

Timeless Quotes - Sadly The Late Paul Shetler - "Its not Your Health Record it's a Government Record Of Your Health Information"

or

H. L. Mencken - "For every complex problem there is an answer that is clear, simple, and wrong."

Sunday, September 04, 2011

I Wonder What Is Going On With the Terminology Function At NEHTA?


First we had this appear.
Subject: NEHTA - National Clinical Terminology & Information Service - AMT monthly releases between August and December 2011
Date: Thu, 25 Aug 2011 09:30:17 +1000
From: "NCTIS ServiceDesk" <nctis.servicedesk@nehta.gov.au>
NEHTA - National Clinical Terminology & Information Service - AMT monthly releases between August and December 2011
Since 2007 the NCTIS has been issuing monthly national releases of the Australian Medicines Terminology (AMT) incorporating new items registered with the TGA.
While the AMT has been implemented in clinical systems in Australia, NCTIS has been consulting with software vendors and the terminology community about how we can increase uptake. As a result of this consultation and our recent AMT license holder’s survey we have identified a range of materials that would assist people implementing AMT. These materials will be defined in an AMT Implementation Plan that will be published on NEHTA’s website shortly. The AMT Implementation Plan will set out pathways to implement AMT, including integration guidance and include timeframes for deliverables such as AMT v3.
In order to respond to the market and deliver material requested by implementers over the coming months, NCTIS is pausing AMT monthly releases until December 2011. This decision has come following consultation with healthcare organisations and vendors that have already completed large scale AMT implementations, and have confirmed their operational use of AMT in clinical settings will not be impacted by this action.
We look forward to continuing our work with you to use AMT throughout the healthcare sector to support consistent terminology in medications management. Should you have any questions, comments or feedback, please direct these to terminologies@nehta.gov.au.
Regards,
NCTIS Service Desk
National Clinical Terminology & Information Service
nehta - National E-Health Transition Authority
PO Box 5190
West End, Queensland 4101 
Then a day or so ago we had a follow-up.
Subject: Request for feedback - SNOMED CT-AU
Date:        Thu, 1 Sep 2011 13:00:03 +1000
From:       NCTIS ServiceDesk
To:             undisclosed-recipients:;
Request for feedback on your dependency on SNOMED CT-AU release date
NCTIS released the first national release of SNOMED CT-AU in December 2009 and has been following a six monthly release cycle since then. To support the national Personally Controlled Electronic Health Record, NEHTA is undertaking additional development of detailed clinical models and associated SNOMED CT-AU reference sets.  At the same time, we are also investing resources in the development of AMTv3 which will more closely align with SNOMED CT-AU, and to develop other implementation resources to help vendors and implementers use AMT.
To allocate NEHTA resources to this work, the NCTIS is considering delaying the November 2011 release. We are also investigating aligning Terminology releases with other Specification releases, such as Specialist letters, to provide a more integrated release to implementers. While we need to consult with users of NEHTA’s specifications and terminology to determine the viability of this, we can confirm that the next SNOMED CT-AU will be no later than May 2012.
In order to evaluate the significance this may have we are now writing to license holders to determine whether a delay would have an impact on any clinical application using SNOMED CT-AU. We have already written to all license holders who downloaded any of our last three releases, and have not had received feedback that any clinical application would be impacted, and the results of our recent user survey indicate this is not the case more broadly with other license holders. However, we would like to confirm this directly with you.   
If you have implemented SNOMED CT-AU in a clinical setting and would be impacted by a delay in the November release, please advise us by COB Tuesday 6th September at terminologies@nehta.gov.au or 07 3023 8400.
We look forward to continuing our work with you to use SNOMED CT-AU throughout the healthcare sector to support consistent terminology in clinical management.
Regards,
NCTIS Service Desk
National Clinical Terminology & Information Service
nehta - National E-Health Transition Authority
PO Box 5190
West End, Queensland   4101 
What NEHTA seems to be saying here is that we can’t actually walk and chew gum and so do you mind if we stop chewing so we don’t fall over.
More amazingly it seems after working for at least six years on the Australian Medicines Terminology only now have they discovered a need to actually have an implementation plan!
Surely an organisation that is being funded as well as NEHTA is can continue its ‘business as usual’ functions while undertaking some planning and improved service delivery changes?
It is interesting that Sept 1, 2011 the Health Minister announced some updates to the Pharmaceutical Benefits Scheme Coverage:
See here:

Erbitux, Gilenya and Other Medicines Listed on the PBS

More than 400,000 Australians will benefit from new subsidised medicines from 1 September 2011, including patients suffering multiple sclerosis, cystic fibrosis and various forms of cancer.
.....
In all, twenty-one (21) new listings on the PBS come into effect from 1 September. More detail on these is available on the PBS website at www.pbs.gov.au.
-----
So already the current AMT is way out of date!
Also remember that SNOMED-CT is also meant to be the terminology underpinning the PCEHR - but it is now on hold until the end of the year apparently. I suspect that will need an implementation plan as well!
Bottom line is that under NEHTA’s management we see major initiatives such as clinical terminology simply floundering and not being used as might have been hoped.
(Of course we all know there are a few fundamental issues with SNOMED-CT - I wonder how that fits into this ‘pause’? See here:
I wonder when the powers that be will notice all this is not exactly going as planned?
Soon I hope!
David.

21 comments:

Anonymous said...

David, I would like NEHTA to provide a list of the sites where “the AMT has been implemented in clinical systems in Australia” – by implemented, I mean implemented and currently in routine day-to-day use.

There are two reasons for this:
(1)we would like to consider collaborating with one or more of those sites

(2)despite our best efforts we have been unable to ascertain where the AMT has been implemented and to what extend it is being used

There seems to be no good reason why NEHTA does not make this information publicly available on its web site.

How many different clinical system vendors are using the AMT and in how many sites is it being used?

Can anyone tell me please? The answer, I believe, lies at the heart of NEHTA’s credibility.

Anonymous said...

AMT is a key deliverable for the 'Vendor Panel' members. Based on the timeframes they are working towards, it's likely AMT functionality will start to emerge in clinical systems early next year, at least in the Wave 1 sites.

Anonymous said...

Sunday, September 04, 2011 8:12:00 PM said "it's likely AMT will start to emerge early next year".

So when NEHTA says in the above release "While the AMT has been implemented in clinical systems in Australia... " .... it is distorting the truth.

The correct position from NEHTA therefore should have been "The AMT has yet to be implemented in clinical system in Australia".

Anonymous said...

AMT is another Nehta stuff up. Its not fit for use in a clinical system and is not really SNOMED-CT either, as most of the functionality a terminology should offer is absent. Its use in a clinical system would be a token gesture, as it would have to be mapped to a real terminology to allow the functionality required to allow safe drug use via an EHR. eg Allergy checking is not really possible using AMT alone.

Anonymous said...

If we have to have anonymous comments, can we at least have good quality ones?
- what makes AMT not fit for use?
- what terminology functionality is missing?
- What real terminology would it be mapped to?
- why would it's use be a token gesture?
- what terminology would you use for allergy checking? (do you mean just drug checking or more?)

Another note, while the desktop vendors are "implementing" AMT - probably by mapping the codes they use now to it - and this will be operational, there are other users of it, and they use it now. Someone else should clarify how they use it.

Andrew Patterson said...

I think you will find that any current implementations of AMT in clinical environments are in hospitals that aren't operating on a monthly drug list update cycle anyway (nor would they be particularly interested in additions to the PBS). Hence a lack of updates for 4 months won't affect them.

I don't think AMT has been implemented in any GP or community prescribing clinical system - which is where the release schedule for AMT would most have an impact.

(of course, the PBS not the same as being listed for sale, and so it is quite possible that some of the new drugs on the PBS have already been available in Australia - just not subsidised. Indeed in appears that Erbitux is in the AMT already)

Dr David G More MB PhD said...

Interestingly, a correspondent has pointed out that the planned pause in SNOMED-CT and AMT delivery might have the purpose of permitting a review of where all this is going, what is working and what is not and how the issues (perceived or real)can be addressed.

If indeed the pause is to permit a 'strategic review and rethink' this could only be a good thing IMVHO. No point 'throwing good money after bad' as they say.

David.

Andrew McIntyre said...

I would agree that AMT lacks the semantics to function in a way that allows for proper allergy checking, short of manually flagging every drug that is a type of penicillin, for example.

If you look at RXNorm in the US its done much better and even has free rest based APIs to use if you want to. In the US they have added semantics to allow for meaningfull use, where AMT has removed 90% of the Semantics that were in the SNOMED drug hierarchy which precludes meaningfull use. Its very frustrating, and I have tried to fix the issues with Nehta but its hard to get any traction with Pencil Necks.

Anonymous said...

