Agencies like the Commonwealth’s Australian Digital Health Agency (which has just lost it CEO in unexplained circumstances) have a mantra that essentially suggests that if we can get full adoption and use of Digital Health tools by clinical professionals (doctors nurses and the like) that we will suddenly arrive at some sort of clinical Nirvana where there are dramatic improvements in the quality, safely and connectedness of care delivery all of which will be to the ultimate benefit of the patient and their lives.
They point out that all sorts of major transformations and improvements have been seen in areas from banking to logistics and transport and it can thus only be a matter of time before the same improvements in quality and efficiency will be seen in healthcare.
If this were true then surely, by now, we would have working exemplars where most of the problems have been definitively been solved and all that is left is to dot the ‘i’s and cross the ‘t’s and roll out a highly successful model globally.
That process is also yet to begin because, to be frank, no one has got there yet!
If we look around the world we see all sorts of things happening and we see evidence of real benefit on a significant scale from Digital Health to be rather hard to find.
In the U.S. we see harm, clinical errors and clinician burnout and frustration on an unexpected scale and we see consumer frustration with the lack of access they have to their own clinical information. Worse they often find the different clinicians they visit cannot access their records from other clinicians leading to frustration at the least and real harm at worst.
Interestingly some smaller Scandinavian countries, Scotland and New Zealand have made more progress but are still working on issues while it seems clear that what they are doing does not scale well as England continues to struggle after near 20 years of effort.
This brings us to Australia. We need to break Australia into three areas.
In the Public Hospital Sector we have had decade long processes to get basic clinical and administrative systems in place with some sort of success in NSW and Victoria with considerably less success in all the other States. This has been associated with considerable clinician annoyance and alienation. The private Hospital Sector is about 5 years behind the public system.
In the GP sector – and to a much lesser extent for specialists – basic record keeping, billing and other admin functions are commonly computerised with some communication functions also in place at a practice level. Inter-practice information flows are still very limited.
Australia also has a one-of-its-kind health information repository – the My Health Record – which collects all sorts of clinical and administrative information into a huge document repository. The thought is that this repository will be looked up on the off chance it has some useful information and that the time spent looking it up will be repaid. To date there is no evidence I have seen that this is actually happening. Despite just on 90% of the population having a record at least half are still empty and many others contain old and ageing data as best one can find out. No usage statistics of any value are published by the Government, but it is known that while automatic contributions to the repository continue to mount the same cannot be said for clinician curated content. The system has failed in over six years of operation to attract much clinician usage or interest.
Importantly, as far as I know, there are anecdotes but no compelling evidence that the investment and work so far has been worth it.
It would be fair to say that globally, advanced economies struggle to progress Digital Health initiatives and that issues that were widely discussed a decade are still seeking solution. In developing countries some basics have made a little progress and need to be fostered to come to richer fruition.
Having watched on since 1987 I have to say I have concluded a number of things.
1. Successful computerised support of clinical practice is very much harder than was imagined 30 years ago and is still being worked on.
2. Small scale initiatives can often work very well but seem very rarely to scale.
3. The track record of large centralised government solutions is abysmal – none have really succeeded.
4. Unless clinicians see value in the use of technology for themselves and their patients adoption is an extremely hard slog without significant financial incentives – as have been used in Australia.
5.Trying to turn skilled clinicians into data entry clerks does not end well.
6. Simplistic thinking about what will and won’t work for clinicians and patients seems always to reveal the problem being address is very much harder than it seemed initially.
So, in summary we are not there yet, the problem of obtaining real value from Digital Health is a truly wicked one but maybe, just maybe the next five years will be different. Some think so and I guess time will tell.
David.
Same story; nothing changes.
ReplyDeleteIn your December 11 blog titled - “The Tim Kelsey Legacy – Has Been A Failure For Australian Digital Health.” Tom Bowden @5.36 AM suggested that “we need to look at the evidence of what does and does not work around the world, at what technology is doing and how well our citizenry are likely accept some sharing of their personal data and come up with another plan entirely.”
He then asked if “Anyone was keen to work together on this?”
I wonder what response he got.
I wonder too whether, and if so how, he expects such an exercise to be funded or is he asking for unpaid volunteers? !!!!!
Well said David. Digital health has seen some excellent innovative developments since 1987, but sadly these have not been always been taken up by the people who have the power to do so, because of their lack of knowledge/experience, unwillingness to listen to experts, being driven by political deadlines, and the influence of large consulting firms. And yet if Australia had embraced these innovations (many of them home-grown!), some of the issues you have raised may have been addressed.
