The power of a unified data infrastructure: Covid-19 and beyond

The new COVID algorithm can predict which people are at the greatest risk of developing complications after contracting coronavirus. By vaccinating these people first, a study has shown that the risk of hospital intakes could be cut in half after administering just one million vaccinations – potentially saving lives. So, what’s required to scale this data system across the Netherlands? And how can these efforts work to inspire the Amsterdam Economic Board’s goal of creating a unified healthcare data infrastructure?

Dr Ron Herings was applying large amounts of data to solve healthcare questions long before the term “big data” even existed. And, since the beginning of the COVID-19 pandemic, he’s been involved in developing the COVID DATA NL-consortium’s COVID Early Warning System, which unites the efforts of various scientists, hospitals, universities, research institutes and businesses. This quickly evolving infrastructure is updated daily using anonymised patient data to give insights into the current state of corona infections, hospitals intakes and deaths.

Meanwhile, the associated COVID algorithm calculates the chance of someone getting seriously ill from the coronavirus. A recent study showed that if those patients are fast-tracked for vaccination, the results would be dramatic. With the algorithm, it found that a vaccination rate of 8% (one million vaccines) would result in a 50% reduction in hospital admissions and mortality. Without the algorithm, the report states that the same results would only be achieved with a rate of 50% (8.5 million vaccines) in a random approach, and 24% (about 4 million vaccines) if people aged above 60 were prioritised. Currently undergoing a final field test, the system is almost ready to be rolled out.

Follow the numbers… What numbers?

As director of PHARMO Institute for Drug Outcomes Research and Professor in Pharmaco-Epidemiology and Healthcare Optimisation at Amsterdam UMC, Dr Herings has many titles. But, essentially, he sees himself as a “pharmaco-detective”. He says: “As investigators, we study adverse drug reactions, and circumstantial evidence of a drug being bad and then we try to find the proof that it’s true. The most important challenge the past 30 years was to enable access to the needed data, which is dramatically complex, time-consuming and costly in the Netherlands.”

In this quest, Dr Herings has become internationally recognised as a data pioneer – having linked and analysed countless research databases. “It all started as a student in Geneva where I wanted to do a study to find out how many people used a particular drug and whether they were cured by it,” Herings explains. “Nobody knew the numbers. This struck me as very strange. We spend all this money on healthcare but most often we don’t know if these interventions are safe and effective in real life.”

Little has changed since this time, Herings says. “We still don’t know how many people had a specific disease or really used a particular drug in the Netherlands and how many people are infected with SARS-Cov-2 virus? We can’t give you a proper number. There are also many questions that we need to answer in the future about the efficiency of vaccination and the effectiveness and safety of vaccines. So, we really need a daily-updated database to help patients, caregivers, authorities, government and health insurers with better-informed decisions.”

Zen and the art of connecting databases

Dr Herings has made it his business to find such numbers – and to connect these numbers with other relevant numbers to come up with answers. “You start with a research question, such as ‘How many people using oral contraception have cancer?’ The pharmacies know how many people take contraceptives. But for outcomes, like cancer, you need to go to the hospitals. So, you have to ask the hospitals to link their data to the pharmacy data. And that’s where it gets complicated: time, money, lawyers, privacy issues, conflict of interests, etcetera.”

“So maybe after a couple of years you’ve achieved this. But then you need to check if that cancer is a first manifestation or a recurring one – so we need to link to the cancer registry for a long period of time. And to find out more about type, staging and how bad that cancer was, we have also to link to the pathology registry.” This must all be done while adapting to the latest technologies and regulations.

“Of course, after all this, you want to keep the data for the next question. But you need a very good reason to keep this data due to privacy regulations. To be successful at this work, you really have to be a passionate idiot,” Herings laughs.

