Healthcare IT has been underfunded in Europe. Now the money is here, but the countries are not ready
This is the transcript of the Friday Weekly Healthtech room at Clubhouse titled“Google vs. OpenEHR in Solving Healthcare Interoperability”, which was organised as part of the Health Podcast Network Club.
In the session on 9 April, Better’s CEO Tomaž Gornik shared his views about data standards in healthcare, the digitalisation of healthcare in Europe, the entrance of Google into healthcare and the progress made in the last 10 years around healthcare data and interoperability.
Google is taking ambitious strides to make healthcare data sharable and searchable without changing the data structure. openEHR specifications, on the other hand, are based on open standards. When used, the basic infrastructure of the data is unified across institutions that use openEHR, and the content is separate from the technology. Perhaps we can start by learning why openEHR is dear to you. And apart from running Better, you’re also a member of the management board of the openEHR Foundation.
Better is a company that builds software for healthcare. Our mission is to help care teams in three different ways. The first one is to try to simplify their work. The second is to reduce the backlog of applications, meaning that we try to build tools to assemble applications quickly. Third, we believe in storing data in an open format, and we don’t compromise on that. Everything we do is based on openEHR.
Can you elaborate on what exactly it means for data to be built on open standards, because that can be misinterpreted and misunderstood in terms of who can access the data?
To clarify: we’re not talking about data being freely accessible, there are several legal issues tied to that. We’re talking about data being in a format that is open, published, and free to use. When you, as a healthcare provider, have the legal rights and the need to access data, you should be able to access it, understand it, and use it – without having to talk to your vendor. This is the biggest problem we are facing in healthcare IT today. Even though it belongs to them, healthcare providers are struggling to get their data and put it all together. Often, vendors want to charge for access to this data, or it can just be in a format that is not possible to understand without the vendor.
When it comes to interoperability, the first broad association is FHIR standards for the exchange of data. Can we clarify the difference between openEHR and FHIR? How broad is the use of open EHR data models? How many countries use openEHR?
openEHR development started more than 25 years ago. At first, it was pretty much an academic exercise. Then, about ten years ago, companies started realising the benefits of open standards. Today, we have over 2,000 clinical modellers from 100 different countries, and over 1,000 clinical models are currently in use and available for free. The key thing here is that these data models are being designed by clinicians.
The question about the difference between openEHR and FHIR is a part of every interoperability discussion today. FHIR is an evolution of HL7 standards, basically HL7 version four. HL7 standards have always been about interoperability, referring to operability between systems. With HL7, the data stays where it is, and when you need it, you access it through FHIR APIs. We don’t believe that is the right concept. I’m not saying that FHIR doesn’t have its uses, after all, we do use it for data exchange. But we are more in the camp of saying, “Look, if you’re going to store clinical data, there is no good reason to store it in a proprietary format unless you’re looking to lock the customer in for business reasons.” There’s no good technical reason why you wouldn’t store data in an open format from the start. And this is the big difference. OpenEHR is all about the persistence of data, whereas FHIR is about the exchange of data. Now, obviously, you need both, but I have never met a customer who didn’t want data in an open format. I disagree with the idea that “if we don’t need to move the source data, we can access it through APIs when needed.” We’ve been doing this for a very long time, and it doesn’t work all that well.
How do you see new interoperability initiatives for healthcare? The last 10 years have shown us, especially in the United States, that interoperability is very hard to achieve. But now Google is coming in with their solutions, such as their Care Studio. In essence, they are trying to convince hospitals to put their data into the cloud, where Google will allow them to search through specific patient data with their AI-based healthcare search engine. Thoughts?
Countries are different, as are their health systems. If I start with interoperability in the United States, where the health system is predominantly private, the institutions don’t want to give patients their data, as this protects them to a degree from patients going to another provider and other hospitals. The reasons for the lack of interoperability are business-related. It’s slightly different with large systems like Kaiser Permanente – inside that system, the data does flow. But when you try to go outside of a system, even if it’s a big one, you will always have issues, because the business implications of sharing data are not quite clear. Contrast that to many places in Europe, in the NHS for instance, where the trusts do not compete for patients. In Europe, the lack of interoperability is primarily a technical issue.
The last thing you mentioned was Google and the idea of not needing to structure data because you can put some AI on top of it. This concept of putting some AI on top of existing data and trying to make it somehow appear in the right format is quite attractive of course, because it’s automatic. You don’t need to do much about it. And, of course, Google has done a lot of great work here. I believe that in the future, AI will also be used to capture data, and it will mostly capture data in a structured format. Hence, the structure will be less and less of a problem going forward. The challenge is the legacy data that we already have in our systems and has to be somehow converted. You can do it with AI, and trust Google to do it for you, or you can spend a lot of effort doing it yourself. In the end, what you want is a consistent record about a patient from cradle to grave in a format that you can use. I think we will achieve that with a combination of the two approaches.