ReplyDeleteFor example, the Australian SNOMED CT and Australian Medicines Terminologies are fundamental to enabling interoperability and meaningful analytics, but the My Health Record system does not yet use this capability. Thus it drags all of its connecting systems down to its lowest common denominator. CSIRO has now taken the lead on the terminology server and this is already delivering benefits elsewhere than Australia. Also HL7’s FHIR standards work – developed with Australian expertise, experience and knowledge (with its beginnings in the now disenfranchised Australian standards development community) is not being taken up in Australia with the same level of enthusiasm that it is internationally.
Decades ago, Australia was active in the development of the openEHR standard for how health data is structured and defined, and this work has gone on to develop European health informatics standards and systems - but was virtually ignored here in Australia.
None of these technologies have been used in the architecture of the My Health Record – a missed opportunity to set an example and lead national (and international) developments.
Let’s hope that it isn’t too late to change.
The Department has given no indication it understands. The ADHA Board has demonstrated its new Directors have been powerless.
ReplyDeleteSNAFU.
ReplyDeleteI think the "addition" of government $$ has been the problem all along. eHealth was flying along in Australia with the initial invention of PIT, followed by adoption of HL7V2 with active participation of laboratory IT people and local standards....
ReplyDeleteBut then we had healthconnect who said no..no we are going to invent something new and the NEHTA I and II who did the same, and the ADHA who just did public relations as far as I can see?
We have had working PKI, working individual PKI signing, working standards processes, working message testing and vendor/lab engagement but all that has been blown away by inept government national eHealth authorities none of whom had a clue what they were doing, but everyone was paid to do it their way or the highway.
They have driven us off the highway, into the back blocks, pushed us out of the bus and are in the process of driving away as fast as possible. The $$ have all dissipated into consulting firms or multinational corporations... remember the $20 million IBM made off with for not doing NASH? That is the tip of the iceberg.
The problem is incompetent bureaucrats, who have no idea what good governance is. Small governance around standards compliance for patient safety around the year 2000 would have resulted in a much different landscape. The government is inept and they just need to exit the scene and let technical people try and find a way back to civilization from where we are out in the weeds.
Clinicians will pay for useful software, and any software that interacts with the outside world should be tested and standards compliant for the sake of public safety. We need to redevelop technical ability within the eHealth community, rather than promote used car salesman who can extract $$ from taxpayers for no return. Its a mess and the mess has been engineered by the hubris of the bureaucracy. Its time they left the building.
Andrew nailed it when he said "the government is inept and just needs to leave the scene".
DeleteThat unfortunately will not happen. The cycle will once more be repeated under a new CEO. Winding down the ADHA and starting again is too big an embarrassment, politically.
Those of us with the insight and ability to bring new thinking to the fore are unjustly feared by those in government, who more than anything else should be reaching out to genuinely and honestly embrace the new thinkers with open arms.
It's not difficult to do and it's not something to fear. It's the only way they can step forward to everyone's benefit.
Many have tried, one way or another and not been heard or welcomed. It will always be so. As you and Andrew have said ... they just need to leave the scene and get out of the way.
DeleteGreetings, Tom Bowden here.
ReplyDeleteResponding to commentators above - General comment: it'd be great to lose the cloak of anonymity so we know who we are talking to. Any chance of doing that?
Andrew McIntyre: I totally agree with your comments apart from your reverence for PIT. That was a blatant (and highly successful) attempt by the major path labs' oligopoly to derail commercial messaging services and tightly control their GP referral bases -something that has cost the taxpayer dearly and held up widespread development of pan-sector messaging. Though there is no point in crying over spilt milk. Otherwise agree on absolutely everything.
Annonymous at 9.15pm - I think the ADHA board has been quite forceful, whilst knowing few details, I understand they have cleared out the Augean stables quite effectively. Big question is - where to from here?
Annonymous @ 5.48pm - I have had no response, which worries me. The participants in this blog are some of very few people that have a reasonable understanding of this stufff and we need to put forward a plan to replace the current MHR before efforts to carry it on are put in place. We need to lobby the ADHA Board and government. But first we need an alternative plan.
We can't just sit here criticising. The time for doing that is long past. We have won the argument. It is not working. Let's move on.
We must propose an alternative strategy and I am quite keen to participate in doing that or even to lead it.
Right now (today/tomorrow) I have a paper to submit on a case study of shared electronic records in ED; 5 years of Clinicians looking at GP records but by early next week I should be clear of that and will post an alternative view on this Blog of how we should be sharing records.
Let's have a really thoughtful and vigorous debate and see if we, the disempowered, can rise up and seize the wheel.It's our turn.