Every penicillin maps down to one of 8 concepts (procaine penicillin, benzylpenicillin sodium, benzathine penicillin, phenoxymethylpenicillin potassium, benzylpenicillin, phenoxymethylpenicillin, phenoxymethylpenicillin benzathine, benzathine benzylpenicillin). But those 8 are not linked to a single concept for penicillin. That's certainly a lacking semantic, but it's not like you have to flag every drug.

btw, it's very hard to get traction with people when you call them pencil necks in public.

Dr David G More MB PhD said...

"btw, it's very hard to get traction with people when you call them pencil necks in public."

I think that just might be a reflection of extreme frustration - but then how would one know. I know for certain being nice has not often been successful!

David.

Eric Browne said...

Anonymous said: "Every penicillin maps down to one of 8 concepts". Could you please explain the relevance to Andrew McIntyre's post.
The AMT supports no mechanism for determining if a given ingredient of a product (Trade or medicinal) is a penicillin. e.g. amoxycillin contains amoxycillin is as good as it gets. An additional linked knowledge source would be needed. This is well known; has been known from the very outset; was known by the Medicines Coding Council of Australia that predated NEHTA. What we don't know is the path and timetable by which NEHTA proposes this linkage be met. Moreover, documentation regarding the AMT talks of using Australian Preferred Names, and of providing synonyms, neither of which have ever been true of any release of the AMT to date.
e.g. The Australian Preferred Name is 'amoxicillin'. The International Non-proprietary Name is 'amoxicillin'. The AMT has no entry 'amoxicillin'. To my knowledge, the AMT provides no synonyms.
The same Anonymous also wrote "btw, it's very hard to get traction with people when you call them pencil necks in public".
Well I'd suggest that it's very hard to get traction when you try hard! In fact, I'd suggest it's nigh on impossible to get traction full stop! I totally concur with David when he suggests that it "just might be a reflection of extreme frustration - but then how would one know."

Anonymous said...

On NEHTAs website, search for "Victoria AMT". You will find 2 press releases. AMT has been implemented within HealthSMART clinical system (Cerner Millennium) for Acute Hospitals.

A core component of prescribing within Cerner Millennium is the Multum drug information that enables appropriate medication interaction (drug-drug, drug-allergy, drug-disease) and therapeutic duplication checking. AMT is embedded in the Multum catalogue.
The medications component of HealthSMARTs clinical system is being used to support outpatient and discharge prescribing at this point in time (inpatient prescribing is next).

The HealthSMART website clearly tells you which hospitals / health services are live with the clinical system.
http://www.health.vic.gov.au/healthsmart/clinsysupdate.htm

As for the impact, if AMT delays releases then Multum does not update the data either. Therefore any temporary code additions, deletions and modifications will need to be performed manually until Cerner Multum and NEHTA AMT return to sync. Saying that, temporary codes are required now as AMT is not all encompassing, but this will add some manual effort to the HealthSMART maintenance process (eg. 21 PBS additions for Sep that will not be in Multum until released by AMT).

Eric Browne said...

Anonymous said of the HealthSMART AMT implementation "Therefore any temporary code additions, deletions and modifications will need to be performed manually until Cerner Multum and NEHTA AMT return to sync".

Two points:
1. What evidence is there that Cerner Multum and NEHTA AMT have ever been in sync?
2. Why would HealthSMART be pumping out identifiers for medications based on AMT, in HL7 messages when no recipient system is currently using those identifiers? Surely the names of the prescribed discharge medications are far more important?

Grahame Grieve said...

> The AMT supports no mechanism for
> determining if a given ingredient
> of a product (Trade or medicinal) is a penicillin

That's not entirely true. Or maybe I should ask whether it's entirely untrue. Let's take a containered product pack:
------------
19917011000036103: LPV (phenoxymethylpenicillin (as potassium) 500 mg) capsule: hard, 50 capsules, blister pack (containered trade product pack)
is-a
13162011000036100: LPV (phenoxymethylpenicillin (as potassium) 500 mg) capsule: hard, 50 capsules (trade product pack)
is-a
27621011000036105: phenoxymethylpenicillin 500 mg capsule, 50 capsules (medicinal product pack)
is-a
21370011000036105: phenoxymethylpenicillin (medicinal product)
has-ingredient
2473011000036107: phenoxymethylpenicillin (substance) -----------------
If this isn't "a mechanism for determining if a given ingredient of a product (Trade or medicinal) is a penicillin", then what do you want?

Andrew Patterson said...