In Europe, I see that a lot of hospitals have issues trusting the mega-vendors with their health data – in some cases, it’s even illegal to do so. As I said, if you think about the future, most data will be collected from devices and sensors. This is already structured data, so there’s no issue. We will be doing speech-to-text and text-to-structure, and Al will definitely help with that. Systems will have to trust a data model which they understand. I think we still have a long way to go to get to the point where systems will be able to discern unstructured data on their own in the ways Google is demonstrating now.
You said that AI is going to capture structured data. How far away do you think we still are from that? Healthcare data capture is tricky. For example, the use of natural language processing for medication prescribing. Medication names are not so easy to understand, spell, and even pronounce. Hence, if models are not trained well enough, you can very quickly end up with the same problem that we have with paper, where a prescription can be illegible. Though in this case, it wouldn’t be illegible but just incorrect.
Accuracy is still a problem. Engines are getting better, but I don’t think we’re going to reach the stage anytime soon where they will be used without reserve. I think what’s going to happen is that the systems, the devices, and the sensors will prepare the data, and somebody will have to authorise the input before it goes into the system. In fact, for instance, we capture vitals data from ICU devices in our systems. But, instead of putting the data directly into the system, we present it to a nurse, and she or he just confirms that the data is OK. This raises another interesting question, which we don’t initially think about. If you automate everything and the data starts flowing into the systems, nobody’s going to watch over that, which means that things could go in the wrong direction very quickly.
In the beginning, I think that AI is going to be a helper more than a decision maker. Speech recognition is very strong in radiology where it can actually significantly improve data entry by providing very simple suggestions which pop up on the screen and can be selected in place of having to type whole sentences. But the issue is, of course, that this is a domain with well-defined terminology.
With the rise of technology, connectivity among devices, and automated data capture, cybersecurity is becoming increasingly important. How big is the hunger for healthcare data? We’ve already seen breaches happen in vaccine development.
I’m not an expert on security or cybersecurity, so I can’t comment on that. I don’t think we can avoid being attacked. Attacks will happen; we just need to somehow protect the data. It’s the same for self-driving cars, and for basically everything. I know there is a big interest in health data. And one of the things people don’t realise about health data is that, for instance, if somebody hacks into your bank account, you might lose some money, but you can open a new bank account, or maybe even transfer some of the money and keep on working. If somebody gets into your health data, you have information there that is immutable. Your genome, your allergies, sometimes chronic illnesses – all of these are permanent. I think this is part of the problem that makes health data different from data in other domains.
One more question regarding how much Google can actually achieve, given that healthcare institutions don’t trust the tech giants. Google is moving forward with respectable partnerships in the United States. The culture and the healthcare system there are different in terms of how they regulate, and how hungry they are for innovation when it’s weighted against things like privacy. Europe is very conservative in this department. To what extent do you think that the progress which might be made in the United States is going to be so appealing that it will put pressure on E.U. legislators to change their criteria of what matters and what doesn’t, which could open the door for Google in Europe?
This again is a broad topic. The power of the big five, or however you call them, is a problem not just for Europe but for the whole world. Even in the United States, we can see a lot of discussions about whether this is good for the economy, individuals, or society. It always goes back and forth. Initially, everybody was very excited about Google. It gave you stuff for free, or at least that’s how it looked. But then there was a backlash, and the momentum went the other way.
A lot of companies provide appealing technology. The clinical profession reacts to new innovation in the same excited way as consumers in other sectors, but they often don’t understand the broader implications. That is why we need regulators and people who understand what’s behind it all. Something might look fantastic for an end user, but all the implications – including what you mentioned about cybersecurity – are not self-evident. I won’t speculate on how this is going to turn out. I’ve seen countries start to think about health data as a strategic asset; a national security issue. And this changes the way you deal with the data. This is something I’ve seen with European governments.
Since I raised the question of values and differences in different countries... Among other places, Better is present in Russia, Slovenia, Malta, the United Kingdom, Germany, and is moving into the Spanish market. Do you see any differences in terms of the attitude towards data, data management, and how health data is perceived in these countries?
To be clear, we sell software – we don’t have anything to do with holding data. Our customers manage their own data. That’s an important distinction. But we absolutely see country-related differences. It all depends on whom people trust. It seems that in the West, it used to be that people trusted companies more than governments. In the East, I would say it used to be a little bit different. You would prefer to have a public entity hold your data. In Russia, by law, they cannot store data outside of their country, and that’s why they don’t have LinkedIn. And it goes without saying that the same is true for health data.