Will be in touch,
Tom
I suspect this blog is the last remaining forum for those that might have the history of what has worked and what hasn't to share ideas. Congratulations on your contribution David and to the community who continue to try to do good in this domain.
ReplyDeleteBy way of setting the record straight and picking up on Tom's comment re PIT then I offer a little history. The first electronic pathology reports in Australia were sent in 1969 by Preventicare on IBM CAll 360 and later by Medicheck and the Shepherd Foundation. PIT was 'invented' by Sullivan Nicolaides and QML in 1993 for modem to modem transfer - well before there were 'commercial messaging services'. In 1996 all pathology labs, public and private including SNP & QML, agreed that PIT should not be the Australian Standard and that HL7v2 should be used and we set about producing what is now called a localisation in what was then a working Standards Australia. Despite this PIT continued to be used because it was adopted by the then formative Clinical Information Systems and it (almost) did the job. The take home message from this is that it is really hard to change direction once something is adopted without the right drivers.
Supporting Andrew's commentary and contrary to Tom's view of 'the pathology oligopoly' costing the taxpayer, in 2006, Australia came tantalizingly close to getting standardised reliable interoperability of pathology reporting. With the support of all public and private pathology practices together with the College of Pathologists a proposal was worked up, fully costed and presented to Government for a universal pathology messaging platform for a cost that turns out to be less than 1% of what it has has cost to date for the myHR. That proposal was rejected. Lobbying from those whose interests were most threatened, commercial messaging vendors, may just have played a part.
Like Andrew and David, I believe the way forward is to focus on getting the fundamentals of existing clinical communications right. If we were to have interoperability of pathology, radiology and medications we would have a much better foundation for digital health. Because the little evidence that we have suggests that most harm is done by diagnostic and medication error, we would also be focusing on what matters to patients and the system. This is the same view that I have been putting in National forums and to anyone else who will listen for three decades now and it is why I continue to beaver away on what is needed to do this at the grass roots.
I believe a lot of good has been done by 50 years of electronic transmission of pathology reports but we don't have the research to prove it one way or the other. We should have that evidence and so another thing I would do if I had control would be to fund the research. This is a place in my view where Government has an important and legitimate role to play!
@4:05 PM Quite frankly if I was still working at ADHA I would be looking for a new job rather than hanging around to see what might happen. Best to get out ahead of the rush.
ReplyDeleteResponse to Michael Legg
ReplyDeleteWhile I do not wish to head down a rabbit hole to endlessly debate how virtuous or otherwise the large pathology companies are, I would neverthess like to make a few points:
Pathology services cost the Australian taxpayer 50-100% more per capita than in other OECD countries. This has been the case for most of the past 20 years.
The Phillip Fox review into Australia’s Pathology Services found that: “in the course of this review pathology industry representatives have stated that they receive reports of cash payments and gifts of expensive motor vehicles being made in return for redirection of referral streams”
Placement of collection rooms at inflated rentals…Major political party donors... the list goes on. It is not hard to work out why private Pathology services cost Australian taxpayers dearly.
In terms of technology: Countries that have solved the messaging question (Denmark, Holland, New Zealand) have largely done so by using pathology messaging as the basis of a range of messaging applications. The Australian pathology companies have not allowed this to occur, instead they have set up their own private point to point messaging systems. This is legitimate business practice but has had a very detrimental effect on messaging uptake overall.
Purchasing of shares in GP systems – until recently each of the two major lab companies in Australia owned shares in each of the two major GP systems.
I am not up-to-date with things but until recently at least very little effort was made to ensure full standardisation of messaging – certainly not to the extent that any truly independent party would ensure happened and I note that NATA’s recommendation that the labs have their messaging outputs validated independently on an annual basis was consistently ignored.
Happy to hear your response Michael but then let’s get back to designing a shared record system that will actually work.
Correction:
ReplyDeleteThat should have been:
... until recently each of the two major lab companies in Australia owned shares in one of the two major GP systems.
Mmmmm easily said - "designing a shared record system that will actually work". Who will fund this work? Please don't say government. Too unreliable. Please don't say volunteers. Well meaning, genuine, motivated and committed, though they may be is just not enough.
ReplyDeleteGovernment funds pilots and then discontinues funding at its whim. Government meddles, bureaucrats change all too frequently, bureaucrats have limited knowledge and understanding of the issues involved and their complexity. Their egos disrupt and derail, and government agendas conflict with the overall goal of a shared record system.
Andrew @11.22 PM is not alone in his view that government "just need to leave the scene and get out of the way". This is the first of many hurdles to be overcome before rushing into the sandstorm ahead.
