Swedish registers offer unique opportunities for epidemiological research

It is Monday morning and, unusually, Jonas F. Ludvigsson is working from home. He woke up with a runny nose and decided not to take the early morning commute to Stockholm and Karolinska Institutet.

“I don’t want to infect my fellow passengers on buses and trains. Since COVID, I’ve become more careful about things like that,” he says.

During the pandemic, he became a familiar face to many who followed the news coverage and the debate on restrictions and lockdowns. To another audience, he is perhaps best known as the TV doctor on Malou on Swedish TV4. But in his day-to-day work however, Jonas F. Ludvigsson is a paediatrician at Örebro University Hospital and Professor of Epidemiology at Karolinska Institutet.

Much of his research has focused on gluten intolerance, i.e. coeliac disease, but it also spans several other areas.

“My research mainly focuses on three areas: gastrointestinal diseases, paediatric conditions, and pregnancy. Sometimes two or even all three areas are included in the same study, but not always,” he says.

That epidemiology became Jonas F. Ludvigsson’s field was partly a matter of chance, he explains. He completed his doctoral thesis at Linköping University on the ABIS cohort (All Children in Southeast Sweden). The opponent at his public thesis defence was Professor Anders Ekbom from Karolinska Institutet.

“Anders was a remarkable person who later became my mentor for a quarter of a century. He had enormous enthusiasm for epidemiology and was able to pass that enthusiasm on to others. It is largely thanks to him that I came to enjoy this field so much and chose to remain in it,” says Jonas F. Ludvigsson.

Sweden a world leader

Today, he is himself the senior professor seeking to inspire new generations of researchers to take an interest in register-based research. When lecturing madical students, he often emphasises how important epidemiology is for developing and improving healthcare.

“I actually wrote an opinion piece about it just this morning.”

He is also keen to convey to students and colleagues just how favourable Sweden’s conditions are for epidemiological research, with its system of personal identity numbers and high-quality population registers.

“Epidemiology is one of the few fields of research in which Sweden is a world leader. As a researcher, it is both rewarding and advantageous to work in such a field,” he says. “It makes it easier to publish in leading journals, secure research funding ang gain attention for your results.”

“Today, Denmark and Norway are at a similar level. However, Sweden has roughly twice the population, which still gives us a competitive advantage,” he adds with a smile.

Which registers have been most important for your research?

“The ones I’ve used most are the National Patient Register, the Cause of Death Register, the Cancer Register, the Medical Birth Register, the Prescribed Drug Register and, to some extent, the Military Conscription Register. In addition, Statistics Sweden’s LISA database – the Longitudinal Integrated Database for Health Insurance and Labour Market Studies – which allows us to supplement the data with socioeconomic variables such as education and, in some cases income.”

What makes Swedish registers so uniquely valuable?

“There are several reasons. First and foremost, they cover the entire population, which is important when drawing conclusions. There are very large datasetsin other countries as well, but they are usually not population-based, which increases the risk of selection bias affecting the results.

Secondly, Swedish registers extend far back in time, Ludvigsson emphasises. Instead of just a few years of follow-up, researchers can access several decades of data or, in the case of the Multi-Generation Register, nearly a century.

“Thirdly, they are based on the personal identity number, which enables straightforward linkage between different registers. This also make it possible to follow individuals over time, ensuring they are not lost to follow-up if, for example, they move. We can also link people to their parents, siblings and children, and study heredity in a way that is not possible in many other countries.

“Fourthly, data collection is carried out by government agencies,” he adds. “This means the data are not influenced by individual researchers’ hypotheses, as the organisations collecting the data do not know how they will be used. The centralised data collection also reduces costs dramatically.”

“It might cost me SEK 100,000 to 150,000 to commission a data extract from Statistics Sweden or the National Board of Health and Welfare. If I had funded the collection of these data myself, even SEK 150 million would not have been sufficient. So it is a highly cost-effective way of conducting research,” he says.

Advice for beginners

What practical advice would you give to someone new to epidemiology? What do they need to bear in mind?

“I discuss this in some detail in my book Getting Started in Medical and Health Research. I’ve tried to include everything I wish someone had explained to me when I was a doctoral student. A new edition of the book was published recently.”

One very concrete piece of advice for early-career epidemiologists is to carefully check the code submitted to data holders for processing. Jonas F. Ludvigsson learned this the hard way.

“It was one of the first times I ordered data from the National Board of Health and Welfare. I made a single-character eroor – a full stop instead of a comma in a diagnosis code – which meant the extraction missed thousands of individuals. I only discovered the mistake several months later, and everything had to be redone. When you have paid SEK 100,000 a second time to obtain a correct dataset, you learn to be careful with codes.”

Another piece of advice is to be awareof the time coverage of different registers. Sweden’s register tradition may be long-standing, but individual registers are sometimes relatively new.

“The Prescribed Drug Register was established in July 2005 – only 20 years ago. If you want to study a disease historically over several decades, t that source is not suitable.”

Sometimes the level of detail in the registers is not sufficient. In such cases, some researchers choose to establish their own cohorts. Jonas F. Ludvigsson leads the ESPRESSO cohort, Epidemiology Strengthened by histoPathology Reports in Sweden.

“My view is that you can go very far using only national register data. Epidemiologists do not necessarily need to create their own cohorts. However, if you plan to conduct extensive research and have a strong interest in something not captured in standard registers, there may be good reason to do so. Tissue is one such interest for me. What does intestinal tissue look like in disease? Is it inflamed? That type of information is available in the ESPRESSO cohort, which I have linked to national registers. It is this combined dataset that underpins most of my current research,” he says.

International collaboration

Jonas F. Ludvigsson has described the advantages of Swedish register-based research to an international audience, including in a widely cited article on the Swedish personal identity number. He is also frequently asked by colleagues abroad about the possibility of accessing Swedish register data.

“I encountered such a case only recently: could we contribute Swedish data to an international clinical study? Unfortunately, I had to say no – we cannot transfer Swedish data abroad.”

However, international researchers can still make use of Swedish data, he points out.

“There are two main approaches. Either foreign researchers come to Sweden and analyse the data on site, or we conduct the analyses here on their behalf for a fee. We do that quite often. They then receive only the results, while all individual-level data remain in Sweden.”

A straightforward study design may sometimes be sufficient to answer an epidemiological research question. At other times, more advanced methods are required. One example that Jonas F. Ludvigsson highlights is a study examining the effects of swine flu vaccination during pregnancy on long-term health outcomes in offspring.

“There were strong emotions surrounding swine flu vaccination during pregnancy. Many people were worried, so it was important to provide answers. At the same time, we know there are systematic differences between women who choose to be vaccinated and those who do not. These groups cannot simply be compared directly. To address this, we compared siblings born before and after their mother was vaccinated. I think that is an elegant solution.”

Today, Jonas F. Ludvigsson is involved to varying degrees in many other researchers’ projects, ranging from cancer to cardiovascular disease and neurology. His role is usually to act as an epidemiological adviser, suggesting ways to minimise bias and confounding.

Finally, looking ahead, how do you see the future of epidemiology?

“I recently gave a lecture on that – it’s a pity you weren’t there! The main point I emphasised most strongly was the importance of involving clinicians, those who work with patients on a daily basis. It is easy to assume that epidemiological and biostatistical expertise within a research team is sufficient, but there is then a considerable risk of asking the wrong questions or selecting inappropriate variables. For results to be meaningful, research teams must be grounded in clinical reality.”

Text: Anders Nilsson