In Germany, most healthcare institutions do not trust cloud solutions. They’re lagging behind on the use of cloud technology because they want to keep their data on the premises. In the case of Germany, a lack of trust in centralised data repositories stems from historical reasons. I am generalising, but in Asia, for instance, nobody thinks about this and they’re more than happy to put their data on AWS.
Let’s also talk a little bit about the use of data platforms on national or regional levels to connect different institutions and enable centralised records. This is something we have in Slovenia, where patients can access their own discharge letters. But, of course, not all healthcare providers put data on the platform. From your experience, to what extent do these ideas have the backbones and connectivity data to share records that are still being further developed today? If we look at Australia and their national My Health Record project, not all providers are putting their data in the repository, which makes the care records incomplete. And if the records are incomplete, doctors don’t feel comfortable relying on the available data. The next question regarding connectivity and access to data is, can we as patients ever expect easy access to the data? Are patients going to be the custodians of their data by adding their medical records to a solution on their own? What do you see in terms of development?
There are two different questions here. The first one is: Is the data being shared, and can you access it? And the second one is: What is the amount, type, or quality of the data? For instance, Estonia is the poster child for succeeding in the first case, but not in the second: they actually share data, but their problem is that the data coming from the health institutions is actually very poor because their healthcare institutions are not very digital. So, while you may have a fantastic facility and engine, the source data may not be good enough to be very useful. Now, obviously, in single payer systems, where the payer is pushing digital, policies can be put in place where institutions don’t get paid unless they submit data. But, again, the big question here is the quality of the data. One way is through regulation. We’ve seen a lot of work being done in the United States demanding that patients have the ability to obtain all their data. In Europe, we have GDPR, which means that you as a patient can demand access to all your data. But in reality, we also have to understand that a lot of people don’t want to deal with this. They would like data fluidity, and that a certain physician, who is part of their care process, can have access to their data when he or she needs it. The fact that patients would be the custodians of their own data sounds logical, but if people actually have to do something for this, I don’t think it’s going to take off. Clearly, it depends on the type of patient. If you’re a cancer patient who has frequent contact with the healthcare system, it’s different, but if you are young and healthy, you will have less interest in this.
In many cases, I would say we will see a combination of approaches. To give you some good examples, Scandinavia has been doing this for 25 years. I think there are two different things to think about here. One is whether the patient has access. And, actually, the patient may be given access to their care provider. But then there’s the population health aspect as well, where you want policy, you want to research, and you want to have access to data to make big advances in care.
I think in most systems, especially the ones with a single payer, this can be done, and the patient has the right to revoke consent to certain clinicians and implicitly give consent to another clinician who is treating him or her. And this is a practical solution. The problem they had in France years ago, for instance, was that they spent a lot of money, hundreds of millions, and got only 50,000 people to sign up for the shared record, the Dossier. That was the consequence of an opt-in system where people had to do something to get access. In Slovenia, where we were involved, we took a different approach. You are included by default unless you want to opt out. Once systems started doing that, things improved dramatically. I think it’s really about informing people what the data will be used for, why it’s important to share it, how the data sharing is going to be managed and governed, and that it will not be widely shared without control. I also think we’re past this big thought that people are going to make money off their data by giving it to pharma, for example. I don’t think that’s going to happen. Today we already donate organs, and I don’t believe that we wouldn’t donate our data if we understood the reasons behind doing so.
We started our Friday weekly debates with that question: do patients really want their data? Being fully in charge of your data is a huge responsibility that you as a patient might not want to have. We also need to separate two ideas, which are: Do you own your data, or do you just want to have control over it?
The data is basically an abstraction, a view of you as either a person or your condition. So, I think people underestimate the importance of the accuracy of this data. Things can really go wrong if this data is not accurate. One thing is making it accessible, but the other is making sure it’s an accurate description of the state of your health.
In all the debates around data and interoperability, what is in your view on the progress that has been made in the last 10 years, and what do you anticipate is going to happen in the next 10 years with the rising amount of data coming into healthcare?
Unfortunately, I don’t think much progress has been made in the last 10 years, but I’m very optimistic about the next 10 years. I think COVID is a big part of that. If you think about COVID, it actually shrunk the next 10 years into one. Last year we saw the same amount of activity as in the last 10 years combined. COVID has really demonstrated the value of data, and the issues you can have without it. We can’t even make sound decisions on when to shut down the economy or start it up again, and these are huge decisions that are often still made blindly because there is not enough data available. So, I’m really optimistic about the future. But looking back, to be honest, I couldn’t say that any huge progress was made until the pandemic.
Can you elaborate on that a little bit more?