I agree. First step is government must step aside completely. And if, or when, they do will someone please tell me - Who will fund it?
DeleteI was once given some advice from a mentor to entrepreneurs. Start with something small but which meets a need and/or delivers value. e.g facebook, amazon, google
ReplyDeleteThen grow the idea, always monitoring each feature/benefit.
A big bang approach - boil the ocean - very rarely works.
In the case of health data, it would have been a good idea to develop/promote mechanisms/standards that allow existing clinical systems to be accessed by other healthcare providers via a portal. The mechanism should be health data agnostic. After the care.data debacle, that's exactly the strategy in the UK.
The equivalent of today's health record interoperability problem is that it is seen as two problems where both need to be solved - technical and symantic communication
It's like building a voice phone system that does not just connect two handsets but can translate the conversation between two different languages.
There are systems available to commercial enterprises such that when you call them up, you get a series of menu options that facilitate you paying a bill, querying an account balance, etc or speaking to a particular department. The enterprise probably has a web portal equivalent.
Put the smarts in the endpoints, not in the infrastructure.
The medical problem is access to and use of relevant information, not the exchange, accumulation and management of mostly historical information.
Once you have access to the right information, you can then acquire it and use it.
ReplyDeleteThe first step is to answer the question; Which problems need solving most?
"The first step is to answer the question; Which problems need solving most?"
ReplyDeleteHere's a candidate:
The reduction in diagnostic error, as discussed in a report in 2015
Improving Diagnosis in Health Care.
National Academies of Sciences, Engineering, and Medicine 2015.
Washington, DC: The National Academies Press.
https://doi.org/10.17226/21794.
or how about addressing problems with evidence based medicine. The British Medical Journal has a good primer:
Evidence based medicine: a movement in crisis?
http://www.bmj.com/content/348/bmj.g3725
If technology can't be used to help address a problem that the medical profession perceives as important, then it's probably not a problem worth solving.
So Tom @ 8:30 AM said "let's get back to designing a shared record system that will actually work".
ReplyDeleteThen @5:28PM he asked "which problems need solving the most."
So he has moved from designing a shared record system to identifying the problems that need to be solved.
This sounds like he's playing shove ha'penny or parlour darts. Get focussed, start with a blank canvass, stop suggesting solutions. List the issues.
Issue 1. Sharing relevant clinical information about an individual with and between the individual's health care providers.
ReplyDeleteIssue 2. Ensuring an individual can have an accurate, current, summary record of their essential health information when required which can always be trusted and relied upon by the individuals health care providers.
ReplyDeleteIssue 3. Health information; absolutely essential, non-essential, maybe helpful, maybe unhelpful, may confuse, may overload, may hide the relevant, maybe too sensitive and too private, maybe impracticable.
ReplyDeleteBernard Robertson-Dunn said... "Put the smarts in the endpoints, not in the infrastructure."
ReplyDeleteThis is the crux of the issue with the governments view of messaging interoperability. Currently there are a lot of smarts in the messaging layer, to make it work because of compliance issues wrt both the message creation and the handling/display of compliant messages.
This makes messaging system interoperability virtually impossible as the incidence of errors at endpoints would skyrocket unless you enforced the simplest dumbed down messages, and this is what the ADHA have done to try and make MyHR work, converting atomic data into pdfs. There are already systems using the atomic data that exists and they are not going to accept this, and there is no way it should be allowed as it defeats the reason behind eHealth. If you want that, buy a fax, rather than axe them.
However there are issues with systems of different capabilities and this is the issue Michael Legg alluded to - Systems implemented PIT and are stuck on it as its actually not bad compared with pdf.
This is also the issue wrt "Moving to something new" which is the common solution proposed when systems have to support PIT, HL7V2 and CDA and now FHIR for some things. Unless you do a wholesale move to something new it just increases the number of formats people have to handle and actually dilutes the quality of the support for any one format. A wholesale move to something new is unlikely to work with a huge number of systems plugged into the current formats and little appetite for doing this type of development.
Its these considerations that have been baked into the current Australian HL7V2 standards. ie.
1. Messages should have a text based display format which could be converted to PIT if need be.
2. You can send richer display formats like PDF but should also have the same data as atomic text based data, and ideally a text based display format as well.
3. The richer Semantic atomic content uses Archetypes in V2 which is an extension that allows rich clinical models but the message remains compliant and should render correctly to systems that are oblivious to the semantic content.
This combination allows support for very old PIT based systems, but allows for rich display and complex clinical models can be developed without systems having to understand them, just display the data according to the display rules. Anyone can develop a clinical model and anyone can choose to use or ignore it.