Grahame, as I understand it, the complaint is that because phenoxymethylpenicillin is an 'AU substance' and not the actual international SNOMED phenoxymethylpenicillin concept (i.e. the AMT is a self contained terminology set) - we miss out on the upper part of the substance hierarchy that goes


372725003|Penicillin V (substance)|
is-a
372598006|Penicillinase sensitive penicillins (substance)|
is-a
373270004|Penicillin -class of antibiotic- (substance)|
is-a
373297006|Beta-lactam antibiotic (substance)|

These upper concepts could be used to record allergies in a more general way (e.g. 294492003|
Penicillinase-sensitive penicillins allergy (disorder)|) and have that encompass all similarly structured penicillins.

And that's fine and dandy, and I totally understand the use case, but I think describing that as making AMT 'not being fit for use' is a tad strong.

So I would classify the complaint as not entirely untrue - but perhaps unfair to single out this one use case as a deal-breaker for adoption of the AMT.

Andrew McIntyre said...

Hi Graham,

Its a totally inadequate mechanism because there is no substance hierarchy to compute the result using the terminology. I should be able to say Beta lactam allergy and pick up cephalothin as well but you just can't.

Compare AMT to the real SNOMED-CT relationships and you will realise that AMT is totally inadequate for decision support.

I have tried hard to get traction from Nehta on this one but hit a brick wall. The people also keep changing and it seems the terminology employment plan at Nehta is a revolving door. I wonder why?

You cannot possibly build a Medicines Terminology without Decision Support use cases, but this is what has been done and it shows.

This just needs to be added to long list of failures I am afraid. I have given up hoping that it will improve, Nehta is a dead man walking.

Eric Browne said...

Grahame Grieve wrote:
"If this isn't "a mechanism for determining if a given ingredient of a product (Trade or medicinal) is a penicillin", then what do you want?"

Well I don't think so.
a. Who is to say that "phenoxymethylpenicillin" is a "penicillin"? Simply because it is a superstring?
b. It is a totally unreliable mechanism. I cited an example "amoxycillin" where the AMT has no associations to "penicillin"!

That's why people have long wanted, and NEHTA has promised, that the AMT is to be linked to the SNOMED substance hierarchy. The AMT currently has no substance hierarchy. Ergo, minimal support for allergy/adverse reaction checking or a whole raft other useful functionality. This goes to the very rationale for using a clinical terminology of the complexity of SNOMED.

Grahame Grieve said...

#Eric and Andrew M

The specific complaints "AMT should extend the snomed-CT hierarchy" (which would be nice) or "AMT doesn't relate penicillin and amoxycillin anyway" are rather different from "The AMT supports no mechanism for determining if a given ingredient of a product (Trade or medicinal) is a penicillin". btw, I haven't explored RxNorm or the UK Snomed - how do they solve this problem?

Eric Browne said...

#Grahame

I stand by my assertion that the AMT supports no mechanism for determining if a given ingredient of a product is a penicillin.

Things are even worse than that. Several days ago NEHTA published a number of Detailed Clinical Models, including one for Adverse Reaction, in which, on page 18, the value set to be used ( I assume in the PCEHR) for the agent to which one has had an adverse reaction, is specified. It should either be a product from the AMT or a substance from the SNOMED-based Reference set. In other words, one is forced to reference the SNOMED substance terms, which don't correspond to the substance list in the AMT!! So if the PCEHR has a list of medications based on the AMT, then their ingredients can't be reliably and simply matched against substances recorded in adverse reactions according to NEHTA's own specifications.
Again, I use the example "amoxicillin" vs "amoxycillin". How big an issue this might be I don't know. But then again, who would know?

Grahame Grieve said...

#Eric. I think it's more accurate to say that the mechanism exists but is incompletely implemented.
Obviously the outcome - having two substance lists that are not mapped - is not good. After digging around, I find that certainly wasn't the original intent - The problem is that the Snomed substance lists was believed to unstable, and this has not been resolved (as far as I can tell)

Andrew McIntyre said...

Graham,

Its most accurate to say that compared to the SNOMED-CT core AMT lacks the functionality to make it useful for decision support. The allergy question is one serious shortcoming, but the drugs themselves are in quite a flat list as well with no upper level hierarchy. SMOMED-CT is about computability and AMT is devoid of this ability, apart from generic <-> trade name mappings.

You can choose to call it "incompletely implemented" if you like but it is "non-functional" which is a huge problem. If you cannot compute that a penicillin allergy raises an alert when amoxil is prescribed it makes a mockery of Nehtas own slides, which actually use that as an example!!!!