By Astrid Helene Mortensen and Niels Landbo Krogh

How do we determine whether a new drug works as intended? The answer lies in clinical trials—often large, complex, and expensive undertakings. But they don’t have to be necessarily. Emilie Højbjerre-Frandsen, a PhD student in Statistics at Aalborg University’s Department of Mathematical Sciences, is finalizing a research project that makes it possible to design trials with fewer participants and still maintain the same precision.

We spoke with Emilie about how this works in practice when conducting clinical, randomized studies within the pharmaceutical industry—and how she, as a curious student, ended up contributing to the extension of a method already getting acknowledgement from the European Medicines Agency.

How do you use mathematics in your PhD project?

"I work on how to use data from previous clinical trials to optimize new ones. It’s about increasing or holding the same level of precision—or ‘power,’ as we call it—of the study, thereby improving the chance of detecting an effect of the drug, if one exists. Using these tools we can reduce the number of participants while still obtaining reliable results."

What are the advantages of the method you’re developing?

"We can potentially bring medicine to patients faster because we’re able to demonstrate effectiveness more quickly in trials.  Our simulation studies show that the number of participants can be reduced by 10–20 percent," says Emilie Højbjerre-Frandsen.

"It may also lead to fewer dropouts, as more patients can receive the new treatment, which is believed to be better. Patients given placebo may have higher risk of dropping out if no effect of treatment is experienced."

The PhD student has analyzed data from diabetes patients, but the method can be applied broadly to other disease areas.

A Path to Meaningful Research

Have you seen growing interest in your method?

"Yes. Several regulatory agencies are starting to look at the method and have included it in some of their guidelines, indicating that it can be used. There’s also strong interest from the pharmaceutical industry, with whom I collaborate in my PhD project (Novo Nordisk)."

How did you end up working on this specific method?

"During my Master’s thesis, I worked with prognostic covariate adjustment in linear models. These are models using continuous data—like blood glucose levels. It was a new method at the time and hadn’t been peer-reviewed yet back then, which made it a compelling area to investigate. In my PhD project, I extended the method to include non-linear models—used for count data or binary outcomes. For example, how many times a patient with haemophilia bleeds, or if a patient goes from overweight to normal weight.

Tell us a bit about your background. Did you always know you wanted to work with statistics?

"No. I had some doubts during my studies, but in my fifth semester, I had a course in statistics, and it suddenly became more interesting for me. Then I got a student job at Novo Nordisk, where I saw how statistics could be used in real life—to help bring medicine to people."

How did your PhD project get started?

"I wrote my Master’s thesis with two others in collaboration with Novo Nordisk. Someone at the company had heard of the method but hadn’t had time to explore it much. We thought it was exciting. To start an industrial PhD, we applied for funding from Innovation Fund Denmark. While waiting for a response, I worked as a statistician at Novo Nordisk. Once we received the grant, I could begin the project," Emilie Højbjerre-Frandsen explains.

Could your approach be the main method in coming years?

"It depends on the context. The EMA (European Medicines Agency) has issued a so-called Qualification Opinion on an experimental method I’ve worked on and expanded. My extended version has the same properties, so in some clinical trials, it would have great potential to use it."