Essentially this results in a cleaner layered design:
1. Messaging deals with privacy, authentication and routing by reading a compliant message to get addressing details, message type etc
2. HL7V2 carries data that can be rendered by systems for human display, but also carries atomic data values for machine use
3. A clinical modelling layer allows rich models (eg The cancer reporting models) to be developed without changing the HL7V2 standard and these models are transparent to systems that don't understand them.
To make this work we need:
1. A robust PKI in place that can be trusted and easily renewed in software and includes individual keys for signing
2. Tested Standards compliance for HL7V2 covering both message construction and rendering/handling of messages by endpoints along with automated production of both messages and acks. Systems should be required to be able to produce and consume HL7V2 files as a standard feature, and not locked up in commercial deals
3. A format agnostic clinical modelling effort to capture and document what data should be collected with a general ability to downgrade reporting to text if need be. These models data instances can then be translated into HL7V2 automatically (or FHIR etc)
The 3 layers have well defined interfaces and each can be tested and developed within their layers without requiring a lot of coordination with the other layers.
These concepts have been built into current HL7 Australia standards, but the ADHA probably don't get that?
Andrew, I think we are in agreement about the approach.
ReplyDeleteHowever, an observation.
You may not have been around during the OSI/GOSIP vs TCP/IP wars on the 1980-1990s
The OSI model is a well engineered layered model, TCP/IP can only be described as a Nike model - just do it - a "make technology work" model.
Well engineered systems take longer to architect/design/implement but there's a good reason for that - they are usually very complex problems.
We now have a dystopian internet world where the stupidities of "just do it" have come home to roost.
Governments and technologists are incapable of understanding the complexities of certain problems and want solutions ASAP.
The architected, layered solutions are essential but not acceptable to those who have the money and to those who think they know the solution.
The only thing the ADHA got was the "if everyone has a myhr, it will get used" message. They've done everything they can to register Australians, they've now run out of options. Hence the exodus.
Issue 4. Prioritise multiple vendors, multiple applications, multiple domains, multiple stakeholders, multiple service providers.
ReplyDeleteIssue 5. Any contributions?
ReplyDeleteI understanding funding models are a recurring theme that need addressing. Proper adoption and implementation of the HI service and there are some very useful nuggets in a flying blind publication from a few years ago and AHHA blueprint from two years ago is also worth looking at.
ReplyDeleteYes government has a roll, standards and legislation, both critical enablers of innovation and interoperability. The federal government should not build systems and stay very far away from UX and UI.
Happy to formalise a simple contribution later in the week
Issue 5. Prioritise and weight applications by usage, benefit, relevance, complexity, workflow impact.
ReplyDeleteIssue 6. List, prioritise and compare, what clinicians and patients need and what they think they want.
ReplyDeleteIssue 7. Define User Interface design tool funtionality requirements and features, evaluate existing tools and identify their design constraints
ReplyDeleteJust an observation:
ReplyDeleteAn issue isn't a problem. You need to be able to express a problem in terms of a question that you can then attempt to answer. You can't answer an issue.
Issue 8: I think part of the problem is people trying solve the problem without having a deep understanding of the problem domain. Health workers know what useful software is and people have created standards to help them solve problems they understand.
ReplyDeleteLeaving end users to decide whats useful to them and when data is communicated outside the software application or comes in, it must be compliant with any standards in use and must be displayed correctly. Imposing solutions from a position of ignorance is the problem. Governance is saying you will comply with standards once data leaves/enters your application and providing some infrastructure such as PKI and Terminology/identifiers. Anything more than that is the cause of the failures in my view.
@4:55 PM OK. Is this question a suitable place to start?
ReplyDeleteQuestion 1. Is there a need to be able to share relevant clinical information about an individual with and between the individual's health care providers? If not, why not? If yes how can that be done?
8:29pm
ReplyDeleteBefore you ask that question you need to define "relevant clinical information about an individual"
Relevance, currency, redundant, obsolete, trivial— all very good data questions. As the question was ‘share’ rather than ‘send or exchange’ the attention changes to how to ensure data is made discoverable, maintained and how then to tackle duplication.
ReplyDeleteOr is it that some are focused on maintaining a nation level view for a localised need, and maintain a centralised database is the answer to a distributed problem?
@10:42 PM & 8:07 AM. Do not make the .mistake of expanding the scope of the question. Stay focussed.
ReplyDelete8:07 AM's comments need to be dealt with later, not now.
The first question has now become:
1. What is relevant clinical information?
And the second question should be:
ReplyDeleteQ2. What is the most essential problem you are trying to solve?
Agree 11:22.
ReplyDelete