COVID-19 as a cutter of red tape. Or not

As a global emergency, COVID-19 has worked to cut a lot of red tape for healthcare innovations – just look at how quickly effective vaccines are being created and approved. But for Dr Herings and his team, there were no shortcuts. As a new disease, the coronavirus initially had no special code that could be easily tracked and traced – doctors simply noted it as a random text remark in their records. And with no tests at the start, there was also no way to truly confirm if someone had the virus.

“We sent a questionnaire to almost 500 GP practices in the Netherlands to ask whether they had a coronavirus patient, whether this patient was hospitalised or died, and how were they sure it was COVID-19. We could link this up to other registries to collect age, gender, underlying diseases, where they lived – and this is what we used to make the algorithm – a living algorithm to be updated as frequently as possible.”

Currently, the larger alert system is being implemented in Amsterdam, as well as across the provinces of North Holland and Flevoland. After a final field test in the second week of March, it should be ready for roll-out – enabling the involved doctors to, for example, start making their vaccination priority lists.

“It’s time to go with that banana”

So, what’s required to make the process all fall into place and scale across the Netherlands? “Commitment. Commitment on a political level. As the Dutch phrase goes: Gaan met die banaan! – it’s time to go with that banana!”

In the ideal scenario, all 8,000 GPs in the Netherlands will fill in questionnaires each day to record what’s going on in the 400,000 address locations around the Netherlands. We’ll be able to really see where there are people with COVID. You can then take measures: whether it’s shielding or vaccinating. It’s a no-brainer. Especially now we know the Corona app doesn’t work.”

Dr Herings knew that the app wouldn’t be very effective from the beginning. “It’s a basic fact in epidemiology that we need to focus, measure and learn from people who are ill. And a GP knows them best. A focus on counting people who could become ill is interesting but less useful to predict hospitalisations. With most of the COVID infections being asymptomatic and the rather random approach to testing, it’s simply impossible to know who will become ill. So, it’s very hard to effectively stop the spread. Our challenge is to stop complications – which is a totally different way of thinking.”

Beyond COVID-19: bigger, better infrastructures

Dr Herings is now only more convinced of the importance of developing a unified health data infrastructure. “On a technical level, we’re actually already there. It’s a Lego system. All the parts are there. Nothing needs to be invented,” he says.

“But what COVID-19 has taught us is we need the system already in place – whether it’s for COVID-20, Q fever, Epstein-Barr, or something else. And, by keeping it organised by region, we keep the data safe in terms of privacy. The Nordic countries are already doing it. The UK also has much better data access thanks to their NHS. I think the Netherlands can do it.”

“You don’t have to start from scratch and invent a new system. Use our data infrastructure and network approach. Just ask!” says Dr Herings. “And remember to give something back to those that provide the data. As with our COVID-19 system, we are providing a service to GPs: helping them identify their patients that are at the greatest risk. It’s a contractual exchange of data and knowledge.”

First Amsterdam, then the world

Herings sees Amsterdam as a great place to start. “As I said, it can be very frustrating when you have to build from nothing. So, you really need to have the support of the people in charge, those who dare to make decisions that unite different interests.”

“The Amsterdam Economic Board already has that support in place. They actually have the agreement with everyone stating that they are committing to setting up a regional data infrastructure. People also realise you have to spend money on it – and in fact it’s not very expensive yet very much needed. You can really create unique knowledge that can improve healthcare based on outcomes. And, most importantly, they recognise that you can do it fast but not in a week.”

Invitation to collaborate

COVID-19 is just one example. The potential benefits of the meaningful use of data are much greater. For example, we can use data to add value to healthcare, helping to prevent hospital admissions or overtreatments, for personalised medicine, to be quicker to diagnose conditions, and, in the case of many lifestyle diseases, we can focus on prevention so that citizens remain healthy  If you would like to contribute to the Health data infrastructure, contact Jeroen Maas, Lead Health at Amsterdam Economic Board. Read more about the Health data infrastructure initiative here.

#slimgroengezond

If you want to receive updates you are invited to join Smart Health Amsterdam. It is the network for data- and AI-driven innovation in Amsterdam’s life sciences and health sector. Join the community here.