11 May 2023

On another $429m for a self-declared data repository

Government Political

If you really wanted people to believe the My Health Record could somehow morph into a real-time data-sharing platform you probably shouldn’t have rebranded it as a “national repository platform”.

There are lots of positive noises wafting out of various parts of the Canberra health reform ecosystem to indicate that the federal government is starting to get its head around just how far off the mark we are in terms of a digital health infrastructure that would be able to underpin much needed system reform.

Given that the budget seems to have signalled that the government is much more serious about long-term systemic health system reform, we suddenly don’t have much time to get our act together on building digital health infrastructure that can underpin such reform.

The bottom line for such infrastructure would be that we could actually share meaningful data seamlessly across the system and with patients in real time.

Health Minister Mark Butler gave the MHR project a bollocking before the budget by saying variously it was a clunky, out-of-date system that still doesn’t do what it was designed to do – share data in real time – after throwing a mass of money at it and therefore needed a full overhaul.

It’s a good start if Mr Butler understands that the MHR is, so far, a pretty big fail and we need to change this dynamic quickly if we are going to move on broader systemic reform.

But if anyone reads the words in the budget that relate to how we might go about rapidly changing this picture, it feels like we all still have a lot to worry about.

For starters, the budget describes how the next phase for the MHR will be the creation of a new national repository platform to support “easier, more secure data sharing across all healthcare settings”.

What exactly is a national repository platform?

I’m not sure who is in charge of marketing in the Department of Health and Aged Care but relabelling the MHR as a “national repository” isn’t a great way convince people that you are going to open the floodgates on seamless real-time data sharing, even if you add the word “platform” after it.

Repository is defined in the Webster Dictionary as “a place or container where something is deposited or stored; one that contains or stores something immaterial”.

Ouch, they should have looked that up before headlining it as how we should underpin the future of a digital health strategy focussed on real time seamless data sharing.

If you are going to come out and say the MHR is a giant fail in the manner Butler has, surely that failure should induce some changes in how you’ve been doing things.

But if you read what detail there is on the MHR in the budget – there is very little detail — two things stand out:

  • We aren’t going to make any meaningful changes to the plan that got us into this mess, nor to who got us into it (or, perhaps more fairly, who didn’t get us out of it, as it would be hard to blame the current administration in digital health for getting us into this mess);
  • The MHR, which is a centralised hub-and-spoke database model of sharing (something which most other countries abandoned as the logical model of real-time data sharing in healthcare by 2015), is somehow still the centrepiece of how we intend to share data moving forward in Australia.

How do you change our 20-year-old, go-nowhere trajectory if you don’t actually change anything?

What we see in the budget is a plan to persist with the MHR as the centrepiece of our data-sharing efforts, and that plan and its execution are all coming out of the Australian Digital Health Agency. All business as usual.

Nothing to see here, apparently. Things have been going so well, we’re sticking with the team and the plan.

Mr Butler surely can’t have it both ways. The MHR is a mess that we need to spend a further $429m over the next two years to fix apparently (after already spending $2.2 billion and getting nowhere), but we are okay with sticking with the people, plan and organisation that hasn’t made even the smallest dint in problem over the last three and half years?

Those same people seem to want to now triple down on a centralised database model of sharing that no country which has been successful moving the dial on interoperability uses any more.

Does Mr Butler, who has openly admitted the whole thing is a fail, really accept that the ADHA is functional and has a good plan we should all still have faith in?

Read this National Digital Health Strategy from the Agency, produced around 2019 and what you will see is that every single major objective has been missed by a very wide margin. Check out the sections in orange under each major objective of “What will be delivered by 2022”.

In fairness, the current administration didn’t set these targets, but after more than three years having a go, they still managed to miss them all.

Should we then be alarmed that what is being suggested to fix interoperability and create real-time effective data sharing in the system is pretty much BAU at the ADHA with a bit of a tweak – let’s now try to make the central data repository we’ve created more accessible by building some cool new stuff around it to make data flow in and out better?

To be a broken record about this, even if we do build lots of cool FHIR-based APIs around our prized centralised database, the fundamentals of efficient data flow in a modern distributed web-enabled ecosystem are still being bypassed in a big way here.

Here’s how (yet again):

  • If you insist on a giant centralised “repository” you insist that all points of the system have to take the time and effort to get the data out of their system and into the repository first. Then, when someone needs that data, they have to get it back out of the repository, in a form that is meaningful for what they want to do.
  • In this model you don’t know exactly what any other part of the system is going to need, so you have to insist that every bit of data that is generated everywhere in the system that might be meaningful is sent to the middle, and organised enough in the centre so it isn’t too confusing when someone wants to pick some of it out later. This creates the need for massive, costly infrastructure and logistics and also a giant co-ordination issue for the government.
  • In a distributed model of data sharing, as occurs in the US, Israel, Denmark and other advanced countries, meaningful data flows directly from one point of the system (say a database in a GP’s patient management system) to another, or a patient, when it is needed in real time.
  • You don’t have to be good at maths to work out that if you cut out the big centralised database in the middle of the centralised model, more than 50% of the chain of data flow is not needed. That avoids all the cost of building and running a giant centralised honeypot of data that will never have all the real-time data that exists in such a complex system anyway because there is a lot of latency in having to send data to the middle all the time and get it back.
  • There are also massive issues associated with polling such a big database to find what you actually need, especially if you need it quickly and without any error. In the distributed system the data is polled directly from one part of the system which has the data live in a small and discrete set (say a GP PMS system), and sends it to the other part (say a hospital EMR across the country, or a local allied health provider), thus avoiding the whole middle bit thing.
  • How hard is that to understand?

It’s of course feasible that beneath the political waterline, the people who matter now understand the above distinction between the technology and data-sharing models and that somehow we need to change course a lot more than the budget is indicating we intend too.

If that’s the case the MHR isn’t going to suddenly disappear. It’s a useful database and will fulfil some useful functions moving forward in what will be a long transition to a properly functional data sharing ecosystem.

But in this scenario the MHR won’t be the centrepiece of moving our system to meaningful real-time data sharing at the end of the day.

At last week’s Wild Health Leaders Summit in Canberra, one of the interoperability panelists had an intriguing way of thinking about this possible problem for the powers that be.

This panelist did not profess to know anything about what was actually going on politically beneath the waterline on the MHR when questioned about it, but he did understand the serious structural issues of the MHR data-sharing model, and posited a possible way things might move forward given the tricky political and people issues at play.

He said that maybe we should be thinking of the MHR as brand, not as an actual thing. And that the brand could be, say, “a patient-centric healthcare platform powered by data liquidity”.

In that brand both a modern distributed model can be built and the old MHR can still live.

The idea goes like this:

The objective of the MHR was always good. It was, in essence, patient empowerment over their health data through creating data liquidity – the ability of meaningful data to move seamlessly across the system in real time.

The product and the execution was never right.

So, stick with the objective, and the “political brand”, but change the execution and the product over time.

It’s a bit “don’t rock the boat just make your way to the wheelhouse and quietly start steering it in a completely new direction”.

Maybe this is what’s going on behind the scenes at DHoAC.

If Mr Butler had chosen to disband the ADHA and stop the MHR, as he could have, he almost certainly would have been mired in a giant and very visible mess for at least the next year or so.

So maybe, just maybe, the government is being clever about this and it’s going to take its time working out how best to cut this very complex problem of transitioning from a $2.1 billion repository that doesn’t work and can’t ever work well, to something much more fit for purpose to underpin systemic reform in the future.