I’ll give you an example. A few years ago, I went into this meeting of people in medical informatics, and they were discussing how to store lab results. I thought to myself, “For the last 30 years we’ve been debating on how to store lab results. Don’t tell me this hasn’t been solved yet!” They were having issues storing lab results, which is the first thing that was digitised in healthcare and, to be honest, it happened 30 years ago. Large advances have been made in some institutions in certain countries, such as digitising hospitals in the United States, but if we’re talking about data exchange or sharing, which was your question, I didn’t see all that big of a difference. Of course, there are some countries and cities that went from paper to digital, like Moscow City, for instance. What I’m trying to say is that, in terms of awareness about the need for quality data, last year was game-changing.
This is also why with the entrance of tech giants, such as Google and Amazon, into the healthcare space, it is important to look at the successes they’ve had so far, despite being new entrants in the market.
We also saw them fail in many of these big initiatives.
The Heaven Project?
Yes. I’ll give you a very practical example. About 15 years ago, I put my team which was dealing with telecoms – really complex systems, high throughput, and a lot of transactions – into healthcare, and the initial thought was, this is going to be easy. We were doing telecoms, so how difficult could healthcare be? Well, if you ask any of them today, they will tell you: it is much more difficult. For example, there is the longevity of data, and the complexity of data. Also, a healthcare CEO has a hard time telling his users what to use. The solutions have to be useful, usable, user-friendly, and accepted by clinicians. This is different from other industries. Now, the reason I am optimistic is not just because COVID has changed things. I see a new generation of employees coming into healthcare. They are demanding a good user experience because this is what they are used to with the other digital tools they use. The same is true for patients. Finally, all across Europe, huge investments are coming to healthcare: 20 billion euros in Spain, 30 billion in France, and so on. This is for healthcare in general, but a part of this will be for digitalization. Most European countries are going to invest not millions, but billions in healthcare.
Since you’ve got some insight, can you elaborate a bit further? Do these countries have anything in common in terms of how they want to do things, because just putting money into the space is not enough. We’ve seen how the UK failed years ago when they tried to unify electronic health records by offering healthcare institutions a range of services that providers could choose from, but in the end, it still wasn’t enough for success.
That is a big worry – can countries use the money effectively? If you’re trying to say that sometimes too much money is worse than no money, I couldn’t agree with that, but there is a big concern that people will say we should just do something instead of nothing, so that we can collect the funding. I’ll give you an example from Catalonia. Catalonia has 12 biobanks for a population of 8 million people. The 12 biobanks exist because of different funding initiatives. Once the funding ran out, they started struggling to maintain these biobanks, because obviously, it would be more efficient to have just one or two. And it’s the same thing with this IT funding. It’s going to be critical to decide what type of solutions it will be used for, that these solutions are actually sustainable in the future – when there will be a lot less funding. I don’t think we have time for this discussion because it’s a really long one, but suffice it to say, it’s a really big challenge for Europe. Most of these countries are not ready, and they are being forced, as we know in Slovenia as well, to define strategies and projects. And it’s a struggle. I’m sure the big consultancies will take a large part of that money. But what will be the long-term outcome? I’m not sure.
In a way, you might have answered the last question that I still have. When it comes to innovation and technology adoption, Europe is often seen as the laggard. When looking for inspiration, everybody turns to the United States, because it’s the largest healthcare market. China’s progress in AI is impressive, but it’s a different market with very different regulations. In your view, are the critiques of Europe regarding the speed of change justifiable?
You have to look at the United States holistically. There are pockets of fantastic innovation where things are being done really quickly. But there are also pockets that look like a disaster; basically barren land in a digital sense, with nothing. The United States represents 50% of global healthcare spending, and 80% of global health IT spending. So, are they using the money efficiently? It is clear that is not the case.
Europe, on the other hand, was underfunded, except for a few countries. Whether Europe is capable of doing what needs to be done is another question. We’re seeing examples of exemplary projects from Catalonia, which I believe has a really good vision and digital strategy. We see Scandinavia, maybe not going as fast as people thought they would, but definitely going in the right direction. Culturally, they have this sharing mentality which really helps in this case.
Unfortunately, I don’t think this is easy to do at the European level. In large countries, change is happening at regional levels. I think improvements will be made in many cases, some of them will be great, some of them will be just a start. For years, health IT was underfunded compared to IT spending in other industries. This funding will now come, but we’re not ready. I always point out the NHS digital Academy, which trained 500 CCIOs and CIOs to understand the concepts, what needs to be done and how to effect change. This was a fantastic effort. When we now go to market, we can actually talk to people who understand the new approaches, which is not the case in a lot of other countries. Things are very different across geographies.